commit 8a8da1872199f1309243b13e107db5cf4751c9f0 Author: ModelHub XC Date: Sat Jun 13 09:45:16 2026 +0800 初始化项目,由ModelHub XC社区提供模型 Model: boradorish/qwen3-8b-finetuned-train Source: Original Platform diff --git a/.gitattributes b/.gitattributes new file mode 100644 index 0000000..52373fe --- /dev/null +++ b/.gitattributes @@ -0,0 +1,36 @@ +*.7z filter=lfs diff=lfs merge=lfs -text +*.arrow filter=lfs diff=lfs merge=lfs -text +*.bin filter=lfs diff=lfs merge=lfs -text +*.bz2 filter=lfs diff=lfs merge=lfs -text +*.ckpt filter=lfs diff=lfs merge=lfs -text +*.ftz filter=lfs diff=lfs merge=lfs -text +*.gz filter=lfs diff=lfs merge=lfs -text +*.h5 filter=lfs diff=lfs merge=lfs -text +*.joblib filter=lfs diff=lfs merge=lfs -text +*.lfs.* filter=lfs diff=lfs merge=lfs -text +*.mlmodel filter=lfs diff=lfs merge=lfs -text +*.model filter=lfs diff=lfs merge=lfs -text +*.msgpack filter=lfs diff=lfs merge=lfs -text +*.npy filter=lfs diff=lfs merge=lfs -text +*.npz filter=lfs diff=lfs merge=lfs -text +*.onnx filter=lfs diff=lfs merge=lfs -text +*.ot filter=lfs diff=lfs merge=lfs -text +*.parquet filter=lfs diff=lfs merge=lfs -text +*.pb filter=lfs diff=lfs merge=lfs -text +*.pickle filter=lfs diff=lfs merge=lfs -text +*.pkl filter=lfs diff=lfs merge=lfs -text +*.pt filter=lfs diff=lfs merge=lfs -text +*.pth filter=lfs diff=lfs merge=lfs -text +*.rar filter=lfs diff=lfs merge=lfs -text +*.safetensors filter=lfs diff=lfs merge=lfs -text +saved_model/**/* filter=lfs diff=lfs merge=lfs -text +*.tar.* filter=lfs diff=lfs merge=lfs -text +*.tar filter=lfs diff=lfs merge=lfs -text +*.tflite filter=lfs diff=lfs merge=lfs -text +*.tgz filter=lfs diff=lfs merge=lfs -text +*.wasm filter=lfs diff=lfs merge=lfs -text +*.xz filter=lfs diff=lfs merge=lfs -text +*.zip filter=lfs diff=lfs merge=lfs -text +*.zst filter=lfs diff=lfs merge=lfs -text +*tfevents* filter=lfs diff=lfs merge=lfs -text +tokenizer.json filter=lfs diff=lfs merge=lfs -text diff --git a/README.md b/README.md new file mode 100644 index 0000000..9f7438e --- /dev/null +++ b/README.md @@ -0,0 +1,61 @@ +--- +library_name: transformers +license: other +base_model: Qwen/Qwen3-8B +tags: +- llama-factory +- full +- generated_from_trainer +model-index: +- name: sft + results: [] +--- + + + +# sft + +This model is a fine-tuned version of [Qwen/Qwen3-8B](https://huggingface.co/Qwen/Qwen3-8B) on the sunny_reasoning dataset. + +## Model description + +More information needed + +## Intended uses & limitations + +More information needed + +## Training and evaluation data + +More information needed + +## Training procedure + +### Training hyperparameters + +The following hyperparameters were used during training: +- learning_rate: 2e-05 +- train_batch_size: 1 +- eval_batch_size: 8 +- seed: 42 +- distributed_type: multi-GPU +- num_devices: 2 +- gradient_accumulation_steps: 4 +- total_train_batch_size: 8 +- total_eval_batch_size: 16 +- optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments +- lr_scheduler_type: cosine +- lr_scheduler_warmup_steps: 0.1 +- num_epochs: 3.0 + +### Training results + + + +### Framework versions + +- Transformers 5.2.0 +- Pytorch 2.9.1+cu130 +- Datasets 4.0.0 +- Tokenizers 0.22.2 diff --git a/all_results.json b/all_results.json new file mode 100644 index 0000000..38a9764 --- /dev/null +++ b/all_results.json @@ -0,0 +1,8 @@ +{ + "epoch": 3.0, + "total_flos": 4.717303430077481e+17, + "train_loss": 0.15186350211219263, + "train_runtime": 38487.8681, + "train_samples_per_second": 1.115, + "train_steps_per_second": 0.139 +} \ No newline at end of file diff --git a/chat_template.jinja b/chat_template.jinja new file mode 100644 index 0000000..01be9b3 --- /dev/null +++ b/chat_template.jinja @@ -0,0 +1,89 @@ +{%- if tools %} + {{- '<|im_start|>system\n' }} + {%- if messages[0].role == 'system' %} + {{- messages[0].content + '\n\n' }} + {%- endif %} + {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n" }} + {%- for tool in tools %} + {{- "\n" }} + {{- tool | tojson }} + {%- endfor %} + {{- "\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n<|im_end|>\n" }} +{%- else %} + {%- if messages[0].role == 'system' %} + {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }} + {%- endif %} +{%- endif %} +{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %} +{%- for message in messages[::-1] %} + {%- set index = (messages|length - 1) - loop.index0 %} + {%- if ns.multi_step_tool and message.role == "user" and message.content is string and not(message.content.startswith('') and message.content.endswith('')) %} + {%- set ns.multi_step_tool = false %} + {%- set ns.last_query_index = index %} + {%- endif %} +{%- endfor %} +{%- for message in messages %} + {%- if message.content is string %} + {%- set content = message.content %} + {%- else %} + {%- set content = '' %} + {%- endif %} + {%- if (message.role == "user") or (message.role == "system" and not loop.first) %} + {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }} + {%- elif message.role == "assistant" %} + {%- set reasoning_content = '' %} + {%- if message.reasoning_content is string %} + {%- set reasoning_content = message.reasoning_content %} + {%- else %} + {%- if '' in content %} + {%- set reasoning_content = content.split('')[0].rstrip('\n').split('')[-1].lstrip('\n') %} + {%- set content = content.split('')[-1].lstrip('\n') %} + {%- endif %} + {%- endif %} + {%- if loop.index0 > ns.last_query_index %} + {%- if loop.last or (not loop.last and reasoning_content) %} + {{- '<|im_start|>' + message.role + '\n\n' + reasoning_content.strip('\n') + '\n\n\n' + content.lstrip('\n') }} + {%- else %} + {{- '<|im_start|>' + message.role + '\n' + content }} + {%- endif %} + {%- else %} + {{- '<|im_start|>' + message.role + '\n' + content }} + {%- endif %} + {%- if message.tool_calls %} + {%- for tool_call in message.tool_calls %} + {%- if (loop.first and content) or (not loop.first) %} + {{- '\n' }} + {%- endif %} + {%- if tool_call.function %} + {%- set tool_call = tool_call.function %} + {%- endif %} + {{- '\n{"name": "' }} + {{- tool_call.name }} + {{- '", "arguments": ' }} + {%- if tool_call.arguments is string %} + {{- tool_call.arguments }} + {%- else %} + {{- tool_call.arguments | tojson }} + {%- endif %} + {{- '}\n' }} + {%- endfor %} + {%- endif %} + {{- '<|im_end|>\n' }} + {%- elif message.role == "tool" %} + {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %} + {{- '<|im_start|>user' }} + {%- endif %} + {{- '\n\n' }} + {{- content }} + {{- '\n' }} + {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %} + {{- '<|im_end|>\n' }} + {%- endif %} + {%- endif %} +{%- endfor %} +{%- if add_generation_prompt %} + {{- '<|im_start|>assistant\n' }} + {%- if enable_thinking is defined and enable_thinking is false %} + {{- '\n\n\n\n' }} + {%- endif %} +{%- endif %} \ No newline at end of file diff --git a/config.json b/config.json new file mode 100644 index 0000000..a8e7a96 --- /dev/null +++ b/config.json @@ -0,0 +1,71 @@ +{ + "architectures": [ + "Qwen3ForCausalLM" + ], + "attention_bias": false, + "attention_dropout": 0.0, + "bos_token_id": null, + "dtype": "bfloat16", + "eos_token_id": 151645, + "head_dim": 128, + "hidden_act": "silu", + "hidden_size": 4096, + "initializer_range": 0.02, + "intermediate_size": 12288, + "layer_types": [ + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention", + "full_attention" + ], + "max_position_embeddings": 40960, + "max_window_layers": 36, + "model_type": "qwen3", + "num_attention_heads": 32, + "num_hidden_layers": 36, + "num_key_value_heads": 8, + "pad_token_id": 151643, + "rms_norm_eps": 1e-06, + "rope_parameters": { + "rope_theta": 1000000, + "rope_type": "default" + }, + "sliding_window": null, + "tie_word_embeddings": false, + "transformers_version": "5.2.0", + "use_cache": false, + "use_sliding_window": false, + "vocab_size": 151936 +} diff --git a/generation_config.json b/generation_config.json new file mode 100644 index 0000000..1701c94 --- /dev/null +++ b/generation_config.json @@ -0,0 +1,12 @@ +{ + "do_sample": true, + "eos_token_id": [ + 151645, + 151643 + ], + "pad_token_id": 151643, + "temperature": 0.6, + "top_k": 20, + "top_p": 0.95, + "transformers_version": "5.2.0" +} diff --git a/model.safetensors b/model.safetensors new file mode 100644 index 0000000..7eb8acc --- /dev/null +++ b/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f7ad827c7864c69f3c4510ea22b3191c37f24162f70ec57f5cf31128b69e191d +size 16381517208 diff --git a/tokenizer.json b/tokenizer.json new file mode 100644 index 0000000..c7afbed --- /dev/null +++ b/tokenizer.json @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:be75606093db2094d7cd20f3c2f385c212750648bd6ea4fb2bf507a6a4c55506 +size 11422650 diff --git a/tokenizer_config.json b/tokenizer_config.json new file mode 100644 index 0000000..145e2c7 --- /dev/null +++ b/tokenizer_config.json @@ -0,0 +1,30 @@ +{ + "add_prefix_space": false, + "backend": "tokenizers", + "bos_token": null, + "clean_up_tokenization_spaces": false, + "eos_token": "<|im_end|>", + "errors": "replace", + "extra_special_tokens": [ + "<|im_start|>", + "<|im_end|>", + "<|object_ref_start|>", + "<|object_ref_end|>", + "<|box_start|>", + "<|box_end|>", + "<|quad_start|>", + "<|quad_end|>", + "<|vision_start|>", + "<|vision_end|>", + "<|vision_pad|>", + "<|image_pad|>", + "<|video_pad|>" + ], + "is_local": false, + "model_max_length": 131072, + "pad_token": "<|endoftext|>", + "padding_side": "right", + "split_special_tokens": false, + "tokenizer_class": "Qwen2Tokenizer", + "unk_token": null +} diff --git a/train_results.json b/train_results.json new file mode 100644 index 0000000..38a9764 --- /dev/null +++ b/train_results.json @@ -0,0 +1,8 @@ +{ + "epoch": 3.0, + "total_flos": 4.717303430077481e+17, + "train_loss": 0.15186350211219263, + "train_runtime": 38487.8681, + "train_samples_per_second": 1.115, + "train_steps_per_second": 0.139 +} \ No newline at end of file diff --git a/trainer_log.jsonl b/trainer_log.jsonl new file mode 100644 index 0000000..2c9613e --- /dev/null +++ b/trainer_log.jsonl @@ -0,0 +1,5365 @@ +{"current_steps": 1, "total_steps": 5364, "loss": 0.1708594113588333, "lr": 0.0, "epoch": 0.0005594405594405594, "percentage": 0.02, "elapsed_time": "0:00:09", "remaining_time": "14:02:42"} +{"current_steps": 2, "total_steps": 5364, "loss": 0.2891339659690857, "lr": 3.7243947858473e-08, "epoch": 0.0011188811188811189, "percentage": 0.04, "elapsed_time": "0:00:16", "remaining_time": "12:06:20"} +{"current_steps": 3, "total_steps": 5364, "loss": 0.4187517762184143, "lr": 7.4487895716946e-08, "epoch": 0.0016783216783216783, "percentage": 0.06, "elapsed_time": "0:00:23", "remaining_time": "11:25:29"} +{"current_steps": 4, "total_steps": 5364, "loss": 0.13779963552951813, "lr": 1.11731843575419e-07, "epoch": 0.0022377622377622378, "percentage": 0.07, "elapsed_time": "0:00:30", "remaining_time": "11:20:06"} +{"current_steps": 5, "total_steps": 5364, "loss": 0.31178635358810425, "lr": 1.48975791433892e-07, "epoch": 0.002797202797202797, "percentage": 0.09, "elapsed_time": "0:00:37", "remaining_time": "11:06:55"} +{"current_steps": 6, "total_steps": 5364, "loss": 0.31994569301605225, "lr": 1.8621973929236503e-07, "epoch": 0.0033566433566433566, "percentage": 0.11, "elapsed_time": "0:00:44", "remaining_time": "10:58:29"} +{"current_steps": 7, "total_steps": 5364, "loss": 0.3239685893058777, "lr": 2.23463687150838e-07, "epoch": 0.003916083916083916, "percentage": 0.13, "elapsed_time": "0:00:52", "remaining_time": "11:03:44"} +{"current_steps": 8, "total_steps": 5364, "loss": 0.40604960918426514, "lr": 2.60707635009311e-07, "epoch": 0.0044755244755244755, "percentage": 0.15, "elapsed_time": "0:00:58", "remaining_time": "10:56:45"} +{"current_steps": 9, "total_steps": 5364, "loss": 0.26543134450912476, "lr": 2.97951582867784e-07, "epoch": 0.005034965034965035, "percentage": 0.17, "elapsed_time": "0:01:05", "remaining_time": "10:52:57"} +{"current_steps": 10, "total_steps": 5364, "loss": 0.15973007678985596, "lr": 3.3519553072625703e-07, "epoch": 0.005594405594405594, "percentage": 0.19, "elapsed_time": "0:01:13", "remaining_time": "10:55:53"} +{"current_steps": 11, "total_steps": 5364, "loss": 0.24497920274734497, "lr": 3.7243947858473006e-07, "epoch": 0.006153846153846154, "percentage": 0.21, "elapsed_time": "0:01:20", "remaining_time": "10:55:06"} +{"current_steps": 12, "total_steps": 5364, "loss": 0.507112443447113, "lr": 4.09683426443203e-07, "epoch": 0.006713286713286713, "percentage": 0.22, "elapsed_time": "0:01:27", "remaining_time": "10:54:02"} +{"current_steps": 13, "total_steps": 5364, "loss": 0.36340224742889404, "lr": 4.46927374301676e-07, "epoch": 0.007272727272727273, "percentage": 0.24, "elapsed_time": "0:01:35", "remaining_time": "10:56:43"} +{"current_steps": 14, "total_steps": 5364, "loss": 0.29563137888908386, "lr": 4.84171322160149e-07, "epoch": 0.007832167832167832, "percentage": 0.26, "elapsed_time": "0:01:42", "remaining_time": "10:55:15"} +{"current_steps": 15, "total_steps": 5364, "loss": 0.3542381227016449, "lr": 5.21415270018622e-07, "epoch": 0.008391608391608392, "percentage": 0.28, "elapsed_time": "0:01:50", "remaining_time": "10:55:44"} +{"current_steps": 16, "total_steps": 5364, "loss": 0.2901057004928589, "lr": 5.58659217877095e-07, "epoch": 0.008951048951048951, "percentage": 0.3, "elapsed_time": "0:01:57", "remaining_time": "10:53:09"} +{"current_steps": 17, "total_steps": 5364, "loss": 0.3724617660045624, "lr": 5.95903165735568e-07, "epoch": 0.00951048951048951, "percentage": 0.32, "elapsed_time": "0:02:05", "remaining_time": "10:58:00"} +{"current_steps": 18, "total_steps": 5364, "loss": 0.20881980657577515, "lr": 6.331471135940411e-07, "epoch": 0.01006993006993007, "percentage": 0.34, "elapsed_time": "0:02:12", "remaining_time": "10:55:06"} +{"current_steps": 19, "total_steps": 5364, "loss": 0.1372430920600891, "lr": 6.703910614525141e-07, "epoch": 0.01062937062937063, "percentage": 0.35, "elapsed_time": "0:02:19", "remaining_time": "10:55:06"} +{"current_steps": 20, "total_steps": 5364, "loss": 0.4805028736591339, "lr": 7.076350093109869e-07, "epoch": 0.011188811188811189, "percentage": 0.37, "elapsed_time": "0:02:26", "remaining_time": "10:52:15"} +{"current_steps": 21, "total_steps": 5364, "loss": 0.28077012300491333, "lr": 7.448789571694601e-07, "epoch": 0.011748251748251748, "percentage": 0.39, "elapsed_time": "0:02:33", "remaining_time": "10:52:14"} +{"current_steps": 22, "total_steps": 5364, "loss": 0.3277529776096344, "lr": 7.82122905027933e-07, "epoch": 0.012307692307692308, "percentage": 0.41, "elapsed_time": "0:02:40", "remaining_time": "10:51:21"} +{"current_steps": 23, "total_steps": 5364, "loss": 0.4050637483596802, "lr": 8.19366852886406e-07, "epoch": 0.012867132867132867, "percentage": 0.43, "elapsed_time": "0:02:47", "remaining_time": "10:50:00"} +{"current_steps": 24, "total_steps": 5364, "loss": 0.22474241256713867, "lr": 8.566108007448791e-07, "epoch": 0.013426573426573427, "percentage": 0.45, "elapsed_time": "0:02:54", "remaining_time": "10:48:15"} +{"current_steps": 25, "total_steps": 5364, "loss": 0.20507575571537018, "lr": 8.93854748603352e-07, "epoch": 0.013986013986013986, "percentage": 0.47, "elapsed_time": "0:03:01", "remaining_time": "10:46:34"} +{"current_steps": 26, "total_steps": 5364, "loss": 0.26187247037887573, "lr": 9.31098696461825e-07, "epoch": 0.014545454545454545, "percentage": 0.48, "elapsed_time": "0:03:08", "remaining_time": "10:44:58"} +{"current_steps": 27, "total_steps": 5364, "loss": 0.28988659381866455, "lr": 9.68342644320298e-07, "epoch": 0.015104895104895105, "percentage": 0.5, "elapsed_time": "0:03:15", "remaining_time": "10:43:27"} +{"current_steps": 28, "total_steps": 5364, "loss": 0.5565254092216492, "lr": 1.005586592178771e-06, "epoch": 0.015664335664335664, "percentage": 0.52, "elapsed_time": "0:03:22", "remaining_time": "10:42:06"} +{"current_steps": 29, "total_steps": 5364, "loss": 0.3537827432155609, "lr": 1.042830540037244e-06, "epoch": 0.016223776223776225, "percentage": 0.54, "elapsed_time": "0:03:29", "remaining_time": "10:42:03"} +{"current_steps": 30, "total_steps": 5364, "loss": 0.33376017212867737, "lr": 1.080074487895717e-06, "epoch": 0.016783216783216783, "percentage": 0.56, "elapsed_time": "0:03:36", "remaining_time": "10:42:35"} +{"current_steps": 31, "total_steps": 5364, "loss": 0.3507142663002014, "lr": 1.11731843575419e-06, "epoch": 0.017342657342657344, "percentage": 0.58, "elapsed_time": "0:03:43", "remaining_time": "10:41:21"} +{"current_steps": 32, "total_steps": 5364, "loss": 0.11607243120670319, "lr": 1.1545623836126631e-06, "epoch": 0.017902097902097902, "percentage": 0.6, "elapsed_time": "0:03:51", "remaining_time": "10:42:19"} +{"current_steps": 33, "total_steps": 5364, "loss": 0.24526657164096832, "lr": 1.191806331471136e-06, "epoch": 0.018461538461538463, "percentage": 0.62, "elapsed_time": "0:03:58", "remaining_time": "10:42:24"} +{"current_steps": 34, "total_steps": 5364, "loss": 0.35029327869415283, "lr": 1.229050279329609e-06, "epoch": 0.01902097902097902, "percentage": 0.63, "elapsed_time": "0:04:05", "remaining_time": "10:41:21"} +{"current_steps": 35, "total_steps": 5364, "loss": 0.2289755493402481, "lr": 1.2662942271880822e-06, "epoch": 0.019580419580419582, "percentage": 0.65, "elapsed_time": "0:04:12", "remaining_time": "10:40:42"} +{"current_steps": 36, "total_steps": 5364, "loss": 0.3760473132133484, "lr": 1.303538175046555e-06, "epoch": 0.02013986013986014, "percentage": 0.67, "elapsed_time": "0:04:19", "remaining_time": "10:39:41"} +{"current_steps": 37, "total_steps": 5364, "loss": 0.3920930325984955, "lr": 1.3407821229050281e-06, "epoch": 0.0206993006993007, "percentage": 0.69, "elapsed_time": "0:04:26", "remaining_time": "10:39:15"} +{"current_steps": 38, "total_steps": 5364, "loss": 0.22879764437675476, "lr": 1.378026070763501e-06, "epoch": 0.02125874125874126, "percentage": 0.71, "elapsed_time": "0:04:33", "remaining_time": "10:38:31"} +{"current_steps": 39, "total_steps": 5364, "loss": 0.09033755958080292, "lr": 1.4152700186219739e-06, "epoch": 0.02181818181818182, "percentage": 0.73, "elapsed_time": "0:04:40", "remaining_time": "10:38:49"} +{"current_steps": 40, "total_steps": 5364, "loss": 0.46498849987983704, "lr": 1.4525139664804472e-06, "epoch": 0.022377622377622378, "percentage": 0.75, "elapsed_time": "0:04:47", "remaining_time": "10:38:42"} +{"current_steps": 41, "total_steps": 5364, "loss": 0.3834105134010315, "lr": 1.4897579143389202e-06, "epoch": 0.02293706293706294, "percentage": 0.76, "elapsed_time": "0:04:54", "remaining_time": "10:37:52"} +{"current_steps": 42, "total_steps": 5364, "loss": 0.45146605372428894, "lr": 1.5270018621973931e-06, "epoch": 0.023496503496503496, "percentage": 0.78, "elapsed_time": "0:05:01", "remaining_time": "10:37:01"} +{"current_steps": 43, "total_steps": 5364, "loss": 0.3595430850982666, "lr": 1.564245810055866e-06, "epoch": 0.024055944055944058, "percentage": 0.8, "elapsed_time": "0:05:08", "remaining_time": "10:36:16"} +{"current_steps": 44, "total_steps": 5364, "loss": 0.14026623964309692, "lr": 1.601489757914339e-06, "epoch": 0.024615384615384615, "percentage": 0.82, "elapsed_time": "0:05:15", "remaining_time": "10:36:14"} +{"current_steps": 45, "total_steps": 5364, "loss": 0.2923401892185211, "lr": 1.638733705772812e-06, "epoch": 0.025174825174825177, "percentage": 0.84, "elapsed_time": "0:05:22", "remaining_time": "10:35:27"} +{"current_steps": 46, "total_steps": 5364, "loss": 0.263121098279953, "lr": 1.675977653631285e-06, "epoch": 0.025734265734265734, "percentage": 0.86, "elapsed_time": "0:05:29", "remaining_time": "10:35:12"} +{"current_steps": 47, "total_steps": 5364, "loss": 0.29029422998428345, "lr": 1.7132216014897581e-06, "epoch": 0.026293706293706295, "percentage": 0.88, "elapsed_time": "0:05:36", "remaining_time": "10:34:38"} +{"current_steps": 48, "total_steps": 5364, "loss": 0.07528115808963776, "lr": 1.7504655493482312e-06, "epoch": 0.026853146853146853, "percentage": 0.89, "elapsed_time": "0:05:43", "remaining_time": "10:34:48"} +{"current_steps": 49, "total_steps": 5364, "loss": 0.3030151128768921, "lr": 1.787709497206704e-06, "epoch": 0.027412587412587414, "percentage": 0.91, "elapsed_time": "0:05:50", "remaining_time": "10:34:06"} +{"current_steps": 50, "total_steps": 5364, "loss": 0.21776533126831055, "lr": 1.8249534450651772e-06, "epoch": 0.027972027972027972, "percentage": 0.93, "elapsed_time": "0:05:57", "remaining_time": "10:33:28"} +{"current_steps": 51, "total_steps": 5364, "loss": 0.1647198349237442, "lr": 1.86219739292365e-06, "epoch": 0.028531468531468533, "percentage": 0.95, "elapsed_time": "0:06:05", "remaining_time": "10:34:06"} +{"current_steps": 52, "total_steps": 5364, "loss": 0.39184725284576416, "lr": 1.899441340782123e-06, "epoch": 0.02909090909090909, "percentage": 0.97, "elapsed_time": "0:06:12", "remaining_time": "10:33:30"} +{"current_steps": 53, "total_steps": 5364, "loss": 0.22568494081497192, "lr": 1.936685288640596e-06, "epoch": 0.029650349650349652, "percentage": 0.99, "elapsed_time": "0:06:18", "remaining_time": "10:32:52"} +{"current_steps": 54, "total_steps": 5364, "loss": 0.3494471311569214, "lr": 1.9739292364990693e-06, "epoch": 0.03020979020979021, "percentage": 1.01, "elapsed_time": "0:06:25", "remaining_time": "10:32:22"} +{"current_steps": 55, "total_steps": 5364, "loss": 0.15478362143039703, "lr": 2.011173184357542e-06, "epoch": 0.03076923076923077, "percentage": 1.03, "elapsed_time": "0:06:33", "remaining_time": "10:33:12"} +{"current_steps": 56, "total_steps": 5364, "loss": 0.31515175104141235, "lr": 2.048417132216015e-06, "epoch": 0.03132867132867133, "percentage": 1.04, "elapsed_time": "0:06:40", "remaining_time": "10:32:39"} +{"current_steps": 57, "total_steps": 5364, "loss": 0.2727990746498108, "lr": 2.085661080074488e-06, "epoch": 0.031888111888111886, "percentage": 1.06, "elapsed_time": "0:06:47", "remaining_time": "10:32:02"} +{"current_steps": 58, "total_steps": 5364, "loss": 0.08602505922317505, "lr": 2.1229050279329612e-06, "epoch": 0.03244755244755245, "percentage": 1.08, "elapsed_time": "0:06:54", "remaining_time": "10:32:04"} +{"current_steps": 59, "total_steps": 5364, "loss": 0.3949652910232544, "lr": 2.160148975791434e-06, "epoch": 0.03300699300699301, "percentage": 1.1, "elapsed_time": "0:07:01", "remaining_time": "10:31:46"} +{"current_steps": 60, "total_steps": 5364, "loss": 0.2999691367149353, "lr": 2.197392923649907e-06, "epoch": 0.033566433566433566, "percentage": 1.12, "elapsed_time": "0:07:08", "remaining_time": "10:31:14"} +{"current_steps": 61, "total_steps": 5364, "loss": 0.33670473098754883, "lr": 2.23463687150838e-06, "epoch": 0.034125874125874124, "percentage": 1.14, "elapsed_time": "0:07:15", "remaining_time": "10:30:50"} +{"current_steps": 62, "total_steps": 5364, "loss": 0.10987913608551025, "lr": 2.271880819366853e-06, "epoch": 0.03468531468531469, "percentage": 1.16, "elapsed_time": "0:07:22", "remaining_time": "10:30:22"} +{"current_steps": 63, "total_steps": 5364, "loss": 0.170758455991745, "lr": 2.3091247672253262e-06, "epoch": 0.035244755244755246, "percentage": 1.17, "elapsed_time": "0:07:29", "remaining_time": "10:30:54"} +{"current_steps": 64, "total_steps": 5364, "loss": 0.27042537927627563, "lr": 2.3463687150837993e-06, "epoch": 0.035804195804195804, "percentage": 1.19, "elapsed_time": "0:07:36", "remaining_time": "10:30:23"} +{"current_steps": 65, "total_steps": 5364, "loss": 0.18736714124679565, "lr": 2.383612662942272e-06, "epoch": 0.03636363636363636, "percentage": 1.21, "elapsed_time": "0:07:44", "remaining_time": "10:30:29"} +{"current_steps": 66, "total_steps": 5364, "loss": 0.24357211589813232, "lr": 2.420856610800745e-06, "epoch": 0.036923076923076927, "percentage": 1.23, "elapsed_time": "0:07:51", "remaining_time": "10:30:45"} +{"current_steps": 67, "total_steps": 5364, "loss": 0.5307877659797668, "lr": 2.458100558659218e-06, "epoch": 0.037482517482517484, "percentage": 1.25, "elapsed_time": "0:07:58", "remaining_time": "10:30:16"} +{"current_steps": 68, "total_steps": 5364, "loss": 0.33759641647338867, "lr": 2.495344506517691e-06, "epoch": 0.03804195804195804, "percentage": 1.27, "elapsed_time": "0:08:05", "remaining_time": "10:30:31"} +{"current_steps": 69, "total_steps": 5364, "loss": 0.06529378145933151, "lr": 2.5325884543761643e-06, "epoch": 0.0386013986013986, "percentage": 1.29, "elapsed_time": "0:08:13", "remaining_time": "10:31:04"} +{"current_steps": 70, "total_steps": 5364, "loss": 0.09254449605941772, "lr": 2.569832402234637e-06, "epoch": 0.039160839160839164, "percentage": 1.3, "elapsed_time": "0:08:20", "remaining_time": "10:30:35"} +{"current_steps": 71, "total_steps": 5364, "loss": 0.17956393957138062, "lr": 2.60707635009311e-06, "epoch": 0.03972027972027972, "percentage": 1.32, "elapsed_time": "0:08:27", "remaining_time": "10:30:07"} +{"current_steps": 72, "total_steps": 5364, "loss": 0.17691554129123688, "lr": 2.644320297951583e-06, "epoch": 0.04027972027972028, "percentage": 1.34, "elapsed_time": "0:08:34", "remaining_time": "10:29:47"} +{"current_steps": 73, "total_steps": 5364, "loss": 0.4376865327358246, "lr": 2.6815642458100562e-06, "epoch": 0.04083916083916084, "percentage": 1.36, "elapsed_time": "0:08:40", "remaining_time": "10:29:21"} +{"current_steps": 74, "total_steps": 5364, "loss": 0.09415476024150848, "lr": 2.718808193668529e-06, "epoch": 0.0413986013986014, "percentage": 1.38, "elapsed_time": "0:08:48", "remaining_time": "10:29:52"} +{"current_steps": 75, "total_steps": 5364, "loss": 0.16237586736679077, "lr": 2.756052141527002e-06, "epoch": 0.04195804195804196, "percentage": 1.4, "elapsed_time": "0:08:55", "remaining_time": "10:29:26"} +{"current_steps": 76, "total_steps": 5364, "loss": 0.0929182767868042, "lr": 2.793296089385475e-06, "epoch": 0.04251748251748252, "percentage": 1.42, "elapsed_time": "0:09:02", "remaining_time": "10:29:00"} +{"current_steps": 77, "total_steps": 5364, "loss": 0.2737734913825989, "lr": 2.8305400372439477e-06, "epoch": 0.043076923076923075, "percentage": 1.44, "elapsed_time": "0:09:09", "remaining_time": "10:28:31"} +{"current_steps": 78, "total_steps": 5364, "loss": 0.1599085032939911, "lr": 2.867783985102421e-06, "epoch": 0.04363636363636364, "percentage": 1.45, "elapsed_time": "0:09:16", "remaining_time": "10:28:08"} +{"current_steps": 79, "total_steps": 5364, "loss": 0.3695093095302582, "lr": 2.9050279329608943e-06, "epoch": 0.0441958041958042, "percentage": 1.47, "elapsed_time": "0:09:22", "remaining_time": "10:27:43"} +{"current_steps": 80, "total_steps": 5364, "loss": 0.34134459495544434, "lr": 2.9422718808193674e-06, "epoch": 0.044755244755244755, "percentage": 1.49, "elapsed_time": "0:09:30", "remaining_time": "10:27:54"} +{"current_steps": 81, "total_steps": 5364, "loss": 0.28330808877944946, "lr": 2.9795158286778405e-06, "epoch": 0.04531468531468531, "percentage": 1.51, "elapsed_time": "0:09:38", "remaining_time": "10:28:18"} +{"current_steps": 82, "total_steps": 5364, "loss": 0.15364162623882294, "lr": 3.016759776536313e-06, "epoch": 0.04587412587412588, "percentage": 1.53, "elapsed_time": "0:09:44", "remaining_time": "10:28:01"} +{"current_steps": 83, "total_steps": 5364, "loss": 0.07387331873178482, "lr": 3.0540037243947862e-06, "epoch": 0.046433566433566435, "percentage": 1.55, "elapsed_time": "0:09:52", "remaining_time": "10:28:33"} +{"current_steps": 84, "total_steps": 5364, "loss": 0.14365491271018982, "lr": 3.0912476722532593e-06, "epoch": 0.04699300699300699, "percentage": 1.57, "elapsed_time": "0:09:59", "remaining_time": "10:28:08"} +{"current_steps": 85, "total_steps": 5364, "loss": 0.11993853747844696, "lr": 3.128491620111732e-06, "epoch": 0.04755244755244755, "percentage": 1.58, "elapsed_time": "0:10:06", "remaining_time": "10:27:55"} +{"current_steps": 86, "total_steps": 5364, "loss": 0.3427037000656128, "lr": 3.165735567970205e-06, "epoch": 0.048111888111888115, "percentage": 1.6, "elapsed_time": "0:10:13", "remaining_time": "10:27:30"} +{"current_steps": 87, "total_steps": 5364, "loss": 0.17792752385139465, "lr": 3.202979515828678e-06, "epoch": 0.04867132867132867, "percentage": 1.62, "elapsed_time": "0:10:20", "remaining_time": "10:27:05"} +{"current_steps": 88, "total_steps": 5364, "loss": 0.17775408923625946, "lr": 3.240223463687151e-06, "epoch": 0.04923076923076923, "percentage": 1.64, "elapsed_time": "0:10:27", "remaining_time": "10:27:18"} +{"current_steps": 89, "total_steps": 5364, "loss": 0.3558940887451172, "lr": 3.277467411545624e-06, "epoch": 0.04979020979020979, "percentage": 1.66, "elapsed_time": "0:10:35", "remaining_time": "10:27:22"} +{"current_steps": 90, "total_steps": 5364, "loss": 0.1892475187778473, "lr": 3.314711359404097e-06, "epoch": 0.05034965034965035, "percentage": 1.68, "elapsed_time": "0:10:42", "remaining_time": "10:27:18"} +{"current_steps": 91, "total_steps": 5364, "loss": 0.2643950581550598, "lr": 3.35195530726257e-06, "epoch": 0.05090909090909091, "percentage": 1.7, "elapsed_time": "0:10:49", "remaining_time": "10:27:23"} +{"current_steps": 92, "total_steps": 5364, "loss": 0.34897392988204956, "lr": 3.3891992551210427e-06, "epoch": 0.05146853146853147, "percentage": 1.72, "elapsed_time": "0:10:57", "remaining_time": "10:27:31"} +{"current_steps": 93, "total_steps": 5364, "loss": 0.3010078966617584, "lr": 3.4264432029795162e-06, "epoch": 0.052027972027972026, "percentage": 1.73, "elapsed_time": "0:11:04", "remaining_time": "10:27:16"} +{"current_steps": 94, "total_steps": 5364, "loss": 0.47209683060646057, "lr": 3.4636871508379893e-06, "epoch": 0.05258741258741259, "percentage": 1.75, "elapsed_time": "0:11:11", "remaining_time": "10:27:11"} +{"current_steps": 95, "total_steps": 5364, "loss": 0.12591159343719482, "lr": 3.5009310986964624e-06, "epoch": 0.05314685314685315, "percentage": 1.77, "elapsed_time": "0:11:18", "remaining_time": "10:27:01"} +{"current_steps": 96, "total_steps": 5364, "loss": 0.2532845139503479, "lr": 3.538175046554935e-06, "epoch": 0.053706293706293706, "percentage": 1.79, "elapsed_time": "0:11:25", "remaining_time": "10:26:49"} +{"current_steps": 97, "total_steps": 5364, "loss": 0.10538607090711594, "lr": 3.575418994413408e-06, "epoch": 0.054265734265734264, "percentage": 1.81, "elapsed_time": "0:11:32", "remaining_time": "10:26:57"} +{"current_steps": 98, "total_steps": 5364, "loss": 0.07029303908348083, "lr": 3.6126629422718813e-06, "epoch": 0.05482517482517483, "percentage": 1.83, "elapsed_time": "0:11:39", "remaining_time": "10:26:35"} +{"current_steps": 99, "total_steps": 5364, "loss": 0.2196342945098877, "lr": 3.6499068901303543e-06, "epoch": 0.055384615384615386, "percentage": 1.85, "elapsed_time": "0:11:46", "remaining_time": "10:26:15"} +{"current_steps": 100, "total_steps": 5364, "loss": 0.3075711131095886, "lr": 3.687150837988827e-06, "epoch": 0.055944055944055944, "percentage": 1.86, "elapsed_time": "0:11:53", "remaining_time": "10:26:21"} +{"current_steps": 101, "total_steps": 5364, "loss": 0.16532987356185913, "lr": 3.7243947858473e-06, "epoch": 0.0565034965034965, "percentage": 1.88, "elapsed_time": "0:12:01", "remaining_time": "10:26:55"} +{"current_steps": 102, "total_steps": 5364, "loss": 0.10717526078224182, "lr": 3.761638733705773e-06, "epoch": 0.057062937062937066, "percentage": 1.9, "elapsed_time": "0:12:08", "remaining_time": "10:26:47"} +{"current_steps": 103, "total_steps": 5364, "loss": 0.2941814661026001, "lr": 3.798882681564246e-06, "epoch": 0.057622377622377624, "percentage": 1.92, "elapsed_time": "0:12:16", "remaining_time": "10:26:35"} +{"current_steps": 104, "total_steps": 5364, "loss": 0.31177815794944763, "lr": 3.836126629422719e-06, "epoch": 0.05818181818181818, "percentage": 1.94, "elapsed_time": "0:12:22", "remaining_time": "10:26:11"} +{"current_steps": 105, "total_steps": 5364, "loss": 0.05996173247694969, "lr": 3.873370577281192e-06, "epoch": 0.05874125874125874, "percentage": 1.96, "elapsed_time": "0:12:29", "remaining_time": "10:25:59"} +{"current_steps": 106, "total_steps": 5364, "loss": 0.3034001290798187, "lr": 3.910614525139665e-06, "epoch": 0.059300699300699304, "percentage": 1.98, "elapsed_time": "0:12:37", "remaining_time": "10:25:59"} +{"current_steps": 107, "total_steps": 5364, "loss": 0.18240708112716675, "lr": 3.947858472998139e-06, "epoch": 0.05986013986013986, "percentage": 1.99, "elapsed_time": "0:12:44", "remaining_time": "10:25:38"} +{"current_steps": 108, "total_steps": 5364, "loss": 0.17454060912132263, "lr": 3.985102420856611e-06, "epoch": 0.06041958041958042, "percentage": 2.01, "elapsed_time": "0:12:51", "remaining_time": "10:25:30"} +{"current_steps": 109, "total_steps": 5364, "loss": 0.4517461955547333, "lr": 4.022346368715084e-06, "epoch": 0.06097902097902098, "percentage": 2.03, "elapsed_time": "0:12:58", "remaining_time": "10:25:35"} +{"current_steps": 110, "total_steps": 5364, "loss": 0.36985060572624207, "lr": 4.0595903165735574e-06, "epoch": 0.06153846153846154, "percentage": 2.05, "elapsed_time": "0:13:05", "remaining_time": "10:25:15"} +{"current_steps": 111, "total_steps": 5364, "loss": 0.13078774511814117, "lr": 4.09683426443203e-06, "epoch": 0.0620979020979021, "percentage": 2.07, "elapsed_time": "0:13:12", "remaining_time": "10:24:56"} +{"current_steps": 112, "total_steps": 5364, "loss": 0.23888853192329407, "lr": 4.134078212290504e-06, "epoch": 0.06265734265734266, "percentage": 2.09, "elapsed_time": "0:13:19", "remaining_time": "10:24:37"} +{"current_steps": 113, "total_steps": 5364, "loss": 0.2135712057352066, "lr": 4.171322160148976e-06, "epoch": 0.06321678321678321, "percentage": 2.11, "elapsed_time": "0:13:26", "remaining_time": "10:24:20"} +{"current_steps": 114, "total_steps": 5364, "loss": 0.223187655210495, "lr": 4.208566108007449e-06, "epoch": 0.06377622377622377, "percentage": 2.13, "elapsed_time": "0:13:33", "remaining_time": "10:24:01"} +{"current_steps": 115, "total_steps": 5364, "loss": 0.11557766050100327, "lr": 4.2458100558659224e-06, "epoch": 0.06433566433566433, "percentage": 2.14, "elapsed_time": "0:13:40", "remaining_time": "10:23:56"} +{"current_steps": 116, "total_steps": 5364, "loss": 0.37016069889068604, "lr": 4.283054003724395e-06, "epoch": 0.0648951048951049, "percentage": 2.16, "elapsed_time": "0:13:47", "remaining_time": "10:24:03"} +{"current_steps": 117, "total_steps": 5364, "loss": 0.21168112754821777, "lr": 4.320297951582868e-06, "epoch": 0.06545454545454546, "percentage": 2.18, "elapsed_time": "0:13:54", "remaining_time": "10:23:44"} +{"current_steps": 118, "total_steps": 5364, "loss": 0.2518339455127716, "lr": 4.357541899441341e-06, "epoch": 0.06601398601398602, "percentage": 2.2, "elapsed_time": "0:14:01", "remaining_time": "10:23:31"} +{"current_steps": 119, "total_steps": 5364, "loss": 0.3767576217651367, "lr": 4.394785847299814e-06, "epoch": 0.06657342657342658, "percentage": 2.22, "elapsed_time": "0:14:08", "remaining_time": "10:23:18"} +{"current_steps": 120, "total_steps": 5364, "loss": 0.21407866477966309, "lr": 4.432029795158287e-06, "epoch": 0.06713286713286713, "percentage": 2.24, "elapsed_time": "0:14:15", "remaining_time": "10:23:00"} +{"current_steps": 121, "total_steps": 5364, "loss": 0.1839624047279358, "lr": 4.46927374301676e-06, "epoch": 0.06769230769230769, "percentage": 2.26, "elapsed_time": "0:14:22", "remaining_time": "10:22:43"} +{"current_steps": 122, "total_steps": 5364, "loss": 0.4413369297981262, "lr": 4.506517690875234e-06, "epoch": 0.06825174825174825, "percentage": 2.27, "elapsed_time": "0:14:29", "remaining_time": "10:22:28"} +{"current_steps": 123, "total_steps": 5364, "loss": 0.21440665423870087, "lr": 4.543761638733706e-06, "epoch": 0.0688111888111888, "percentage": 2.29, "elapsed_time": "0:14:36", "remaining_time": "10:22:45"} +{"current_steps": 124, "total_steps": 5364, "loss": 0.259691059589386, "lr": 4.581005586592179e-06, "epoch": 0.06937062937062938, "percentage": 2.31, "elapsed_time": "0:14:43", "remaining_time": "10:22:31"} +{"current_steps": 125, "total_steps": 5364, "loss": 0.4421902298927307, "lr": 4.6182495344506524e-06, "epoch": 0.06993006993006994, "percentage": 2.33, "elapsed_time": "0:14:51", "remaining_time": "10:22:28"} +{"current_steps": 126, "total_steps": 5364, "loss": 0.30450770258903503, "lr": 4.655493482309125e-06, "epoch": 0.07048951048951049, "percentage": 2.35, "elapsed_time": "0:14:58", "remaining_time": "10:22:18"} +{"current_steps": 127, "total_steps": 5364, "loss": 0.14847812056541443, "lr": 4.692737430167599e-06, "epoch": 0.07104895104895105, "percentage": 2.37, "elapsed_time": "0:15:05", "remaining_time": "10:22:20"} +{"current_steps": 128, "total_steps": 5364, "loss": 0.3133907914161682, "lr": 4.729981378026071e-06, "epoch": 0.07160839160839161, "percentage": 2.39, "elapsed_time": "0:15:12", "remaining_time": "10:22:02"} +{"current_steps": 129, "total_steps": 5364, "loss": 0.24020129442214966, "lr": 4.767225325884544e-06, "epoch": 0.07216783216783217, "percentage": 2.4, "elapsed_time": "0:15:19", "remaining_time": "10:22:11"} +{"current_steps": 130, "total_steps": 5364, "loss": 0.4030919075012207, "lr": 4.8044692737430175e-06, "epoch": 0.07272727272727272, "percentage": 2.42, "elapsed_time": "0:15:27", "remaining_time": "10:22:12"} +{"current_steps": 131, "total_steps": 5364, "loss": 0.44573646783828735, "lr": 4.84171322160149e-06, "epoch": 0.07328671328671328, "percentage": 2.44, "elapsed_time": "0:15:34", "remaining_time": "10:21:54"} +{"current_steps": 132, "total_steps": 5364, "loss": 0.09619377553462982, "lr": 4.878957169459963e-06, "epoch": 0.07384615384615385, "percentage": 2.46, "elapsed_time": "0:15:40", "remaining_time": "10:21:36"} +{"current_steps": 133, "total_steps": 5364, "loss": 0.23731550574302673, "lr": 4.916201117318436e-06, "epoch": 0.07440559440559441, "percentage": 2.48, "elapsed_time": "0:15:47", "remaining_time": "10:21:19"} +{"current_steps": 134, "total_steps": 5364, "loss": 0.24186912178993225, "lr": 4.953445065176909e-06, "epoch": 0.07496503496503497, "percentage": 2.5, "elapsed_time": "0:15:55", "remaining_time": "10:21:24"} +{"current_steps": 135, "total_steps": 5364, "loss": 0.2194354236125946, "lr": 4.990689013035382e-06, "epoch": 0.07552447552447553, "percentage": 2.52, "elapsed_time": "0:16:02", "remaining_time": "10:21:07"} +{"current_steps": 136, "total_steps": 5364, "loss": 0.2353523075580597, "lr": 5.027932960893855e-06, "epoch": 0.07608391608391608, "percentage": 2.54, "elapsed_time": "0:16:09", "remaining_time": "10:20:50"} +{"current_steps": 137, "total_steps": 5364, "loss": 0.19667869806289673, "lr": 5.065176908752329e-06, "epoch": 0.07664335664335664, "percentage": 2.55, "elapsed_time": "0:16:15", "remaining_time": "10:20:33"} +{"current_steps": 138, "total_steps": 5364, "loss": 0.19589757919311523, "lr": 5.102420856610801e-06, "epoch": 0.0772027972027972, "percentage": 2.57, "elapsed_time": "0:16:22", "remaining_time": "10:20:23"} +{"current_steps": 139, "total_steps": 5364, "loss": 0.11321631073951721, "lr": 5.139664804469274e-06, "epoch": 0.07776223776223776, "percentage": 2.59, "elapsed_time": "0:16:30", "remaining_time": "10:20:28"} +{"current_steps": 140, "total_steps": 5364, "loss": 0.1956067979335785, "lr": 5.1769087523277475e-06, "epoch": 0.07832167832167833, "percentage": 2.61, "elapsed_time": "0:16:37", "remaining_time": "10:20:38"} +{"current_steps": 141, "total_steps": 5364, "loss": 0.1621169149875641, "lr": 5.21415270018622e-06, "epoch": 0.07888111888111889, "percentage": 2.63, "elapsed_time": "0:16:44", "remaining_time": "10:20:25"} +{"current_steps": 142, "total_steps": 5364, "loss": 0.07715841382741928, "lr": 5.251396648044693e-06, "epoch": 0.07944055944055944, "percentage": 2.65, "elapsed_time": "0:16:52", "remaining_time": "10:20:30"} +{"current_steps": 143, "total_steps": 5364, "loss": 0.24442407488822937, "lr": 5.288640595903166e-06, "epoch": 0.08, "percentage": 2.67, "elapsed_time": "0:16:59", "remaining_time": "10:20:18"} +{"current_steps": 144, "total_steps": 5364, "loss": 0.16002245247364044, "lr": 5.325884543761639e-06, "epoch": 0.08055944055944056, "percentage": 2.68, "elapsed_time": "0:17:06", "remaining_time": "10:20:21"} +{"current_steps": 145, "total_steps": 5364, "loss": 0.24429768323898315, "lr": 5.3631284916201125e-06, "epoch": 0.08111888111888112, "percentage": 2.7, "elapsed_time": "0:17:13", "remaining_time": "10:20:07"} +{"current_steps": 146, "total_steps": 5364, "loss": 0.35507357120513916, "lr": 5.400372439478585e-06, "epoch": 0.08167832167832167, "percentage": 2.72, "elapsed_time": "0:17:20", "remaining_time": "10:20:04"} +{"current_steps": 147, "total_steps": 5364, "loss": 0.3768399953842163, "lr": 5.437616387337058e-06, "epoch": 0.08223776223776223, "percentage": 2.74, "elapsed_time": "0:17:28", "remaining_time": "10:19:59"} +{"current_steps": 148, "total_steps": 5364, "loss": 0.04639776796102524, "lr": 5.474860335195531e-06, "epoch": 0.0827972027972028, "percentage": 2.76, "elapsed_time": "0:17:35", "remaining_time": "10:20:06"} +{"current_steps": 149, "total_steps": 5364, "loss": 0.22011643648147583, "lr": 5.512104283054004e-06, "epoch": 0.08335664335664336, "percentage": 2.78, "elapsed_time": "0:17:42", "remaining_time": "10:19:50"} +{"current_steps": 150, "total_steps": 5364, "loss": 0.153517946600914, "lr": 5.549348230912477e-06, "epoch": 0.08391608391608392, "percentage": 2.8, "elapsed_time": "0:17:49", "remaining_time": "10:19:49"} +{"current_steps": 151, "total_steps": 5364, "loss": 0.19159212708473206, "lr": 5.58659217877095e-06, "epoch": 0.08447552447552448, "percentage": 2.82, "elapsed_time": "0:17:56", "remaining_time": "10:19:34"} +{"current_steps": 152, "total_steps": 5364, "loss": 0.12672311067581177, "lr": 5.623836126629423e-06, "epoch": 0.08503496503496503, "percentage": 2.83, "elapsed_time": "0:18:03", "remaining_time": "10:19:18"} +{"current_steps": 153, "total_steps": 5364, "loss": 0.09087824821472168, "lr": 5.6610800744878955e-06, "epoch": 0.08559440559440559, "percentage": 2.85, "elapsed_time": "0:18:10", "remaining_time": "10:19:14"} +{"current_steps": 154, "total_steps": 5364, "loss": 0.06624268740415573, "lr": 5.698324022346369e-06, "epoch": 0.08615384615384615, "percentage": 2.87, "elapsed_time": "0:18:18", "remaining_time": "10:19:17"} +{"current_steps": 155, "total_steps": 5364, "loss": 0.2487543672323227, "lr": 5.735567970204842e-06, "epoch": 0.08671328671328671, "percentage": 2.89, "elapsed_time": "0:18:25", "remaining_time": "10:19:01"} +{"current_steps": 156, "total_steps": 5364, "loss": 0.3196878731250763, "lr": 5.772811918063314e-06, "epoch": 0.08727272727272728, "percentage": 2.91, "elapsed_time": "0:18:31", "remaining_time": "10:18:36"} +{"current_steps": 157, "total_steps": 5364, "loss": 0.19816963374614716, "lr": 5.810055865921789e-06, "epoch": 0.08783216783216784, "percentage": 2.93, "elapsed_time": "0:18:38", "remaining_time": "10:18:30"} +{"current_steps": 158, "total_steps": 5364, "loss": 0.20571818947792053, "lr": 5.847299813780261e-06, "epoch": 0.0883916083916084, "percentage": 2.95, "elapsed_time": "0:18:45", "remaining_time": "10:18:20"} +{"current_steps": 159, "total_steps": 5364, "loss": 0.24458524584770203, "lr": 5.884543761638735e-06, "epoch": 0.08895104895104895, "percentage": 2.96, "elapsed_time": "0:18:53", "remaining_time": "10:18:37"} +{"current_steps": 160, "total_steps": 5364, "loss": 0.11320653557777405, "lr": 5.9217877094972075e-06, "epoch": 0.08951048951048951, "percentage": 2.98, "elapsed_time": "0:19:01", "remaining_time": "10:18:31"} +{"current_steps": 161, "total_steps": 5364, "loss": 0.22428768873214722, "lr": 5.959031657355681e-06, "epoch": 0.09006993006993007, "percentage": 3.0, "elapsed_time": "0:19:08", "remaining_time": "10:18:35"} +{"current_steps": 162, "total_steps": 5364, "loss": 0.669741690158844, "lr": 5.996275605214154e-06, "epoch": 0.09062937062937063, "percentage": 3.02, "elapsed_time": "0:19:15", "remaining_time": "10:18:34"} +{"current_steps": 163, "total_steps": 5364, "loss": 0.27598151564598083, "lr": 6.033519553072626e-06, "epoch": 0.09118881118881118, "percentage": 3.04, "elapsed_time": "0:19:23", "remaining_time": "10:18:30"} +{"current_steps": 164, "total_steps": 5364, "loss": 0.22851765155792236, "lr": 6.0707635009311e-06, "epoch": 0.09174825174825176, "percentage": 3.06, "elapsed_time": "0:19:29", "remaining_time": "10:18:15"} +{"current_steps": 165, "total_steps": 5364, "loss": 0.20998801290988922, "lr": 6.1080074487895725e-06, "epoch": 0.09230769230769231, "percentage": 3.08, "elapsed_time": "0:19:36", "remaining_time": "10:18:00"} +{"current_steps": 166, "total_steps": 5364, "loss": 0.1636764258146286, "lr": 6.145251396648045e-06, "epoch": 0.09286713286713287, "percentage": 3.09, "elapsed_time": "0:19:43", "remaining_time": "10:17:50"} +{"current_steps": 167, "total_steps": 5364, "loss": 0.1978091597557068, "lr": 6.182495344506519e-06, "epoch": 0.09342657342657343, "percentage": 3.11, "elapsed_time": "0:19:50", "remaining_time": "10:17:35"} +{"current_steps": 168, "total_steps": 5364, "loss": 0.381095826625824, "lr": 6.219739292364991e-06, "epoch": 0.09398601398601399, "percentage": 3.13, "elapsed_time": "0:19:57", "remaining_time": "10:17:30"} +{"current_steps": 169, "total_steps": 5364, "loss": 0.19355298578739166, "lr": 6.256983240223464e-06, "epoch": 0.09454545454545454, "percentage": 3.15, "elapsed_time": "0:20:05", "remaining_time": "10:17:26"} +{"current_steps": 170, "total_steps": 5364, "loss": 0.46214985847473145, "lr": 6.2942271880819375e-06, "epoch": 0.0951048951048951, "percentage": 3.17, "elapsed_time": "0:20:12", "remaining_time": "10:17:12"} +{"current_steps": 171, "total_steps": 5364, "loss": 0.3472660183906555, "lr": 6.33147113594041e-06, "epoch": 0.09566433566433566, "percentage": 3.19, "elapsed_time": "0:20:18", "remaining_time": "10:16:56"} +{"current_steps": 172, "total_steps": 5364, "loss": 0.26900821924209595, "lr": 6.368715083798883e-06, "epoch": 0.09622377622377623, "percentage": 3.21, "elapsed_time": "0:20:25", "remaining_time": "10:16:45"} +{"current_steps": 173, "total_steps": 5364, "loss": 0.17796164751052856, "lr": 6.405959031657356e-06, "epoch": 0.09678321678321679, "percentage": 3.23, "elapsed_time": "0:20:33", "remaining_time": "10:16:49"} +{"current_steps": 174, "total_steps": 5364, "loss": 0.47886112332344055, "lr": 6.443202979515829e-06, "epoch": 0.09734265734265735, "percentage": 3.24, "elapsed_time": "0:20:40", "remaining_time": "10:16:39"} +{"current_steps": 175, "total_steps": 5364, "loss": 0.1489739865064621, "lr": 6.480446927374302e-06, "epoch": 0.0979020979020979, "percentage": 3.26, "elapsed_time": "0:20:48", "remaining_time": "10:16:55"} +{"current_steps": 176, "total_steps": 5364, "loss": 0.30581769347190857, "lr": 6.517690875232775e-06, "epoch": 0.09846153846153846, "percentage": 3.28, "elapsed_time": "0:20:55", "remaining_time": "10:16:41"} +{"current_steps": 177, "total_steps": 5364, "loss": 0.2701963484287262, "lr": 6.554934823091248e-06, "epoch": 0.09902097902097902, "percentage": 3.3, "elapsed_time": "0:21:02", "remaining_time": "10:16:28"} +{"current_steps": 178, "total_steps": 5364, "loss": 0.0773768424987793, "lr": 6.592178770949721e-06, "epoch": 0.09958041958041958, "percentage": 3.32, "elapsed_time": "0:21:09", "remaining_time": "10:16:15"} +{"current_steps": 179, "total_steps": 5364, "loss": 0.24532026052474976, "lr": 6.629422718808194e-06, "epoch": 0.10013986013986013, "percentage": 3.34, "elapsed_time": "0:21:16", "remaining_time": "10:16:03"} +{"current_steps": 180, "total_steps": 5364, "loss": 0.2108195722103119, "lr": 6.666666666666667e-06, "epoch": 0.1006993006993007, "percentage": 3.36, "elapsed_time": "0:21:23", "remaining_time": "10:16:10"} +{"current_steps": 181, "total_steps": 5364, "loss": 0.20850294828414917, "lr": 6.70391061452514e-06, "epoch": 0.10125874125874126, "percentage": 3.37, "elapsed_time": "0:21:30", "remaining_time": "10:15:57"} +{"current_steps": 182, "total_steps": 5364, "loss": 0.18542012572288513, "lr": 6.741154562383613e-06, "epoch": 0.10181818181818182, "percentage": 3.39, "elapsed_time": "0:21:37", "remaining_time": "10:15:35"} +{"current_steps": 183, "total_steps": 5364, "loss": 0.1984855681657791, "lr": 6.7783985102420855e-06, "epoch": 0.10237762237762238, "percentage": 3.41, "elapsed_time": "0:21:44", "remaining_time": "10:15:21"} +{"current_steps": 184, "total_steps": 5364, "loss": 0.142118439078331, "lr": 6.815642458100559e-06, "epoch": 0.10293706293706294, "percentage": 3.43, "elapsed_time": "0:21:51", "remaining_time": "10:15:27"} +{"current_steps": 185, "total_steps": 5364, "loss": 0.1488180309534073, "lr": 6.8528864059590325e-06, "epoch": 0.1034965034965035, "percentage": 3.45, "elapsed_time": "0:21:58", "remaining_time": "10:15:14"} +{"current_steps": 186, "total_steps": 5364, "loss": 0.10070636868476868, "lr": 6.890130353817506e-06, "epoch": 0.10405594405594405, "percentage": 3.47, "elapsed_time": "0:22:05", "remaining_time": "10:15:02"} +{"current_steps": 187, "total_steps": 5364, "loss": 0.21599172055721283, "lr": 6.927374301675979e-06, "epoch": 0.10461538461538461, "percentage": 3.49, "elapsed_time": "0:22:13", "remaining_time": "10:15:04"} +{"current_steps": 188, "total_steps": 5364, "loss": 0.2776916027069092, "lr": 6.964618249534451e-06, "epoch": 0.10517482517482518, "percentage": 3.5, "elapsed_time": "0:22:20", "remaining_time": "10:14:53"} +{"current_steps": 189, "total_steps": 5364, "loss": 0.26190677285194397, "lr": 7.001862197392925e-06, "epoch": 0.10573426573426574, "percentage": 3.52, "elapsed_time": "0:22:26", "remaining_time": "10:14:40"} +{"current_steps": 190, "total_steps": 5364, "loss": 0.45778489112854004, "lr": 7.0391061452513975e-06, "epoch": 0.1062937062937063, "percentage": 3.54, "elapsed_time": "0:22:33", "remaining_time": "10:14:27"} +{"current_steps": 191, "total_steps": 5364, "loss": 0.23035813868045807, "lr": 7.07635009310987e-06, "epoch": 0.10685314685314685, "percentage": 3.56, "elapsed_time": "0:22:40", "remaining_time": "10:14:12"} +{"current_steps": 192, "total_steps": 5364, "loss": 0.13400602340698242, "lr": 7.113594040968344e-06, "epoch": 0.10741258741258741, "percentage": 3.58, "elapsed_time": "0:22:47", "remaining_time": "10:14:03"} +{"current_steps": 193, "total_steps": 5364, "loss": 0.107833132147789, "lr": 7.150837988826816e-06, "epoch": 0.10797202797202797, "percentage": 3.6, "elapsed_time": "0:22:55", "remaining_time": "10:14:06"} +{"current_steps": 194, "total_steps": 5364, "loss": 0.10887670516967773, "lr": 7.188081936685289e-06, "epoch": 0.10853146853146853, "percentage": 3.62, "elapsed_time": "0:23:02", "remaining_time": "10:13:56"} +{"current_steps": 195, "total_steps": 5364, "loss": 0.45634034276008606, "lr": 7.2253258845437625e-06, "epoch": 0.10909090909090909, "percentage": 3.64, "elapsed_time": "0:23:09", "remaining_time": "10:13:45"} +{"current_steps": 196, "total_steps": 5364, "loss": 0.13844041526317596, "lr": 7.262569832402235e-06, "epoch": 0.10965034965034966, "percentage": 3.65, "elapsed_time": "0:23:17", "remaining_time": "10:13:55"} +{"current_steps": 197, "total_steps": 5364, "loss": 0.24760495126247406, "lr": 7.299813780260709e-06, "epoch": 0.11020979020979021, "percentage": 3.67, "elapsed_time": "0:23:24", "remaining_time": "10:13:50"} +{"current_steps": 198, "total_steps": 5364, "loss": 0.2501208186149597, "lr": 7.337057728119181e-06, "epoch": 0.11076923076923077, "percentage": 3.69, "elapsed_time": "0:23:31", "remaining_time": "10:13:45"} +{"current_steps": 199, "total_steps": 5364, "loss": 0.3180907368659973, "lr": 7.374301675977654e-06, "epoch": 0.11132867132867133, "percentage": 3.71, "elapsed_time": "0:23:38", "remaining_time": "10:13:32"} +{"current_steps": 200, "total_steps": 5364, "loss": 0.4095803499221802, "lr": 7.4115456238361275e-06, "epoch": 0.11188811188811189, "percentage": 3.73, "elapsed_time": "0:23:45", "remaining_time": "10:13:32"} +{"current_steps": 201, "total_steps": 5364, "loss": 0.06699497997760773, "lr": 7.4487895716946e-06, "epoch": 0.11244755244755245, "percentage": 3.75, "elapsed_time": "0:23:52", "remaining_time": "10:13:19"} +{"current_steps": 202, "total_steps": 5364, "loss": 0.14195895195007324, "lr": 7.486033519553073e-06, "epoch": 0.113006993006993, "percentage": 3.77, "elapsed_time": "0:23:59", "remaining_time": "10:13:13"} +{"current_steps": 203, "total_steps": 5364, "loss": 0.3649834990501404, "lr": 7.523277467411546e-06, "epoch": 0.11356643356643356, "percentage": 3.78, "elapsed_time": "0:24:07", "remaining_time": "10:13:23"} +{"current_steps": 204, "total_steps": 5364, "loss": 0.15401607751846313, "lr": 7.560521415270019e-06, "epoch": 0.11412587412587413, "percentage": 3.8, "elapsed_time": "0:24:14", "remaining_time": "10:13:14"} +{"current_steps": 205, "total_steps": 5364, "loss": 0.4673020839691162, "lr": 7.597765363128492e-06, "epoch": 0.11468531468531469, "percentage": 3.82, "elapsed_time": "0:24:21", "remaining_time": "10:13:05"} +{"current_steps": 206, "total_steps": 5364, "loss": 0.19694717228412628, "lr": 7.635009310986964e-06, "epoch": 0.11524475524475525, "percentage": 3.84, "elapsed_time": "0:24:28", "remaining_time": "10:12:52"} +{"current_steps": 207, "total_steps": 5364, "loss": 0.38036859035491943, "lr": 7.672253258845438e-06, "epoch": 0.1158041958041958, "percentage": 3.86, "elapsed_time": "0:24:35", "remaining_time": "10:12:39"} +{"current_steps": 208, "total_steps": 5364, "loss": 0.20669475197792053, "lr": 7.709497206703911e-06, "epoch": 0.11636363636363636, "percentage": 3.88, "elapsed_time": "0:24:42", "remaining_time": "10:12:35"} +{"current_steps": 209, "total_steps": 5364, "loss": 0.30299532413482666, "lr": 7.746741154562383e-06, "epoch": 0.11692307692307692, "percentage": 3.9, "elapsed_time": "0:24:49", "remaining_time": "10:12:28"} +{"current_steps": 210, "total_steps": 5364, "loss": 0.13257357478141785, "lr": 7.783985102420857e-06, "epoch": 0.11748251748251748, "percentage": 3.91, "elapsed_time": "0:24:56", "remaining_time": "10:12:06"} +{"current_steps": 211, "total_steps": 5364, "loss": 0.17287683486938477, "lr": 7.82122905027933e-06, "epoch": 0.11804195804195804, "percentage": 3.93, "elapsed_time": "0:25:03", "remaining_time": "10:11:56"} +{"current_steps": 212, "total_steps": 5364, "loss": 0.4957885444164276, "lr": 7.858472998137802e-06, "epoch": 0.11860139860139861, "percentage": 3.95, "elapsed_time": "0:25:10", "remaining_time": "10:11:52"} +{"current_steps": 213, "total_steps": 5364, "loss": 0.14581745862960815, "lr": 7.895716945996277e-06, "epoch": 0.11916083916083917, "percentage": 3.97, "elapsed_time": "0:25:17", "remaining_time": "10:11:40"} +{"current_steps": 214, "total_steps": 5364, "loss": 0.21463730931282043, "lr": 7.932960893854749e-06, "epoch": 0.11972027972027972, "percentage": 3.99, "elapsed_time": "0:25:24", "remaining_time": "10:11:28"} +{"current_steps": 215, "total_steps": 5364, "loss": 0.20221847295761108, "lr": 7.970204841713223e-06, "epoch": 0.12027972027972028, "percentage": 4.01, "elapsed_time": "0:25:31", "remaining_time": "10:11:14"} +{"current_steps": 216, "total_steps": 5364, "loss": 0.258322149515152, "lr": 8.007448789571696e-06, "epoch": 0.12083916083916084, "percentage": 4.03, "elapsed_time": "0:25:38", "remaining_time": "10:11:00"} +{"current_steps": 217, "total_steps": 5364, "loss": 0.22753967344760895, "lr": 8.044692737430168e-06, "epoch": 0.1213986013986014, "percentage": 4.05, "elapsed_time": "0:25:45", "remaining_time": "10:10:48"} +{"current_steps": 218, "total_steps": 5364, "loss": 0.2552938759326935, "lr": 8.081936685288641e-06, "epoch": 0.12195804195804195, "percentage": 4.06, "elapsed_time": "0:25:52", "remaining_time": "10:10:51"} +{"current_steps": 219, "total_steps": 5364, "loss": 0.19879496097564697, "lr": 8.119180633147115e-06, "epoch": 0.12251748251748251, "percentage": 4.08, "elapsed_time": "0:26:00", "remaining_time": "10:10:51"} +{"current_steps": 220, "total_steps": 5364, "loss": 0.167031928896904, "lr": 8.156424581005588e-06, "epoch": 0.12307692307692308, "percentage": 4.1, "elapsed_time": "0:26:06", "remaining_time": "10:10:38"} +{"current_steps": 221, "total_steps": 5364, "loss": 0.09976990520954132, "lr": 8.19366852886406e-06, "epoch": 0.12363636363636364, "percentage": 4.12, "elapsed_time": "0:26:14", "remaining_time": "10:10:30"} +{"current_steps": 222, "total_steps": 5364, "loss": 0.25700443983078003, "lr": 8.230912476722534e-06, "epoch": 0.1241958041958042, "percentage": 4.14, "elapsed_time": "0:26:21", "remaining_time": "10:10:21"} +{"current_steps": 223, "total_steps": 5364, "loss": 0.06720789521932602, "lr": 8.268156424581007e-06, "epoch": 0.12475524475524476, "percentage": 4.16, "elapsed_time": "0:26:27", "remaining_time": "10:10:08"} +{"current_steps": 224, "total_steps": 5364, "loss": 0.1633196324110031, "lr": 8.305400372439479e-06, "epoch": 0.12531468531468531, "percentage": 4.18, "elapsed_time": "0:26:35", "remaining_time": "10:10:17"} +{"current_steps": 225, "total_steps": 5364, "loss": 0.07755030691623688, "lr": 8.342644320297953e-06, "epoch": 0.1258741258741259, "percentage": 4.19, "elapsed_time": "0:26:42", "remaining_time": "10:10:04"} +{"current_steps": 226, "total_steps": 5364, "loss": 0.22566697001457214, "lr": 8.379888268156426e-06, "epoch": 0.12643356643356643, "percentage": 4.21, "elapsed_time": "0:26:49", "remaining_time": "10:09:51"} +{"current_steps": 227, "total_steps": 5364, "loss": 0.14982138574123383, "lr": 8.417132216014898e-06, "epoch": 0.126993006993007, "percentage": 4.23, "elapsed_time": "0:26:56", "remaining_time": "10:09:42"} +{"current_steps": 228, "total_steps": 5364, "loss": 0.2582307457923889, "lr": 8.454376163873371e-06, "epoch": 0.12755244755244755, "percentage": 4.25, "elapsed_time": "0:27:04", "remaining_time": "10:09:44"} +{"current_steps": 229, "total_steps": 5364, "loss": 0.32020890712738037, "lr": 8.491620111731845e-06, "epoch": 0.12811188811188812, "percentage": 4.27, "elapsed_time": "0:27:11", "remaining_time": "10:09:35"} +{"current_steps": 230, "total_steps": 5364, "loss": 0.20898734033107758, "lr": 8.528864059590317e-06, "epoch": 0.12867132867132866, "percentage": 4.29, "elapsed_time": "0:27:18", "remaining_time": "10:09:33"} +{"current_steps": 231, "total_steps": 5364, "loss": 0.07202384620904922, "lr": 8.56610800744879e-06, "epoch": 0.12923076923076923, "percentage": 4.31, "elapsed_time": "0:27:25", "remaining_time": "10:09:27"} +{"current_steps": 232, "total_steps": 5364, "loss": 0.08093154430389404, "lr": 8.603351955307264e-06, "epoch": 0.1297902097902098, "percentage": 4.33, "elapsed_time": "0:27:32", "remaining_time": "10:09:15"} +{"current_steps": 233, "total_steps": 5364, "loss": 0.13062073290348053, "lr": 8.640595903165736e-06, "epoch": 0.13034965034965035, "percentage": 4.34, "elapsed_time": "0:27:39", "remaining_time": "10:09:08"} +{"current_steps": 234, "total_steps": 5364, "loss": 0.09924593567848206, "lr": 8.677839851024209e-06, "epoch": 0.13090909090909092, "percentage": 4.36, "elapsed_time": "0:27:46", "remaining_time": "10:09:03"} +{"current_steps": 235, "total_steps": 5364, "loss": 0.23715560138225555, "lr": 8.715083798882683e-06, "epoch": 0.13146853146853146, "percentage": 4.38, "elapsed_time": "0:27:53", "remaining_time": "10:08:50"} +{"current_steps": 236, "total_steps": 5364, "loss": 0.20219574868679047, "lr": 8.752327746741154e-06, "epoch": 0.13202797202797203, "percentage": 4.4, "elapsed_time": "0:28:00", "remaining_time": "10:08:38"} +{"current_steps": 237, "total_steps": 5364, "loss": 0.17427216470241547, "lr": 8.789571694599628e-06, "epoch": 0.13258741258741258, "percentage": 4.42, "elapsed_time": "0:28:08", "remaining_time": "10:08:37"} +{"current_steps": 238, "total_steps": 5364, "loss": 0.38720574975013733, "lr": 8.826815642458101e-06, "epoch": 0.13314685314685315, "percentage": 4.44, "elapsed_time": "0:28:14", "remaining_time": "10:08:25"} +{"current_steps": 239, "total_steps": 5364, "loss": 0.32538652420043945, "lr": 8.864059590316573e-06, "epoch": 0.1337062937062937, "percentage": 4.46, "elapsed_time": "0:28:22", "remaining_time": "10:08:19"} +{"current_steps": 240, "total_steps": 5364, "loss": 0.23696181178092957, "lr": 8.901303538175047e-06, "epoch": 0.13426573426573427, "percentage": 4.47, "elapsed_time": "0:28:29", "remaining_time": "10:08:21"} +{"current_steps": 241, "total_steps": 5364, "loss": 0.25718289613723755, "lr": 8.93854748603352e-06, "epoch": 0.13482517482517484, "percentage": 4.49, "elapsed_time": "0:28:36", "remaining_time": "10:08:04"} +{"current_steps": 242, "total_steps": 5364, "loss": 0.14556273818016052, "lr": 8.975791433891994e-06, "epoch": 0.13538461538461538, "percentage": 4.51, "elapsed_time": "0:28:43", "remaining_time": "10:07:58"} +{"current_steps": 243, "total_steps": 5364, "loss": 0.16508439183235168, "lr": 9.013035381750467e-06, "epoch": 0.13594405594405595, "percentage": 4.53, "elapsed_time": "0:28:50", "remaining_time": "10:07:46"} +{"current_steps": 244, "total_steps": 5364, "loss": 0.1431969553232193, "lr": 9.050279329608939e-06, "epoch": 0.1365034965034965, "percentage": 4.55, "elapsed_time": "0:28:58", "remaining_time": "10:07:53"} +{"current_steps": 245, "total_steps": 5364, "loss": 0.12685249745845795, "lr": 9.087523277467413e-06, "epoch": 0.13706293706293707, "percentage": 4.57, "elapsed_time": "0:29:05", "remaining_time": "10:07:47"} +{"current_steps": 246, "total_steps": 5364, "loss": 0.24326680600643158, "lr": 9.124767225325886e-06, "epoch": 0.1376223776223776, "percentage": 4.59, "elapsed_time": "0:29:13", "remaining_time": "10:07:51"} +{"current_steps": 247, "total_steps": 5364, "loss": 0.18389342725276947, "lr": 9.162011173184358e-06, "epoch": 0.13818181818181818, "percentage": 4.6, "elapsed_time": "0:29:19", "remaining_time": "10:07:39"} +{"current_steps": 248, "total_steps": 5364, "loss": 0.387996643781662, "lr": 9.199255121042831e-06, "epoch": 0.13874125874125876, "percentage": 4.62, "elapsed_time": "0:29:26", "remaining_time": "10:07:30"} +{"current_steps": 249, "total_steps": 5364, "loss": 0.1683967113494873, "lr": 9.236499068901305e-06, "epoch": 0.1393006993006993, "percentage": 4.64, "elapsed_time": "0:29:34", "remaining_time": "10:07:21"} +{"current_steps": 250, "total_steps": 5364, "loss": 0.38224977254867554, "lr": 9.273743016759777e-06, "epoch": 0.13986013986013987, "percentage": 4.66, "elapsed_time": "0:29:40", "remaining_time": "10:07:11"} +{"current_steps": 251, "total_steps": 5364, "loss": 0.2202875018119812, "lr": 9.31098696461825e-06, "epoch": 0.14041958041958041, "percentage": 4.68, "elapsed_time": "0:29:47", "remaining_time": "10:07:02"} +{"current_steps": 252, "total_steps": 5364, "loss": 0.0943504124879837, "lr": 9.348230912476724e-06, "epoch": 0.14097902097902099, "percentage": 4.7, "elapsed_time": "0:29:56", "remaining_time": "10:07:13"} +{"current_steps": 253, "total_steps": 5364, "loss": 0.051278285682201385, "lr": 9.385474860335197e-06, "epoch": 0.14153846153846153, "percentage": 4.72, "elapsed_time": "0:30:03", "remaining_time": "10:07:19"} +{"current_steps": 254, "total_steps": 5364, "loss": 0.16502758860588074, "lr": 9.422718808193669e-06, "epoch": 0.1420979020979021, "percentage": 4.74, "elapsed_time": "0:30:10", "remaining_time": "10:07:07"} +{"current_steps": 255, "total_steps": 5364, "loss": 0.2247346043586731, "lr": 9.459962756052143e-06, "epoch": 0.14265734265734265, "percentage": 4.75, "elapsed_time": "0:30:17", "remaining_time": "10:06:58"} +{"current_steps": 256, "total_steps": 5364, "loss": 0.18743066489696503, "lr": 9.497206703910616e-06, "epoch": 0.14321678321678322, "percentage": 4.77, "elapsed_time": "0:30:24", "remaining_time": "10:06:47"} +{"current_steps": 257, "total_steps": 5364, "loss": 0.13635236024856567, "lr": 9.534450651769088e-06, "epoch": 0.1437762237762238, "percentage": 4.79, "elapsed_time": "0:30:31", "remaining_time": "10:06:35"} +{"current_steps": 258, "total_steps": 5364, "loss": 0.29340460896492004, "lr": 9.571694599627561e-06, "epoch": 0.14433566433566433, "percentage": 4.81, "elapsed_time": "0:30:38", "remaining_time": "10:06:23"} +{"current_steps": 259, "total_steps": 5364, "loss": 0.2556872069835663, "lr": 9.608938547486035e-06, "epoch": 0.1448951048951049, "percentage": 4.83, "elapsed_time": "0:30:45", "remaining_time": "10:06:20"} +{"current_steps": 260, "total_steps": 5364, "loss": 0.27386772632598877, "lr": 9.646182495344507e-06, "epoch": 0.14545454545454545, "percentage": 4.85, "elapsed_time": "0:30:53", "remaining_time": "10:06:23"} +{"current_steps": 261, "total_steps": 5364, "loss": 0.03233690559864044, "lr": 9.68342644320298e-06, "epoch": 0.14601398601398602, "percentage": 4.87, "elapsed_time": "0:31:00", "remaining_time": "10:06:10"} +{"current_steps": 262, "total_steps": 5364, "loss": 0.26094627380371094, "lr": 9.720670391061454e-06, "epoch": 0.14657342657342656, "percentage": 4.88, "elapsed_time": "0:31:07", "remaining_time": "10:05:58"} +{"current_steps": 263, "total_steps": 5364, "loss": 0.1413864940404892, "lr": 9.757914338919926e-06, "epoch": 0.14713286713286713, "percentage": 4.9, "elapsed_time": "0:31:14", "remaining_time": "10:05:47"} +{"current_steps": 264, "total_steps": 5364, "loss": 0.1462213695049286, "lr": 9.795158286778399e-06, "epoch": 0.1476923076923077, "percentage": 4.92, "elapsed_time": "0:31:21", "remaining_time": "10:05:44"} +{"current_steps": 265, "total_steps": 5364, "loss": 0.44167637825012207, "lr": 9.832402234636873e-06, "epoch": 0.14825174825174825, "percentage": 4.94, "elapsed_time": "0:31:28", "remaining_time": "10:05:33"} +{"current_steps": 266, "total_steps": 5364, "loss": 0.4117547869682312, "lr": 9.869646182495344e-06, "epoch": 0.14881118881118882, "percentage": 4.96, "elapsed_time": "0:31:35", "remaining_time": "10:05:32"} +{"current_steps": 267, "total_steps": 5364, "loss": 0.3823855519294739, "lr": 9.906890130353818e-06, "epoch": 0.14937062937062937, "percentage": 4.98, "elapsed_time": "0:31:43", "remaining_time": "10:05:30"} +{"current_steps": 268, "total_steps": 5364, "loss": 0.14567437767982483, "lr": 9.944134078212291e-06, "epoch": 0.14993006993006994, "percentage": 5.0, "elapsed_time": "0:31:50", "remaining_time": "10:05:24"} +{"current_steps": 269, "total_steps": 5364, "loss": 0.32547974586486816, "lr": 9.981378026070763e-06, "epoch": 0.15048951048951048, "percentage": 5.01, "elapsed_time": "0:31:57", "remaining_time": "10:05:20"} +{"current_steps": 270, "total_steps": 5364, "loss": 0.21595175564289093, "lr": 1.0018621973929237e-05, "epoch": 0.15104895104895105, "percentage": 5.03, "elapsed_time": "0:32:04", "remaining_time": "10:05:17"} +{"current_steps": 271, "total_steps": 5364, "loss": 0.2322026938199997, "lr": 1.005586592178771e-05, "epoch": 0.1516083916083916, "percentage": 5.05, "elapsed_time": "0:32:12", "remaining_time": "10:05:16"} +{"current_steps": 272, "total_steps": 5364, "loss": 0.14051656424999237, "lr": 1.0093109869646182e-05, "epoch": 0.15216783216783217, "percentage": 5.07, "elapsed_time": "0:32:19", "remaining_time": "10:05:07"} +{"current_steps": 273, "total_steps": 5364, "loss": 0.30927422642707825, "lr": 1.0130353817504657e-05, "epoch": 0.15272727272727274, "percentage": 5.09, "elapsed_time": "0:32:26", "remaining_time": "10:04:56"} +{"current_steps": 274, "total_steps": 5364, "loss": 0.14183266460895538, "lr": 1.0167597765363129e-05, "epoch": 0.15328671328671328, "percentage": 5.11, "elapsed_time": "0:32:33", "remaining_time": "10:04:44"} +{"current_steps": 275, "total_steps": 5364, "loss": 0.11704777181148529, "lr": 1.0204841713221603e-05, "epoch": 0.15384615384615385, "percentage": 5.13, "elapsed_time": "0:32:40", "remaining_time": "10:04:39"} +{"current_steps": 276, "total_steps": 5364, "loss": 0.17365817725658417, "lr": 1.0242085661080074e-05, "epoch": 0.1544055944055944, "percentage": 5.15, "elapsed_time": "0:32:47", "remaining_time": "10:04:27"} +{"current_steps": 277, "total_steps": 5364, "loss": 0.14538846909999847, "lr": 1.0279329608938548e-05, "epoch": 0.15496503496503497, "percentage": 5.16, "elapsed_time": "0:32:54", "remaining_time": "10:04:17"} +{"current_steps": 278, "total_steps": 5364, "loss": 0.2588534653186798, "lr": 1.031657355679702e-05, "epoch": 0.15552447552447551, "percentage": 5.18, "elapsed_time": "0:33:01", "remaining_time": "10:04:07"} +{"current_steps": 279, "total_steps": 5364, "loss": 0.2705457806587219, "lr": 1.0353817504655495e-05, "epoch": 0.15608391608391609, "percentage": 5.2, "elapsed_time": "0:33:08", "remaining_time": "10:03:58"} +{"current_steps": 280, "total_steps": 5364, "loss": 0.24899393320083618, "lr": 1.0391061452513967e-05, "epoch": 0.15664335664335666, "percentage": 5.22, "elapsed_time": "0:33:15", "remaining_time": "10:03:54"} +{"current_steps": 281, "total_steps": 5364, "loss": 0.20160573720932007, "lr": 1.042830540037244e-05, "epoch": 0.1572027972027972, "percentage": 5.24, "elapsed_time": "0:33:22", "remaining_time": "10:03:48"} +{"current_steps": 282, "total_steps": 5364, "loss": 0.25672706961631775, "lr": 1.0465549348230912e-05, "epoch": 0.15776223776223777, "percentage": 5.26, "elapsed_time": "0:33:30", "remaining_time": "10:03:46"} +{"current_steps": 283, "total_steps": 5364, "loss": 0.16140006482601166, "lr": 1.0502793296089386e-05, "epoch": 0.15832167832167832, "percentage": 5.28, "elapsed_time": "0:33:37", "remaining_time": "10:03:47"} +{"current_steps": 284, "total_steps": 5364, "loss": 0.30654963850975037, "lr": 1.054003724394786e-05, "epoch": 0.1588811188811189, "percentage": 5.29, "elapsed_time": "0:33:44", "remaining_time": "10:03:35"} +{"current_steps": 285, "total_steps": 5364, "loss": 0.15222902595996857, "lr": 1.0577281191806333e-05, "epoch": 0.15944055944055943, "percentage": 5.31, "elapsed_time": "0:33:52", "remaining_time": "10:03:33"} +{"current_steps": 286, "total_steps": 5364, "loss": 0.2799169719219208, "lr": 1.0614525139664806e-05, "epoch": 0.16, "percentage": 5.33, "elapsed_time": "0:33:58", "remaining_time": "10:03:22"} +{"current_steps": 287, "total_steps": 5364, "loss": 0.2820596396923065, "lr": 1.0651769087523278e-05, "epoch": 0.16055944055944055, "percentage": 5.35, "elapsed_time": "0:34:05", "remaining_time": "10:03:13"} +{"current_steps": 288, "total_steps": 5364, "loss": 0.26518529653549194, "lr": 1.0689013035381753e-05, "epoch": 0.16111888111888112, "percentage": 5.37, "elapsed_time": "0:34:12", "remaining_time": "10:03:01"} +{"current_steps": 289, "total_steps": 5364, "loss": 0.25247806310653687, "lr": 1.0726256983240225e-05, "epoch": 0.1616783216783217, "percentage": 5.39, "elapsed_time": "0:34:19", "remaining_time": "10:02:51"} +{"current_steps": 290, "total_steps": 5364, "loss": 0.3577049970626831, "lr": 1.0763500931098698e-05, "epoch": 0.16223776223776223, "percentage": 5.41, "elapsed_time": "0:34:26", "remaining_time": "10:02:40"} +{"current_steps": 291, "total_steps": 5364, "loss": 0.31357356905937195, "lr": 1.080074487895717e-05, "epoch": 0.1627972027972028, "percentage": 5.43, "elapsed_time": "0:34:33", "remaining_time": "10:02:28"} +{"current_steps": 292, "total_steps": 5364, "loss": 0.17181327939033508, "lr": 1.0837988826815644e-05, "epoch": 0.16335664335664335, "percentage": 5.44, "elapsed_time": "0:34:40", "remaining_time": "10:02:17"} +{"current_steps": 293, "total_steps": 5364, "loss": 0.38565880060195923, "lr": 1.0875232774674116e-05, "epoch": 0.16391608391608392, "percentage": 5.46, "elapsed_time": "0:34:47", "remaining_time": "10:02:04"} +{"current_steps": 294, "total_steps": 5364, "loss": 0.4313648045063019, "lr": 1.091247672253259e-05, "epoch": 0.16447552447552446, "percentage": 5.48, "elapsed_time": "0:34:54", "remaining_time": "10:01:58"} +{"current_steps": 295, "total_steps": 5364, "loss": 0.2349538952112198, "lr": 1.0949720670391063e-05, "epoch": 0.16503496503496504, "percentage": 5.5, "elapsed_time": "0:35:01", "remaining_time": "10:01:47"} +{"current_steps": 296, "total_steps": 5364, "loss": 0.08374534547328949, "lr": 1.0986964618249536e-05, "epoch": 0.1655944055944056, "percentage": 5.52, "elapsed_time": "0:35:08", "remaining_time": "10:01:46"} +{"current_steps": 297, "total_steps": 5364, "loss": 0.237091526389122, "lr": 1.1024208566108008e-05, "epoch": 0.16615384615384615, "percentage": 5.54, "elapsed_time": "0:35:15", "remaining_time": "10:01:37"} +{"current_steps": 298, "total_steps": 5364, "loss": 0.16771084070205688, "lr": 1.1061452513966481e-05, "epoch": 0.16671328671328672, "percentage": 5.56, "elapsed_time": "0:35:22", "remaining_time": "10:01:26"} +{"current_steps": 299, "total_steps": 5364, "loss": 0.2835752069950104, "lr": 1.1098696461824953e-05, "epoch": 0.16727272727272727, "percentage": 5.57, "elapsed_time": "0:35:29", "remaining_time": "10:01:16"} +{"current_steps": 300, "total_steps": 5364, "loss": 0.06887897849082947, "lr": 1.1135940409683428e-05, "epoch": 0.16783216783216784, "percentage": 5.59, "elapsed_time": "0:35:36", "remaining_time": "10:01:05"} +{"current_steps": 301, "total_steps": 5364, "loss": 0.3984430730342865, "lr": 1.11731843575419e-05, "epoch": 0.16839160839160838, "percentage": 5.61, "elapsed_time": "0:35:43", "remaining_time": "10:00:56"} +{"current_steps": 302, "total_steps": 5364, "loss": 0.35076266527175903, "lr": 1.1210428305400374e-05, "epoch": 0.16895104895104895, "percentage": 5.63, "elapsed_time": "0:35:50", "remaining_time": "10:00:45"} +{"current_steps": 303, "total_steps": 5364, "loss": 0.21225865185260773, "lr": 1.1247672253258846e-05, "epoch": 0.1695104895104895, "percentage": 5.65, "elapsed_time": "0:35:57", "remaining_time": "10:00:38"} +{"current_steps": 304, "total_steps": 5364, "loss": 0.15718327462673187, "lr": 1.1284916201117319e-05, "epoch": 0.17006993006993007, "percentage": 5.67, "elapsed_time": "0:36:05", "remaining_time": "10:00:36"} +{"current_steps": 305, "total_steps": 5364, "loss": 0.22824551165103912, "lr": 1.1322160148975791e-05, "epoch": 0.17062937062937064, "percentage": 5.69, "elapsed_time": "0:36:12", "remaining_time": "10:00:32"} +{"current_steps": 306, "total_steps": 5364, "loss": 0.6170090436935425, "lr": 1.1359404096834266e-05, "epoch": 0.17118881118881119, "percentage": 5.7, "elapsed_time": "0:36:19", "remaining_time": "10:00:26"} +{"current_steps": 307, "total_steps": 5364, "loss": 0.11314629763364792, "lr": 1.1396648044692738e-05, "epoch": 0.17174825174825176, "percentage": 5.72, "elapsed_time": "0:36:26", "remaining_time": "10:00:20"} +{"current_steps": 308, "total_steps": 5364, "loss": 0.29253992438316345, "lr": 1.1433891992551211e-05, "epoch": 0.1723076923076923, "percentage": 5.74, "elapsed_time": "0:36:33", "remaining_time": "10:00:09"} +{"current_steps": 309, "total_steps": 5364, "loss": 0.16079992055892944, "lr": 1.1471135940409683e-05, "epoch": 0.17286713286713287, "percentage": 5.76, "elapsed_time": "0:36:40", "remaining_time": "9:59:58"} +{"current_steps": 310, "total_steps": 5364, "loss": 0.23877716064453125, "lr": 1.1508379888268157e-05, "epoch": 0.17342657342657342, "percentage": 5.78, "elapsed_time": "0:36:47", "remaining_time": "9:59:53"} +{"current_steps": 311, "total_steps": 5364, "loss": 0.09725035727024078, "lr": 1.1545623836126629e-05, "epoch": 0.173986013986014, "percentage": 5.8, "elapsed_time": "0:36:54", "remaining_time": "9:59:44"} +{"current_steps": 312, "total_steps": 5364, "loss": 0.3041437268257141, "lr": 1.1582867783985104e-05, "epoch": 0.17454545454545456, "percentage": 5.82, "elapsed_time": "0:37:01", "remaining_time": "9:59:32"} +{"current_steps": 313, "total_steps": 5364, "loss": 0.14455457031726837, "lr": 1.1620111731843577e-05, "epoch": 0.1751048951048951, "percentage": 5.84, "elapsed_time": "0:37:08", "remaining_time": "9:59:29"} +{"current_steps": 314, "total_steps": 5364, "loss": 0.13114410638809204, "lr": 1.1657355679702049e-05, "epoch": 0.17566433566433567, "percentage": 5.85, "elapsed_time": "0:37:15", "remaining_time": "9:59:20"} +{"current_steps": 315, "total_steps": 5364, "loss": 0.1356728971004486, "lr": 1.1694599627560523e-05, "epoch": 0.17622377622377622, "percentage": 5.87, "elapsed_time": "0:37:22", "remaining_time": "9:59:09"} +{"current_steps": 316, "total_steps": 5364, "loss": 0.2697957754135132, "lr": 1.1731843575418994e-05, "epoch": 0.1767832167832168, "percentage": 5.89, "elapsed_time": "0:37:29", "remaining_time": "9:59:01"} +{"current_steps": 317, "total_steps": 5364, "loss": 0.41352230310440063, "lr": 1.176908752327747e-05, "epoch": 0.17734265734265733, "percentage": 5.91, "elapsed_time": "0:37:36", "remaining_time": "9:58:50"} +{"current_steps": 318, "total_steps": 5364, "loss": 0.1763618290424347, "lr": 1.1806331471135941e-05, "epoch": 0.1779020979020979, "percentage": 5.93, "elapsed_time": "0:37:44", "remaining_time": "9:58:46"} +{"current_steps": 319, "total_steps": 5364, "loss": 0.06649969518184662, "lr": 1.1843575418994415e-05, "epoch": 0.17846153846153845, "percentage": 5.95, "elapsed_time": "0:37:50", "remaining_time": "9:58:35"} +{"current_steps": 320, "total_steps": 5364, "loss": 0.2024984359741211, "lr": 1.1880819366852887e-05, "epoch": 0.17902097902097902, "percentage": 5.97, "elapsed_time": "0:37:58", "remaining_time": "9:58:33"} +{"current_steps": 321, "total_steps": 5364, "loss": 0.18501436710357666, "lr": 1.1918063314711362e-05, "epoch": 0.1795804195804196, "percentage": 5.98, "elapsed_time": "0:38:05", "remaining_time": "9:58:31"} +{"current_steps": 322, "total_steps": 5364, "loss": 0.28460073471069336, "lr": 1.1955307262569834e-05, "epoch": 0.18013986013986014, "percentage": 6.0, "elapsed_time": "0:38:12", "remaining_time": "9:58:20"} +{"current_steps": 323, "total_steps": 5364, "loss": 0.18380779027938843, "lr": 1.1992551210428307e-05, "epoch": 0.1806993006993007, "percentage": 6.02, "elapsed_time": "0:38:19", "remaining_time": "9:58:09"} +{"current_steps": 324, "total_steps": 5364, "loss": 0.4374958574771881, "lr": 1.2029795158286779e-05, "epoch": 0.18125874125874125, "percentage": 6.04, "elapsed_time": "0:38:26", "remaining_time": "9:57:59"} +{"current_steps": 325, "total_steps": 5364, "loss": 0.21650944650173187, "lr": 1.2067039106145253e-05, "epoch": 0.18181818181818182, "percentage": 6.06, "elapsed_time": "0:38:33", "remaining_time": "9:57:48"} +{"current_steps": 326, "total_steps": 5364, "loss": 0.21292313933372498, "lr": 1.2104283054003724e-05, "epoch": 0.18237762237762237, "percentage": 6.08, "elapsed_time": "0:38:40", "remaining_time": "9:57:39"} +{"current_steps": 327, "total_steps": 5364, "loss": 0.3288453221321106, "lr": 1.21415270018622e-05, "epoch": 0.18293706293706294, "percentage": 6.1, "elapsed_time": "0:38:47", "remaining_time": "9:57:33"} +{"current_steps": 328, "total_steps": 5364, "loss": 0.1771489679813385, "lr": 1.2178770949720671e-05, "epoch": 0.1834965034965035, "percentage": 6.11, "elapsed_time": "0:38:54", "remaining_time": "9:57:21"} +{"current_steps": 329, "total_steps": 5364, "loss": 0.11406119167804718, "lr": 1.2216014897579145e-05, "epoch": 0.18405594405594405, "percentage": 6.13, "elapsed_time": "0:39:01", "remaining_time": "9:57:19"} +{"current_steps": 330, "total_steps": 5364, "loss": 0.24953454732894897, "lr": 1.2253258845437617e-05, "epoch": 0.18461538461538463, "percentage": 6.15, "elapsed_time": "0:39:08", "remaining_time": "9:57:10"} +{"current_steps": 331, "total_steps": 5364, "loss": 0.07181106507778168, "lr": 1.229050279329609e-05, "epoch": 0.18517482517482517, "percentage": 6.17, "elapsed_time": "0:39:16", "remaining_time": "9:57:05"} +{"current_steps": 332, "total_steps": 5364, "loss": 0.2392260581254959, "lr": 1.2327746741154562e-05, "epoch": 0.18573426573426574, "percentage": 6.19, "elapsed_time": "0:39:22", "remaining_time": "9:56:54"} +{"current_steps": 333, "total_steps": 5364, "loss": 0.10333205759525299, "lr": 1.2364990689013037e-05, "epoch": 0.18629370629370628, "percentage": 6.21, "elapsed_time": "0:39:30", "remaining_time": "9:56:54"} +{"current_steps": 334, "total_steps": 5364, "loss": 0.1868247389793396, "lr": 1.2402234636871509e-05, "epoch": 0.18685314685314686, "percentage": 6.23, "elapsed_time": "0:39:37", "remaining_time": "9:56:43"} +{"current_steps": 335, "total_steps": 5364, "loss": 0.0466235987842083, "lr": 1.2439478584729983e-05, "epoch": 0.1874125874125874, "percentage": 6.25, "elapsed_time": "0:39:44", "remaining_time": "9:56:41"} +{"current_steps": 336, "total_steps": 5364, "loss": 0.2161976397037506, "lr": 1.2476722532588454e-05, "epoch": 0.18797202797202797, "percentage": 6.26, "elapsed_time": "0:39:51", "remaining_time": "9:56:30"} +{"current_steps": 337, "total_steps": 5364, "loss": 0.2055448591709137, "lr": 1.2513966480446928e-05, "epoch": 0.18853146853146854, "percentage": 6.28, "elapsed_time": "0:39:58", "remaining_time": "9:56:24"} +{"current_steps": 338, "total_steps": 5364, "loss": 0.25013861060142517, "lr": 1.25512104283054e-05, "epoch": 0.1890909090909091, "percentage": 6.3, "elapsed_time": "0:40:06", "remaining_time": "9:56:24"} +{"current_steps": 339, "total_steps": 5364, "loss": 0.1826154589653015, "lr": 1.2588454376163875e-05, "epoch": 0.18965034965034966, "percentage": 6.32, "elapsed_time": "0:40:13", "remaining_time": "9:56:13"} +{"current_steps": 340, "total_steps": 5364, "loss": 0.4009782671928406, "lr": 1.2625698324022347e-05, "epoch": 0.1902097902097902, "percentage": 6.34, "elapsed_time": "0:40:20", "remaining_time": "9:56:04"} +{"current_steps": 341, "total_steps": 5364, "loss": 0.2914050221443176, "lr": 1.266294227188082e-05, "epoch": 0.19076923076923077, "percentage": 6.36, "elapsed_time": "0:40:27", "remaining_time": "9:55:57"} +{"current_steps": 342, "total_steps": 5364, "loss": 0.11122561991214752, "lr": 1.2700186219739294e-05, "epoch": 0.19132867132867132, "percentage": 6.38, "elapsed_time": "0:40:35", "remaining_time": "9:55:58"} +{"current_steps": 343, "total_steps": 5364, "loss": 0.18652702867984772, "lr": 1.2737430167597766e-05, "epoch": 0.1918881118881119, "percentage": 6.39, "elapsed_time": "0:40:42", "remaining_time": "9:55:48"} +{"current_steps": 344, "total_steps": 5364, "loss": 0.13408324122428894, "lr": 1.277467411545624e-05, "epoch": 0.19244755244755246, "percentage": 6.41, "elapsed_time": "0:40:49", "remaining_time": "9:55:46"} +{"current_steps": 345, "total_steps": 5364, "loss": 0.13975383341312408, "lr": 1.2811918063314713e-05, "epoch": 0.193006993006993, "percentage": 6.43, "elapsed_time": "0:40:56", "remaining_time": "9:55:35"} +{"current_steps": 346, "total_steps": 5364, "loss": 0.15843935310840607, "lr": 1.2849162011173186e-05, "epoch": 0.19356643356643358, "percentage": 6.45, "elapsed_time": "0:41:03", "remaining_time": "9:55:24"} +{"current_steps": 347, "total_steps": 5364, "loss": 0.37123599648475647, "lr": 1.2886405959031658e-05, "epoch": 0.19412587412587412, "percentage": 6.47, "elapsed_time": "0:41:10", "remaining_time": "9:55:18"} +{"current_steps": 348, "total_steps": 5364, "loss": 0.17427772283554077, "lr": 1.2923649906890131e-05, "epoch": 0.1946853146853147, "percentage": 6.49, "elapsed_time": "0:41:17", "remaining_time": "9:55:09"} +{"current_steps": 349, "total_steps": 5364, "loss": 0.4043179452419281, "lr": 1.2960893854748603e-05, "epoch": 0.19524475524475524, "percentage": 6.51, "elapsed_time": "0:41:24", "remaining_time": "9:54:58"} +{"current_steps": 350, "total_steps": 5364, "loss": 0.22299587726593018, "lr": 1.2998137802607078e-05, "epoch": 0.1958041958041958, "percentage": 6.52, "elapsed_time": "0:41:31", "remaining_time": "9:54:49"} +{"current_steps": 351, "total_steps": 5364, "loss": 0.30284884572029114, "lr": 1.303538175046555e-05, "epoch": 0.19636363636363635, "percentage": 6.54, "elapsed_time": "0:41:38", "remaining_time": "9:54:41"} +{"current_steps": 352, "total_steps": 5364, "loss": 0.3166622221469879, "lr": 1.3072625698324024e-05, "epoch": 0.19692307692307692, "percentage": 6.56, "elapsed_time": "0:41:45", "remaining_time": "9:54:30"} +{"current_steps": 353, "total_steps": 5364, "loss": 0.395719051361084, "lr": 1.3109869646182496e-05, "epoch": 0.1974825174825175, "percentage": 6.58, "elapsed_time": "0:41:52", "remaining_time": "9:54:20"} +{"current_steps": 354, "total_steps": 5364, "loss": 0.16517296433448792, "lr": 1.314711359404097e-05, "epoch": 0.19804195804195804, "percentage": 6.6, "elapsed_time": "0:41:59", "remaining_time": "9:54:12"} +{"current_steps": 355, "total_steps": 5364, "loss": 0.14390204846858978, "lr": 1.3184357541899443e-05, "epoch": 0.1986013986013986, "percentage": 6.62, "elapsed_time": "0:42:06", "remaining_time": "9:54:08"} +{"current_steps": 356, "total_steps": 5364, "loss": 0.15885721147060394, "lr": 1.3221601489757916e-05, "epoch": 0.19916083916083915, "percentage": 6.64, "elapsed_time": "0:42:13", "remaining_time": "9:53:59"} +{"current_steps": 357, "total_steps": 5364, "loss": 0.18029023706912994, "lr": 1.3258845437616388e-05, "epoch": 0.19972027972027973, "percentage": 6.66, "elapsed_time": "0:42:20", "remaining_time": "9:53:53"} +{"current_steps": 358, "total_steps": 5364, "loss": 0.24055664241313934, "lr": 1.3296089385474861e-05, "epoch": 0.20027972027972027, "percentage": 6.67, "elapsed_time": "0:42:27", "remaining_time": "9:53:44"} +{"current_steps": 359, "total_steps": 5364, "loss": 0.40840989351272583, "lr": 1.3333333333333333e-05, "epoch": 0.20083916083916084, "percentage": 6.69, "elapsed_time": "0:42:34", "remaining_time": "9:53:38"} +{"current_steps": 360, "total_steps": 5364, "loss": 0.3558077812194824, "lr": 1.3370577281191808e-05, "epoch": 0.2013986013986014, "percentage": 6.71, "elapsed_time": "0:42:41", "remaining_time": "9:53:24"} +{"current_steps": 361, "total_steps": 5364, "loss": 0.10893086344003677, "lr": 1.340782122905028e-05, "epoch": 0.20195804195804196, "percentage": 6.73, "elapsed_time": "0:42:48", "remaining_time": "9:53:13"} +{"current_steps": 362, "total_steps": 5364, "loss": 0.26309388875961304, "lr": 1.3445065176908754e-05, "epoch": 0.20251748251748253, "percentage": 6.75, "elapsed_time": "0:42:55", "remaining_time": "9:53:03"} +{"current_steps": 363, "total_steps": 5364, "loss": 0.1265127807855606, "lr": 1.3482309124767226e-05, "epoch": 0.20307692307692307, "percentage": 6.77, "elapsed_time": "0:43:02", "remaining_time": "9:52:52"} +{"current_steps": 364, "total_steps": 5364, "loss": 0.23097705841064453, "lr": 1.3519553072625699e-05, "epoch": 0.20363636363636364, "percentage": 6.79, "elapsed_time": "0:43:09", "remaining_time": "9:52:48"} +{"current_steps": 365, "total_steps": 5364, "loss": 0.06925821304321289, "lr": 1.3556797020484171e-05, "epoch": 0.2041958041958042, "percentage": 6.8, "elapsed_time": "0:43:16", "remaining_time": "9:52:41"} +{"current_steps": 366, "total_steps": 5364, "loss": 0.16943222284317017, "lr": 1.3594040968342646e-05, "epoch": 0.20475524475524476, "percentage": 6.82, "elapsed_time": "0:43:24", "remaining_time": "9:52:45"} +{"current_steps": 367, "total_steps": 5364, "loss": 0.0987367033958435, "lr": 1.3631284916201118e-05, "epoch": 0.2053146853146853, "percentage": 6.84, "elapsed_time": "0:43:32", "remaining_time": "9:52:44"} +{"current_steps": 368, "total_steps": 5364, "loss": 0.17131972312927246, "lr": 1.3668528864059591e-05, "epoch": 0.20587412587412587, "percentage": 6.86, "elapsed_time": "0:43:38", "remaining_time": "9:52:35"} +{"current_steps": 369, "total_steps": 5364, "loss": 0.16666218638420105, "lr": 1.3705772811918065e-05, "epoch": 0.20643356643356645, "percentage": 6.88, "elapsed_time": "0:43:45", "remaining_time": "9:52:24"} +{"current_steps": 370, "total_steps": 5364, "loss": 0.1290956735610962, "lr": 1.3743016759776537e-05, "epoch": 0.206993006993007, "percentage": 6.9, "elapsed_time": "0:43:52", "remaining_time": "9:52:15"} +{"current_steps": 371, "total_steps": 5364, "loss": 0.4316938519477844, "lr": 1.3780260707635012e-05, "epoch": 0.20755244755244756, "percentage": 6.92, "elapsed_time": "0:44:00", "remaining_time": "9:52:09"} +{"current_steps": 372, "total_steps": 5364, "loss": 0.31795668601989746, "lr": 1.3817504655493484e-05, "epoch": 0.2081118881118881, "percentage": 6.94, "elapsed_time": "0:44:06", "remaining_time": "9:51:59"} +{"current_steps": 373, "total_steps": 5364, "loss": 0.10897572338581085, "lr": 1.3854748603351957e-05, "epoch": 0.20867132867132868, "percentage": 6.95, "elapsed_time": "0:44:14", "remaining_time": "9:51:56"} +{"current_steps": 374, "total_steps": 5364, "loss": 0.22178541123867035, "lr": 1.389199255121043e-05, "epoch": 0.20923076923076922, "percentage": 6.97, "elapsed_time": "0:44:21", "remaining_time": "9:51:52"} +{"current_steps": 375, "total_steps": 5364, "loss": 0.15575313568115234, "lr": 1.3929236499068903e-05, "epoch": 0.2097902097902098, "percentage": 6.99, "elapsed_time": "0:44:28", "remaining_time": "9:51:41"} +{"current_steps": 376, "total_steps": 5364, "loss": 0.3155873119831085, "lr": 1.3966480446927374e-05, "epoch": 0.21034965034965036, "percentage": 7.01, "elapsed_time": "0:44:35", "remaining_time": "9:51:31"} +{"current_steps": 377, "total_steps": 5364, "loss": 0.25676119327545166, "lr": 1.400372439478585e-05, "epoch": 0.2109090909090909, "percentage": 7.03, "elapsed_time": "0:44:42", "remaining_time": "9:51:21"} +{"current_steps": 378, "total_steps": 5364, "loss": 0.18600091338157654, "lr": 1.4040968342644321e-05, "epoch": 0.21146853146853148, "percentage": 7.05, "elapsed_time": "0:44:49", "remaining_time": "9:51:17"} +{"current_steps": 379, "total_steps": 5364, "loss": 0.1625310480594635, "lr": 1.4078212290502795e-05, "epoch": 0.21202797202797202, "percentage": 7.07, "elapsed_time": "0:44:56", "remaining_time": "9:51:06"} +{"current_steps": 380, "total_steps": 5364, "loss": 0.42895030975341797, "lr": 1.4115456238361267e-05, "epoch": 0.2125874125874126, "percentage": 7.08, "elapsed_time": "0:45:04", "remaining_time": "9:51:09"} +{"current_steps": 381, "total_steps": 5364, "loss": 0.29572248458862305, "lr": 1.415270018621974e-05, "epoch": 0.21314685314685314, "percentage": 7.1, "elapsed_time": "0:45:11", "remaining_time": "9:50:59"} +{"current_steps": 382, "total_steps": 5364, "loss": 0.21049325168132782, "lr": 1.4189944134078212e-05, "epoch": 0.2137062937062937, "percentage": 7.12, "elapsed_time": "0:45:18", "remaining_time": "9:50:53"} +{"current_steps": 383, "total_steps": 5364, "loss": 0.4587564468383789, "lr": 1.4227188081936687e-05, "epoch": 0.21426573426573425, "percentage": 7.14, "elapsed_time": "0:45:25", "remaining_time": "9:50:47"} +{"current_steps": 384, "total_steps": 5364, "loss": 0.19551566243171692, "lr": 1.426443202979516e-05, "epoch": 0.21482517482517482, "percentage": 7.16, "elapsed_time": "0:45:32", "remaining_time": "9:50:42"} +{"current_steps": 385, "total_steps": 5364, "loss": 0.30103886127471924, "lr": 1.4301675977653633e-05, "epoch": 0.2153846153846154, "percentage": 7.18, "elapsed_time": "0:45:39", "remaining_time": "9:50:31"} +{"current_steps": 386, "total_steps": 5364, "loss": 0.3774051070213318, "lr": 1.4338919925512104e-05, "epoch": 0.21594405594405594, "percentage": 7.2, "elapsed_time": "0:45:46", "remaining_time": "9:50:24"} +{"current_steps": 387, "total_steps": 5364, "loss": 0.12075591832399368, "lr": 1.4376163873370578e-05, "epoch": 0.2165034965034965, "percentage": 7.21, "elapsed_time": "0:45:53", "remaining_time": "9:50:16"} +{"current_steps": 388, "total_steps": 5364, "loss": 0.2473677098751068, "lr": 1.4413407821229052e-05, "epoch": 0.21706293706293706, "percentage": 7.23, "elapsed_time": "0:46:00", "remaining_time": "9:50:09"} +{"current_steps": 389, "total_steps": 5364, "loss": 0.26568669080734253, "lr": 1.4450651769087525e-05, "epoch": 0.21762237762237763, "percentage": 7.25, "elapsed_time": "0:46:07", "remaining_time": "9:50:00"} +{"current_steps": 390, "total_steps": 5364, "loss": 0.22834305465221405, "lr": 1.4487895716945997e-05, "epoch": 0.21818181818181817, "percentage": 7.27, "elapsed_time": "0:46:14", "remaining_time": "9:49:50"} +{"current_steps": 391, "total_steps": 5364, "loss": 0.19060564041137695, "lr": 1.452513966480447e-05, "epoch": 0.21874125874125874, "percentage": 7.29, "elapsed_time": "0:46:22", "remaining_time": "9:49:46"} +{"current_steps": 392, "total_steps": 5364, "loss": 0.3759705424308777, "lr": 1.4562383612662942e-05, "epoch": 0.21930069930069931, "percentage": 7.31, "elapsed_time": "0:46:29", "remaining_time": "9:49:36"} +{"current_steps": 393, "total_steps": 5364, "loss": 0.1936790645122528, "lr": 1.4599627560521417e-05, "epoch": 0.21986013986013986, "percentage": 7.33, "elapsed_time": "0:46:36", "remaining_time": "9:49:27"} +{"current_steps": 394, "total_steps": 5364, "loss": 0.21975558996200562, "lr": 1.463687150837989e-05, "epoch": 0.22041958041958043, "percentage": 7.35, "elapsed_time": "0:46:43", "remaining_time": "9:49:18"} +{"current_steps": 395, "total_steps": 5364, "loss": 0.2661827504634857, "lr": 1.4674115456238363e-05, "epoch": 0.22097902097902097, "percentage": 7.36, "elapsed_time": "0:46:49", "remaining_time": "9:49:08"} +{"current_steps": 396, "total_steps": 5364, "loss": 0.41784244775772095, "lr": 1.4711359404096834e-05, "epoch": 0.22153846153846155, "percentage": 7.38, "elapsed_time": "0:46:57", "remaining_time": "9:49:04"} +{"current_steps": 397, "total_steps": 5364, "loss": 0.375749409198761, "lr": 1.4748603351955308e-05, "epoch": 0.2220979020979021, "percentage": 7.4, "elapsed_time": "0:47:04", "remaining_time": "9:48:54"} +{"current_steps": 398, "total_steps": 5364, "loss": 0.05348823219537735, "lr": 1.4785847299813783e-05, "epoch": 0.22265734265734266, "percentage": 7.42, "elapsed_time": "0:47:11", "remaining_time": "9:48:53"} +{"current_steps": 399, "total_steps": 5364, "loss": 0.1662866175174713, "lr": 1.4823091247672255e-05, "epoch": 0.2232167832167832, "percentage": 7.44, "elapsed_time": "0:47:19", "remaining_time": "9:48:49"} +{"current_steps": 400, "total_steps": 5364, "loss": 0.1660555601119995, "lr": 1.4860335195530729e-05, "epoch": 0.22377622377622378, "percentage": 7.46, "elapsed_time": "0:47:26", "remaining_time": "9:48:42"} +{"current_steps": 401, "total_steps": 5364, "loss": 0.18232253193855286, "lr": 1.48975791433892e-05, "epoch": 0.22433566433566435, "percentage": 7.48, "elapsed_time": "0:47:33", "remaining_time": "9:48:35"} +{"current_steps": 402, "total_steps": 5364, "loss": 0.2067871391773224, "lr": 1.4934823091247674e-05, "epoch": 0.2248951048951049, "percentage": 7.49, "elapsed_time": "0:47:40", "remaining_time": "9:48:32"} +{"current_steps": 403, "total_steps": 5364, "loss": 0.2378668189048767, "lr": 1.4972067039106146e-05, "epoch": 0.22545454545454546, "percentage": 7.51, "elapsed_time": "0:47:47", "remaining_time": "9:48:22"} +{"current_steps": 404, "total_steps": 5364, "loss": 0.10804593563079834, "lr": 1.5009310986964621e-05, "epoch": 0.226013986013986, "percentage": 7.53, "elapsed_time": "0:47:54", "remaining_time": "9:48:12"} +{"current_steps": 405, "total_steps": 5364, "loss": 0.21366216242313385, "lr": 1.5046554934823093e-05, "epoch": 0.22657342657342658, "percentage": 7.55, "elapsed_time": "0:48:01", "remaining_time": "9:48:06"} +{"current_steps": 406, "total_steps": 5364, "loss": 0.3075847327709198, "lr": 1.5083798882681566e-05, "epoch": 0.22713286713286712, "percentage": 7.57, "elapsed_time": "0:48:09", "remaining_time": "9:48:07"} +{"current_steps": 407, "total_steps": 5364, "loss": 0.21630480885505676, "lr": 1.5121042830540038e-05, "epoch": 0.2276923076923077, "percentage": 7.59, "elapsed_time": "0:48:16", "remaining_time": "9:47:59"} +{"current_steps": 408, "total_steps": 5364, "loss": 0.0948122888803482, "lr": 1.5158286778398512e-05, "epoch": 0.22825174825174827, "percentage": 7.61, "elapsed_time": "0:48:23", "remaining_time": "9:47:46"} +{"current_steps": 409, "total_steps": 5364, "loss": 0.0394311286509037, "lr": 1.5195530726256983e-05, "epoch": 0.2288111888111888, "percentage": 7.62, "elapsed_time": "0:48:30", "remaining_time": "9:47:40"} +{"current_steps": 410, "total_steps": 5364, "loss": 0.004848596174269915, "lr": 1.5232774674115459e-05, "epoch": 0.22937062937062938, "percentage": 7.64, "elapsed_time": "0:48:37", "remaining_time": "9:47:28"} +{"current_steps": 411, "total_steps": 5364, "loss": 0.16655394434928894, "lr": 1.527001862197393e-05, "epoch": 0.22993006993006992, "percentage": 7.66, "elapsed_time": "0:48:44", "remaining_time": "9:47:21"} +{"current_steps": 412, "total_steps": 5364, "loss": 0.4191613793373108, "lr": 1.5307262569832404e-05, "epoch": 0.2304895104895105, "percentage": 7.68, "elapsed_time": "0:48:51", "remaining_time": "9:47:11"} +{"current_steps": 413, "total_steps": 5364, "loss": 0.22198426723480225, "lr": 1.5344506517690876e-05, "epoch": 0.23104895104895104, "percentage": 7.7, "elapsed_time": "0:48:58", "remaining_time": "9:47:01"} +{"current_steps": 414, "total_steps": 5364, "loss": 0.11012548953294754, "lr": 1.538175046554935e-05, "epoch": 0.2316083916083916, "percentage": 7.72, "elapsed_time": "0:49:05", "remaining_time": "9:47:03"} +{"current_steps": 415, "total_steps": 5364, "loss": 0.18286389112472534, "lr": 1.5418994413407823e-05, "epoch": 0.23216783216783216, "percentage": 7.74, "elapsed_time": "0:49:12", "remaining_time": "9:46:53"} +{"current_steps": 416, "total_steps": 5364, "loss": 0.11632704734802246, "lr": 1.5456238361266295e-05, "epoch": 0.23272727272727273, "percentage": 7.76, "elapsed_time": "0:49:19", "remaining_time": "9:46:45"} +{"current_steps": 417, "total_steps": 5364, "loss": 0.18329191207885742, "lr": 1.5493482309124766e-05, "epoch": 0.2332867132867133, "percentage": 7.77, "elapsed_time": "0:49:27", "remaining_time": "9:46:39"} +{"current_steps": 418, "total_steps": 5364, "loss": 0.1778550148010254, "lr": 1.553072625698324e-05, "epoch": 0.23384615384615384, "percentage": 7.79, "elapsed_time": "0:49:34", "remaining_time": "9:46:30"} +{"current_steps": 419, "total_steps": 5364, "loss": 0.1682322919368744, "lr": 1.5567970204841713e-05, "epoch": 0.23440559440559441, "percentage": 7.81, "elapsed_time": "0:49:40", "remaining_time": "9:46:21"} +{"current_steps": 420, "total_steps": 5364, "loss": 0.20202751457691193, "lr": 1.560521415270019e-05, "epoch": 0.23496503496503496, "percentage": 7.83, "elapsed_time": "0:49:47", "remaining_time": "9:46:11"} +{"current_steps": 421, "total_steps": 5364, "loss": 0.1698126345872879, "lr": 1.564245810055866e-05, "epoch": 0.23552447552447553, "percentage": 7.85, "elapsed_time": "0:49:55", "remaining_time": "9:46:11"} +{"current_steps": 422, "total_steps": 5364, "loss": 0.22727744281291962, "lr": 1.5679702048417132e-05, "epoch": 0.23608391608391607, "percentage": 7.87, "elapsed_time": "0:50:02", "remaining_time": "9:46:03"} +{"current_steps": 423, "total_steps": 5364, "loss": 0.20175057649612427, "lr": 1.5716945996275604e-05, "epoch": 0.23664335664335664, "percentage": 7.89, "elapsed_time": "0:50:09", "remaining_time": "9:45:53"} +{"current_steps": 424, "total_steps": 5364, "loss": 0.1681423783302307, "lr": 1.575418994413408e-05, "epoch": 0.23720279720279722, "percentage": 7.9, "elapsed_time": "0:50:16", "remaining_time": "9:45:44"} +{"current_steps": 425, "total_steps": 5364, "loss": 0.36445632576942444, "lr": 1.5791433891992554e-05, "epoch": 0.23776223776223776, "percentage": 7.92, "elapsed_time": "0:50:23", "remaining_time": "9:45:35"} +{"current_steps": 426, "total_steps": 5364, "loss": 0.26764899492263794, "lr": 1.5828677839851026e-05, "epoch": 0.23832167832167833, "percentage": 7.94, "elapsed_time": "0:50:30", "remaining_time": "9:45:23"} +{"current_steps": 427, "total_steps": 5364, "loss": 0.1031881645321846, "lr": 1.5865921787709498e-05, "epoch": 0.23888111888111888, "percentage": 7.96, "elapsed_time": "0:50:37", "remaining_time": "9:45:15"} +{"current_steps": 428, "total_steps": 5364, "loss": 0.0615655779838562, "lr": 1.590316573556797e-05, "epoch": 0.23944055944055945, "percentage": 7.98, "elapsed_time": "0:50:44", "remaining_time": "9:45:09"} +{"current_steps": 429, "total_steps": 5364, "loss": 0.1426311582326889, "lr": 1.5940409683426445e-05, "epoch": 0.24, "percentage": 8.0, "elapsed_time": "0:50:51", "remaining_time": "9:45:03"} +{"current_steps": 430, "total_steps": 5364, "loss": 0.5700697898864746, "lr": 1.5977653631284917e-05, "epoch": 0.24055944055944056, "percentage": 8.02, "elapsed_time": "0:50:58", "remaining_time": "9:44:59"} +{"current_steps": 431, "total_steps": 5364, "loss": 0.19980737566947937, "lr": 1.6014897579143392e-05, "epoch": 0.2411188811188811, "percentage": 8.04, "elapsed_time": "0:51:06", "remaining_time": "9:44:54"} +{"current_steps": 432, "total_steps": 5364, "loss": 0.22451502084732056, "lr": 1.6052141527001864e-05, "epoch": 0.24167832167832168, "percentage": 8.05, "elapsed_time": "0:51:13", "remaining_time": "9:44:44"} +{"current_steps": 433, "total_steps": 5364, "loss": 0.33274221420288086, "lr": 1.6089385474860336e-05, "epoch": 0.24223776223776225, "percentage": 8.07, "elapsed_time": "0:51:19", "remaining_time": "9:44:34"} +{"current_steps": 434, "total_steps": 5364, "loss": 0.19224220514297485, "lr": 1.6126629422718808e-05, "epoch": 0.2427972027972028, "percentage": 8.09, "elapsed_time": "0:51:27", "remaining_time": "9:44:27"} +{"current_steps": 435, "total_steps": 5364, "loss": 0.14506609737873077, "lr": 1.6163873370577283e-05, "epoch": 0.24335664335664337, "percentage": 8.11, "elapsed_time": "0:51:33", "remaining_time": "9:44:17"} +{"current_steps": 436, "total_steps": 5364, "loss": 0.049820952117443085, "lr": 1.6201117318435755e-05, "epoch": 0.2439160839160839, "percentage": 8.13, "elapsed_time": "0:51:41", "remaining_time": "9:44:20"} +{"current_steps": 437, "total_steps": 5364, "loss": 0.19768625497817993, "lr": 1.623836126629423e-05, "epoch": 0.24447552447552448, "percentage": 8.15, "elapsed_time": "0:51:48", "remaining_time": "9:44:12"} +{"current_steps": 438, "total_steps": 5364, "loss": 0.23779600858688354, "lr": 1.62756052141527e-05, "epoch": 0.24503496503496502, "percentage": 8.17, "elapsed_time": "0:51:55", "remaining_time": "9:44:03"} +{"current_steps": 439, "total_steps": 5364, "loss": 0.4842148423194885, "lr": 1.6312849162011177e-05, "epoch": 0.2455944055944056, "percentage": 8.18, "elapsed_time": "0:52:02", "remaining_time": "9:43:55"} +{"current_steps": 440, "total_steps": 5364, "loss": 0.11237426102161407, "lr": 1.635009310986965e-05, "epoch": 0.24615384615384617, "percentage": 8.2, "elapsed_time": "0:52:09", "remaining_time": "9:43:47"} +{"current_steps": 441, "total_steps": 5364, "loss": 0.14319780468940735, "lr": 1.638733705772812e-05, "epoch": 0.2467132867132867, "percentage": 8.22, "elapsed_time": "0:52:17", "remaining_time": "9:43:46"} +{"current_steps": 442, "total_steps": 5364, "loss": 0.35914701223373413, "lr": 1.6424581005586592e-05, "epoch": 0.24727272727272728, "percentage": 8.24, "elapsed_time": "0:52:24", "remaining_time": "9:43:36"} +{"current_steps": 443, "total_steps": 5364, "loss": 0.12121884524822235, "lr": 1.6461824953445067e-05, "epoch": 0.24783216783216783, "percentage": 8.26, "elapsed_time": "0:52:31", "remaining_time": "9:43:28"} +{"current_steps": 444, "total_steps": 5364, "loss": 0.34015312790870667, "lr": 1.649906890130354e-05, "epoch": 0.2483916083916084, "percentage": 8.28, "elapsed_time": "0:52:38", "remaining_time": "9:43:20"} +{"current_steps": 445, "total_steps": 5364, "loss": 0.2301608771085739, "lr": 1.6536312849162014e-05, "epoch": 0.24895104895104894, "percentage": 8.3, "elapsed_time": "0:52:45", "remaining_time": "9:43:09"} +{"current_steps": 446, "total_steps": 5364, "loss": 0.3519575893878937, "lr": 1.6573556797020486e-05, "epoch": 0.2495104895104895, "percentage": 8.31, "elapsed_time": "0:52:52", "remaining_time": "9:43:05"} +{"current_steps": 447, "total_steps": 5364, "loss": 0.08999969810247421, "lr": 1.6610800744878958e-05, "epoch": 0.25006993006993006, "percentage": 8.33, "elapsed_time": "0:52:59", "remaining_time": "9:42:55"} +{"current_steps": 448, "total_steps": 5364, "loss": 0.28024739027023315, "lr": 1.664804469273743e-05, "epoch": 0.25062937062937063, "percentage": 8.35, "elapsed_time": "0:53:07", "remaining_time": "9:42:51"} +{"current_steps": 449, "total_steps": 5364, "loss": 0.43207600712776184, "lr": 1.6685288640595905e-05, "epoch": 0.2511888111888112, "percentage": 8.37, "elapsed_time": "0:53:13", "remaining_time": "9:42:41"} +{"current_steps": 450, "total_steps": 5364, "loss": 0.2294520139694214, "lr": 1.6722532588454377e-05, "epoch": 0.2517482517482518, "percentage": 8.39, "elapsed_time": "0:53:21", "remaining_time": "9:42:42"} +{"current_steps": 451, "total_steps": 5364, "loss": 0.1383638232946396, "lr": 1.6759776536312852e-05, "epoch": 0.2523076923076923, "percentage": 8.41, "elapsed_time": "0:53:28", "remaining_time": "9:42:34"} +{"current_steps": 452, "total_steps": 5364, "loss": 0.25563934445381165, "lr": 1.6797020484171324e-05, "epoch": 0.25286713286713286, "percentage": 8.43, "elapsed_time": "0:53:36", "remaining_time": "9:42:29"} +{"current_steps": 453, "total_steps": 5364, "loss": 0.21258139610290527, "lr": 1.6834264432029796e-05, "epoch": 0.25342657342657343, "percentage": 8.45, "elapsed_time": "0:53:42", "remaining_time": "9:42:19"} +{"current_steps": 454, "total_steps": 5364, "loss": 0.24583888053894043, "lr": 1.687150837988827e-05, "epoch": 0.253986013986014, "percentage": 8.46, "elapsed_time": "0:53:49", "remaining_time": "9:42:11"} +{"current_steps": 455, "total_steps": 5364, "loss": 0.49477094411849976, "lr": 1.6908752327746743e-05, "epoch": 0.2545454545454545, "percentage": 8.48, "elapsed_time": "0:53:56", "remaining_time": "9:42:03"} +{"current_steps": 456, "total_steps": 5364, "loss": 0.19425439834594727, "lr": 1.6945996275605218e-05, "epoch": 0.2551048951048951, "percentage": 8.5, "elapsed_time": "0:54:04", "remaining_time": "9:41:57"} +{"current_steps": 457, "total_steps": 5364, "loss": 0.1849464774131775, "lr": 1.698324022346369e-05, "epoch": 0.25566433566433566, "percentage": 8.52, "elapsed_time": "0:54:10", "remaining_time": "9:41:47"} +{"current_steps": 458, "total_steps": 5364, "loss": 0.2405935525894165, "lr": 1.702048417132216e-05, "epoch": 0.25622377622377623, "percentage": 8.54, "elapsed_time": "0:54:17", "remaining_time": "9:41:37"} +{"current_steps": 459, "total_steps": 5364, "loss": 0.21965207159519196, "lr": 1.7057728119180633e-05, "epoch": 0.2567832167832168, "percentage": 8.56, "elapsed_time": "0:54:24", "remaining_time": "9:41:27"} +{"current_steps": 460, "total_steps": 5364, "loss": 0.25860822200775146, "lr": 1.709497206703911e-05, "epoch": 0.2573426573426573, "percentage": 8.58, "elapsed_time": "0:54:31", "remaining_time": "9:41:19"} +{"current_steps": 461, "total_steps": 5364, "loss": 0.08967152237892151, "lr": 1.713221601489758e-05, "epoch": 0.2579020979020979, "percentage": 8.59, "elapsed_time": "0:54:39", "remaining_time": "9:41:20"} +{"current_steps": 462, "total_steps": 5364, "loss": 0.21769946813583374, "lr": 1.7169459962756056e-05, "epoch": 0.25846153846153846, "percentage": 8.61, "elapsed_time": "0:54:46", "remaining_time": "9:41:13"} +{"current_steps": 463, "total_steps": 5364, "loss": 0.1710788607597351, "lr": 1.7206703910614527e-05, "epoch": 0.25902097902097904, "percentage": 8.63, "elapsed_time": "0:54:53", "remaining_time": "9:41:04"} +{"current_steps": 464, "total_steps": 5364, "loss": 0.3484511077404022, "lr": 1.7243947858473e-05, "epoch": 0.2595804195804196, "percentage": 8.65, "elapsed_time": "0:55:00", "remaining_time": "9:40:55"} +{"current_steps": 465, "total_steps": 5364, "loss": 0.2597725987434387, "lr": 1.728119180633147e-05, "epoch": 0.2601398601398601, "percentage": 8.67, "elapsed_time": "0:55:08", "remaining_time": "9:40:53"} +{"current_steps": 466, "total_steps": 5364, "loss": 0.2551834285259247, "lr": 1.7318435754189946e-05, "epoch": 0.2606993006993007, "percentage": 8.69, "elapsed_time": "0:55:15", "remaining_time": "9:40:46"} +{"current_steps": 467, "total_steps": 5364, "loss": 0.0673862025141716, "lr": 1.7355679702048418e-05, "epoch": 0.26125874125874127, "percentage": 8.71, "elapsed_time": "0:55:23", "remaining_time": "9:40:47"} +{"current_steps": 468, "total_steps": 5364, "loss": 0.12457968294620514, "lr": 1.7392923649906893e-05, "epoch": 0.26181818181818184, "percentage": 8.72, "elapsed_time": "0:55:30", "remaining_time": "9:40:37"} +{"current_steps": 469, "total_steps": 5364, "loss": 0.20733694732189178, "lr": 1.7430167597765365e-05, "epoch": 0.26237762237762235, "percentage": 8.74, "elapsed_time": "0:55:37", "remaining_time": "9:40:36"} +{"current_steps": 470, "total_steps": 5364, "loss": 0.2562675178050995, "lr": 1.7467411545623837e-05, "epoch": 0.2629370629370629, "percentage": 8.76, "elapsed_time": "0:55:44", "remaining_time": "9:40:26"} +{"current_steps": 471, "total_steps": 5364, "loss": 0.39981701970100403, "lr": 1.750465549348231e-05, "epoch": 0.2634965034965035, "percentage": 8.78, "elapsed_time": "0:55:51", "remaining_time": "9:40:19"} +{"current_steps": 472, "total_steps": 5364, "loss": 0.1550302356481552, "lr": 1.7541899441340784e-05, "epoch": 0.26405594405594407, "percentage": 8.8, "elapsed_time": "0:55:58", "remaining_time": "9:40:12"} +{"current_steps": 473, "total_steps": 5364, "loss": 0.3247944712638855, "lr": 1.7579143389199256e-05, "epoch": 0.26461538461538464, "percentage": 8.82, "elapsed_time": "0:56:05", "remaining_time": "9:40:02"} +{"current_steps": 474, "total_steps": 5364, "loss": 0.14899270236492157, "lr": 1.761638733705773e-05, "epoch": 0.26517482517482516, "percentage": 8.84, "elapsed_time": "0:56:13", "remaining_time": "9:39:57"} +{"current_steps": 475, "total_steps": 5364, "loss": 0.10161678493022919, "lr": 1.7653631284916203e-05, "epoch": 0.26573426573426573, "percentage": 8.86, "elapsed_time": "0:56:19", "remaining_time": "9:39:47"} +{"current_steps": 476, "total_steps": 5364, "loss": 0.21037091314792633, "lr": 1.7690875232774675e-05, "epoch": 0.2662937062937063, "percentage": 8.87, "elapsed_time": "0:56:26", "remaining_time": "9:39:39"} +{"current_steps": 477, "total_steps": 5364, "loss": 0.12281668931245804, "lr": 1.7728119180633146e-05, "epoch": 0.26685314685314687, "percentage": 8.89, "elapsed_time": "0:56:34", "remaining_time": "9:39:39"} +{"current_steps": 478, "total_steps": 5364, "loss": 0.3243931531906128, "lr": 1.776536312849162e-05, "epoch": 0.2674125874125874, "percentage": 8.91, "elapsed_time": "0:56:41", "remaining_time": "9:39:30"} +{"current_steps": 479, "total_steps": 5364, "loss": 0.07544611394405365, "lr": 1.7802607076350093e-05, "epoch": 0.26797202797202796, "percentage": 8.93, "elapsed_time": "0:56:48", "remaining_time": "9:39:24"} +{"current_steps": 480, "total_steps": 5364, "loss": 0.13213147222995758, "lr": 1.783985102420857e-05, "epoch": 0.26853146853146853, "percentage": 8.95, "elapsed_time": "0:56:55", "remaining_time": "9:39:17"} +{"current_steps": 481, "total_steps": 5364, "loss": 0.14075420796871185, "lr": 1.787709497206704e-05, "epoch": 0.2690909090909091, "percentage": 8.97, "elapsed_time": "0:57:03", "remaining_time": "9:39:12"} +{"current_steps": 482, "total_steps": 5364, "loss": 0.3751899003982544, "lr": 1.7914338919925512e-05, "epoch": 0.2696503496503497, "percentage": 8.99, "elapsed_time": "0:57:10", "remaining_time": "9:39:02"} +{"current_steps": 483, "total_steps": 5364, "loss": 0.21884986758232117, "lr": 1.7951582867783987e-05, "epoch": 0.2702097902097902, "percentage": 9.0, "elapsed_time": "0:57:16", "remaining_time": "9:38:52"} +{"current_steps": 484, "total_steps": 5364, "loss": 0.4682178497314453, "lr": 1.798882681564246e-05, "epoch": 0.27076923076923076, "percentage": 9.02, "elapsed_time": "0:57:23", "remaining_time": "9:38:42"} +{"current_steps": 485, "total_steps": 5364, "loss": 0.2147725522518158, "lr": 1.8026070763500934e-05, "epoch": 0.27132867132867133, "percentage": 9.04, "elapsed_time": "0:57:30", "remaining_time": "9:38:35"} +{"current_steps": 486, "total_steps": 5364, "loss": 0.10786425322294235, "lr": 1.8063314711359406e-05, "epoch": 0.2718881118881119, "percentage": 9.06, "elapsed_time": "0:57:38", "remaining_time": "9:38:35"} +{"current_steps": 487, "total_steps": 5364, "loss": 0.23253406584262848, "lr": 1.8100558659217878e-05, "epoch": 0.2724475524475524, "percentage": 9.08, "elapsed_time": "0:57:45", "remaining_time": "9:38:28"} +{"current_steps": 488, "total_steps": 5364, "loss": 0.2840860188007355, "lr": 1.813780260707635e-05, "epoch": 0.273006993006993, "percentage": 9.1, "elapsed_time": "0:57:52", "remaining_time": "9:38:18"} +{"current_steps": 489, "total_steps": 5364, "loss": 0.20700715482234955, "lr": 1.8175046554934825e-05, "epoch": 0.27356643356643356, "percentage": 9.12, "elapsed_time": "0:57:59", "remaining_time": "9:38:08"} +{"current_steps": 490, "total_steps": 5364, "loss": 0.2722780406475067, "lr": 1.8212290502793297e-05, "epoch": 0.27412587412587414, "percentage": 9.13, "elapsed_time": "0:58:06", "remaining_time": "9:38:04"} +{"current_steps": 491, "total_steps": 5364, "loss": 0.26255929470062256, "lr": 1.8249534450651772e-05, "epoch": 0.2746853146853147, "percentage": 9.15, "elapsed_time": "0:58:14", "remaining_time": "9:37:57"} +{"current_steps": 492, "total_steps": 5364, "loss": 0.3739148676395416, "lr": 1.8286778398510244e-05, "epoch": 0.2752447552447552, "percentage": 9.17, "elapsed_time": "0:58:20", "remaining_time": "9:37:48"} +{"current_steps": 493, "total_steps": 5364, "loss": 0.3016756474971771, "lr": 1.8324022346368716e-05, "epoch": 0.2758041958041958, "percentage": 9.19, "elapsed_time": "0:58:28", "remaining_time": "9:37:41"} +{"current_steps": 494, "total_steps": 5364, "loss": 0.3016898036003113, "lr": 1.8361266294227188e-05, "epoch": 0.27636363636363637, "percentage": 9.21, "elapsed_time": "0:58:35", "remaining_time": "9:37:34"} +{"current_steps": 495, "total_steps": 5364, "loss": 0.036744993180036545, "lr": 1.8398510242085663e-05, "epoch": 0.27692307692307694, "percentage": 9.23, "elapsed_time": "0:58:42", "remaining_time": "9:37:31"} +{"current_steps": 496, "total_steps": 5364, "loss": 0.20957455039024353, "lr": 1.8435754189944135e-05, "epoch": 0.2774825174825175, "percentage": 9.25, "elapsed_time": "0:58:50", "remaining_time": "9:37:27"} +{"current_steps": 497, "total_steps": 5364, "loss": 0.4254513382911682, "lr": 1.847299813780261e-05, "epoch": 0.278041958041958, "percentage": 9.27, "elapsed_time": "0:58:57", "remaining_time": "9:37:20"} +{"current_steps": 498, "total_steps": 5364, "loss": 0.020748628303408623, "lr": 1.851024208566108e-05, "epoch": 0.2786013986013986, "percentage": 9.28, "elapsed_time": "0:59:04", "remaining_time": "9:37:13"} +{"current_steps": 499, "total_steps": 5364, "loss": 0.36569589376449585, "lr": 1.8547486033519553e-05, "epoch": 0.27916083916083917, "percentage": 9.3, "elapsed_time": "0:59:11", "remaining_time": "9:37:05"} +{"current_steps": 500, "total_steps": 5364, "loss": 0.3316744863986969, "lr": 1.8584729981378025e-05, "epoch": 0.27972027972027974, "percentage": 9.32, "elapsed_time": "0:59:18", "remaining_time": "9:37:00"} +{"current_steps": 501, "total_steps": 5364, "loss": 0.23616641759872437, "lr": 1.86219739292365e-05, "epoch": 0.28027972027972026, "percentage": 9.34, "elapsed_time": "0:59:55", "remaining_time": "9:41:35"} +{"current_steps": 502, "total_steps": 5364, "loss": 0.03917497396469116, "lr": 1.8659217877094972e-05, "epoch": 0.28083916083916083, "percentage": 9.36, "elapsed_time": "1:00:02", "remaining_time": "9:41:28"} +{"current_steps": 503, "total_steps": 5364, "loss": 0.22358393669128418, "lr": 1.8696461824953447e-05, "epoch": 0.2813986013986014, "percentage": 9.38, "elapsed_time": "1:00:08", "remaining_time": "9:41:17"} +{"current_steps": 504, "total_steps": 5364, "loss": 0.3953176736831665, "lr": 1.873370577281192e-05, "epoch": 0.28195804195804197, "percentage": 9.4, "elapsed_time": "1:00:15", "remaining_time": "9:41:07"} +{"current_steps": 505, "total_steps": 5364, "loss": 0.26574385166168213, "lr": 1.8770949720670394e-05, "epoch": 0.28251748251748254, "percentage": 9.41, "elapsed_time": "1:00:22", "remaining_time": "9:40:57"} +{"current_steps": 506, "total_steps": 5364, "loss": 0.24285589158535004, "lr": 1.8808193668528866e-05, "epoch": 0.28307692307692306, "percentage": 9.43, "elapsed_time": "1:00:30", "remaining_time": "9:40:51"} +{"current_steps": 507, "total_steps": 5364, "loss": 0.1264358013868332, "lr": 1.8845437616387338e-05, "epoch": 0.28363636363636363, "percentage": 9.45, "elapsed_time": "1:00:37", "remaining_time": "9:40:45"} +{"current_steps": 508, "total_steps": 5364, "loss": 0.20133408904075623, "lr": 1.888268156424581e-05, "epoch": 0.2841958041958042, "percentage": 9.47, "elapsed_time": "1:00:44", "remaining_time": "9:40:36"} +{"current_steps": 509, "total_steps": 5364, "loss": 0.26330941915512085, "lr": 1.8919925512104285e-05, "epoch": 0.2847552447552448, "percentage": 9.49, "elapsed_time": "1:00:51", "remaining_time": "9:40:32"} +{"current_steps": 510, "total_steps": 5364, "loss": 0.20988181233406067, "lr": 1.895716945996276e-05, "epoch": 0.2853146853146853, "percentage": 9.51, "elapsed_time": "1:00:58", "remaining_time": "9:40:23"} +{"current_steps": 511, "total_steps": 5364, "loss": 0.18932560086250305, "lr": 1.8994413407821232e-05, "epoch": 0.28587412587412586, "percentage": 9.53, "elapsed_time": "1:01:06", "remaining_time": "9:40:16"} +{"current_steps": 512, "total_steps": 5364, "loss": 0.2920285165309906, "lr": 1.9031657355679704e-05, "epoch": 0.28643356643356643, "percentage": 9.55, "elapsed_time": "1:01:12", "remaining_time": "9:40:06"} +{"current_steps": 513, "total_steps": 5364, "loss": 0.20274771749973297, "lr": 1.9068901303538176e-05, "epoch": 0.286993006993007, "percentage": 9.56, "elapsed_time": "1:01:19", "remaining_time": "9:39:57"} +{"current_steps": 514, "total_steps": 5364, "loss": 0.11821774393320084, "lr": 1.910614525139665e-05, "epoch": 0.2875524475524476, "percentage": 9.58, "elapsed_time": "1:01:26", "remaining_time": "9:39:49"} +{"current_steps": 515, "total_steps": 5364, "loss": 0.17717942595481873, "lr": 1.9143389199255123e-05, "epoch": 0.2881118881118881, "percentage": 9.6, "elapsed_time": "1:01:34", "remaining_time": "9:39:46"} +{"current_steps": 516, "total_steps": 5364, "loss": 0.12990251183509827, "lr": 1.9180633147113598e-05, "epoch": 0.28867132867132866, "percentage": 9.62, "elapsed_time": "1:01:41", "remaining_time": "9:39:38"} +{"current_steps": 517, "total_steps": 5364, "loss": 0.08428557962179184, "lr": 1.921787709497207e-05, "epoch": 0.28923076923076924, "percentage": 9.64, "elapsed_time": "1:01:49", "remaining_time": "9:39:37"} +{"current_steps": 518, "total_steps": 5364, "loss": 0.16661573946475983, "lr": 1.925512104283054e-05, "epoch": 0.2897902097902098, "percentage": 9.66, "elapsed_time": "1:01:57", "remaining_time": "9:39:37"} +{"current_steps": 519, "total_steps": 5364, "loss": 0.07727187126874924, "lr": 1.9292364990689013e-05, "epoch": 0.2903496503496503, "percentage": 9.68, "elapsed_time": "1:02:04", "remaining_time": "9:39:32"} +{"current_steps": 520, "total_steps": 5364, "loss": 0.08176373690366745, "lr": 1.932960893854749e-05, "epoch": 0.2909090909090909, "percentage": 9.69, "elapsed_time": "1:02:12", "remaining_time": "9:39:28"} +{"current_steps": 521, "total_steps": 5364, "loss": 0.29716911911964417, "lr": 1.936685288640596e-05, "epoch": 0.29146853146853147, "percentage": 9.71, "elapsed_time": "1:02:19", "remaining_time": "9:39:18"} +{"current_steps": 522, "total_steps": 5364, "loss": 0.18725723028182983, "lr": 1.9404096834264436e-05, "epoch": 0.29202797202797204, "percentage": 9.73, "elapsed_time": "1:02:26", "remaining_time": "9:39:09"} +{"current_steps": 523, "total_steps": 5364, "loss": 0.2251569777727127, "lr": 1.9441340782122907e-05, "epoch": 0.2925874125874126, "percentage": 9.75, "elapsed_time": "1:02:33", "remaining_time": "9:39:03"} +{"current_steps": 524, "total_steps": 5364, "loss": 0.17864307761192322, "lr": 1.947858472998138e-05, "epoch": 0.2931468531468531, "percentage": 9.77, "elapsed_time": "1:02:40", "remaining_time": "9:38:54"} +{"current_steps": 525, "total_steps": 5364, "loss": 0.21538743376731873, "lr": 1.951582867783985e-05, "epoch": 0.2937062937062937, "percentage": 9.79, "elapsed_time": "1:02:47", "remaining_time": "9:38:44"} +{"current_steps": 526, "total_steps": 5364, "loss": 0.30963966250419617, "lr": 1.9553072625698326e-05, "epoch": 0.29426573426573427, "percentage": 9.81, "elapsed_time": "1:02:54", "remaining_time": "9:38:37"} +{"current_steps": 527, "total_steps": 5364, "loss": 0.3740716278553009, "lr": 1.9590316573556798e-05, "epoch": 0.29482517482517484, "percentage": 9.82, "elapsed_time": "1:03:01", "remaining_time": "9:38:29"} +{"current_steps": 528, "total_steps": 5364, "loss": 0.29379579424858093, "lr": 1.9627560521415273e-05, "epoch": 0.2953846153846154, "percentage": 9.84, "elapsed_time": "1:03:09", "remaining_time": "9:38:24"} +{"current_steps": 529, "total_steps": 5364, "loss": 0.1428535431623459, "lr": 1.9664804469273745e-05, "epoch": 0.29594405594405593, "percentage": 9.86, "elapsed_time": "1:03:16", "remaining_time": "9:38:16"} +{"current_steps": 530, "total_steps": 5364, "loss": 0.09543010592460632, "lr": 1.9702048417132217e-05, "epoch": 0.2965034965034965, "percentage": 9.88, "elapsed_time": "1:03:23", "remaining_time": "9:38:11"} +{"current_steps": 531, "total_steps": 5364, "loss": 0.2161623239517212, "lr": 1.973929236499069e-05, "epoch": 0.29706293706293707, "percentage": 9.9, "elapsed_time": "1:03:30", "remaining_time": "9:38:02"} +{"current_steps": 532, "total_steps": 5364, "loss": 0.23159924149513245, "lr": 1.9776536312849164e-05, "epoch": 0.29762237762237764, "percentage": 9.92, "elapsed_time": "1:03:37", "remaining_time": "9:37:52"} +{"current_steps": 533, "total_steps": 5364, "loss": 0.17287680506706238, "lr": 1.9813780260707636e-05, "epoch": 0.29818181818181816, "percentage": 9.94, "elapsed_time": "1:03:44", "remaining_time": "9:37:43"} +{"current_steps": 534, "total_steps": 5364, "loss": 0.4449688792228699, "lr": 1.985102420856611e-05, "epoch": 0.29874125874125873, "percentage": 9.96, "elapsed_time": "1:03:51", "remaining_time": "9:37:33"} +{"current_steps": 535, "total_steps": 5364, "loss": 0.2960590422153473, "lr": 1.9888268156424583e-05, "epoch": 0.2993006993006993, "percentage": 9.97, "elapsed_time": "1:03:58", "remaining_time": "9:37:24"} +{"current_steps": 536, "total_steps": 5364, "loss": 0.3170793652534485, "lr": 1.9925512104283055e-05, "epoch": 0.2998601398601399, "percentage": 9.99, "elapsed_time": "1:04:05", "remaining_time": "9:37:20"} +{"current_steps": 537, "total_steps": 5364, "loss": 0.357006311416626, "lr": 1.9962756052141526e-05, "epoch": 0.30041958041958045, "percentage": 10.01, "elapsed_time": "1:04:12", "remaining_time": "9:37:10"} +{"current_steps": 538, "total_steps": 5364, "loss": 0.05219828709959984, "lr": 2e-05, "epoch": 0.30097902097902096, "percentage": 10.03, "elapsed_time": "1:04:20", "remaining_time": "9:37:05"} +{"current_steps": 539, "total_steps": 5364, "loss": 0.005741588771343231, "lr": 1.9999997882052766e-05, "epoch": 0.30153846153846153, "percentage": 10.05, "elapsed_time": "1:04:27", "remaining_time": "9:36:59"} +{"current_steps": 540, "total_steps": 5364, "loss": 0.2549275755882263, "lr": 1.999999152821195e-05, "epoch": 0.3020979020979021, "percentage": 10.07, "elapsed_time": "1:04:34", "remaining_time": "9:36:49"} +{"current_steps": 541, "total_steps": 5364, "loss": 0.28893324732780457, "lr": 1.9999980938480244e-05, "epoch": 0.3026573426573427, "percentage": 10.09, "elapsed_time": "1:04:41", "remaining_time": "9:36:43"} +{"current_steps": 542, "total_steps": 5364, "loss": 0.3215692639350891, "lr": 1.999996611286214e-05, "epoch": 0.3032167832167832, "percentage": 10.1, "elapsed_time": "1:04:48", "remaining_time": "9:36:33"} +{"current_steps": 543, "total_steps": 5364, "loss": 0.21357734501361847, "lr": 1.999994705136392e-05, "epoch": 0.30377622377622376, "percentage": 10.12, "elapsed_time": "1:04:55", "remaining_time": "9:36:29"} +{"current_steps": 544, "total_steps": 5364, "loss": 0.08483588695526123, "lr": 1.9999923753993643e-05, "epoch": 0.30433566433566434, "percentage": 10.14, "elapsed_time": "1:05:03", "remaining_time": "9:36:22"} +{"current_steps": 545, "total_steps": 5364, "loss": 0.20176199078559875, "lr": 1.9999896220761195e-05, "epoch": 0.3048951048951049, "percentage": 10.16, "elapsed_time": "1:05:10", "remaining_time": "9:36:15"} +{"current_steps": 546, "total_steps": 5364, "loss": 0.2352019101381302, "lr": 1.999986445167823e-05, "epoch": 0.3054545454545455, "percentage": 10.18, "elapsed_time": "1:05:17", "remaining_time": "9:36:10"} +{"current_steps": 547, "total_steps": 5364, "loss": 0.16775330901145935, "lr": 1.9999828446758205e-05, "epoch": 0.306013986013986, "percentage": 10.2, "elapsed_time": "1:05:25", "remaining_time": "9:36:08"} +{"current_steps": 548, "total_steps": 5364, "loss": 0.21414825320243835, "lr": 1.9999788206016375e-05, "epoch": 0.30657342657342657, "percentage": 10.22, "elapsed_time": "1:05:32", "remaining_time": "9:36:03"} +{"current_steps": 549, "total_steps": 5364, "loss": 0.18333928287029266, "lr": 1.9999743729469784e-05, "epoch": 0.30713286713286714, "percentage": 10.23, "elapsed_time": "1:05:40", "remaining_time": "9:35:56"} +{"current_steps": 550, "total_steps": 5364, "loss": 0.3981580436229706, "lr": 1.9999695017137266e-05, "epoch": 0.3076923076923077, "percentage": 10.25, "elapsed_time": "1:05:47", "remaining_time": "9:35:48"} +{"current_steps": 551, "total_steps": 5364, "loss": 0.23577269911766052, "lr": 1.9999642069039466e-05, "epoch": 0.3082517482517482, "percentage": 10.27, "elapsed_time": "1:05:54", "remaining_time": "9:35:39"} +{"current_steps": 552, "total_steps": 5364, "loss": 0.2632318437099457, "lr": 1.9999584885198802e-05, "epoch": 0.3088111888111888, "percentage": 10.29, "elapsed_time": "1:06:01", "remaining_time": "9:35:30"} +{"current_steps": 553, "total_steps": 5364, "loss": 0.2891254425048828, "lr": 1.99995234656395e-05, "epoch": 0.30937062937062937, "percentage": 10.31, "elapsed_time": "1:06:08", "remaining_time": "9:35:23"} +{"current_steps": 554, "total_steps": 5364, "loss": 0.08629722893238068, "lr": 1.999945781038758e-05, "epoch": 0.30993006993006994, "percentage": 10.33, "elapsed_time": "1:06:15", "remaining_time": "9:35:13"} +{"current_steps": 555, "total_steps": 5364, "loss": 0.07846999913454056, "lr": 1.999938791947085e-05, "epoch": 0.3104895104895105, "percentage": 10.35, "elapsed_time": "1:06:22", "remaining_time": "9:35:05"} +{"current_steps": 556, "total_steps": 5364, "loss": 0.11268086731433868, "lr": 1.9999313792918918e-05, "epoch": 0.31104895104895103, "percentage": 10.37, "elapsed_time": "1:06:29", "remaining_time": "9:35:03"} +{"current_steps": 557, "total_steps": 5364, "loss": 0.16018052399158478, "lr": 1.999923543076318e-05, "epoch": 0.3116083916083916, "percentage": 10.38, "elapsed_time": "1:06:36", "remaining_time": "9:34:53"} +{"current_steps": 558, "total_steps": 5364, "loss": 0.08534029126167297, "lr": 1.999915283303683e-05, "epoch": 0.31216783216783217, "percentage": 10.4, "elapsed_time": "1:06:43", "remaining_time": "9:34:45"} +{"current_steps": 559, "total_steps": 5364, "loss": 0.2227429896593094, "lr": 1.9999065999774855e-05, "epoch": 0.31272727272727274, "percentage": 10.42, "elapsed_time": "1:06:51", "remaining_time": "9:34:40"} +{"current_steps": 560, "total_steps": 5364, "loss": 0.07833819836378098, "lr": 1.9998974931014036e-05, "epoch": 0.3132867132867133, "percentage": 10.44, "elapsed_time": "1:06:58", "remaining_time": "9:34:30"} +{"current_steps": 561, "total_steps": 5364, "loss": 0.21360847353935242, "lr": 1.999887962679295e-05, "epoch": 0.31384615384615383, "percentage": 10.46, "elapsed_time": "1:07:05", "remaining_time": "9:34:22"} +{"current_steps": 562, "total_steps": 5364, "loss": 0.26231253147125244, "lr": 1.999878008715197e-05, "epoch": 0.3144055944055944, "percentage": 10.48, "elapsed_time": "1:07:12", "remaining_time": "9:34:12"} +{"current_steps": 563, "total_steps": 5364, "loss": 0.18054431676864624, "lr": 1.9998676312133254e-05, "epoch": 0.314965034965035, "percentage": 10.5, "elapsed_time": "1:07:19", "remaining_time": "9:34:03"} +{"current_steps": 564, "total_steps": 5364, "loss": 0.17667227983474731, "lr": 1.9998568301780764e-05, "epoch": 0.31552447552447555, "percentage": 10.51, "elapsed_time": "1:07:26", "remaining_time": "9:34:01"} +{"current_steps": 565, "total_steps": 5364, "loss": 0.09886716306209564, "lr": 1.999845605614025e-05, "epoch": 0.31608391608391606, "percentage": 10.53, "elapsed_time": "1:07:34", "remaining_time": "9:34:00"} +{"current_steps": 566, "total_steps": 5364, "loss": 0.0950867235660553, "lr": 1.999833957525926e-05, "epoch": 0.31664335664335663, "percentage": 10.55, "elapsed_time": "1:07:41", "remaining_time": "9:33:53"} +{"current_steps": 567, "total_steps": 5364, "loss": 0.10726246982812881, "lr": 1.9998218859187132e-05, "epoch": 0.3172027972027972, "percentage": 10.57, "elapsed_time": "1:07:49", "remaining_time": "9:33:51"} +{"current_steps": 568, "total_steps": 5364, "loss": 0.22885611653327942, "lr": 1.9998093907975e-05, "epoch": 0.3177622377622378, "percentage": 10.59, "elapsed_time": "1:07:56", "remaining_time": "9:33:41"} +{"current_steps": 569, "total_steps": 5364, "loss": 0.16578838229179382, "lr": 1.9997964721675797e-05, "epoch": 0.31832167832167835, "percentage": 10.61, "elapsed_time": "1:08:03", "remaining_time": "9:33:32"} +{"current_steps": 570, "total_steps": 5364, "loss": 0.25660938024520874, "lr": 1.999783130034424e-05, "epoch": 0.31888111888111886, "percentage": 10.63, "elapsed_time": "1:08:10", "remaining_time": "9:33:21"} +{"current_steps": 571, "total_steps": 5364, "loss": 0.20382758975028992, "lr": 1.9997693644036844e-05, "epoch": 0.31944055944055944, "percentage": 10.65, "elapsed_time": "1:08:17", "remaining_time": "9:33:13"} +{"current_steps": 572, "total_steps": 5364, "loss": 0.15027384459972382, "lr": 1.9997551752811922e-05, "epoch": 0.32, "percentage": 10.66, "elapsed_time": "1:08:24", "remaining_time": "9:33:03"} +{"current_steps": 573, "total_steps": 5364, "loss": 0.30086737871170044, "lr": 1.9997405626729577e-05, "epoch": 0.3205594405594406, "percentage": 10.68, "elapsed_time": "1:08:31", "remaining_time": "9:32:54"} +{"current_steps": 574, "total_steps": 5364, "loss": 0.22597002983093262, "lr": 1.9997255265851705e-05, "epoch": 0.3211188811188811, "percentage": 10.7, "elapsed_time": "1:08:38", "remaining_time": "9:32:45"} +{"current_steps": 575, "total_steps": 5364, "loss": 0.3176121115684509, "lr": 1.9997100670241994e-05, "epoch": 0.32167832167832167, "percentage": 10.72, "elapsed_time": "1:08:44", "remaining_time": "9:32:35"} +{"current_steps": 576, "total_steps": 5364, "loss": 0.20588722825050354, "lr": 1.999694183996594e-05, "epoch": 0.32223776223776224, "percentage": 10.74, "elapsed_time": "1:08:51", "remaining_time": "9:32:26"} +{"current_steps": 577, "total_steps": 5364, "loss": 0.23704226315021515, "lr": 1.999677877509081e-05, "epoch": 0.3227972027972028, "percentage": 10.76, "elapsed_time": "1:08:58", "remaining_time": "9:32:16"} +{"current_steps": 578, "total_steps": 5364, "loss": 0.3305433392524719, "lr": 1.9996611475685685e-05, "epoch": 0.3233566433566434, "percentage": 10.78, "elapsed_time": "1:09:05", "remaining_time": "9:32:08"} +{"current_steps": 579, "total_steps": 5364, "loss": 0.2543655037879944, "lr": 1.9996439941821427e-05, "epoch": 0.3239160839160839, "percentage": 10.79, "elapsed_time": "1:09:12", "remaining_time": "9:32:00"} +{"current_steps": 580, "total_steps": 5364, "loss": 0.1492166519165039, "lr": 1.9996264173570697e-05, "epoch": 0.32447552447552447, "percentage": 10.81, "elapsed_time": "1:09:20", "remaining_time": "9:31:57"} +{"current_steps": 581, "total_steps": 5364, "loss": 0.2557552754878998, "lr": 1.999608417100795e-05, "epoch": 0.32503496503496504, "percentage": 10.83, "elapsed_time": "1:09:27", "remaining_time": "9:31:49"} +{"current_steps": 582, "total_steps": 5364, "loss": 0.29245346784591675, "lr": 1.9995899934209422e-05, "epoch": 0.3255944055944056, "percentage": 10.85, "elapsed_time": "1:09:34", "remaining_time": "9:31:39"} +{"current_steps": 583, "total_steps": 5364, "loss": 0.2192707657814026, "lr": 1.9995711463253174e-05, "epoch": 0.3261538461538461, "percentage": 10.87, "elapsed_time": "1:09:42", "remaining_time": "9:31:35"} +{"current_steps": 584, "total_steps": 5364, "loss": 0.34399017691612244, "lr": 1.9995518758219027e-05, "epoch": 0.3267132867132867, "percentage": 10.89, "elapsed_time": "1:09:48", "remaining_time": "9:31:25"} +{"current_steps": 585, "total_steps": 5364, "loss": 0.23436295986175537, "lr": 1.9995321819188608e-05, "epoch": 0.32727272727272727, "percentage": 10.91, "elapsed_time": "1:09:55", "remaining_time": "9:31:15"} +{"current_steps": 586, "total_steps": 5364, "loss": 0.19554248452186584, "lr": 1.9995120646245345e-05, "epoch": 0.32783216783216784, "percentage": 10.92, "elapsed_time": "1:10:02", "remaining_time": "9:31:06"} +{"current_steps": 587, "total_steps": 5364, "loss": 0.10892355442047119, "lr": 1.9994915239474447e-05, "epoch": 0.3283916083916084, "percentage": 10.94, "elapsed_time": "1:10:09", "remaining_time": "9:30:57"} +{"current_steps": 588, "total_steps": 5364, "loss": 0.022315168753266335, "lr": 1.9994705598962927e-05, "epoch": 0.32895104895104893, "percentage": 10.96, "elapsed_time": "1:10:16", "remaining_time": "9:30:49"} +{"current_steps": 589, "total_steps": 5364, "loss": 0.212662011384964, "lr": 1.9994491724799583e-05, "epoch": 0.3295104895104895, "percentage": 10.98, "elapsed_time": "1:10:23", "remaining_time": "9:30:41"} +{"current_steps": 590, "total_steps": 5364, "loss": 0.23674353957176208, "lr": 1.9994273617075007e-05, "epoch": 0.3300699300699301, "percentage": 11.0, "elapsed_time": "1:10:30", "remaining_time": "9:30:34"} +{"current_steps": 591, "total_steps": 5364, "loss": 0.23225592076778412, "lr": 1.9994051275881597e-05, "epoch": 0.33062937062937064, "percentage": 11.02, "elapsed_time": "1:10:37", "remaining_time": "9:30:25"} +{"current_steps": 592, "total_steps": 5364, "loss": 0.21238228678703308, "lr": 1.9993824701313523e-05, "epoch": 0.3311888111888112, "percentage": 11.04, "elapsed_time": "1:10:44", "remaining_time": "9:30:15"} +{"current_steps": 593, "total_steps": 5364, "loss": 0.20713210105895996, "lr": 1.999359389346677e-05, "epoch": 0.33174825174825173, "percentage": 11.06, "elapsed_time": "1:10:51", "remaining_time": "9:30:09"} +{"current_steps": 594, "total_steps": 5364, "loss": 0.23903657495975494, "lr": 1.99933588524391e-05, "epoch": 0.3323076923076923, "percentage": 11.07, "elapsed_time": "1:10:58", "remaining_time": "9:29:59"} +{"current_steps": 595, "total_steps": 5364, "loss": 0.33758509159088135, "lr": 1.9993119578330073e-05, "epoch": 0.3328671328671329, "percentage": 11.09, "elapsed_time": "1:11:06", "remaining_time": "9:29:54"} +{"current_steps": 596, "total_steps": 5364, "loss": 0.2791229784488678, "lr": 1.999287607124105e-05, "epoch": 0.33342657342657345, "percentage": 11.11, "elapsed_time": "1:11:13", "remaining_time": "9:29:48"} +{"current_steps": 597, "total_steps": 5364, "loss": 0.22226542234420776, "lr": 1.9992628331275164e-05, "epoch": 0.33398601398601396, "percentage": 11.13, "elapsed_time": "1:11:20", "remaining_time": "9:29:39"} +{"current_steps": 598, "total_steps": 5364, "loss": 0.28454047441482544, "lr": 1.9992376358537367e-05, "epoch": 0.33454545454545453, "percentage": 11.15, "elapsed_time": "1:11:28", "remaining_time": "9:29:36"} +{"current_steps": 599, "total_steps": 5364, "loss": 0.10577382892370224, "lr": 1.999212015313439e-05, "epoch": 0.3351048951048951, "percentage": 11.17, "elapsed_time": "1:11:35", "remaining_time": "9:29:29"} +{"current_steps": 600, "total_steps": 5364, "loss": 0.27060481905937195, "lr": 1.9991859715174758e-05, "epoch": 0.3356643356643357, "percentage": 11.19, "elapsed_time": "1:11:42", "remaining_time": "9:29:22"} +{"current_steps": 601, "total_steps": 5364, "loss": 0.13188296556472778, "lr": 1.9991595044768788e-05, "epoch": 0.33622377622377625, "percentage": 11.2, "elapsed_time": "1:11:49", "remaining_time": "9:29:13"} +{"current_steps": 602, "total_steps": 5364, "loss": 0.18403789401054382, "lr": 1.9991326142028595e-05, "epoch": 0.33678321678321677, "percentage": 11.22, "elapsed_time": "1:11:57", "remaining_time": "9:29:09"} +{"current_steps": 603, "total_steps": 5364, "loss": 0.11701376736164093, "lr": 1.9991053007068077e-05, "epoch": 0.33734265734265734, "percentage": 11.24, "elapsed_time": "1:12:04", "remaining_time": "9:29:04"} +{"current_steps": 604, "total_steps": 5364, "loss": 0.16472938656806946, "lr": 1.999077564000294e-05, "epoch": 0.3379020979020979, "percentage": 11.26, "elapsed_time": "1:12:11", "remaining_time": "9:28:58"} +{"current_steps": 605, "total_steps": 5364, "loss": 0.24946929514408112, "lr": 1.9990494040950666e-05, "epoch": 0.3384615384615385, "percentage": 11.28, "elapsed_time": "1:12:18", "remaining_time": "9:28:48"} +{"current_steps": 606, "total_steps": 5364, "loss": 0.01795024238526821, "lr": 1.9990208210030543e-05, "epoch": 0.339020979020979, "percentage": 11.3, "elapsed_time": "1:12:25", "remaining_time": "9:28:41"} +{"current_steps": 607, "total_steps": 5364, "loss": 0.058757804334163666, "lr": 1.9989918147363644e-05, "epoch": 0.33958041958041957, "percentage": 11.32, "elapsed_time": "1:12:32", "remaining_time": "9:28:33"} +{"current_steps": 608, "total_steps": 5364, "loss": 0.40756332874298096, "lr": 1.9989623853072833e-05, "epoch": 0.34013986013986014, "percentage": 11.33, "elapsed_time": "1:12:39", "remaining_time": "9:28:24"} +{"current_steps": 609, "total_steps": 5364, "loss": 0.2548201382160187, "lr": 1.9989325327282777e-05, "epoch": 0.3406993006993007, "percentage": 11.35, "elapsed_time": "1:12:47", "remaining_time": "9:28:17"} +{"current_steps": 610, "total_steps": 5364, "loss": 0.2671986520290375, "lr": 1.998902257011992e-05, "epoch": 0.3412587412587413, "percentage": 11.37, "elapsed_time": "1:12:54", "remaining_time": "9:28:08"} +{"current_steps": 611, "total_steps": 5364, "loss": 0.5934276580810547, "lr": 1.9988715581712516e-05, "epoch": 0.3418181818181818, "percentage": 11.39, "elapsed_time": "1:13:01", "remaining_time": "9:28:03"} +{"current_steps": 612, "total_steps": 5364, "loss": 0.17480790615081787, "lr": 1.998840436219059e-05, "epoch": 0.34237762237762237, "percentage": 11.41, "elapsed_time": "1:13:08", "remaining_time": "9:27:59"} +{"current_steps": 613, "total_steps": 5364, "loss": 0.09544789046049118, "lr": 1.9988088911685986e-05, "epoch": 0.34293706293706294, "percentage": 11.43, "elapsed_time": "1:13:16", "remaining_time": "9:27:54"} +{"current_steps": 614, "total_steps": 5364, "loss": 0.32014167308807373, "lr": 1.998776923033231e-05, "epoch": 0.3434965034965035, "percentage": 11.45, "elapsed_time": "1:13:23", "remaining_time": "9:27:44"} +{"current_steps": 615, "total_steps": 5364, "loss": 0.16702398657798767, "lr": 1.9987445318264994e-05, "epoch": 0.34405594405594403, "percentage": 11.47, "elapsed_time": "1:13:30", "remaining_time": "9:27:37"} +{"current_steps": 616, "total_steps": 5364, "loss": 0.4500359892845154, "lr": 1.9987117175621223e-05, "epoch": 0.3446153846153846, "percentage": 11.48, "elapsed_time": "1:13:37", "remaining_time": "9:27:30"} +{"current_steps": 617, "total_steps": 5364, "loss": 0.18182170391082764, "lr": 1.9986784802540014e-05, "epoch": 0.3451748251748252, "percentage": 11.5, "elapsed_time": "1:13:44", "remaining_time": "9:27:22"} +{"current_steps": 618, "total_steps": 5364, "loss": 0.2514946758747101, "lr": 1.9986448199162142e-05, "epoch": 0.34573426573426574, "percentage": 11.52, "elapsed_time": "1:13:53", "remaining_time": "9:27:25"} +{"current_steps": 619, "total_steps": 5364, "loss": 0.21140369772911072, "lr": 1.99861073656302e-05, "epoch": 0.3462937062937063, "percentage": 11.54, "elapsed_time": "1:14:00", "remaining_time": "9:27:19"} +{"current_steps": 620, "total_steps": 5364, "loss": 0.22062942385673523, "lr": 1.9985762302088548e-05, "epoch": 0.34685314685314683, "percentage": 11.56, "elapsed_time": "1:14:07", "remaining_time": "9:27:11"} +{"current_steps": 621, "total_steps": 5364, "loss": 0.18166667222976685, "lr": 1.9985413008683364e-05, "epoch": 0.3474125874125874, "percentage": 11.58, "elapsed_time": "1:14:14", "remaining_time": "9:27:02"} +{"current_steps": 622, "total_steps": 5364, "loss": 0.21900129318237305, "lr": 1.99850594855626e-05, "epoch": 0.347972027972028, "percentage": 11.6, "elapsed_time": "1:14:20", "remaining_time": "9:26:49"} +{"current_steps": 623, "total_steps": 5364, "loss": 0.008729547262191772, "lr": 1.9984701732876008e-05, "epoch": 0.34853146853146855, "percentage": 11.61, "elapsed_time": "1:14:27", "remaining_time": "9:26:40"} +{"current_steps": 624, "total_steps": 5364, "loss": 0.19723348319530487, "lr": 1.9984339750775118e-05, "epoch": 0.3490909090909091, "percentage": 11.63, "elapsed_time": "1:14:35", "remaining_time": "9:26:33"} +{"current_steps": 625, "total_steps": 5364, "loss": 0.14206211268901825, "lr": 1.9983973539413274e-05, "epoch": 0.34965034965034963, "percentage": 11.65, "elapsed_time": "1:14:41", "remaining_time": "9:26:23"} +{"current_steps": 626, "total_steps": 5364, "loss": 0.11825959384441376, "lr": 1.9983603098945596e-05, "epoch": 0.3502097902097902, "percentage": 11.67, "elapsed_time": "1:14:48", "remaining_time": "9:26:14"} +{"current_steps": 627, "total_steps": 5364, "loss": 0.15087342262268066, "lr": 1.998322842952899e-05, "epoch": 0.3507692307692308, "percentage": 11.69, "elapsed_time": "1:14:56", "remaining_time": "9:26:09"} +{"current_steps": 628, "total_steps": 5364, "loss": 0.12859410047531128, "lr": 1.9982849531322174e-05, "epoch": 0.35132867132867135, "percentage": 11.71, "elapsed_time": "1:15:03", "remaining_time": "9:26:02"} +{"current_steps": 629, "total_steps": 5364, "loss": 0.2304256111383438, "lr": 1.998246640448564e-05, "epoch": 0.35188811188811187, "percentage": 11.73, "elapsed_time": "1:15:10", "remaining_time": "9:25:56"} +{"current_steps": 630, "total_steps": 5364, "loss": 0.1558963656425476, "lr": 1.9982079049181676e-05, "epoch": 0.35244755244755244, "percentage": 11.74, "elapsed_time": "1:15:17", "remaining_time": "9:25:46"} +{"current_steps": 631, "total_steps": 5364, "loss": 0.24547037482261658, "lr": 1.9981687465574363e-05, "epoch": 0.353006993006993, "percentage": 11.76, "elapsed_time": "1:15:24", "remaining_time": "9:25:37"} +{"current_steps": 632, "total_steps": 5364, "loss": 0.0876728892326355, "lr": 1.998129165382957e-05, "epoch": 0.3535664335664336, "percentage": 11.78, "elapsed_time": "1:15:31", "remaining_time": "9:25:27"} +{"current_steps": 633, "total_steps": 5364, "loss": 0.09364181756973267, "lr": 1.9980891614114956e-05, "epoch": 0.35412587412587415, "percentage": 11.8, "elapsed_time": "1:15:38", "remaining_time": "9:25:17"} +{"current_steps": 634, "total_steps": 5364, "loss": 0.15264952182769775, "lr": 1.998048734659998e-05, "epoch": 0.35468531468531467, "percentage": 11.82, "elapsed_time": "1:15:45", "remaining_time": "9:25:08"} +{"current_steps": 635, "total_steps": 5364, "loss": 0.024990465492010117, "lr": 1.9980078851455888e-05, "epoch": 0.35524475524475524, "percentage": 11.84, "elapsed_time": "1:15:52", "remaining_time": "9:25:01"} +{"current_steps": 636, "total_steps": 5364, "loss": 0.30937856435775757, "lr": 1.9979666128855705e-05, "epoch": 0.3558041958041958, "percentage": 11.86, "elapsed_time": "1:15:59", "remaining_time": "9:24:53"} +{"current_steps": 637, "total_steps": 5364, "loss": 0.25523439049720764, "lr": 1.9979249178974258e-05, "epoch": 0.3563636363636364, "percentage": 11.88, "elapsed_time": "1:16:06", "remaining_time": "9:24:43"} +{"current_steps": 638, "total_steps": 5364, "loss": 0.2529276907444, "lr": 1.9978828001988165e-05, "epoch": 0.3569230769230769, "percentage": 11.89, "elapsed_time": "1:16:13", "remaining_time": "9:24:35"} +{"current_steps": 639, "total_steps": 5364, "loss": 0.12054543197154999, "lr": 1.9978402598075833e-05, "epoch": 0.35748251748251747, "percentage": 11.91, "elapsed_time": "1:16:20", "remaining_time": "9:24:26"} +{"current_steps": 640, "total_steps": 5364, "loss": 0.22388938069343567, "lr": 1.997797296741746e-05, "epoch": 0.35804195804195804, "percentage": 11.93, "elapsed_time": "1:16:26", "remaining_time": "9:24:16"} +{"current_steps": 641, "total_steps": 5364, "loss": 0.13563752174377441, "lr": 1.9977539110195025e-05, "epoch": 0.3586013986013986, "percentage": 11.95, "elapsed_time": "1:16:34", "remaining_time": "9:24:10"} +{"current_steps": 642, "total_steps": 5364, "loss": 0.13494934141635895, "lr": 1.9977101026592313e-05, "epoch": 0.3591608391608392, "percentage": 11.97, "elapsed_time": "1:16:41", "remaining_time": "9:24:01"} +{"current_steps": 643, "total_steps": 5364, "loss": 0.05932402238249779, "lr": 1.997665871679489e-05, "epoch": 0.3597202797202797, "percentage": 11.99, "elapsed_time": "1:16:48", "remaining_time": "9:23:56"} +{"current_steps": 644, "total_steps": 5364, "loss": 0.18919488787651062, "lr": 1.997621218099011e-05, "epoch": 0.3602797202797203, "percentage": 12.01, "elapsed_time": "1:16:55", "remaining_time": "9:23:47"} +{"current_steps": 645, "total_steps": 5364, "loss": 0.2577393352985382, "lr": 1.9975761419367126e-05, "epoch": 0.36083916083916084, "percentage": 12.02, "elapsed_time": "1:17:02", "remaining_time": "9:23:38"} +{"current_steps": 646, "total_steps": 5364, "loss": 0.2787085771560669, "lr": 1.997530643211687e-05, "epoch": 0.3613986013986014, "percentage": 12.04, "elapsed_time": "1:17:09", "remaining_time": "9:23:30"} +{"current_steps": 647, "total_steps": 5364, "loss": 0.43514931201934814, "lr": 1.997484721943208e-05, "epoch": 0.36195804195804193, "percentage": 12.06, "elapsed_time": "1:17:16", "remaining_time": "9:23:20"} +{"current_steps": 648, "total_steps": 5364, "loss": 0.2881569266319275, "lr": 1.9974383781507264e-05, "epoch": 0.3625174825174825, "percentage": 12.08, "elapsed_time": "1:17:23", "remaining_time": "9:23:14"} +{"current_steps": 649, "total_steps": 5364, "loss": 0.11132467538118362, "lr": 1.997391611853873e-05, "epoch": 0.3630769230769231, "percentage": 12.1, "elapsed_time": "1:17:30", "remaining_time": "9:23:07"} +{"current_steps": 650, "total_steps": 5364, "loss": 0.29948365688323975, "lr": 1.997344423072458e-05, "epoch": 0.36363636363636365, "percentage": 12.12, "elapsed_time": "1:17:37", "remaining_time": "9:22:58"} +{"current_steps": 651, "total_steps": 5364, "loss": 0.44342485070228577, "lr": 1.9972968118264698e-05, "epoch": 0.3641958041958042, "percentage": 12.14, "elapsed_time": "1:17:44", "remaining_time": "9:22:48"} +{"current_steps": 652, "total_steps": 5364, "loss": 0.3593648076057434, "lr": 1.9972487781360762e-05, "epoch": 0.36475524475524473, "percentage": 12.16, "elapsed_time": "1:17:51", "remaining_time": "9:22:39"} +{"current_steps": 653, "total_steps": 5364, "loss": 0.4287704825401306, "lr": 1.9972003220216233e-05, "epoch": 0.3653146853146853, "percentage": 12.17, "elapsed_time": "1:17:58", "remaining_time": "9:22:30"} +{"current_steps": 654, "total_steps": 5364, "loss": 0.2837563455104828, "lr": 1.997151443503637e-05, "epoch": 0.3658741258741259, "percentage": 12.19, "elapsed_time": "1:18:05", "remaining_time": "9:22:23"} +{"current_steps": 655, "total_steps": 5364, "loss": 0.5405474305152893, "lr": 1.9971021426028218e-05, "epoch": 0.36643356643356645, "percentage": 12.21, "elapsed_time": "1:18:12", "remaining_time": "9:22:17"} +{"current_steps": 656, "total_steps": 5364, "loss": 0.09462281316518784, "lr": 1.9970524193400607e-05, "epoch": 0.366993006993007, "percentage": 12.23, "elapsed_time": "1:18:19", "remaining_time": "9:22:11"} +{"current_steps": 657, "total_steps": 5364, "loss": 0.32398372888565063, "lr": 1.9970022737364167e-05, "epoch": 0.36755244755244754, "percentage": 12.25, "elapsed_time": "1:18:26", "remaining_time": "9:22:01"} +{"current_steps": 658, "total_steps": 5364, "loss": 0.1537763923406601, "lr": 1.9969517058131294e-05, "epoch": 0.3681118881118881, "percentage": 12.27, "elapsed_time": "1:18:34", "remaining_time": "9:21:55"} +{"current_steps": 659, "total_steps": 5364, "loss": 0.22327053546905518, "lr": 1.9969007155916205e-05, "epoch": 0.3686713286713287, "percentage": 12.29, "elapsed_time": "1:18:41", "remaining_time": "9:21:47"} +{"current_steps": 660, "total_steps": 5364, "loss": 0.21873745322227478, "lr": 1.996849303093488e-05, "epoch": 0.36923076923076925, "percentage": 12.3, "elapsed_time": "1:18:48", "remaining_time": "9:21:39"} +{"current_steps": 661, "total_steps": 5364, "loss": 0.17564524710178375, "lr": 1.99679746834051e-05, "epoch": 0.36979020979020977, "percentage": 12.32, "elapsed_time": "1:18:55", "remaining_time": "9:21:30"} +{"current_steps": 662, "total_steps": 5364, "loss": 0.00918425340205431, "lr": 1.996745211354643e-05, "epoch": 0.37034965034965034, "percentage": 12.34, "elapsed_time": "1:19:02", "remaining_time": "9:21:21"} +{"current_steps": 663, "total_steps": 5364, "loss": 0.30388692021369934, "lr": 1.9966925321580228e-05, "epoch": 0.3709090909090909, "percentage": 12.36, "elapsed_time": "1:19:09", "remaining_time": "9:21:13"} +{"current_steps": 664, "total_steps": 5364, "loss": 0.20366382598876953, "lr": 1.9966394307729634e-05, "epoch": 0.3714685314685315, "percentage": 12.38, "elapsed_time": "1:19:16", "remaining_time": "9:21:04"} +{"current_steps": 665, "total_steps": 5364, "loss": 0.16192631423473358, "lr": 1.9965859072219583e-05, "epoch": 0.37202797202797205, "percentage": 12.4, "elapsed_time": "1:19:22", "remaining_time": "9:20:55"} +{"current_steps": 666, "total_steps": 5364, "loss": 0.38440436124801636, "lr": 1.996531961527679e-05, "epoch": 0.37258741258741257, "percentage": 12.42, "elapsed_time": "1:19:29", "remaining_time": "9:20:45"} +{"current_steps": 667, "total_steps": 5364, "loss": 0.17998406291007996, "lr": 1.996477593712977e-05, "epoch": 0.37314685314685314, "percentage": 12.43, "elapsed_time": "1:19:36", "remaining_time": "9:20:36"} +{"current_steps": 668, "total_steps": 5364, "loss": 0.05032619461417198, "lr": 1.9964228038008812e-05, "epoch": 0.3737062937062937, "percentage": 12.45, "elapsed_time": "1:19:44", "remaining_time": "9:20:32"} +{"current_steps": 669, "total_steps": 5364, "loss": 0.36880049109458923, "lr": 1.996367591814601e-05, "epoch": 0.3742657342657343, "percentage": 12.47, "elapsed_time": "1:19:51", "remaining_time": "9:20:24"} +{"current_steps": 670, "total_steps": 5364, "loss": 0.23061102628707886, "lr": 1.9963119577775225e-05, "epoch": 0.3748251748251748, "percentage": 12.49, "elapsed_time": "1:19:58", "remaining_time": "9:20:17"} +{"current_steps": 671, "total_steps": 5364, "loss": 0.19010961055755615, "lr": 1.9962559017132124e-05, "epoch": 0.37538461538461537, "percentage": 12.51, "elapsed_time": "1:20:05", "remaining_time": "9:20:08"} +{"current_steps": 672, "total_steps": 5364, "loss": 0.13563646376132965, "lr": 1.9961994236454153e-05, "epoch": 0.37594405594405594, "percentage": 12.53, "elapsed_time": "1:20:12", "remaining_time": "9:20:04"} +{"current_steps": 673, "total_steps": 5364, "loss": 0.23564599454402924, "lr": 1.9961425235980547e-05, "epoch": 0.3765034965034965, "percentage": 12.55, "elapsed_time": "1:20:20", "remaining_time": "9:20:00"} +{"current_steps": 674, "total_steps": 5364, "loss": 0.1854414939880371, "lr": 1.996085201595233e-05, "epoch": 0.3770629370629371, "percentage": 12.57, "elapsed_time": "1:20:27", "remaining_time": "9:19:51"} +{"current_steps": 675, "total_steps": 5364, "loss": 0.11217008531093597, "lr": 1.9960274576612307e-05, "epoch": 0.3776223776223776, "percentage": 12.58, "elapsed_time": "1:20:34", "remaining_time": "9:19:41"} +{"current_steps": 676, "total_steps": 5364, "loss": 0.353920042514801, "lr": 1.9959692918205083e-05, "epoch": 0.3781818181818182, "percentage": 12.6, "elapsed_time": "1:20:41", "remaining_time": "9:19:32"} +{"current_steps": 677, "total_steps": 5364, "loss": 0.35348647832870483, "lr": 1.9959107040977037e-05, "epoch": 0.37874125874125875, "percentage": 12.62, "elapsed_time": "1:20:48", "remaining_time": "9:19:25"} +{"current_steps": 678, "total_steps": 5364, "loss": 0.1647416651248932, "lr": 1.995851694517634e-05, "epoch": 0.3793006993006993, "percentage": 12.64, "elapsed_time": "1:20:55", "remaining_time": "9:19:16"} +{"current_steps": 679, "total_steps": 5364, "loss": 0.1425113081932068, "lr": 1.9957922631052954e-05, "epoch": 0.37986013986013983, "percentage": 12.66, "elapsed_time": "1:21:02", "remaining_time": "9:19:09"} +{"current_steps": 680, "total_steps": 5364, "loss": 0.3788657784461975, "lr": 1.995732409885862e-05, "epoch": 0.3804195804195804, "percentage": 12.68, "elapsed_time": "1:21:09", "remaining_time": "9:19:00"} +{"current_steps": 681, "total_steps": 5364, "loss": 0.14348703622817993, "lr": 1.9956721348846876e-05, "epoch": 0.380979020979021, "percentage": 12.7, "elapsed_time": "1:21:16", "remaining_time": "9:18:55"} +{"current_steps": 682, "total_steps": 5364, "loss": 0.18510159850120544, "lr": 1.995611438127303e-05, "epoch": 0.38153846153846155, "percentage": 12.71, "elapsed_time": "1:21:24", "remaining_time": "9:18:49"} +{"current_steps": 683, "total_steps": 5364, "loss": 0.03736811503767967, "lr": 1.9955503196394198e-05, "epoch": 0.3820979020979021, "percentage": 12.73, "elapsed_time": "1:21:31", "remaining_time": "9:18:46"} +{"current_steps": 684, "total_steps": 5364, "loss": 0.3070823550224304, "lr": 1.9954887794469266e-05, "epoch": 0.38265734265734264, "percentage": 12.75, "elapsed_time": "1:21:38", "remaining_time": "9:18:37"} +{"current_steps": 685, "total_steps": 5364, "loss": 0.11352913826704025, "lr": 1.9954268175758915e-05, "epoch": 0.3832167832167832, "percentage": 12.77, "elapsed_time": "1:21:45", "remaining_time": "9:18:29"} +{"current_steps": 686, "total_steps": 5364, "loss": 0.12527580559253693, "lr": 1.9953644340525607e-05, "epoch": 0.3837762237762238, "percentage": 12.79, "elapsed_time": "1:21:52", "remaining_time": "9:18:19"} +{"current_steps": 687, "total_steps": 5364, "loss": 0.4368479549884796, "lr": 1.995301628903359e-05, "epoch": 0.38433566433566435, "percentage": 12.81, "elapsed_time": "1:21:59", "remaining_time": "9:18:10"} +{"current_steps": 688, "total_steps": 5364, "loss": 0.32364341616630554, "lr": 1.99523840215489e-05, "epoch": 0.3848951048951049, "percentage": 12.83, "elapsed_time": "1:22:06", "remaining_time": "9:18:04"} +{"current_steps": 689, "total_steps": 5364, "loss": 0.25932592153549194, "lr": 1.9951747538339365e-05, "epoch": 0.38545454545454544, "percentage": 12.84, "elapsed_time": "1:22:14", "remaining_time": "9:17:58"} +{"current_steps": 690, "total_steps": 5364, "loss": 0.19032754004001617, "lr": 1.9951106839674585e-05, "epoch": 0.386013986013986, "percentage": 12.86, "elapsed_time": "1:22:20", "remaining_time": "9:17:48"} +{"current_steps": 691, "total_steps": 5364, "loss": 0.21685388684272766, "lr": 1.9950461925825956e-05, "epoch": 0.3865734265734266, "percentage": 12.88, "elapsed_time": "1:22:27", "remaining_time": "9:17:41"} +{"current_steps": 692, "total_steps": 5364, "loss": 0.19696876406669617, "lr": 1.9949812797066656e-05, "epoch": 0.38713286713286715, "percentage": 12.9, "elapsed_time": "1:22:34", "remaining_time": "9:17:32"} +{"current_steps": 693, "total_steps": 5364, "loss": 0.26852601766586304, "lr": 1.9949159453671653e-05, "epoch": 0.38769230769230767, "percentage": 12.92, "elapsed_time": "1:22:42", "remaining_time": "9:17:26"} +{"current_steps": 694, "total_steps": 5364, "loss": 0.08862755447626114, "lr": 1.9948501895917695e-05, "epoch": 0.38825174825174824, "percentage": 12.94, "elapsed_time": "1:22:49", "remaining_time": "9:17:17"} +{"current_steps": 695, "total_steps": 5364, "loss": 0.020724620670080185, "lr": 1.9947840124083312e-05, "epoch": 0.3888111888111888, "percentage": 12.96, "elapsed_time": "1:22:56", "remaining_time": "9:17:11"} +{"current_steps": 696, "total_steps": 5364, "loss": 0.03882191702723503, "lr": 1.994717413844883e-05, "epoch": 0.3893706293706294, "percentage": 12.98, "elapsed_time": "1:23:03", "remaining_time": "9:17:03"} +{"current_steps": 697, "total_steps": 5364, "loss": 0.32231131196022034, "lr": 1.9946503939296348e-05, "epoch": 0.38993006993006996, "percentage": 12.99, "elapsed_time": "1:23:10", "remaining_time": "9:16:53"} +{"current_steps": 698, "total_steps": 5364, "loss": 0.10666343569755554, "lr": 1.9945829526909758e-05, "epoch": 0.39048951048951047, "percentage": 13.01, "elapsed_time": "1:23:17", "remaining_time": "9:16:47"} +{"current_steps": 699, "total_steps": 5364, "loss": 0.1684054136276245, "lr": 1.994515090157473e-05, "epoch": 0.39104895104895104, "percentage": 13.03, "elapsed_time": "1:23:24", "remaining_time": "9:16:38"} +{"current_steps": 700, "total_steps": 5364, "loss": 0.13398505747318268, "lr": 1.994446806357873e-05, "epoch": 0.3916083916083916, "percentage": 13.05, "elapsed_time": "1:23:31", "remaining_time": "9:16:30"} +{"current_steps": 701, "total_steps": 5364, "loss": 0.3183130621910095, "lr": 1.9943781013210995e-05, "epoch": 0.3921678321678322, "percentage": 13.07, "elapsed_time": "1:23:38", "remaining_time": "9:16:20"} +{"current_steps": 702, "total_steps": 5364, "loss": 0.15352673828601837, "lr": 1.9943089750762556e-05, "epoch": 0.3927272727272727, "percentage": 13.09, "elapsed_time": "1:23:45", "remaining_time": "9:16:16"} +{"current_steps": 703, "total_steps": 5364, "loss": 0.245853990316391, "lr": 1.9942394276526222e-05, "epoch": 0.3932867132867133, "percentage": 13.11, "elapsed_time": "1:23:52", "remaining_time": "9:16:09"} +{"current_steps": 704, "total_steps": 5364, "loss": 0.0638028085231781, "lr": 1.9941694590796586e-05, "epoch": 0.39384615384615385, "percentage": 13.12, "elapsed_time": "1:24:00", "remaining_time": "9:16:05"} +{"current_steps": 705, "total_steps": 5364, "loss": 0.14579376578330994, "lr": 1.9940990693870038e-05, "epoch": 0.3944055944055944, "percentage": 13.14, "elapsed_time": "1:24:07", "remaining_time": "9:15:58"} +{"current_steps": 706, "total_steps": 5364, "loss": 0.17308682203292847, "lr": 1.994028258604473e-05, "epoch": 0.394965034965035, "percentage": 13.16, "elapsed_time": "1:24:14", "remaining_time": "9:15:50"} +{"current_steps": 707, "total_steps": 5364, "loss": 0.3714531660079956, "lr": 1.993957026762061e-05, "epoch": 0.3955244755244755, "percentage": 13.18, "elapsed_time": "1:24:21", "remaining_time": "9:15:43"} +{"current_steps": 708, "total_steps": 5364, "loss": 0.2819252908229828, "lr": 1.9938853738899415e-05, "epoch": 0.3960839160839161, "percentage": 13.2, "elapsed_time": "1:24:29", "remaining_time": "9:15:35"} +{"current_steps": 709, "total_steps": 5364, "loss": 0.29829293489456177, "lr": 1.9938133000184655e-05, "epoch": 0.39664335664335665, "percentage": 13.22, "elapsed_time": "1:24:35", "remaining_time": "9:15:26"} +{"current_steps": 710, "total_steps": 5364, "loss": 0.09194601327180862, "lr": 1.993740805178163e-05, "epoch": 0.3972027972027972, "percentage": 13.24, "elapsed_time": "1:24:44", "remaining_time": "9:15:26"} +{"current_steps": 711, "total_steps": 5364, "loss": 0.4935469627380371, "lr": 1.9936678893997416e-05, "epoch": 0.39776223776223774, "percentage": 13.26, "elapsed_time": "1:24:51", "remaining_time": "9:15:17"} +{"current_steps": 712, "total_steps": 5364, "loss": 0.3046504855155945, "lr": 1.9935945527140878e-05, "epoch": 0.3983216783216783, "percentage": 13.27, "elapsed_time": "1:24:58", "remaining_time": "9:15:11"} +{"current_steps": 713, "total_steps": 5364, "loss": 0.208089217543602, "lr": 1.9935207951522665e-05, "epoch": 0.3988811188811189, "percentage": 13.29, "elapsed_time": "1:25:05", "remaining_time": "9:15:02"} +{"current_steps": 714, "total_steps": 5364, "loss": 0.2735172510147095, "lr": 1.9934466167455206e-05, "epoch": 0.39944055944055945, "percentage": 13.31, "elapsed_time": "1:25:12", "remaining_time": "9:14:53"} +{"current_steps": 715, "total_steps": 5364, "loss": 0.07762259989976883, "lr": 1.993372017525271e-05, "epoch": 0.4, "percentage": 13.33, "elapsed_time": "1:25:18", "remaining_time": "9:14:44"} +{"current_steps": 716, "total_steps": 5364, "loss": 0.14039283990859985, "lr": 1.9932969975231173e-05, "epoch": 0.40055944055944054, "percentage": 13.35, "elapsed_time": "1:25:25", "remaining_time": "9:14:35"} +{"current_steps": 717, "total_steps": 5364, "loss": 0.046993188560009, "lr": 1.993221556770837e-05, "epoch": 0.4011188811188811, "percentage": 13.37, "elapsed_time": "1:25:32", "remaining_time": "9:14:27"} +{"current_steps": 718, "total_steps": 5364, "loss": 0.12122257053852081, "lr": 1.9931456953003864e-05, "epoch": 0.4016783216783217, "percentage": 13.39, "elapsed_time": "1:25:40", "remaining_time": "9:14:21"} +{"current_steps": 719, "total_steps": 5364, "loss": 0.022285472601652145, "lr": 1.9930694131438994e-05, "epoch": 0.40223776223776225, "percentage": 13.4, "elapsed_time": "1:25:46", "remaining_time": "9:14:10"} +{"current_steps": 720, "total_steps": 5364, "loss": 0.24089959263801575, "lr": 1.9929927103336882e-05, "epoch": 0.4027972027972028, "percentage": 13.42, "elapsed_time": "1:25:53", "remaining_time": "9:14:01"} +{"current_steps": 721, "total_steps": 5364, "loss": 0.06587815284729004, "lr": 1.9929155869022437e-05, "epoch": 0.40335664335664334, "percentage": 13.44, "elapsed_time": "1:26:01", "remaining_time": "9:13:55"} +{"current_steps": 722, "total_steps": 5364, "loss": 0.048804886639118195, "lr": 1.992838042882234e-05, "epoch": 0.4039160839160839, "percentage": 13.46, "elapsed_time": "1:26:07", "remaining_time": "9:13:46"} +{"current_steps": 723, "total_steps": 5364, "loss": 0.5738763809204102, "lr": 1.9927600783065065e-05, "epoch": 0.4044755244755245, "percentage": 13.48, "elapsed_time": "1:26:14", "remaining_time": "9:13:37"} +{"current_steps": 724, "total_steps": 5364, "loss": 0.29001617431640625, "lr": 1.9926816932080853e-05, "epoch": 0.40503496503496506, "percentage": 13.5, "elapsed_time": "1:26:21", "remaining_time": "9:13:28"} +{"current_steps": 725, "total_steps": 5364, "loss": 0.1561538130044937, "lr": 1.992602887620175e-05, "epoch": 0.40559440559440557, "percentage": 13.52, "elapsed_time": "1:26:28", "remaining_time": "9:13:20"} +{"current_steps": 726, "total_steps": 5364, "loss": 0.24576209485530853, "lr": 1.992523661576155e-05, "epoch": 0.40615384615384614, "percentage": 13.53, "elapsed_time": "1:26:35", "remaining_time": "9:13:12"} +{"current_steps": 727, "total_steps": 5364, "loss": 0.1025722399353981, "lr": 1.9924440151095863e-05, "epoch": 0.4067132867132867, "percentage": 13.55, "elapsed_time": "1:26:42", "remaining_time": "9:13:03"} +{"current_steps": 728, "total_steps": 5364, "loss": 0.21115200221538544, "lr": 1.9923639482542053e-05, "epoch": 0.4072727272727273, "percentage": 13.57, "elapsed_time": "1:26:50", "remaining_time": "9:13:00"} +{"current_steps": 729, "total_steps": 5364, "loss": 0.20292046666145325, "lr": 1.9922834610439276e-05, "epoch": 0.40783216783216786, "percentage": 13.59, "elapsed_time": "1:26:57", "remaining_time": "9:12:53"} +{"current_steps": 730, "total_steps": 5364, "loss": 0.2252373993396759, "lr": 1.9922025535128465e-05, "epoch": 0.4083916083916084, "percentage": 13.61, "elapsed_time": "1:27:04", "remaining_time": "9:12:44"} +{"current_steps": 731, "total_steps": 5364, "loss": 0.15693120658397675, "lr": 1.9921212256952345e-05, "epoch": 0.40895104895104895, "percentage": 13.63, "elapsed_time": "1:27:11", "remaining_time": "9:12:36"} +{"current_steps": 732, "total_steps": 5364, "loss": 0.1647944301366806, "lr": 1.9920394776255403e-05, "epoch": 0.4095104895104895, "percentage": 13.65, "elapsed_time": "1:27:18", "remaining_time": "9:12:31"} +{"current_steps": 733, "total_steps": 5364, "loss": 0.26393818855285645, "lr": 1.9919573093383923e-05, "epoch": 0.4100699300699301, "percentage": 13.67, "elapsed_time": "1:27:25", "remaining_time": "9:12:22"} +{"current_steps": 734, "total_steps": 5364, "loss": 0.32554638385772705, "lr": 1.9918747208685948e-05, "epoch": 0.4106293706293706, "percentage": 13.68, "elapsed_time": "1:27:33", "remaining_time": "9:12:17"} +{"current_steps": 735, "total_steps": 5364, "loss": 0.3616198003292084, "lr": 1.991791712251133e-05, "epoch": 0.4111888111888112, "percentage": 13.7, "elapsed_time": "1:27:40", "remaining_time": "9:12:08"} +{"current_steps": 736, "total_steps": 5364, "loss": 0.2198535054922104, "lr": 1.9917082835211678e-05, "epoch": 0.41174825174825175, "percentage": 13.72, "elapsed_time": "1:27:47", "remaining_time": "9:12:01"} +{"current_steps": 737, "total_steps": 5364, "loss": 0.31540998816490173, "lr": 1.991624434714038e-05, "epoch": 0.4123076923076923, "percentage": 13.74, "elapsed_time": "1:27:54", "remaining_time": "9:11:52"} +{"current_steps": 738, "total_steps": 5364, "loss": 0.2003219723701477, "lr": 1.9915401658652622e-05, "epoch": 0.4128671328671329, "percentage": 13.76, "elapsed_time": "1:28:01", "remaining_time": "9:11:46"} +{"current_steps": 739, "total_steps": 5364, "loss": 0.19488200545310974, "lr": 1.9914554770105352e-05, "epoch": 0.4134265734265734, "percentage": 13.78, "elapsed_time": "1:28:08", "remaining_time": "9:11:38"} +{"current_steps": 740, "total_steps": 5364, "loss": 0.09745018184185028, "lr": 1.9913703681857305e-05, "epoch": 0.413986013986014, "percentage": 13.8, "elapsed_time": "1:28:16", "remaining_time": "9:11:34"} +{"current_steps": 741, "total_steps": 5364, "loss": 0.059455666691064835, "lr": 1.991284839426899e-05, "epoch": 0.41454545454545455, "percentage": 13.81, "elapsed_time": "1:28:23", "remaining_time": "9:11:29"} +{"current_steps": 742, "total_steps": 5364, "loss": 0.341095507144928, "lr": 1.9911988907702704e-05, "epoch": 0.4151048951048951, "percentage": 13.83, "elapsed_time": "1:28:30", "remaining_time": "9:11:20"} +{"current_steps": 743, "total_steps": 5364, "loss": 0.31905606389045715, "lr": 1.9911125222522513e-05, "epoch": 0.41566433566433564, "percentage": 13.85, "elapsed_time": "1:28:37", "remaining_time": "9:11:14"} +{"current_steps": 744, "total_steps": 5364, "loss": 0.3375856280326843, "lr": 1.991025733909426e-05, "epoch": 0.4162237762237762, "percentage": 13.87, "elapsed_time": "1:28:45", "remaining_time": "9:11:07"} +{"current_steps": 745, "total_steps": 5364, "loss": 0.24612030386924744, "lr": 1.990938525778558e-05, "epoch": 0.4167832167832168, "percentage": 13.89, "elapsed_time": "1:28:52", "remaining_time": "9:10:58"} +{"current_steps": 746, "total_steps": 5364, "loss": 0.20706714689731598, "lr": 1.990850897896587e-05, "epoch": 0.41734265734265735, "percentage": 13.91, "elapsed_time": "1:28:58", "remaining_time": "9:10:49"} +{"current_steps": 747, "total_steps": 5364, "loss": 0.24323835968971252, "lr": 1.9907628503006317e-05, "epoch": 0.4179020979020979, "percentage": 13.93, "elapsed_time": "1:29:05", "remaining_time": "9:10:41"} +{"current_steps": 748, "total_steps": 5364, "loss": 0.26747220754623413, "lr": 1.990674383027988e-05, "epoch": 0.41846153846153844, "percentage": 13.94, "elapsed_time": "1:29:13", "remaining_time": "9:10:34"} +{"current_steps": 749, "total_steps": 5364, "loss": 0.24660389125347137, "lr": 1.99058549611613e-05, "epoch": 0.419020979020979, "percentage": 13.96, "elapsed_time": "1:29:20", "remaining_time": "9:10:28"} +{"current_steps": 750, "total_steps": 5364, "loss": 0.19418266415596008, "lr": 1.9904961896027085e-05, "epoch": 0.4195804195804196, "percentage": 13.98, "elapsed_time": "1:29:27", "remaining_time": "9:10:19"} +{"current_steps": 751, "total_steps": 5364, "loss": 0.33401572704315186, "lr": 1.9904064635255533e-05, "epoch": 0.42013986013986016, "percentage": 14.0, "elapsed_time": "1:29:34", "remaining_time": "9:10:13"} +{"current_steps": 752, "total_steps": 5364, "loss": 0.3395441770553589, "lr": 1.9903163179226715e-05, "epoch": 0.4206993006993007, "percentage": 14.02, "elapsed_time": "1:29:41", "remaining_time": "9:10:04"} +{"current_steps": 753, "total_steps": 5364, "loss": 0.37687867879867554, "lr": 1.990225752832248e-05, "epoch": 0.42125874125874124, "percentage": 14.04, "elapsed_time": "1:29:48", "remaining_time": "9:09:56"} +{"current_steps": 754, "total_steps": 5364, "loss": 0.1782049685716629, "lr": 1.9901347682926447e-05, "epoch": 0.4218181818181818, "percentage": 14.06, "elapsed_time": "1:29:55", "remaining_time": "9:09:47"} +{"current_steps": 755, "total_steps": 5364, "loss": 0.18066397309303284, "lr": 1.9900433643424018e-05, "epoch": 0.4223776223776224, "percentage": 14.08, "elapsed_time": "1:30:02", "remaining_time": "9:09:40"} +{"current_steps": 756, "total_steps": 5364, "loss": 0.33274298906326294, "lr": 1.9899515410202372e-05, "epoch": 0.42293706293706296, "percentage": 14.09, "elapsed_time": "1:30:09", "remaining_time": "9:09:33"} +{"current_steps": 757, "total_steps": 5364, "loss": 0.2913612723350525, "lr": 1.9898592983650462e-05, "epoch": 0.4234965034965035, "percentage": 14.11, "elapsed_time": "1:30:17", "remaining_time": "9:09:27"} +{"current_steps": 758, "total_steps": 5364, "loss": 0.2726729214191437, "lr": 1.9897666364159018e-05, "epoch": 0.42405594405594405, "percentage": 14.13, "elapsed_time": "1:30:23", "remaining_time": "9:09:18"} +{"current_steps": 759, "total_steps": 5364, "loss": 0.12482152134180069, "lr": 1.989673555212055e-05, "epoch": 0.4246153846153846, "percentage": 14.15, "elapsed_time": "1:30:31", "remaining_time": "9:09:15"} +{"current_steps": 760, "total_steps": 5364, "loss": 0.2711922526359558, "lr": 1.9895800547929337e-05, "epoch": 0.4251748251748252, "percentage": 14.17, "elapsed_time": "1:30:39", "remaining_time": "9:09:10"} +{"current_steps": 761, "total_steps": 5364, "loss": 0.36687546968460083, "lr": 1.9894861351981432e-05, "epoch": 0.42573426573426576, "percentage": 14.19, "elapsed_time": "1:30:46", "remaining_time": "9:09:01"} +{"current_steps": 762, "total_steps": 5364, "loss": 0.12306220829486847, "lr": 1.989391796467468e-05, "epoch": 0.4262937062937063, "percentage": 14.21, "elapsed_time": "1:30:53", "remaining_time": "9:08:54"} +{"current_steps": 763, "total_steps": 5364, "loss": 0.21053528785705566, "lr": 1.989297038640868e-05, "epoch": 0.42685314685314685, "percentage": 14.22, "elapsed_time": "1:31:00", "remaining_time": "9:08:47"} +{"current_steps": 764, "total_steps": 5364, "loss": 0.2965262532234192, "lr": 1.989201861758482e-05, "epoch": 0.4274125874125874, "percentage": 14.24, "elapsed_time": "1:31:07", "remaining_time": "9:08:38"} +{"current_steps": 765, "total_steps": 5364, "loss": 0.17599058151245117, "lr": 1.989106265860626e-05, "epoch": 0.427972027972028, "percentage": 14.26, "elapsed_time": "1:31:14", "remaining_time": "9:08:29"} +{"current_steps": 766, "total_steps": 5364, "loss": 0.2206752896308899, "lr": 1.989010250987793e-05, "epoch": 0.4285314685314685, "percentage": 14.28, "elapsed_time": "1:31:21", "remaining_time": "9:08:21"} +{"current_steps": 767, "total_steps": 5364, "loss": 0.31108683347702026, "lr": 1.9889138171806545e-05, "epoch": 0.4290909090909091, "percentage": 14.3, "elapsed_time": "1:31:28", "remaining_time": "9:08:15"} +{"current_steps": 768, "total_steps": 5364, "loss": 0.4014092683792114, "lr": 1.9888169644800584e-05, "epoch": 0.42965034965034965, "percentage": 14.32, "elapsed_time": "1:31:35", "remaining_time": "9:08:08"} +{"current_steps": 769, "total_steps": 5364, "loss": 0.40258675813674927, "lr": 1.9887196929270306e-05, "epoch": 0.4302097902097902, "percentage": 14.34, "elapsed_time": "1:31:43", "remaining_time": "9:08:03"} +{"current_steps": 770, "total_steps": 5364, "loss": 0.046959392726421356, "lr": 1.988622002562774e-05, "epoch": 0.4307692307692308, "percentage": 14.35, "elapsed_time": "1:31:50", "remaining_time": "9:07:57"} +{"current_steps": 771, "total_steps": 5364, "loss": 0.12582331895828247, "lr": 1.98852389342867e-05, "epoch": 0.4313286713286713, "percentage": 14.37, "elapsed_time": "1:31:58", "remaining_time": "9:07:52"} +{"current_steps": 772, "total_steps": 5364, "loss": 0.18430662155151367, "lr": 1.988425365566276e-05, "epoch": 0.4318881118881119, "percentage": 14.39, "elapsed_time": "1:32:05", "remaining_time": "9:07:48"} +{"current_steps": 773, "total_steps": 5364, "loss": 0.25388067960739136, "lr": 1.9883264190173274e-05, "epoch": 0.43244755244755245, "percentage": 14.41, "elapsed_time": "1:32:12", "remaining_time": "9:07:41"} +{"current_steps": 774, "total_steps": 5364, "loss": 0.17330634593963623, "lr": 1.9882270538237368e-05, "epoch": 0.433006993006993, "percentage": 14.43, "elapsed_time": "1:32:20", "remaining_time": "9:07:34"} +{"current_steps": 775, "total_steps": 5364, "loss": 0.11140122264623642, "lr": 1.9881272700275947e-05, "epoch": 0.43356643356643354, "percentage": 14.45, "elapsed_time": "1:32:27", "remaining_time": "9:07:30"} +{"current_steps": 776, "total_steps": 5364, "loss": 0.17463473975658417, "lr": 1.9880270676711682e-05, "epoch": 0.4341258741258741, "percentage": 14.47, "elapsed_time": "1:32:34", "remaining_time": "9:07:22"} +{"current_steps": 777, "total_steps": 5364, "loss": 0.1907040774822235, "lr": 1.9879264467969016e-05, "epoch": 0.4346853146853147, "percentage": 14.49, "elapsed_time": "1:32:42", "remaining_time": "9:07:17"} +{"current_steps": 778, "total_steps": 5364, "loss": 0.35833752155303955, "lr": 1.9878254074474176e-05, "epoch": 0.43524475524475525, "percentage": 14.5, "elapsed_time": "1:32:49", "remaining_time": "9:07:11"} +{"current_steps": 779, "total_steps": 5364, "loss": 0.14887432754039764, "lr": 1.9877239496655146e-05, "epoch": 0.4358041958041958, "percentage": 14.52, "elapsed_time": "1:32:57", "remaining_time": "9:07:06"} +{"current_steps": 780, "total_steps": 5364, "loss": 0.3146654963493347, "lr": 1.98762207349417e-05, "epoch": 0.43636363636363634, "percentage": 14.54, "elapsed_time": "1:33:04", "remaining_time": "9:06:59"} +{"current_steps": 781, "total_steps": 5364, "loss": 0.06458631157875061, "lr": 1.9875197789765366e-05, "epoch": 0.4369230769230769, "percentage": 14.56, "elapsed_time": "1:33:11", "remaining_time": "9:06:51"} +{"current_steps": 782, "total_steps": 5364, "loss": 0.23434588313102722, "lr": 1.9874170661559456e-05, "epoch": 0.4374825174825175, "percentage": 14.58, "elapsed_time": "1:33:18", "remaining_time": "9:06:42"} +{"current_steps": 783, "total_steps": 5364, "loss": 0.23638221621513367, "lr": 1.987313935075905e-05, "epoch": 0.43804195804195806, "percentage": 14.6, "elapsed_time": "1:33:25", "remaining_time": "9:06:37"} +{"current_steps": 784, "total_steps": 5364, "loss": 0.2840083837509155, "lr": 1.9872103857801003e-05, "epoch": 0.43860139860139863, "percentage": 14.62, "elapsed_time": "1:33:32", "remaining_time": "9:06:28"} +{"current_steps": 785, "total_steps": 5364, "loss": 0.1863371878862381, "lr": 1.987106418312393e-05, "epoch": 0.43916083916083914, "percentage": 14.63, "elapsed_time": "1:33:39", "remaining_time": "9:06:20"} +{"current_steps": 786, "total_steps": 5364, "loss": 0.144439235329628, "lr": 1.987002032716824e-05, "epoch": 0.4397202797202797, "percentage": 14.65, "elapsed_time": "1:33:47", "remaining_time": "9:06:16"} +{"current_steps": 787, "total_steps": 5364, "loss": 0.20231716334819794, "lr": 1.9868972290376094e-05, "epoch": 0.4402797202797203, "percentage": 14.67, "elapsed_time": "1:33:54", "remaining_time": "9:06:07"} +{"current_steps": 788, "total_steps": 5364, "loss": 0.32276082038879395, "lr": 1.9867920073191424e-05, "epoch": 0.44083916083916086, "percentage": 14.69, "elapsed_time": "1:34:01", "remaining_time": "9:05:58"} +{"current_steps": 789, "total_steps": 5364, "loss": 0.133274644613266, "lr": 1.9866863676059943e-05, "epoch": 0.4413986013986014, "percentage": 14.71, "elapsed_time": "1:34:08", "remaining_time": "9:05:50"} +{"current_steps": 790, "total_steps": 5364, "loss": 0.28316041827201843, "lr": 1.986580309942913e-05, "epoch": 0.44195804195804195, "percentage": 14.73, "elapsed_time": "1:34:15", "remaining_time": "9:05:41"} +{"current_steps": 791, "total_steps": 5364, "loss": 0.35311996936798096, "lr": 1.986473834374823e-05, "epoch": 0.4425174825174825, "percentage": 14.75, "elapsed_time": "1:34:21", "remaining_time": "9:05:32"} +{"current_steps": 792, "total_steps": 5364, "loss": 0.3782188296318054, "lr": 1.9863669409468262e-05, "epoch": 0.4430769230769231, "percentage": 14.77, "elapsed_time": "1:34:28", "remaining_time": "9:05:24"} +{"current_steps": 793, "total_steps": 5364, "loss": 0.31462928652763367, "lr": 1.986259629704202e-05, "epoch": 0.44363636363636366, "percentage": 14.78, "elapsed_time": "1:34:36", "remaining_time": "9:05:18"} +{"current_steps": 794, "total_steps": 5364, "loss": 0.2784261703491211, "lr": 1.9861519006924066e-05, "epoch": 0.4441958041958042, "percentage": 14.8, "elapsed_time": "1:34:43", "remaining_time": "9:05:10"} +{"current_steps": 795, "total_steps": 5364, "loss": 0.2810211479663849, "lr": 1.986043753957072e-05, "epoch": 0.44475524475524475, "percentage": 14.82, "elapsed_time": "1:34:50", "remaining_time": "9:05:03"} +{"current_steps": 796, "total_steps": 5364, "loss": 0.49437206983566284, "lr": 1.9859351895440085e-05, "epoch": 0.4453146853146853, "percentage": 14.84, "elapsed_time": "1:34:57", "remaining_time": "9:04:55"} +{"current_steps": 797, "total_steps": 5364, "loss": 0.37909960746765137, "lr": 1.9858262074992034e-05, "epoch": 0.4458741258741259, "percentage": 14.86, "elapsed_time": "1:35:04", "remaining_time": "9:04:47"} +{"current_steps": 798, "total_steps": 5364, "loss": 0.18024909496307373, "lr": 1.9857168078688188e-05, "epoch": 0.4464335664335664, "percentage": 14.88, "elapsed_time": "1:35:11", "remaining_time": "9:04:39"} +{"current_steps": 799, "total_steps": 5364, "loss": 0.32987910509109497, "lr": 1.9856069906991963e-05, "epoch": 0.446993006993007, "percentage": 14.9, "elapsed_time": "1:35:18", "remaining_time": "9:04:33"} +{"current_steps": 800, "total_steps": 5364, "loss": 0.22431644797325134, "lr": 1.9854967560368533e-05, "epoch": 0.44755244755244755, "percentage": 14.91, "elapsed_time": "1:35:25", "remaining_time": "9:04:26"} +{"current_steps": 801, "total_steps": 5364, "loss": 0.27222999930381775, "lr": 1.985386103928484e-05, "epoch": 0.4481118881118881, "percentage": 14.93, "elapsed_time": "1:35:33", "remaining_time": "9:04:19"} +{"current_steps": 802, "total_steps": 5364, "loss": 0.161219984292984, "lr": 1.985275034420959e-05, "epoch": 0.4486713286713287, "percentage": 14.95, "elapsed_time": "1:35:39", "remaining_time": "9:04:10"} +{"current_steps": 803, "total_steps": 5364, "loss": 0.22425110638141632, "lr": 1.9851635475613266e-05, "epoch": 0.4492307692307692, "percentage": 14.97, "elapsed_time": "1:35:46", "remaining_time": "9:04:02"} +{"current_steps": 804, "total_steps": 5364, "loss": 0.0884673148393631, "lr": 1.9850516433968114e-05, "epoch": 0.4497902097902098, "percentage": 14.99, "elapsed_time": "1:35:53", "remaining_time": "9:03:53"} +{"current_steps": 805, "total_steps": 5364, "loss": 0.025272320955991745, "lr": 1.9849393219748147e-05, "epoch": 0.45034965034965035, "percentage": 15.01, "elapsed_time": "1:36:00", "remaining_time": "9:03:45"} +{"current_steps": 806, "total_steps": 5364, "loss": 0.23998430371284485, "lr": 1.984826583342915e-05, "epoch": 0.4509090909090909, "percentage": 15.03, "elapsed_time": "1:36:07", "remaining_time": "9:03:36"} +{"current_steps": 807, "total_steps": 5364, "loss": 0.3951619863510132, "lr": 1.9847134275488665e-05, "epoch": 0.45146853146853144, "percentage": 15.04, "elapsed_time": "1:36:14", "remaining_time": "9:03:27"} +{"current_steps": 808, "total_steps": 5364, "loss": 0.3042892515659332, "lr": 1.9845998546406017e-05, "epoch": 0.452027972027972, "percentage": 15.06, "elapsed_time": "1:36:21", "remaining_time": "9:03:21"} +{"current_steps": 809, "total_steps": 5364, "loss": 0.3718934655189514, "lr": 1.9844858646662285e-05, "epoch": 0.4525874125874126, "percentage": 15.08, "elapsed_time": "1:36:28", "remaining_time": "9:03:12"} +{"current_steps": 810, "total_steps": 5364, "loss": 0.25356805324554443, "lr": 1.984371457674031e-05, "epoch": 0.45314685314685316, "percentage": 15.1, "elapsed_time": "1:36:35", "remaining_time": "9:03:04"} +{"current_steps": 811, "total_steps": 5364, "loss": 0.12882554531097412, "lr": 1.9842566337124723e-05, "epoch": 0.45370629370629373, "percentage": 15.12, "elapsed_time": "1:36:42", "remaining_time": "9:02:55"} +{"current_steps": 812, "total_steps": 5364, "loss": 0.11873381584882736, "lr": 1.9841413928301898e-05, "epoch": 0.45426573426573424, "percentage": 15.14, "elapsed_time": "1:36:49", "remaining_time": "9:02:48"} +{"current_steps": 813, "total_steps": 5364, "loss": 0.45210009813308716, "lr": 1.9840257350759985e-05, "epoch": 0.4548251748251748, "percentage": 15.16, "elapsed_time": "1:36:57", "remaining_time": "9:02:44"} +{"current_steps": 814, "total_steps": 5364, "loss": 0.37297913432121277, "lr": 1.9839096604988892e-05, "epoch": 0.4553846153846154, "percentage": 15.18, "elapsed_time": "1:37:04", "remaining_time": "9:02:38"} +{"current_steps": 815, "total_steps": 5364, "loss": 0.2697679400444031, "lr": 1.9837931691480307e-05, "epoch": 0.45594405594405596, "percentage": 15.19, "elapsed_time": "1:37:11", "remaining_time": "9:02:29"} +{"current_steps": 816, "total_steps": 5364, "loss": 0.06176023930311203, "lr": 1.9836762610727674e-05, "epoch": 0.45650349650349653, "percentage": 15.21, "elapsed_time": "1:37:19", "remaining_time": "9:02:27"} +{"current_steps": 817, "total_steps": 5364, "loss": 0.19568443298339844, "lr": 1.98355893632262e-05, "epoch": 0.45706293706293705, "percentage": 15.23, "elapsed_time": "1:37:26", "remaining_time": "9:02:18"} +{"current_steps": 818, "total_steps": 5364, "loss": 0.11257348209619522, "lr": 1.983441194947286e-05, "epoch": 0.4576223776223776, "percentage": 15.25, "elapsed_time": "1:37:33", "remaining_time": "9:02:13"} +{"current_steps": 819, "total_steps": 5364, "loss": 0.07968845963478088, "lr": 1.9833230369966397e-05, "epoch": 0.4581818181818182, "percentage": 15.27, "elapsed_time": "1:37:41", "remaining_time": "9:02:05"} +{"current_steps": 820, "total_steps": 5364, "loss": 0.16019681096076965, "lr": 1.9832044625207313e-05, "epoch": 0.45874125874125876, "percentage": 15.29, "elapsed_time": "1:37:47", "remaining_time": "9:01:56"} +{"current_steps": 821, "total_steps": 5364, "loss": 0.20029059052467346, "lr": 1.9830854715697878e-05, "epoch": 0.4593006993006993, "percentage": 15.31, "elapsed_time": "1:37:54", "remaining_time": "9:01:48"} +{"current_steps": 822, "total_steps": 5364, "loss": 0.22813692688941956, "lr": 1.9829660641942125e-05, "epoch": 0.45986013986013985, "percentage": 15.32, "elapsed_time": "1:38:02", "remaining_time": "9:01:43"} +{"current_steps": 823, "total_steps": 5364, "loss": 0.10862059891223907, "lr": 1.9828462404445853e-05, "epoch": 0.4604195804195804, "percentage": 15.34, "elapsed_time": "1:38:09", "remaining_time": "9:01:36"} +{"current_steps": 824, "total_steps": 5364, "loss": 0.32558006048202515, "lr": 1.982726000371662e-05, "epoch": 0.460979020979021, "percentage": 15.36, "elapsed_time": "1:38:16", "remaining_time": "9:01:28"} +{"current_steps": 825, "total_steps": 5364, "loss": 0.21844086050987244, "lr": 1.982605344026375e-05, "epoch": 0.46153846153846156, "percentage": 15.38, "elapsed_time": "1:38:23", "remaining_time": "9:01:22"} +{"current_steps": 826, "total_steps": 5364, "loss": 0.18136177957057953, "lr": 1.9824842714598335e-05, "epoch": 0.4620979020979021, "percentage": 15.4, "elapsed_time": "1:38:31", "remaining_time": "9:01:15"} +{"current_steps": 827, "total_steps": 5364, "loss": 0.07972593605518341, "lr": 1.982362782723322e-05, "epoch": 0.46265734265734265, "percentage": 15.42, "elapsed_time": "1:38:38", "remaining_time": "9:01:08"} +{"current_steps": 828, "total_steps": 5364, "loss": 0.04100705683231354, "lr": 1.9822408778683016e-05, "epoch": 0.4632167832167832, "percentage": 15.44, "elapsed_time": "1:38:45", "remaining_time": "9:01:00"} +{"current_steps": 829, "total_steps": 5364, "loss": 0.12353743612766266, "lr": 1.982118556946411e-05, "epoch": 0.4637762237762238, "percentage": 15.45, "elapsed_time": "1:38:52", "remaining_time": "9:00:54"} +{"current_steps": 830, "total_steps": 5364, "loss": 0.2682718336582184, "lr": 1.9819958200094636e-05, "epoch": 0.4643356643356643, "percentage": 15.47, "elapsed_time": "1:38:59", "remaining_time": "9:00:45"} +{"current_steps": 831, "total_steps": 5364, "loss": 0.2304733246564865, "lr": 1.9818726671094492e-05, "epoch": 0.4648951048951049, "percentage": 15.49, "elapsed_time": "1:39:06", "remaining_time": "9:00:37"} +{"current_steps": 832, "total_steps": 5364, "loss": 0.1034315824508667, "lr": 1.9817490982985337e-05, "epoch": 0.46545454545454545, "percentage": 15.51, "elapsed_time": "1:39:13", "remaining_time": "9:00:29"} +{"current_steps": 833, "total_steps": 5364, "loss": 0.32902252674102783, "lr": 1.9816251136290608e-05, "epoch": 0.466013986013986, "percentage": 15.53, "elapsed_time": "1:39:20", "remaining_time": "9:00:20"} +{"current_steps": 834, "total_steps": 5364, "loss": 0.1359844207763672, "lr": 1.9815007131535477e-05, "epoch": 0.4665734265734266, "percentage": 15.55, "elapsed_time": "1:39:27", "remaining_time": "9:00:15"} +{"current_steps": 835, "total_steps": 5364, "loss": 0.15279315412044525, "lr": 1.98137589692469e-05, "epoch": 0.4671328671328671, "percentage": 15.57, "elapsed_time": "1:39:35", "remaining_time": "9:00:08"} +{"current_steps": 836, "total_steps": 5364, "loss": 0.212006613612175, "lr": 1.981250664995358e-05, "epoch": 0.4676923076923077, "percentage": 15.59, "elapsed_time": "1:39:42", "remaining_time": "9:00:00"} +{"current_steps": 837, "total_steps": 5364, "loss": 0.07839135825634003, "lr": 1.9811250174185998e-05, "epoch": 0.46825174825174826, "percentage": 15.6, "elapsed_time": "1:39:49", "remaining_time": "8:59:52"} +{"current_steps": 838, "total_steps": 5364, "loss": 0.04277295991778374, "lr": 1.9809989542476368e-05, "epoch": 0.46881118881118883, "percentage": 15.62, "elapsed_time": "1:39:57", "remaining_time": "8:59:50"} +{"current_steps": 839, "total_steps": 5364, "loss": 0.09534847736358643, "lr": 1.9808724755358688e-05, "epoch": 0.46937062937062934, "percentage": 15.64, "elapsed_time": "1:40:03", "remaining_time": "8:59:41"} +{"current_steps": 840, "total_steps": 5364, "loss": 0.3481186628341675, "lr": 1.980745581336871e-05, "epoch": 0.4699300699300699, "percentage": 15.66, "elapsed_time": "1:40:11", "remaining_time": "8:59:33"} +{"current_steps": 841, "total_steps": 5364, "loss": 0.2360180914402008, "lr": 1.980618271704394e-05, "epoch": 0.4704895104895105, "percentage": 15.68, "elapsed_time": "1:40:17", "remaining_time": "8:59:24"} +{"current_steps": 842, "total_steps": 5364, "loss": 0.4861489534378052, "lr": 1.980490546692365e-05, "epoch": 0.47104895104895106, "percentage": 15.7, "elapsed_time": "1:40:24", "remaining_time": "8:59:16"} +{"current_steps": 843, "total_steps": 5364, "loss": 0.237271249294281, "lr": 1.980362406354887e-05, "epoch": 0.47160839160839163, "percentage": 15.72, "elapsed_time": "1:40:31", "remaining_time": "8:59:08"} +{"current_steps": 844, "total_steps": 5364, "loss": 0.27513962984085083, "lr": 1.9802338507462395e-05, "epoch": 0.47216783216783215, "percentage": 15.73, "elapsed_time": "1:40:39", "remaining_time": "8:59:02"} +{"current_steps": 845, "total_steps": 5364, "loss": 0.07421436160802841, "lr": 1.9801048799208763e-05, "epoch": 0.4727272727272727, "percentage": 15.75, "elapsed_time": "1:40:46", "remaining_time": "8:58:54"} +{"current_steps": 846, "total_steps": 5364, "loss": 0.1646387130022049, "lr": 1.979975493933428e-05, "epoch": 0.4732867132867133, "percentage": 15.77, "elapsed_time": "1:40:53", "remaining_time": "8:58:48"} +{"current_steps": 847, "total_steps": 5364, "loss": 0.10142280906438828, "lr": 1.9798456928387023e-05, "epoch": 0.47384615384615386, "percentage": 15.79, "elapsed_time": "1:41:00", "remaining_time": "8:58:41"} +{"current_steps": 848, "total_steps": 5364, "loss": 0.2740744352340698, "lr": 1.9797154766916805e-05, "epoch": 0.47440559440559443, "percentage": 15.81, "elapsed_time": "1:41:07", "remaining_time": "8:58:32"} +{"current_steps": 849, "total_steps": 5364, "loss": 0.07237780839204788, "lr": 1.9795848455475216e-05, "epoch": 0.47496503496503495, "percentage": 15.83, "elapsed_time": "1:41:14", "remaining_time": "8:58:26"} +{"current_steps": 850, "total_steps": 5364, "loss": 0.05178440734744072, "lr": 1.9794537994615587e-05, "epoch": 0.4755244755244755, "percentage": 15.85, "elapsed_time": "1:41:22", "remaining_time": "8:58:20"} +{"current_steps": 851, "total_steps": 5364, "loss": 0.191196471452713, "lr": 1.9793223384893023e-05, "epoch": 0.4760839160839161, "percentage": 15.87, "elapsed_time": "1:41:29", "remaining_time": "8:58:15"} +{"current_steps": 852, "total_steps": 5364, "loss": 0.34000539779663086, "lr": 1.9791904626864378e-05, "epoch": 0.47664335664335666, "percentage": 15.88, "elapsed_time": "1:41:36", "remaining_time": "8:58:08"} +{"current_steps": 853, "total_steps": 5364, "loss": 0.0698869526386261, "lr": 1.9790581721088257e-05, "epoch": 0.4772027972027972, "percentage": 15.9, "elapsed_time": "1:41:44", "remaining_time": "8:58:04"} +{"current_steps": 854, "total_steps": 5364, "loss": 0.141836479306221, "lr": 1.9789254668125035e-05, "epoch": 0.47776223776223775, "percentage": 15.92, "elapsed_time": "1:41:52", "remaining_time": "8:57:57"} +{"current_steps": 855, "total_steps": 5364, "loss": 0.1997612714767456, "lr": 1.9787923468536838e-05, "epoch": 0.4783216783216783, "percentage": 15.94, "elapsed_time": "1:41:59", "remaining_time": "8:57:51"} +{"current_steps": 856, "total_steps": 5364, "loss": 0.2586309015750885, "lr": 1.9786588122887545e-05, "epoch": 0.4788811188811189, "percentage": 15.96, "elapsed_time": "1:42:06", "remaining_time": "8:57:42"} +{"current_steps": 857, "total_steps": 5364, "loss": 0.37913331389427185, "lr": 1.9785248631742796e-05, "epoch": 0.47944055944055947, "percentage": 15.98, "elapsed_time": "1:42:13", "remaining_time": "8:57:33"} +{"current_steps": 858, "total_steps": 5364, "loss": 0.249455526471138, "lr": 1.9783904995669987e-05, "epoch": 0.48, "percentage": 16.0, "elapsed_time": "1:42:20", "remaining_time": "8:57:27"} +{"current_steps": 859, "total_steps": 5364, "loss": 0.20057329535484314, "lr": 1.9782557215238263e-05, "epoch": 0.48055944055944055, "percentage": 16.01, "elapsed_time": "1:42:27", "remaining_time": "8:57:20"} +{"current_steps": 860, "total_steps": 5364, "loss": 0.14626863598823547, "lr": 1.978120529101853e-05, "epoch": 0.4811188811188811, "percentage": 16.03, "elapsed_time": "1:42:34", "remaining_time": "8:57:11"} +{"current_steps": 861, "total_steps": 5364, "loss": 0.2326146960258484, "lr": 1.977984922358346e-05, "epoch": 0.4816783216783217, "percentage": 16.05, "elapsed_time": "1:42:41", "remaining_time": "8:57:04"} +{"current_steps": 862, "total_steps": 5364, "loss": 0.2547037899494171, "lr": 1.9778489013507455e-05, "epoch": 0.4822377622377622, "percentage": 16.07, "elapsed_time": "1:42:49", "remaining_time": "8:56:59"} +{"current_steps": 863, "total_steps": 5364, "loss": 0.25627368688583374, "lr": 1.9777124661366693e-05, "epoch": 0.4827972027972028, "percentage": 16.09, "elapsed_time": "1:42:56", "remaining_time": "8:56:54"} +{"current_steps": 864, "total_steps": 5364, "loss": 0.42010629177093506, "lr": 1.9775756167739098e-05, "epoch": 0.48335664335664336, "percentage": 16.11, "elapsed_time": "1:43:04", "remaining_time": "8:56:51"} +{"current_steps": 865, "total_steps": 5364, "loss": 0.0896216481924057, "lr": 1.9774383533204345e-05, "epoch": 0.48391608391608393, "percentage": 16.13, "elapsed_time": "1:43:12", "remaining_time": "8:56:46"} +{"current_steps": 866, "total_steps": 5364, "loss": 0.13138100504875183, "lr": 1.9773006758343873e-05, "epoch": 0.4844755244755245, "percentage": 16.14, "elapsed_time": "1:43:19", "remaining_time": "8:56:38"} +{"current_steps": 867, "total_steps": 5364, "loss": 0.12429220229387283, "lr": 1.977162584374087e-05, "epoch": 0.485034965034965, "percentage": 16.16, "elapsed_time": "1:43:26", "remaining_time": "8:56:30"} +{"current_steps": 868, "total_steps": 5364, "loss": 0.16589125990867615, "lr": 1.977024078998027e-05, "epoch": 0.4855944055944056, "percentage": 16.18, "elapsed_time": "1:43:32", "remaining_time": "8:56:21"} +{"current_steps": 869, "total_steps": 5364, "loss": 0.2439456731081009, "lr": 1.9768851597648774e-05, "epoch": 0.48615384615384616, "percentage": 16.2, "elapsed_time": "1:43:39", "remaining_time": "8:56:12"} +{"current_steps": 870, "total_steps": 5364, "loss": 0.3458921015262604, "lr": 1.9767458267334828e-05, "epoch": 0.48671328671328673, "percentage": 16.22, "elapsed_time": "1:43:46", "remaining_time": "8:56:04"} +{"current_steps": 871, "total_steps": 5364, "loss": 0.25526830554008484, "lr": 1.9766060799628627e-05, "epoch": 0.48727272727272725, "percentage": 16.24, "elapsed_time": "1:43:54", "remaining_time": "8:55:59"} +{"current_steps": 872, "total_steps": 5364, "loss": 0.2836688756942749, "lr": 1.976465919512213e-05, "epoch": 0.4878321678321678, "percentage": 16.26, "elapsed_time": "1:44:01", "remaining_time": "8:55:51"} +{"current_steps": 873, "total_steps": 5364, "loss": 0.049724068492650986, "lr": 1.9763253454409038e-05, "epoch": 0.4883916083916084, "percentage": 16.28, "elapsed_time": "1:44:08", "remaining_time": "8:55:43"} +{"current_steps": 874, "total_steps": 5364, "loss": 0.19707101583480835, "lr": 1.9761843578084813e-05, "epoch": 0.48895104895104896, "percentage": 16.29, "elapsed_time": "1:44:15", "remaining_time": "8:55:34"} +{"current_steps": 875, "total_steps": 5364, "loss": 0.29132080078125, "lr": 1.9760429566746656e-05, "epoch": 0.48951048951048953, "percentage": 16.31, "elapsed_time": "1:44:22", "remaining_time": "8:55:27"} +{"current_steps": 876, "total_steps": 5364, "loss": 0.19825905561447144, "lr": 1.9759011420993527e-05, "epoch": 0.49006993006993005, "percentage": 16.33, "elapsed_time": "1:44:29", "remaining_time": "8:55:20"} +{"current_steps": 877, "total_steps": 5364, "loss": 0.2920805513858795, "lr": 1.975758914142615e-05, "epoch": 0.4906293706293706, "percentage": 16.35, "elapsed_time": "1:44:36", "remaining_time": "8:55:11"} +{"current_steps": 878, "total_steps": 5364, "loss": 0.41261839866638184, "lr": 1.9756162728646972e-05, "epoch": 0.4911888111888112, "percentage": 16.37, "elapsed_time": "1:44:43", "remaining_time": "8:55:03"} +{"current_steps": 879, "total_steps": 5364, "loss": 0.14197523891925812, "lr": 1.975473218326022e-05, "epoch": 0.49174825174825176, "percentage": 16.39, "elapsed_time": "1:44:50", "remaining_time": "8:54:56"} +{"current_steps": 880, "total_steps": 5364, "loss": 0.16157715022563934, "lr": 1.975329750587185e-05, "epoch": 0.49230769230769234, "percentage": 16.41, "elapsed_time": "1:44:57", "remaining_time": "8:54:48"} +{"current_steps": 881, "total_steps": 5364, "loss": 0.3838087320327759, "lr": 1.9751858697089577e-05, "epoch": 0.49286713286713285, "percentage": 16.42, "elapsed_time": "1:45:04", "remaining_time": "8:54:41"} +{"current_steps": 882, "total_steps": 5364, "loss": 0.19504405558109283, "lr": 1.9750415757522863e-05, "epoch": 0.4934265734265734, "percentage": 16.44, "elapsed_time": "1:45:11", "remaining_time": "8:54:32"} +{"current_steps": 883, "total_steps": 5364, "loss": 0.2133583277463913, "lr": 1.974896868778293e-05, "epoch": 0.493986013986014, "percentage": 16.46, "elapsed_time": "1:45:18", "remaining_time": "8:54:25"} +{"current_steps": 884, "total_steps": 5364, "loss": 0.2611194849014282, "lr": 1.9747517488482732e-05, "epoch": 0.49454545454545457, "percentage": 16.48, "elapsed_time": "1:45:26", "remaining_time": "8:54:19"} +{"current_steps": 885, "total_steps": 5364, "loss": 0.04857818782329559, "lr": 1.974606216023699e-05, "epoch": 0.4951048951048951, "percentage": 16.5, "elapsed_time": "1:45:33", "remaining_time": "8:54:13"} +{"current_steps": 886, "total_steps": 5364, "loss": 0.3626279830932617, "lr": 1.9744602703662158e-05, "epoch": 0.49566433566433565, "percentage": 16.52, "elapsed_time": "1:45:40", "remaining_time": "8:54:05"} +{"current_steps": 887, "total_steps": 5364, "loss": 0.4035492539405823, "lr": 1.9743139119376457e-05, "epoch": 0.4962237762237762, "percentage": 16.54, "elapsed_time": "1:45:47", "remaining_time": "8:53:57"} +{"current_steps": 888, "total_steps": 5364, "loss": 0.3036898076534271, "lr": 1.9741671407999835e-05, "epoch": 0.4967832167832168, "percentage": 16.55, "elapsed_time": "1:45:54", "remaining_time": "8:53:50"} +{"current_steps": 889, "total_steps": 5364, "loss": 0.12967118620872498, "lr": 1.9740199570154003e-05, "epoch": 0.49734265734265737, "percentage": 16.57, "elapsed_time": "1:46:01", "remaining_time": "8:53:43"} +{"current_steps": 890, "total_steps": 5364, "loss": 0.23828613758087158, "lr": 1.9738723606462415e-05, "epoch": 0.4979020979020979, "percentage": 16.59, "elapsed_time": "1:46:08", "remaining_time": "8:53:34"} +{"current_steps": 891, "total_steps": 5364, "loss": 0.2823325991630554, "lr": 1.973724351755028e-05, "epoch": 0.49846153846153846, "percentage": 16.61, "elapsed_time": "1:46:15", "remaining_time": "8:53:26"} +{"current_steps": 892, "total_steps": 5364, "loss": 0.26219749450683594, "lr": 1.9735759304044536e-05, "epoch": 0.499020979020979, "percentage": 16.63, "elapsed_time": "1:46:22", "remaining_time": "8:53:18"} +{"current_steps": 893, "total_steps": 5364, "loss": 0.25803399085998535, "lr": 1.973427096657389e-05, "epoch": 0.4995804195804196, "percentage": 16.65, "elapsed_time": "1:46:29", "remaining_time": "8:53:10"} +{"current_steps": 894, "total_steps": 5364, "loss": 0.2967890799045563, "lr": 1.9732778505768785e-05, "epoch": 0.5001398601398601, "percentage": 16.67, "elapsed_time": "1:46:36", "remaining_time": "8:53:02"} +{"current_steps": 895, "total_steps": 5364, "loss": 0.2366620898246765, "lr": 1.9731281922261406e-05, "epoch": 0.5006993006993007, "percentage": 16.69, "elapsed_time": "1:46:43", "remaining_time": "8:52:54"} +{"current_steps": 896, "total_steps": 5364, "loss": 0.21914657950401306, "lr": 1.9729781216685695e-05, "epoch": 0.5012587412587413, "percentage": 16.7, "elapsed_time": "1:46:50", "remaining_time": "8:52:47"} +{"current_steps": 897, "total_steps": 5364, "loss": 0.11656968295574188, "lr": 1.9728276389677338e-05, "epoch": 0.5018181818181818, "percentage": 16.72, "elapsed_time": "1:46:57", "remaining_time": "8:52:41"} +{"current_steps": 898, "total_steps": 5364, "loss": 0.16053169965744019, "lr": 1.9726767441873753e-05, "epoch": 0.5023776223776224, "percentage": 16.74, "elapsed_time": "1:47:04", "remaining_time": "8:52:33"} +{"current_steps": 899, "total_steps": 5364, "loss": 0.06676869839429855, "lr": 1.9725254373914126e-05, "epoch": 0.502937062937063, "percentage": 16.76, "elapsed_time": "1:47:12", "remaining_time": "8:52:25"} +{"current_steps": 900, "total_steps": 5364, "loss": 0.3024398982524872, "lr": 1.9723737186439372e-05, "epoch": 0.5034965034965035, "percentage": 16.78, "elapsed_time": "1:47:18", "remaining_time": "8:52:16"} +{"current_steps": 901, "total_steps": 5364, "loss": 0.2132788747549057, "lr": 1.9722215880092152e-05, "epoch": 0.504055944055944, "percentage": 16.8, "elapsed_time": "1:47:25", "remaining_time": "8:52:08"} +{"current_steps": 902, "total_steps": 5364, "loss": 0.24151788651943207, "lr": 1.972069045551688e-05, "epoch": 0.5046153846153846, "percentage": 16.82, "elapsed_time": "1:47:33", "remaining_time": "8:52:01"} +{"current_steps": 903, "total_steps": 5364, "loss": 0.01287270151078701, "lr": 1.9719160913359706e-05, "epoch": 0.5051748251748251, "percentage": 16.83, "elapsed_time": "1:47:40", "remaining_time": "8:51:55"} +{"current_steps": 904, "total_steps": 5364, "loss": 0.12392720580101013, "lr": 1.9717627254268533e-05, "epoch": 0.5057342657342657, "percentage": 16.85, "elapsed_time": "1:47:47", "remaining_time": "8:51:48"} +{"current_steps": 905, "total_steps": 5364, "loss": 0.26476046442985535, "lr": 1.9716089478893e-05, "epoch": 0.5062937062937063, "percentage": 16.87, "elapsed_time": "1:47:54", "remaining_time": "8:51:39"} +{"current_steps": 906, "total_steps": 5364, "loss": 0.15284022688865662, "lr": 1.9714547587884492e-05, "epoch": 0.5068531468531469, "percentage": 16.89, "elapsed_time": "1:48:01", "remaining_time": "8:51:33"} +{"current_steps": 907, "total_steps": 5364, "loss": 0.2212950438261032, "lr": 1.9713001581896136e-05, "epoch": 0.5074125874125874, "percentage": 16.91, "elapsed_time": "1:48:08", "remaining_time": "8:51:24"} +{"current_steps": 908, "total_steps": 5364, "loss": 0.033081285655498505, "lr": 1.971145146158281e-05, "epoch": 0.507972027972028, "percentage": 16.93, "elapsed_time": "1:48:15", "remaining_time": "8:51:18"} +{"current_steps": 909, "total_steps": 5364, "loss": 0.37525105476379395, "lr": 1.970989722760112e-05, "epoch": 0.5085314685314686, "percentage": 16.95, "elapsed_time": "1:48:22", "remaining_time": "8:51:09"} +{"current_steps": 910, "total_steps": 5364, "loss": 0.1389291137456894, "lr": 1.970833888060943e-05, "epoch": 0.509090909090909, "percentage": 16.96, "elapsed_time": "1:48:29", "remaining_time": "8:51:02"} +{"current_steps": 911, "total_steps": 5364, "loss": 0.1689014732837677, "lr": 1.970677642126784e-05, "epoch": 0.5096503496503496, "percentage": 16.98, "elapsed_time": "1:48:36", "remaining_time": "8:50:54"} +{"current_steps": 912, "total_steps": 5364, "loss": 0.3219829201698303, "lr": 1.9705209850238185e-05, "epoch": 0.5102097902097902, "percentage": 17.0, "elapsed_time": "1:48:43", "remaining_time": "8:50:47"} +{"current_steps": 913, "total_steps": 5364, "loss": 0.030895326286554337, "lr": 1.9703639168184052e-05, "epoch": 0.5107692307692308, "percentage": 17.02, "elapsed_time": "1:48:50", "remaining_time": "8:50:37"} +{"current_steps": 914, "total_steps": 5364, "loss": 0.11441054195165634, "lr": 1.9702064375770765e-05, "epoch": 0.5113286713286713, "percentage": 17.04, "elapsed_time": "1:48:58", "remaining_time": "8:50:32"} +{"current_steps": 915, "total_steps": 5364, "loss": 0.20935362577438354, "lr": 1.9700485473665386e-05, "epoch": 0.5118881118881119, "percentage": 17.06, "elapsed_time": "1:49:04", "remaining_time": "8:50:23"} +{"current_steps": 916, "total_steps": 5364, "loss": 0.31683462858200073, "lr": 1.969890246253673e-05, "epoch": 0.5124475524475525, "percentage": 17.08, "elapsed_time": "1:49:11", "remaining_time": "8:50:15"} +{"current_steps": 917, "total_steps": 5364, "loss": 0.2438500076532364, "lr": 1.9697315343055336e-05, "epoch": 0.513006993006993, "percentage": 17.1, "elapsed_time": "1:49:19", "remaining_time": "8:50:08"} +{"current_steps": 918, "total_steps": 5364, "loss": 0.10532652586698532, "lr": 1.969572411589349e-05, "epoch": 0.5135664335664336, "percentage": 17.11, "elapsed_time": "1:49:25", "remaining_time": "8:49:59"} +{"current_steps": 919, "total_steps": 5364, "loss": 0.48565253615379333, "lr": 1.9694128781725225e-05, "epoch": 0.5141258741258741, "percentage": 17.13, "elapsed_time": "1:49:32", "remaining_time": "8:49:51"} +{"current_steps": 920, "total_steps": 5364, "loss": 0.3974907696247101, "lr": 1.969252934122631e-05, "epoch": 0.5146853146853146, "percentage": 17.15, "elapsed_time": "1:49:39", "remaining_time": "8:49:42"} +{"current_steps": 921, "total_steps": 5364, "loss": 0.2859296202659607, "lr": 1.969092579507424e-05, "epoch": 0.5152447552447552, "percentage": 17.17, "elapsed_time": "1:49:46", "remaining_time": "8:49:34"} +{"current_steps": 922, "total_steps": 5364, "loss": 0.22826267778873444, "lr": 1.9689318143948266e-05, "epoch": 0.5158041958041958, "percentage": 17.19, "elapsed_time": "1:49:53", "remaining_time": "8:49:26"} +{"current_steps": 923, "total_steps": 5364, "loss": 0.26515689492225647, "lr": 1.9687706388529376e-05, "epoch": 0.5163636363636364, "percentage": 17.21, "elapsed_time": "1:50:00", "remaining_time": "8:49:18"} +{"current_steps": 924, "total_steps": 5364, "loss": 0.22250650823116302, "lr": 1.9686090529500282e-05, "epoch": 0.5169230769230769, "percentage": 17.23, "elapsed_time": "1:50:07", "remaining_time": "8:49:12"} +{"current_steps": 925, "total_steps": 5364, "loss": 0.17744609713554382, "lr": 1.968447056754546e-05, "epoch": 0.5174825174825175, "percentage": 17.24, "elapsed_time": "1:50:15", "remaining_time": "8:49:06"} +{"current_steps": 926, "total_steps": 5364, "loss": 0.1091318279504776, "lr": 1.9682846503351097e-05, "epoch": 0.5180419580419581, "percentage": 17.26, "elapsed_time": "1:50:22", "remaining_time": "8:48:59"} +{"current_steps": 927, "total_steps": 5364, "loss": 0.22607238590717316, "lr": 1.9681218337605134e-05, "epoch": 0.5186013986013986, "percentage": 17.28, "elapsed_time": "1:50:29", "remaining_time": "8:48:52"} +{"current_steps": 928, "total_steps": 5364, "loss": 0.3530200719833374, "lr": 1.9679586070997243e-05, "epoch": 0.5191608391608392, "percentage": 17.3, "elapsed_time": "1:50:36", "remaining_time": "8:48:44"} +{"current_steps": 929, "total_steps": 5364, "loss": 0.2328314483165741, "lr": 1.9677949704218836e-05, "epoch": 0.5197202797202797, "percentage": 17.32, "elapsed_time": "1:50:44", "remaining_time": "8:48:38"} +{"current_steps": 930, "total_steps": 5364, "loss": 0.10878923535346985, "lr": 1.967630923796306e-05, "epoch": 0.5202797202797202, "percentage": 17.34, "elapsed_time": "1:50:51", "remaining_time": "8:48:30"} +{"current_steps": 931, "total_steps": 5364, "loss": 0.1710374653339386, "lr": 1.96746646729248e-05, "epoch": 0.5208391608391608, "percentage": 17.36, "elapsed_time": "1:50:57", "remaining_time": "8:48:21"} +{"current_steps": 932, "total_steps": 5364, "loss": 0.36729955673217773, "lr": 1.967301600980068e-05, "epoch": 0.5213986013986014, "percentage": 17.38, "elapsed_time": "1:51:05", "remaining_time": "8:48:15"} +{"current_steps": 933, "total_steps": 5364, "loss": 0.06783082336187363, "lr": 1.967136324928905e-05, "epoch": 0.521958041958042, "percentage": 17.39, "elapsed_time": "1:51:12", "remaining_time": "8:48:10"} +{"current_steps": 934, "total_steps": 5364, "loss": 0.2012096643447876, "lr": 1.966970639209e-05, "epoch": 0.5225174825174825, "percentage": 17.41, "elapsed_time": "1:51:19", "remaining_time": "8:48:01"} +{"current_steps": 935, "total_steps": 5364, "loss": 0.05832952633500099, "lr": 1.9668045438905367e-05, "epoch": 0.5230769230769231, "percentage": 17.43, "elapsed_time": "1:51:27", "remaining_time": "8:47:56"} +{"current_steps": 936, "total_steps": 5364, "loss": 0.37049317359924316, "lr": 1.9666380390438706e-05, "epoch": 0.5236363636363637, "percentage": 17.45, "elapsed_time": "1:51:34", "remaining_time": "8:47:49"} +{"current_steps": 937, "total_steps": 5364, "loss": 0.10587126016616821, "lr": 1.9664711247395313e-05, "epoch": 0.5241958041958042, "percentage": 17.47, "elapsed_time": "1:51:41", "remaining_time": "8:47:43"} +{"current_steps": 938, "total_steps": 5364, "loss": 0.39503955841064453, "lr": 1.9663038010482224e-05, "epoch": 0.5247552447552447, "percentage": 17.49, "elapsed_time": "1:51:48", "remaining_time": "8:47:35"} +{"current_steps": 939, "total_steps": 5364, "loss": 0.20081260800361633, "lr": 1.9661360680408202e-05, "epoch": 0.5253146853146853, "percentage": 17.51, "elapsed_time": "1:51:55", "remaining_time": "8:47:27"} +{"current_steps": 940, "total_steps": 5364, "loss": 0.4869721829891205, "lr": 1.965967925788375e-05, "epoch": 0.5258741258741259, "percentage": 17.52, "elapsed_time": "1:52:02", "remaining_time": "8:47:17"} +{"current_steps": 941, "total_steps": 5364, "loss": 0.17658472061157227, "lr": 1.9657993743621098e-05, "epoch": 0.5264335664335664, "percentage": 17.54, "elapsed_time": "1:52:09", "remaining_time": "8:47:09"} +{"current_steps": 942, "total_steps": 5364, "loss": 0.3039664626121521, "lr": 1.965630413833421e-05, "epoch": 0.526993006993007, "percentage": 17.56, "elapsed_time": "1:52:16", "remaining_time": "8:47:01"} +{"current_steps": 943, "total_steps": 5364, "loss": 0.43305009603500366, "lr": 1.9654610442738786e-05, "epoch": 0.5275524475524476, "percentage": 17.58, "elapsed_time": "1:52:22", "remaining_time": "8:46:52"} +{"current_steps": 944, "total_steps": 5364, "loss": 0.14372022449970245, "lr": 1.965291265755226e-05, "epoch": 0.5281118881118881, "percentage": 17.6, "elapsed_time": "1:52:30", "remaining_time": "8:46:45"} +{"current_steps": 945, "total_steps": 5364, "loss": 0.21755558252334595, "lr": 1.9651210783493794e-05, "epoch": 0.5286713286713287, "percentage": 17.62, "elapsed_time": "1:52:37", "remaining_time": "8:46:41"} +{"current_steps": 946, "total_steps": 5364, "loss": 0.2760806083679199, "lr": 1.964950482128428e-05, "epoch": 0.5292307692307693, "percentage": 17.64, "elapsed_time": "1:52:45", "remaining_time": "8:46:34"} +{"current_steps": 947, "total_steps": 5364, "loss": 0.13141363859176636, "lr": 1.9647794771646356e-05, "epoch": 0.5297902097902097, "percentage": 17.65, "elapsed_time": "1:52:52", "remaining_time": "8:46:26"} +{"current_steps": 948, "total_steps": 5364, "loss": 0.2564341127872467, "lr": 1.9646080635304374e-05, "epoch": 0.5303496503496503, "percentage": 17.67, "elapsed_time": "1:52:59", "remaining_time": "8:46:18"} +{"current_steps": 949, "total_steps": 5364, "loss": 0.37522318959236145, "lr": 1.964436241298442e-05, "epoch": 0.5309090909090909, "percentage": 17.69, "elapsed_time": "1:53:06", "remaining_time": "8:46:10"} +{"current_steps": 950, "total_steps": 5364, "loss": 0.4672318696975708, "lr": 1.9642640105414324e-05, "epoch": 0.5314685314685315, "percentage": 17.71, "elapsed_time": "1:53:13", "remaining_time": "8:46:02"} +{"current_steps": 951, "total_steps": 5364, "loss": 0.08840466290712357, "lr": 1.9640913713323632e-05, "epoch": 0.532027972027972, "percentage": 17.73, "elapsed_time": "1:53:19", "remaining_time": "8:45:54"} +{"current_steps": 952, "total_steps": 5364, "loss": 0.3974948525428772, "lr": 1.9639183237443623e-05, "epoch": 0.5325874125874126, "percentage": 17.75, "elapsed_time": "1:53:26", "remaining_time": "8:45:46"} +{"current_steps": 953, "total_steps": 5364, "loss": 0.08519744873046875, "lr": 1.9637448678507317e-05, "epoch": 0.5331468531468532, "percentage": 17.77, "elapsed_time": "1:53:34", "remaining_time": "8:45:39"} +{"current_steps": 954, "total_steps": 5364, "loss": 0.22603653371334076, "lr": 1.9635710037249446e-05, "epoch": 0.5337062937062937, "percentage": 17.79, "elapsed_time": "1:53:41", "remaining_time": "8:45:33"} +{"current_steps": 955, "total_steps": 5364, "loss": 0.11233668029308319, "lr": 1.9633967314406485e-05, "epoch": 0.5342657342657343, "percentage": 17.8, "elapsed_time": "1:53:47", "remaining_time": "8:45:23"} +{"current_steps": 956, "total_steps": 5364, "loss": 0.30219048261642456, "lr": 1.9632220510716635e-05, "epoch": 0.5348251748251748, "percentage": 17.82, "elapsed_time": "1:53:54", "remaining_time": "8:45:14"} +{"current_steps": 957, "total_steps": 5364, "loss": 0.3165651559829712, "lr": 1.9630469626919814e-05, "epoch": 0.5353846153846153, "percentage": 17.84, "elapsed_time": "1:54:01", "remaining_time": "8:45:06"} +{"current_steps": 958, "total_steps": 5364, "loss": 0.14552021026611328, "lr": 1.962871466375769e-05, "epoch": 0.5359440559440559, "percentage": 17.86, "elapsed_time": "1:54:08", "remaining_time": "8:44:58"} +{"current_steps": 959, "total_steps": 5364, "loss": 0.31409138441085815, "lr": 1.9626955621973634e-05, "epoch": 0.5365034965034965, "percentage": 17.88, "elapsed_time": "1:54:15", "remaining_time": "8:44:49"} +{"current_steps": 960, "total_steps": 5364, "loss": 0.13729244470596313, "lr": 1.9625192502312772e-05, "epoch": 0.5370629370629371, "percentage": 17.9, "elapsed_time": "1:54:22", "remaining_time": "8:44:41"} +{"current_steps": 961, "total_steps": 5364, "loss": 0.21573972702026367, "lr": 1.9623425305521932e-05, "epoch": 0.5376223776223776, "percentage": 17.92, "elapsed_time": "1:54:29", "remaining_time": "8:44:33"} +{"current_steps": 962, "total_steps": 5364, "loss": 0.2081308215856552, "lr": 1.9621654032349685e-05, "epoch": 0.5381818181818182, "percentage": 17.93, "elapsed_time": "1:54:36", "remaining_time": "8:44:24"} +{"current_steps": 963, "total_steps": 5364, "loss": 0.2961956262588501, "lr": 1.961987868354632e-05, "epoch": 0.5387412587412588, "percentage": 17.95, "elapsed_time": "1:54:43", "remaining_time": "8:44:17"} +{"current_steps": 964, "total_steps": 5364, "loss": 0.12377064675092697, "lr": 1.9618099259863862e-05, "epoch": 0.5393006993006993, "percentage": 17.97, "elapsed_time": "1:54:50", "remaining_time": "8:44:09"} +{"current_steps": 965, "total_steps": 5364, "loss": 0.043612588196992874, "lr": 1.961631576205605e-05, "epoch": 0.5398601398601398, "percentage": 17.99, "elapsed_time": "1:54:57", "remaining_time": "8:44:02"} +{"current_steps": 966, "total_steps": 5364, "loss": 0.24353444576263428, "lr": 1.9614528190878356e-05, "epoch": 0.5404195804195804, "percentage": 18.01, "elapsed_time": "1:55:04", "remaining_time": "8:43:54"} +{"current_steps": 967, "total_steps": 5364, "loss": 0.21760807931423187, "lr": 1.9612736547087977e-05, "epoch": 0.540979020979021, "percentage": 18.03, "elapsed_time": "1:55:11", "remaining_time": "8:43:45"} +{"current_steps": 968, "total_steps": 5364, "loss": 0.26910704374313354, "lr": 1.9610940831443836e-05, "epoch": 0.5415384615384615, "percentage": 18.05, "elapsed_time": "1:55:18", "remaining_time": "8:43:37"} +{"current_steps": 969, "total_steps": 5364, "loss": 0.2254650592803955, "lr": 1.9609141044706576e-05, "epoch": 0.5420979020979021, "percentage": 18.06, "elapsed_time": "1:55:25", "remaining_time": "8:43:29"} +{"current_steps": 970, "total_steps": 5364, "loss": 0.10507339239120483, "lr": 1.9607337187638573e-05, "epoch": 0.5426573426573427, "percentage": 18.08, "elapsed_time": "1:55:32", "remaining_time": "8:43:22"} +{"current_steps": 971, "total_steps": 5364, "loss": 0.19645673036575317, "lr": 1.9605529261003916e-05, "epoch": 0.5432167832167832, "percentage": 18.1, "elapsed_time": "1:55:39", "remaining_time": "8:43:15"} +{"current_steps": 972, "total_steps": 5364, "loss": 0.32230716943740845, "lr": 1.9603717265568427e-05, "epoch": 0.5437762237762238, "percentage": 18.12, "elapsed_time": "1:55:46", "remaining_time": "8:43:06"} +{"current_steps": 973, "total_steps": 5364, "loss": 0.15786448121070862, "lr": 1.9601901202099648e-05, "epoch": 0.5443356643356644, "percentage": 18.14, "elapsed_time": "1:55:53", "remaining_time": "8:43:00"} +{"current_steps": 974, "total_steps": 5364, "loss": 0.144511878490448, "lr": 1.960008107136684e-05, "epoch": 0.5448951048951048, "percentage": 18.16, "elapsed_time": "1:56:01", "remaining_time": "8:42:55"} +{"current_steps": 975, "total_steps": 5364, "loss": 0.19467437267303467, "lr": 1.9598256874140995e-05, "epoch": 0.5454545454545454, "percentage": 18.18, "elapsed_time": "1:56:08", "remaining_time": "8:42:47"} +{"current_steps": 976, "total_steps": 5364, "loss": 0.16548840701580048, "lr": 1.9596428611194825e-05, "epoch": 0.546013986013986, "percentage": 18.2, "elapsed_time": "1:56:15", "remaining_time": "8:42:43"} +{"current_steps": 977, "total_steps": 5364, "loss": 0.14353393018245697, "lr": 1.9594596283302762e-05, "epoch": 0.5465734265734266, "percentage": 18.21, "elapsed_time": "1:56:22", "remaining_time": "8:42:34"} +{"current_steps": 978, "total_steps": 5364, "loss": 0.15446056425571442, "lr": 1.959275989124096e-05, "epoch": 0.5471328671328671, "percentage": 18.23, "elapsed_time": "1:56:30", "remaining_time": "8:42:28"} +{"current_steps": 979, "total_steps": 5364, "loss": 0.1718764752149582, "lr": 1.9590919435787295e-05, "epoch": 0.5476923076923077, "percentage": 18.25, "elapsed_time": "1:56:37", "remaining_time": "8:42:20"} +{"current_steps": 980, "total_steps": 5364, "loss": 0.16891226172447205, "lr": 1.9589074917721364e-05, "epoch": 0.5482517482517483, "percentage": 18.27, "elapsed_time": "1:56:44", "remaining_time": "8:42:13"} +{"current_steps": 981, "total_steps": 5364, "loss": 0.18720705807209015, "lr": 1.958722633782449e-05, "epoch": 0.5488111888111888, "percentage": 18.29, "elapsed_time": "1:56:51", "remaining_time": "8:42:04"} +{"current_steps": 982, "total_steps": 5364, "loss": 0.29135066270828247, "lr": 1.9585373696879705e-05, "epoch": 0.5493706293706294, "percentage": 18.31, "elapsed_time": "1:56:58", "remaining_time": "8:41:56"} +{"current_steps": 983, "total_steps": 5364, "loss": 0.3132430911064148, "lr": 1.9583516995671773e-05, "epoch": 0.5499300699300699, "percentage": 18.33, "elapsed_time": "1:57:04", "remaining_time": "8:41:48"} +{"current_steps": 984, "total_steps": 5364, "loss": 0.21503329277038574, "lr": 1.958165623498717e-05, "epoch": 0.5504895104895104, "percentage": 18.34, "elapsed_time": "1:57:12", "remaining_time": "8:41:41"} +{"current_steps": 985, "total_steps": 5364, "loss": 0.04449233412742615, "lr": 1.9579791415614094e-05, "epoch": 0.551048951048951, "percentage": 18.36, "elapsed_time": "1:57:20", "remaining_time": "8:41:38"} +{"current_steps": 986, "total_steps": 5364, "loss": 0.28126204013824463, "lr": 1.9577922538342467e-05, "epoch": 0.5516083916083916, "percentage": 18.38, "elapsed_time": "1:57:26", "remaining_time": "8:41:29"} +{"current_steps": 987, "total_steps": 5364, "loss": 0.14415481686592102, "lr": 1.9576049603963922e-05, "epoch": 0.5521678321678322, "percentage": 18.4, "elapsed_time": "1:57:34", "remaining_time": "8:41:22"} +{"current_steps": 988, "total_steps": 5364, "loss": 0.2581061124801636, "lr": 1.9574172613271816e-05, "epoch": 0.5527272727272727, "percentage": 18.42, "elapsed_time": "1:57:41", "remaining_time": "8:41:16"} +{"current_steps": 989, "total_steps": 5364, "loss": 0.17576652765274048, "lr": 1.9572291567061224e-05, "epoch": 0.5532867132867133, "percentage": 18.44, "elapsed_time": "1:57:48", "remaining_time": "8:41:08"} +{"current_steps": 990, "total_steps": 5364, "loss": 0.0734764039516449, "lr": 1.9570406466128934e-05, "epoch": 0.5538461538461539, "percentage": 18.46, "elapsed_time": "1:57:55", "remaining_time": "8:41:02"} +{"current_steps": 991, "total_steps": 5364, "loss": 0.2866566777229309, "lr": 1.956851731127346e-05, "epoch": 0.5544055944055944, "percentage": 18.48, "elapsed_time": "1:58:02", "remaining_time": "8:40:54"} +{"current_steps": 992, "total_steps": 5364, "loss": 0.3186303973197937, "lr": 1.956662410329502e-05, "epoch": 0.554965034965035, "percentage": 18.49, "elapsed_time": "1:58:10", "remaining_time": "8:40:48"} +{"current_steps": 993, "total_steps": 5364, "loss": 0.11688470840454102, "lr": 1.956472684299556e-05, "epoch": 0.5555244755244755, "percentage": 18.51, "elapsed_time": "1:58:17", "remaining_time": "8:40:40"} +{"current_steps": 994, "total_steps": 5364, "loss": 0.17378711700439453, "lr": 1.9562825531178746e-05, "epoch": 0.556083916083916, "percentage": 18.53, "elapsed_time": "1:58:24", "remaining_time": "8:40:34"} +{"current_steps": 995, "total_steps": 5364, "loss": 0.3526480793952942, "lr": 1.9560920168649944e-05, "epoch": 0.5566433566433566, "percentage": 18.55, "elapsed_time": "1:58:31", "remaining_time": "8:40:27"} +{"current_steps": 996, "total_steps": 5364, "loss": 0.18774738907814026, "lr": 1.955901075621625e-05, "epoch": 0.5572027972027972, "percentage": 18.57, "elapsed_time": "1:58:38", "remaining_time": "8:40:20"} +{"current_steps": 997, "total_steps": 5364, "loss": 0.12678533792495728, "lr": 1.9557097294686477e-05, "epoch": 0.5577622377622378, "percentage": 18.59, "elapsed_time": "1:58:45", "remaining_time": "8:40:12"} +{"current_steps": 998, "total_steps": 5364, "loss": 0.20538268983364105, "lr": 1.9555179784871132e-05, "epoch": 0.5583216783216783, "percentage": 18.61, "elapsed_time": "1:58:52", "remaining_time": "8:40:04"} +{"current_steps": 999, "total_steps": 5364, "loss": 0.2257968932390213, "lr": 1.955325822758247e-05, "epoch": 0.5588811188811189, "percentage": 18.62, "elapsed_time": "1:58:59", "remaining_time": "8:39:57"} +{"current_steps": 1000, "total_steps": 5364, "loss": 0.3862190246582031, "lr": 1.9551332623634426e-05, "epoch": 0.5594405594405595, "percentage": 18.64, "elapsed_time": "1:59:06", "remaining_time": "8:39:48"} +{"current_steps": 1001, "total_steps": 5364, "loss": 0.07053911685943604, "lr": 1.954940297384267e-05, "epoch": 0.56, "percentage": 18.66, "elapsed_time": "1:59:41", "remaining_time": "8:41:43"} +{"current_steps": 1002, "total_steps": 5364, "loss": 0.32261836528778076, "lr": 1.9547469279024593e-05, "epoch": 0.5605594405594405, "percentage": 18.68, "elapsed_time": "1:59:48", "remaining_time": "8:41:35"} +{"current_steps": 1003, "total_steps": 5364, "loss": 0.15613330900669098, "lr": 1.9545531539999275e-05, "epoch": 0.5611188811188811, "percentage": 18.7, "elapsed_time": "1:59:56", "remaining_time": "8:41:29"} +{"current_steps": 1004, "total_steps": 5364, "loss": 0.17701421678066254, "lr": 1.9543589757587525e-05, "epoch": 0.5616783216783217, "percentage": 18.72, "elapsed_time": "2:00:03", "remaining_time": "8:41:21"} +{"current_steps": 1005, "total_steps": 5364, "loss": 0.3297882676124573, "lr": 1.954164393261186e-05, "epoch": 0.5622377622377622, "percentage": 18.74, "elapsed_time": "2:00:10", "remaining_time": "8:41:13"} +{"current_steps": 1006, "total_steps": 5364, "loss": 0.06182924285531044, "lr": 1.9539694065896516e-05, "epoch": 0.5627972027972028, "percentage": 18.75, "elapsed_time": "2:00:17", "remaining_time": "8:41:05"} +{"current_steps": 1007, "total_steps": 5364, "loss": 0.1331361085176468, "lr": 1.9537740158267432e-05, "epoch": 0.5633566433566434, "percentage": 18.77, "elapsed_time": "2:00:25", "remaining_time": "8:41:00"} +{"current_steps": 1008, "total_steps": 5364, "loss": 0.13544823229312897, "lr": 1.9535782210552266e-05, "epoch": 0.5639160839160839, "percentage": 18.79, "elapsed_time": "2:00:31", "remaining_time": "8:40:52"} +{"current_steps": 1009, "total_steps": 5364, "loss": 0.16716547310352325, "lr": 1.953382022358038e-05, "epoch": 0.5644755244755245, "percentage": 18.81, "elapsed_time": "2:00:39", "remaining_time": "8:40:45"} +{"current_steps": 1010, "total_steps": 5364, "loss": 0.33848369121551514, "lr": 1.9531854198182853e-05, "epoch": 0.5650349650349651, "percentage": 18.83, "elapsed_time": "2:00:46", "remaining_time": "8:40:38"} +{"current_steps": 1011, "total_steps": 5364, "loss": 0.4353000819683075, "lr": 1.952988413519247e-05, "epoch": 0.5655944055944055, "percentage": 18.85, "elapsed_time": "2:00:53", "remaining_time": "8:40:31"} +{"current_steps": 1012, "total_steps": 5364, "loss": 0.05286436527967453, "lr": 1.9527910035443735e-05, "epoch": 0.5661538461538461, "percentage": 18.87, "elapsed_time": "2:01:00", "remaining_time": "8:40:24"} +{"current_steps": 1013, "total_steps": 5364, "loss": 0.24054157733917236, "lr": 1.9525931899772848e-05, "epoch": 0.5667132867132867, "percentage": 18.89, "elapsed_time": "2:01:07", "remaining_time": "8:40:16"} +{"current_steps": 1014, "total_steps": 5364, "loss": 0.14250990748405457, "lr": 1.952394972901773e-05, "epoch": 0.5672727272727273, "percentage": 18.9, "elapsed_time": "2:01:14", "remaining_time": "8:40:08"} +{"current_steps": 1015, "total_steps": 5364, "loss": 0.30289068818092346, "lr": 1.952196352401801e-05, "epoch": 0.5678321678321678, "percentage": 18.92, "elapsed_time": "2:01:21", "remaining_time": "8:39:59"} +{"current_steps": 1016, "total_steps": 5364, "loss": 0.27133095264434814, "lr": 1.9519973285615023e-05, "epoch": 0.5683916083916084, "percentage": 18.94, "elapsed_time": "2:01:28", "remaining_time": "8:39:51"} +{"current_steps": 1017, "total_steps": 5364, "loss": 0.1966119110584259, "lr": 1.9517979014651808e-05, "epoch": 0.568951048951049, "percentage": 18.96, "elapsed_time": "2:01:35", "remaining_time": "8:39:43"} +{"current_steps": 1018, "total_steps": 5364, "loss": 0.17500001192092896, "lr": 1.9515980711973123e-05, "epoch": 0.5695104895104895, "percentage": 18.98, "elapsed_time": "2:01:42", "remaining_time": "8:39:35"} +{"current_steps": 1019, "total_steps": 5364, "loss": 0.1158447340130806, "lr": 1.9513978378425427e-05, "epoch": 0.5700699300699301, "percentage": 19.0, "elapsed_time": "2:01:49", "remaining_time": "8:39:27"} +{"current_steps": 1020, "total_steps": 5364, "loss": 0.15415272116661072, "lr": 1.9511972014856882e-05, "epoch": 0.5706293706293706, "percentage": 19.02, "elapsed_time": "2:01:56", "remaining_time": "8:39:18"} +{"current_steps": 1021, "total_steps": 5364, "loss": 0.2781256139278412, "lr": 1.9509961622117366e-05, "epoch": 0.5711888111888112, "percentage": 19.03, "elapsed_time": "2:02:03", "remaining_time": "8:39:10"} +{"current_steps": 1022, "total_steps": 5364, "loss": 0.2514978051185608, "lr": 1.9507947201058464e-05, "epoch": 0.5717482517482517, "percentage": 19.05, "elapsed_time": "2:02:10", "remaining_time": "8:39:01"} +{"current_steps": 1023, "total_steps": 5364, "loss": 0.09165353327989578, "lr": 1.9505928752533458e-05, "epoch": 0.5723076923076923, "percentage": 19.07, "elapsed_time": "2:02:17", "remaining_time": "8:38:54"} +{"current_steps": 1024, "total_steps": 5364, "loss": 0.14220130443572998, "lr": 1.9503906277397342e-05, "epoch": 0.5728671328671329, "percentage": 19.09, "elapsed_time": "2:02:23", "remaining_time": "8:38:45"} +{"current_steps": 1025, "total_steps": 5364, "loss": 0.315986692905426, "lr": 1.950187977650682e-05, "epoch": 0.5734265734265734, "percentage": 19.11, "elapsed_time": "2:02:31", "remaining_time": "8:38:38"} +{"current_steps": 1026, "total_steps": 5364, "loss": 0.11382804065942764, "lr": 1.9499849250720292e-05, "epoch": 0.573986013986014, "percentage": 19.13, "elapsed_time": "2:02:38", "remaining_time": "8:38:30"} +{"current_steps": 1027, "total_steps": 5364, "loss": 0.12490015476942062, "lr": 1.9497814700897866e-05, "epoch": 0.5745454545454546, "percentage": 19.15, "elapsed_time": "2:02:45", "remaining_time": "8:38:22"} +{"current_steps": 1028, "total_steps": 5364, "loss": 0.0684887021780014, "lr": 1.9495776127901357e-05, "epoch": 0.5751048951048952, "percentage": 19.16, "elapsed_time": "2:02:52", "remaining_time": "8:38:14"} +{"current_steps": 1029, "total_steps": 5364, "loss": 0.12830427289009094, "lr": 1.9493733532594286e-05, "epoch": 0.5756643356643356, "percentage": 19.18, "elapsed_time": "2:02:59", "remaining_time": "8:38:07"} +{"current_steps": 1030, "total_steps": 5364, "loss": 0.19304972887039185, "lr": 1.9491686915841876e-05, "epoch": 0.5762237762237762, "percentage": 19.2, "elapsed_time": "2:03:06", "remaining_time": "8:38:00"} +{"current_steps": 1031, "total_steps": 5364, "loss": 0.34067511558532715, "lr": 1.9489636278511046e-05, "epoch": 0.5767832167832168, "percentage": 19.22, "elapsed_time": "2:03:13", "remaining_time": "8:37:52"} +{"current_steps": 1032, "total_steps": 5364, "loss": 0.24409878253936768, "lr": 1.9487581621470426e-05, "epoch": 0.5773426573426573, "percentage": 19.24, "elapsed_time": "2:03:20", "remaining_time": "8:37:43"} +{"current_steps": 1033, "total_steps": 5364, "loss": 0.11452331393957138, "lr": 1.948552294559035e-05, "epoch": 0.5779020979020979, "percentage": 19.26, "elapsed_time": "2:03:27", "remaining_time": "8:37:38"} +{"current_steps": 1034, "total_steps": 5364, "loss": 0.2811158001422882, "lr": 1.9483460251742852e-05, "epoch": 0.5784615384615385, "percentage": 19.28, "elapsed_time": "2:03:34", "remaining_time": "8:37:30"} +{"current_steps": 1035, "total_steps": 5364, "loss": 0.15953418612480164, "lr": 1.9481393540801664e-05, "epoch": 0.579020979020979, "percentage": 19.3, "elapsed_time": "2:03:42", "remaining_time": "8:37:23"} +{"current_steps": 1036, "total_steps": 5364, "loss": 0.12728016078472137, "lr": 1.947932281364222e-05, "epoch": 0.5795804195804196, "percentage": 19.31, "elapsed_time": "2:03:49", "remaining_time": "8:37:17"} +{"current_steps": 1037, "total_steps": 5364, "loss": 0.4229060113430023, "lr": 1.947724807114167e-05, "epoch": 0.5801398601398602, "percentage": 19.33, "elapsed_time": "2:03:56", "remaining_time": "8:37:09"} +{"current_steps": 1038, "total_steps": 5364, "loss": 0.37713998556137085, "lr": 1.947516931417884e-05, "epoch": 0.5806993006993006, "percentage": 19.35, "elapsed_time": "2:04:03", "remaining_time": "8:37:03"} +{"current_steps": 1039, "total_steps": 5364, "loss": 0.1380643993616104, "lr": 1.9473086543634277e-05, "epoch": 0.5812587412587412, "percentage": 19.37, "elapsed_time": "2:04:10", "remaining_time": "8:36:55"} +{"current_steps": 1040, "total_steps": 5364, "loss": 0.2617342174053192, "lr": 1.9470999760390216e-05, "epoch": 0.5818181818181818, "percentage": 19.39, "elapsed_time": "2:04:17", "remaining_time": "8:36:47"} +{"current_steps": 1041, "total_steps": 5364, "loss": 0.07876032590866089, "lr": 1.9468908965330603e-05, "epoch": 0.5823776223776224, "percentage": 19.41, "elapsed_time": "2:04:24", "remaining_time": "8:36:39"} +{"current_steps": 1042, "total_steps": 5364, "loss": 0.3294464647769928, "lr": 1.9466814159341066e-05, "epoch": 0.5829370629370629, "percentage": 19.43, "elapsed_time": "2:04:31", "remaining_time": "8:36:31"} +{"current_steps": 1043, "total_steps": 5364, "loss": 0.37307578325271606, "lr": 1.9464715343308957e-05, "epoch": 0.5834965034965035, "percentage": 19.44, "elapsed_time": "2:04:38", "remaining_time": "8:36:23"} +{"current_steps": 1044, "total_steps": 5364, "loss": 0.11747735738754272, "lr": 1.94626125181233e-05, "epoch": 0.5840559440559441, "percentage": 19.46, "elapsed_time": "2:04:45", "remaining_time": "8:36:15"} +{"current_steps": 1045, "total_steps": 5364, "loss": 0.20038610696792603, "lr": 1.9460505684674836e-05, "epoch": 0.5846153846153846, "percentage": 19.48, "elapsed_time": "2:04:52", "remaining_time": "8:36:07"} +{"current_steps": 1046, "total_steps": 5364, "loss": 0.08299502730369568, "lr": 1.9458394843855998e-05, "epoch": 0.5851748251748252, "percentage": 19.5, "elapsed_time": "2:04:59", "remaining_time": "8:35:58"} +{"current_steps": 1047, "total_steps": 5364, "loss": 0.4093180298805237, "lr": 1.9456279996560908e-05, "epoch": 0.5857342657342657, "percentage": 19.52, "elapsed_time": "2:05:06", "remaining_time": "8:35:51"} +{"current_steps": 1048, "total_steps": 5364, "loss": 0.31745290756225586, "lr": 1.94541611436854e-05, "epoch": 0.5862937062937063, "percentage": 19.54, "elapsed_time": "2:05:13", "remaining_time": "8:35:43"} +{"current_steps": 1049, "total_steps": 5364, "loss": 0.2111779898405075, "lr": 1.9452038286127e-05, "epoch": 0.5868531468531468, "percentage": 19.56, "elapsed_time": "2:05:20", "remaining_time": "8:35:35"} +{"current_steps": 1050, "total_steps": 5364, "loss": 0.18930718302726746, "lr": 1.944991142478492e-05, "epoch": 0.5874125874125874, "percentage": 19.57, "elapsed_time": "2:05:27", "remaining_time": "8:35:27"} +{"current_steps": 1051, "total_steps": 5364, "loss": 0.08649227023124695, "lr": 1.9447780560560085e-05, "epoch": 0.587972027972028, "percentage": 19.59, "elapsed_time": "2:05:35", "remaining_time": "8:35:24"} +{"current_steps": 1052, "total_steps": 5364, "loss": 0.06979280710220337, "lr": 1.9445645694355098e-05, "epoch": 0.5885314685314685, "percentage": 19.61, "elapsed_time": "2:05:43", "remaining_time": "8:35:17"} +{"current_steps": 1053, "total_steps": 5364, "loss": 0.064674511551857, "lr": 1.9443506827074274e-05, "epoch": 0.5890909090909091, "percentage": 19.63, "elapsed_time": "2:05:50", "remaining_time": "8:35:10"} +{"current_steps": 1054, "total_steps": 5364, "loss": 0.11316889524459839, "lr": 1.94413639596236e-05, "epoch": 0.5896503496503497, "percentage": 19.65, "elapsed_time": "2:05:57", "remaining_time": "8:35:03"} +{"current_steps": 1055, "total_steps": 5364, "loss": 0.30373477935791016, "lr": 1.9439217092910795e-05, "epoch": 0.5902097902097903, "percentage": 19.67, "elapsed_time": "2:06:04", "remaining_time": "8:34:57"} +{"current_steps": 1056, "total_steps": 5364, "loss": 0.21477210521697998, "lr": 1.943706622784523e-05, "epoch": 0.5907692307692308, "percentage": 19.69, "elapsed_time": "2:06:12", "remaining_time": "8:34:52"} +{"current_steps": 1057, "total_steps": 5364, "loss": 0.15463876724243164, "lr": 1.9434911365337995e-05, "epoch": 0.5913286713286713, "percentage": 19.71, "elapsed_time": "2:06:19", "remaining_time": "8:34:44"} +{"current_steps": 1058, "total_steps": 5364, "loss": 0.24767793715000153, "lr": 1.943275250630187e-05, "epoch": 0.5918881118881119, "percentage": 19.72, "elapsed_time": "2:06:26", "remaining_time": "8:34:36"} +{"current_steps": 1059, "total_steps": 5364, "loss": 0.18484212458133698, "lr": 1.943058965165132e-05, "epoch": 0.5924475524475524, "percentage": 19.74, "elapsed_time": "2:06:33", "remaining_time": "8:34:28"} +{"current_steps": 1060, "total_steps": 5364, "loss": 0.29778236150741577, "lr": 1.9428422802302508e-05, "epoch": 0.593006993006993, "percentage": 19.76, "elapsed_time": "2:06:40", "remaining_time": "8:34:22"} +{"current_steps": 1061, "total_steps": 5364, "loss": 0.24156953394412994, "lr": 1.9426251959173293e-05, "epoch": 0.5935664335664336, "percentage": 19.78, "elapsed_time": "2:06:47", "remaining_time": "8:34:13"} +{"current_steps": 1062, "total_steps": 5364, "loss": 0.1564061939716339, "lr": 1.942407712318322e-05, "epoch": 0.5941258741258741, "percentage": 19.8, "elapsed_time": "2:06:55", "remaining_time": "8:34:08"} +{"current_steps": 1063, "total_steps": 5364, "loss": 0.3132760524749756, "lr": 1.942189829525352e-05, "epoch": 0.5946853146853147, "percentage": 19.82, "elapsed_time": "2:07:02", "remaining_time": "8:34:00"} +{"current_steps": 1064, "total_steps": 5364, "loss": 0.35332822799682617, "lr": 1.9419715476307132e-05, "epoch": 0.5952447552447553, "percentage": 19.84, "elapsed_time": "2:07:09", "remaining_time": "8:33:52"} +{"current_steps": 1065, "total_steps": 5364, "loss": 0.4086112976074219, "lr": 1.9417528667268664e-05, "epoch": 0.5958041958041959, "percentage": 19.85, "elapsed_time": "2:07:16", "remaining_time": "8:33:45"} +{"current_steps": 1066, "total_steps": 5364, "loss": 0.11487323045730591, "lr": 1.941533786906443e-05, "epoch": 0.5963636363636363, "percentage": 19.87, "elapsed_time": "2:07:23", "remaining_time": "8:33:38"} +{"current_steps": 1067, "total_steps": 5364, "loss": 0.2311253547668457, "lr": 1.9413143082622433e-05, "epoch": 0.5969230769230769, "percentage": 19.89, "elapsed_time": "2:07:30", "remaining_time": "8:33:31"} +{"current_steps": 1068, "total_steps": 5364, "loss": 0.3118816018104553, "lr": 1.9410944308872358e-05, "epoch": 0.5974825174825175, "percentage": 19.91, "elapsed_time": "2:07:37", "remaining_time": "8:33:23"} +{"current_steps": 1069, "total_steps": 5364, "loss": 0.15288785099983215, "lr": 1.9408741548745584e-05, "epoch": 0.598041958041958, "percentage": 19.93, "elapsed_time": "2:07:44", "remaining_time": "8:33:16"} +{"current_steps": 1070, "total_steps": 5364, "loss": 0.08625251054763794, "lr": 1.940653480317517e-05, "epoch": 0.5986013986013986, "percentage": 19.95, "elapsed_time": "2:07:52", "remaining_time": "8:33:10"} +{"current_steps": 1071, "total_steps": 5364, "loss": 0.31451061367988586, "lr": 1.9404324073095883e-05, "epoch": 0.5991608391608392, "percentage": 19.97, "elapsed_time": "2:07:59", "remaining_time": "8:33:03"} +{"current_steps": 1072, "total_steps": 5364, "loss": 0.3542405962944031, "lr": 1.940210935944415e-05, "epoch": 0.5997202797202797, "percentage": 19.99, "elapsed_time": "2:08:06", "remaining_time": "8:32:55"} +{"current_steps": 1073, "total_steps": 5364, "loss": 0.18369606137275696, "lr": 1.9399890663158113e-05, "epoch": 0.6002797202797203, "percentage": 20.0, "elapsed_time": "2:08:14", "remaining_time": "8:32:49"} +{"current_steps": 1074, "total_steps": 5364, "loss": 0.19230841100215912, "lr": 1.939766798517758e-05, "epoch": 0.6008391608391609, "percentage": 20.02, "elapsed_time": "2:08:21", "remaining_time": "8:32:43"} +{"current_steps": 1075, "total_steps": 5364, "loss": 0.06281548738479614, "lr": 1.939544132644406e-05, "epoch": 0.6013986013986014, "percentage": 20.04, "elapsed_time": "2:08:28", "remaining_time": "8:32:36"} +{"current_steps": 1076, "total_steps": 5364, "loss": 0.18373005092144012, "lr": 1.9393210687900735e-05, "epoch": 0.6019580419580419, "percentage": 20.06, "elapsed_time": "2:08:35", "remaining_time": "8:32:28"} +{"current_steps": 1077, "total_steps": 5364, "loss": 0.18567317724227905, "lr": 1.9390976070492486e-05, "epoch": 0.6025174825174825, "percentage": 20.08, "elapsed_time": "2:08:42", "remaining_time": "8:32:20"} +{"current_steps": 1078, "total_steps": 5364, "loss": 0.05393151938915253, "lr": 1.9388737475165874e-05, "epoch": 0.6030769230769231, "percentage": 20.1, "elapsed_time": "2:08:49", "remaining_time": "8:32:13"} +{"current_steps": 1079, "total_steps": 5364, "loss": 0.11506589502096176, "lr": 1.938649490286914e-05, "epoch": 0.6036363636363636, "percentage": 20.12, "elapsed_time": "2:08:57", "remaining_time": "8:32:06"} +{"current_steps": 1080, "total_steps": 5364, "loss": 0.142998605966568, "lr": 1.9384248354552214e-05, "epoch": 0.6041958041958042, "percentage": 20.13, "elapsed_time": "2:09:04", "remaining_time": "8:32:00"} +{"current_steps": 1081, "total_steps": 5364, "loss": 0.21724161505699158, "lr": 1.9381997831166716e-05, "epoch": 0.6047552447552448, "percentage": 20.15, "elapsed_time": "2:09:12", "remaining_time": "8:31:57"} +{"current_steps": 1082, "total_steps": 5364, "loss": 0.09619180113077164, "lr": 1.9379743333665937e-05, "epoch": 0.6053146853146854, "percentage": 20.17, "elapsed_time": "2:09:20", "remaining_time": "8:31:51"} +{"current_steps": 1083, "total_steps": 5364, "loss": 0.12282408028841019, "lr": 1.9377484863004863e-05, "epoch": 0.6058741258741259, "percentage": 20.19, "elapsed_time": "2:09:27", "remaining_time": "8:31:45"} +{"current_steps": 1084, "total_steps": 5364, "loss": 0.32464998960494995, "lr": 1.9375222420140157e-05, "epoch": 0.6064335664335664, "percentage": 20.21, "elapsed_time": "2:09:34", "remaining_time": "8:31:37"} +{"current_steps": 1085, "total_steps": 5364, "loss": 0.21114413440227509, "lr": 1.9372956006030162e-05, "epoch": 0.606993006993007, "percentage": 20.23, "elapsed_time": "2:09:41", "remaining_time": "8:31:30"} +{"current_steps": 1086, "total_steps": 5364, "loss": 0.047926660627126694, "lr": 1.937068562163491e-05, "epoch": 0.6075524475524475, "percentage": 20.25, "elapsed_time": "2:09:49", "remaining_time": "8:31:23"} +{"current_steps": 1087, "total_steps": 5364, "loss": 0.3124011158943176, "lr": 1.9368411267916117e-05, "epoch": 0.6081118881118881, "percentage": 20.26, "elapsed_time": "2:09:56", "remaining_time": "8:31:17"} +{"current_steps": 1088, "total_steps": 5364, "loss": 0.3309420943260193, "lr": 1.936613294583717e-05, "epoch": 0.6086713286713287, "percentage": 20.28, "elapsed_time": "2:10:04", "remaining_time": "8:31:11"} +{"current_steps": 1089, "total_steps": 5364, "loss": 0.3941347002983093, "lr": 1.936385065636314e-05, "epoch": 0.6092307692307692, "percentage": 20.3, "elapsed_time": "2:10:11", "remaining_time": "8:31:03"} +{"current_steps": 1090, "total_steps": 5364, "loss": 0.25723814964294434, "lr": 1.936156440046079e-05, "epoch": 0.6097902097902098, "percentage": 20.32, "elapsed_time": "2:10:18", "remaining_time": "8:30:55"} +{"current_steps": 1091, "total_steps": 5364, "loss": 0.27008870244026184, "lr": 1.935927417909854e-05, "epoch": 0.6103496503496504, "percentage": 20.34, "elapsed_time": "2:10:25", "remaining_time": "8:30:48"} +{"current_steps": 1092, "total_steps": 5364, "loss": 0.2866484522819519, "lr": 1.9356979993246515e-05, "epoch": 0.610909090909091, "percentage": 20.36, "elapsed_time": "2:10:32", "remaining_time": "8:30:41"} +{"current_steps": 1093, "total_steps": 5364, "loss": 0.2076454907655716, "lr": 1.9354681843876504e-05, "epoch": 0.6114685314685314, "percentage": 20.38, "elapsed_time": "2:10:40", "remaining_time": "8:30:36"} +{"current_steps": 1094, "total_steps": 5364, "loss": 0.3817736804485321, "lr": 1.935237973196198e-05, "epoch": 0.612027972027972, "percentage": 20.4, "elapsed_time": "2:10:47", "remaining_time": "8:30:30"} +{"current_steps": 1095, "total_steps": 5364, "loss": 0.43747881054878235, "lr": 1.935007365847809e-05, "epoch": 0.6125874125874126, "percentage": 20.41, "elapsed_time": "2:10:54", "remaining_time": "8:30:22"} +{"current_steps": 1096, "total_steps": 5364, "loss": 0.2928038239479065, "lr": 1.9347763624401664e-05, "epoch": 0.6131468531468531, "percentage": 20.43, "elapsed_time": "2:11:01", "remaining_time": "8:30:15"} +{"current_steps": 1097, "total_steps": 5364, "loss": 0.2194501757621765, "lr": 1.934544963071121e-05, "epoch": 0.6137062937062937, "percentage": 20.45, "elapsed_time": "2:11:09", "remaining_time": "8:30:09"} +{"current_steps": 1098, "total_steps": 5364, "loss": 0.337615430355072, "lr": 1.9343131678386908e-05, "epoch": 0.6142657342657343, "percentage": 20.47, "elapsed_time": "2:11:16", "remaining_time": "8:30:01"} +{"current_steps": 1099, "total_steps": 5364, "loss": 0.2631508708000183, "lr": 1.934080976841062e-05, "epoch": 0.6148251748251748, "percentage": 20.49, "elapsed_time": "2:11:23", "remaining_time": "8:29:52"} +{"current_steps": 1100, "total_steps": 5364, "loss": 0.22559913992881775, "lr": 1.9338483901765887e-05, "epoch": 0.6153846153846154, "percentage": 20.51, "elapsed_time": "2:11:30", "remaining_time": "8:29:44"} +{"current_steps": 1101, "total_steps": 5364, "loss": 0.043556440621614456, "lr": 1.9336154079437915e-05, "epoch": 0.615944055944056, "percentage": 20.53, "elapsed_time": "2:11:37", "remaining_time": "8:29:38"} +{"current_steps": 1102, "total_steps": 5364, "loss": 0.26283159852027893, "lr": 1.9333820302413594e-05, "epoch": 0.6165034965034965, "percentage": 20.54, "elapsed_time": "2:11:44", "remaining_time": "8:29:31"} +{"current_steps": 1103, "total_steps": 5364, "loss": 0.19678513705730438, "lr": 1.933148257168149e-05, "epoch": 0.617062937062937, "percentage": 20.56, "elapsed_time": "2:11:51", "remaining_time": "8:29:23"} +{"current_steps": 1104, "total_steps": 5364, "loss": 0.49419569969177246, "lr": 1.9329140888231834e-05, "epoch": 0.6176223776223776, "percentage": 20.58, "elapsed_time": "2:11:58", "remaining_time": "8:29:15"} +{"current_steps": 1105, "total_steps": 5364, "loss": 0.22792497277259827, "lr": 1.9326795253056543e-05, "epoch": 0.6181818181818182, "percentage": 20.6, "elapsed_time": "2:12:05", "remaining_time": "8:29:07"} +{"current_steps": 1106, "total_steps": 5364, "loss": 0.5005597472190857, "lr": 1.9324445667149208e-05, "epoch": 0.6187412587412587, "percentage": 20.62, "elapsed_time": "2:12:12", "remaining_time": "8:28:58"} +{"current_steps": 1107, "total_steps": 5364, "loss": 0.1787535697221756, "lr": 1.9322092131505084e-05, "epoch": 0.6193006993006993, "percentage": 20.64, "elapsed_time": "2:12:19", "remaining_time": "8:28:52"} +{"current_steps": 1108, "total_steps": 5364, "loss": 0.26390495896339417, "lr": 1.9319734647121104e-05, "epoch": 0.6198601398601399, "percentage": 20.66, "elapsed_time": "2:12:26", "remaining_time": "8:28:43"} +{"current_steps": 1109, "total_steps": 5364, "loss": 0.2035454958677292, "lr": 1.9317373214995873e-05, "epoch": 0.6204195804195805, "percentage": 20.67, "elapsed_time": "2:12:33", "remaining_time": "8:28:37"} +{"current_steps": 1110, "total_steps": 5364, "loss": 0.25793570280075073, "lr": 1.9315007836129664e-05, "epoch": 0.620979020979021, "percentage": 20.69, "elapsed_time": "2:12:41", "remaining_time": "8:28:30"} +{"current_steps": 1111, "total_steps": 5364, "loss": 0.12532973289489746, "lr": 1.9312638511524436e-05, "epoch": 0.6215384615384615, "percentage": 20.71, "elapsed_time": "2:12:48", "remaining_time": "8:28:22"} +{"current_steps": 1112, "total_steps": 5364, "loss": 0.35972028970718384, "lr": 1.931026524218381e-05, "epoch": 0.6220979020979021, "percentage": 20.73, "elapsed_time": "2:12:55", "remaining_time": "8:28:14"} +{"current_steps": 1113, "total_steps": 5364, "loss": 0.09086612612009048, "lr": 1.930788802911307e-05, "epoch": 0.6226573426573426, "percentage": 20.75, "elapsed_time": "2:13:02", "remaining_time": "8:28:06"} +{"current_steps": 1114, "total_steps": 5364, "loss": 0.3678678870201111, "lr": 1.930550687331918e-05, "epoch": 0.6232167832167832, "percentage": 20.77, "elapsed_time": "2:13:09", "remaining_time": "8:27:59"} +{"current_steps": 1115, "total_steps": 5364, "loss": 0.12469181418418884, "lr": 1.9303121775810777e-05, "epoch": 0.6237762237762238, "percentage": 20.79, "elapsed_time": "2:13:16", "remaining_time": "8:27:54"} +{"current_steps": 1116, "total_steps": 5364, "loss": 0.31603944301605225, "lr": 1.930073273759816e-05, "epoch": 0.6243356643356643, "percentage": 20.81, "elapsed_time": "2:13:23", "remaining_time": "8:27:46"} +{"current_steps": 1117, "total_steps": 5364, "loss": 0.2397618293762207, "lr": 1.9298339759693304e-05, "epoch": 0.6248951048951049, "percentage": 20.82, "elapsed_time": "2:13:30", "remaining_time": "8:27:37"} +{"current_steps": 1118, "total_steps": 5364, "loss": 0.3813819885253906, "lr": 1.9295942843109844e-05, "epoch": 0.6254545454545455, "percentage": 20.84, "elapsed_time": "2:13:38", "remaining_time": "8:27:31"} +{"current_steps": 1119, "total_steps": 5364, "loss": 0.18809424340724945, "lr": 1.9293541988863088e-05, "epoch": 0.6260139860139861, "percentage": 20.86, "elapsed_time": "2:13:45", "remaining_time": "8:27:23"} +{"current_steps": 1120, "total_steps": 5364, "loss": 0.17729797959327698, "lr": 1.9291137197970018e-05, "epoch": 0.6265734265734266, "percentage": 20.88, "elapsed_time": "2:13:52", "remaining_time": "8:27:15"} +{"current_steps": 1121, "total_steps": 5364, "loss": 0.18533581495285034, "lr": 1.9288728471449272e-05, "epoch": 0.6271328671328671, "percentage": 20.9, "elapsed_time": "2:13:59", "remaining_time": "8:27:07"} +{"current_steps": 1122, "total_steps": 5364, "loss": 0.13926270604133606, "lr": 1.928631581032117e-05, "epoch": 0.6276923076923077, "percentage": 20.92, "elapsed_time": "2:14:06", "remaining_time": "8:27:00"} +{"current_steps": 1123, "total_steps": 5364, "loss": 0.12458394467830658, "lr": 1.9283899215607684e-05, "epoch": 0.6282517482517482, "percentage": 20.94, "elapsed_time": "2:14:13", "remaining_time": "8:26:54"} +{"current_steps": 1124, "total_steps": 5364, "loss": 0.3207188546657562, "lr": 1.9281478688332456e-05, "epoch": 0.6288111888111888, "percentage": 20.95, "elapsed_time": "2:14:20", "remaining_time": "8:26:48"} +{"current_steps": 1125, "total_steps": 5364, "loss": 0.20889319479465485, "lr": 1.9279054229520796e-05, "epoch": 0.6293706293706294, "percentage": 20.97, "elapsed_time": "2:14:27", "remaining_time": "8:26:40"} +{"current_steps": 1126, "total_steps": 5364, "loss": 0.11756350845098495, "lr": 1.9276625840199685e-05, "epoch": 0.62993006993007, "percentage": 20.99, "elapsed_time": "2:14:35", "remaining_time": "8:26:32"} +{"current_steps": 1127, "total_steps": 5364, "loss": 0.3940296173095703, "lr": 1.9274193521397758e-05, "epoch": 0.6304895104895105, "percentage": 21.01, "elapsed_time": "2:14:41", "remaining_time": "8:26:24"} +{"current_steps": 1128, "total_steps": 5364, "loss": 0.2546404004096985, "lr": 1.927175727414532e-05, "epoch": 0.6310489510489511, "percentage": 21.03, "elapsed_time": "2:14:48", "remaining_time": "8:26:15"} +{"current_steps": 1129, "total_steps": 5364, "loss": 0.3997475802898407, "lr": 1.926931709947434e-05, "epoch": 0.6316083916083917, "percentage": 21.05, "elapsed_time": "2:14:55", "remaining_time": "8:26:07"} +{"current_steps": 1130, "total_steps": 5364, "loss": 0.17789559066295624, "lr": 1.9266872998418454e-05, "epoch": 0.6321678321678321, "percentage": 21.07, "elapsed_time": "2:15:02", "remaining_time": "8:26:00"} +{"current_steps": 1131, "total_steps": 5364, "loss": 0.16646352410316467, "lr": 1.926442497201295e-05, "epoch": 0.6327272727272727, "percentage": 21.09, "elapsed_time": "2:15:09", "remaining_time": "8:25:51"} +{"current_steps": 1132, "total_steps": 5364, "loss": 0.17646440863609314, "lr": 1.9261973021294797e-05, "epoch": 0.6332867132867133, "percentage": 21.1, "elapsed_time": "2:15:16", "remaining_time": "8:25:44"} +{"current_steps": 1133, "total_steps": 5364, "loss": 0.11284934729337692, "lr": 1.92595171473026e-05, "epoch": 0.6338461538461538, "percentage": 21.12, "elapsed_time": "2:15:23", "remaining_time": "8:25:36"} +{"current_steps": 1134, "total_steps": 5364, "loss": 0.34719952940940857, "lr": 1.925705735107666e-05, "epoch": 0.6344055944055944, "percentage": 21.14, "elapsed_time": "2:15:30", "remaining_time": "8:25:29"} +{"current_steps": 1135, "total_steps": 5364, "loss": 0.21634995937347412, "lr": 1.92545936336589e-05, "epoch": 0.634965034965035, "percentage": 21.16, "elapsed_time": "2:15:38", "remaining_time": "8:25:22"} +{"current_steps": 1136, "total_steps": 5364, "loss": 0.17749923467636108, "lr": 1.9252125996092945e-05, "epoch": 0.6355244755244756, "percentage": 21.18, "elapsed_time": "2:15:45", "remaining_time": "8:25:14"} +{"current_steps": 1137, "total_steps": 5364, "loss": 0.19224539399147034, "lr": 1.9249654439424044e-05, "epoch": 0.6360839160839161, "percentage": 21.2, "elapsed_time": "2:15:51", "remaining_time": "8:25:06"} +{"current_steps": 1138, "total_steps": 5364, "loss": 0.15738870203495026, "lr": 1.9247178964699133e-05, "epoch": 0.6366433566433567, "percentage": 21.22, "elapsed_time": "2:15:58", "remaining_time": "8:24:58"} +{"current_steps": 1139, "total_steps": 5364, "loss": 0.32128435373306274, "lr": 1.924469957296679e-05, "epoch": 0.6372027972027972, "percentage": 21.23, "elapsed_time": "2:16:05", "remaining_time": "8:24:50"} +{"current_steps": 1140, "total_steps": 5364, "loss": 0.47951018810272217, "lr": 1.9242216265277265e-05, "epoch": 0.6377622377622377, "percentage": 21.25, "elapsed_time": "2:16:12", "remaining_time": "8:24:42"} +{"current_steps": 1141, "total_steps": 5364, "loss": 0.3300664722919464, "lr": 1.923972904268246e-05, "epoch": 0.6383216783216783, "percentage": 21.27, "elapsed_time": "2:16:19", "remaining_time": "8:24:34"} +{"current_steps": 1142, "total_steps": 5364, "loss": 0.17956653237342834, "lr": 1.9237237906235933e-05, "epoch": 0.6388811188811189, "percentage": 21.29, "elapsed_time": "2:16:26", "remaining_time": "8:24:26"} +{"current_steps": 1143, "total_steps": 5364, "loss": 0.13673678040504456, "lr": 1.92347428569929e-05, "epoch": 0.6394405594405594, "percentage": 21.31, "elapsed_time": "2:16:34", "remaining_time": "8:24:20"} +{"current_steps": 1144, "total_steps": 5364, "loss": 0.22357279062271118, "lr": 1.923224389601025e-05, "epoch": 0.64, "percentage": 21.33, "elapsed_time": "2:16:41", "remaining_time": "8:24:12"} +{"current_steps": 1145, "total_steps": 5364, "loss": 0.18866273760795593, "lr": 1.92297410243465e-05, "epoch": 0.6405594405594406, "percentage": 21.35, "elapsed_time": "2:16:48", "remaining_time": "8:24:04"} +{"current_steps": 1146, "total_steps": 5364, "loss": 0.3767423629760742, "lr": 1.9227234243061855e-05, "epoch": 0.6411188811188812, "percentage": 21.36, "elapsed_time": "2:16:54", "remaining_time": "8:23:56"} +{"current_steps": 1147, "total_steps": 5364, "loss": 0.2464519441127777, "lr": 1.9224723553218152e-05, "epoch": 0.6416783216783217, "percentage": 21.38, "elapsed_time": "2:17:02", "remaining_time": "8:23:49"} +{"current_steps": 1148, "total_steps": 5364, "loss": 0.12272395193576813, "lr": 1.9222208955878896e-05, "epoch": 0.6422377622377622, "percentage": 21.4, "elapsed_time": "2:17:09", "remaining_time": "8:23:41"} +{"current_steps": 1149, "total_steps": 5364, "loss": 0.09784625470638275, "lr": 1.9219690452109244e-05, "epoch": 0.6427972027972028, "percentage": 21.42, "elapsed_time": "2:17:15", "remaining_time": "8:23:32"} +{"current_steps": 1150, "total_steps": 5364, "loss": 0.08572151511907578, "lr": 1.9217168042976008e-05, "epoch": 0.6433566433566433, "percentage": 21.44, "elapsed_time": "2:17:23", "remaining_time": "8:23:25"} +{"current_steps": 1151, "total_steps": 5364, "loss": 0.18448996543884277, "lr": 1.9214641729547647e-05, "epoch": 0.6439160839160839, "percentage": 21.46, "elapsed_time": "2:17:30", "remaining_time": "8:23:17"} +{"current_steps": 1152, "total_steps": 5364, "loss": 0.15158987045288086, "lr": 1.921211151289429e-05, "epoch": 0.6444755244755245, "percentage": 21.48, "elapsed_time": "2:17:37", "remaining_time": "8:23:11"} +{"current_steps": 1153, "total_steps": 5364, "loss": 0.1011272668838501, "lr": 1.920957739408771e-05, "epoch": 0.645034965034965, "percentage": 21.5, "elapsed_time": "2:17:44", "remaining_time": "8:23:03"} +{"current_steps": 1154, "total_steps": 5364, "loss": 0.2921428382396698, "lr": 1.920703937420133e-05, "epoch": 0.6455944055944056, "percentage": 21.51, "elapsed_time": "2:17:51", "remaining_time": "8:22:55"} +{"current_steps": 1155, "total_steps": 5364, "loss": 0.48127010464668274, "lr": 1.920449745431022e-05, "epoch": 0.6461538461538462, "percentage": 21.53, "elapsed_time": "2:17:58", "remaining_time": "8:22:48"} +{"current_steps": 1156, "total_steps": 5364, "loss": 0.225082665681839, "lr": 1.9201951635491123e-05, "epoch": 0.6467132867132868, "percentage": 21.55, "elapsed_time": "2:18:05", "remaining_time": "8:22:40"} +{"current_steps": 1157, "total_steps": 5364, "loss": 0.2524714171886444, "lr": 1.919940191882242e-05, "epoch": 0.6472727272727272, "percentage": 21.57, "elapsed_time": "2:18:12", "remaining_time": "8:22:32"} +{"current_steps": 1158, "total_steps": 5364, "loss": 0.24488875269889832, "lr": 1.9196848305384136e-05, "epoch": 0.6478321678321678, "percentage": 21.59, "elapsed_time": "2:18:19", "remaining_time": "8:22:24"} +{"current_steps": 1159, "total_steps": 5364, "loss": 0.29227587580680847, "lr": 1.9194290796257963e-05, "epoch": 0.6483916083916084, "percentage": 21.61, "elapsed_time": "2:18:26", "remaining_time": "8:22:16"} +{"current_steps": 1160, "total_steps": 5364, "loss": 0.2693462371826172, "lr": 1.919172939252723e-05, "epoch": 0.6489510489510489, "percentage": 21.63, "elapsed_time": "2:18:33", "remaining_time": "8:22:10"} +{"current_steps": 1161, "total_steps": 5364, "loss": 0.18121157586574554, "lr": 1.9189164095276916e-05, "epoch": 0.6495104895104895, "percentage": 21.64, "elapsed_time": "2:18:41", "remaining_time": "8:22:03"} +{"current_steps": 1162, "total_steps": 5364, "loss": 0.18202300369739532, "lr": 1.9186594905593665e-05, "epoch": 0.6500699300699301, "percentage": 21.66, "elapsed_time": "2:18:48", "remaining_time": "8:21:58"} +{"current_steps": 1163, "total_steps": 5364, "loss": 0.29729363322257996, "lr": 1.9184021824565753e-05, "epoch": 0.6506293706293707, "percentage": 21.68, "elapsed_time": "2:18:55", "remaining_time": "8:21:51"} +{"current_steps": 1164, "total_steps": 5364, "loss": 0.3485714793205261, "lr": 1.918144485328311e-05, "epoch": 0.6511888111888112, "percentage": 21.7, "elapsed_time": "2:19:02", "remaining_time": "8:21:42"} +{"current_steps": 1165, "total_steps": 5364, "loss": 0.248091921210289, "lr": 1.917886399283731e-05, "epoch": 0.6517482517482518, "percentage": 21.72, "elapsed_time": "2:19:09", "remaining_time": "8:21:34"} +{"current_steps": 1166, "total_steps": 5364, "loss": 0.2711426019668579, "lr": 1.917627924432159e-05, "epoch": 0.6523076923076923, "percentage": 21.74, "elapsed_time": "2:19:16", "remaining_time": "8:21:27"} +{"current_steps": 1167, "total_steps": 5364, "loss": 0.10632821917533875, "lr": 1.917369060883081e-05, "epoch": 0.6528671328671328, "percentage": 21.76, "elapsed_time": "2:19:23", "remaining_time": "8:21:18"} +{"current_steps": 1168, "total_steps": 5364, "loss": 0.11299716681241989, "lr": 1.9171098087461496e-05, "epoch": 0.6534265734265734, "percentage": 21.77, "elapsed_time": "2:19:30", "remaining_time": "8:21:11"} +{"current_steps": 1169, "total_steps": 5364, "loss": 0.39953672885894775, "lr": 1.9168501681311806e-05, "epoch": 0.653986013986014, "percentage": 21.79, "elapsed_time": "2:19:37", "remaining_time": "8:21:03"} +{"current_steps": 1170, "total_steps": 5364, "loss": 0.26359349489212036, "lr": 1.9165901391481557e-05, "epoch": 0.6545454545454545, "percentage": 21.81, "elapsed_time": "2:19:44", "remaining_time": "8:20:55"} +{"current_steps": 1171, "total_steps": 5364, "loss": 0.07769016176462173, "lr": 1.9163297219072195e-05, "epoch": 0.6551048951048951, "percentage": 21.83, "elapsed_time": "2:19:51", "remaining_time": "8:20:47"} +{"current_steps": 1172, "total_steps": 5364, "loss": 0.18297234177589417, "lr": 1.9160689165186835e-05, "epoch": 0.6556643356643357, "percentage": 21.85, "elapsed_time": "2:19:58", "remaining_time": "8:20:39"} +{"current_steps": 1173, "total_steps": 5364, "loss": 0.1556541621685028, "lr": 1.9158077230930207e-05, "epoch": 0.6562237762237763, "percentage": 21.87, "elapsed_time": "2:20:05", "remaining_time": "8:20:31"} +{"current_steps": 1174, "total_steps": 5364, "loss": 0.2870713174343109, "lr": 1.915546141740871e-05, "epoch": 0.6567832167832168, "percentage": 21.89, "elapsed_time": "2:20:12", "remaining_time": "8:20:25"} +{"current_steps": 1175, "total_steps": 5364, "loss": 0.29671698808670044, "lr": 1.915284172573036e-05, "epoch": 0.6573426573426573, "percentage": 21.91, "elapsed_time": "2:20:20", "remaining_time": "8:20:19"} +{"current_steps": 1176, "total_steps": 5364, "loss": 0.26493242383003235, "lr": 1.915021815700485e-05, "epoch": 0.6579020979020979, "percentage": 21.92, "elapsed_time": "2:20:27", "remaining_time": "8:20:11"} +{"current_steps": 1177, "total_steps": 5364, "loss": 0.22414883971214294, "lr": 1.914759071234348e-05, "epoch": 0.6584615384615384, "percentage": 21.94, "elapsed_time": "2:20:34", "remaining_time": "8:20:04"} +{"current_steps": 1178, "total_steps": 5364, "loss": 0.1610300987958908, "lr": 1.9144959392859216e-05, "epoch": 0.659020979020979, "percentage": 21.96, "elapsed_time": "2:20:41", "remaining_time": "8:19:57"} +{"current_steps": 1179, "total_steps": 5364, "loss": 0.14239132404327393, "lr": 1.9142324199666652e-05, "epoch": 0.6595804195804196, "percentage": 21.98, "elapsed_time": "2:20:49", "remaining_time": "8:19:51"} +{"current_steps": 1180, "total_steps": 5364, "loss": 0.31461817026138306, "lr": 1.9139685133882036e-05, "epoch": 0.6601398601398601, "percentage": 22.0, "elapsed_time": "2:20:55", "remaining_time": "8:19:42"} +{"current_steps": 1181, "total_steps": 5364, "loss": 0.20869994163513184, "lr": 1.913704219662324e-05, "epoch": 0.6606993006993007, "percentage": 22.02, "elapsed_time": "2:21:02", "remaining_time": "8:19:34"} +{"current_steps": 1182, "total_steps": 5364, "loss": 0.2910383641719818, "lr": 1.913439538900979e-05, "epoch": 0.6612587412587413, "percentage": 22.04, "elapsed_time": "2:21:09", "remaining_time": "8:19:26"} +{"current_steps": 1183, "total_steps": 5364, "loss": 0.08348032087087631, "lr": 1.913174471216284e-05, "epoch": 0.6618181818181819, "percentage": 22.05, "elapsed_time": "2:21:16", "remaining_time": "8:19:18"} +{"current_steps": 1184, "total_steps": 5364, "loss": 0.30902546644210815, "lr": 1.9129090167205197e-05, "epoch": 0.6623776223776224, "percentage": 22.07, "elapsed_time": "2:21:23", "remaining_time": "8:19:10"} +{"current_steps": 1185, "total_steps": 5364, "loss": 0.14539474248886108, "lr": 1.912643175526129e-05, "epoch": 0.6629370629370629, "percentage": 22.09, "elapsed_time": "2:21:31", "remaining_time": "8:19:05"} +{"current_steps": 1186, "total_steps": 5364, "loss": 0.2426593154668808, "lr": 1.91237694774572e-05, "epoch": 0.6634965034965035, "percentage": 22.11, "elapsed_time": "2:21:38", "remaining_time": "8:18:57"} +{"current_steps": 1187, "total_steps": 5364, "loss": 0.2552768588066101, "lr": 1.9121103334920636e-05, "epoch": 0.664055944055944, "percentage": 22.13, "elapsed_time": "2:21:45", "remaining_time": "8:18:50"} +{"current_steps": 1188, "total_steps": 5364, "loss": 0.2865668535232544, "lr": 1.911843332878095e-05, "epoch": 0.6646153846153846, "percentage": 22.15, "elapsed_time": "2:21:52", "remaining_time": "8:18:43"} +{"current_steps": 1189, "total_steps": 5364, "loss": 0.2804175913333893, "lr": 1.911575946016913e-05, "epoch": 0.6651748251748252, "percentage": 22.17, "elapsed_time": "2:21:59", "remaining_time": "8:18:35"} +{"current_steps": 1190, "total_steps": 5364, "loss": 0.10971076786518097, "lr": 1.9113081730217793e-05, "epoch": 0.6657342657342658, "percentage": 22.18, "elapsed_time": "2:22:06", "remaining_time": "8:18:28"} +{"current_steps": 1191, "total_steps": 5364, "loss": 0.1628573089838028, "lr": 1.9110400140061203e-05, "epoch": 0.6662937062937063, "percentage": 22.2, "elapsed_time": "2:22:13", "remaining_time": "8:18:20"} +{"current_steps": 1192, "total_steps": 5364, "loss": 0.05909005552530289, "lr": 1.9107714690835252e-05, "epoch": 0.6668531468531469, "percentage": 22.22, "elapsed_time": "2:22:21", "remaining_time": "8:18:13"} +{"current_steps": 1193, "total_steps": 5364, "loss": 0.09491445869207382, "lr": 1.9105025383677464e-05, "epoch": 0.6674125874125875, "percentage": 22.24, "elapsed_time": "2:22:28", "remaining_time": "8:18:06"} +{"current_steps": 1194, "total_steps": 5364, "loss": 0.11123067140579224, "lr": 1.9102332219727003e-05, "epoch": 0.6679720279720279, "percentage": 22.26, "elapsed_time": "2:22:35", "remaining_time": "8:17:59"} +{"current_steps": 1195, "total_steps": 5364, "loss": 0.13746190071105957, "lr": 1.9099635200124668e-05, "epoch": 0.6685314685314685, "percentage": 22.28, "elapsed_time": "2:22:42", "remaining_time": "8:17:52"} +{"current_steps": 1196, "total_steps": 5364, "loss": 0.12544505298137665, "lr": 1.9096934326012885e-05, "epoch": 0.6690909090909091, "percentage": 22.3, "elapsed_time": "2:22:50", "remaining_time": "8:17:46"} +{"current_steps": 1197, "total_steps": 5364, "loss": 0.18139979243278503, "lr": 1.9094229598535717e-05, "epoch": 0.6696503496503496, "percentage": 22.32, "elapsed_time": "2:22:57", "remaining_time": "8:17:38"} +{"current_steps": 1198, "total_steps": 5364, "loss": 0.18799258768558502, "lr": 1.9091521018838856e-05, "epoch": 0.6702097902097902, "percentage": 22.33, "elapsed_time": "2:23:04", "remaining_time": "8:17:31"} +{"current_steps": 1199, "total_steps": 5364, "loss": 0.22460326552391052, "lr": 1.9088808588069633e-05, "epoch": 0.6707692307692308, "percentage": 22.35, "elapsed_time": "2:23:11", "remaining_time": "8:17:22"} +{"current_steps": 1200, "total_steps": 5364, "loss": 0.356277197599411, "lr": 1.9086092307377e-05, "epoch": 0.6713286713286714, "percentage": 22.37, "elapsed_time": "2:23:17", "remaining_time": "8:17:14"} +{"current_steps": 1201, "total_steps": 5364, "loss": 0.2535674571990967, "lr": 1.9083372177911543e-05, "epoch": 0.6718881118881119, "percentage": 22.39, "elapsed_time": "2:23:24", "remaining_time": "8:17:05"} +{"current_steps": 1202, "total_steps": 5364, "loss": 0.20889690518379211, "lr": 1.9080648200825487e-05, "epoch": 0.6724475524475525, "percentage": 22.41, "elapsed_time": "2:23:31", "remaining_time": "8:16:59"} +{"current_steps": 1203, "total_steps": 5364, "loss": 0.16600444912910461, "lr": 1.9077920377272673e-05, "epoch": 0.673006993006993, "percentage": 22.43, "elapsed_time": "2:23:38", "remaining_time": "8:16:51"} +{"current_steps": 1204, "total_steps": 5364, "loss": 0.2649494707584381, "lr": 1.9075188708408586e-05, "epoch": 0.6735664335664335, "percentage": 22.45, "elapsed_time": "2:23:46", "remaining_time": "8:16:44"} +{"current_steps": 1205, "total_steps": 5364, "loss": 0.09610937535762787, "lr": 1.9072453195390327e-05, "epoch": 0.6741258741258741, "percentage": 22.46, "elapsed_time": "2:23:52", "remaining_time": "8:16:36"} +{"current_steps": 1206, "total_steps": 5364, "loss": 0.15152233839035034, "lr": 1.906971383937663e-05, "epoch": 0.6746853146853147, "percentage": 22.48, "elapsed_time": "2:24:00", "remaining_time": "8:16:29"} +{"current_steps": 1207, "total_steps": 5364, "loss": 0.27731889486312866, "lr": 1.9066970641527857e-05, "epoch": 0.6752447552447552, "percentage": 22.5, "elapsed_time": "2:24:07", "remaining_time": "8:16:23"} +{"current_steps": 1208, "total_steps": 5364, "loss": 0.3000246286392212, "lr": 1.9064223603006e-05, "epoch": 0.6758041958041958, "percentage": 22.52, "elapsed_time": "2:24:14", "remaining_time": "8:16:15"} +{"current_steps": 1209, "total_steps": 5364, "loss": 0.145631343126297, "lr": 1.9061472724974677e-05, "epoch": 0.6763636363636364, "percentage": 22.54, "elapsed_time": "2:24:21", "remaining_time": "8:16:07"} +{"current_steps": 1210, "total_steps": 5364, "loss": 0.257687509059906, "lr": 1.9058718008599126e-05, "epoch": 0.676923076923077, "percentage": 22.56, "elapsed_time": "2:24:29", "remaining_time": "8:16:03"} +{"current_steps": 1211, "total_steps": 5364, "loss": 0.08178170025348663, "lr": 1.9055959455046218e-05, "epoch": 0.6774825174825175, "percentage": 22.58, "elapsed_time": "2:24:36", "remaining_time": "8:15:55"} +{"current_steps": 1212, "total_steps": 5364, "loss": 0.11994513124227524, "lr": 1.9053197065484447e-05, "epoch": 0.678041958041958, "percentage": 22.6, "elapsed_time": "2:24:43", "remaining_time": "8:15:47"} +{"current_steps": 1213, "total_steps": 5364, "loss": 0.0022237920202314854, "lr": 1.905043084108394e-05, "epoch": 0.6786013986013986, "percentage": 22.61, "elapsed_time": "2:24:50", "remaining_time": "8:15:40"} +{"current_steps": 1214, "total_steps": 5364, "loss": 0.44033315777778625, "lr": 1.9047660783016424e-05, "epoch": 0.6791608391608391, "percentage": 22.63, "elapsed_time": "2:24:58", "remaining_time": "8:15:33"} +{"current_steps": 1215, "total_steps": 5364, "loss": 0.09359294921159744, "lr": 1.904488689245528e-05, "epoch": 0.6797202797202797, "percentage": 22.65, "elapsed_time": "2:25:04", "remaining_time": "8:15:25"} +{"current_steps": 1216, "total_steps": 5364, "loss": 0.14086344838142395, "lr": 1.904210917057549e-05, "epoch": 0.6802797202797203, "percentage": 22.67, "elapsed_time": "2:25:12", "remaining_time": "8:15:18"} +{"current_steps": 1217, "total_steps": 5364, "loss": 0.05836724489927292, "lr": 1.9039327618553677e-05, "epoch": 0.6808391608391609, "percentage": 22.69, "elapsed_time": "2:25:19", "remaining_time": "8:15:11"} +{"current_steps": 1218, "total_steps": 5364, "loss": 0.2508912980556488, "lr": 1.903654223756807e-05, "epoch": 0.6813986013986014, "percentage": 22.71, "elapsed_time": "2:25:26", "remaining_time": "8:15:03"} +{"current_steps": 1219, "total_steps": 5364, "loss": 0.22959785163402557, "lr": 1.903375302879853e-05, "epoch": 0.681958041958042, "percentage": 22.73, "elapsed_time": "2:25:33", "remaining_time": "8:14:56"} +{"current_steps": 1220, "total_steps": 5364, "loss": 0.31373026967048645, "lr": 1.903095999342653e-05, "epoch": 0.6825174825174826, "percentage": 22.74, "elapsed_time": "2:25:40", "remaining_time": "8:14:50"} +{"current_steps": 1221, "total_steps": 5364, "loss": 0.25159138441085815, "lr": 1.9028163132635182e-05, "epoch": 0.683076923076923, "percentage": 22.76, "elapsed_time": "2:25:48", "remaining_time": "8:14:43"} +{"current_steps": 1222, "total_steps": 5364, "loss": 0.07931622862815857, "lr": 1.90253624476092e-05, "epoch": 0.6836363636363636, "percentage": 22.78, "elapsed_time": "2:25:55", "remaining_time": "8:14:38"} +{"current_steps": 1223, "total_steps": 5364, "loss": 0.18305863440036774, "lr": 1.9022557939534924e-05, "epoch": 0.6841958041958042, "percentage": 22.8, "elapsed_time": "2:26:02", "remaining_time": "8:14:30"} +{"current_steps": 1224, "total_steps": 5364, "loss": 0.24432870745658875, "lr": 1.9019749609600313e-05, "epoch": 0.6847552447552447, "percentage": 22.82, "elapsed_time": "2:26:10", "remaining_time": "8:14:24"} +{"current_steps": 1225, "total_steps": 5364, "loss": 0.12469576299190521, "lr": 1.9016937458994952e-05, "epoch": 0.6853146853146853, "percentage": 22.84, "elapsed_time": "2:26:17", "remaining_time": "8:14:17"} +{"current_steps": 1226, "total_steps": 5364, "loss": 0.2719348967075348, "lr": 1.9014121488910028e-05, "epoch": 0.6858741258741259, "percentage": 22.86, "elapsed_time": "2:26:24", "remaining_time": "8:14:09"} +{"current_steps": 1227, "total_steps": 5364, "loss": 0.27376121282577515, "lr": 1.9011301700538367e-05, "epoch": 0.6864335664335665, "percentage": 22.87, "elapsed_time": "2:26:31", "remaining_time": "8:14:01"} +{"current_steps": 1228, "total_steps": 5364, "loss": 0.3813643455505371, "lr": 1.9008478095074396e-05, "epoch": 0.686993006993007, "percentage": 22.89, "elapsed_time": "2:26:38", "remaining_time": "8:13:53"} +{"current_steps": 1229, "total_steps": 5364, "loss": 0.21032418310642242, "lr": 1.9005650673714163e-05, "epoch": 0.6875524475524476, "percentage": 22.91, "elapsed_time": "2:26:45", "remaining_time": "8:13:46"} +{"current_steps": 1230, "total_steps": 5364, "loss": 0.1845524162054062, "lr": 1.9002819437655343e-05, "epoch": 0.6881118881118881, "percentage": 22.93, "elapsed_time": "2:26:52", "remaining_time": "8:13:38"} +{"current_steps": 1231, "total_steps": 5364, "loss": 0.048418037593364716, "lr": 1.899998438809721e-05, "epoch": 0.6886713286713286, "percentage": 22.95, "elapsed_time": "2:27:00", "remaining_time": "8:13:32"} +{"current_steps": 1232, "total_steps": 5364, "loss": 0.13670310378074646, "lr": 1.8997145526240656e-05, "epoch": 0.6892307692307692, "percentage": 22.97, "elapsed_time": "2:27:07", "remaining_time": "8:13:25"} +{"current_steps": 1233, "total_steps": 5364, "loss": 0.14143714308738708, "lr": 1.8994302853288205e-05, "epoch": 0.6897902097902098, "percentage": 22.99, "elapsed_time": "2:27:14", "remaining_time": "8:13:17"} +{"current_steps": 1234, "total_steps": 5364, "loss": 0.03789675235748291, "lr": 1.8991456370443973e-05, "epoch": 0.6903496503496503, "percentage": 23.01, "elapsed_time": "2:27:21", "remaining_time": "8:13:09"} +{"current_steps": 1235, "total_steps": 5364, "loss": 0.152780219912529, "lr": 1.898860607891371e-05, "epoch": 0.6909090909090909, "percentage": 23.02, "elapsed_time": "2:27:28", "remaining_time": "8:13:01"} +{"current_steps": 1236, "total_steps": 5364, "loss": 0.1169387623667717, "lr": 1.898575197990476e-05, "epoch": 0.6914685314685315, "percentage": 23.04, "elapsed_time": "2:27:34", "remaining_time": "8:12:53"} +{"current_steps": 1237, "total_steps": 5364, "loss": 0.15557576715946198, "lr": 1.8982894074626094e-05, "epoch": 0.6920279720279721, "percentage": 23.06, "elapsed_time": "2:27:42", "remaining_time": "8:12:47"} +{"current_steps": 1238, "total_steps": 5364, "loss": 0.14882495999336243, "lr": 1.8980032364288286e-05, "epoch": 0.6925874125874126, "percentage": 23.08, "elapsed_time": "2:27:49", "remaining_time": "8:12:39"} +{"current_steps": 1239, "total_steps": 5364, "loss": 0.2935725450515747, "lr": 1.8977166850103535e-05, "epoch": 0.6931468531468531, "percentage": 23.1, "elapsed_time": "2:27:56", "remaining_time": "8:12:31"} +{"current_steps": 1240, "total_steps": 5364, "loss": 0.25928008556365967, "lr": 1.8974297533285636e-05, "epoch": 0.6937062937062937, "percentage": 23.12, "elapsed_time": "2:28:02", "remaining_time": "8:12:23"} +{"current_steps": 1241, "total_steps": 5364, "loss": 0.2947520613670349, "lr": 1.8971424415050005e-05, "epoch": 0.6942657342657342, "percentage": 23.14, "elapsed_time": "2:28:09", "remaining_time": "8:12:14"} +{"current_steps": 1242, "total_steps": 5364, "loss": 0.13924340903759003, "lr": 1.8968547496613657e-05, "epoch": 0.6948251748251748, "percentage": 23.15, "elapsed_time": "2:28:16", "remaining_time": "8:12:07"} +{"current_steps": 1243, "total_steps": 5364, "loss": 0.22265732288360596, "lr": 1.8965666779195235e-05, "epoch": 0.6953846153846154, "percentage": 23.17, "elapsed_time": "2:28:24", "remaining_time": "8:12:00"} +{"current_steps": 1244, "total_steps": 5364, "loss": 0.3004734516143799, "lr": 1.896278226401497e-05, "epoch": 0.695944055944056, "percentage": 23.19, "elapsed_time": "2:28:31", "remaining_time": "8:11:52"} +{"current_steps": 1245, "total_steps": 5364, "loss": 0.2936972975730896, "lr": 1.895989395229472e-05, "epoch": 0.6965034965034965, "percentage": 23.21, "elapsed_time": "2:28:38", "remaining_time": "8:11:44"} +{"current_steps": 1246, "total_steps": 5364, "loss": 0.1826832890510559, "lr": 1.8957001845257947e-05, "epoch": 0.6970629370629371, "percentage": 23.23, "elapsed_time": "2:28:44", "remaining_time": "8:11:36"} +{"current_steps": 1247, "total_steps": 5364, "loss": 0.15641947090625763, "lr": 1.8954105944129702e-05, "epoch": 0.6976223776223777, "percentage": 23.25, "elapsed_time": "2:28:51", "remaining_time": "8:11:28"} +{"current_steps": 1248, "total_steps": 5364, "loss": 0.22361494600772858, "lr": 1.895120625013667e-05, "epoch": 0.6981818181818182, "percentage": 23.27, "elapsed_time": "2:28:59", "remaining_time": "8:11:21"} +{"current_steps": 1249, "total_steps": 5364, "loss": 0.08146579563617706, "lr": 1.894830276450713e-05, "epoch": 0.6987412587412587, "percentage": 23.28, "elapsed_time": "2:29:06", "remaining_time": "8:11:15"} +{"current_steps": 1250, "total_steps": 5364, "loss": 0.16098377108573914, "lr": 1.894539548847096e-05, "epoch": 0.6993006993006993, "percentage": 23.3, "elapsed_time": "2:29:13", "remaining_time": "8:11:07"} +{"current_steps": 1251, "total_steps": 5364, "loss": 0.04757336527109146, "lr": 1.8942484423259663e-05, "epoch": 0.6998601398601398, "percentage": 23.32, "elapsed_time": "2:29:21", "remaining_time": "8:11:02"} +{"current_steps": 1252, "total_steps": 5364, "loss": 0.23201589286327362, "lr": 1.8939569570106326e-05, "epoch": 0.7004195804195804, "percentage": 23.34, "elapsed_time": "2:29:28", "remaining_time": "8:10:55"} +{"current_steps": 1253, "total_steps": 5364, "loss": 0.09793180972337723, "lr": 1.8936650930245655e-05, "epoch": 0.700979020979021, "percentage": 23.36, "elapsed_time": "2:29:36", "remaining_time": "8:10:49"} +{"current_steps": 1254, "total_steps": 5364, "loss": 0.15101146697998047, "lr": 1.8933728504913953e-05, "epoch": 0.7015384615384616, "percentage": 23.38, "elapsed_time": "2:29:43", "remaining_time": "8:10:42"} +{"current_steps": 1255, "total_steps": 5364, "loss": 0.25147831439971924, "lr": 1.893080229534913e-05, "epoch": 0.7020979020979021, "percentage": 23.4, "elapsed_time": "2:29:50", "remaining_time": "8:10:36"} +{"current_steps": 1256, "total_steps": 5364, "loss": 0.29706037044525146, "lr": 1.8927872302790696e-05, "epoch": 0.7026573426573427, "percentage": 23.42, "elapsed_time": "2:29:57", "remaining_time": "8:10:28"} +{"current_steps": 1257, "total_steps": 5364, "loss": 0.20549073815345764, "lr": 1.8924938528479766e-05, "epoch": 0.7032167832167833, "percentage": 23.43, "elapsed_time": "2:30:04", "remaining_time": "8:10:20"} +{"current_steps": 1258, "total_steps": 5364, "loss": 0.14198213815689087, "lr": 1.8922000973659056e-05, "epoch": 0.7037762237762237, "percentage": 23.45, "elapsed_time": "2:30:11", "remaining_time": "8:10:14"} +{"current_steps": 1259, "total_steps": 5364, "loss": 0.23146186769008636, "lr": 1.8919059639572882e-05, "epoch": 0.7043356643356643, "percentage": 23.47, "elapsed_time": "2:30:19", "remaining_time": "8:10:07"} +{"current_steps": 1260, "total_steps": 5364, "loss": 0.3991163372993469, "lr": 1.891611452746716e-05, "epoch": 0.7048951048951049, "percentage": 23.49, "elapsed_time": "2:30:26", "remaining_time": "8:09:59"} +{"current_steps": 1261, "total_steps": 5364, "loss": 0.19554615020751953, "lr": 1.8913165638589416e-05, "epoch": 0.7054545454545454, "percentage": 23.51, "elapsed_time": "2:30:32", "remaining_time": "8:09:51"} +{"current_steps": 1262, "total_steps": 5364, "loss": 0.29548728466033936, "lr": 1.891021297418876e-05, "epoch": 0.706013986013986, "percentage": 23.53, "elapsed_time": "2:30:39", "remaining_time": "8:09:43"} +{"current_steps": 1263, "total_steps": 5364, "loss": 0.2255270779132843, "lr": 1.8907256535515914e-05, "epoch": 0.7065734265734266, "percentage": 23.55, "elapsed_time": "2:30:46", "remaining_time": "8:09:35"} +{"current_steps": 1264, "total_steps": 5364, "loss": 0.37839800119400024, "lr": 1.8904296323823193e-05, "epoch": 0.7071328671328672, "percentage": 23.56, "elapsed_time": "2:30:54", "remaining_time": "8:09:28"} +{"current_steps": 1265, "total_steps": 5364, "loss": 0.27291709184646606, "lr": 1.8901332340364512e-05, "epoch": 0.7076923076923077, "percentage": 23.58, "elapsed_time": "2:31:00", "remaining_time": "8:09:20"} +{"current_steps": 1266, "total_steps": 5364, "loss": 0.3188212215900421, "lr": 1.8898364586395383e-05, "epoch": 0.7082517482517483, "percentage": 23.6, "elapsed_time": "2:31:08", "remaining_time": "8:09:12"} +{"current_steps": 1267, "total_steps": 5364, "loss": 0.25226080417633057, "lr": 1.8895393063172916e-05, "epoch": 0.7088111888111888, "percentage": 23.62, "elapsed_time": "2:31:15", "remaining_time": "8:09:05"} +{"current_steps": 1268, "total_steps": 5364, "loss": 0.154170960187912, "lr": 1.8892417771955815e-05, "epoch": 0.7093706293706293, "percentage": 23.64, "elapsed_time": "2:31:22", "remaining_time": "8:08:58"} +{"current_steps": 1269, "total_steps": 5364, "loss": 0.22951939702033997, "lr": 1.888943871400438e-05, "epoch": 0.7099300699300699, "percentage": 23.66, "elapsed_time": "2:31:29", "remaining_time": "8:08:50"} +{"current_steps": 1270, "total_steps": 5364, "loss": 0.1874401867389679, "lr": 1.8886455890580513e-05, "epoch": 0.7104895104895105, "percentage": 23.68, "elapsed_time": "2:31:36", "remaining_time": "8:08:44"} +{"current_steps": 1271, "total_steps": 5364, "loss": 0.4291166067123413, "lr": 1.8883469302947705e-05, "epoch": 0.711048951048951, "percentage": 23.7, "elapsed_time": "2:31:43", "remaining_time": "8:08:36"} +{"current_steps": 1272, "total_steps": 5364, "loss": 0.01633700728416443, "lr": 1.8880478952371042e-05, "epoch": 0.7116083916083916, "percentage": 23.71, "elapsed_time": "2:31:50", "remaining_time": "8:08:28"} +{"current_steps": 1273, "total_steps": 5364, "loss": 0.2168062925338745, "lr": 1.8877484840117208e-05, "epoch": 0.7121678321678322, "percentage": 23.73, "elapsed_time": "2:31:57", "remaining_time": "8:08:21"} +{"current_steps": 1274, "total_steps": 5364, "loss": 0.31777673959732056, "lr": 1.887448696745447e-05, "epoch": 0.7127272727272728, "percentage": 23.75, "elapsed_time": "2:32:04", "remaining_time": "8:08:13"} +{"current_steps": 1275, "total_steps": 5364, "loss": 0.3538969159126282, "lr": 1.8871485335652705e-05, "epoch": 0.7132867132867133, "percentage": 23.77, "elapsed_time": "2:32:11", "remaining_time": "8:08:05"} +{"current_steps": 1276, "total_steps": 5364, "loss": 0.06914018094539642, "lr": 1.8868479945983367e-05, "epoch": 0.7138461538461538, "percentage": 23.79, "elapsed_time": "2:32:19", "remaining_time": "8:08:00"} +{"current_steps": 1277, "total_steps": 5364, "loss": 0.039529550820589066, "lr": 1.886547079971951e-05, "epoch": 0.7144055944055944, "percentage": 23.81, "elapsed_time": "2:32:26", "remaining_time": "8:07:52"} +{"current_steps": 1278, "total_steps": 5364, "loss": 0.10326577723026276, "lr": 1.886245789813577e-05, "epoch": 0.7149650349650349, "percentage": 23.83, "elapsed_time": "2:32:33", "remaining_time": "8:07:44"} +{"current_steps": 1279, "total_steps": 5364, "loss": 0.25542083382606506, "lr": 1.885944124250839e-05, "epoch": 0.7155244755244755, "percentage": 23.84, "elapsed_time": "2:32:39", "remaining_time": "8:07:36"} +{"current_steps": 1280, "total_steps": 5364, "loss": 0.35049593448638916, "lr": 1.8856420834115182e-05, "epoch": 0.7160839160839161, "percentage": 23.86, "elapsed_time": "2:32:46", "remaining_time": "8:07:27"} +{"current_steps": 1281, "total_steps": 5364, "loss": 0.26107078790664673, "lr": 1.885339667423557e-05, "epoch": 0.7166433566433567, "percentage": 23.88, "elapsed_time": "2:32:54", "remaining_time": "8:07:20"} +{"current_steps": 1282, "total_steps": 5364, "loss": 0.1546381711959839, "lr": 1.8850368764150554e-05, "epoch": 0.7172027972027972, "percentage": 23.9, "elapsed_time": "2:33:00", "remaining_time": "8:07:12"} +{"current_steps": 1283, "total_steps": 5364, "loss": 0.13336928188800812, "lr": 1.884733710514272e-05, "epoch": 0.7177622377622378, "percentage": 23.92, "elapsed_time": "2:33:08", "remaining_time": "8:07:06"} +{"current_steps": 1284, "total_steps": 5364, "loss": 0.38645532727241516, "lr": 1.884430169849625e-05, "epoch": 0.7183216783216784, "percentage": 23.94, "elapsed_time": "2:33:15", "remaining_time": "8:06:59"} +{"current_steps": 1285, "total_steps": 5364, "loss": 0.5598582029342651, "lr": 1.884126254549691e-05, "epoch": 0.7188811188811188, "percentage": 23.96, "elapsed_time": "2:33:22", "remaining_time": "8:06:52"} +{"current_steps": 1286, "total_steps": 5364, "loss": 0.21529540419578552, "lr": 1.8838219647432053e-05, "epoch": 0.7194405594405594, "percentage": 23.97, "elapsed_time": "2:33:29", "remaining_time": "8:06:44"} +{"current_steps": 1287, "total_steps": 5364, "loss": 0.09676319360733032, "lr": 1.8835173005590616e-05, "epoch": 0.72, "percentage": 23.99, "elapsed_time": "2:33:36", "remaining_time": "8:06:37"} +{"current_steps": 1288, "total_steps": 5364, "loss": 0.3112836480140686, "lr": 1.883212262126313e-05, "epoch": 0.7205594405594405, "percentage": 24.01, "elapsed_time": "2:33:43", "remaining_time": "8:06:28"} +{"current_steps": 1289, "total_steps": 5364, "loss": 0.2938607931137085, "lr": 1.88290684957417e-05, "epoch": 0.7211188811188811, "percentage": 24.03, "elapsed_time": "2:33:50", "remaining_time": "8:06:22"} +{"current_steps": 1290, "total_steps": 5364, "loss": 0.230940580368042, "lr": 1.8826010630320025e-05, "epoch": 0.7216783216783217, "percentage": 24.05, "elapsed_time": "2:33:58", "remaining_time": "8:06:16"} +{"current_steps": 1291, "total_steps": 5364, "loss": 0.3394428491592407, "lr": 1.8822949026293383e-05, "epoch": 0.7222377622377623, "percentage": 24.07, "elapsed_time": "2:34:05", "remaining_time": "8:06:08"} +{"current_steps": 1292, "total_steps": 5364, "loss": 0.21149572730064392, "lr": 1.881988368495864e-05, "epoch": 0.7227972027972028, "percentage": 24.09, "elapsed_time": "2:34:12", "remaining_time": "8:06:00"} +{"current_steps": 1293, "total_steps": 5364, "loss": 0.21099887788295746, "lr": 1.8816814607614234e-05, "epoch": 0.7233566433566434, "percentage": 24.11, "elapsed_time": "2:34:19", "remaining_time": "8:05:52"} +{"current_steps": 1294, "total_steps": 5364, "loss": 0.16308756172657013, "lr": 1.8813741795560203e-05, "epoch": 0.7239160839160839, "percentage": 24.12, "elapsed_time": "2:34:26", "remaining_time": "8:05:44"} +{"current_steps": 1295, "total_steps": 5364, "loss": 0.12714090943336487, "lr": 1.8810665250098153e-05, "epoch": 0.7244755244755244, "percentage": 24.14, "elapsed_time": "2:34:33", "remaining_time": "8:05:37"} +{"current_steps": 1296, "total_steps": 5364, "loss": 0.25265365839004517, "lr": 1.880758497253128e-05, "epoch": 0.725034965034965, "percentage": 24.16, "elapsed_time": "2:34:40", "remaining_time": "8:05:30"} +{"current_steps": 1297, "total_steps": 5364, "loss": 0.13695551455020905, "lr": 1.880450096416435e-05, "epoch": 0.7255944055944056, "percentage": 24.18, "elapsed_time": "2:34:47", "remaining_time": "8:05:22"} +{"current_steps": 1298, "total_steps": 5364, "loss": 0.08582916855812073, "lr": 1.8801413226303722e-05, "epoch": 0.7261538461538461, "percentage": 24.2, "elapsed_time": "2:34:54", "remaining_time": "8:05:16"} +{"current_steps": 1299, "total_steps": 5364, "loss": 0.130242258310318, "lr": 1.879832176025733e-05, "epoch": 0.7267132867132867, "percentage": 24.22, "elapsed_time": "2:35:01", "remaining_time": "8:05:08"} +{"current_steps": 1300, "total_steps": 5364, "loss": 0.22602578997612, "lr": 1.8795226567334683e-05, "epoch": 0.7272727272727273, "percentage": 24.24, "elapsed_time": "2:35:09", "remaining_time": "8:05:01"} +{"current_steps": 1301, "total_steps": 5364, "loss": 0.3310128450393677, "lr": 1.879212764884687e-05, "epoch": 0.7278321678321679, "percentage": 24.25, "elapsed_time": "2:35:16", "remaining_time": "8:04:55"} +{"current_steps": 1302, "total_steps": 5364, "loss": 0.17944839596748352, "lr": 1.878902500610657e-05, "epoch": 0.7283916083916084, "percentage": 24.27, "elapsed_time": "2:35:23", "remaining_time": "8:04:47"} +{"current_steps": 1303, "total_steps": 5364, "loss": 0.12377354502677917, "lr": 1.878591864042802e-05, "epoch": 0.7289510489510489, "percentage": 24.29, "elapsed_time": "2:35:30", "remaining_time": "8:04:40"} +{"current_steps": 1304, "total_steps": 5364, "loss": 0.23134580254554749, "lr": 1.8782808553127044e-05, "epoch": 0.7295104895104895, "percentage": 24.31, "elapsed_time": "2:35:37", "remaining_time": "8:04:32"} +{"current_steps": 1305, "total_steps": 5364, "loss": 0.38295817375183105, "lr": 1.8779694745521052e-05, "epoch": 0.73006993006993, "percentage": 24.33, "elapsed_time": "2:35:44", "remaining_time": "8:04:24"} +{"current_steps": 1306, "total_steps": 5364, "loss": 0.37209686636924744, "lr": 1.8776577218929007e-05, "epoch": 0.7306293706293706, "percentage": 24.35, "elapsed_time": "2:35:51", "remaining_time": "8:04:17"} +{"current_steps": 1307, "total_steps": 5364, "loss": 0.19278863072395325, "lr": 1.877345597467147e-05, "epoch": 0.7311888111888112, "percentage": 24.37, "elapsed_time": "2:35:58", "remaining_time": "8:04:09"} +{"current_steps": 1308, "total_steps": 5364, "loss": 0.29413846135139465, "lr": 1.8770331014070563e-05, "epoch": 0.7317482517482518, "percentage": 24.38, "elapsed_time": "2:36:05", "remaining_time": "8:04:01"} +{"current_steps": 1309, "total_steps": 5364, "loss": 0.33733540773391724, "lr": 1.876720233844999e-05, "epoch": 0.7323076923076923, "percentage": 24.4, "elapsed_time": "2:36:12", "remaining_time": "8:03:53"} +{"current_steps": 1310, "total_steps": 5364, "loss": 0.18579035997390747, "lr": 1.8764069949135018e-05, "epoch": 0.7328671328671329, "percentage": 24.42, "elapsed_time": "2:36:19", "remaining_time": "8:03:45"} +{"current_steps": 1311, "total_steps": 5364, "loss": 0.24365222454071045, "lr": 1.87609338474525e-05, "epoch": 0.7334265734265735, "percentage": 24.44, "elapsed_time": "2:36:26", "remaining_time": "8:03:39"} +{"current_steps": 1312, "total_steps": 5364, "loss": 0.046830154955387115, "lr": 1.875779403473085e-05, "epoch": 0.733986013986014, "percentage": 24.46, "elapsed_time": "2:36:33", "remaining_time": "8:03:32"} +{"current_steps": 1313, "total_steps": 5364, "loss": 0.19351214170455933, "lr": 1.8754650512300067e-05, "epoch": 0.7345454545454545, "percentage": 24.48, "elapsed_time": "2:36:41", "remaining_time": "8:03:25"} +{"current_steps": 1314, "total_steps": 5364, "loss": 0.03813890367746353, "lr": 1.8751503281491708e-05, "epoch": 0.7351048951048951, "percentage": 24.5, "elapsed_time": "2:36:48", "remaining_time": "8:03:17"} +{"current_steps": 1315, "total_steps": 5364, "loss": 0.10832241177558899, "lr": 1.8748352343638912e-05, "epoch": 0.7356643356643356, "percentage": 24.52, "elapsed_time": "2:36:55", "remaining_time": "8:03:10"} +{"current_steps": 1316, "total_steps": 5364, "loss": 0.2444397658109665, "lr": 1.8745197700076376e-05, "epoch": 0.7362237762237762, "percentage": 24.53, "elapsed_time": "2:37:02", "remaining_time": "8:03:02"} +{"current_steps": 1317, "total_steps": 5364, "loss": 0.2845626771450043, "lr": 1.874203935214038e-05, "epoch": 0.7367832167832168, "percentage": 24.55, "elapsed_time": "2:37:09", "remaining_time": "8:02:55"} +{"current_steps": 1318, "total_steps": 5364, "loss": 0.286207914352417, "lr": 1.8738877301168766e-05, "epoch": 0.7373426573426574, "percentage": 24.57, "elapsed_time": "2:37:16", "remaining_time": "8:02:47"} +{"current_steps": 1319, "total_steps": 5364, "loss": 0.18193387985229492, "lr": 1.873571154850094e-05, "epoch": 0.7379020979020979, "percentage": 24.59, "elapsed_time": "2:37:22", "remaining_time": "8:02:38"} +{"current_steps": 1320, "total_steps": 5364, "loss": 0.09196147322654724, "lr": 1.8732542095477888e-05, "epoch": 0.7384615384615385, "percentage": 24.61, "elapsed_time": "2:37:29", "remaining_time": "8:02:30"} +{"current_steps": 1321, "total_steps": 5364, "loss": 0.12661802768707275, "lr": 1.8729368943442156e-05, "epoch": 0.7390209790209791, "percentage": 24.63, "elapsed_time": "2:37:36", "remaining_time": "8:02:23"} +{"current_steps": 1322, "total_steps": 5364, "loss": 0.14102940261363983, "lr": 1.872619209373785e-05, "epoch": 0.7395804195804195, "percentage": 24.65, "elapsed_time": "2:37:43", "remaining_time": "8:02:15"} +{"current_steps": 1323, "total_steps": 5364, "loss": 0.1998492330312729, "lr": 1.872301154771066e-05, "epoch": 0.7401398601398601, "percentage": 24.66, "elapsed_time": "2:37:50", "remaining_time": "8:02:07"} +{"current_steps": 1324, "total_steps": 5364, "loss": 0.2623226046562195, "lr": 1.871982730670783e-05, "epoch": 0.7406993006993007, "percentage": 24.68, "elapsed_time": "2:37:57", "remaining_time": "8:01:58"} +{"current_steps": 1325, "total_steps": 5364, "loss": 0.2594923973083496, "lr": 1.871663937207816e-05, "epoch": 0.7412587412587412, "percentage": 24.7, "elapsed_time": "2:38:04", "remaining_time": "8:01:51"} +{"current_steps": 1326, "total_steps": 5364, "loss": 0.3071717619895935, "lr": 1.8713447745172043e-05, "epoch": 0.7418181818181818, "percentage": 24.72, "elapsed_time": "2:38:11", "remaining_time": "8:01:43"} +{"current_steps": 1327, "total_steps": 5364, "loss": 0.2433147132396698, "lr": 1.8710252427341408e-05, "epoch": 0.7423776223776224, "percentage": 24.74, "elapsed_time": "2:38:18", "remaining_time": "8:01:35"} +{"current_steps": 1328, "total_steps": 5364, "loss": 0.2626168131828308, "lr": 1.8707053419939756e-05, "epoch": 0.742937062937063, "percentage": 24.76, "elapsed_time": "2:38:25", "remaining_time": "8:01:27"} +{"current_steps": 1329, "total_steps": 5364, "loss": 0.22473716735839844, "lr": 1.8703850724322158e-05, "epoch": 0.7434965034965035, "percentage": 24.78, "elapsed_time": "2:38:32", "remaining_time": "8:01:20"} +{"current_steps": 1330, "total_steps": 5364, "loss": 0.09228631854057312, "lr": 1.870064434184524e-05, "epoch": 0.7440559440559441, "percentage": 24.79, "elapsed_time": "2:38:39", "remaining_time": "8:01:14"} +{"current_steps": 1331, "total_steps": 5364, "loss": 0.12133573740720749, "lr": 1.8697434273867193e-05, "epoch": 0.7446153846153846, "percentage": 24.81, "elapsed_time": "2:38:46", "remaining_time": "8:01:06"} +{"current_steps": 1332, "total_steps": 5364, "loss": 0.11321327090263367, "lr": 1.8694220521747765e-05, "epoch": 0.7451748251748251, "percentage": 24.83, "elapsed_time": "2:38:53", "remaining_time": "8:00:59"} +{"current_steps": 1333, "total_steps": 5364, "loss": 0.09745318442583084, "lr": 1.8691003086848273e-05, "epoch": 0.7457342657342657, "percentage": 24.85, "elapsed_time": "2:39:00", "remaining_time": "8:00:51"} +{"current_steps": 1334, "total_steps": 5364, "loss": 0.31245553493499756, "lr": 1.868778197053158e-05, "epoch": 0.7462937062937063, "percentage": 24.87, "elapsed_time": "2:39:08", "remaining_time": "8:00:46"} +{"current_steps": 1335, "total_steps": 5364, "loss": 0.2549764811992645, "lr": 1.8684557174162123e-05, "epoch": 0.7468531468531469, "percentage": 24.89, "elapsed_time": "2:39:15", "remaining_time": "8:00:38"} +{"current_steps": 1336, "total_steps": 5364, "loss": 0.3100712299346924, "lr": 1.868132869910589e-05, "epoch": 0.7474125874125874, "percentage": 24.91, "elapsed_time": "2:39:22", "remaining_time": "8:00:30"} +{"current_steps": 1337, "total_steps": 5364, "loss": 0.12030473351478577, "lr": 1.8678096546730432e-05, "epoch": 0.747972027972028, "percentage": 24.93, "elapsed_time": "2:39:29", "remaining_time": "8:00:22"} +{"current_steps": 1338, "total_steps": 5364, "loss": 0.2433275729417801, "lr": 1.8674860718404852e-05, "epoch": 0.7485314685314686, "percentage": 24.94, "elapsed_time": "2:39:36", "remaining_time": "8:00:14"} +{"current_steps": 1339, "total_steps": 5364, "loss": 0.2903353273868561, "lr": 1.8671621215499812e-05, "epoch": 0.7490909090909091, "percentage": 24.96, "elapsed_time": "2:39:43", "remaining_time": "8:00:06"} +{"current_steps": 1340, "total_steps": 5364, "loss": 0.35398098826408386, "lr": 1.8668378039387532e-05, "epoch": 0.7496503496503496, "percentage": 24.98, "elapsed_time": "2:39:50", "remaining_time": "7:59:59"} +{"current_steps": 1341, "total_steps": 5364, "loss": 0.21113407611846924, "lr": 1.866513119144179e-05, "epoch": 0.7502097902097902, "percentage": 25.0, "elapsed_time": "2:39:57", "remaining_time": "7:59:52"} +{"current_steps": 1342, "total_steps": 5364, "loss": 0.1938326060771942, "lr": 1.8661880673037913e-05, "epoch": 0.7507692307692307, "percentage": 25.02, "elapsed_time": "2:40:05", "remaining_time": "7:59:46"} +{"current_steps": 1343, "total_steps": 5364, "loss": 0.131469264626503, "lr": 1.8658626485552783e-05, "epoch": 0.7513286713286713, "percentage": 25.04, "elapsed_time": "2:40:12", "remaining_time": "7:59:39"} +{"current_steps": 1344, "total_steps": 5364, "loss": 0.09293493628501892, "lr": 1.8655368630364847e-05, "epoch": 0.7518881118881119, "percentage": 25.06, "elapsed_time": "2:40:19", "remaining_time": "7:59:33"} +{"current_steps": 1345, "total_steps": 5364, "loss": 0.13842858374118805, "lr": 1.8652107108854094e-05, "epoch": 0.7524475524475525, "percentage": 25.07, "elapsed_time": "2:40:26", "remaining_time": "7:59:25"} +{"current_steps": 1346, "total_steps": 5364, "loss": 0.18156328797340393, "lr": 1.864884192240207e-05, "epoch": 0.753006993006993, "percentage": 25.09, "elapsed_time": "2:40:33", "remaining_time": "7:59:17"} +{"current_steps": 1347, "total_steps": 5364, "loss": 0.14162813127040863, "lr": 1.8645573072391877e-05, "epoch": 0.7535664335664336, "percentage": 25.11, "elapsed_time": "2:40:40", "remaining_time": "7:59:10"} +{"current_steps": 1348, "total_steps": 5364, "loss": 0.21421772241592407, "lr": 1.8642300560208156e-05, "epoch": 0.7541258741258742, "percentage": 25.13, "elapsed_time": "2:40:47", "remaining_time": "7:59:02"} +{"current_steps": 1349, "total_steps": 5364, "loss": 0.2547898292541504, "lr": 1.8639024387237122e-05, "epoch": 0.7546853146853146, "percentage": 25.15, "elapsed_time": "2:40:54", "remaining_time": "7:58:55"} +{"current_steps": 1350, "total_steps": 5364, "loss": 0.15476490557193756, "lr": 1.863574455486652e-05, "epoch": 0.7552447552447552, "percentage": 25.17, "elapsed_time": "2:41:01", "remaining_time": "7:58:48"} +{"current_steps": 1351, "total_steps": 5364, "loss": 0.1261695921421051, "lr": 1.863246106448565e-05, "epoch": 0.7558041958041958, "percentage": 25.19, "elapsed_time": "2:41:09", "remaining_time": "7:58:42"} +{"current_steps": 1352, "total_steps": 5364, "loss": 0.2383459210395813, "lr": 1.8629173917485367e-05, "epoch": 0.7563636363636363, "percentage": 25.21, "elapsed_time": "2:41:16", "remaining_time": "7:58:34"} +{"current_steps": 1353, "total_steps": 5364, "loss": 0.18813957273960114, "lr": 1.8625883115258075e-05, "epoch": 0.7569230769230769, "percentage": 25.22, "elapsed_time": "2:41:23", "remaining_time": "7:58:26"} +{"current_steps": 1354, "total_steps": 5364, "loss": 0.08599026501178741, "lr": 1.8622588659197717e-05, "epoch": 0.7574825174825175, "percentage": 25.24, "elapsed_time": "2:41:30", "remaining_time": "7:58:19"} +{"current_steps": 1355, "total_steps": 5364, "loss": 0.13507288694381714, "lr": 1.8619290550699794e-05, "epoch": 0.7580419580419581, "percentage": 25.26, "elapsed_time": "2:41:37", "remaining_time": "7:58:13"} +{"current_steps": 1356, "total_steps": 5364, "loss": 0.328048437833786, "lr": 1.8615988791161346e-05, "epoch": 0.7586013986013986, "percentage": 25.28, "elapsed_time": "2:41:44", "remaining_time": "7:58:05"} +{"current_steps": 1357, "total_steps": 5364, "loss": 0.17863020300865173, "lr": 1.861268338198097e-05, "epoch": 0.7591608391608392, "percentage": 25.3, "elapsed_time": "2:41:51", "remaining_time": "7:57:57"} +{"current_steps": 1358, "total_steps": 5364, "loss": 0.10626602917909622, "lr": 1.8609374324558794e-05, "epoch": 0.7597202797202797, "percentage": 25.32, "elapsed_time": "2:41:58", "remaining_time": "7:57:50"} +{"current_steps": 1359, "total_steps": 5364, "loss": 0.13862758874893188, "lr": 1.8606061620296506e-05, "epoch": 0.7602797202797202, "percentage": 25.34, "elapsed_time": "2:42:05", "remaining_time": "7:57:41"} +{"current_steps": 1360, "total_steps": 5364, "loss": 0.2208954244852066, "lr": 1.860274527059733e-05, "epoch": 0.7608391608391608, "percentage": 25.35, "elapsed_time": "2:42:13", "remaining_time": "7:57:35"} +{"current_steps": 1361, "total_steps": 5364, "loss": 0.24852490425109863, "lr": 1.8599425276866036e-05, "epoch": 0.7613986013986014, "percentage": 25.37, "elapsed_time": "2:42:20", "remaining_time": "7:57:28"} +{"current_steps": 1362, "total_steps": 5364, "loss": 0.16396808624267578, "lr": 1.8596101640508943e-05, "epoch": 0.761958041958042, "percentage": 25.39, "elapsed_time": "2:42:27", "remaining_time": "7:57:21"} +{"current_steps": 1363, "total_steps": 5364, "loss": 0.1367586851119995, "lr": 1.8592774362933904e-05, "epoch": 0.7625174825174825, "percentage": 25.41, "elapsed_time": "2:42:34", "remaining_time": "7:57:15"} +{"current_steps": 1364, "total_steps": 5364, "loss": 0.310060977935791, "lr": 1.8589443445550324e-05, "epoch": 0.7630769230769231, "percentage": 25.43, "elapsed_time": "2:42:41", "remaining_time": "7:57:07"} +{"current_steps": 1365, "total_steps": 5364, "loss": 0.05615415796637535, "lr": 1.8586108889769137e-05, "epoch": 0.7636363636363637, "percentage": 25.45, "elapsed_time": "2:42:48", "remaining_time": "7:56:59"} +{"current_steps": 1366, "total_steps": 5364, "loss": 0.10863764584064484, "lr": 1.858277069700283e-05, "epoch": 0.7641958041958042, "percentage": 25.47, "elapsed_time": "2:42:57", "remaining_time": "7:56:56"} +{"current_steps": 1367, "total_steps": 5364, "loss": 0.07619760185480118, "lr": 1.857942886866542e-05, "epoch": 0.7647552447552447, "percentage": 25.48, "elapsed_time": "2:43:04", "remaining_time": "7:56:49"} +{"current_steps": 1368, "total_steps": 5364, "loss": 0.30702123045921326, "lr": 1.8576083406172478e-05, "epoch": 0.7653146853146853, "percentage": 25.5, "elapsed_time": "2:43:11", "remaining_time": "7:56:42"} +{"current_steps": 1369, "total_steps": 5364, "loss": 0.15862104296684265, "lr": 1.8572734310941105e-05, "epoch": 0.7658741258741258, "percentage": 25.52, "elapsed_time": "2:43:18", "remaining_time": "7:56:34"} +{"current_steps": 1370, "total_steps": 5364, "loss": 0.15880060195922852, "lr": 1.856938158438994e-05, "epoch": 0.7664335664335664, "percentage": 25.54, "elapsed_time": "2:43:25", "remaining_time": "7:56:26"} +{"current_steps": 1371, "total_steps": 5364, "loss": 0.27035221457481384, "lr": 1.8566025227939168e-05, "epoch": 0.766993006993007, "percentage": 25.56, "elapsed_time": "2:43:32", "remaining_time": "7:56:18"} +{"current_steps": 1372, "total_steps": 5364, "loss": 0.31578513979911804, "lr": 1.85626652430105e-05, "epoch": 0.7675524475524476, "percentage": 25.58, "elapsed_time": "2:43:39", "remaining_time": "7:56:11"} +{"current_steps": 1373, "total_steps": 5364, "loss": 0.2574416697025299, "lr": 1.8559301631027192e-05, "epoch": 0.7681118881118881, "percentage": 25.6, "elapsed_time": "2:43:46", "remaining_time": "7:56:03"} +{"current_steps": 1374, "total_steps": 5364, "loss": 0.16276581585407257, "lr": 1.8555934393414033e-05, "epoch": 0.7686713286713287, "percentage": 25.62, "elapsed_time": "2:43:53", "remaining_time": "7:55:56"} +{"current_steps": 1375, "total_steps": 5364, "loss": 0.3703039586544037, "lr": 1.855256353159735e-05, "epoch": 0.7692307692307693, "percentage": 25.63, "elapsed_time": "2:44:00", "remaining_time": "7:55:48"} +{"current_steps": 1376, "total_steps": 5364, "loss": 0.11179596185684204, "lr": 1.854918904700501e-05, "epoch": 0.7697902097902098, "percentage": 25.65, "elapsed_time": "2:44:07", "remaining_time": "7:55:41"} +{"current_steps": 1377, "total_steps": 5364, "loss": 0.22255843877792358, "lr": 1.8545810941066403e-05, "epoch": 0.7703496503496503, "percentage": 25.67, "elapsed_time": "2:44:15", "remaining_time": "7:55:34"} +{"current_steps": 1378, "total_steps": 5364, "loss": 0.1687420904636383, "lr": 1.8542429215212464e-05, "epoch": 0.7709090909090909, "percentage": 25.69, "elapsed_time": "2:44:22", "remaining_time": "7:55:26"} +{"current_steps": 1379, "total_steps": 5364, "loss": 0.3099929988384247, "lr": 1.8539043870875648e-05, "epoch": 0.7714685314685314, "percentage": 25.71, "elapsed_time": "2:44:28", "remaining_time": "7:55:18"} +{"current_steps": 1380, "total_steps": 5364, "loss": 0.23256784677505493, "lr": 1.853565490948996e-05, "epoch": 0.772027972027972, "percentage": 25.73, "elapsed_time": "2:44:35", "remaining_time": "7:55:11"} +{"current_steps": 1381, "total_steps": 5364, "loss": 0.16178445518016815, "lr": 1.853226233249092e-05, "epoch": 0.7725874125874126, "percentage": 25.75, "elapsed_time": "2:44:42", "remaining_time": "7:55:03"} +{"current_steps": 1382, "total_steps": 5364, "loss": 0.20867958664894104, "lr": 1.8528866141315597e-05, "epoch": 0.7731468531468532, "percentage": 25.76, "elapsed_time": "2:44:50", "remaining_time": "7:54:57"} +{"current_steps": 1383, "total_steps": 5364, "loss": 0.390109658241272, "lr": 1.8525466337402572e-05, "epoch": 0.7737062937062937, "percentage": 25.78, "elapsed_time": "2:44:57", "remaining_time": "7:54:49"} +{"current_steps": 1384, "total_steps": 5364, "loss": 0.3983703553676605, "lr": 1.8522062922191976e-05, "epoch": 0.7742657342657343, "percentage": 25.8, "elapsed_time": "2:45:04", "remaining_time": "7:54:42"} +{"current_steps": 1385, "total_steps": 5364, "loss": 0.3265753388404846, "lr": 1.8518655897125452e-05, "epoch": 0.7748251748251749, "percentage": 25.82, "elapsed_time": "2:45:11", "remaining_time": "7:54:35"} +{"current_steps": 1386, "total_steps": 5364, "loss": 0.18475818634033203, "lr": 1.8515245263646186e-05, "epoch": 0.7753846153846153, "percentage": 25.84, "elapsed_time": "2:45:18", "remaining_time": "7:54:28"} +{"current_steps": 1387, "total_steps": 5364, "loss": 0.5182363390922546, "lr": 1.851183102319888e-05, "epoch": 0.7759440559440559, "percentage": 25.86, "elapsed_time": "2:45:25", "remaining_time": "7:54:20"} +{"current_steps": 1388, "total_steps": 5364, "loss": 0.08971129357814789, "lr": 1.8508413177229775e-05, "epoch": 0.7765034965034965, "percentage": 25.88, "elapsed_time": "2:45:32", "remaining_time": "7:54:12"} +{"current_steps": 1389, "total_steps": 5364, "loss": 0.3412243127822876, "lr": 1.8504991727186634e-05, "epoch": 0.777062937062937, "percentage": 25.89, "elapsed_time": "2:45:39", "remaining_time": "7:54:04"} +{"current_steps": 1390, "total_steps": 5364, "loss": 0.2338174283504486, "lr": 1.8501566674518743e-05, "epoch": 0.7776223776223776, "percentage": 25.91, "elapsed_time": "2:45:47", "remaining_time": "7:53:59"} +{"current_steps": 1391, "total_steps": 5364, "loss": 0.2681596577167511, "lr": 1.8498138020676924e-05, "epoch": 0.7781818181818182, "percentage": 25.93, "elapsed_time": "2:45:54", "remaining_time": "7:53:52"} +{"current_steps": 1392, "total_steps": 5364, "loss": 0.29711586236953735, "lr": 1.8494705767113513e-05, "epoch": 0.7787412587412588, "percentage": 25.95, "elapsed_time": "2:46:01", "remaining_time": "7:53:44"} +{"current_steps": 1393, "total_steps": 5364, "loss": 0.11838226765394211, "lr": 1.8491269915282382e-05, "epoch": 0.7793006993006993, "percentage": 25.97, "elapsed_time": "2:46:09", "remaining_time": "7:53:39"} +{"current_steps": 1394, "total_steps": 5364, "loss": 0.14415180683135986, "lr": 1.8487830466638916e-05, "epoch": 0.7798601398601399, "percentage": 25.99, "elapsed_time": "2:46:16", "remaining_time": "7:53:32"} +{"current_steps": 1395, "total_steps": 5364, "loss": 0.0834486111998558, "lr": 1.8484387422640035e-05, "epoch": 0.7804195804195804, "percentage": 26.01, "elapsed_time": "2:46:24", "remaining_time": "7:53:26"} +{"current_steps": 1396, "total_steps": 5364, "loss": 0.23688632249832153, "lr": 1.848094078474417e-05, "epoch": 0.7809790209790209, "percentage": 26.03, "elapsed_time": "2:46:31", "remaining_time": "7:53:18"} +{"current_steps": 1397, "total_steps": 5364, "loss": 0.21789711713790894, "lr": 1.8477490554411284e-05, "epoch": 0.7815384615384615, "percentage": 26.04, "elapsed_time": "2:46:38", "remaining_time": "7:53:11"} +{"current_steps": 1398, "total_steps": 5364, "loss": 0.10302956402301788, "lr": 1.8474036733102857e-05, "epoch": 0.7820979020979021, "percentage": 26.06, "elapsed_time": "2:46:45", "remaining_time": "7:53:04"} +{"current_steps": 1399, "total_steps": 5364, "loss": 0.23543362319469452, "lr": 1.8470579322281893e-05, "epoch": 0.7826573426573427, "percentage": 26.08, "elapsed_time": "2:46:52", "remaining_time": "7:52:57"} +{"current_steps": 1400, "total_steps": 5364, "loss": 0.1557886153459549, "lr": 1.8467118323412914e-05, "epoch": 0.7832167832167832, "percentage": 26.1, "elapsed_time": "2:46:59", "remaining_time": "7:52:49"} +{"current_steps": 1401, "total_steps": 5364, "loss": 0.21970303356647491, "lr": 1.846365373796196e-05, "epoch": 0.7837762237762238, "percentage": 26.12, "elapsed_time": "2:47:06", "remaining_time": "7:52:41"} +{"current_steps": 1402, "total_steps": 5364, "loss": 0.1444120854139328, "lr": 1.84601855673966e-05, "epoch": 0.7843356643356644, "percentage": 26.14, "elapsed_time": "2:47:13", "remaining_time": "7:52:34"} +{"current_steps": 1403, "total_steps": 5364, "loss": 0.33197301626205444, "lr": 1.8456713813185905e-05, "epoch": 0.784895104895105, "percentage": 26.16, "elapsed_time": "2:47:20", "remaining_time": "7:52:26"} +{"current_steps": 1404, "total_steps": 5364, "loss": 0.17940428853034973, "lr": 1.845323847680048e-05, "epoch": 0.7854545454545454, "percentage": 26.17, "elapsed_time": "2:47:27", "remaining_time": "7:52:20"} +{"current_steps": 1405, "total_steps": 5364, "loss": 0.28178954124450684, "lr": 1.8449759559712438e-05, "epoch": 0.786013986013986, "percentage": 26.19, "elapsed_time": "2:47:34", "remaining_time": "7:52:12"} +{"current_steps": 1406, "total_steps": 5364, "loss": 0.07722143828868866, "lr": 1.844627706339541e-05, "epoch": 0.7865734265734265, "percentage": 26.21, "elapsed_time": "2:47:41", "remaining_time": "7:52:04"} +{"current_steps": 1407, "total_steps": 5364, "loss": 0.24841883778572083, "lr": 1.844279098932455e-05, "epoch": 0.7871328671328671, "percentage": 26.23, "elapsed_time": "2:47:48", "remaining_time": "7:51:56"} +{"current_steps": 1408, "total_steps": 5364, "loss": 0.3509570360183716, "lr": 1.8439301338976518e-05, "epoch": 0.7876923076923077, "percentage": 26.25, "elapsed_time": "2:47:55", "remaining_time": "7:51:48"} +{"current_steps": 1409, "total_steps": 5364, "loss": 0.1569032073020935, "lr": 1.8435808113829494e-05, "epoch": 0.7882517482517483, "percentage": 26.27, "elapsed_time": "2:48:02", "remaining_time": "7:51:41"} +{"current_steps": 1410, "total_steps": 5364, "loss": 0.08288843184709549, "lr": 1.8432311315363173e-05, "epoch": 0.7888111888111888, "percentage": 26.29, "elapsed_time": "2:48:09", "remaining_time": "7:51:34"} +{"current_steps": 1411, "total_steps": 5364, "loss": 0.2277572751045227, "lr": 1.842881094505876e-05, "epoch": 0.7893706293706294, "percentage": 26.3, "elapsed_time": "2:48:17", "remaining_time": "7:51:28"} +{"current_steps": 1412, "total_steps": 5364, "loss": 0.24825136363506317, "lr": 1.8425307004398975e-05, "epoch": 0.78993006993007, "percentage": 26.32, "elapsed_time": "2:48:24", "remaining_time": "7:51:21"} +{"current_steps": 1413, "total_steps": 5364, "loss": 0.248886838555336, "lr": 1.842179949486805e-05, "epoch": 0.7904895104895104, "percentage": 26.34, "elapsed_time": "2:48:32", "remaining_time": "7:51:15"} +{"current_steps": 1414, "total_steps": 5364, "loss": 0.25317031145095825, "lr": 1.841828841795173e-05, "epoch": 0.791048951048951, "percentage": 26.36, "elapsed_time": "2:48:39", "remaining_time": "7:51:07"} +{"current_steps": 1415, "total_steps": 5364, "loss": 0.1791454702615738, "lr": 1.841477377513727e-05, "epoch": 0.7916083916083916, "percentage": 26.38, "elapsed_time": "2:48:46", "remaining_time": "7:51:00"} +{"current_steps": 1416, "total_steps": 5364, "loss": 0.038264572620391846, "lr": 1.8411255567913437e-05, "epoch": 0.7921678321678322, "percentage": 26.4, "elapsed_time": "2:48:53", "remaining_time": "7:50:54"} +{"current_steps": 1417, "total_steps": 5364, "loss": 0.11317572742700577, "lr": 1.8407733797770504e-05, "epoch": 0.7927272727272727, "percentage": 26.42, "elapsed_time": "2:49:01", "remaining_time": "7:50:47"} +{"current_steps": 1418, "total_steps": 5364, "loss": 0.11327424645423889, "lr": 1.8404208466200254e-05, "epoch": 0.7932867132867133, "percentage": 26.44, "elapsed_time": "2:49:08", "remaining_time": "7:50:40"} +{"current_steps": 1419, "total_steps": 5364, "loss": 0.26368314027786255, "lr": 1.840067957469599e-05, "epoch": 0.7938461538461539, "percentage": 26.45, "elapsed_time": "2:49:15", "remaining_time": "7:50:32"} +{"current_steps": 1420, "total_steps": 5364, "loss": 0.565875768661499, "lr": 1.8397147124752495e-05, "epoch": 0.7944055944055944, "percentage": 26.47, "elapsed_time": "2:49:22", "remaining_time": "7:50:26"} +{"current_steps": 1421, "total_steps": 5364, "loss": 0.11382953822612762, "lr": 1.8393611117866098e-05, "epoch": 0.794965034965035, "percentage": 26.49, "elapsed_time": "2:49:29", "remaining_time": "7:50:18"} +{"current_steps": 1422, "total_steps": 5364, "loss": 0.3906436264514923, "lr": 1.83900715555346e-05, "epoch": 0.7955244755244755, "percentage": 26.51, "elapsed_time": "2:49:36", "remaining_time": "7:50:10"} +{"current_steps": 1423, "total_steps": 5364, "loss": 0.318826824426651, "lr": 1.8386528439257328e-05, "epoch": 0.796083916083916, "percentage": 26.53, "elapsed_time": "2:49:43", "remaining_time": "7:50:03"} +{"current_steps": 1424, "total_steps": 5364, "loss": 0.09145613759756088, "lr": 1.838298177053511e-05, "epoch": 0.7966433566433566, "percentage": 26.55, "elapsed_time": "2:49:50", "remaining_time": "7:49:55"} +{"current_steps": 1425, "total_steps": 5364, "loss": 0.1778845340013504, "lr": 1.8379431550870273e-05, "epoch": 0.7972027972027972, "percentage": 26.57, "elapsed_time": "2:49:57", "remaining_time": "7:49:47"} +{"current_steps": 1426, "total_steps": 5364, "loss": 0.08287844806909561, "lr": 1.8375877781766655e-05, "epoch": 0.7977622377622378, "percentage": 26.58, "elapsed_time": "2:50:04", "remaining_time": "7:49:40"} +{"current_steps": 1427, "total_steps": 5364, "loss": 0.2466338872909546, "lr": 1.8372320464729595e-05, "epoch": 0.7983216783216783, "percentage": 26.6, "elapsed_time": "2:50:11", "remaining_time": "7:49:32"} +{"current_steps": 1428, "total_steps": 5364, "loss": 0.20674027502536774, "lr": 1.8368759601265935e-05, "epoch": 0.7988811188811189, "percentage": 26.62, "elapsed_time": "2:50:18", "remaining_time": "7:49:26"} +{"current_steps": 1429, "total_steps": 5364, "loss": 0.1120634526014328, "lr": 1.8365195192884016e-05, "epoch": 0.7994405594405595, "percentage": 26.64, "elapsed_time": "2:50:25", "remaining_time": "7:49:18"} +{"current_steps": 1430, "total_steps": 5364, "loss": 0.20542174577713013, "lr": 1.836162724109369e-05, "epoch": 0.8, "percentage": 26.66, "elapsed_time": "2:50:33", "remaining_time": "7:49:12"} +{"current_steps": 1431, "total_steps": 5364, "loss": 0.11126194894313812, "lr": 1.83580557474063e-05, "epoch": 0.8005594405594405, "percentage": 26.68, "elapsed_time": "2:50:40", "remaining_time": "7:49:05"} +{"current_steps": 1432, "total_steps": 5364, "loss": 0.13972947001457214, "lr": 1.8354480713334696e-05, "epoch": 0.8011188811188811, "percentage": 26.7, "elapsed_time": "2:50:48", "remaining_time": "7:48:59"} +{"current_steps": 1433, "total_steps": 5364, "loss": 0.049743738025426865, "lr": 1.8350902140393217e-05, "epoch": 0.8016783216783216, "percentage": 26.72, "elapsed_time": "2:50:55", "remaining_time": "7:48:51"} +{"current_steps": 1434, "total_steps": 5364, "loss": 0.1878945678472519, "lr": 1.8347320030097714e-05, "epoch": 0.8022377622377622, "percentage": 26.73, "elapsed_time": "2:51:02", "remaining_time": "7:48:45"} +{"current_steps": 1435, "total_steps": 5364, "loss": 0.14915086328983307, "lr": 1.8343734383965536e-05, "epoch": 0.8027972027972028, "percentage": 26.75, "elapsed_time": "2:51:09", "remaining_time": "7:48:38"} +{"current_steps": 1436, "total_steps": 5364, "loss": 0.22911271452903748, "lr": 1.8340145203515516e-05, "epoch": 0.8033566433566434, "percentage": 26.77, "elapsed_time": "2:51:16", "remaining_time": "7:48:30"} +{"current_steps": 1437, "total_steps": 5364, "loss": 0.32970792055130005, "lr": 1.8336552490268e-05, "epoch": 0.8039160839160839, "percentage": 26.79, "elapsed_time": "2:51:23", "remaining_time": "7:48:22"} +{"current_steps": 1438, "total_steps": 5364, "loss": 0.31875988841056824, "lr": 1.833295624574482e-05, "epoch": 0.8044755244755245, "percentage": 26.81, "elapsed_time": "2:51:30", "remaining_time": "7:48:16"} +{"current_steps": 1439, "total_steps": 5364, "loss": 0.27195048332214355, "lr": 1.8329356471469303e-05, "epoch": 0.8050349650349651, "percentage": 26.83, "elapsed_time": "2:51:37", "remaining_time": "7:48:08"} +{"current_steps": 1440, "total_steps": 5364, "loss": 0.210830420255661, "lr": 1.8325753168966286e-05, "epoch": 0.8055944055944056, "percentage": 26.85, "elapsed_time": "2:51:44", "remaining_time": "7:48:00"} +{"current_steps": 1441, "total_steps": 5364, "loss": 0.19733984768390656, "lr": 1.8322146339762077e-05, "epoch": 0.8061538461538461, "percentage": 26.86, "elapsed_time": "2:51:52", "remaining_time": "7:47:54"} +{"current_steps": 1442, "total_steps": 5364, "loss": 0.17760130763053894, "lr": 1.8318535985384503e-05, "epoch": 0.8067132867132867, "percentage": 26.88, "elapsed_time": "2:51:59", "remaining_time": "7:47:47"} +{"current_steps": 1443, "total_steps": 5364, "loss": 0.12815436720848083, "lr": 1.8314922107362866e-05, "epoch": 0.8072727272727273, "percentage": 26.9, "elapsed_time": "2:52:06", "remaining_time": "7:47:40"} +{"current_steps": 1444, "total_steps": 5364, "loss": 0.41048890352249146, "lr": 1.8311304707227963e-05, "epoch": 0.8078321678321678, "percentage": 26.92, "elapsed_time": "2:52:13", "remaining_time": "7:47:33"} +{"current_steps": 1445, "total_steps": 5364, "loss": 0.0654655247926712, "lr": 1.8307683786512095e-05, "epoch": 0.8083916083916084, "percentage": 26.94, "elapsed_time": "2:52:20", "remaining_time": "7:47:25"} +{"current_steps": 1446, "total_steps": 5364, "loss": 0.13886767625808716, "lr": 1.8304059346749042e-05, "epoch": 0.808951048951049, "percentage": 26.96, "elapsed_time": "2:52:27", "remaining_time": "7:47:17"} +{"current_steps": 1447, "total_steps": 5364, "loss": 0.11193230748176575, "lr": 1.8300431389474077e-05, "epoch": 0.8095104895104895, "percentage": 26.98, "elapsed_time": "2:52:34", "remaining_time": "7:47:10"} +{"current_steps": 1448, "total_steps": 5364, "loss": 0.06599109619855881, "lr": 1.8296799916223963e-05, "epoch": 0.8100699300699301, "percentage": 26.99, "elapsed_time": "2:52:42", "remaining_time": "7:47:03"} +{"current_steps": 1449, "total_steps": 5364, "loss": 0.19477437436580658, "lr": 1.8293164928536962e-05, "epoch": 0.8106293706293707, "percentage": 27.01, "elapsed_time": "2:52:48", "remaining_time": "7:46:55"} +{"current_steps": 1450, "total_steps": 5364, "loss": 0.19513902068138123, "lr": 1.8289526427952805e-05, "epoch": 0.8111888111888111, "percentage": 27.03, "elapsed_time": "2:52:56", "remaining_time": "7:46:48"} +{"current_steps": 1451, "total_steps": 5364, "loss": 0.31178590655326843, "lr": 1.8285884416012727e-05, "epoch": 0.8117482517482517, "percentage": 27.05, "elapsed_time": "2:53:02", "remaining_time": "7:46:40"} +{"current_steps": 1452, "total_steps": 5364, "loss": 0.4289611577987671, "lr": 1.8282238894259455e-05, "epoch": 0.8123076923076923, "percentage": 27.07, "elapsed_time": "2:53:09", "remaining_time": "7:46:32"} +{"current_steps": 1453, "total_steps": 5364, "loss": 0.08483538776636124, "lr": 1.827858986423718e-05, "epoch": 0.8128671328671329, "percentage": 27.09, "elapsed_time": "2:53:17", "remaining_time": "7:46:26"} +{"current_steps": 1454, "total_steps": 5364, "loss": 0.11833000183105469, "lr": 1.82749373274916e-05, "epoch": 0.8134265734265734, "percentage": 27.11, "elapsed_time": "2:53:24", "remaining_time": "7:46:19"} +{"current_steps": 1455, "total_steps": 5364, "loss": 0.18328005075454712, "lr": 1.8271281285569884e-05, "epoch": 0.813986013986014, "percentage": 27.13, "elapsed_time": "2:53:31", "remaining_time": "7:46:11"} +{"current_steps": 1456, "total_steps": 5364, "loss": 0.27575016021728516, "lr": 1.8267621740020704e-05, "epoch": 0.8145454545454546, "percentage": 27.14, "elapsed_time": "2:53:38", "remaining_time": "7:46:04"} +{"current_steps": 1457, "total_steps": 5364, "loss": 0.24062323570251465, "lr": 1.8263958692394196e-05, "epoch": 0.8151048951048951, "percentage": 27.16, "elapsed_time": "2:53:45", "remaining_time": "7:45:57"} +{"current_steps": 1458, "total_steps": 5364, "loss": 0.32751497626304626, "lr": 1.826029214424199e-05, "epoch": 0.8156643356643357, "percentage": 27.18, "elapsed_time": "2:53:52", "remaining_time": "7:45:49"} +{"current_steps": 1459, "total_steps": 5364, "loss": 0.23958967626094818, "lr": 1.8256622097117197e-05, "epoch": 0.8162237762237762, "percentage": 27.2, "elapsed_time": "2:53:59", "remaining_time": "7:45:41"} +{"current_steps": 1460, "total_steps": 5364, "loss": 0.16753628849983215, "lr": 1.8252948552574416e-05, "epoch": 0.8167832167832167, "percentage": 27.22, "elapsed_time": "2:54:06", "remaining_time": "7:45:33"} +{"current_steps": 1461, "total_steps": 5364, "loss": 0.08340223133563995, "lr": 1.8249271512169712e-05, "epoch": 0.8173426573426573, "percentage": 27.24, "elapsed_time": "2:54:14", "remaining_time": "7:45:27"} +{"current_steps": 1462, "total_steps": 5364, "loss": 0.16506993770599365, "lr": 1.824559097746065e-05, "epoch": 0.8179020979020979, "percentage": 27.26, "elapsed_time": "2:54:21", "remaining_time": "7:45:21"} +{"current_steps": 1463, "total_steps": 5364, "loss": 0.183750718832016, "lr": 1.8241906950006257e-05, "epoch": 0.8184615384615385, "percentage": 27.27, "elapsed_time": "2:54:28", "remaining_time": "7:45:14"} +{"current_steps": 1464, "total_steps": 5364, "loss": 0.2130127102136612, "lr": 1.8238219431367056e-05, "epoch": 0.819020979020979, "percentage": 27.29, "elapsed_time": "2:54:35", "remaining_time": "7:45:07"} +{"current_steps": 1465, "total_steps": 5364, "loss": 0.1270238757133484, "lr": 1.823452842310504e-05, "epoch": 0.8195804195804196, "percentage": 27.31, "elapsed_time": "2:54:42", "remaining_time": "7:44:59"} +{"current_steps": 1466, "total_steps": 5364, "loss": 0.06742765009403229, "lr": 1.8230833926783673e-05, "epoch": 0.8201398601398602, "percentage": 27.33, "elapsed_time": "2:54:49", "remaining_time": "7:44:51"} +{"current_steps": 1467, "total_steps": 5364, "loss": 0.2227836698293686, "lr": 1.8227135943967912e-05, "epoch": 0.8206993006993007, "percentage": 27.35, "elapsed_time": "2:54:56", "remaining_time": "7:44:44"} +{"current_steps": 1468, "total_steps": 5364, "loss": 0.3019767999649048, "lr": 1.8223434476224178e-05, "epoch": 0.8212587412587412, "percentage": 27.37, "elapsed_time": "2:55:04", "remaining_time": "7:44:37"} +{"current_steps": 1469, "total_steps": 5364, "loss": 0.08236641436815262, "lr": 1.8219729525120383e-05, "epoch": 0.8218181818181818, "percentage": 27.39, "elapsed_time": "2:55:11", "remaining_time": "7:44:30"} +{"current_steps": 1470, "total_steps": 5364, "loss": 0.13357867300510406, "lr": 1.82160210922259e-05, "epoch": 0.8223776223776224, "percentage": 27.4, "elapsed_time": "2:55:18", "remaining_time": "7:44:23"} +{"current_steps": 1471, "total_steps": 5364, "loss": 0.17603634297847748, "lr": 1.8212309179111578e-05, "epoch": 0.8229370629370629, "percentage": 27.42, "elapsed_time": "2:55:25", "remaining_time": "7:44:16"} +{"current_steps": 1472, "total_steps": 5364, "loss": 0.20842991769313812, "lr": 1.820859378734975e-05, "epoch": 0.8234965034965035, "percentage": 27.44, "elapsed_time": "2:55:32", "remaining_time": "7:44:08"} +{"current_steps": 1473, "total_steps": 5364, "loss": 0.04474640265107155, "lr": 1.820487491851421e-05, "epoch": 0.8240559440559441, "percentage": 27.46, "elapsed_time": "2:55:39", "remaining_time": "7:44:01"} +{"current_steps": 1474, "total_steps": 5364, "loss": 0.3176015019416809, "lr": 1.8201152574180243e-05, "epoch": 0.8246153846153846, "percentage": 27.48, "elapsed_time": "2:55:46", "remaining_time": "7:43:53"} +{"current_steps": 1475, "total_steps": 5364, "loss": 0.16510044038295746, "lr": 1.8197426755924587e-05, "epoch": 0.8251748251748252, "percentage": 27.5, "elapsed_time": "2:55:53", "remaining_time": "7:43:45"} +{"current_steps": 1476, "total_steps": 5364, "loss": 0.03591936081647873, "lr": 1.8193697465325458e-05, "epoch": 0.8257342657342658, "percentage": 27.52, "elapsed_time": "2:56:01", "remaining_time": "7:43:39"} +{"current_steps": 1477, "total_steps": 5364, "loss": 0.09115429222583771, "lr": 1.8189964703962545e-05, "epoch": 0.8262937062937062, "percentage": 27.54, "elapsed_time": "2:56:08", "remaining_time": "7:43:32"} +{"current_steps": 1478, "total_steps": 5364, "loss": 0.055043987929821014, "lr": 1.818622847341701e-05, "epoch": 0.8268531468531468, "percentage": 27.55, "elapsed_time": "2:56:15", "remaining_time": "7:43:26"} +{"current_steps": 1479, "total_steps": 5364, "loss": 0.03424611687660217, "lr": 1.8182488775271477e-05, "epoch": 0.8274125874125874, "percentage": 27.57, "elapsed_time": "2:56:23", "remaining_time": "7:43:19"} +{"current_steps": 1480, "total_steps": 5364, "loss": 0.1455080360174179, "lr": 1.8178745611110045e-05, "epoch": 0.827972027972028, "percentage": 27.59, "elapsed_time": "2:56:30", "remaining_time": "7:43:11"} +{"current_steps": 1481, "total_steps": 5364, "loss": 0.09734778851270676, "lr": 1.817499898251828e-05, "epoch": 0.8285314685314685, "percentage": 27.61, "elapsed_time": "2:56:37", "remaining_time": "7:43:04"} +{"current_steps": 1482, "total_steps": 5364, "loss": 0.3973001539707184, "lr": 1.8171248891083212e-05, "epoch": 0.8290909090909091, "percentage": 27.63, "elapsed_time": "2:56:44", "remaining_time": "7:42:56"} +{"current_steps": 1483, "total_steps": 5364, "loss": 0.2914142608642578, "lr": 1.8167495338393337e-05, "epoch": 0.8296503496503497, "percentage": 27.65, "elapsed_time": "2:56:51", "remaining_time": "7:42:50"} +{"current_steps": 1484, "total_steps": 5364, "loss": 0.39465975761413574, "lr": 1.8163738326038625e-05, "epoch": 0.8302097902097902, "percentage": 27.67, "elapsed_time": "2:56:58", "remaining_time": "7:42:42"} +{"current_steps": 1485, "total_steps": 5364, "loss": 0.30928659439086914, "lr": 1.8159977855610507e-05, "epoch": 0.8307692307692308, "percentage": 27.68, "elapsed_time": "2:57:05", "remaining_time": "7:42:36"} +{"current_steps": 1486, "total_steps": 5364, "loss": 0.20367246866226196, "lr": 1.8156213928701878e-05, "epoch": 0.8313286713286713, "percentage": 27.7, "elapsed_time": "2:57:13", "remaining_time": "7:42:30"} +{"current_steps": 1487, "total_steps": 5364, "loss": 0.06059672683477402, "lr": 1.8152446546907095e-05, "epoch": 0.8318881118881118, "percentage": 27.72, "elapsed_time": "2:57:20", "remaining_time": "7:42:23"} +{"current_steps": 1488, "total_steps": 5364, "loss": 0.15735286474227905, "lr": 1.8148675711821982e-05, "epoch": 0.8324475524475524, "percentage": 27.74, "elapsed_time": "2:57:27", "remaining_time": "7:42:15"} +{"current_steps": 1489, "total_steps": 5364, "loss": 0.16312754154205322, "lr": 1.814490142504383e-05, "epoch": 0.833006993006993, "percentage": 27.76, "elapsed_time": "2:57:34", "remaining_time": "7:42:07"} +{"current_steps": 1490, "total_steps": 5364, "loss": 0.24480485916137695, "lr": 1.814112368817138e-05, "epoch": 0.8335664335664336, "percentage": 27.78, "elapsed_time": "2:57:42", "remaining_time": "7:42:02"} +{"current_steps": 1491, "total_steps": 5364, "loss": 0.23900210857391357, "lr": 1.8137342502804848e-05, "epoch": 0.8341258741258741, "percentage": 27.8, "elapsed_time": "2:57:49", "remaining_time": "7:41:55"} +{"current_steps": 1492, "total_steps": 5364, "loss": 0.0961148738861084, "lr": 1.81335578705459e-05, "epoch": 0.8346853146853147, "percentage": 27.82, "elapsed_time": "2:57:57", "remaining_time": "7:41:49"} +{"current_steps": 1493, "total_steps": 5364, "loss": 0.22839057445526123, "lr": 1.8129769792997662e-05, "epoch": 0.8352447552447553, "percentage": 27.83, "elapsed_time": "2:58:04", "remaining_time": "7:41:42"} +{"current_steps": 1494, "total_steps": 5364, "loss": 0.2662620544433594, "lr": 1.8125978271764736e-05, "epoch": 0.8358041958041958, "percentage": 27.85, "elapsed_time": "2:58:11", "remaining_time": "7:41:35"} +{"current_steps": 1495, "total_steps": 5364, "loss": 0.22938987612724304, "lr": 1.8122183308453157e-05, "epoch": 0.8363636363636363, "percentage": 27.87, "elapsed_time": "2:58:18", "remaining_time": "7:41:27"} +{"current_steps": 1496, "total_steps": 5364, "loss": 0.09429021179676056, "lr": 1.811838490467044e-05, "epoch": 0.8369230769230769, "percentage": 27.89, "elapsed_time": "2:58:26", "remaining_time": "7:41:22"} +{"current_steps": 1497, "total_steps": 5364, "loss": 0.24409779906272888, "lr": 1.8114583062025547e-05, "epoch": 0.8374825174825175, "percentage": 27.91, "elapsed_time": "2:58:33", "remaining_time": "7:41:14"} +{"current_steps": 1498, "total_steps": 5364, "loss": 0.18156293034553528, "lr": 1.81107777821289e-05, "epoch": 0.838041958041958, "percentage": 27.93, "elapsed_time": "2:58:40", "remaining_time": "7:41:08"} +{"current_steps": 1499, "total_steps": 5364, "loss": 0.25297877192497253, "lr": 1.8106969066592368e-05, "epoch": 0.8386013986013986, "percentage": 27.95, "elapsed_time": "2:58:47", "remaining_time": "7:41:00"} +{"current_steps": 1500, "total_steps": 5364, "loss": 0.297002375125885, "lr": 1.8103156917029293e-05, "epoch": 0.8391608391608392, "percentage": 27.96, "elapsed_time": "2:58:54", "remaining_time": "7:40:52"} +{"current_steps": 1501, "total_steps": 5364, "loss": 0.1789444386959076, "lr": 1.8099341335054453e-05, "epoch": 0.8397202797202797, "percentage": 27.98, "elapsed_time": "2:59:29", "remaining_time": "7:41:56"} +{"current_steps": 1502, "total_steps": 5364, "loss": 0.043852709233760834, "lr": 1.809552232228409e-05, "epoch": 0.8402797202797203, "percentage": 28.0, "elapsed_time": "2:59:36", "remaining_time": "7:41:48"} +{"current_steps": 1503, "total_steps": 5364, "loss": 0.11034829914569855, "lr": 1.8091699880335903e-05, "epoch": 0.8408391608391609, "percentage": 28.02, "elapsed_time": "2:59:43", "remaining_time": "7:41:42"} +{"current_steps": 1504, "total_steps": 5364, "loss": 0.2640632390975952, "lr": 1.808787401082903e-05, "epoch": 0.8413986013986015, "percentage": 28.04, "elapsed_time": "2:59:50", "remaining_time": "7:41:34"} +{"current_steps": 1505, "total_steps": 5364, "loss": 0.3162291646003723, "lr": 1.808404471538408e-05, "epoch": 0.8419580419580419, "percentage": 28.06, "elapsed_time": "2:59:57", "remaining_time": "7:41:27"} +{"current_steps": 1506, "total_steps": 5364, "loss": 0.4285929501056671, "lr": 1.808021199562309e-05, "epoch": 0.8425174825174825, "percentage": 28.08, "elapsed_time": "3:00:04", "remaining_time": "7:41:19"} +{"current_steps": 1507, "total_steps": 5364, "loss": 0.2947182357311249, "lr": 1.8076375853169562e-05, "epoch": 0.8430769230769231, "percentage": 28.09, "elapsed_time": "3:00:11", "remaining_time": "7:41:11"} +{"current_steps": 1508, "total_steps": 5364, "loss": 0.12361947447061539, "lr": 1.8072536289648455e-05, "epoch": 0.8436363636363636, "percentage": 28.11, "elapsed_time": "3:00:18", "remaining_time": "7:41:03"} +{"current_steps": 1509, "total_steps": 5364, "loss": 0.2050374448299408, "lr": 1.8068693306686157e-05, "epoch": 0.8441958041958042, "percentage": 28.13, "elapsed_time": "3:00:25", "remaining_time": "7:40:55"} +{"current_steps": 1510, "total_steps": 5364, "loss": 0.0773276835680008, "lr": 1.806484690591052e-05, "epoch": 0.8447552447552448, "percentage": 28.15, "elapsed_time": "3:00:32", "remaining_time": "7:40:47"} +{"current_steps": 1511, "total_steps": 5364, "loss": 0.13120685517787933, "lr": 1.8060997088950837e-05, "epoch": 0.8453146853146853, "percentage": 28.17, "elapsed_time": "3:00:39", "remaining_time": "7:40:41"} +{"current_steps": 1512, "total_steps": 5364, "loss": 0.24504128098487854, "lr": 1.805714385743785e-05, "epoch": 0.8458741258741259, "percentage": 28.19, "elapsed_time": "3:00:46", "remaining_time": "7:40:33"} +{"current_steps": 1513, "total_steps": 5364, "loss": 0.2342679500579834, "lr": 1.805328721300375e-05, "epoch": 0.8464335664335665, "percentage": 28.21, "elapsed_time": "3:00:53", "remaining_time": "7:40:25"} +{"current_steps": 1514, "total_steps": 5364, "loss": 0.15123756229877472, "lr": 1.8049427157282164e-05, "epoch": 0.846993006993007, "percentage": 28.23, "elapsed_time": "3:01:01", "remaining_time": "7:40:18"} +{"current_steps": 1515, "total_steps": 5364, "loss": 0.21666626632213593, "lr": 1.8045563691908178e-05, "epoch": 0.8475524475524475, "percentage": 28.24, "elapsed_time": "3:01:08", "remaining_time": "7:40:11"} +{"current_steps": 1516, "total_steps": 5364, "loss": 0.16559508442878723, "lr": 1.8041696818518314e-05, "epoch": 0.8481118881118881, "percentage": 28.26, "elapsed_time": "3:01:15", "remaining_time": "7:40:04"} +{"current_steps": 1517, "total_steps": 5364, "loss": 0.20894457399845123, "lr": 1.803782653875054e-05, "epoch": 0.8486713286713287, "percentage": 28.28, "elapsed_time": "3:01:22", "remaining_time": "7:39:56"} +{"current_steps": 1518, "total_steps": 5364, "loss": 0.14789265394210815, "lr": 1.8033952854244256e-05, "epoch": 0.8492307692307692, "percentage": 28.3, "elapsed_time": "3:01:29", "remaining_time": "7:39:48"} +{"current_steps": 1519, "total_steps": 5364, "loss": 0.16713255643844604, "lr": 1.8030075766640326e-05, "epoch": 0.8497902097902098, "percentage": 28.32, "elapsed_time": "3:01:36", "remaining_time": "7:39:42"} +{"current_steps": 1520, "total_steps": 5364, "loss": 0.32784438133239746, "lr": 1.8026195277581035e-05, "epoch": 0.8503496503496504, "percentage": 28.34, "elapsed_time": "3:01:43", "remaining_time": "7:39:35"} +{"current_steps": 1521, "total_steps": 5364, "loss": 0.44883227348327637, "lr": 1.8022311388710127e-05, "epoch": 0.850909090909091, "percentage": 28.36, "elapsed_time": "3:01:50", "remaining_time": "7:39:27"} +{"current_steps": 1522, "total_steps": 5364, "loss": 0.11389957368373871, "lr": 1.8018424101672762e-05, "epoch": 0.8514685314685315, "percentage": 28.37, "elapsed_time": "3:01:58", "remaining_time": "7:39:20"} +{"current_steps": 1523, "total_steps": 5364, "loss": 0.20411473512649536, "lr": 1.8014533418115567e-05, "epoch": 0.852027972027972, "percentage": 28.39, "elapsed_time": "3:02:05", "remaining_time": "7:39:15"} +{"current_steps": 1524, "total_steps": 5364, "loss": 0.11225219070911407, "lr": 1.8010639339686585e-05, "epoch": 0.8525874125874126, "percentage": 28.41, "elapsed_time": "3:02:12", "remaining_time": "7:39:07"} +{"current_steps": 1525, "total_steps": 5364, "loss": 0.36463797092437744, "lr": 1.8006741868035313e-05, "epoch": 0.8531468531468531, "percentage": 28.43, "elapsed_time": "3:02:20", "remaining_time": "7:39:00"} +{"current_steps": 1526, "total_steps": 5364, "loss": 0.2551735043525696, "lr": 1.8002841004812677e-05, "epoch": 0.8537062937062937, "percentage": 28.45, "elapsed_time": "3:02:27", "remaining_time": "7:38:52"} +{"current_steps": 1527, "total_steps": 5364, "loss": 0.4101211428642273, "lr": 1.799893675167104e-05, "epoch": 0.8542657342657343, "percentage": 28.47, "elapsed_time": "3:02:34", "remaining_time": "7:38:44"} +{"current_steps": 1528, "total_steps": 5364, "loss": 0.1648799180984497, "lr": 1.7995029110264207e-05, "epoch": 0.8548251748251748, "percentage": 28.49, "elapsed_time": "3:02:41", "remaining_time": "7:38:37"} +{"current_steps": 1529, "total_steps": 5364, "loss": 0.10228888690471649, "lr": 1.7991118082247405e-05, "epoch": 0.8553846153846154, "percentage": 28.5, "elapsed_time": "3:02:47", "remaining_time": "7:38:28"} +{"current_steps": 1530, "total_steps": 5364, "loss": 0.2177266925573349, "lr": 1.7987203669277314e-05, "epoch": 0.855944055944056, "percentage": 28.52, "elapsed_time": "3:02:54", "remaining_time": "7:38:20"} +{"current_steps": 1531, "total_steps": 5364, "loss": 0.1733018308877945, "lr": 1.7983285873012027e-05, "epoch": 0.8565034965034966, "percentage": 28.54, "elapsed_time": "3:03:01", "remaining_time": "7:38:12"} +{"current_steps": 1532, "total_steps": 5364, "loss": 0.16278310120105743, "lr": 1.797936469511109e-05, "epoch": 0.857062937062937, "percentage": 28.56, "elapsed_time": "3:03:08", "remaining_time": "7:38:04"} +{"current_steps": 1533, "total_steps": 5364, "loss": 0.14814062416553497, "lr": 1.7975440137235475e-05, "epoch": 0.8576223776223776, "percentage": 28.58, "elapsed_time": "3:03:15", "remaining_time": "7:37:56"} +{"current_steps": 1534, "total_steps": 5364, "loss": 0.26308244466781616, "lr": 1.7971512201047575e-05, "epoch": 0.8581818181818182, "percentage": 28.6, "elapsed_time": "3:03:22", "remaining_time": "7:37:49"} +{"current_steps": 1535, "total_steps": 5364, "loss": 0.3105056583881378, "lr": 1.7967580888211226e-05, "epoch": 0.8587412587412587, "percentage": 28.62, "elapsed_time": "3:03:29", "remaining_time": "7:37:41"} +{"current_steps": 1536, "total_steps": 5364, "loss": 0.2541734576225281, "lr": 1.7963646200391693e-05, "epoch": 0.8593006993006993, "percentage": 28.64, "elapsed_time": "3:03:36", "remaining_time": "7:37:33"} +{"current_steps": 1537, "total_steps": 5364, "loss": 0.29362165927886963, "lr": 1.7959708139255658e-05, "epoch": 0.8598601398601399, "percentage": 28.65, "elapsed_time": "3:03:43", "remaining_time": "7:37:26"} +{"current_steps": 1538, "total_steps": 5364, "loss": 0.14905357360839844, "lr": 1.7955766706471257e-05, "epoch": 0.8604195804195804, "percentage": 28.67, "elapsed_time": "3:03:49", "remaining_time": "7:37:18"} +{"current_steps": 1539, "total_steps": 5364, "loss": 0.48252275586128235, "lr": 1.7951821903708027e-05, "epoch": 0.860979020979021, "percentage": 28.69, "elapsed_time": "3:03:57", "remaining_time": "7:37:11"} +{"current_steps": 1540, "total_steps": 5364, "loss": 0.09573337435722351, "lr": 1.7947873732636954e-05, "epoch": 0.8615384615384616, "percentage": 28.71, "elapsed_time": "3:04:04", "remaining_time": "7:37:04"} +{"current_steps": 1541, "total_steps": 5364, "loss": 0.09844426810741425, "lr": 1.7943922194930436e-05, "epoch": 0.862097902097902, "percentage": 28.73, "elapsed_time": "3:04:11", "remaining_time": "7:36:57"} +{"current_steps": 1542, "total_steps": 5364, "loss": 0.46954768896102905, "lr": 1.7939967292262305e-05, "epoch": 0.8626573426573426, "percentage": 28.75, "elapsed_time": "3:04:18", "remaining_time": "7:36:49"} +{"current_steps": 1543, "total_steps": 5364, "loss": 0.21179917454719543, "lr": 1.7936009026307814e-05, "epoch": 0.8632167832167832, "percentage": 28.77, "elapsed_time": "3:04:25", "remaining_time": "7:36:41"} +{"current_steps": 1544, "total_steps": 5364, "loss": 0.09582630544900894, "lr": 1.7932047398743642e-05, "epoch": 0.8637762237762238, "percentage": 28.78, "elapsed_time": "3:04:32", "remaining_time": "7:36:35"} +{"current_steps": 1545, "total_steps": 5364, "loss": 0.11422031372785568, "lr": 1.79280824112479e-05, "epoch": 0.8643356643356643, "percentage": 28.8, "elapsed_time": "3:04:39", "remaining_time": "7:36:27"} +{"current_steps": 1546, "total_steps": 5364, "loss": 0.054923612624406815, "lr": 1.7924114065500107e-05, "epoch": 0.8648951048951049, "percentage": 28.82, "elapsed_time": "3:04:46", "remaining_time": "7:36:19"} +{"current_steps": 1547, "total_steps": 5364, "loss": 0.4437777101993561, "lr": 1.7920142363181212e-05, "epoch": 0.8654545454545455, "percentage": 28.84, "elapsed_time": "3:04:53", "remaining_time": "7:36:11"} +{"current_steps": 1548, "total_steps": 5364, "loss": 0.26439574360847473, "lr": 1.7916167305973592e-05, "epoch": 0.866013986013986, "percentage": 28.86, "elapsed_time": "3:05:00", "remaining_time": "7:36:03"} +{"current_steps": 1549, "total_steps": 5364, "loss": 0.19346833229064941, "lr": 1.7912188895561035e-05, "epoch": 0.8665734265734266, "percentage": 28.88, "elapsed_time": "3:05:07", "remaining_time": "7:35:55"} +{"current_steps": 1550, "total_steps": 5364, "loss": 0.25684428215026855, "lr": 1.7908207133628756e-05, "epoch": 0.8671328671328671, "percentage": 28.9, "elapsed_time": "3:05:14", "remaining_time": "7:35:47"} +{"current_steps": 1551, "total_steps": 5364, "loss": 0.3001222312450409, "lr": 1.7904222021863385e-05, "epoch": 0.8676923076923077, "percentage": 28.91, "elapsed_time": "3:05:20", "remaining_time": "7:35:39"} +{"current_steps": 1552, "total_steps": 5364, "loss": 0.27886486053466797, "lr": 1.7900233561952976e-05, "epoch": 0.8682517482517482, "percentage": 28.93, "elapsed_time": "3:05:28", "remaining_time": "7:35:32"} +{"current_steps": 1553, "total_steps": 5364, "loss": 0.17085976898670197, "lr": 1.7896241755586994e-05, "epoch": 0.8688111888111888, "percentage": 28.95, "elapsed_time": "3:05:35", "remaining_time": "7:35:24"} +{"current_steps": 1554, "total_steps": 5364, "loss": 0.10421441495418549, "lr": 1.7892246604456335e-05, "epoch": 0.8693706293706294, "percentage": 28.97, "elapsed_time": "3:05:42", "remaining_time": "7:35:18"} +{"current_steps": 1555, "total_steps": 5364, "loss": 0.19437983632087708, "lr": 1.7888248110253292e-05, "epoch": 0.8699300699300699, "percentage": 28.99, "elapsed_time": "3:05:49", "remaining_time": "7:35:10"} +{"current_steps": 1556, "total_steps": 5364, "loss": 0.11298567801713943, "lr": 1.7884246274671592e-05, "epoch": 0.8704895104895105, "percentage": 29.01, "elapsed_time": "3:05:56", "remaining_time": "7:35:03"} +{"current_steps": 1557, "total_steps": 5364, "loss": 0.1516723930835724, "lr": 1.7880241099406368e-05, "epoch": 0.8710489510489511, "percentage": 29.03, "elapsed_time": "3:06:03", "remaining_time": "7:34:55"} +{"current_steps": 1558, "total_steps": 5364, "loss": 0.29406455159187317, "lr": 1.787623258615417e-05, "epoch": 0.8716083916083917, "percentage": 29.05, "elapsed_time": "3:06:10", "remaining_time": "7:34:49"} +{"current_steps": 1559, "total_steps": 5364, "loss": 0.10896499454975128, "lr": 1.7872220736612963e-05, "epoch": 0.8721678321678321, "percentage": 29.06, "elapsed_time": "3:06:17", "remaining_time": "7:34:40"} +{"current_steps": 1560, "total_steps": 5364, "loss": 0.21315889060497284, "lr": 1.7868205552482122e-05, "epoch": 0.8727272727272727, "percentage": 29.08, "elapsed_time": "3:06:24", "remaining_time": "7:34:32"} +{"current_steps": 1561, "total_steps": 5364, "loss": 0.27381691336631775, "lr": 1.786418703546244e-05, "epoch": 0.8732867132867133, "percentage": 29.1, "elapsed_time": "3:06:31", "remaining_time": "7:34:24"} +{"current_steps": 1562, "total_steps": 5364, "loss": 0.2666417956352234, "lr": 1.786016518725611e-05, "epoch": 0.8738461538461538, "percentage": 29.12, "elapsed_time": "3:06:38", "remaining_time": "7:34:18"} +{"current_steps": 1563, "total_steps": 5364, "loss": 0.18479011952877045, "lr": 1.785614000956676e-05, "epoch": 0.8744055944055944, "percentage": 29.14, "elapsed_time": "3:06:45", "remaining_time": "7:34:11"} +{"current_steps": 1564, "total_steps": 5364, "loss": 0.20558267831802368, "lr": 1.7852111504099395e-05, "epoch": 0.874965034965035, "percentage": 29.16, "elapsed_time": "3:06:52", "remaining_time": "7:34:03"} +{"current_steps": 1565, "total_steps": 5364, "loss": 0.1752689778804779, "lr": 1.7848079672560462e-05, "epoch": 0.8755244755244755, "percentage": 29.18, "elapsed_time": "3:06:59", "remaining_time": "7:33:56"} +{"current_steps": 1566, "total_steps": 5364, "loss": 0.12320493161678314, "lr": 1.7844044516657796e-05, "epoch": 0.8760839160839161, "percentage": 29.19, "elapsed_time": "3:07:07", "remaining_time": "7:33:49"} +{"current_steps": 1567, "total_steps": 5364, "loss": 0.3106265068054199, "lr": 1.7840006038100645e-05, "epoch": 0.8766433566433567, "percentage": 29.21, "elapsed_time": "3:07:14", "remaining_time": "7:33:41"} +{"current_steps": 1568, "total_steps": 5364, "loss": 0.3688770830631256, "lr": 1.7835964238599666e-05, "epoch": 0.8772027972027973, "percentage": 29.23, "elapsed_time": "3:07:20", "remaining_time": "7:33:33"} +{"current_steps": 1569, "total_steps": 5364, "loss": 0.06713690608739853, "lr": 1.783191911986693e-05, "epoch": 0.8777622377622377, "percentage": 29.25, "elapsed_time": "3:07:28", "remaining_time": "7:33:26"} +{"current_steps": 1570, "total_steps": 5364, "loss": 0.2924523651599884, "lr": 1.7827870683615894e-05, "epoch": 0.8783216783216783, "percentage": 29.27, "elapsed_time": "3:07:35", "remaining_time": "7:33:19"} +{"current_steps": 1571, "total_steps": 5364, "loss": 0.21796932816505432, "lr": 1.7823818931561444e-05, "epoch": 0.8788811188811189, "percentage": 29.29, "elapsed_time": "3:07:42", "remaining_time": "7:33:11"} +{"current_steps": 1572, "total_steps": 5364, "loss": 0.11356610804796219, "lr": 1.781976386541985e-05, "epoch": 0.8794405594405594, "percentage": 29.31, "elapsed_time": "3:07:49", "remaining_time": "7:33:03"} +{"current_steps": 1573, "total_steps": 5364, "loss": 0.24290059506893158, "lr": 1.781570548690881e-05, "epoch": 0.88, "percentage": 29.33, "elapsed_time": "3:07:56", "remaining_time": "7:32:56"} +{"current_steps": 1574, "total_steps": 5364, "loss": 0.13034270703792572, "lr": 1.781164379774739e-05, "epoch": 0.8805594405594406, "percentage": 29.34, "elapsed_time": "3:08:03", "remaining_time": "7:32:50"} +{"current_steps": 1575, "total_steps": 5364, "loss": 0.07894525676965714, "lr": 1.78075787996561e-05, "epoch": 0.8811188811188811, "percentage": 29.36, "elapsed_time": "3:08:10", "remaining_time": "7:32:42"} +{"current_steps": 1576, "total_steps": 5364, "loss": 0.13240502774715424, "lr": 1.780351049435681e-05, "epoch": 0.8816783216783217, "percentage": 29.38, "elapsed_time": "3:08:17", "remaining_time": "7:32:35"} +{"current_steps": 1577, "total_steps": 5364, "loss": 0.21378064155578613, "lr": 1.7799438883572825e-05, "epoch": 0.8822377622377623, "percentage": 29.4, "elapsed_time": "3:08:24", "remaining_time": "7:32:27"} +{"current_steps": 1578, "total_steps": 5364, "loss": 0.1334388554096222, "lr": 1.7795363969028834e-05, "epoch": 0.8827972027972028, "percentage": 29.42, "elapsed_time": "3:08:31", "remaining_time": "7:32:19"} +{"current_steps": 1579, "total_steps": 5364, "loss": 0.2582058906555176, "lr": 1.7791285752450923e-05, "epoch": 0.8833566433566433, "percentage": 29.44, "elapsed_time": "3:08:38", "remaining_time": "7:32:11"} +{"current_steps": 1580, "total_steps": 5364, "loss": 0.4457191526889801, "lr": 1.7787204235566587e-05, "epoch": 0.8839160839160839, "percentage": 29.46, "elapsed_time": "3:08:45", "remaining_time": "7:32:04"} +{"current_steps": 1581, "total_steps": 5364, "loss": 0.12105165421962738, "lr": 1.7783119420104707e-05, "epoch": 0.8844755244755245, "percentage": 29.47, "elapsed_time": "3:08:52", "remaining_time": "7:31:56"} +{"current_steps": 1582, "total_steps": 5364, "loss": 0.2794908881187439, "lr": 1.7779031307795576e-05, "epoch": 0.885034965034965, "percentage": 29.49, "elapsed_time": "3:08:59", "remaining_time": "7:31:49"} +{"current_steps": 1583, "total_steps": 5364, "loss": 0.3630887269973755, "lr": 1.7774939900370868e-05, "epoch": 0.8855944055944056, "percentage": 29.51, "elapsed_time": "3:09:06", "remaining_time": "7:31:41"} +{"current_steps": 1584, "total_steps": 5364, "loss": 0.44180774688720703, "lr": 1.7770845199563665e-05, "epoch": 0.8861538461538462, "percentage": 29.53, "elapsed_time": "3:09:13", "remaining_time": "7:31:33"} +{"current_steps": 1585, "total_steps": 5364, "loss": 0.13922330737113953, "lr": 1.776674720710844e-05, "epoch": 0.8867132867132868, "percentage": 29.55, "elapsed_time": "3:09:20", "remaining_time": "7:31:26"} +{"current_steps": 1586, "total_steps": 5364, "loss": 0.01429430115967989, "lr": 1.7762645924741052e-05, "epoch": 0.8872727272727273, "percentage": 29.57, "elapsed_time": "3:09:27", "remaining_time": "7:31:19"} +{"current_steps": 1587, "total_steps": 5364, "loss": 0.4742003083229065, "lr": 1.7758541354198765e-05, "epoch": 0.8878321678321678, "percentage": 29.59, "elapsed_time": "3:09:34", "remaining_time": "7:31:11"} +{"current_steps": 1588, "total_steps": 5364, "loss": 0.19136425852775574, "lr": 1.7754433497220233e-05, "epoch": 0.8883916083916084, "percentage": 29.6, "elapsed_time": "3:09:42", "remaining_time": "7:31:05"} +{"current_steps": 1589, "total_steps": 5364, "loss": 0.1977563202381134, "lr": 1.77503223555455e-05, "epoch": 0.8889510489510489, "percentage": 29.62, "elapsed_time": "3:09:50", "remaining_time": "7:30:59"} +{"current_steps": 1590, "total_steps": 5364, "loss": 0.18498551845550537, "lr": 1.7746207930916005e-05, "epoch": 0.8895104895104895, "percentage": 29.64, "elapsed_time": "3:09:57", "remaining_time": "7:30:53"} +{"current_steps": 1591, "total_steps": 5364, "loss": 0.12361852824687958, "lr": 1.7742090225074567e-05, "epoch": 0.8900699300699301, "percentage": 29.66, "elapsed_time": "3:10:05", "remaining_time": "7:30:47"} +{"current_steps": 1592, "total_steps": 5364, "loss": 0.2083572894334793, "lr": 1.7737969239765412e-05, "epoch": 0.8906293706293706, "percentage": 29.68, "elapsed_time": "3:10:13", "remaining_time": "7:30:42"} +{"current_steps": 1593, "total_steps": 5364, "loss": 0.3200020492076874, "lr": 1.773384497673414e-05, "epoch": 0.8911888111888112, "percentage": 29.7, "elapsed_time": "3:10:20", "remaining_time": "7:30:34"} +{"current_steps": 1594, "total_steps": 5364, "loss": 0.12671269476413727, "lr": 1.7729717437727746e-05, "epoch": 0.8917482517482518, "percentage": 29.72, "elapsed_time": "3:10:27", "remaining_time": "7:30:27"} +{"current_steps": 1595, "total_steps": 5364, "loss": 0.2997719943523407, "lr": 1.7725586624494614e-05, "epoch": 0.8923076923076924, "percentage": 29.74, "elapsed_time": "3:10:34", "remaining_time": "7:30:20"} +{"current_steps": 1596, "total_steps": 5364, "loss": 0.2988595962524414, "lr": 1.772145253878451e-05, "epoch": 0.8928671328671328, "percentage": 29.75, "elapsed_time": "3:10:41", "remaining_time": "7:30:12"} +{"current_steps": 1597, "total_steps": 5364, "loss": 0.14170129597187042, "lr": 1.7717315182348597e-05, "epoch": 0.8934265734265734, "percentage": 29.77, "elapsed_time": "3:10:48", "remaining_time": "7:30:04"} +{"current_steps": 1598, "total_steps": 5364, "loss": 0.14661243557929993, "lr": 1.7713174556939403e-05, "epoch": 0.893986013986014, "percentage": 29.79, "elapsed_time": "3:10:55", "remaining_time": "7:29:58"} +{"current_steps": 1599, "total_steps": 5364, "loss": 0.31579482555389404, "lr": 1.7709030664310862e-05, "epoch": 0.8945454545454545, "percentage": 29.81, "elapsed_time": "3:11:02", "remaining_time": "7:29:50"} +{"current_steps": 1600, "total_steps": 5364, "loss": 0.17461243271827698, "lr": 1.770488350621828e-05, "epoch": 0.8951048951048951, "percentage": 29.83, "elapsed_time": "3:11:09", "remaining_time": "7:29:42"} +{"current_steps": 1601, "total_steps": 5364, "loss": 0.3058391213417053, "lr": 1.7700733084418355e-05, "epoch": 0.8956643356643357, "percentage": 29.85, "elapsed_time": "3:11:17", "remaining_time": "7:29:35"} +{"current_steps": 1602, "total_steps": 5364, "loss": 0.2317260205745697, "lr": 1.769657940066915e-05, "epoch": 0.8962237762237762, "percentage": 29.87, "elapsed_time": "3:11:24", "remaining_time": "7:29:28"} +{"current_steps": 1603, "total_steps": 5364, "loss": 0.16295269131660461, "lr": 1.7692422456730133e-05, "epoch": 0.8967832167832168, "percentage": 29.88, "elapsed_time": "3:11:31", "remaining_time": "7:29:21"} +{"current_steps": 1604, "total_steps": 5364, "loss": 0.22177726030349731, "lr": 1.7688262254362135e-05, "epoch": 0.8973426573426574, "percentage": 29.9, "elapsed_time": "3:11:38", "remaining_time": "7:29:13"} +{"current_steps": 1605, "total_steps": 5364, "loss": 0.2447076141834259, "lr": 1.768409879532738e-05, "epoch": 0.8979020979020979, "percentage": 29.92, "elapsed_time": "3:11:45", "remaining_time": "7:29:07"} +{"current_steps": 1606, "total_steps": 5364, "loss": 0.2966054677963257, "lr": 1.767993208138946e-05, "epoch": 0.8984615384615384, "percentage": 29.94, "elapsed_time": "3:11:53", "remaining_time": "7:29:00"} +{"current_steps": 1607, "total_steps": 5364, "loss": 0.37276527285575867, "lr": 1.7675762114313348e-05, "epoch": 0.899020979020979, "percentage": 29.96, "elapsed_time": "3:12:00", "remaining_time": "7:28:52"} +{"current_steps": 1608, "total_steps": 5364, "loss": 0.2815719246864319, "lr": 1.767158889586541e-05, "epoch": 0.8995804195804196, "percentage": 29.98, "elapsed_time": "3:12:07", "remaining_time": "7:28:45"} +{"current_steps": 1609, "total_steps": 5364, "loss": 0.12265042215585709, "lr": 1.7667412427813362e-05, "epoch": 0.9001398601398601, "percentage": 30.0, "elapsed_time": "3:12:13", "remaining_time": "7:28:37"} +{"current_steps": 1610, "total_steps": 5364, "loss": 0.08063453435897827, "lr": 1.7663232711926325e-05, "epoch": 0.9006993006993007, "percentage": 30.01, "elapsed_time": "3:12:20", "remaining_time": "7:28:29"} +{"current_steps": 1611, "total_steps": 5364, "loss": 0.02443072944879532, "lr": 1.7659049749974778e-05, "epoch": 0.9012587412587413, "percentage": 30.03, "elapsed_time": "3:12:28", "remaining_time": "7:28:23"} +{"current_steps": 1612, "total_steps": 5364, "loss": 0.21359990537166595, "lr": 1.7654863543730576e-05, "epoch": 0.9018181818181819, "percentage": 30.05, "elapsed_time": "3:12:35", "remaining_time": "7:28:16"} +{"current_steps": 1613, "total_steps": 5364, "loss": 0.2204931527376175, "lr": 1.7650674094966955e-05, "epoch": 0.9023776223776224, "percentage": 30.07, "elapsed_time": "3:12:42", "remaining_time": "7:28:09"} +{"current_steps": 1614, "total_steps": 5364, "loss": 0.08348868787288666, "lr": 1.7646481405458523e-05, "epoch": 0.9029370629370629, "percentage": 30.09, "elapsed_time": "3:12:50", "remaining_time": "7:28:02"} +{"current_steps": 1615, "total_steps": 5364, "loss": 0.1931324303150177, "lr": 1.7642285476981252e-05, "epoch": 0.9034965034965035, "percentage": 30.11, "elapsed_time": "3:12:57", "remaining_time": "7:27:56"} +{"current_steps": 1616, "total_steps": 5364, "loss": 0.46750667691230774, "lr": 1.76380863113125e-05, "epoch": 0.904055944055944, "percentage": 30.13, "elapsed_time": "3:13:04", "remaining_time": "7:27:48"} +{"current_steps": 1617, "total_steps": 5364, "loss": 0.19260993599891663, "lr": 1.763388391023099e-05, "epoch": 0.9046153846153846, "percentage": 30.15, "elapsed_time": "3:13:11", "remaining_time": "7:27:40"} +{"current_steps": 1618, "total_steps": 5364, "loss": 0.18408820033073425, "lr": 1.762967827551681e-05, "epoch": 0.9051748251748252, "percentage": 30.16, "elapsed_time": "3:13:19", "remaining_time": "7:27:34"} +{"current_steps": 1619, "total_steps": 5364, "loss": 0.20373623073101044, "lr": 1.7625469408951425e-05, "epoch": 0.9057342657342657, "percentage": 30.18, "elapsed_time": "3:13:26", "remaining_time": "7:27:26"} +{"current_steps": 1620, "total_steps": 5364, "loss": 0.14490383863449097, "lr": 1.7621257312317666e-05, "epoch": 0.9062937062937063, "percentage": 30.2, "elapsed_time": "3:13:32", "remaining_time": "7:27:18"} +{"current_steps": 1621, "total_steps": 5364, "loss": 0.07406635582447052, "lr": 1.761704198739973e-05, "epoch": 0.9068531468531469, "percentage": 30.22, "elapsed_time": "3:13:39", "remaining_time": "7:27:11"} +{"current_steps": 1622, "total_steps": 5364, "loss": 0.07868371903896332, "lr": 1.7612823435983187e-05, "epoch": 0.9074125874125875, "percentage": 30.24, "elapsed_time": "3:13:46", "remaining_time": "7:27:03"} +{"current_steps": 1623, "total_steps": 5364, "loss": 0.18978124856948853, "lr": 1.7608601659854975e-05, "epoch": 0.9079720279720279, "percentage": 30.26, "elapsed_time": "3:13:54", "remaining_time": "7:26:57"} +{"current_steps": 1624, "total_steps": 5364, "loss": 0.12689727544784546, "lr": 1.7604376660803387e-05, "epoch": 0.9085314685314685, "percentage": 30.28, "elapsed_time": "3:14:01", "remaining_time": "7:26:49"} +{"current_steps": 1625, "total_steps": 5364, "loss": 0.0913744792342186, "lr": 1.760014844061809e-05, "epoch": 0.9090909090909091, "percentage": 30.29, "elapsed_time": "3:14:08", "remaining_time": "7:26:42"} +{"current_steps": 1626, "total_steps": 5364, "loss": 0.12943418323993683, "lr": 1.7595917001090113e-05, "epoch": 0.9096503496503496, "percentage": 30.31, "elapsed_time": "3:14:15", "remaining_time": "7:26:34"} +{"current_steps": 1627, "total_steps": 5364, "loss": 0.08174778521060944, "lr": 1.759168234401185e-05, "epoch": 0.9102097902097902, "percentage": 30.33, "elapsed_time": "3:14:22", "remaining_time": "7:26:26"} +{"current_steps": 1628, "total_steps": 5364, "loss": 0.10777575522661209, "lr": 1.758744447117706e-05, "epoch": 0.9107692307692308, "percentage": 30.35, "elapsed_time": "3:14:29", "remaining_time": "7:26:19"} +{"current_steps": 1629, "total_steps": 5364, "loss": 0.18899346888065338, "lr": 1.7583203384380857e-05, "epoch": 0.9113286713286713, "percentage": 30.37, "elapsed_time": "3:14:36", "remaining_time": "7:26:12"} +{"current_steps": 1630, "total_steps": 5364, "loss": 0.474851131439209, "lr": 1.757895908541972e-05, "epoch": 0.9118881118881119, "percentage": 30.39, "elapsed_time": "3:14:43", "remaining_time": "7:26:04"} +{"current_steps": 1631, "total_steps": 5364, "loss": 0.1819888949394226, "lr": 1.757471157609149e-05, "epoch": 0.9124475524475525, "percentage": 30.41, "elapsed_time": "3:14:50", "remaining_time": "7:25:57"} +{"current_steps": 1632, "total_steps": 5364, "loss": 0.2196742594242096, "lr": 1.7570460858195372e-05, "epoch": 0.9130069930069931, "percentage": 30.43, "elapsed_time": "3:14:57", "remaining_time": "7:25:49"} +{"current_steps": 1633, "total_steps": 5364, "loss": 0.14556241035461426, "lr": 1.756620693353192e-05, "epoch": 0.9135664335664335, "percentage": 30.44, "elapsed_time": "3:15:05", "remaining_time": "7:25:43"} +{"current_steps": 1634, "total_steps": 5364, "loss": 0.04626625403761864, "lr": 1.7561949803903052e-05, "epoch": 0.9141258741258741, "percentage": 30.46, "elapsed_time": "3:15:12", "remaining_time": "7:25:36"} +{"current_steps": 1635, "total_steps": 5364, "loss": 0.17609430849552155, "lr": 1.7557689471112044e-05, "epoch": 0.9146853146853147, "percentage": 30.48, "elapsed_time": "3:15:19", "remaining_time": "7:25:28"} +{"current_steps": 1636, "total_steps": 5364, "loss": 0.19454896450042725, "lr": 1.7553425936963532e-05, "epoch": 0.9152447552447552, "percentage": 30.5, "elapsed_time": "3:15:26", "remaining_time": "7:25:20"} +{"current_steps": 1637, "total_steps": 5364, "loss": 0.26230350136756897, "lr": 1.75491592032635e-05, "epoch": 0.9158041958041958, "percentage": 30.52, "elapsed_time": "3:15:33", "remaining_time": "7:25:13"} +{"current_steps": 1638, "total_steps": 5364, "loss": 0.34506380558013916, "lr": 1.7544889271819285e-05, "epoch": 0.9163636363636364, "percentage": 30.54, "elapsed_time": "3:15:40", "remaining_time": "7:25:06"} +{"current_steps": 1639, "total_steps": 5364, "loss": 0.18003425002098083, "lr": 1.75406161444396e-05, "epoch": 0.916923076923077, "percentage": 30.56, "elapsed_time": "3:15:47", "remaining_time": "7:24:58"} +{"current_steps": 1640, "total_steps": 5364, "loss": 0.4571908116340637, "lr": 1.7536339822934484e-05, "epoch": 0.9174825174825175, "percentage": 30.57, "elapsed_time": "3:15:54", "remaining_time": "7:24:50"} +{"current_steps": 1641, "total_steps": 5364, "loss": 0.27332603931427, "lr": 1.7532060309115345e-05, "epoch": 0.9180419580419581, "percentage": 30.59, "elapsed_time": "3:16:01", "remaining_time": "7:24:44"} +{"current_steps": 1642, "total_steps": 5364, "loss": 0.2583649754524231, "lr": 1.7527777604794944e-05, "epoch": 0.9186013986013986, "percentage": 30.61, "elapsed_time": "3:16:09", "remaining_time": "7:24:37"} +{"current_steps": 1643, "total_steps": 5364, "loss": 0.36741504073143005, "lr": 1.7523491711787384e-05, "epoch": 0.9191608391608391, "percentage": 30.63, "elapsed_time": "3:16:16", "remaining_time": "7:24:29"} +{"current_steps": 1644, "total_steps": 5364, "loss": 0.13514187932014465, "lr": 1.7519202631908126e-05, "epoch": 0.9197202797202797, "percentage": 30.65, "elapsed_time": "3:16:22", "remaining_time": "7:24:22"} +{"current_steps": 1645, "total_steps": 5364, "loss": 0.18971621990203857, "lr": 1.751491036697398e-05, "epoch": 0.9202797202797203, "percentage": 30.67, "elapsed_time": "3:16:29", "remaining_time": "7:24:14"} +{"current_steps": 1646, "total_steps": 5364, "loss": 0.27957165241241455, "lr": 1.7510614918803102e-05, "epoch": 0.9208391608391608, "percentage": 30.69, "elapsed_time": "3:16:36", "remaining_time": "7:24:06"} +{"current_steps": 1647, "total_steps": 5364, "loss": 0.06857086718082428, "lr": 1.7506316289215005e-05, "epoch": 0.9213986013986014, "percentage": 30.7, "elapsed_time": "3:16:44", "remaining_time": "7:24:00"} +{"current_steps": 1648, "total_steps": 5364, "loss": 0.4726221561431885, "lr": 1.7502014480030533e-05, "epoch": 0.921958041958042, "percentage": 30.72, "elapsed_time": "3:16:52", "remaining_time": "7:23:54"} +{"current_steps": 1649, "total_steps": 5364, "loss": 0.29996445775032043, "lr": 1.7497709493071893e-05, "epoch": 0.9225174825174826, "percentage": 30.74, "elapsed_time": "3:16:59", "remaining_time": "7:23:47"} +{"current_steps": 1650, "total_steps": 5364, "loss": 0.33131396770477295, "lr": 1.749340133016263e-05, "epoch": 0.9230769230769231, "percentage": 30.76, "elapsed_time": "3:17:06", "remaining_time": "7:23:39"} +{"current_steps": 1651, "total_steps": 5364, "loss": 0.10618998110294342, "lr": 1.748908999312764e-05, "epoch": 0.9236363636363636, "percentage": 30.78, "elapsed_time": "3:17:14", "remaining_time": "7:23:34"} +{"current_steps": 1652, "total_steps": 5364, "loss": 0.2302563637495041, "lr": 1.7484775483793158e-05, "epoch": 0.9241958041958042, "percentage": 30.8, "elapsed_time": "3:17:21", "remaining_time": "7:23:27"} +{"current_steps": 1653, "total_steps": 5364, "loss": 0.4540604054927826, "lr": 1.7480457803986762e-05, "epoch": 0.9247552447552447, "percentage": 30.82, "elapsed_time": "3:17:28", "remaining_time": "7:23:20"} +{"current_steps": 1654, "total_steps": 5364, "loss": 0.349995881319046, "lr": 1.747613695553738e-05, "epoch": 0.9253146853146853, "percentage": 30.84, "elapsed_time": "3:17:35", "remaining_time": "7:23:13"} +{"current_steps": 1655, "total_steps": 5364, "loss": 0.284639835357666, "lr": 1.747181294027527e-05, "epoch": 0.9258741258741259, "percentage": 30.85, "elapsed_time": "3:17:43", "remaining_time": "7:23:06"} +{"current_steps": 1656, "total_steps": 5364, "loss": 0.19863250851631165, "lr": 1.746748576003205e-05, "epoch": 0.9264335664335664, "percentage": 30.87, "elapsed_time": "3:17:50", "remaining_time": "7:22:59"} +{"current_steps": 1657, "total_steps": 5364, "loss": 0.03898322209715843, "lr": 1.746315541664066e-05, "epoch": 0.926993006993007, "percentage": 30.89, "elapsed_time": "3:17:57", "remaining_time": "7:22:53"} +{"current_steps": 1658, "total_steps": 5364, "loss": 0.19983133673667908, "lr": 1.7458821911935387e-05, "epoch": 0.9275524475524476, "percentage": 30.91, "elapsed_time": "3:18:05", "remaining_time": "7:22:47"} +{"current_steps": 1659, "total_steps": 5364, "loss": 0.06280991435050964, "lr": 1.7454485247751868e-05, "epoch": 0.9281118881118882, "percentage": 30.93, "elapsed_time": "3:18:13", "remaining_time": "7:22:41"} +{"current_steps": 1660, "total_steps": 5364, "loss": 0.1316758096218109, "lr": 1.7450145425927055e-05, "epoch": 0.9286713286713286, "percentage": 30.95, "elapsed_time": "3:18:20", "remaining_time": "7:22:33"} +{"current_steps": 1661, "total_steps": 5364, "loss": 0.14309416711330414, "lr": 1.744580244829926e-05, "epoch": 0.9292307692307692, "percentage": 30.97, "elapsed_time": "3:18:27", "remaining_time": "7:22:25"} +{"current_steps": 1662, "total_steps": 5364, "loss": 0.17261527478694916, "lr": 1.7441456316708118e-05, "epoch": 0.9297902097902098, "percentage": 30.98, "elapsed_time": "3:18:34", "remaining_time": "7:22:19"} +{"current_steps": 1663, "total_steps": 5364, "loss": 0.20808139443397522, "lr": 1.743710703299461e-05, "epoch": 0.9303496503496503, "percentage": 31.0, "elapsed_time": "3:18:42", "remaining_time": "7:22:12"} +{"current_steps": 1664, "total_steps": 5364, "loss": 0.23567354679107666, "lr": 1.7432754599001037e-05, "epoch": 0.9309090909090909, "percentage": 31.02, "elapsed_time": "3:18:49", "remaining_time": "7:22:06"} +{"current_steps": 1665, "total_steps": 5364, "loss": 0.40661072731018066, "lr": 1.7428399016571052e-05, "epoch": 0.9314685314685315, "percentage": 31.04, "elapsed_time": "3:18:56", "remaining_time": "7:21:58"} +{"current_steps": 1666, "total_steps": 5364, "loss": 0.2870525121688843, "lr": 1.742404028754963e-05, "epoch": 0.932027972027972, "percentage": 31.06, "elapsed_time": "3:19:03", "remaining_time": "7:21:50"} +{"current_steps": 1667, "total_steps": 5364, "loss": 0.014616936445236206, "lr": 1.7419678413783088e-05, "epoch": 0.9325874125874126, "percentage": 31.08, "elapsed_time": "3:19:11", "remaining_time": "7:21:44"} +{"current_steps": 1668, "total_steps": 5364, "loss": 0.19322210550308228, "lr": 1.7415313397119066e-05, "epoch": 0.9331468531468532, "percentage": 31.1, "elapsed_time": "3:19:18", "remaining_time": "7:21:38"} +{"current_steps": 1669, "total_steps": 5364, "loss": 0.16863524913787842, "lr": 1.7410945239406533e-05, "epoch": 0.9337062937062937, "percentage": 31.11, "elapsed_time": "3:19:25", "remaining_time": "7:21:30"} +{"current_steps": 1670, "total_steps": 5364, "loss": 0.20633713901042938, "lr": 1.7406573942495803e-05, "epoch": 0.9342657342657342, "percentage": 31.13, "elapsed_time": "3:19:32", "remaining_time": "7:21:22"} +{"current_steps": 1671, "total_steps": 5364, "loss": 0.402421236038208, "lr": 1.740219950823851e-05, "epoch": 0.9348251748251748, "percentage": 31.15, "elapsed_time": "3:19:39", "remaining_time": "7:21:15"} +{"current_steps": 1672, "total_steps": 5364, "loss": 0.3429017663002014, "lr": 1.7397821938487615e-05, "epoch": 0.9353846153846154, "percentage": 31.17, "elapsed_time": "3:19:46", "remaining_time": "7:21:07"} +{"current_steps": 1673, "total_steps": 5364, "loss": 0.08628326654434204, "lr": 1.7393441235097417e-05, "epoch": 0.9359440559440559, "percentage": 31.19, "elapsed_time": "3:19:53", "remaining_time": "7:21:00"} +{"current_steps": 1674, "total_steps": 5364, "loss": 0.1812334507703781, "lr": 1.7389057399923524e-05, "epoch": 0.9365034965034965, "percentage": 31.21, "elapsed_time": "3:20:00", "remaining_time": "7:20:53"} +{"current_steps": 1675, "total_steps": 5364, "loss": 0.23956313729286194, "lr": 1.7384670434822894e-05, "epoch": 0.9370629370629371, "percentage": 31.23, "elapsed_time": "3:20:07", "remaining_time": "7:20:45"} +{"current_steps": 1676, "total_steps": 5364, "loss": 0.19388212263584137, "lr": 1.738028034165379e-05, "epoch": 0.9376223776223777, "percentage": 31.25, "elapsed_time": "3:20:14", "remaining_time": "7:20:37"} +{"current_steps": 1677, "total_steps": 5364, "loss": 0.07962428033351898, "lr": 1.7375887122275818e-05, "epoch": 0.9381818181818182, "percentage": 31.26, "elapsed_time": "3:20:21", "remaining_time": "7:20:30"} +{"current_steps": 1678, "total_steps": 5364, "loss": 0.20845860242843628, "lr": 1.7371490778549894e-05, "epoch": 0.9387412587412587, "percentage": 31.28, "elapsed_time": "3:20:28", "remaining_time": "7:20:22"} +{"current_steps": 1679, "total_steps": 5364, "loss": 0.21996483206748962, "lr": 1.7367091312338263e-05, "epoch": 0.9393006993006993, "percentage": 31.3, "elapsed_time": "3:20:35", "remaining_time": "7:20:15"} +{"current_steps": 1680, "total_steps": 5364, "loss": 0.35951757431030273, "lr": 1.736268872550449e-05, "epoch": 0.9398601398601398, "percentage": 31.32, "elapsed_time": "3:20:42", "remaining_time": "7:20:08"} +{"current_steps": 1681, "total_steps": 5364, "loss": 0.29795268177986145, "lr": 1.7358283019913473e-05, "epoch": 0.9404195804195804, "percentage": 31.34, "elapsed_time": "3:20:49", "remaining_time": "7:20:00"} +{"current_steps": 1682, "total_steps": 5364, "loss": 0.11087039858102798, "lr": 1.7353874197431412e-05, "epoch": 0.940979020979021, "percentage": 31.36, "elapsed_time": "3:20:56", "remaining_time": "7:19:52"} +{"current_steps": 1683, "total_steps": 5364, "loss": 0.10425370931625366, "lr": 1.7349462259925846e-05, "epoch": 0.9415384615384615, "percentage": 31.38, "elapsed_time": "3:21:03", "remaining_time": "7:19:44"} +{"current_steps": 1684, "total_steps": 5364, "loss": 0.12108875811100006, "lr": 1.7345047209265618e-05, "epoch": 0.9420979020979021, "percentage": 31.39, "elapsed_time": "3:21:10", "remaining_time": "7:19:37"} +{"current_steps": 1685, "total_steps": 5364, "loss": 0.32281485199928284, "lr": 1.7340629047320906e-05, "epoch": 0.9426573426573427, "percentage": 31.41, "elapsed_time": "3:21:17", "remaining_time": "7:19:30"} +{"current_steps": 1686, "total_steps": 5364, "loss": 0.12829357385635376, "lr": 1.7336207775963184e-05, "epoch": 0.9432167832167833, "percentage": 31.43, "elapsed_time": "3:21:24", "remaining_time": "7:19:22"} +{"current_steps": 1687, "total_steps": 5364, "loss": 0.19423940777778625, "lr": 1.7331783397065267e-05, "epoch": 0.9437762237762237, "percentage": 31.45, "elapsed_time": "3:21:31", "remaining_time": "7:19:14"} +{"current_steps": 1688, "total_steps": 5364, "loss": 0.09307831525802612, "lr": 1.7327355912501272e-05, "epoch": 0.9443356643356643, "percentage": 31.47, "elapsed_time": "3:21:38", "remaining_time": "7:19:07"} +{"current_steps": 1689, "total_steps": 5364, "loss": 0.20622305572032928, "lr": 1.7322925324146632e-05, "epoch": 0.9448951048951049, "percentage": 31.49, "elapsed_time": "3:21:46", "remaining_time": "7:19:00"} +{"current_steps": 1690, "total_steps": 5364, "loss": 0.16810685396194458, "lr": 1.73184916338781e-05, "epoch": 0.9454545454545454, "percentage": 31.51, "elapsed_time": "3:21:52", "remaining_time": "7:18:53"} +{"current_steps": 1691, "total_steps": 5364, "loss": 0.1370825469493866, "lr": 1.7314054843573744e-05, "epoch": 0.946013986013986, "percentage": 31.52, "elapsed_time": "3:22:00", "remaining_time": "7:18:45"} +{"current_steps": 1692, "total_steps": 5364, "loss": 0.1516319215297699, "lr": 1.7309614955112934e-05, "epoch": 0.9465734265734266, "percentage": 31.54, "elapsed_time": "3:22:06", "remaining_time": "7:18:37"} +{"current_steps": 1693, "total_steps": 5364, "loss": 0.1676393747329712, "lr": 1.7305171970376363e-05, "epoch": 0.9471328671328672, "percentage": 31.56, "elapsed_time": "3:22:13", "remaining_time": "7:18:30"} +{"current_steps": 1694, "total_steps": 5364, "loss": 0.24507887661457062, "lr": 1.7300725891246034e-05, "epoch": 0.9476923076923077, "percentage": 31.58, "elapsed_time": "3:22:21", "remaining_time": "7:18:23"} +{"current_steps": 1695, "total_steps": 5364, "loss": 0.17468124628067017, "lr": 1.729627671960526e-05, "epoch": 0.9482517482517483, "percentage": 31.6, "elapsed_time": "3:22:28", "remaining_time": "7:18:15"} +{"current_steps": 1696, "total_steps": 5364, "loss": 0.32875490188598633, "lr": 1.7291824457338657e-05, "epoch": 0.9488111888111889, "percentage": 31.62, "elapsed_time": "3:22:34", "remaining_time": "7:18:07"} +{"current_steps": 1697, "total_steps": 5364, "loss": 0.27696138620376587, "lr": 1.7287369106332164e-05, "epoch": 0.9493706293706293, "percentage": 31.64, "elapsed_time": "3:22:42", "remaining_time": "7:18:00"} +{"current_steps": 1698, "total_steps": 5364, "loss": 0.15715940296649933, "lr": 1.7282910668473016e-05, "epoch": 0.9499300699300699, "percentage": 31.66, "elapsed_time": "3:22:49", "remaining_time": "7:17:53"} +{"current_steps": 1699, "total_steps": 5364, "loss": 0.27758514881134033, "lr": 1.7278449145649763e-05, "epoch": 0.9504895104895105, "percentage": 31.67, "elapsed_time": "3:22:56", "remaining_time": "7:17:47"} +{"current_steps": 1700, "total_steps": 5364, "loss": 0.21390266716480255, "lr": 1.7273984539752254e-05, "epoch": 0.951048951048951, "percentage": 31.69, "elapsed_time": "3:23:03", "remaining_time": "7:17:39"} +{"current_steps": 1701, "total_steps": 5364, "loss": 0.14004793763160706, "lr": 1.7269516852671654e-05, "epoch": 0.9516083916083916, "percentage": 31.71, "elapsed_time": "3:23:11", "remaining_time": "7:17:33"} +{"current_steps": 1702, "total_steps": 5364, "loss": 0.15599679946899414, "lr": 1.7265046086300427e-05, "epoch": 0.9521678321678322, "percentage": 31.73, "elapsed_time": "3:23:18", "remaining_time": "7:17:25"} +{"current_steps": 1703, "total_steps": 5364, "loss": 0.11955723166465759, "lr": 1.7260572242532345e-05, "epoch": 0.9527272727272728, "percentage": 31.75, "elapsed_time": "3:23:25", "remaining_time": "7:17:17"} +{"current_steps": 1704, "total_steps": 5364, "loss": 0.28404897451400757, "lr": 1.7256095323262476e-05, "epoch": 0.9532867132867133, "percentage": 31.77, "elapsed_time": "3:23:32", "remaining_time": "7:17:10"} +{"current_steps": 1705, "total_steps": 5364, "loss": 0.14715871214866638, "lr": 1.7251615330387194e-05, "epoch": 0.9538461538461539, "percentage": 31.79, "elapsed_time": "3:23:38", "remaining_time": "7:17:02"} +{"current_steps": 1706, "total_steps": 5364, "loss": 0.3182292878627777, "lr": 1.7247132265804184e-05, "epoch": 0.9544055944055944, "percentage": 31.8, "elapsed_time": "3:23:46", "remaining_time": "7:16:55"} +{"current_steps": 1707, "total_steps": 5364, "loss": 0.1009145975112915, "lr": 1.724264613141242e-05, "epoch": 0.9549650349650349, "percentage": 31.82, "elapsed_time": "3:23:53", "remaining_time": "7:16:48"} +{"current_steps": 1708, "total_steps": 5364, "loss": 0.28568771481513977, "lr": 1.7238156929112185e-05, "epoch": 0.9555244755244755, "percentage": 31.84, "elapsed_time": "3:24:00", "remaining_time": "7:16:41"} +{"current_steps": 1709, "total_steps": 5364, "loss": 0.1518329530954361, "lr": 1.7233664660805055e-05, "epoch": 0.9560839160839161, "percentage": 31.86, "elapsed_time": "3:24:08", "remaining_time": "7:16:34"} +{"current_steps": 1710, "total_steps": 5364, "loss": 0.28611087799072266, "lr": 1.7229169328393904e-05, "epoch": 0.9566433566433566, "percentage": 31.88, "elapsed_time": "3:24:14", "remaining_time": "7:16:26"} +{"current_steps": 1711, "total_steps": 5364, "loss": 0.155736044049263, "lr": 1.7224670933782913e-05, "epoch": 0.9572027972027972, "percentage": 31.9, "elapsed_time": "3:24:21", "remaining_time": "7:16:19"} +{"current_steps": 1712, "total_steps": 5364, "loss": 0.07055732607841492, "lr": 1.7220169478877554e-05, "epoch": 0.9577622377622378, "percentage": 31.92, "elapsed_time": "3:24:28", "remaining_time": "7:16:11"} +{"current_steps": 1713, "total_steps": 5364, "loss": 0.12250161170959473, "lr": 1.7215664965584593e-05, "epoch": 0.9583216783216784, "percentage": 31.94, "elapsed_time": "3:24:36", "remaining_time": "7:16:05"} +{"current_steps": 1714, "total_steps": 5364, "loss": 0.2894744873046875, "lr": 1.7211157395812094e-05, "epoch": 0.9588811188811189, "percentage": 31.95, "elapsed_time": "3:24:43", "remaining_time": "7:15:57"} +{"current_steps": 1715, "total_steps": 5364, "loss": 0.2420210987329483, "lr": 1.720664677146942e-05, "epoch": 0.9594405594405594, "percentage": 31.97, "elapsed_time": "3:24:50", "remaining_time": "7:15:49"} +{"current_steps": 1716, "total_steps": 5364, "loss": 0.30071884393692017, "lr": 1.7202133094467223e-05, "epoch": 0.96, "percentage": 31.99, "elapsed_time": "3:24:57", "remaining_time": "7:15:42"} +{"current_steps": 1717, "total_steps": 5364, "loss": 0.06418092548847198, "lr": 1.7197616366717446e-05, "epoch": 0.9605594405594405, "percentage": 32.01, "elapsed_time": "3:25:04", "remaining_time": "7:15:35"} +{"current_steps": 1718, "total_steps": 5364, "loss": 0.23366746306419373, "lr": 1.7193096590133325e-05, "epoch": 0.9611188811188811, "percentage": 32.03, "elapsed_time": "3:25:11", "remaining_time": "7:15:27"} +{"current_steps": 1719, "total_steps": 5364, "loss": 0.13986369967460632, "lr": 1.7188573766629398e-05, "epoch": 0.9616783216783217, "percentage": 32.05, "elapsed_time": "3:25:19", "remaining_time": "7:15:21"} +{"current_steps": 1720, "total_steps": 5364, "loss": 0.10744775831699371, "lr": 1.7184047898121477e-05, "epoch": 0.9622377622377623, "percentage": 32.07, "elapsed_time": "3:25:26", "remaining_time": "7:15:14"} +{"current_steps": 1721, "total_steps": 5364, "loss": 0.18372637033462524, "lr": 1.7179518986526676e-05, "epoch": 0.9627972027972028, "percentage": 32.08, "elapsed_time": "3:25:33", "remaining_time": "7:15:06"} +{"current_steps": 1722, "total_steps": 5364, "loss": 0.17784248292446136, "lr": 1.71749870337634e-05, "epoch": 0.9633566433566434, "percentage": 32.1, "elapsed_time": "3:25:40", "remaining_time": "7:14:59"} +{"current_steps": 1723, "total_steps": 5364, "loss": 0.17018987238407135, "lr": 1.7170452041751328e-05, "epoch": 0.963916083916084, "percentage": 32.12, "elapsed_time": "3:25:47", "remaining_time": "7:14:52"} +{"current_steps": 1724, "total_steps": 5364, "loss": 0.1566029191017151, "lr": 1.716591401241143e-05, "epoch": 0.9644755244755244, "percentage": 32.14, "elapsed_time": "3:25:54", "remaining_time": "7:14:44"} +{"current_steps": 1725, "total_steps": 5364, "loss": 0.2766360640525818, "lr": 1.7161372947665982e-05, "epoch": 0.965034965034965, "percentage": 32.16, "elapsed_time": "3:26:01", "remaining_time": "7:14:36"} +{"current_steps": 1726, "total_steps": 5364, "loss": 0.08040576428174973, "lr": 1.715682884943852e-05, "epoch": 0.9655944055944056, "percentage": 32.18, "elapsed_time": "3:26:08", "remaining_time": "7:14:29"} +{"current_steps": 1727, "total_steps": 5364, "loss": 0.19021615386009216, "lr": 1.715228171965388e-05, "epoch": 0.9661538461538461, "percentage": 32.2, "elapsed_time": "3:26:15", "remaining_time": "7:14:21"} +{"current_steps": 1728, "total_steps": 5364, "loss": 0.204390287399292, "lr": 1.714773156023818e-05, "epoch": 0.9667132867132867, "percentage": 32.21, "elapsed_time": "3:26:22", "remaining_time": "7:14:14"} +{"current_steps": 1729, "total_steps": 5364, "loss": 0.3924446105957031, "lr": 1.7143178373118815e-05, "epoch": 0.9672727272727273, "percentage": 32.23, "elapsed_time": "3:26:29", "remaining_time": "7:14:07"} +{"current_steps": 1730, "total_steps": 5364, "loss": 0.3880074620246887, "lr": 1.713862216022447e-05, "epoch": 0.9678321678321679, "percentage": 32.25, "elapsed_time": "3:26:36", "remaining_time": "7:13:59"} +{"current_steps": 1731, "total_steps": 5364, "loss": 0.22814473509788513, "lr": 1.7134062923485106e-05, "epoch": 0.9683916083916084, "percentage": 32.27, "elapsed_time": "3:26:43", "remaining_time": "7:13:51"} +{"current_steps": 1732, "total_steps": 5364, "loss": 0.2966257631778717, "lr": 1.7129500664831972e-05, "epoch": 0.968951048951049, "percentage": 32.29, "elapsed_time": "3:26:50", "remaining_time": "7:13:43"} +{"current_steps": 1733, "total_steps": 5364, "loss": 0.2558753192424774, "lr": 1.712493538619759e-05, "epoch": 0.9695104895104895, "percentage": 32.31, "elapsed_time": "3:26:56", "remaining_time": "7:13:36"} +{"current_steps": 1734, "total_steps": 5364, "loss": 0.3099202513694763, "lr": 1.7120367089515763e-05, "epoch": 0.97006993006993, "percentage": 32.33, "elapsed_time": "3:27:03", "remaining_time": "7:13:28"} +{"current_steps": 1735, "total_steps": 5364, "loss": 0.09802938997745514, "lr": 1.7115795776721575e-05, "epoch": 0.9706293706293706, "percentage": 32.35, "elapsed_time": "3:27:11", "remaining_time": "7:13:21"} +{"current_steps": 1736, "total_steps": 5364, "loss": 0.10171899944543839, "lr": 1.7111221449751385e-05, "epoch": 0.9711888111888112, "percentage": 32.36, "elapsed_time": "3:27:18", "remaining_time": "7:13:14"} +{"current_steps": 1737, "total_steps": 5364, "loss": 0.12753914296627045, "lr": 1.710664411054283e-05, "epoch": 0.9717482517482517, "percentage": 32.38, "elapsed_time": "3:27:25", "remaining_time": "7:13:06"} +{"current_steps": 1738, "total_steps": 5364, "loss": 0.2561277151107788, "lr": 1.7102063761034823e-05, "epoch": 0.9723076923076923, "percentage": 32.4, "elapsed_time": "3:27:32", "remaining_time": "7:13:00"} +{"current_steps": 1739, "total_steps": 5364, "loss": 0.28080281615257263, "lr": 1.709748040316755e-05, "epoch": 0.9728671328671329, "percentage": 32.42, "elapsed_time": "3:27:39", "remaining_time": "7:12:52"} +{"current_steps": 1740, "total_steps": 5364, "loss": 0.22109544277191162, "lr": 1.7092894038882475e-05, "epoch": 0.9734265734265735, "percentage": 32.44, "elapsed_time": "3:27:46", "remaining_time": "7:12:44"} +{"current_steps": 1741, "total_steps": 5364, "loss": 0.21845363080501556, "lr": 1.708830467012233e-05, "epoch": 0.973986013986014, "percentage": 32.46, "elapsed_time": "3:27:53", "remaining_time": "7:12:37"} +{"current_steps": 1742, "total_steps": 5364, "loss": 0.13033252954483032, "lr": 1.708371229883113e-05, "epoch": 0.9745454545454545, "percentage": 32.48, "elapsed_time": "3:28:00", "remaining_time": "7:12:30"} +{"current_steps": 1743, "total_steps": 5364, "loss": 0.15591920912265778, "lr": 1.7079116926954146e-05, "epoch": 0.9751048951048951, "percentage": 32.49, "elapsed_time": "3:28:08", "remaining_time": "7:12:23"} +{"current_steps": 1744, "total_steps": 5364, "loss": 0.2553369402885437, "lr": 1.707451855643794e-05, "epoch": 0.9756643356643356, "percentage": 32.51, "elapsed_time": "3:28:15", "remaining_time": "7:12:16"} +{"current_steps": 1745, "total_steps": 5364, "loss": 0.3380037546157837, "lr": 1.7069917189230322e-05, "epoch": 0.9762237762237762, "percentage": 32.53, "elapsed_time": "3:28:23", "remaining_time": "7:12:10"} +{"current_steps": 1746, "total_steps": 5364, "loss": 0.0386936292052269, "lr": 1.706531282728039e-05, "epoch": 0.9767832167832168, "percentage": 32.55, "elapsed_time": "3:28:31", "remaining_time": "7:12:05"} +{"current_steps": 1747, "total_steps": 5364, "loss": 0.428783118724823, "lr": 1.70607054725385e-05, "epoch": 0.9773426573426574, "percentage": 32.57, "elapsed_time": "3:28:38", "remaining_time": "7:11:58"} +{"current_steps": 1748, "total_steps": 5364, "loss": 0.2550695538520813, "lr": 1.705609512695628e-05, "epoch": 0.9779020979020979, "percentage": 32.59, "elapsed_time": "3:28:45", "remaining_time": "7:11:50"} +{"current_steps": 1749, "total_steps": 5364, "loss": 0.1777503788471222, "lr": 1.7051481792486627e-05, "epoch": 0.9784615384615385, "percentage": 32.61, "elapsed_time": "3:28:52", "remaining_time": "7:11:43"} +{"current_steps": 1750, "total_steps": 5364, "loss": 0.20227137207984924, "lr": 1.704686547108369e-05, "epoch": 0.9790209790209791, "percentage": 32.62, "elapsed_time": "3:29:00", "remaining_time": "7:11:36"} +{"current_steps": 1751, "total_steps": 5364, "loss": 0.10227454453706741, "lr": 1.7042246164702904e-05, "epoch": 0.9795804195804195, "percentage": 32.64, "elapsed_time": "3:29:07", "remaining_time": "7:11:29"} +{"current_steps": 1752, "total_steps": 5364, "loss": 0.29930809140205383, "lr": 1.7037623875300958e-05, "epoch": 0.9801398601398601, "percentage": 32.66, "elapsed_time": "3:29:14", "remaining_time": "7:11:22"} +{"current_steps": 1753, "total_steps": 5364, "loss": 0.03869568556547165, "lr": 1.70329986048358e-05, "epoch": 0.9806993006993007, "percentage": 32.68, "elapsed_time": "3:29:21", "remaining_time": "7:11:14"} +{"current_steps": 1754, "total_steps": 5364, "loss": 0.17714332044124603, "lr": 1.702837035526665e-05, "epoch": 0.9812587412587412, "percentage": 32.7, "elapsed_time": "3:29:28", "remaining_time": "7:11:07"} +{"current_steps": 1755, "total_steps": 5364, "loss": 0.04542902857065201, "lr": 1.7023739128553988e-05, "epoch": 0.9818181818181818, "percentage": 32.72, "elapsed_time": "3:29:35", "remaining_time": "7:11:00"} +{"current_steps": 1756, "total_steps": 5364, "loss": 0.16884341835975647, "lr": 1.7019104926659542e-05, "epoch": 0.9823776223776224, "percentage": 32.74, "elapsed_time": "3:29:42", "remaining_time": "7:10:52"} +{"current_steps": 1757, "total_steps": 5364, "loss": 0.11557105928659439, "lr": 1.701446775154632e-05, "epoch": 0.982937062937063, "percentage": 32.76, "elapsed_time": "3:29:49", "remaining_time": "7:10:45"} +{"current_steps": 1758, "total_steps": 5364, "loss": 0.2557441294193268, "lr": 1.7009827605178578e-05, "epoch": 0.9834965034965035, "percentage": 32.77, "elapsed_time": "3:29:56", "remaining_time": "7:10:37"} +{"current_steps": 1759, "total_steps": 5364, "loss": 0.18762089312076569, "lr": 1.700518448952184e-05, "epoch": 0.9840559440559441, "percentage": 32.79, "elapsed_time": "3:30:03", "remaining_time": "7:10:29"} +{"current_steps": 1760, "total_steps": 5364, "loss": 0.12941545248031616, "lr": 1.7000538406542863e-05, "epoch": 0.9846153846153847, "percentage": 32.81, "elapsed_time": "3:30:10", "remaining_time": "7:10:22"} +{"current_steps": 1761, "total_steps": 5364, "loss": 0.19491036236286163, "lr": 1.69958893582097e-05, "epoch": 0.9851748251748251, "percentage": 32.83, "elapsed_time": "3:30:17", "remaining_time": "7:10:14"} +{"current_steps": 1762, "total_steps": 5364, "loss": 0.18471840023994446, "lr": 1.6991237346491623e-05, "epoch": 0.9857342657342657, "percentage": 32.85, "elapsed_time": "3:30:23", "remaining_time": "7:10:06"} +{"current_steps": 1763, "total_steps": 5364, "loss": 0.3501903712749481, "lr": 1.6986582373359182e-05, "epoch": 0.9862937062937063, "percentage": 32.87, "elapsed_time": "3:30:31", "remaining_time": "7:10:00"} +{"current_steps": 1764, "total_steps": 5364, "loss": 0.11872689425945282, "lr": 1.6981924440784175e-05, "epoch": 0.9868531468531468, "percentage": 32.89, "elapsed_time": "3:30:38", "remaining_time": "7:09:52"} +{"current_steps": 1765, "total_steps": 5364, "loss": 0.19359099864959717, "lr": 1.697726355073965e-05, "epoch": 0.9874125874125874, "percentage": 32.9, "elapsed_time": "3:30:45", "remaining_time": "7:09:45"} +{"current_steps": 1766, "total_steps": 5364, "loss": 0.13528797030448914, "lr": 1.697259970519991e-05, "epoch": 0.987972027972028, "percentage": 32.92, "elapsed_time": "3:30:52", "remaining_time": "7:09:37"} +{"current_steps": 1767, "total_steps": 5364, "loss": 0.1479535698890686, "lr": 1.6967932906140517e-05, "epoch": 0.9885314685314686, "percentage": 32.94, "elapsed_time": "3:30:59", "remaining_time": "7:09:30"} +{"current_steps": 1768, "total_steps": 5364, "loss": 0.23039215803146362, "lr": 1.696326315553827e-05, "epoch": 0.9890909090909091, "percentage": 32.96, "elapsed_time": "3:31:06", "remaining_time": "7:09:23"} +{"current_steps": 1769, "total_steps": 5364, "loss": 0.09387499839067459, "lr": 1.695859045537123e-05, "epoch": 0.9896503496503497, "percentage": 32.98, "elapsed_time": "3:31:13", "remaining_time": "7:09:15"} +{"current_steps": 1770, "total_steps": 5364, "loss": 0.233891099691391, "lr": 1.6953914807618707e-05, "epoch": 0.9902097902097902, "percentage": 33.0, "elapsed_time": "3:31:20", "remaining_time": "7:09:07"} +{"current_steps": 1771, "total_steps": 5364, "loss": 0.20743700861930847, "lr": 1.6949236214261252e-05, "epoch": 0.9907692307692307, "percentage": 33.02, "elapsed_time": "3:31:27", "remaining_time": "7:09:00"} +{"current_steps": 1772, "total_steps": 5364, "loss": 0.20164258778095245, "lr": 1.6944554677280668e-05, "epoch": 0.9913286713286713, "percentage": 33.04, "elapsed_time": "3:31:34", "remaining_time": "7:08:52"} +{"current_steps": 1773, "total_steps": 5364, "loss": 0.15108591318130493, "lr": 1.6939870198660005e-05, "epoch": 0.9918881118881119, "percentage": 33.05, "elapsed_time": "3:31:41", "remaining_time": "7:08:44"} +{"current_steps": 1774, "total_steps": 5364, "loss": 0.21661114692687988, "lr": 1.6935182780383558e-05, "epoch": 0.9924475524475525, "percentage": 33.07, "elapsed_time": "3:31:48", "remaining_time": "7:08:38"} +{"current_steps": 1775, "total_steps": 5364, "loss": 0.48865288496017456, "lr": 1.6930492424436864e-05, "epoch": 0.993006993006993, "percentage": 33.09, "elapsed_time": "3:31:55", "remaining_time": "7:08:30"} +{"current_steps": 1776, "total_steps": 5364, "loss": 0.17697106301784515, "lr": 1.692579913280672e-05, "epoch": 0.9935664335664336, "percentage": 33.11, "elapsed_time": "3:32:03", "remaining_time": "7:08:24"} +{"current_steps": 1777, "total_steps": 5364, "loss": 0.2752719223499298, "lr": 1.692110290748114e-05, "epoch": 0.9941258741258742, "percentage": 33.13, "elapsed_time": "3:32:10", "remaining_time": "7:08:16"} +{"current_steps": 1778, "total_steps": 5364, "loss": 0.12802013754844666, "lr": 1.691640375044941e-05, "epoch": 0.9946853146853147, "percentage": 33.15, "elapsed_time": "3:32:17", "remaining_time": "7:08:09"} +{"current_steps": 1779, "total_steps": 5364, "loss": 0.25361230969429016, "lr": 1.6911701663702033e-05, "epoch": 0.9952447552447552, "percentage": 33.17, "elapsed_time": "3:32:24", "remaining_time": "7:08:01"} +{"current_steps": 1780, "total_steps": 5364, "loss": 0.2693482041358948, "lr": 1.6906996649230766e-05, "epoch": 0.9958041958041958, "percentage": 33.18, "elapsed_time": "3:32:31", "remaining_time": "7:07:53"} +{"current_steps": 1781, "total_steps": 5364, "loss": 0.34110140800476074, "lr": 1.690228870902861e-05, "epoch": 0.9963636363636363, "percentage": 33.2, "elapsed_time": "3:32:37", "remaining_time": "7:07:46"} +{"current_steps": 1782, "total_steps": 5364, "loss": 0.45697373151779175, "lr": 1.6897577845089785e-05, "epoch": 0.9969230769230769, "percentage": 33.22, "elapsed_time": "3:32:44", "remaining_time": "7:07:38"} +{"current_steps": 1783, "total_steps": 5364, "loss": 0.33918535709381104, "lr": 1.689286405940978e-05, "epoch": 0.9974825174825175, "percentage": 33.24, "elapsed_time": "3:32:51", "remaining_time": "7:07:31"} +{"current_steps": 1784, "total_steps": 5364, "loss": 0.11391283571720123, "lr": 1.6888147353985288e-05, "epoch": 0.998041958041958, "percentage": 33.26, "elapsed_time": "3:32:59", "remaining_time": "7:07:24"} +{"current_steps": 1785, "total_steps": 5364, "loss": 0.0903647318482399, "lr": 1.688342773081427e-05, "epoch": 0.9986013986013986, "percentage": 33.28, "elapsed_time": "3:33:06", "remaining_time": "7:07:17"} +{"current_steps": 1786, "total_steps": 5364, "loss": 0.122248575091362, "lr": 1.6878705191895895e-05, "epoch": 0.9991608391608392, "percentage": 33.3, "elapsed_time": "3:33:13", "remaining_time": "7:07:09"} +{"current_steps": 1787, "total_steps": 5364, "loss": 0.2232385128736496, "lr": 1.6873979739230594e-05, "epoch": 0.9997202797202798, "percentage": 33.31, "elapsed_time": "3:33:20", "remaining_time": "7:07:01"} +{"current_steps": 1788, "total_steps": 5364, "loss": 0.25163209438323975, "lr": 1.686925137482001e-05, "epoch": 1.0, "percentage": 33.33, "elapsed_time": "3:33:23", "remaining_time": "7:06:47"} +{"current_steps": 1789, "total_steps": 5364, "loss": 0.08295510709285736, "lr": 1.6864520100667036e-05, "epoch": 1.0005594405594405, "percentage": 33.35, "elapsed_time": "3:33:31", "remaining_time": "7:06:40"} +{"current_steps": 1790, "total_steps": 5364, "loss": 0.1294628232717514, "lr": 1.685978591877578e-05, "epoch": 1.0011188811188811, "percentage": 33.37, "elapsed_time": "3:33:38", "remaining_time": "7:06:33"} +{"current_steps": 1791, "total_steps": 5364, "loss": 0.11114399135112762, "lr": 1.6855048831151596e-05, "epoch": 1.0016783216783216, "percentage": 33.39, "elapsed_time": "3:33:45", "remaining_time": "7:06:26"} +{"current_steps": 1792, "total_steps": 5364, "loss": 0.12066492438316345, "lr": 1.6850308839801067e-05, "epoch": 1.0022377622377623, "percentage": 33.41, "elapsed_time": "3:33:52", "remaining_time": "7:06:19"} +{"current_steps": 1793, "total_steps": 5364, "loss": 0.20680563151836395, "lr": 1.6845565946732003e-05, "epoch": 1.0027972027972027, "percentage": 33.43, "elapsed_time": "3:33:59", "remaining_time": "7:06:11"} +{"current_steps": 1794, "total_steps": 5364, "loss": 0.20539617538452148, "lr": 1.684082015395344e-05, "epoch": 1.0033566433566434, "percentage": 33.45, "elapsed_time": "3:34:06", "remaining_time": "7:06:04"} +{"current_steps": 1795, "total_steps": 5364, "loss": 0.09164722263813019, "lr": 1.6836071463475646e-05, "epoch": 1.003916083916084, "percentage": 33.46, "elapsed_time": "3:34:13", "remaining_time": "7:05:57"} +{"current_steps": 1796, "total_steps": 5364, "loss": 0.13413432240486145, "lr": 1.6831319877310118e-05, "epoch": 1.0044755244755246, "percentage": 33.48, "elapsed_time": "3:34:21", "remaining_time": "7:05:50"} +{"current_steps": 1797, "total_steps": 5364, "loss": 0.21825465559959412, "lr": 1.6826565397469577e-05, "epoch": 1.005034965034965, "percentage": 33.5, "elapsed_time": "3:34:28", "remaining_time": "7:05:43"} +{"current_steps": 1798, "total_steps": 5364, "loss": 0.007302472833544016, "lr": 1.682180802596797e-05, "epoch": 1.0055944055944055, "percentage": 33.52, "elapsed_time": "3:34:35", "remaining_time": "7:05:36"} +{"current_steps": 1799, "total_steps": 5364, "loss": 0.18644356727600098, "lr": 1.6817047764820472e-05, "epoch": 1.0061538461538462, "percentage": 33.54, "elapsed_time": "3:34:43", "remaining_time": "7:05:29"} +{"current_steps": 1800, "total_steps": 5364, "loss": 0.1517438292503357, "lr": 1.6812284616043476e-05, "epoch": 1.0067132867132866, "percentage": 33.56, "elapsed_time": "3:34:50", "remaining_time": "7:05:22"} +{"current_steps": 1801, "total_steps": 5364, "loss": 0.10193042457103729, "lr": 1.6807518581654605e-05, "epoch": 1.0072727272727273, "percentage": 33.58, "elapsed_time": "3:34:57", "remaining_time": "7:05:15"} +{"current_steps": 1802, "total_steps": 5364, "loss": 0.19434235990047455, "lr": 1.6802749663672698e-05, "epoch": 1.0078321678321678, "percentage": 33.59, "elapsed_time": "3:35:04", "remaining_time": "7:05:08"} +{"current_steps": 1803, "total_steps": 5364, "loss": 0.06855184584856033, "lr": 1.679797786411782e-05, "epoch": 1.0083916083916085, "percentage": 33.61, "elapsed_time": "3:35:11", "remaining_time": "7:05:01"} +{"current_steps": 1804, "total_steps": 5364, "loss": 0.08057352155447006, "lr": 1.6793203185011252e-05, "epoch": 1.008951048951049, "percentage": 33.63, "elapsed_time": "3:35:19", "remaining_time": "7:04:54"} +{"current_steps": 1805, "total_steps": 5364, "loss": 0.06162555515766144, "lr": 1.6788425628375504e-05, "epoch": 1.0095104895104896, "percentage": 33.65, "elapsed_time": "3:35:27", "remaining_time": "7:04:49"} +{"current_steps": 1806, "total_steps": 5364, "loss": 0.4912581443786621, "lr": 1.678364519623429e-05, "epoch": 1.01006993006993, "percentage": 33.67, "elapsed_time": "3:35:34", "remaining_time": "7:04:41"} +{"current_steps": 1807, "total_steps": 5364, "loss": 0.1503697633743286, "lr": 1.6778861890612558e-05, "epoch": 1.0106293706293705, "percentage": 33.69, "elapsed_time": "3:35:41", "remaining_time": "7:04:33"} +{"current_steps": 1808, "total_steps": 5364, "loss": 0.20396408438682556, "lr": 1.677407571353646e-05, "epoch": 1.0111888111888112, "percentage": 33.71, "elapsed_time": "3:35:48", "remaining_time": "7:04:26"} +{"current_steps": 1809, "total_steps": 5364, "loss": 0.15781229734420776, "lr": 1.6769286667033374e-05, "epoch": 1.0117482517482517, "percentage": 33.72, "elapsed_time": "3:35:54", "remaining_time": "7:04:18"} +{"current_steps": 1810, "total_steps": 5364, "loss": 0.216548889875412, "lr": 1.6764494753131887e-05, "epoch": 1.0123076923076924, "percentage": 33.74, "elapsed_time": "3:36:02", "remaining_time": "7:04:12"} +{"current_steps": 1811, "total_steps": 5364, "loss": 0.05147993564605713, "lr": 1.6759699973861808e-05, "epoch": 1.0128671328671328, "percentage": 33.76, "elapsed_time": "3:36:09", "remaining_time": "7:04:04"} +{"current_steps": 1812, "total_steps": 5364, "loss": 0.14214424788951874, "lr": 1.6754902331254152e-05, "epoch": 1.0134265734265735, "percentage": 33.78, "elapsed_time": "3:36:16", "remaining_time": "7:03:57"} +{"current_steps": 1813, "total_steps": 5364, "loss": 0.17502613365650177, "lr": 1.6750101827341146e-05, "epoch": 1.013986013986014, "percentage": 33.8, "elapsed_time": "3:36:23", "remaining_time": "7:03:50"} +{"current_steps": 1814, "total_steps": 5364, "loss": 0.15899983048439026, "lr": 1.6745298464156235e-05, "epoch": 1.0145454545454546, "percentage": 33.82, "elapsed_time": "3:36:30", "remaining_time": "7:03:42"} +{"current_steps": 1815, "total_steps": 5364, "loss": 0.15133816003799438, "lr": 1.6740492243734078e-05, "epoch": 1.015104895104895, "percentage": 33.84, "elapsed_time": "3:36:37", "remaining_time": "7:03:35"} +{"current_steps": 1816, "total_steps": 5364, "loss": 0.08814603090286255, "lr": 1.6735683168110534e-05, "epoch": 1.0156643356643356, "percentage": 33.86, "elapsed_time": "3:36:44", "remaining_time": "7:03:28"} +{"current_steps": 1817, "total_steps": 5364, "loss": 0.25201737880706787, "lr": 1.6730871239322678e-05, "epoch": 1.0162237762237762, "percentage": 33.87, "elapsed_time": "3:36:51", "remaining_time": "7:03:20"} +{"current_steps": 1818, "total_steps": 5364, "loss": 0.18676358461380005, "lr": 1.6726056459408792e-05, "epoch": 1.0167832167832167, "percentage": 33.89, "elapsed_time": "3:36:58", "remaining_time": "7:03:13"} +{"current_steps": 1819, "total_steps": 5364, "loss": 0.12932777404785156, "lr": 1.6721238830408366e-05, "epoch": 1.0173426573426574, "percentage": 33.91, "elapsed_time": "3:37:06", "remaining_time": "7:03:06"} +{"current_steps": 1820, "total_steps": 5364, "loss": 0.17120352387428284, "lr": 1.67164183543621e-05, "epoch": 1.0179020979020978, "percentage": 33.93, "elapsed_time": "3:37:13", "remaining_time": "7:02:58"} +{"current_steps": 1821, "total_steps": 5364, "loss": 0.07034595310688019, "lr": 1.6711595033311892e-05, "epoch": 1.0184615384615385, "percentage": 33.95, "elapsed_time": "3:37:20", "remaining_time": "7:02:52"} +{"current_steps": 1822, "total_steps": 5364, "loss": 0.12487390637397766, "lr": 1.670676886930085e-05, "epoch": 1.019020979020979, "percentage": 33.97, "elapsed_time": "3:37:27", "remaining_time": "7:02:45"} +{"current_steps": 1823, "total_steps": 5364, "loss": 0.15174169838428497, "lr": 1.670193986437329e-05, "epoch": 1.0195804195804197, "percentage": 33.99, "elapsed_time": "3:37:34", "remaining_time": "7:02:37"} +{"current_steps": 1824, "total_steps": 5364, "loss": 0.08087201416492462, "lr": 1.6697108020574723e-05, "epoch": 1.0201398601398601, "percentage": 34.0, "elapsed_time": "3:37:43", "remaining_time": "7:02:32"} +{"current_steps": 1825, "total_steps": 5364, "loss": 0.10425123572349548, "lr": 1.6692273339951875e-05, "epoch": 1.0206993006993006, "percentage": 34.02, "elapsed_time": "3:37:50", "remaining_time": "7:02:25"} +{"current_steps": 1826, "total_steps": 5364, "loss": 0.20766519010066986, "lr": 1.6687435824552653e-05, "epoch": 1.0212587412587413, "percentage": 34.04, "elapsed_time": "3:37:57", "remaining_time": "7:02:18"} +{"current_steps": 1827, "total_steps": 5364, "loss": 0.09538062661886215, "lr": 1.6682595476426188e-05, "epoch": 1.0218181818181817, "percentage": 34.06, "elapsed_time": "3:38:04", "remaining_time": "7:02:11"} +{"current_steps": 1828, "total_steps": 5364, "loss": 0.1688915640115738, "lr": 1.6677752297622797e-05, "epoch": 1.0223776223776224, "percentage": 34.08, "elapsed_time": "3:38:11", "remaining_time": "7:02:04"} +{"current_steps": 1829, "total_steps": 5364, "loss": 0.254028856754303, "lr": 1.6672906290194003e-05, "epoch": 1.0229370629370629, "percentage": 34.1, "elapsed_time": "3:38:18", "remaining_time": "7:01:56"} +{"current_steps": 1830, "total_steps": 5364, "loss": 0.04055303707718849, "lr": 1.6668057456192514e-05, "epoch": 1.0234965034965036, "percentage": 34.12, "elapsed_time": "3:38:26", "remaining_time": "7:01:49"} +{"current_steps": 1831, "total_steps": 5364, "loss": 0.12503540515899658, "lr": 1.666320579767225e-05, "epoch": 1.024055944055944, "percentage": 34.13, "elapsed_time": "3:38:33", "remaining_time": "7:01:42"} +{"current_steps": 1832, "total_steps": 5364, "loss": 0.1253487914800644, "lr": 1.665835131668833e-05, "epoch": 1.0246153846153847, "percentage": 34.15, "elapsed_time": "3:38:41", "remaining_time": "7:01:36"} +{"current_steps": 1833, "total_steps": 5364, "loss": 0.168176531791687, "lr": 1.6653494015297055e-05, "epoch": 1.0251748251748252, "percentage": 34.17, "elapsed_time": "3:38:48", "remaining_time": "7:01:29"} +{"current_steps": 1834, "total_steps": 5364, "loss": 0.1692991852760315, "lr": 1.664863389555592e-05, "epoch": 1.0257342657342656, "percentage": 34.19, "elapsed_time": "3:38:55", "remaining_time": "7:01:22"} +{"current_steps": 1835, "total_steps": 5364, "loss": 0.09577256441116333, "lr": 1.6643770959523628e-05, "epoch": 1.0262937062937063, "percentage": 34.21, "elapsed_time": "3:39:03", "remaining_time": "7:01:16"} +{"current_steps": 1836, "total_steps": 5364, "loss": 0.04955904185771942, "lr": 1.6638905209260067e-05, "epoch": 1.0268531468531468, "percentage": 34.23, "elapsed_time": "3:39:09", "remaining_time": "7:01:08"} +{"current_steps": 1837, "total_steps": 5364, "loss": 0.13321664929389954, "lr": 1.6634036646826314e-05, "epoch": 1.0274125874125875, "percentage": 34.25, "elapsed_time": "3:39:17", "remaining_time": "7:01:02"} +{"current_steps": 1838, "total_steps": 5364, "loss": 0.14160437881946564, "lr": 1.6629165274284648e-05, "epoch": 1.027972027972028, "percentage": 34.27, "elapsed_time": "3:39:25", "remaining_time": "7:00:57"} +{"current_steps": 1839, "total_steps": 5364, "loss": 0.13516326248645782, "lr": 1.662429109369852e-05, "epoch": 1.0285314685314686, "percentage": 34.28, "elapsed_time": "3:39:33", "remaining_time": "7:00:51"} +{"current_steps": 1840, "total_steps": 5364, "loss": 0.014976169914007187, "lr": 1.661941410713259e-05, "epoch": 1.029090909090909, "percentage": 34.3, "elapsed_time": "3:39:40", "remaining_time": "7:00:43"} +{"current_steps": 1841, "total_steps": 5364, "loss": 0.10721161961555481, "lr": 1.6614534316652694e-05, "epoch": 1.0296503496503497, "percentage": 34.32, "elapsed_time": "3:39:47", "remaining_time": "7:00:36"} +{"current_steps": 1842, "total_steps": 5364, "loss": 0.05441214144229889, "lr": 1.6609651724325863e-05, "epoch": 1.0302097902097902, "percentage": 34.34, "elapsed_time": "3:39:54", "remaining_time": "7:00:28"} +{"current_steps": 1843, "total_steps": 5364, "loss": 0.08860521018505096, "lr": 1.6604766332220306e-05, "epoch": 1.0307692307692307, "percentage": 34.36, "elapsed_time": "3:40:01", "remaining_time": "7:00:20"} +{"current_steps": 1844, "total_steps": 5364, "loss": 0.22397495806217194, "lr": 1.659987814240543e-05, "epoch": 1.0313286713286713, "percentage": 34.38, "elapsed_time": "3:40:08", "remaining_time": "7:00:13"} +{"current_steps": 1845, "total_steps": 5364, "loss": 0.1872205138206482, "lr": 1.659498715695182e-05, "epoch": 1.0318881118881118, "percentage": 34.4, "elapsed_time": "3:40:15", "remaining_time": "7:00:05"} +{"current_steps": 1846, "total_steps": 5364, "loss": 0.0827154889702797, "lr": 1.659009337793124e-05, "epoch": 1.0324475524475525, "percentage": 34.41, "elapsed_time": "3:40:22", "remaining_time": "6:59:58"} +{"current_steps": 1847, "total_steps": 5364, "loss": 0.09823012351989746, "lr": 1.658519680741665e-05, "epoch": 1.033006993006993, "percentage": 34.43, "elapsed_time": "3:40:30", "remaining_time": "6:59:52"} +{"current_steps": 1848, "total_steps": 5364, "loss": 0.09591659903526306, "lr": 1.658029744748218e-05, "epoch": 1.0335664335664336, "percentage": 34.45, "elapsed_time": "3:40:36", "remaining_time": "6:59:44"} +{"current_steps": 1849, "total_steps": 5364, "loss": 0.04960941523313522, "lr": 1.657539530020315e-05, "epoch": 1.034125874125874, "percentage": 34.47, "elapsed_time": "3:40:44", "remaining_time": "6:59:37"} +{"current_steps": 1850, "total_steps": 5364, "loss": 0.10362362861633301, "lr": 1.657049036765606e-05, "epoch": 1.0346853146853148, "percentage": 34.49, "elapsed_time": "3:40:51", "remaining_time": "6:59:30"} +{"current_steps": 1851, "total_steps": 5364, "loss": 0.15144553780555725, "lr": 1.656558265191858e-05, "epoch": 1.0352447552447552, "percentage": 34.51, "elapsed_time": "3:40:58", "remaining_time": "6:59:22"} +{"current_steps": 1852, "total_steps": 5364, "loss": 0.07592298090457916, "lr": 1.6560672155069573e-05, "epoch": 1.0358041958041957, "percentage": 34.53, "elapsed_time": "3:41:05", "remaining_time": "6:59:15"} +{"current_steps": 1853, "total_steps": 5364, "loss": 0.1961602121591568, "lr": 1.6555758879189076e-05, "epoch": 1.0363636363636364, "percentage": 34.55, "elapsed_time": "3:41:12", "remaining_time": "6:59:07"} +{"current_steps": 1854, "total_steps": 5364, "loss": 0.21520818769931793, "lr": 1.6550842826358295e-05, "epoch": 1.0369230769230768, "percentage": 34.56, "elapsed_time": "3:41:19", "remaining_time": "6:59:00"} +{"current_steps": 1855, "total_steps": 5364, "loss": 0.1701516956090927, "lr": 1.654592399865962e-05, "epoch": 1.0374825174825175, "percentage": 34.58, "elapsed_time": "3:41:26", "remaining_time": "6:58:52"} +{"current_steps": 1856, "total_steps": 5364, "loss": 0.034923385828733444, "lr": 1.6541002398176614e-05, "epoch": 1.038041958041958, "percentage": 34.6, "elapsed_time": "3:41:33", "remaining_time": "6:58:45"} +{"current_steps": 1857, "total_steps": 5364, "loss": 0.1235581561923027, "lr": 1.6536078026994017e-05, "epoch": 1.0386013986013987, "percentage": 34.62, "elapsed_time": "3:41:40", "remaining_time": "6:58:38"} +{"current_steps": 1858, "total_steps": 5364, "loss": 0.25293564796447754, "lr": 1.6531150887197744e-05, "epoch": 1.0391608391608391, "percentage": 34.64, "elapsed_time": "3:41:47", "remaining_time": "6:58:30"} +{"current_steps": 1859, "total_steps": 5364, "loss": 0.10524363070726395, "lr": 1.652622098087487e-05, "epoch": 1.0397202797202798, "percentage": 34.66, "elapsed_time": "3:41:54", "remaining_time": "6:58:22"} +{"current_steps": 1860, "total_steps": 5364, "loss": 0.028555942699313164, "lr": 1.6521288310113654e-05, "epoch": 1.0402797202797203, "percentage": 34.68, "elapsed_time": "3:42:01", "remaining_time": "6:58:16"} +{"current_steps": 1861, "total_steps": 5364, "loss": 0.13310854136943817, "lr": 1.651635287700353e-05, "epoch": 1.0408391608391607, "percentage": 34.69, "elapsed_time": "3:42:08", "remaining_time": "6:58:09"} +{"current_steps": 1862, "total_steps": 5364, "loss": 0.2666672468185425, "lr": 1.651141468363509e-05, "epoch": 1.0413986013986014, "percentage": 34.71, "elapsed_time": "3:42:16", "remaining_time": "6:58:02"} +{"current_steps": 1863, "total_steps": 5364, "loss": 0.22971510887145996, "lr": 1.65064737321001e-05, "epoch": 1.0419580419580419, "percentage": 34.73, "elapsed_time": "3:42:23", "remaining_time": "6:57:55"} +{"current_steps": 1864, "total_steps": 5364, "loss": 0.19471578299999237, "lr": 1.6501530024491493e-05, "epoch": 1.0425174825174826, "percentage": 34.75, "elapsed_time": "3:42:30", "remaining_time": "6:57:47"} +{"current_steps": 1865, "total_steps": 5364, "loss": 0.07229534536600113, "lr": 1.6496583562903383e-05, "epoch": 1.043076923076923, "percentage": 34.77, "elapsed_time": "3:42:37", "remaining_time": "6:57:40"} +{"current_steps": 1866, "total_steps": 5364, "loss": 0.1333303153514862, "lr": 1.6491634349431024e-05, "epoch": 1.0436363636363637, "percentage": 34.79, "elapsed_time": "3:42:44", "remaining_time": "6:57:33"} +{"current_steps": 1867, "total_steps": 5364, "loss": 0.22127163410186768, "lr": 1.6486682386170862e-05, "epoch": 1.0441958041958042, "percentage": 34.81, "elapsed_time": "3:42:51", "remaining_time": "6:57:25"} +{"current_steps": 1868, "total_steps": 5364, "loss": 0.05909726768732071, "lr": 1.648172767522049e-05, "epoch": 1.0447552447552448, "percentage": 34.82, "elapsed_time": "3:42:58", "remaining_time": "6:57:18"} +{"current_steps": 1869, "total_steps": 5364, "loss": 0.08577749133110046, "lr": 1.647677021867867e-05, "epoch": 1.0453146853146853, "percentage": 34.84, "elapsed_time": "3:43:06", "remaining_time": "6:57:12"} +{"current_steps": 1870, "total_steps": 5364, "loss": 0.017373185604810715, "lr": 1.6471810018645334e-05, "epoch": 1.045874125874126, "percentage": 34.86, "elapsed_time": "3:43:13", "remaining_time": "6:57:04"} +{"current_steps": 1871, "total_steps": 5364, "loss": 0.22174407541751862, "lr": 1.646684707722157e-05, "epoch": 1.0464335664335664, "percentage": 34.88, "elapsed_time": "3:43:21", "remaining_time": "6:56:59"} +{"current_steps": 1872, "total_steps": 5364, "loss": 0.492086797952652, "lr": 1.6461881396509625e-05, "epoch": 1.046993006993007, "percentage": 34.9, "elapsed_time": "3:43:29", "remaining_time": "6:56:52"} +{"current_steps": 1873, "total_steps": 5364, "loss": 0.052501581609249115, "lr": 1.6456912978612903e-05, "epoch": 1.0475524475524476, "percentage": 34.92, "elapsed_time": "3:43:36", "remaining_time": "6:56:45"} +{"current_steps": 1874, "total_steps": 5364, "loss": 0.11107093095779419, "lr": 1.6451941825635987e-05, "epoch": 1.048111888111888, "percentage": 34.94, "elapsed_time": "3:43:43", "remaining_time": "6:56:38"} +{"current_steps": 1875, "total_steps": 5364, "loss": 0.04314631596207619, "lr": 1.6446967939684592e-05, "epoch": 1.0486713286713287, "percentage": 34.96, "elapsed_time": "3:43:50", "remaining_time": "6:56:31"} +{"current_steps": 1876, "total_steps": 5364, "loss": 0.06144296005368233, "lr": 1.644199132286561e-05, "epoch": 1.0492307692307692, "percentage": 34.97, "elapsed_time": "3:43:57", "remaining_time": "6:56:24"} +{"current_steps": 1877, "total_steps": 5364, "loss": 0.06786086410284042, "lr": 1.6437011977287084e-05, "epoch": 1.0497902097902099, "percentage": 34.99, "elapsed_time": "3:44:04", "remaining_time": "6:56:16"} +{"current_steps": 1878, "total_steps": 5364, "loss": 0.14964109659194946, "lr": 1.643202990505821e-05, "epoch": 1.0503496503496503, "percentage": 35.01, "elapsed_time": "3:44:11", "remaining_time": "6:56:08"} +{"current_steps": 1879, "total_steps": 5364, "loss": 0.33606380224227905, "lr": 1.6427045108289344e-05, "epoch": 1.050909090909091, "percentage": 35.03, "elapsed_time": "3:44:17", "remaining_time": "6:56:00"} +{"current_steps": 1880, "total_steps": 5364, "loss": 0.157263845205307, "lr": 1.642205758909199e-05, "epoch": 1.0514685314685315, "percentage": 35.05, "elapsed_time": "3:44:25", "remaining_time": "6:55:53"} +{"current_steps": 1881, "total_steps": 5364, "loss": 0.0952346995472908, "lr": 1.641706734957881e-05, "epoch": 1.052027972027972, "percentage": 35.07, "elapsed_time": "3:44:31", "remaining_time": "6:55:45"} +{"current_steps": 1882, "total_steps": 5364, "loss": 0.08294466882944107, "lr": 1.641207439186362e-05, "epoch": 1.0525874125874126, "percentage": 35.09, "elapsed_time": "3:44:38", "remaining_time": "6:55:38"} +{"current_steps": 1883, "total_steps": 5364, "loss": 0.26249027252197266, "lr": 1.6407078718061374e-05, "epoch": 1.053146853146853, "percentage": 35.1, "elapsed_time": "3:44:46", "remaining_time": "6:55:30"} +{"current_steps": 1884, "total_steps": 5364, "loss": 0.21906225383281708, "lr": 1.6402080330288198e-05, "epoch": 1.0537062937062938, "percentage": 35.12, "elapsed_time": "3:44:52", "remaining_time": "6:55:23"} +{"current_steps": 1885, "total_steps": 5364, "loss": 0.23156702518463135, "lr": 1.639707923066135e-05, "epoch": 1.0542657342657342, "percentage": 35.14, "elapsed_time": "3:45:00", "remaining_time": "6:55:16"} +{"current_steps": 1886, "total_steps": 5364, "loss": 0.14702865481376648, "lr": 1.6392075421299246e-05, "epoch": 1.054825174825175, "percentage": 35.16, "elapsed_time": "3:45:07", "remaining_time": "6:55:09"} +{"current_steps": 1887, "total_steps": 5364, "loss": 0.058649949729442596, "lr": 1.6387068904321445e-05, "epoch": 1.0553846153846154, "percentage": 35.18, "elapsed_time": "3:45:15", "remaining_time": "6:55:03"} +{"current_steps": 1888, "total_steps": 5364, "loss": 0.2501545548439026, "lr": 1.6382059681848656e-05, "epoch": 1.055944055944056, "percentage": 35.2, "elapsed_time": "3:45:22", "remaining_time": "6:54:56"} +{"current_steps": 1889, "total_steps": 5364, "loss": 0.17257605493068695, "lr": 1.637704775600273e-05, "epoch": 1.0565034965034965, "percentage": 35.22, "elapsed_time": "3:45:29", "remaining_time": "6:54:49"} +{"current_steps": 1890, "total_steps": 5364, "loss": 0.1971706748008728, "lr": 1.637203312890667e-05, "epoch": 1.057062937062937, "percentage": 35.23, "elapsed_time": "3:45:36", "remaining_time": "6:54:41"} +{"current_steps": 1891, "total_steps": 5364, "loss": 0.321458637714386, "lr": 1.6367015802684618e-05, "epoch": 1.0576223776223777, "percentage": 35.25, "elapsed_time": "3:45:44", "remaining_time": "6:54:34"} +{"current_steps": 1892, "total_steps": 5364, "loss": 0.3172048032283783, "lr": 1.6361995779461856e-05, "epoch": 1.0581818181818181, "percentage": 35.27, "elapsed_time": "3:45:51", "remaining_time": "6:54:27"} +{"current_steps": 1893, "total_steps": 5364, "loss": 0.32852277159690857, "lr": 1.6356973061364822e-05, "epoch": 1.0587412587412588, "percentage": 35.29, "elapsed_time": "3:45:58", "remaining_time": "6:54:20"} +{"current_steps": 1894, "total_steps": 5364, "loss": 0.19330886006355286, "lr": 1.6351947650521077e-05, "epoch": 1.0593006993006993, "percentage": 35.31, "elapsed_time": "3:46:04", "remaining_time": "6:54:12"} +{"current_steps": 1895, "total_steps": 5364, "loss": 0.1689222753047943, "lr": 1.634691954905934e-05, "epoch": 1.05986013986014, "percentage": 35.33, "elapsed_time": "3:46:12", "remaining_time": "6:54:05"} +{"current_steps": 1896, "total_steps": 5364, "loss": 0.10688693821430206, "lr": 1.634188875910945e-05, "epoch": 1.0604195804195804, "percentage": 35.35, "elapsed_time": "3:46:19", "remaining_time": "6:53:57"} +{"current_steps": 1897, "total_steps": 5364, "loss": 0.279634028673172, "lr": 1.6336855282802413e-05, "epoch": 1.060979020979021, "percentage": 35.37, "elapsed_time": "3:46:26", "remaining_time": "6:53:51"} +{"current_steps": 1898, "total_steps": 5364, "loss": 0.1641359180212021, "lr": 1.6331819122270345e-05, "epoch": 1.0615384615384615, "percentage": 35.38, "elapsed_time": "3:46:33", "remaining_time": "6:53:43"} +{"current_steps": 1899, "total_steps": 5364, "loss": 0.13418062031269073, "lr": 1.632678027964651e-05, "epoch": 1.062097902097902, "percentage": 35.4, "elapsed_time": "3:46:40", "remaining_time": "6:53:35"} +{"current_steps": 1900, "total_steps": 5364, "loss": 0.031035996973514557, "lr": 1.6321738757065313e-05, "epoch": 1.0626573426573427, "percentage": 35.42, "elapsed_time": "3:46:47", "remaining_time": "6:53:29"} +{"current_steps": 1901, "total_steps": 5364, "loss": 0.1120566800236702, "lr": 1.631669455666229e-05, "epoch": 1.0632167832167831, "percentage": 35.44, "elapsed_time": "3:46:54", "remaining_time": "6:53:21"} +{"current_steps": 1902, "total_steps": 5364, "loss": 0.2700100243091583, "lr": 1.6311647680574107e-05, "epoch": 1.0637762237762238, "percentage": 35.46, "elapsed_time": "3:47:02", "remaining_time": "6:53:14"} +{"current_steps": 1903, "total_steps": 5364, "loss": 0.039188630878925323, "lr": 1.6306598130938573e-05, "epoch": 1.0643356643356643, "percentage": 35.48, "elapsed_time": "3:47:09", "remaining_time": "6:53:07"} +{"current_steps": 1904, "total_steps": 5364, "loss": 0.07795591652393341, "lr": 1.630154590989462e-05, "epoch": 1.064895104895105, "percentage": 35.5, "elapsed_time": "3:47:16", "remaining_time": "6:52:59"} +{"current_steps": 1905, "total_steps": 5364, "loss": 0.1720692217350006, "lr": 1.6296491019582315e-05, "epoch": 1.0654545454545454, "percentage": 35.51, "elapsed_time": "3:47:23", "remaining_time": "6:52:52"} +{"current_steps": 1906, "total_steps": 5364, "loss": 0.10365504026412964, "lr": 1.6291433462142858e-05, "epoch": 1.0660139860139861, "percentage": 35.53, "elapsed_time": "3:47:30", "remaining_time": "6:52:44"} +{"current_steps": 1907, "total_steps": 5364, "loss": 0.19018085300922394, "lr": 1.6286373239718577e-05, "epoch": 1.0665734265734266, "percentage": 35.55, "elapsed_time": "3:47:36", "remaining_time": "6:52:37"} +{"current_steps": 1908, "total_steps": 5364, "loss": 0.3060765564441681, "lr": 1.6281310354452928e-05, "epoch": 1.067132867132867, "percentage": 35.57, "elapsed_time": "3:47:43", "remaining_time": "6:52:29"} +{"current_steps": 1909, "total_steps": 5364, "loss": 0.08396832644939423, "lr": 1.6276244808490497e-05, "epoch": 1.0676923076923077, "percentage": 35.59, "elapsed_time": "3:47:51", "remaining_time": "6:52:24"} +{"current_steps": 1910, "total_steps": 5364, "loss": 0.11355772614479065, "lr": 1.6271176603976994e-05, "epoch": 1.0682517482517482, "percentage": 35.61, "elapsed_time": "3:47:58", "remaining_time": "6:52:16"} +{"current_steps": 1911, "total_steps": 5364, "loss": 0.16688455641269684, "lr": 1.626610574305926e-05, "epoch": 1.0688111888111889, "percentage": 35.63, "elapsed_time": "3:48:05", "remaining_time": "6:52:08"} +{"current_steps": 1912, "total_steps": 5364, "loss": 0.21331313252449036, "lr": 1.6261032227885253e-05, "epoch": 1.0693706293706293, "percentage": 35.65, "elapsed_time": "3:48:13", "remaining_time": "6:52:02"} +{"current_steps": 1913, "total_steps": 5364, "loss": 0.2213878035545349, "lr": 1.6255956060604067e-05, "epoch": 1.06993006993007, "percentage": 35.66, "elapsed_time": "3:48:20", "remaining_time": "6:51:54"} +{"current_steps": 1914, "total_steps": 5364, "loss": 0.07644247263669968, "lr": 1.6250877243365906e-05, "epoch": 1.0704895104895105, "percentage": 35.68, "elapsed_time": "3:48:28", "remaining_time": "6:51:49"} +{"current_steps": 1915, "total_steps": 5364, "loss": 0.10563966631889343, "lr": 1.6245795778322108e-05, "epoch": 1.0710489510489511, "percentage": 35.7, "elapsed_time": "3:48:34", "remaining_time": "6:51:41"} +{"current_steps": 1916, "total_steps": 5364, "loss": 0.08450563997030258, "lr": 1.6240711667625127e-05, "epoch": 1.0716083916083916, "percentage": 35.72, "elapsed_time": "3:48:41", "remaining_time": "6:51:33"} +{"current_steps": 1917, "total_steps": 5364, "loss": 0.1038089245557785, "lr": 1.623562491342854e-05, "epoch": 1.072167832167832, "percentage": 35.74, "elapsed_time": "3:48:48", "remaining_time": "6:51:25"} +{"current_steps": 1918, "total_steps": 5364, "loss": 0.059487588703632355, "lr": 1.623053551788704e-05, "epoch": 1.0727272727272728, "percentage": 35.76, "elapsed_time": "3:48:55", "remaining_time": "6:51:18"} +{"current_steps": 1919, "total_steps": 5364, "loss": 0.10687503963708878, "lr": 1.6225443483156443e-05, "epoch": 1.0732867132867132, "percentage": 35.78, "elapsed_time": "3:49:02", "remaining_time": "6:51:10"} +{"current_steps": 1920, "total_steps": 5364, "loss": 0.17818018794059753, "lr": 1.622034881139368e-05, "epoch": 1.073846153846154, "percentage": 35.79, "elapsed_time": "3:49:09", "remaining_time": "6:51:02"} +{"current_steps": 1921, "total_steps": 5364, "loss": 0.114201121032238, "lr": 1.6215251504756796e-05, "epoch": 1.0744055944055944, "percentage": 35.81, "elapsed_time": "3:49:16", "remaining_time": "6:50:56"} +{"current_steps": 1922, "total_steps": 5364, "loss": 0.176470085978508, "lr": 1.6210151565404967e-05, "epoch": 1.074965034965035, "percentage": 35.83, "elapsed_time": "3:49:23", "remaining_time": "6:50:49"} +{"current_steps": 1923, "total_steps": 5364, "loss": 0.07264681160449982, "lr": 1.6205048995498466e-05, "epoch": 1.0755244755244755, "percentage": 35.85, "elapsed_time": "3:49:30", "remaining_time": "6:50:41"} +{"current_steps": 1924, "total_steps": 5364, "loss": 0.17455509305000305, "lr": 1.6199943797198687e-05, "epoch": 1.0760839160839162, "percentage": 35.87, "elapsed_time": "3:49:37", "remaining_time": "6:50:33"} +{"current_steps": 1925, "total_steps": 5364, "loss": 0.3306233882904053, "lr": 1.6194835972668142e-05, "epoch": 1.0766433566433566, "percentage": 35.89, "elapsed_time": "3:49:44", "remaining_time": "6:50:26"} +{"current_steps": 1926, "total_steps": 5364, "loss": 0.06638755649328232, "lr": 1.6189725524070445e-05, "epoch": 1.077202797202797, "percentage": 35.91, "elapsed_time": "3:49:51", "remaining_time": "6:50:19"} +{"current_steps": 1927, "total_steps": 5364, "loss": 0.11081357300281525, "lr": 1.6184612453570338e-05, "epoch": 1.0777622377622378, "percentage": 35.92, "elapsed_time": "3:49:59", "remaining_time": "6:50:11"} +{"current_steps": 1928, "total_steps": 5364, "loss": 0.10380128026008606, "lr": 1.6179496763333657e-05, "epoch": 1.0783216783216782, "percentage": 35.94, "elapsed_time": "3:50:06", "remaining_time": "6:50:05"} +{"current_steps": 1929, "total_steps": 5364, "loss": 0.2093890905380249, "lr": 1.617437845552736e-05, "epoch": 1.078881118881119, "percentage": 35.96, "elapsed_time": "3:50:14", "remaining_time": "6:49:58"} +{"current_steps": 1930, "total_steps": 5364, "loss": 0.07261718064546585, "lr": 1.61692575323195e-05, "epoch": 1.0794405594405594, "percentage": 35.98, "elapsed_time": "3:50:21", "remaining_time": "6:49:52"} +{"current_steps": 1931, "total_steps": 5364, "loss": 0.1493971049785614, "lr": 1.616413399587925e-05, "epoch": 1.08, "percentage": 36.0, "elapsed_time": "3:50:28", "remaining_time": "6:49:44"} +{"current_steps": 1932, "total_steps": 5364, "loss": 0.21584853529930115, "lr": 1.6159007848376885e-05, "epoch": 1.0805594405594405, "percentage": 36.02, "elapsed_time": "3:50:35", "remaining_time": "6:49:36"} +{"current_steps": 1933, "total_steps": 5364, "loss": 0.14152511954307556, "lr": 1.615387909198379e-05, "epoch": 1.0811188811188812, "percentage": 36.04, "elapsed_time": "3:50:42", "remaining_time": "6:49:29"} +{"current_steps": 1934, "total_steps": 5364, "loss": 0.08901500701904297, "lr": 1.614874772887245e-05, "epoch": 1.0816783216783217, "percentage": 36.06, "elapsed_time": "3:50:49", "remaining_time": "6:49:22"} +{"current_steps": 1935, "total_steps": 5364, "loss": 0.1626087874174118, "lr": 1.6143613761216456e-05, "epoch": 1.0822377622377621, "percentage": 36.07, "elapsed_time": "3:50:56", "remaining_time": "6:49:14"} +{"current_steps": 1936, "total_steps": 5364, "loss": 0.10775715112686157, "lr": 1.6138477191190503e-05, "epoch": 1.0827972027972028, "percentage": 36.09, "elapsed_time": "3:51:03", "remaining_time": "6:49:07"} +{"current_steps": 1937, "total_steps": 5364, "loss": 0.0633353665471077, "lr": 1.6133338020970392e-05, "epoch": 1.0833566433566433, "percentage": 36.11, "elapsed_time": "3:51:10", "remaining_time": "6:48:59"} +{"current_steps": 1938, "total_steps": 5364, "loss": 0.12046018242835999, "lr": 1.6128196252733013e-05, "epoch": 1.083916083916084, "percentage": 36.13, "elapsed_time": "3:51:17", "remaining_time": "6:48:52"} +{"current_steps": 1939, "total_steps": 5364, "loss": 0.16687169671058655, "lr": 1.6123051888656367e-05, "epoch": 1.0844755244755244, "percentage": 36.15, "elapsed_time": "3:51:24", "remaining_time": "6:48:44"} +{"current_steps": 1940, "total_steps": 5364, "loss": 0.1763058304786682, "lr": 1.6117904930919558e-05, "epoch": 1.085034965034965, "percentage": 36.17, "elapsed_time": "3:51:31", "remaining_time": "6:48:36"} +{"current_steps": 1941, "total_steps": 5364, "loss": 0.14062942564487457, "lr": 1.611275538170278e-05, "epoch": 1.0855944055944056, "percentage": 36.19, "elapsed_time": "3:51:37", "remaining_time": "6:48:29"} +{"current_steps": 1942, "total_steps": 5364, "loss": 0.15122312307357788, "lr": 1.6107603243187323e-05, "epoch": 1.0861538461538462, "percentage": 36.2, "elapsed_time": "3:51:44", "remaining_time": "6:48:21"} +{"current_steps": 1943, "total_steps": 5364, "loss": 0.12458480149507523, "lr": 1.6102448517555584e-05, "epoch": 1.0867132867132867, "percentage": 36.22, "elapsed_time": "3:51:51", "remaining_time": "6:48:14"} +{"current_steps": 1944, "total_steps": 5364, "loss": 0.26847195625305176, "lr": 1.6097291206991052e-05, "epoch": 1.0872727272727274, "percentage": 36.24, "elapsed_time": "3:51:58", "remaining_time": "6:48:06"} +{"current_steps": 1945, "total_steps": 5364, "loss": 0.08677637577056885, "lr": 1.6092131313678302e-05, "epoch": 1.0878321678321679, "percentage": 36.26, "elapsed_time": "3:52:05", "remaining_time": "6:47:58"} +{"current_steps": 1946, "total_steps": 5364, "loss": 0.29111695289611816, "lr": 1.6086968839803016e-05, "epoch": 1.0883916083916083, "percentage": 36.28, "elapsed_time": "3:52:12", "remaining_time": "6:47:51"} +{"current_steps": 1947, "total_steps": 5364, "loss": 0.10848572850227356, "lr": 1.6081803787551963e-05, "epoch": 1.088951048951049, "percentage": 36.3, "elapsed_time": "3:52:19", "remaining_time": "6:47:43"} +{"current_steps": 1948, "total_steps": 5364, "loss": 0.14661894738674164, "lr": 1.6076636159113e-05, "epoch": 1.0895104895104895, "percentage": 36.32, "elapsed_time": "3:52:26", "remaining_time": "6:47:36"} +{"current_steps": 1949, "total_steps": 5364, "loss": 0.2794772684574127, "lr": 1.607146595667509e-05, "epoch": 1.0900699300699301, "percentage": 36.33, "elapsed_time": "3:52:33", "remaining_time": "6:47:28"} +{"current_steps": 1950, "total_steps": 5364, "loss": 0.19241252541542053, "lr": 1.6066293182428264e-05, "epoch": 1.0906293706293706, "percentage": 36.35, "elapsed_time": "3:52:40", "remaining_time": "6:47:21"} +{"current_steps": 1951, "total_steps": 5364, "loss": 0.10696657001972198, "lr": 1.6061117838563662e-05, "epoch": 1.0911888111888113, "percentage": 36.37, "elapsed_time": "3:52:47", "remaining_time": "6:47:14"} +{"current_steps": 1952, "total_steps": 5364, "loss": 0.15843608975410461, "lr": 1.6055939927273506e-05, "epoch": 1.0917482517482517, "percentage": 36.39, "elapsed_time": "3:52:54", "remaining_time": "6:47:07"} +{"current_steps": 1953, "total_steps": 5364, "loss": 0.0833943709731102, "lr": 1.6050759450751097e-05, "epoch": 1.0923076923076924, "percentage": 36.41, "elapsed_time": "3:53:02", "remaining_time": "6:47:00"} +{"current_steps": 1954, "total_steps": 5364, "loss": 0.17700690031051636, "lr": 1.604557641119084e-05, "epoch": 1.0928671328671329, "percentage": 36.43, "elapsed_time": "3:53:08", "remaining_time": "6:46:52"} +{"current_steps": 1955, "total_steps": 5364, "loss": 0.18030565977096558, "lr": 1.6040390810788203e-05, "epoch": 1.0934265734265733, "percentage": 36.45, "elapsed_time": "3:53:16", "remaining_time": "6:46:45"} +{"current_steps": 1956, "total_steps": 5364, "loss": 0.048740699887275696, "lr": 1.6035202651739768e-05, "epoch": 1.093986013986014, "percentage": 36.47, "elapsed_time": "3:53:23", "remaining_time": "6:46:37"} +{"current_steps": 1957, "total_steps": 5364, "loss": 0.2627163529396057, "lr": 1.6030011936243173e-05, "epoch": 1.0945454545454545, "percentage": 36.48, "elapsed_time": "3:53:30", "remaining_time": "6:46:30"} +{"current_steps": 1958, "total_steps": 5364, "loss": 0.24802029132843018, "lr": 1.6024818666497153e-05, "epoch": 1.0951048951048952, "percentage": 36.5, "elapsed_time": "3:53:36", "remaining_time": "6:46:22"} +{"current_steps": 1959, "total_steps": 5364, "loss": 0.04002990573644638, "lr": 1.6019622844701525e-05, "epoch": 1.0956643356643356, "percentage": 36.52, "elapsed_time": "3:53:43", "remaining_time": "6:46:15"} +{"current_steps": 1960, "total_steps": 5364, "loss": 0.19142946600914001, "lr": 1.6014424473057177e-05, "epoch": 1.0962237762237763, "percentage": 36.54, "elapsed_time": "3:53:51", "remaining_time": "6:46:08"} +{"current_steps": 1961, "total_steps": 5364, "loss": 0.21755559742450714, "lr": 1.6009223553766094e-05, "epoch": 1.0967832167832168, "percentage": 36.56, "elapsed_time": "3:53:58", "remaining_time": "6:46:01"} +{"current_steps": 1962, "total_steps": 5364, "loss": 0.14702238142490387, "lr": 1.6004020089031324e-05, "epoch": 1.0973426573426575, "percentage": 36.58, "elapsed_time": "3:54:05", "remaining_time": "6:45:53"} +{"current_steps": 1963, "total_steps": 5364, "loss": 0.16499340534210205, "lr": 1.5998814081057e-05, "epoch": 1.097902097902098, "percentage": 36.6, "elapsed_time": "3:54:12", "remaining_time": "6:45:46"} +{"current_steps": 1964, "total_steps": 5364, "loss": 0.23408597707748413, "lr": 1.5993605532048336e-05, "epoch": 1.0984615384615384, "percentage": 36.61, "elapsed_time": "3:54:19", "remaining_time": "6:45:39"} +{"current_steps": 1965, "total_steps": 5364, "loss": 0.257779985666275, "lr": 1.598839444421161e-05, "epoch": 1.099020979020979, "percentage": 36.63, "elapsed_time": "3:54:26", "remaining_time": "6:45:32"} +{"current_steps": 1966, "total_steps": 5364, "loss": 0.09074355661869049, "lr": 1.59831808197542e-05, "epoch": 1.0995804195804195, "percentage": 36.65, "elapsed_time": "3:54:34", "remaining_time": "6:45:25"} +{"current_steps": 1967, "total_steps": 5364, "loss": 0.1712971329689026, "lr": 1.597796466088452e-05, "epoch": 1.1001398601398602, "percentage": 36.67, "elapsed_time": "3:54:41", "remaining_time": "6:45:18"} +{"current_steps": 1968, "total_steps": 5364, "loss": 0.17786143720149994, "lr": 1.59727459698121e-05, "epoch": 1.1006993006993007, "percentage": 36.69, "elapsed_time": "3:54:48", "remaining_time": "6:45:10"} +{"current_steps": 1969, "total_steps": 5364, "loss": 0.11310650408267975, "lr": 1.5967524748747515e-05, "epoch": 1.1012587412587413, "percentage": 36.71, "elapsed_time": "3:54:54", "remaining_time": "6:45:02"} +{"current_steps": 1970, "total_steps": 5364, "loss": 0.26985836029052734, "lr": 1.5962300999902416e-05, "epoch": 1.1018181818181818, "percentage": 36.73, "elapsed_time": "3:55:01", "remaining_time": "6:44:55"} +{"current_steps": 1971, "total_steps": 5364, "loss": 0.10740834474563599, "lr": 1.595707472548953e-05, "epoch": 1.1023776223776225, "percentage": 36.74, "elapsed_time": "3:55:08", "remaining_time": "6:44:47"} +{"current_steps": 1972, "total_steps": 5364, "loss": 0.05938706174492836, "lr": 1.5951845927722653e-05, "epoch": 1.102937062937063, "percentage": 36.76, "elapsed_time": "3:55:16", "remaining_time": "6:44:40"} +{"current_steps": 1973, "total_steps": 5364, "loss": 0.07059267163276672, "lr": 1.594661460881665e-05, "epoch": 1.1034965034965034, "percentage": 36.78, "elapsed_time": "3:55:22", "remaining_time": "6:44:33"} +{"current_steps": 1974, "total_steps": 5364, "loss": 0.11098214983940125, "lr": 1.5941380770987445e-05, "epoch": 1.104055944055944, "percentage": 36.8, "elapsed_time": "3:55:29", "remaining_time": "6:44:25"} +{"current_steps": 1975, "total_steps": 5364, "loss": 0.21214474737644196, "lr": 1.5936144416452048e-05, "epoch": 1.1046153846153846, "percentage": 36.82, "elapsed_time": "3:55:36", "remaining_time": "6:44:17"} +{"current_steps": 1976, "total_steps": 5364, "loss": 0.13516046106815338, "lr": 1.593090554742851e-05, "epoch": 1.1051748251748252, "percentage": 36.84, "elapsed_time": "3:55:43", "remaining_time": "6:44:10"} +{"current_steps": 1977, "total_steps": 5364, "loss": 0.07961449772119522, "lr": 1.5925664166135972e-05, "epoch": 1.1057342657342657, "percentage": 36.86, "elapsed_time": "3:55:50", "remaining_time": "6:44:02"} +{"current_steps": 1978, "total_steps": 5364, "loss": 0.1356368064880371, "lr": 1.592042027479462e-05, "epoch": 1.1062937062937064, "percentage": 36.88, "elapsed_time": "3:55:57", "remaining_time": "6:43:54"} +{"current_steps": 1979, "total_steps": 5364, "loss": 0.18539609014987946, "lr": 1.591517387562572e-05, "epoch": 1.1068531468531468, "percentage": 36.89, "elapsed_time": "3:56:04", "remaining_time": "6:43:47"} +{"current_steps": 1980, "total_steps": 5364, "loss": 0.1736263632774353, "lr": 1.5909924970851584e-05, "epoch": 1.1074125874125875, "percentage": 36.91, "elapsed_time": "3:56:11", "remaining_time": "6:43:40"} +{"current_steps": 1981, "total_steps": 5364, "loss": 0.1802954524755478, "lr": 1.5904673562695594e-05, "epoch": 1.107972027972028, "percentage": 36.93, "elapsed_time": "3:56:18", "remaining_time": "6:43:33"} +{"current_steps": 1982, "total_steps": 5364, "loss": 0.11812008917331696, "lr": 1.5899419653382192e-05, "epoch": 1.1085314685314684, "percentage": 36.95, "elapsed_time": "3:56:25", "remaining_time": "6:43:25"} +{"current_steps": 1983, "total_steps": 5364, "loss": 0.144090473651886, "lr": 1.5894163245136874e-05, "epoch": 1.1090909090909091, "percentage": 36.97, "elapsed_time": "3:56:32", "remaining_time": "6:43:18"} +{"current_steps": 1984, "total_steps": 5364, "loss": 0.07787076383829117, "lr": 1.588890434018621e-05, "epoch": 1.1096503496503496, "percentage": 36.99, "elapsed_time": "3:56:39", "remaining_time": "6:43:10"} +{"current_steps": 1985, "total_steps": 5364, "loss": 0.14424125850200653, "lr": 1.588364294075781e-05, "epoch": 1.1102097902097903, "percentage": 37.01, "elapsed_time": "3:56:46", "remaining_time": "6:43:03"} +{"current_steps": 1986, "total_steps": 5364, "loss": 0.170151025056839, "lr": 1.5878379049080344e-05, "epoch": 1.1107692307692307, "percentage": 37.02, "elapsed_time": "3:56:53", "remaining_time": "6:42:56"} +{"current_steps": 1987, "total_steps": 5364, "loss": 0.19215074181556702, "lr": 1.5873112667383542e-05, "epoch": 1.1113286713286714, "percentage": 37.04, "elapsed_time": "3:57:00", "remaining_time": "6:42:48"} +{"current_steps": 1988, "total_steps": 5364, "loss": 0.08479930460453033, "lr": 1.5867843797898194e-05, "epoch": 1.1118881118881119, "percentage": 37.06, "elapsed_time": "3:57:07", "remaining_time": "6:42:41"} +{"current_steps": 1989, "total_steps": 5364, "loss": 0.1669064462184906, "lr": 1.586257244285613e-05, "epoch": 1.1124475524475526, "percentage": 37.08, "elapsed_time": "3:57:14", "remaining_time": "6:42:34"} +{"current_steps": 1990, "total_steps": 5364, "loss": 0.16119353473186493, "lr": 1.585729860449025e-05, "epoch": 1.113006993006993, "percentage": 37.1, "elapsed_time": "3:57:22", "remaining_time": "6:42:27"} +{"current_steps": 1991, "total_steps": 5364, "loss": 0.11893858015537262, "lr": 1.5852022285034485e-05, "epoch": 1.1135664335664335, "percentage": 37.12, "elapsed_time": "3:57:29", "remaining_time": "6:42:20"} +{"current_steps": 1992, "total_steps": 5364, "loss": 0.22438114881515503, "lr": 1.5846743486723834e-05, "epoch": 1.1141258741258742, "percentage": 37.14, "elapsed_time": "3:57:36", "remaining_time": "6:42:13"} +{"current_steps": 1993, "total_steps": 5364, "loss": 0.21321550011634827, "lr": 1.5841462211794337e-05, "epoch": 1.1146853146853146, "percentage": 37.16, "elapsed_time": "3:57:44", "remaining_time": "6:42:06"} +{"current_steps": 1994, "total_steps": 5364, "loss": 0.11793461441993713, "lr": 1.5836178462483092e-05, "epoch": 1.1152447552447553, "percentage": 37.17, "elapsed_time": "3:57:51", "remaining_time": "6:41:59"} +{"current_steps": 1995, "total_steps": 5364, "loss": 0.2472701221704483, "lr": 1.583089224102824e-05, "epoch": 1.1158041958041958, "percentage": 37.19, "elapsed_time": "3:57:58", "remaining_time": "6:41:53"} +{"current_steps": 1996, "total_steps": 5364, "loss": 0.23547127842903137, "lr": 1.582560354966896e-05, "epoch": 1.1163636363636364, "percentage": 37.21, "elapsed_time": "3:58:05", "remaining_time": "6:41:45"} +{"current_steps": 1997, "total_steps": 5364, "loss": 0.21054032444953918, "lr": 1.5820312390645488e-05, "epoch": 1.116923076923077, "percentage": 37.23, "elapsed_time": "3:58:12", "remaining_time": "6:41:37"} +{"current_steps": 1998, "total_steps": 5364, "loss": 0.04946086183190346, "lr": 1.5815018766199112e-05, "epoch": 1.1174825174825176, "percentage": 37.25, "elapsed_time": "3:58:19", "remaining_time": "6:41:30"} +{"current_steps": 1999, "total_steps": 5364, "loss": 0.11270518600940704, "lr": 1.5809722678572147e-05, "epoch": 1.118041958041958, "percentage": 37.27, "elapsed_time": "3:58:26", "remaining_time": "6:41:22"} +{"current_steps": 2000, "total_steps": 5364, "loss": 0.1681709885597229, "lr": 1.5804424130007963e-05, "epoch": 1.1186013986013985, "percentage": 37.29, "elapsed_time": "3:58:33", "remaining_time": "6:41:16"} +{"current_steps": 2001, "total_steps": 5364, "loss": 0.35935884714126587, "lr": 1.5799123122750967e-05, "epoch": 1.1191608391608392, "percentage": 37.3, "elapsed_time": "3:59:09", "remaining_time": "6:41:56"} +{"current_steps": 2002, "total_steps": 5364, "loss": 0.24719780683517456, "lr": 1.5793819659046615e-05, "epoch": 1.1197202797202797, "percentage": 37.32, "elapsed_time": "3:59:16", "remaining_time": "6:41:48"} +{"current_steps": 2003, "total_steps": 5364, "loss": 0.25767379999160767, "lr": 1.5788513741141393e-05, "epoch": 1.1202797202797203, "percentage": 37.34, "elapsed_time": "3:59:23", "remaining_time": "6:41:41"} +{"current_steps": 2004, "total_steps": 5364, "loss": 0.15381020307540894, "lr": 1.578320537128283e-05, "epoch": 1.1208391608391608, "percentage": 37.36, "elapsed_time": "3:59:30", "remaining_time": "6:41:33"} +{"current_steps": 2005, "total_steps": 5364, "loss": 0.2354571372270584, "lr": 1.57778945517195e-05, "epoch": 1.1213986013986015, "percentage": 37.38, "elapsed_time": "3:59:36", "remaining_time": "6:41:25"} +{"current_steps": 2006, "total_steps": 5364, "loss": 0.4068896770477295, "lr": 1.577258128470101e-05, "epoch": 1.121958041958042, "percentage": 37.4, "elapsed_time": "3:59:43", "remaining_time": "6:41:18"} +{"current_steps": 2007, "total_steps": 5364, "loss": 0.04392913728952408, "lr": 1.5767265572478003e-05, "epoch": 1.1225174825174826, "percentage": 37.42, "elapsed_time": "3:59:50", "remaining_time": "6:41:10"} +{"current_steps": 2008, "total_steps": 5364, "loss": 0.021909723058342934, "lr": 1.5761947417302156e-05, "epoch": 1.123076923076923, "percentage": 37.43, "elapsed_time": "3:59:58", "remaining_time": "6:41:04"} +{"current_steps": 2009, "total_steps": 5364, "loss": 0.077776238322258, "lr": 1.575662682142619e-05, "epoch": 1.1236363636363635, "percentage": 37.45, "elapsed_time": "4:00:06", "remaining_time": "6:40:57"} +{"current_steps": 2010, "total_steps": 5364, "loss": 0.03840276971459389, "lr": 1.5751303787103846e-05, "epoch": 1.1241958041958042, "percentage": 37.47, "elapsed_time": "4:00:12", "remaining_time": "6:40:50"} +{"current_steps": 2011, "total_steps": 5364, "loss": 0.19674792885780334, "lr": 1.574597831658991e-05, "epoch": 1.1247552447552447, "percentage": 37.49, "elapsed_time": "4:00:19", "remaining_time": "6:40:42"} +{"current_steps": 2012, "total_steps": 5364, "loss": 0.16901172697544098, "lr": 1.5740650412140194e-05, "epoch": 1.1253146853146854, "percentage": 37.51, "elapsed_time": "4:00:26", "remaining_time": "6:40:34"} +{"current_steps": 2013, "total_steps": 5364, "loss": 0.1439274549484253, "lr": 1.5735320076011537e-05, "epoch": 1.1258741258741258, "percentage": 37.53, "elapsed_time": "4:00:33", "remaining_time": "6:40:27"} +{"current_steps": 2014, "total_steps": 5364, "loss": 0.07391933351755142, "lr": 1.5729987310461822e-05, "epoch": 1.1264335664335665, "percentage": 37.55, "elapsed_time": "4:00:41", "remaining_time": "6:40:21"} +{"current_steps": 2015, "total_steps": 5364, "loss": 0.2885581851005554, "lr": 1.5724652117749947e-05, "epoch": 1.126993006993007, "percentage": 37.57, "elapsed_time": "4:00:48", "remaining_time": "6:40:14"} +{"current_steps": 2016, "total_steps": 5364, "loss": 0.08786419779062271, "lr": 1.5719314500135844e-05, "epoch": 1.1275524475524477, "percentage": 37.58, "elapsed_time": "4:00:56", "remaining_time": "6:40:07"} +{"current_steps": 2017, "total_steps": 5364, "loss": 0.14879518747329712, "lr": 1.5713974459880467e-05, "epoch": 1.1281118881118881, "percentage": 37.6, "elapsed_time": "4:01:03", "remaining_time": "6:40:00"} +{"current_steps": 2018, "total_steps": 5364, "loss": 0.071931391954422, "lr": 1.570863199924581e-05, "epoch": 1.1286713286713286, "percentage": 37.62, "elapsed_time": "4:01:10", "remaining_time": "6:39:52"} +{"current_steps": 2019, "total_steps": 5364, "loss": 0.213780015707016, "lr": 1.570328712049488e-05, "epoch": 1.1292307692307693, "percentage": 37.64, "elapsed_time": "4:01:17", "remaining_time": "6:39:45"} +{"current_steps": 2020, "total_steps": 5364, "loss": 0.31340765953063965, "lr": 1.5697939825891707e-05, "epoch": 1.1297902097902097, "percentage": 37.66, "elapsed_time": "4:01:23", "remaining_time": "6:39:37"} +{"current_steps": 2021, "total_steps": 5364, "loss": 0.17053598165512085, "lr": 1.569259011770135e-05, "epoch": 1.1303496503496504, "percentage": 37.68, "elapsed_time": "4:01:30", "remaining_time": "6:39:29"} +{"current_steps": 2022, "total_steps": 5364, "loss": 0.10571020841598511, "lr": 1.5687237998189892e-05, "epoch": 1.1309090909090909, "percentage": 37.7, "elapsed_time": "4:01:38", "remaining_time": "6:39:22"} +{"current_steps": 2023, "total_steps": 5364, "loss": 0.22666051983833313, "lr": 1.568188346962443e-05, "epoch": 1.1314685314685315, "percentage": 37.71, "elapsed_time": "4:01:45", "remaining_time": "6:39:15"} +{"current_steps": 2024, "total_steps": 5364, "loss": 0.08405138552188873, "lr": 1.567652653427309e-05, "epoch": 1.132027972027972, "percentage": 37.73, "elapsed_time": "4:01:52", "remaining_time": "6:39:08"} +{"current_steps": 2025, "total_steps": 5364, "loss": 0.1688159853219986, "lr": 1.567116719440501e-05, "epoch": 1.1325874125874127, "percentage": 37.75, "elapsed_time": "4:01:59", "remaining_time": "6:39:01"} +{"current_steps": 2026, "total_steps": 5364, "loss": 0.14032645523548126, "lr": 1.5665805452290354e-05, "epoch": 1.1331468531468532, "percentage": 37.77, "elapsed_time": "4:02:07", "remaining_time": "6:38:54"} +{"current_steps": 2027, "total_steps": 5364, "loss": 0.10113655775785446, "lr": 1.5660441310200293e-05, "epoch": 1.1337062937062936, "percentage": 37.79, "elapsed_time": "4:02:14", "remaining_time": "6:38:48"} +{"current_steps": 2028, "total_steps": 5364, "loss": 0.2282724529504776, "lr": 1.5655074770407027e-05, "epoch": 1.1342657342657343, "percentage": 37.81, "elapsed_time": "4:02:21", "remaining_time": "6:38:41"} +{"current_steps": 2029, "total_steps": 5364, "loss": 0.07945576310157776, "lr": 1.5649705835183767e-05, "epoch": 1.1348251748251748, "percentage": 37.83, "elapsed_time": "4:02:28", "remaining_time": "6:38:33"} +{"current_steps": 2030, "total_steps": 5364, "loss": 0.17717701196670532, "lr": 1.5644334506804727e-05, "epoch": 1.1353846153846154, "percentage": 37.84, "elapsed_time": "4:02:36", "remaining_time": "6:38:26"} +{"current_steps": 2031, "total_steps": 5364, "loss": 0.202545627951622, "lr": 1.5638960787545154e-05, "epoch": 1.135944055944056, "percentage": 37.86, "elapsed_time": "4:02:43", "remaining_time": "6:38:19"} +{"current_steps": 2032, "total_steps": 5364, "loss": 0.1656865030527115, "lr": 1.5633584679681295e-05, "epoch": 1.1365034965034966, "percentage": 37.88, "elapsed_time": "4:02:50", "remaining_time": "6:38:11"} +{"current_steps": 2033, "total_steps": 5364, "loss": 0.19921652972698212, "lr": 1.5628206185490414e-05, "epoch": 1.137062937062937, "percentage": 37.9, "elapsed_time": "4:02:57", "remaining_time": "6:38:04"} +{"current_steps": 2034, "total_steps": 5364, "loss": 0.22966957092285156, "lr": 1.5622825307250787e-05, "epoch": 1.1376223776223777, "percentage": 37.92, "elapsed_time": "4:03:04", "remaining_time": "6:37:56"} +{"current_steps": 2035, "total_steps": 5364, "loss": 0.09429632127285004, "lr": 1.5617442047241693e-05, "epoch": 1.1381818181818182, "percentage": 37.94, "elapsed_time": "4:03:10", "remaining_time": "6:37:48"} +{"current_steps": 2036, "total_steps": 5364, "loss": 0.2207282930612564, "lr": 1.5612056407743425e-05, "epoch": 1.1387412587412586, "percentage": 37.96, "elapsed_time": "4:03:17", "remaining_time": "6:37:41"} +{"current_steps": 2037, "total_steps": 5364, "loss": 0.15288980305194855, "lr": 1.5606668391037283e-05, "epoch": 1.1393006993006993, "percentage": 37.98, "elapsed_time": "4:03:24", "remaining_time": "6:37:33"} +{"current_steps": 2038, "total_steps": 5364, "loss": 0.1533864438533783, "lr": 1.560127799940558e-05, "epoch": 1.1398601398601398, "percentage": 37.99, "elapsed_time": "4:03:31", "remaining_time": "6:37:26"} +{"current_steps": 2039, "total_steps": 5364, "loss": 0.14706389605998993, "lr": 1.5595885235131616e-05, "epoch": 1.1404195804195805, "percentage": 38.01, "elapsed_time": "4:03:38", "remaining_time": "6:37:18"} +{"current_steps": 2040, "total_steps": 5364, "loss": 0.1838333010673523, "lr": 1.5590490100499722e-05, "epoch": 1.140979020979021, "percentage": 38.03, "elapsed_time": "4:03:45", "remaining_time": "6:37:11"} +{"current_steps": 2041, "total_steps": 5364, "loss": 0.23537740111351013, "lr": 1.5585092597795216e-05, "epoch": 1.1415384615384616, "percentage": 38.05, "elapsed_time": "4:03:52", "remaining_time": "6:37:03"} +{"current_steps": 2042, "total_steps": 5364, "loss": 0.12658530473709106, "lr": 1.5579692729304418e-05, "epoch": 1.142097902097902, "percentage": 38.07, "elapsed_time": "4:03:59", "remaining_time": "6:36:56"} +{"current_steps": 2043, "total_steps": 5364, "loss": 0.2904822826385498, "lr": 1.557429049731466e-05, "epoch": 1.1426573426573428, "percentage": 38.09, "elapsed_time": "4:04:06", "remaining_time": "6:36:48"} +{"current_steps": 2044, "total_steps": 5364, "loss": 0.2652897834777832, "lr": 1.556888590411427e-05, "epoch": 1.1432167832167832, "percentage": 38.11, "elapsed_time": "4:04:13", "remaining_time": "6:36:41"} +{"current_steps": 2045, "total_steps": 5364, "loss": 0.08475368469953537, "lr": 1.5563478951992575e-05, "epoch": 1.1437762237762237, "percentage": 38.12, "elapsed_time": "4:04:20", "remaining_time": "6:36:34"} +{"current_steps": 2046, "total_steps": 5364, "loss": 0.1557948887348175, "lr": 1.555806964323991e-05, "epoch": 1.1443356643356644, "percentage": 38.14, "elapsed_time": "4:04:27", "remaining_time": "6:36:26"} +{"current_steps": 2047, "total_steps": 5364, "loss": 0.1669459044933319, "lr": 1.555265798014759e-05, "epoch": 1.1448951048951048, "percentage": 38.16, "elapsed_time": "4:04:34", "remaining_time": "6:36:18"} +{"current_steps": 2048, "total_steps": 5364, "loss": 0.0632094219326973, "lr": 1.5547243965007947e-05, "epoch": 1.1454545454545455, "percentage": 38.18, "elapsed_time": "4:04:41", "remaining_time": "6:36:11"} +{"current_steps": 2049, "total_steps": 5364, "loss": 0.10252562165260315, "lr": 1.5541827600114293e-05, "epoch": 1.146013986013986, "percentage": 38.2, "elapsed_time": "4:04:48", "remaining_time": "6:36:03"} +{"current_steps": 2050, "total_steps": 5364, "loss": 0.06701380014419556, "lr": 1.5536408887760952e-05, "epoch": 1.1465734265734266, "percentage": 38.22, "elapsed_time": "4:04:55", "remaining_time": "6:35:56"} +{"current_steps": 2051, "total_steps": 5364, "loss": 0.05121865123510361, "lr": 1.553098783024323e-05, "epoch": 1.147132867132867, "percentage": 38.24, "elapsed_time": "4:05:02", "remaining_time": "6:35:48"} +{"current_steps": 2052, "total_steps": 5364, "loss": 0.1669970452785492, "lr": 1.5525564429857427e-05, "epoch": 1.1476923076923078, "percentage": 38.26, "elapsed_time": "4:05:09", "remaining_time": "6:35:41"} +{"current_steps": 2053, "total_steps": 5364, "loss": 0.056503891944885254, "lr": 1.552013868890084e-05, "epoch": 1.1482517482517482, "percentage": 38.27, "elapsed_time": "4:05:16", "remaining_time": "6:35:34"} +{"current_steps": 2054, "total_steps": 5364, "loss": 0.028636084869503975, "lr": 1.5514710609671754e-05, "epoch": 1.1488111888111887, "percentage": 38.29, "elapsed_time": "4:05:23", "remaining_time": "6:35:27"} +{"current_steps": 2055, "total_steps": 5364, "loss": 0.0977797657251358, "lr": 1.5509280194469447e-05, "epoch": 1.1493706293706294, "percentage": 38.31, "elapsed_time": "4:05:30", "remaining_time": "6:35:19"} +{"current_steps": 2056, "total_steps": 5364, "loss": 0.14876842498779297, "lr": 1.5503847445594188e-05, "epoch": 1.1499300699300699, "percentage": 38.33, "elapsed_time": "4:05:37", "remaining_time": "6:35:11"} +{"current_steps": 2057, "total_steps": 5364, "loss": 0.29423877596855164, "lr": 1.5498412365347228e-05, "epoch": 1.1504895104895105, "percentage": 38.35, "elapsed_time": "4:05:44", "remaining_time": "6:35:04"} +{"current_steps": 2058, "total_steps": 5364, "loss": 0.22696185111999512, "lr": 1.5492974956030815e-05, "epoch": 1.151048951048951, "percentage": 38.37, "elapsed_time": "4:05:51", "remaining_time": "6:34:56"} +{"current_steps": 2059, "total_steps": 5364, "loss": 0.16424477100372314, "lr": 1.5487535219948174e-05, "epoch": 1.1516083916083917, "percentage": 38.39, "elapsed_time": "4:05:58", "remaining_time": "6:34:49"} +{"current_steps": 2060, "total_steps": 5364, "loss": 0.108802929520607, "lr": 1.548209315940352e-05, "epoch": 1.1521678321678321, "percentage": 38.4, "elapsed_time": "4:06:05", "remaining_time": "6:34:41"} +{"current_steps": 2061, "total_steps": 5364, "loss": 0.06705796718597412, "lr": 1.547664877670205e-05, "epoch": 1.1527272727272728, "percentage": 38.42, "elapsed_time": "4:06:11", "remaining_time": "6:34:33"} +{"current_steps": 2062, "total_steps": 5364, "loss": 0.07843948900699615, "lr": 1.5471202074149956e-05, "epoch": 1.1532867132867133, "percentage": 38.44, "elapsed_time": "4:06:18", "remaining_time": "6:34:26"} +{"current_steps": 2063, "total_steps": 5364, "loss": 0.16249790787696838, "lr": 1.54657530540544e-05, "epoch": 1.1538461538461537, "percentage": 38.46, "elapsed_time": "4:06:25", "remaining_time": "6:34:18"} +{"current_steps": 2064, "total_steps": 5364, "loss": 0.23675481975078583, "lr": 1.546030171872352e-05, "epoch": 1.1544055944055944, "percentage": 38.48, "elapsed_time": "4:06:32", "remaining_time": "6:34:11"} +{"current_steps": 2065, "total_steps": 5364, "loss": 0.2960844039916992, "lr": 1.5454848070466458e-05, "epoch": 1.1549650349650349, "percentage": 38.5, "elapsed_time": "4:06:39", "remaining_time": "6:34:03"} +{"current_steps": 2066, "total_steps": 5364, "loss": 0.33044707775115967, "lr": 1.544939211159331e-05, "epoch": 1.1555244755244756, "percentage": 38.52, "elapsed_time": "4:06:46", "remaining_time": "6:33:55"} +{"current_steps": 2067, "total_steps": 5364, "loss": 0.18616685271263123, "lr": 1.544393384441517e-05, "epoch": 1.156083916083916, "percentage": 38.53, "elapsed_time": "4:06:53", "remaining_time": "6:33:48"} +{"current_steps": 2068, "total_steps": 5364, "loss": 0.18014107644557953, "lr": 1.5438473271244105e-05, "epoch": 1.1566433566433567, "percentage": 38.55, "elapsed_time": "4:07:00", "remaining_time": "6:33:41"} +{"current_steps": 2069, "total_steps": 5364, "loss": 0.08684305101633072, "lr": 1.5433010394393147e-05, "epoch": 1.1572027972027972, "percentage": 38.57, "elapsed_time": "4:07:08", "remaining_time": "6:33:34"} +{"current_steps": 2070, "total_steps": 5364, "loss": 0.3271596431732178, "lr": 1.5427545216176317e-05, "epoch": 1.1577622377622379, "percentage": 38.59, "elapsed_time": "4:07:15", "remaining_time": "6:33:27"} +{"current_steps": 2071, "total_steps": 5364, "loss": 0.06100625544786453, "lr": 1.542207773890861e-05, "epoch": 1.1583216783216783, "percentage": 38.61, "elapsed_time": "4:07:21", "remaining_time": "6:33:19"} +{"current_steps": 2072, "total_steps": 5364, "loss": 0.19233372807502747, "lr": 1.5416607964905984e-05, "epoch": 1.1588811188811188, "percentage": 38.63, "elapsed_time": "4:07:29", "remaining_time": "6:33:13"} +{"current_steps": 2073, "total_steps": 5364, "loss": 0.11356054246425629, "lr": 1.541113589648539e-05, "epoch": 1.1594405594405595, "percentage": 38.65, "elapsed_time": "4:07:36", "remaining_time": "6:33:05"} +{"current_steps": 2074, "total_steps": 5364, "loss": 0.2626035511493683, "lr": 1.5405661535964726e-05, "epoch": 1.16, "percentage": 38.67, "elapsed_time": "4:07:43", "remaining_time": "6:32:58"} +{"current_steps": 2075, "total_steps": 5364, "loss": 0.22842754423618317, "lr": 1.5400184885662883e-05, "epoch": 1.1605594405594406, "percentage": 38.68, "elapsed_time": "4:07:50", "remaining_time": "6:32:50"} +{"current_steps": 2076, "total_steps": 5364, "loss": 0.21133379638195038, "lr": 1.53947059478997e-05, "epoch": 1.161118881118881, "percentage": 38.7, "elapsed_time": "4:07:57", "remaining_time": "6:32:43"} +{"current_steps": 2077, "total_steps": 5364, "loss": 0.2178257703781128, "lr": 1.5389224724996012e-05, "epoch": 1.1616783216783217, "percentage": 38.72, "elapsed_time": "4:08:04", "remaining_time": "6:32:35"} +{"current_steps": 2078, "total_steps": 5364, "loss": 0.14156261086463928, "lr": 1.5383741219273596e-05, "epoch": 1.1622377622377622, "percentage": 38.74, "elapsed_time": "4:08:11", "remaining_time": "6:32:28"} +{"current_steps": 2079, "total_steps": 5364, "loss": 0.2085438370704651, "lr": 1.5378255433055215e-05, "epoch": 1.162797202797203, "percentage": 38.76, "elapsed_time": "4:08:18", "remaining_time": "6:32:21"} +{"current_steps": 2080, "total_steps": 5364, "loss": 0.15259240567684174, "lr": 1.5372767368664584e-05, "epoch": 1.1633566433566433, "percentage": 38.78, "elapsed_time": "4:08:27", "remaining_time": "6:32:16"} +{"current_steps": 2081, "total_steps": 5364, "loss": 0.033811211585998535, "lr": 1.536727702842639e-05, "epoch": 1.1639160839160838, "percentage": 38.8, "elapsed_time": "4:08:34", "remaining_time": "6:32:09"} +{"current_steps": 2082, "total_steps": 5364, "loss": 0.06193928048014641, "lr": 1.5361784414666286e-05, "epoch": 1.1644755244755245, "percentage": 38.81, "elapsed_time": "4:08:41", "remaining_time": "6:32:02"} +{"current_steps": 2083, "total_steps": 5364, "loss": 0.09337253868579865, "lr": 1.5356289529710885e-05, "epoch": 1.165034965034965, "percentage": 38.83, "elapsed_time": "4:08:48", "remaining_time": "6:31:54"} +{"current_steps": 2084, "total_steps": 5364, "loss": 0.11022676527500153, "lr": 1.535079237588776e-05, "epoch": 1.1655944055944056, "percentage": 38.85, "elapsed_time": "4:08:55", "remaining_time": "6:31:47"} +{"current_steps": 2085, "total_steps": 5364, "loss": 0.13584671914577484, "lr": 1.5345292955525452e-05, "epoch": 1.166153846153846, "percentage": 38.87, "elapsed_time": "4:09:02", "remaining_time": "6:31:39"} +{"current_steps": 2086, "total_steps": 5364, "loss": 0.06401629745960236, "lr": 1.533979127095345e-05, "epoch": 1.1667132867132868, "percentage": 38.89, "elapsed_time": "4:09:09", "remaining_time": "6:31:32"} +{"current_steps": 2087, "total_steps": 5364, "loss": 0.16698762774467468, "lr": 1.5334287324502216e-05, "epoch": 1.1672727272727272, "percentage": 38.91, "elapsed_time": "4:09:16", "remaining_time": "6:31:24"} +{"current_steps": 2088, "total_steps": 5364, "loss": 0.16536566615104675, "lr": 1.532878111850316e-05, "epoch": 1.167832167832168, "percentage": 38.93, "elapsed_time": "4:09:23", "remaining_time": "6:31:17"} +{"current_steps": 2089, "total_steps": 5364, "loss": 0.05860889330506325, "lr": 1.5323272655288655e-05, "epoch": 1.1683916083916084, "percentage": 38.94, "elapsed_time": "4:09:30", "remaining_time": "6:31:09"} +{"current_steps": 2090, "total_steps": 5364, "loss": 0.16140541434288025, "lr": 1.5317761937192023e-05, "epoch": 1.1689510489510488, "percentage": 38.96, "elapsed_time": "4:09:37", "remaining_time": "6:31:02"} +{"current_steps": 2091, "total_steps": 5364, "loss": 0.23804160952568054, "lr": 1.5312248966547555e-05, "epoch": 1.1695104895104895, "percentage": 38.98, "elapsed_time": "4:09:44", "remaining_time": "6:30:54"} +{"current_steps": 2092, "total_steps": 5364, "loss": 0.08927642554044724, "lr": 1.530673374569048e-05, "epoch": 1.17006993006993, "percentage": 39.0, "elapsed_time": "4:09:51", "remaining_time": "6:30:47"} +{"current_steps": 2093, "total_steps": 5364, "loss": 0.08895355463027954, "lr": 1.530121627695699e-05, "epoch": 1.1706293706293707, "percentage": 39.02, "elapsed_time": "4:09:58", "remaining_time": "6:30:39"} +{"current_steps": 2094, "total_steps": 5364, "loss": 0.1526639759540558, "lr": 1.5295696562684226e-05, "epoch": 1.1711888111888111, "percentage": 39.04, "elapsed_time": "4:10:05", "remaining_time": "6:30:32"} +{"current_steps": 2095, "total_steps": 5364, "loss": 0.17482587695121765, "lr": 1.529017460521028e-05, "epoch": 1.1717482517482518, "percentage": 39.06, "elapsed_time": "4:10:12", "remaining_time": "6:30:25"} +{"current_steps": 2096, "total_steps": 5364, "loss": 0.09317049384117126, "lr": 1.5284650406874197e-05, "epoch": 1.1723076923076923, "percentage": 39.08, "elapsed_time": "4:10:19", "remaining_time": "6:30:18"} +{"current_steps": 2097, "total_steps": 5364, "loss": 0.18568839132785797, "lr": 1.527912397001597e-05, "epoch": 1.172867132867133, "percentage": 39.09, "elapsed_time": "4:10:26", "remaining_time": "6:30:10"} +{"current_steps": 2098, "total_steps": 5364, "loss": 0.2325122058391571, "lr": 1.527359529697653e-05, "epoch": 1.1734265734265734, "percentage": 39.11, "elapsed_time": "4:10:33", "remaining_time": "6:30:03"} +{"current_steps": 2099, "total_steps": 5364, "loss": 0.17921952903270721, "lr": 1.5268064390097777e-05, "epoch": 1.1739860139860139, "percentage": 39.13, "elapsed_time": "4:10:41", "remaining_time": "6:29:57"} +{"current_steps": 2100, "total_steps": 5364, "loss": 0.1375807523727417, "lr": 1.5262531251722535e-05, "epoch": 1.1745454545454546, "percentage": 39.15, "elapsed_time": "4:10:48", "remaining_time": "6:29:50"} +{"current_steps": 2101, "total_steps": 5364, "loss": 0.15246731042861938, "lr": 1.5256995884194593e-05, "epoch": 1.175104895104895, "percentage": 39.17, "elapsed_time": "4:10:56", "remaining_time": "6:29:43"} +{"current_steps": 2102, "total_steps": 5364, "loss": 0.15595918893814087, "lr": 1.5251458289858665e-05, "epoch": 1.1756643356643357, "percentage": 39.19, "elapsed_time": "4:11:03", "remaining_time": "6:29:35"} +{"current_steps": 2103, "total_steps": 5364, "loss": 0.40096986293792725, "lr": 1.5245918471060423e-05, "epoch": 1.1762237762237762, "percentage": 39.21, "elapsed_time": "4:11:10", "remaining_time": "6:29:28"} +{"current_steps": 2104, "total_steps": 5364, "loss": 0.3100760579109192, "lr": 1.5240376430146473e-05, "epoch": 1.1767832167832168, "percentage": 39.22, "elapsed_time": "4:11:17", "remaining_time": "6:29:21"} +{"current_steps": 2105, "total_steps": 5364, "loss": 0.16883251070976257, "lr": 1.5234832169464367e-05, "epoch": 1.1773426573426573, "percentage": 39.24, "elapsed_time": "4:11:24", "remaining_time": "6:29:14"} +{"current_steps": 2106, "total_steps": 5364, "loss": 0.053090669214725494, "lr": 1.5229285691362594e-05, "epoch": 1.177902097902098, "percentage": 39.26, "elapsed_time": "4:11:31", "remaining_time": "6:29:07"} +{"current_steps": 2107, "total_steps": 5364, "loss": 0.09708268940448761, "lr": 1.5223736998190584e-05, "epoch": 1.1784615384615384, "percentage": 39.28, "elapsed_time": "4:11:38", "remaining_time": "6:28:59"} +{"current_steps": 2108, "total_steps": 5364, "loss": 0.3400949537754059, "lr": 1.5218186092298703e-05, "epoch": 1.179020979020979, "percentage": 39.3, "elapsed_time": "4:11:45", "remaining_time": "6:28:52"} +{"current_steps": 2109, "total_steps": 5364, "loss": 0.11515244096517563, "lr": 1.5212632976038259e-05, "epoch": 1.1795804195804196, "percentage": 39.32, "elapsed_time": "4:11:53", "remaining_time": "6:28:45"} +{"current_steps": 2110, "total_steps": 5364, "loss": 0.20428475737571716, "lr": 1.5207077651761494e-05, "epoch": 1.18013986013986, "percentage": 39.34, "elapsed_time": "4:12:00", "remaining_time": "6:28:37"} +{"current_steps": 2111, "total_steps": 5364, "loss": 0.18052145838737488, "lr": 1.5201520121821583e-05, "epoch": 1.1806993006993007, "percentage": 39.35, "elapsed_time": "4:12:07", "remaining_time": "6:28:30"} +{"current_steps": 2112, "total_steps": 5364, "loss": 0.06973645836114883, "lr": 1.5195960388572636e-05, "epoch": 1.1812587412587412, "percentage": 39.37, "elapsed_time": "4:12:14", "remaining_time": "6:28:23"} +{"current_steps": 2113, "total_steps": 5364, "loss": 0.17916619777679443, "lr": 1.5190398454369702e-05, "epoch": 1.1818181818181819, "percentage": 39.39, "elapsed_time": "4:12:21", "remaining_time": "6:28:15"} +{"current_steps": 2114, "total_steps": 5364, "loss": 0.14442399144172668, "lr": 1.518483432156875e-05, "epoch": 1.1823776223776223, "percentage": 39.41, "elapsed_time": "4:12:28", "remaining_time": "6:28:08"} +{"current_steps": 2115, "total_steps": 5364, "loss": 0.2838698625564575, "lr": 1.5179267992526692e-05, "epoch": 1.182937062937063, "percentage": 39.43, "elapsed_time": "4:12:35", "remaining_time": "6:28:01"} +{"current_steps": 2116, "total_steps": 5364, "loss": 0.14477328956127167, "lr": 1.5173699469601369e-05, "epoch": 1.1834965034965035, "percentage": 39.45, "elapsed_time": "4:12:43", "remaining_time": "6:27:54"} +{"current_steps": 2117, "total_steps": 5364, "loss": 0.14884303510189056, "lr": 1.5168128755151546e-05, "epoch": 1.184055944055944, "percentage": 39.47, "elapsed_time": "4:12:50", "remaining_time": "6:27:47"} +{"current_steps": 2118, "total_steps": 5364, "loss": 0.19063085317611694, "lr": 1.5162555851536916e-05, "epoch": 1.1846153846153846, "percentage": 39.49, "elapsed_time": "4:12:56", "remaining_time": "6:27:39"} +{"current_steps": 2119, "total_steps": 5364, "loss": 0.12622380256652832, "lr": 1.5156980761118107e-05, "epoch": 1.185174825174825, "percentage": 39.5, "elapsed_time": "4:13:03", "remaining_time": "6:27:31"} +{"current_steps": 2120, "total_steps": 5364, "loss": 0.14665603637695312, "lr": 1.5151403486256668e-05, "epoch": 1.1857342657342658, "percentage": 39.52, "elapsed_time": "4:13:10", "remaining_time": "6:27:24"} +{"current_steps": 2121, "total_steps": 5364, "loss": 0.10852214694023132, "lr": 1.5145824029315069e-05, "epoch": 1.1862937062937062, "percentage": 39.54, "elapsed_time": "4:13:18", "remaining_time": "6:27:17"} +{"current_steps": 2122, "total_steps": 5364, "loss": 0.12678593397140503, "lr": 1.5140242392656715e-05, "epoch": 1.186853146853147, "percentage": 39.56, "elapsed_time": "4:13:24", "remaining_time": "6:27:10"} +{"current_steps": 2123, "total_steps": 5364, "loss": 0.09018857777118683, "lr": 1.5134658578645924e-05, "epoch": 1.1874125874125874, "percentage": 39.58, "elapsed_time": "4:13:32", "remaining_time": "6:27:03"} +{"current_steps": 2124, "total_steps": 5364, "loss": 0.11733860522508621, "lr": 1.5129072589647945e-05, "epoch": 1.187972027972028, "percentage": 39.6, "elapsed_time": "4:13:39", "remaining_time": "6:26:56"} +{"current_steps": 2125, "total_steps": 5364, "loss": 0.0967697724699974, "lr": 1.5123484428028938e-05, "epoch": 1.1885314685314685, "percentage": 39.62, "elapsed_time": "4:13:46", "remaining_time": "6:26:48"} +{"current_steps": 2126, "total_steps": 5364, "loss": 0.18317899107933044, "lr": 1.5117894096155992e-05, "epoch": 1.189090909090909, "percentage": 39.63, "elapsed_time": "4:13:54", "remaining_time": "6:26:42"} +{"current_steps": 2127, "total_steps": 5364, "loss": 0.15682217478752136, "lr": 1.5112301596397117e-05, "epoch": 1.1896503496503497, "percentage": 39.65, "elapsed_time": "4:14:01", "remaining_time": "6:26:34"} +{"current_steps": 2128, "total_steps": 5364, "loss": 0.0676366314291954, "lr": 1.510670693112123e-05, "epoch": 1.1902097902097901, "percentage": 39.67, "elapsed_time": "4:14:07", "remaining_time": "6:26:27"} +{"current_steps": 2129, "total_steps": 5364, "loss": 0.11411210894584656, "lr": 1.5101110102698178e-05, "epoch": 1.1907692307692308, "percentage": 39.69, "elapsed_time": "4:14:15", "remaining_time": "6:26:20"} +{"current_steps": 2130, "total_steps": 5364, "loss": 0.15485668182373047, "lr": 1.5095511113498713e-05, "epoch": 1.1913286713286713, "percentage": 39.71, "elapsed_time": "4:14:22", "remaining_time": "6:26:12"} +{"current_steps": 2131, "total_steps": 5364, "loss": 0.2711738348007202, "lr": 1.5089909965894513e-05, "epoch": 1.191888111888112, "percentage": 39.73, "elapsed_time": "4:14:29", "remaining_time": "6:26:05"} +{"current_steps": 2132, "total_steps": 5364, "loss": 0.1966949701309204, "lr": 1.5084306662258162e-05, "epoch": 1.1924475524475524, "percentage": 39.75, "elapsed_time": "4:14:36", "remaining_time": "6:25:58"} +{"current_steps": 2133, "total_steps": 5364, "loss": 0.23829127848148346, "lr": 1.507870120496316e-05, "epoch": 1.193006993006993, "percentage": 39.77, "elapsed_time": "4:14:43", "remaining_time": "6:25:50"} +{"current_steps": 2134, "total_steps": 5364, "loss": 0.06377270817756653, "lr": 1.5073093596383918e-05, "epoch": 1.1935664335664335, "percentage": 39.78, "elapsed_time": "4:14:50", "remaining_time": "6:25:43"} +{"current_steps": 2135, "total_steps": 5364, "loss": 0.29641926288604736, "lr": 1.5067483838895767e-05, "epoch": 1.194125874125874, "percentage": 39.8, "elapsed_time": "4:14:57", "remaining_time": "6:25:36"} +{"current_steps": 2136, "total_steps": 5364, "loss": 0.29890936613082886, "lr": 1.5061871934874934e-05, "epoch": 1.1946853146853147, "percentage": 39.82, "elapsed_time": "4:15:05", "remaining_time": "6:25:29"} +{"current_steps": 2137, "total_steps": 5364, "loss": 0.0704234391450882, "lr": 1.5056257886698566e-05, "epoch": 1.1952447552447552, "percentage": 39.84, "elapsed_time": "4:15:12", "remaining_time": "6:25:22"} +{"current_steps": 2138, "total_steps": 5364, "loss": 0.1706911325454712, "lr": 1.5050641696744711e-05, "epoch": 1.1958041958041958, "percentage": 39.86, "elapsed_time": "4:15:18", "remaining_time": "6:25:14"} +{"current_steps": 2139, "total_steps": 5364, "loss": 0.2793387174606323, "lr": 1.504502336739233e-05, "epoch": 1.1963636363636363, "percentage": 39.88, "elapsed_time": "4:15:25", "remaining_time": "6:25:06"} +{"current_steps": 2140, "total_steps": 5364, "loss": 0.029729411005973816, "lr": 1.5039402901021292e-05, "epoch": 1.196923076923077, "percentage": 39.9, "elapsed_time": "4:15:32", "remaining_time": "6:24:59"} +{"current_steps": 2141, "total_steps": 5364, "loss": 0.23659291863441467, "lr": 1.5033780300012362e-05, "epoch": 1.1974825174825174, "percentage": 39.91, "elapsed_time": "4:15:39", "remaining_time": "6:24:51"} +{"current_steps": 2142, "total_steps": 5364, "loss": 0.10613439232110977, "lr": 1.502815556674721e-05, "epoch": 1.1980419580419581, "percentage": 39.93, "elapsed_time": "4:15:46", "remaining_time": "6:24:44"} +{"current_steps": 2143, "total_steps": 5364, "loss": 0.34353798627853394, "lr": 1.5022528703608425e-05, "epoch": 1.1986013986013986, "percentage": 39.95, "elapsed_time": "4:15:53", "remaining_time": "6:24:36"} +{"current_steps": 2144, "total_steps": 5364, "loss": 0.2048952281475067, "lr": 1.5016899712979478e-05, "epoch": 1.199160839160839, "percentage": 39.97, "elapsed_time": "4:16:00", "remaining_time": "6:24:29"} +{"current_steps": 2145, "total_steps": 5364, "loss": 0.17696765065193176, "lr": 1.5011268597244756e-05, "epoch": 1.1997202797202797, "percentage": 39.99, "elapsed_time": "4:16:07", "remaining_time": "6:24:21"} +{"current_steps": 2146, "total_steps": 5364, "loss": 0.09329492598772049, "lr": 1.5005635358789536e-05, "epoch": 1.2002797202797202, "percentage": 40.01, "elapsed_time": "4:16:14", "remaining_time": "6:24:14"} +{"current_steps": 2147, "total_steps": 5364, "loss": 0.060803212225437164, "lr": 1.5000000000000002e-05, "epoch": 1.2008391608391609, "percentage": 40.03, "elapsed_time": "4:16:21", "remaining_time": "6:24:07"} +{"current_steps": 2148, "total_steps": 5364, "loss": 0.1763313263654709, "lr": 1.4994362523263229e-05, "epoch": 1.2013986013986013, "percentage": 40.04, "elapsed_time": "4:16:28", "remaining_time": "6:24:00"} +{"current_steps": 2149, "total_steps": 5364, "loss": 0.3552679419517517, "lr": 1.4988722930967191e-05, "epoch": 1.201958041958042, "percentage": 40.06, "elapsed_time": "4:16:35", "remaining_time": "6:23:53"} +{"current_steps": 2150, "total_steps": 5364, "loss": 0.08860057592391968, "lr": 1.4983081225500766e-05, "epoch": 1.2025174825174825, "percentage": 40.08, "elapsed_time": "4:16:42", "remaining_time": "6:23:45"} +{"current_steps": 2151, "total_steps": 5364, "loss": 0.06494182348251343, "lr": 1.4977437409253717e-05, "epoch": 1.2030769230769232, "percentage": 40.1, "elapsed_time": "4:16:50", "remaining_time": "6:23:38"} +{"current_steps": 2152, "total_steps": 5364, "loss": 0.2027955800294876, "lr": 1.4971791484616707e-05, "epoch": 1.2036363636363636, "percentage": 40.12, "elapsed_time": "4:16:56", "remaining_time": "6:23:30"} +{"current_steps": 2153, "total_steps": 5364, "loss": 0.17163538932800293, "lr": 1.4966143453981286e-05, "epoch": 1.204195804195804, "percentage": 40.14, "elapsed_time": "4:17:03", "remaining_time": "6:23:22"} +{"current_steps": 2154, "total_steps": 5364, "loss": 0.12215086817741394, "lr": 1.4960493319739905e-05, "epoch": 1.2047552447552448, "percentage": 40.16, "elapsed_time": "4:17:10", "remaining_time": "6:23:15"} +{"current_steps": 2155, "total_steps": 5364, "loss": 0.15273746848106384, "lr": 1.4954841084285899e-05, "epoch": 1.2053146853146852, "percentage": 40.18, "elapsed_time": "4:17:17", "remaining_time": "6:23:07"} +{"current_steps": 2156, "total_steps": 5364, "loss": 0.0647292509675026, "lr": 1.4949186750013493e-05, "epoch": 1.205874125874126, "percentage": 40.19, "elapsed_time": "4:17:24", "remaining_time": "6:23:00"} +{"current_steps": 2157, "total_steps": 5364, "loss": 0.012553836219012737, "lr": 1.4943530319317806e-05, "epoch": 1.2064335664335664, "percentage": 40.21, "elapsed_time": "4:17:31", "remaining_time": "6:22:52"} +{"current_steps": 2158, "total_steps": 5364, "loss": 0.09680251032114029, "lr": 1.4937871794594847e-05, "epoch": 1.206993006993007, "percentage": 40.23, "elapsed_time": "4:17:38", "remaining_time": "6:22:45"} +{"current_steps": 2159, "total_steps": 5364, "loss": 0.12145292013883591, "lr": 1.49322111782415e-05, "epoch": 1.2075524475524475, "percentage": 40.25, "elapsed_time": "4:17:45", "remaining_time": "6:22:38"} +{"current_steps": 2160, "total_steps": 5364, "loss": 0.31201595067977905, "lr": 1.4926548472655545e-05, "epoch": 1.2081118881118882, "percentage": 40.27, "elapsed_time": "4:17:52", "remaining_time": "6:22:30"} +{"current_steps": 2161, "total_steps": 5364, "loss": 0.11184802651405334, "lr": 1.4920883680235643e-05, "epoch": 1.2086713286713286, "percentage": 40.29, "elapsed_time": "4:17:59", "remaining_time": "6:22:22"} +{"current_steps": 2162, "total_steps": 5364, "loss": 0.18981671333312988, "lr": 1.4915216803381341e-05, "epoch": 1.209230769230769, "percentage": 40.31, "elapsed_time": "4:18:06", "remaining_time": "6:22:15"} +{"current_steps": 2163, "total_steps": 5364, "loss": 0.0899486094713211, "lr": 1.4909547844493074e-05, "epoch": 1.2097902097902098, "percentage": 40.32, "elapsed_time": "4:18:12", "remaining_time": "6:22:07"} +{"current_steps": 2164, "total_steps": 5364, "loss": 0.06900287419557571, "lr": 1.4903876805972142e-05, "epoch": 1.2103496503496505, "percentage": 40.34, "elapsed_time": "4:18:19", "remaining_time": "6:22:00"} +{"current_steps": 2165, "total_steps": 5364, "loss": 0.13189850747585297, "lr": 1.4898203690220742e-05, "epoch": 1.210909090909091, "percentage": 40.36, "elapsed_time": "4:18:27", "remaining_time": "6:21:53"} +{"current_steps": 2166, "total_steps": 5364, "loss": 0.10305362939834595, "lr": 1.4892528499641952e-05, "epoch": 1.2114685314685314, "percentage": 40.38, "elapsed_time": "4:18:34", "remaining_time": "6:21:46"} +{"current_steps": 2167, "total_steps": 5364, "loss": 0.25402671098709106, "lr": 1.4886851236639715e-05, "epoch": 1.212027972027972, "percentage": 40.4, "elapsed_time": "4:18:41", "remaining_time": "6:21:39"} +{"current_steps": 2168, "total_steps": 5364, "loss": 0.07016675174236298, "lr": 1.4881171903618863e-05, "epoch": 1.2125874125874125, "percentage": 40.42, "elapsed_time": "4:18:49", "remaining_time": "6:21:32"} +{"current_steps": 2169, "total_steps": 5364, "loss": 0.1338023543357849, "lr": 1.48754905029851e-05, "epoch": 1.2131468531468532, "percentage": 40.44, "elapsed_time": "4:18:56", "remaining_time": "6:21:25"} +{"current_steps": 2170, "total_steps": 5364, "loss": 0.2672594487667084, "lr": 1.4869807037145009e-05, "epoch": 1.2137062937062937, "percentage": 40.45, "elapsed_time": "4:19:03", "remaining_time": "6:21:17"} +{"current_steps": 2171, "total_steps": 5364, "loss": 0.0858537033200264, "lr": 1.4864121508506046e-05, "epoch": 1.2142657342657341, "percentage": 40.47, "elapsed_time": "4:19:10", "remaining_time": "6:21:10"} +{"current_steps": 2172, "total_steps": 5364, "loss": 0.037628188729286194, "lr": 1.4858433919476543e-05, "epoch": 1.2148251748251748, "percentage": 40.49, "elapsed_time": "4:19:17", "remaining_time": "6:21:03"} +{"current_steps": 2173, "total_steps": 5364, "loss": 0.05113109573721886, "lr": 1.4852744272465696e-05, "epoch": 1.2153846153846155, "percentage": 40.51, "elapsed_time": "4:19:24", "remaining_time": "6:20:56"} +{"current_steps": 2174, "total_steps": 5364, "loss": 0.18549975752830505, "lr": 1.4847052569883585e-05, "epoch": 1.215944055944056, "percentage": 40.53, "elapsed_time": "4:19:31", "remaining_time": "6:20:48"} +{"current_steps": 2175, "total_steps": 5364, "loss": 0.04100503772497177, "lr": 1.4841358814141154e-05, "epoch": 1.2165034965034964, "percentage": 40.55, "elapsed_time": "4:19:38", "remaining_time": "6:20:41"} +{"current_steps": 2176, "total_steps": 5364, "loss": 0.1358257234096527, "lr": 1.4835663007650218e-05, "epoch": 1.2170629370629371, "percentage": 40.57, "elapsed_time": "4:19:45", "remaining_time": "6:20:33"} +{"current_steps": 2177, "total_steps": 5364, "loss": 0.16321472823619843, "lr": 1.482996515282346e-05, "epoch": 1.2176223776223776, "percentage": 40.59, "elapsed_time": "4:19:52", "remaining_time": "6:20:26"} +{"current_steps": 2178, "total_steps": 5364, "loss": 0.13497884571552277, "lr": 1.482426525207443e-05, "epoch": 1.2181818181818183, "percentage": 40.6, "elapsed_time": "4:19:59", "remaining_time": "6:20:19"} +{"current_steps": 2179, "total_steps": 5364, "loss": 0.21831157803535461, "lr": 1.4818563307817546e-05, "epoch": 1.2187412587412587, "percentage": 40.62, "elapsed_time": "4:20:07", "remaining_time": "6:20:13"} +{"current_steps": 2180, "total_steps": 5364, "loss": 0.241571843624115, "lr": 1.4812859322468099e-05, "epoch": 1.2193006993006994, "percentage": 40.64, "elapsed_time": "4:20:14", "remaining_time": "6:20:05"} +{"current_steps": 2181, "total_steps": 5364, "loss": 0.029868807643651962, "lr": 1.4807153298442226e-05, "epoch": 1.2198601398601399, "percentage": 40.66, "elapsed_time": "4:20:21", "remaining_time": "6:19:58"} +{"current_steps": 2182, "total_steps": 5364, "loss": 0.10821191221475601, "lr": 1.4801445238156942e-05, "epoch": 1.2204195804195805, "percentage": 40.68, "elapsed_time": "4:20:28", "remaining_time": "6:19:50"} +{"current_steps": 2183, "total_steps": 5364, "loss": 0.0779876857995987, "lr": 1.4795735144030121e-05, "epoch": 1.220979020979021, "percentage": 40.7, "elapsed_time": "4:20:35", "remaining_time": "6:19:43"} +{"current_steps": 2184, "total_steps": 5364, "loss": 0.10421731323003769, "lr": 1.4790023018480502e-05, "epoch": 1.2215384615384615, "percentage": 40.72, "elapsed_time": "4:20:42", "remaining_time": "6:19:35"} +{"current_steps": 2185, "total_steps": 5364, "loss": 0.1073048859834671, "lr": 1.478430886392768e-05, "epoch": 1.2220979020979021, "percentage": 40.73, "elapsed_time": "4:20:49", "remaining_time": "6:19:28"} +{"current_steps": 2186, "total_steps": 5364, "loss": 0.15002557635307312, "lr": 1.477859268279211e-05, "epoch": 1.2226573426573426, "percentage": 40.75, "elapsed_time": "4:20:56", "remaining_time": "6:19:20"} +{"current_steps": 2187, "total_steps": 5364, "loss": 0.38225501775741577, "lr": 1.4772874477495104e-05, "epoch": 1.2232167832167833, "percentage": 40.77, "elapsed_time": "4:21:02", "remaining_time": "6:19:13"} +{"current_steps": 2188, "total_steps": 5364, "loss": 0.12482350319623947, "lr": 1.4767154250458836e-05, "epoch": 1.2237762237762237, "percentage": 40.79, "elapsed_time": "4:21:10", "remaining_time": "6:19:07"} +{"current_steps": 2189, "total_steps": 5364, "loss": 0.10538751631975174, "lr": 1.4761432004106331e-05, "epoch": 1.2243356643356644, "percentage": 40.81, "elapsed_time": "4:21:17", "remaining_time": "6:18:59"} +{"current_steps": 2190, "total_steps": 5364, "loss": 0.06819538027048111, "lr": 1.4755707740861476e-05, "epoch": 1.224895104895105, "percentage": 40.83, "elapsed_time": "4:21:24", "remaining_time": "6:18:51"} +{"current_steps": 2191, "total_steps": 5364, "loss": 0.24481016397476196, "lr": 1.4749981463149005e-05, "epoch": 1.2254545454545456, "percentage": 40.85, "elapsed_time": "4:21:31", "remaining_time": "6:18:44"} +{"current_steps": 2192, "total_steps": 5364, "loss": 0.22804947197437286, "lr": 1.4744253173394509e-05, "epoch": 1.226013986013986, "percentage": 40.87, "elapsed_time": "4:21:38", "remaining_time": "6:18:37"} +{"current_steps": 2193, "total_steps": 5364, "loss": 0.08752113580703735, "lr": 1.4738522874024435e-05, "epoch": 1.2265734265734265, "percentage": 40.88, "elapsed_time": "4:21:45", "remaining_time": "6:18:30"} +{"current_steps": 2194, "total_steps": 5364, "loss": 0.17888322472572327, "lr": 1.4732790567466074e-05, "epoch": 1.2271328671328672, "percentage": 40.9, "elapsed_time": "4:21:53", "remaining_time": "6:18:22"} +{"current_steps": 2195, "total_steps": 5364, "loss": 0.0452457033097744, "lr": 1.4727056256147569e-05, "epoch": 1.2276923076923076, "percentage": 40.92, "elapsed_time": "4:21:59", "remaining_time": "6:18:15"} +{"current_steps": 2196, "total_steps": 5364, "loss": 0.24215678870677948, "lr": 1.4721319942497915e-05, "epoch": 1.2282517482517483, "percentage": 40.94, "elapsed_time": "4:22:06", "remaining_time": "6:18:08"} +{"current_steps": 2197, "total_steps": 5364, "loss": 0.03775346279144287, "lr": 1.471558162894696e-05, "epoch": 1.2288111888111888, "percentage": 40.96, "elapsed_time": "4:22:13", "remaining_time": "6:18:00"} +{"current_steps": 2198, "total_steps": 5364, "loss": 0.030302587896585464, "lr": 1.4709841317925384e-05, "epoch": 1.2293706293706295, "percentage": 40.98, "elapsed_time": "4:22:21", "remaining_time": "6:17:53"} +{"current_steps": 2199, "total_steps": 5364, "loss": 0.08427294343709946, "lr": 1.470409901186473e-05, "epoch": 1.22993006993007, "percentage": 41.0, "elapsed_time": "4:22:28", "remaining_time": "6:17:46"} +{"current_steps": 2200, "total_steps": 5364, "loss": 0.15774285793304443, "lr": 1.4698354713197371e-05, "epoch": 1.2304895104895106, "percentage": 41.01, "elapsed_time": "4:22:35", "remaining_time": "6:17:39"} +{"current_steps": 2201, "total_steps": 5364, "loss": 0.15683630108833313, "lr": 1.4692608424356536e-05, "epoch": 1.231048951048951, "percentage": 41.03, "elapsed_time": "4:22:42", "remaining_time": "6:17:31"} +{"current_steps": 2202, "total_steps": 5364, "loss": 0.13205881416797638, "lr": 1.4686860147776292e-05, "epoch": 1.2316083916083915, "percentage": 41.05, "elapsed_time": "4:22:49", "remaining_time": "6:17:24"} +{"current_steps": 2203, "total_steps": 5364, "loss": 0.16652852296829224, "lr": 1.4681109885891546e-05, "epoch": 1.2321678321678322, "percentage": 41.07, "elapsed_time": "4:22:56", "remaining_time": "6:17:17"} +{"current_steps": 2204, "total_steps": 5364, "loss": 0.17532968521118164, "lr": 1.4675357641138051e-05, "epoch": 1.2327272727272727, "percentage": 41.09, "elapsed_time": "4:23:04", "remaining_time": "6:17:10"} +{"current_steps": 2205, "total_steps": 5364, "loss": 0.0872238501906395, "lr": 1.466960341595239e-05, "epoch": 1.2332867132867134, "percentage": 41.11, "elapsed_time": "4:23:11", "remaining_time": "6:17:03"} +{"current_steps": 2206, "total_steps": 5364, "loss": 0.19453716278076172, "lr": 1.4663847212772005e-05, "epoch": 1.2338461538461538, "percentage": 41.13, "elapsed_time": "4:23:18", "remaining_time": "6:16:56"} +{"current_steps": 2207, "total_steps": 5364, "loss": 0.1096837967634201, "lr": 1.4658089034035152e-05, "epoch": 1.2344055944055945, "percentage": 41.14, "elapsed_time": "4:23:25", "remaining_time": "6:16:49"} +{"current_steps": 2208, "total_steps": 5364, "loss": 0.155201256275177, "lr": 1.4652328882180942e-05, "epoch": 1.234965034965035, "percentage": 41.16, "elapsed_time": "4:23:33", "remaining_time": "6:16:42"} +{"current_steps": 2209, "total_steps": 5364, "loss": 0.06926161050796509, "lr": 1.4646566759649308e-05, "epoch": 1.2355244755244756, "percentage": 41.18, "elapsed_time": "4:23:40", "remaining_time": "6:16:35"} +{"current_steps": 2210, "total_steps": 5364, "loss": 0.13178397715091705, "lr": 1.4640802668881027e-05, "epoch": 1.236083916083916, "percentage": 41.2, "elapsed_time": "4:23:48", "remaining_time": "6:16:29"} +{"current_steps": 2211, "total_steps": 5364, "loss": 0.026109972968697548, "lr": 1.4635036612317708e-05, "epoch": 1.2366433566433566, "percentage": 41.22, "elapsed_time": "4:23:55", "remaining_time": "6:16:22"} +{"current_steps": 2212, "total_steps": 5364, "loss": 0.11033852398395538, "lr": 1.4629268592401794e-05, "epoch": 1.2372027972027972, "percentage": 41.24, "elapsed_time": "4:24:02", "remaining_time": "6:16:14"} +{"current_steps": 2213, "total_steps": 5364, "loss": 0.13531993329524994, "lr": 1.4623498611576549e-05, "epoch": 1.2377622377622377, "percentage": 41.26, "elapsed_time": "4:24:09", "remaining_time": "6:16:07"} +{"current_steps": 2214, "total_steps": 5364, "loss": 0.11963668465614319, "lr": 1.4617726672286088e-05, "epoch": 1.2383216783216784, "percentage": 41.28, "elapsed_time": "4:24:16", "remaining_time": "6:15:59"} +{"current_steps": 2215, "total_steps": 5364, "loss": 0.08609425276517868, "lr": 1.461195277697533e-05, "epoch": 1.2388811188811188, "percentage": 41.29, "elapsed_time": "4:24:23", "remaining_time": "6:15:53"} +{"current_steps": 2216, "total_steps": 5364, "loss": 0.16657137870788574, "lr": 1.460617692809005e-05, "epoch": 1.2394405594405595, "percentage": 41.31, "elapsed_time": "4:24:31", "remaining_time": "6:15:46"} +{"current_steps": 2217, "total_steps": 5364, "loss": 0.23990467190742493, "lr": 1.4600399128076825e-05, "epoch": 1.24, "percentage": 41.33, "elapsed_time": "4:24:38", "remaining_time": "6:15:39"} +{"current_steps": 2218, "total_steps": 5364, "loss": 0.2109047770500183, "lr": 1.4594619379383075e-05, "epoch": 1.2405594405594407, "percentage": 41.35, "elapsed_time": "4:24:45", "remaining_time": "6:15:31"} +{"current_steps": 2219, "total_steps": 5364, "loss": 0.14184972643852234, "lr": 1.4588837684457043e-05, "epoch": 1.2411188811188811, "percentage": 41.37, "elapsed_time": "4:24:52", "remaining_time": "6:15:24"} +{"current_steps": 2220, "total_steps": 5364, "loss": 0.21545779705047607, "lr": 1.4583054045747789e-05, "epoch": 1.2416783216783216, "percentage": 41.39, "elapsed_time": "4:24:59", "remaining_time": "6:15:17"} +{"current_steps": 2221, "total_steps": 5364, "loss": 0.10148483514785767, "lr": 1.4577268465705205e-05, "epoch": 1.2422377622377623, "percentage": 41.41, "elapsed_time": "4:25:07", "remaining_time": "6:15:10"} +{"current_steps": 2222, "total_steps": 5364, "loss": 0.11905775219202042, "lr": 1.457148094678e-05, "epoch": 1.2427972027972027, "percentage": 41.42, "elapsed_time": "4:25:14", "remaining_time": "6:15:03"} +{"current_steps": 2223, "total_steps": 5364, "loss": 0.18257169425487518, "lr": 1.4565691491423705e-05, "epoch": 1.2433566433566434, "percentage": 41.44, "elapsed_time": "4:25:21", "remaining_time": "6:14:56"} +{"current_steps": 2224, "total_steps": 5364, "loss": 0.1472766399383545, "lr": 1.4559900102088676e-05, "epoch": 1.2439160839160839, "percentage": 41.46, "elapsed_time": "4:25:28", "remaining_time": "6:14:48"} +{"current_steps": 2225, "total_steps": 5364, "loss": 0.12070463597774506, "lr": 1.4554106781228084e-05, "epoch": 1.2444755244755246, "percentage": 41.48, "elapsed_time": "4:25:35", "remaining_time": "6:14:41"} +{"current_steps": 2226, "total_steps": 5364, "loss": 0.09103168547153473, "lr": 1.4548311531295913e-05, "epoch": 1.245034965034965, "percentage": 41.5, "elapsed_time": "4:25:42", "remaining_time": "6:14:34"} +{"current_steps": 2227, "total_steps": 5364, "loss": 0.11128625273704529, "lr": 1.4542514354746973e-05, "epoch": 1.2455944055944057, "percentage": 41.52, "elapsed_time": "4:25:50", "remaining_time": "6:14:27"} +{"current_steps": 2228, "total_steps": 5364, "loss": 0.08916842937469482, "lr": 1.4536715254036885e-05, "epoch": 1.2461538461538462, "percentage": 41.54, "elapsed_time": "4:25:57", "remaining_time": "6:14:20"} +{"current_steps": 2229, "total_steps": 5364, "loss": 0.17227885127067566, "lr": 1.4530914231622095e-05, "epoch": 1.2467132867132866, "percentage": 41.55, "elapsed_time": "4:26:04", "remaining_time": "6:14:13"} +{"current_steps": 2230, "total_steps": 5364, "loss": 0.13030363619327545, "lr": 1.4525111289959845e-05, "epoch": 1.2472727272727273, "percentage": 41.57, "elapsed_time": "4:26:11", "remaining_time": "6:14:06"} +{"current_steps": 2231, "total_steps": 5364, "loss": 0.055643633008003235, "lr": 1.4519306431508204e-05, "epoch": 1.2478321678321678, "percentage": 41.59, "elapsed_time": "4:26:18", "remaining_time": "6:13:58"} +{"current_steps": 2232, "total_steps": 5364, "loss": 0.07306533306837082, "lr": 1.4513499658726049e-05, "epoch": 1.2483916083916085, "percentage": 41.61, "elapsed_time": "4:26:25", "remaining_time": "6:13:51"} +{"current_steps": 2233, "total_steps": 5364, "loss": 0.15479454398155212, "lr": 1.4507690974073066e-05, "epoch": 1.248951048951049, "percentage": 41.63, "elapsed_time": "4:26:32", "remaining_time": "6:13:43"} +{"current_steps": 2234, "total_steps": 5364, "loss": 0.1688430905342102, "lr": 1.4501880380009756e-05, "epoch": 1.2495104895104896, "percentage": 41.65, "elapsed_time": "4:26:39", "remaining_time": "6:13:36"} +{"current_steps": 2235, "total_steps": 5364, "loss": 0.25149592757225037, "lr": 1.4496067878997419e-05, "epoch": 1.25006993006993, "percentage": 41.67, "elapsed_time": "4:26:46", "remaining_time": "6:13:29"} +{"current_steps": 2236, "total_steps": 5364, "loss": 0.4366171956062317, "lr": 1.4490253473498177e-05, "epoch": 1.2506293706293707, "percentage": 41.69, "elapsed_time": "4:26:53", "remaining_time": "6:13:22"} +{"current_steps": 2237, "total_steps": 5364, "loss": 0.033375732600688934, "lr": 1.4484437165974945e-05, "epoch": 1.2511888111888112, "percentage": 41.7, "elapsed_time": "4:27:01", "remaining_time": "6:13:15"} +{"current_steps": 2238, "total_steps": 5364, "loss": 0.25928834080696106, "lr": 1.4478618958891452e-05, "epoch": 1.2517482517482517, "percentage": 41.72, "elapsed_time": "4:27:08", "remaining_time": "6:13:07"} +{"current_steps": 2239, "total_steps": 5364, "loss": 0.2190554291009903, "lr": 1.4472798854712229e-05, "epoch": 1.2523076923076923, "percentage": 41.74, "elapsed_time": "4:27:15", "remaining_time": "6:13:00"} +{"current_steps": 2240, "total_steps": 5364, "loss": 0.16690543293952942, "lr": 1.4466976855902609e-05, "epoch": 1.2528671328671328, "percentage": 41.76, "elapsed_time": "4:27:22", "remaining_time": "6:12:52"} +{"current_steps": 2241, "total_steps": 5364, "loss": 0.0784834772348404, "lr": 1.4461152964928729e-05, "epoch": 1.2534265734265735, "percentage": 41.78, "elapsed_time": "4:27:29", "remaining_time": "6:12:45"} +{"current_steps": 2242, "total_steps": 5364, "loss": 0.16247597336769104, "lr": 1.4455327184257533e-05, "epoch": 1.253986013986014, "percentage": 41.8, "elapsed_time": "4:27:36", "remaining_time": "6:12:38"} +{"current_steps": 2243, "total_steps": 5364, "loss": 0.1705142706632614, "lr": 1.4449499516356755e-05, "epoch": 1.2545454545454544, "percentage": 41.82, "elapsed_time": "4:27:43", "remaining_time": "6:12:31"} +{"current_steps": 2244, "total_steps": 5364, "loss": 0.021440725773572922, "lr": 1.4443669963694934e-05, "epoch": 1.255104895104895, "percentage": 41.83, "elapsed_time": "4:27:50", "remaining_time": "6:12:24"} +{"current_steps": 2245, "total_steps": 5364, "loss": 0.18354830145835876, "lr": 1.4437838528741408e-05, "epoch": 1.2556643356643358, "percentage": 41.85, "elapsed_time": "4:27:57", "remaining_time": "6:12:17"} +{"current_steps": 2246, "total_steps": 5364, "loss": 0.1939404457807541, "lr": 1.4432005213966312e-05, "epoch": 1.2562237762237762, "percentage": 41.87, "elapsed_time": "4:28:05", "remaining_time": "6:12:10"} +{"current_steps": 2247, "total_steps": 5364, "loss": 0.07651235908269882, "lr": 1.4426170021840577e-05, "epoch": 1.2567832167832167, "percentage": 41.89, "elapsed_time": "4:28:13", "remaining_time": "6:12:04"} +{"current_steps": 2248, "total_steps": 5364, "loss": 0.046428948640823364, "lr": 1.4420332954835927e-05, "epoch": 1.2573426573426574, "percentage": 41.91, "elapsed_time": "4:28:20", "remaining_time": "6:11:56"} +{"current_steps": 2249, "total_steps": 5364, "loss": 0.1285722702741623, "lr": 1.4414494015424883e-05, "epoch": 1.2579020979020978, "percentage": 41.93, "elapsed_time": "4:28:26", "remaining_time": "6:11:49"} +{"current_steps": 2250, "total_steps": 5364, "loss": 0.06627713143825531, "lr": 1.4408653206080756e-05, "epoch": 1.2584615384615385, "percentage": 41.95, "elapsed_time": "4:28:33", "remaining_time": "6:11:41"} +{"current_steps": 2251, "total_steps": 5364, "loss": 0.23785154521465302, "lr": 1.4402810529277656e-05, "epoch": 1.259020979020979, "percentage": 41.96, "elapsed_time": "4:28:40", "remaining_time": "6:11:34"} +{"current_steps": 2252, "total_steps": 5364, "loss": 0.16897568106651306, "lr": 1.4396965987490472e-05, "epoch": 1.2595804195804197, "percentage": 41.98, "elapsed_time": "4:28:47", "remaining_time": "6:11:26"} +{"current_steps": 2253, "total_steps": 5364, "loss": 0.2043132334947586, "lr": 1.4391119583194899e-05, "epoch": 1.2601398601398601, "percentage": 42.0, "elapsed_time": "4:28:54", "remaining_time": "6:11:19"} +{"current_steps": 2254, "total_steps": 5364, "loss": 0.0723540410399437, "lr": 1.4385271318867405e-05, "epoch": 1.2606993006993008, "percentage": 42.02, "elapsed_time": "4:29:01", "remaining_time": "6:11:12"} +{"current_steps": 2255, "total_steps": 5364, "loss": 0.06294068694114685, "lr": 1.4379421196985258e-05, "epoch": 1.2612587412587413, "percentage": 42.04, "elapsed_time": "4:29:09", "remaining_time": "6:11:05"} +{"current_steps": 2256, "total_steps": 5364, "loss": 0.17759248614311218, "lr": 1.4373569220026504e-05, "epoch": 1.2618181818181817, "percentage": 42.06, "elapsed_time": "4:29:16", "remaining_time": "6:10:57"} +{"current_steps": 2257, "total_steps": 5364, "loss": 0.2183111011981964, "lr": 1.436771539046998e-05, "epoch": 1.2623776223776224, "percentage": 42.08, "elapsed_time": "4:29:23", "remaining_time": "6:10:50"} +{"current_steps": 2258, "total_steps": 5364, "loss": 0.06851965188980103, "lr": 1.4361859710795307e-05, "epoch": 1.2629370629370629, "percentage": 42.1, "elapsed_time": "4:29:30", "remaining_time": "6:10:43"} +{"current_steps": 2259, "total_steps": 5364, "loss": 0.1771463304758072, "lr": 1.435600218348289e-05, "epoch": 1.2634965034965036, "percentage": 42.11, "elapsed_time": "4:29:37", "remaining_time": "6:10:36"} +{"current_steps": 2260, "total_steps": 5364, "loss": 0.06010114401578903, "lr": 1.4350142811013915e-05, "epoch": 1.264055944055944, "percentage": 42.13, "elapsed_time": "4:29:45", "remaining_time": "6:10:30"} +{"current_steps": 2261, "total_steps": 5364, "loss": 0.13397587835788727, "lr": 1.4344281595870349e-05, "epoch": 1.2646153846153847, "percentage": 42.15, "elapsed_time": "4:29:52", "remaining_time": "6:10:22"} +{"current_steps": 2262, "total_steps": 5364, "loss": 0.011446855030953884, "lr": 1.433841854053494e-05, "epoch": 1.2651748251748252, "percentage": 42.17, "elapsed_time": "4:30:00", "remaining_time": "6:10:16"} +{"current_steps": 2263, "total_steps": 5364, "loss": 0.042227908968925476, "lr": 1.4332553647491218e-05, "epoch": 1.2657342657342658, "percentage": 42.19, "elapsed_time": "4:30:07", "remaining_time": "6:10:09"} +{"current_steps": 2264, "total_steps": 5364, "loss": 0.06095722317695618, "lr": 1.4326686919223493e-05, "epoch": 1.2662937062937063, "percentage": 42.21, "elapsed_time": "4:30:14", "remaining_time": "6:10:01"} +{"current_steps": 2265, "total_steps": 5364, "loss": 0.28107377886772156, "lr": 1.4320818358216848e-05, "epoch": 1.2668531468531468, "percentage": 42.23, "elapsed_time": "4:30:21", "remaining_time": "6:09:54"} +{"current_steps": 2266, "total_steps": 5364, "loss": 0.1023840680718422, "lr": 1.4314947966957138e-05, "epoch": 1.2674125874125874, "percentage": 42.24, "elapsed_time": "4:30:28", "remaining_time": "6:09:46"} +{"current_steps": 2267, "total_steps": 5364, "loss": 0.10148217529058456, "lr": 1.4309075747931002e-05, "epoch": 1.267972027972028, "percentage": 42.26, "elapsed_time": "4:30:34", "remaining_time": "6:09:38"} +{"current_steps": 2268, "total_steps": 5364, "loss": 0.11574014276266098, "lr": 1.4303201703625853e-05, "epoch": 1.2685314685314686, "percentage": 42.28, "elapsed_time": "4:30:43", "remaining_time": "6:09:33"} +{"current_steps": 2269, "total_steps": 5364, "loss": 0.1353980004787445, "lr": 1.4297325836529872e-05, "epoch": 1.269090909090909, "percentage": 42.3, "elapsed_time": "4:30:50", "remaining_time": "6:09:25"} +{"current_steps": 2270, "total_steps": 5364, "loss": 0.0474870502948761, "lr": 1.4291448149132015e-05, "epoch": 1.2696503496503497, "percentage": 42.32, "elapsed_time": "4:30:57", "remaining_time": "6:09:19"} +{"current_steps": 2271, "total_steps": 5364, "loss": 0.07048232853412628, "lr": 1.4285568643922004e-05, "epoch": 1.2702097902097902, "percentage": 42.34, "elapsed_time": "4:31:04", "remaining_time": "6:09:12"} +{"current_steps": 2272, "total_steps": 5364, "loss": 0.12050151824951172, "lr": 1.427968732339034e-05, "epoch": 1.2707692307692309, "percentage": 42.36, "elapsed_time": "4:31:11", "remaining_time": "6:09:04"} +{"current_steps": 2273, "total_steps": 5364, "loss": 0.10930024087429047, "lr": 1.4273804190028284e-05, "epoch": 1.2713286713286713, "percentage": 42.38, "elapsed_time": "4:31:18", "remaining_time": "6:08:57"} +{"current_steps": 2274, "total_steps": 5364, "loss": 0.21798774600028992, "lr": 1.4267919246327877e-05, "epoch": 1.2718881118881118, "percentage": 42.39, "elapsed_time": "4:31:25", "remaining_time": "6:08:49"} +{"current_steps": 2275, "total_steps": 5364, "loss": 0.21044299006462097, "lr": 1.4262032494781915e-05, "epoch": 1.2724475524475525, "percentage": 42.41, "elapsed_time": "4:31:33", "remaining_time": "6:08:42"} +{"current_steps": 2276, "total_steps": 5364, "loss": 0.12780770659446716, "lr": 1.425614393788396e-05, "epoch": 1.273006993006993, "percentage": 42.43, "elapsed_time": "4:31:39", "remaining_time": "6:08:35"} +{"current_steps": 2277, "total_steps": 5364, "loss": 0.27876508235931396, "lr": 1.4250253578128343e-05, "epoch": 1.2735664335664336, "percentage": 42.45, "elapsed_time": "4:31:46", "remaining_time": "6:08:27"} +{"current_steps": 2278, "total_steps": 5364, "loss": 0.05382530391216278, "lr": 1.4244361418010169e-05, "epoch": 1.274125874125874, "percentage": 42.47, "elapsed_time": "4:31:54", "remaining_time": "6:08:21"} +{"current_steps": 2279, "total_steps": 5364, "loss": 0.2046329528093338, "lr": 1.423846746002528e-05, "epoch": 1.2746853146853148, "percentage": 42.49, "elapsed_time": "4:32:01", "remaining_time": "6:08:13"} +{"current_steps": 2280, "total_steps": 5364, "loss": 0.3457147777080536, "lr": 1.4232571706670303e-05, "epoch": 1.2752447552447552, "percentage": 42.51, "elapsed_time": "4:32:08", "remaining_time": "6:08:06"} +{"current_steps": 2281, "total_steps": 5364, "loss": 0.1682617962360382, "lr": 1.4226674160442617e-05, "epoch": 1.275804195804196, "percentage": 42.52, "elapsed_time": "4:32:15", "remaining_time": "6:07:58"} +{"current_steps": 2282, "total_steps": 5364, "loss": 0.2219056487083435, "lr": 1.4220774823840355e-05, "epoch": 1.2763636363636364, "percentage": 42.54, "elapsed_time": "4:32:22", "remaining_time": "6:07:51"} +{"current_steps": 2283, "total_steps": 5364, "loss": 0.030912689864635468, "lr": 1.4214873699362422e-05, "epoch": 1.2769230769230768, "percentage": 42.56, "elapsed_time": "4:32:29", "remaining_time": "6:07:43"} +{"current_steps": 2284, "total_steps": 5364, "loss": 0.09763117879629135, "lr": 1.4208970789508463e-05, "epoch": 1.2774825174825175, "percentage": 42.58, "elapsed_time": "4:32:36", "remaining_time": "6:07:36"} +{"current_steps": 2285, "total_steps": 5364, "loss": 0.13511262834072113, "lr": 1.4203066096778893e-05, "epoch": 1.278041958041958, "percentage": 42.6, "elapsed_time": "4:32:42", "remaining_time": "6:07:28"} +{"current_steps": 2286, "total_steps": 5364, "loss": 0.12463656812906265, "lr": 1.419715962367488e-05, "epoch": 1.2786013986013987, "percentage": 42.62, "elapsed_time": "4:32:50", "remaining_time": "6:07:22"} +{"current_steps": 2287, "total_steps": 5364, "loss": 0.12693163752555847, "lr": 1.419125137269834e-05, "epoch": 1.2791608391608391, "percentage": 42.64, "elapsed_time": "4:32:57", "remaining_time": "6:07:14"} +{"current_steps": 2288, "total_steps": 5364, "loss": 0.10232726484537125, "lr": 1.4185341346351947e-05, "epoch": 1.2797202797202798, "percentage": 42.65, "elapsed_time": "4:33:04", "remaining_time": "6:07:06"} +{"current_steps": 2289, "total_steps": 5364, "loss": 0.09676478058099747, "lr": 1.417942954713912e-05, "epoch": 1.2802797202797203, "percentage": 42.67, "elapsed_time": "4:33:11", "remaining_time": "6:06:59"} +{"current_steps": 2290, "total_steps": 5364, "loss": 0.14718085527420044, "lr": 1.4173515977564045e-05, "epoch": 1.280839160839161, "percentage": 42.69, "elapsed_time": "4:33:18", "remaining_time": "6:06:52"} +{"current_steps": 2291, "total_steps": 5364, "loss": 0.28424859046936035, "lr": 1.4167600640131643e-05, "epoch": 1.2813986013986014, "percentage": 42.71, "elapsed_time": "4:33:25", "remaining_time": "6:06:45"} +{"current_steps": 2292, "total_steps": 5364, "loss": 0.12832795083522797, "lr": 1.4161683537347589e-05, "epoch": 1.2819580419580419, "percentage": 42.73, "elapsed_time": "4:33:32", "remaining_time": "6:06:38"} +{"current_steps": 2293, "total_steps": 5364, "loss": 0.0876542329788208, "lr": 1.4155764671718302e-05, "epoch": 1.2825174825174825, "percentage": 42.75, "elapsed_time": "4:33:40", "remaining_time": "6:06:31"} +{"current_steps": 2294, "total_steps": 5364, "loss": 0.20542249083518982, "lr": 1.4149844045750953e-05, "epoch": 1.283076923076923, "percentage": 42.77, "elapsed_time": "4:33:47", "remaining_time": "6:06:24"} +{"current_steps": 2295, "total_steps": 5364, "loss": 0.13296125829219818, "lr": 1.414392166195346e-05, "epoch": 1.2836363636363637, "percentage": 42.79, "elapsed_time": "4:33:55", "remaining_time": "6:06:17"} +{"current_steps": 2296, "total_steps": 5364, "loss": 0.29220935702323914, "lr": 1.4137997522834477e-05, "epoch": 1.2841958041958041, "percentage": 42.8, "elapsed_time": "4:34:02", "remaining_time": "6:06:10"} +{"current_steps": 2297, "total_steps": 5364, "loss": 0.07354001700878143, "lr": 1.413207163090341e-05, "epoch": 1.2847552447552448, "percentage": 42.82, "elapsed_time": "4:34:09", "remaining_time": "6:06:03"} +{"current_steps": 2298, "total_steps": 5364, "loss": 0.049216367304325104, "lr": 1.4126143988670406e-05, "epoch": 1.2853146853146853, "percentage": 42.84, "elapsed_time": "4:34:16", "remaining_time": "6:05:56"} +{"current_steps": 2299, "total_steps": 5364, "loss": 0.15531830489635468, "lr": 1.4120214598646347e-05, "epoch": 1.285874125874126, "percentage": 42.86, "elapsed_time": "4:34:23", "remaining_time": "6:05:49"} +{"current_steps": 2300, "total_steps": 5364, "loss": 0.2570089101791382, "lr": 1.4114283463342867e-05, "epoch": 1.2864335664335664, "percentage": 42.88, "elapsed_time": "4:34:30", "remaining_time": "6:05:41"} +{"current_steps": 2301, "total_steps": 5364, "loss": 0.017475413158535957, "lr": 1.4108350585272321e-05, "epoch": 1.286993006993007, "percentage": 42.9, "elapsed_time": "4:34:37", "remaining_time": "6:05:34"} +{"current_steps": 2302, "total_steps": 5364, "loss": 0.2496194839477539, "lr": 1.4102415966947822e-05, "epoch": 1.2875524475524476, "percentage": 42.92, "elapsed_time": "4:34:44", "remaining_time": "6:05:26"} +{"current_steps": 2303, "total_steps": 5364, "loss": 0.10613247752189636, "lr": 1.4096479610883213e-05, "epoch": 1.288111888111888, "percentage": 42.93, "elapsed_time": "4:34:51", "remaining_time": "6:05:19"} +{"current_steps": 2304, "total_steps": 5364, "loss": 0.24276016652584076, "lr": 1.4090541519593067e-05, "epoch": 1.2886713286713287, "percentage": 42.95, "elapsed_time": "4:34:58", "remaining_time": "6:05:12"} +{"current_steps": 2305, "total_steps": 5364, "loss": 0.3647720217704773, "lr": 1.4084601695592699e-05, "epoch": 1.2892307692307692, "percentage": 42.97, "elapsed_time": "4:35:05", "remaining_time": "6:05:04"} +{"current_steps": 2306, "total_steps": 5364, "loss": 0.06622689217329025, "lr": 1.4078660141398154e-05, "epoch": 1.2897902097902099, "percentage": 42.99, "elapsed_time": "4:35:12", "remaining_time": "6:04:57"} +{"current_steps": 2307, "total_steps": 5364, "loss": 0.19807204604148865, "lr": 1.4072716859526214e-05, "epoch": 1.2903496503496503, "percentage": 43.01, "elapsed_time": "4:35:19", "remaining_time": "6:04:50"} +{"current_steps": 2308, "total_steps": 5364, "loss": 0.03155851364135742, "lr": 1.406677185249439e-05, "epoch": 1.290909090909091, "percentage": 43.03, "elapsed_time": "4:35:26", "remaining_time": "6:04:43"} +{"current_steps": 2309, "total_steps": 5364, "loss": 0.2588830888271332, "lr": 1.4060825122820925e-05, "epoch": 1.2914685314685315, "percentage": 43.05, "elapsed_time": "4:35:34", "remaining_time": "6:04:36"} +{"current_steps": 2310, "total_steps": 5364, "loss": 0.05991354212164879, "lr": 1.405487667302479e-05, "epoch": 1.292027972027972, "percentage": 43.06, "elapsed_time": "4:35:41", "remaining_time": "6:04:29"} +{"current_steps": 2311, "total_steps": 5364, "loss": 0.10276368260383606, "lr": 1.4048926505625682e-05, "epoch": 1.2925874125874126, "percentage": 43.08, "elapsed_time": "4:35:48", "remaining_time": "6:04:22"} +{"current_steps": 2312, "total_steps": 5364, "loss": 0.09990692883729935, "lr": 1.404297462314403e-05, "epoch": 1.293146853146853, "percentage": 43.1, "elapsed_time": "4:35:55", "remaining_time": "6:04:14"} +{"current_steps": 2313, "total_steps": 5364, "loss": 0.15268345177173615, "lr": 1.4037021028100996e-05, "epoch": 1.2937062937062938, "percentage": 43.12, "elapsed_time": "4:36:02", "remaining_time": "6:04:07"} +{"current_steps": 2314, "total_steps": 5364, "loss": 0.1448216736316681, "lr": 1.4031065723018455e-05, "epoch": 1.2942657342657342, "percentage": 43.14, "elapsed_time": "4:36:10", "remaining_time": "6:04:00"} +{"current_steps": 2315, "total_steps": 5364, "loss": 0.022810693830251694, "lr": 1.402510871041901e-05, "epoch": 1.294825174825175, "percentage": 43.16, "elapsed_time": "4:36:17", "remaining_time": "6:03:53"} +{"current_steps": 2316, "total_steps": 5364, "loss": 0.20819419622421265, "lr": 1.4019149992825991e-05, "epoch": 1.2953846153846154, "percentage": 43.18, "elapsed_time": "4:36:24", "remaining_time": "6:03:45"} +{"current_steps": 2317, "total_steps": 5364, "loss": 0.1419975608587265, "lr": 1.4013189572763443e-05, "epoch": 1.295944055944056, "percentage": 43.2, "elapsed_time": "4:36:31", "remaining_time": "6:03:38"} +{"current_steps": 2318, "total_steps": 5364, "loss": 0.09433316439390182, "lr": 1.4007227452756147e-05, "epoch": 1.2965034965034965, "percentage": 43.21, "elapsed_time": "4:36:38", "remaining_time": "6:03:30"} +{"current_steps": 2319, "total_steps": 5364, "loss": 0.1789778769016266, "lr": 1.4001263635329584e-05, "epoch": 1.297062937062937, "percentage": 43.23, "elapsed_time": "4:36:45", "remaining_time": "6:03:24"} +{"current_steps": 2320, "total_steps": 5364, "loss": 0.1831597089767456, "lr": 1.3995298123009971e-05, "epoch": 1.2976223776223776, "percentage": 43.25, "elapsed_time": "4:36:52", "remaining_time": "6:03:16"} +{"current_steps": 2321, "total_steps": 5364, "loss": 0.17378035187721252, "lr": 1.3989330918324231e-05, "epoch": 1.298181818181818, "percentage": 43.27, "elapsed_time": "4:36:59", "remaining_time": "6:03:08"} +{"current_steps": 2322, "total_steps": 5364, "loss": 0.08113575726747513, "lr": 1.3983362023800011e-05, "epoch": 1.2987412587412588, "percentage": 43.29, "elapsed_time": "4:37:06", "remaining_time": "6:03:02"} +{"current_steps": 2323, "total_steps": 5364, "loss": 0.14387616515159607, "lr": 1.3977391441965676e-05, "epoch": 1.2993006993006992, "percentage": 43.31, "elapsed_time": "4:37:13", "remaining_time": "6:02:54"} +{"current_steps": 2324, "total_steps": 5364, "loss": 0.08410822600126266, "lr": 1.3971419175350294e-05, "epoch": 1.29986013986014, "percentage": 43.33, "elapsed_time": "4:37:20", "remaining_time": "6:02:47"} +{"current_steps": 2325, "total_steps": 5364, "loss": 0.15957938134670258, "lr": 1.3965445226483655e-05, "epoch": 1.3004195804195804, "percentage": 43.34, "elapsed_time": "4:37:27", "remaining_time": "6:02:40"} +{"current_steps": 2326, "total_steps": 5364, "loss": 0.07351817190647125, "lr": 1.3959469597896269e-05, "epoch": 1.300979020979021, "percentage": 43.36, "elapsed_time": "4:37:35", "remaining_time": "6:02:33"} +{"current_steps": 2327, "total_steps": 5364, "loss": 0.10773013532161713, "lr": 1.3953492292119343e-05, "epoch": 1.3015384615384615, "percentage": 43.38, "elapsed_time": "4:37:42", "remaining_time": "6:02:26"} +{"current_steps": 2328, "total_steps": 5364, "loss": 0.13009023666381836, "lr": 1.3947513311684798e-05, "epoch": 1.302097902097902, "percentage": 43.4, "elapsed_time": "4:37:49", "remaining_time": "6:02:19"} +{"current_steps": 2329, "total_steps": 5364, "loss": 0.07286220788955688, "lr": 1.394153265912527e-05, "epoch": 1.3026573426573427, "percentage": 43.42, "elapsed_time": "4:37:56", "remaining_time": "6:02:12"} +{"current_steps": 2330, "total_steps": 5364, "loss": 0.2962875962257385, "lr": 1.3935550336974102e-05, "epoch": 1.3032167832167831, "percentage": 43.44, "elapsed_time": "4:38:03", "remaining_time": "6:02:04"} +{"current_steps": 2331, "total_steps": 5364, "loss": 0.1040998324751854, "lr": 1.3929566347765341e-05, "epoch": 1.3037762237762238, "percentage": 43.46, "elapsed_time": "4:38:10", "remaining_time": "6:01:57"} +{"current_steps": 2332, "total_steps": 5364, "loss": 0.1252475082874298, "lr": 1.3923580694033745e-05, "epoch": 1.3043356643356643, "percentage": 43.48, "elapsed_time": "4:38:18", "remaining_time": "6:01:50"} +{"current_steps": 2333, "total_steps": 5364, "loss": 0.15631037950515747, "lr": 1.3917593378314767e-05, "epoch": 1.304895104895105, "percentage": 43.49, "elapsed_time": "4:38:25", "remaining_time": "6:01:43"} +{"current_steps": 2334, "total_steps": 5364, "loss": 0.1472390592098236, "lr": 1.3911604403144572e-05, "epoch": 1.3054545454545454, "percentage": 43.51, "elapsed_time": "4:38:33", "remaining_time": "6:01:37"} +{"current_steps": 2335, "total_steps": 5364, "loss": 0.08398403972387314, "lr": 1.390561377106003e-05, "epoch": 1.306013986013986, "percentage": 43.53, "elapsed_time": "4:38:40", "remaining_time": "6:01:30"} +{"current_steps": 2336, "total_steps": 5364, "loss": 0.16348221898078918, "lr": 1.389962148459871e-05, "epoch": 1.3065734265734266, "percentage": 43.55, "elapsed_time": "4:38:48", "remaining_time": "6:01:23"} +{"current_steps": 2337, "total_steps": 5364, "loss": 0.08259494602680206, "lr": 1.3893627546298882e-05, "epoch": 1.307132867132867, "percentage": 43.57, "elapsed_time": "4:38:55", "remaining_time": "6:01:16"} +{"current_steps": 2338, "total_steps": 5364, "loss": 0.21214108169078827, "lr": 1.388763195869951e-05, "epoch": 1.3076923076923077, "percentage": 43.59, "elapsed_time": "4:39:02", "remaining_time": "6:01:09"} +{"current_steps": 2339, "total_steps": 5364, "loss": 0.18700185418128967, "lr": 1.3881634724340263e-05, "epoch": 1.3082517482517482, "percentage": 43.61, "elapsed_time": "4:39:09", "remaining_time": "6:01:01"} +{"current_steps": 2340, "total_steps": 5364, "loss": 0.2143566608428955, "lr": 1.387563584576151e-05, "epoch": 1.3088111888111889, "percentage": 43.62, "elapsed_time": "4:39:16", "remaining_time": "6:00:54"} +{"current_steps": 2341, "total_steps": 5364, "loss": 0.0645679384469986, "lr": 1.3869635325504307e-05, "epoch": 1.3093706293706293, "percentage": 43.64, "elapsed_time": "4:39:23", "remaining_time": "6:00:47"} +{"current_steps": 2342, "total_steps": 5364, "loss": 0.12910711765289307, "lr": 1.386363316611042e-05, "epoch": 1.30993006993007, "percentage": 43.66, "elapsed_time": "4:39:30", "remaining_time": "6:00:40"} +{"current_steps": 2343, "total_steps": 5364, "loss": 0.03921044245362282, "lr": 1.385762937012229e-05, "epoch": 1.3104895104895105, "percentage": 43.68, "elapsed_time": "4:39:38", "remaining_time": "6:00:33"} +{"current_steps": 2344, "total_steps": 5364, "loss": 0.14040474593639374, "lr": 1.3851623940083068e-05, "epoch": 1.3110489510489511, "percentage": 43.7, "elapsed_time": "4:39:45", "remaining_time": "6:00:25"} +{"current_steps": 2345, "total_steps": 5364, "loss": 0.09478513151407242, "lr": 1.3845616878536588e-05, "epoch": 1.3116083916083916, "percentage": 43.72, "elapsed_time": "4:39:52", "remaining_time": "6:00:18"} +{"current_steps": 2346, "total_steps": 5364, "loss": 0.15669310092926025, "lr": 1.383960818802738e-05, "epoch": 1.312167832167832, "percentage": 43.74, "elapsed_time": "4:39:59", "remaining_time": "6:00:11"} +{"current_steps": 2347, "total_steps": 5364, "loss": 0.07649362087249756, "lr": 1.383359787110066e-05, "epoch": 1.3127272727272727, "percentage": 43.75, "elapsed_time": "4:40:05", "remaining_time": "6:00:03"} +{"current_steps": 2348, "total_steps": 5364, "loss": 0.08076094090938568, "lr": 1.3827585930302336e-05, "epoch": 1.3132867132867134, "percentage": 43.77, "elapsed_time": "4:40:13", "remaining_time": "5:59:56"} +{"current_steps": 2349, "total_steps": 5364, "loss": 0.0926351398229599, "lr": 1.3821572368179007e-05, "epoch": 1.3138461538461539, "percentage": 43.79, "elapsed_time": "4:40:20", "remaining_time": "5:59:48"} +{"current_steps": 2350, "total_steps": 5364, "loss": 0.28519099950790405, "lr": 1.3815557187277944e-05, "epoch": 1.3144055944055943, "percentage": 43.81, "elapsed_time": "4:40:26", "remaining_time": "5:59:41"} +{"current_steps": 2351, "total_steps": 5364, "loss": 0.14812952280044556, "lr": 1.3809540390147121e-05, "epoch": 1.314965034965035, "percentage": 43.83, "elapsed_time": "4:40:33", "remaining_time": "5:59:34"} +{"current_steps": 2352, "total_steps": 5364, "loss": 0.1255715787410736, "lr": 1.3803521979335189e-05, "epoch": 1.3155244755244755, "percentage": 43.85, "elapsed_time": "4:40:41", "remaining_time": "5:59:27"} +{"current_steps": 2353, "total_steps": 5364, "loss": 0.0979311466217041, "lr": 1.3797501957391483e-05, "epoch": 1.3160839160839162, "percentage": 43.87, "elapsed_time": "4:40:48", "remaining_time": "5:59:19"} +{"current_steps": 2354, "total_steps": 5364, "loss": 0.02696126326918602, "lr": 1.3791480326866022e-05, "epoch": 1.3166433566433566, "percentage": 43.89, "elapsed_time": "4:40:55", "remaining_time": "5:59:12"} +{"current_steps": 2355, "total_steps": 5364, "loss": 0.27828383445739746, "lr": 1.3785457090309504e-05, "epoch": 1.317202797202797, "percentage": 43.9, "elapsed_time": "4:41:02", "remaining_time": "5:59:05"} +{"current_steps": 2356, "total_steps": 5364, "loss": 0.07887479662895203, "lr": 1.3779432250273306e-05, "epoch": 1.3177622377622378, "percentage": 43.92, "elapsed_time": "4:41:09", "remaining_time": "5:58:58"} +{"current_steps": 2357, "total_steps": 5364, "loss": 0.20843224227428436, "lr": 1.3773405809309491e-05, "epoch": 1.3183216783216785, "percentage": 43.94, "elapsed_time": "4:41:16", "remaining_time": "5:58:50"} +{"current_steps": 2358, "total_steps": 5364, "loss": 0.22880637645721436, "lr": 1.3767377769970796e-05, "epoch": 1.318881118881119, "percentage": 43.96, "elapsed_time": "4:41:23", "remaining_time": "5:58:43"} +{"current_steps": 2359, "total_steps": 5364, "loss": 0.22276586294174194, "lr": 1.3761348134810631e-05, "epoch": 1.3194405594405594, "percentage": 43.98, "elapsed_time": "4:41:30", "remaining_time": "5:58:36"} +{"current_steps": 2360, "total_steps": 5364, "loss": 0.08788898587226868, "lr": 1.3755316906383085e-05, "epoch": 1.32, "percentage": 44.0, "elapsed_time": "4:41:37", "remaining_time": "5:58:28"} +{"current_steps": 2361, "total_steps": 5364, "loss": 0.05440403148531914, "lr": 1.3749284087242927e-05, "epoch": 1.3205594405594405, "percentage": 44.02, "elapsed_time": "4:41:45", "remaining_time": "5:58:22"} +{"current_steps": 2362, "total_steps": 5364, "loss": 0.0408187098801136, "lr": 1.3743249679945597e-05, "epoch": 1.3211188811188812, "percentage": 44.03, "elapsed_time": "4:41:52", "remaining_time": "5:58:14"} +{"current_steps": 2363, "total_steps": 5364, "loss": 0.15900954604148865, "lr": 1.3737213687047199e-05, "epoch": 1.3216783216783217, "percentage": 44.05, "elapsed_time": "4:41:59", "remaining_time": "5:58:07"} +{"current_steps": 2364, "total_steps": 5364, "loss": 0.2538132667541504, "lr": 1.373117611110452e-05, "epoch": 1.3222377622377621, "percentage": 44.07, "elapsed_time": "4:42:05", "remaining_time": "5:57:59"} +{"current_steps": 2365, "total_steps": 5364, "loss": 0.11166108399629593, "lr": 1.3725136954675016e-05, "epoch": 1.3227972027972028, "percentage": 44.09, "elapsed_time": "4:42:12", "remaining_time": "5:57:52"} +{"current_steps": 2366, "total_steps": 5364, "loss": 0.20830011367797852, "lr": 1.3719096220316805e-05, "epoch": 1.3233566433566435, "percentage": 44.11, "elapsed_time": "4:42:20", "remaining_time": "5:57:45"} +{"current_steps": 2367, "total_steps": 5364, "loss": 0.2259216010570526, "lr": 1.3713053910588683e-05, "epoch": 1.323916083916084, "percentage": 44.13, "elapsed_time": "4:42:27", "remaining_time": "5:57:37"} +{"current_steps": 2368, "total_steps": 5364, "loss": 0.23482930660247803, "lr": 1.3707010028050104e-05, "epoch": 1.3244755244755244, "percentage": 44.15, "elapsed_time": "4:42:34", "remaining_time": "5:57:30"} +{"current_steps": 2369, "total_steps": 5364, "loss": 0.30894967913627625, "lr": 1.3700964575261198e-05, "epoch": 1.325034965034965, "percentage": 44.16, "elapsed_time": "4:42:41", "remaining_time": "5:57:23"} +{"current_steps": 2370, "total_steps": 5364, "loss": 0.019383564591407776, "lr": 1.3694917554782752e-05, "epoch": 1.3255944055944056, "percentage": 44.18, "elapsed_time": "4:42:48", "remaining_time": "5:57:15"} +{"current_steps": 2371, "total_steps": 5364, "loss": 0.1656731367111206, "lr": 1.3688868969176217e-05, "epoch": 1.3261538461538462, "percentage": 44.2, "elapsed_time": "4:42:55", "remaining_time": "5:57:08"} +{"current_steps": 2372, "total_steps": 5364, "loss": 0.11655168235301971, "lr": 1.3682818821003716e-05, "epoch": 1.3267132867132867, "percentage": 44.22, "elapsed_time": "4:43:02", "remaining_time": "5:57:01"} +{"current_steps": 2373, "total_steps": 5364, "loss": 0.15491896867752075, "lr": 1.3676767112828025e-05, "epoch": 1.3272727272727272, "percentage": 44.24, "elapsed_time": "4:43:09", "remaining_time": "5:56:54"} +{"current_steps": 2374, "total_steps": 5364, "loss": 0.09042191505432129, "lr": 1.3670713847212583e-05, "epoch": 1.3278321678321678, "percentage": 44.26, "elapsed_time": "4:43:16", "remaining_time": "5:56:46"} +{"current_steps": 2375, "total_steps": 5364, "loss": 0.12956613302230835, "lr": 1.3664659026721492e-05, "epoch": 1.3283916083916085, "percentage": 44.28, "elapsed_time": "4:43:24", "remaining_time": "5:56:40"} +{"current_steps": 2376, "total_steps": 5364, "loss": 0.16507470607757568, "lr": 1.3658602653919509e-05, "epoch": 1.328951048951049, "percentage": 44.3, "elapsed_time": "4:43:31", "remaining_time": "5:56:33"} +{"current_steps": 2377, "total_steps": 5364, "loss": 0.08841853588819504, "lr": 1.3652544731372046e-05, "epoch": 1.3295104895104894, "percentage": 44.31, "elapsed_time": "4:43:39", "remaining_time": "5:56:26"} +{"current_steps": 2378, "total_steps": 5364, "loss": 0.1448720246553421, "lr": 1.3646485261645177e-05, "epoch": 1.3300699300699301, "percentage": 44.33, "elapsed_time": "4:43:46", "remaining_time": "5:56:19"} +{"current_steps": 2379, "total_steps": 5364, "loss": 0.08173331618309021, "lr": 1.3640424247305633e-05, "epoch": 1.3306293706293706, "percentage": 44.35, "elapsed_time": "4:43:53", "remaining_time": "5:56:12"} +{"current_steps": 2380, "total_steps": 5364, "loss": 0.2008461356163025, "lr": 1.3634361690920797e-05, "epoch": 1.3311888111888113, "percentage": 44.37, "elapsed_time": "4:44:00", "remaining_time": "5:56:05"} +{"current_steps": 2381, "total_steps": 5364, "loss": 0.31576189398765564, "lr": 1.3628297595058695e-05, "epoch": 1.3317482517482517, "percentage": 44.39, "elapsed_time": "4:44:07", "remaining_time": "5:55:58"} +{"current_steps": 2382, "total_steps": 5364, "loss": 0.07697045803070068, "lr": 1.3622231962288022e-05, "epoch": 1.3323076923076922, "percentage": 44.41, "elapsed_time": "4:44:14", "remaining_time": "5:55:50"} +{"current_steps": 2383, "total_steps": 5364, "loss": 0.09908680617809296, "lr": 1.361616479517811e-05, "epoch": 1.3328671328671329, "percentage": 44.43, "elapsed_time": "4:44:21", "remaining_time": "5:55:43"} +{"current_steps": 2384, "total_steps": 5364, "loss": 0.232676699757576, "lr": 1.3610096096298953e-05, "epoch": 1.3334265734265736, "percentage": 44.44, "elapsed_time": "4:44:29", "remaining_time": "5:55:36"} +{"current_steps": 2385, "total_steps": 5364, "loss": 0.12593509256839752, "lr": 1.3604025868221184e-05, "epoch": 1.333986013986014, "percentage": 44.46, "elapsed_time": "4:44:36", "remaining_time": "5:55:29"} +{"current_steps": 2386, "total_steps": 5364, "loss": 0.20413318276405334, "lr": 1.359795411351609e-05, "epoch": 1.3345454545454545, "percentage": 44.48, "elapsed_time": "4:44:43", "remaining_time": "5:55:22"} +{"current_steps": 2387, "total_steps": 5364, "loss": 0.17946910858154297, "lr": 1.3591880834755599e-05, "epoch": 1.3351048951048952, "percentage": 44.5, "elapsed_time": "4:44:50", "remaining_time": "5:55:14"} +{"current_steps": 2388, "total_steps": 5364, "loss": 0.27608245611190796, "lr": 1.358580603451229e-05, "epoch": 1.3356643356643356, "percentage": 44.52, "elapsed_time": "4:44:58", "remaining_time": "5:55:08"} +{"current_steps": 2389, "total_steps": 5364, "loss": 0.2822544276714325, "lr": 1.3579729715359386e-05, "epoch": 1.3362237762237763, "percentage": 44.54, "elapsed_time": "4:45:05", "remaining_time": "5:55:01"} +{"current_steps": 2390, "total_steps": 5364, "loss": 0.2236463725566864, "lr": 1.3573651879870745e-05, "epoch": 1.3367832167832168, "percentage": 44.56, "elapsed_time": "4:45:12", "remaining_time": "5:54:53"} +{"current_steps": 2391, "total_steps": 5364, "loss": 0.1276034265756607, "lr": 1.3567572530620883e-05, "epoch": 1.3373426573426572, "percentage": 44.57, "elapsed_time": "4:45:19", "remaining_time": "5:54:47"} +{"current_steps": 2392, "total_steps": 5364, "loss": 0.22586891055107117, "lr": 1.356149167018494e-05, "epoch": 1.337902097902098, "percentage": 44.59, "elapsed_time": "4:45:27", "remaining_time": "5:54:40"} +{"current_steps": 2393, "total_steps": 5364, "loss": 0.11792099475860596, "lr": 1.3555409301138714e-05, "epoch": 1.3384615384615386, "percentage": 44.61, "elapsed_time": "4:45:34", "remaining_time": "5:54:32"} +{"current_steps": 2394, "total_steps": 5364, "loss": 0.07946892082691193, "lr": 1.3549325426058622e-05, "epoch": 1.339020979020979, "percentage": 44.63, "elapsed_time": "4:45:41", "remaining_time": "5:54:26"} +{"current_steps": 2395, "total_steps": 5364, "loss": 0.08701953291893005, "lr": 1.3543240047521734e-05, "epoch": 1.3395804195804195, "percentage": 44.65, "elapsed_time": "4:45:49", "remaining_time": "5:54:19"} +{"current_steps": 2396, "total_steps": 5364, "loss": 0.06015447527170181, "lr": 1.353715316810575e-05, "epoch": 1.3401398601398602, "percentage": 44.67, "elapsed_time": "4:45:56", "remaining_time": "5:54:11"} +{"current_steps": 2397, "total_steps": 5364, "loss": 0.092784583568573, "lr": 1.353106479038901e-05, "epoch": 1.3406993006993007, "percentage": 44.69, "elapsed_time": "4:46:03", "remaining_time": "5:54:04"} +{"current_steps": 2398, "total_steps": 5364, "loss": 0.12548783421516418, "lr": 1.3524974916950493e-05, "epoch": 1.3412587412587413, "percentage": 44.71, "elapsed_time": "4:46:09", "remaining_time": "5:53:56"} +{"current_steps": 2399, "total_steps": 5364, "loss": 0.23377709090709686, "lr": 1.3518883550369792e-05, "epoch": 1.3418181818181818, "percentage": 44.72, "elapsed_time": "4:46:16", "remaining_time": "5:53:49"} +{"current_steps": 2400, "total_steps": 5364, "loss": 0.1764272153377533, "lr": 1.3512790693227151e-05, "epoch": 1.3423776223776223, "percentage": 44.74, "elapsed_time": "4:46:23", "remaining_time": "5:53:42"} +{"current_steps": 2401, "total_steps": 5364, "loss": 0.176019549369812, "lr": 1.3506696348103443e-05, "epoch": 1.342937062937063, "percentage": 44.76, "elapsed_time": "4:46:30", "remaining_time": "5:53:34"} +{"current_steps": 2402, "total_steps": 5364, "loss": 0.04816877096891403, "lr": 1.3500600517580168e-05, "epoch": 1.3434965034965036, "percentage": 44.78, "elapsed_time": "4:46:37", "remaining_time": "5:53:27"} +{"current_steps": 2403, "total_steps": 5364, "loss": 0.10454526543617249, "lr": 1.3494503204239451e-05, "epoch": 1.344055944055944, "percentage": 44.8, "elapsed_time": "4:46:45", "remaining_time": "5:53:20"} +{"current_steps": 2404, "total_steps": 5364, "loss": 0.20336461067199707, "lr": 1.3488404410664054e-05, "epoch": 1.3446153846153845, "percentage": 44.82, "elapsed_time": "4:46:52", "remaining_time": "5:53:12"} +{"current_steps": 2405, "total_steps": 5364, "loss": 0.06843090057373047, "lr": 1.3482304139437358e-05, "epoch": 1.3451748251748252, "percentage": 44.84, "elapsed_time": "4:46:59", "remaining_time": "5:53:05"} +{"current_steps": 2406, "total_steps": 5364, "loss": 0.15271905064582825, "lr": 1.3476202393143377e-05, "epoch": 1.3457342657342657, "percentage": 44.85, "elapsed_time": "4:47:06", "remaining_time": "5:52:58"} +{"current_steps": 2407, "total_steps": 5364, "loss": 0.09866753220558167, "lr": 1.3470099174366744e-05, "epoch": 1.3462937062937064, "percentage": 44.87, "elapsed_time": "4:47:13", "remaining_time": "5:52:51"} +{"current_steps": 2408, "total_steps": 5364, "loss": 0.21042823791503906, "lr": 1.346399448569272e-05, "epoch": 1.3468531468531468, "percentage": 44.89, "elapsed_time": "4:47:21", "remaining_time": "5:52:44"} +{"current_steps": 2409, "total_steps": 5364, "loss": 0.11054321378469467, "lr": 1.3457888329707184e-05, "epoch": 1.3474125874125873, "percentage": 44.91, "elapsed_time": "4:47:28", "remaining_time": "5:52:37"} +{"current_steps": 2410, "total_steps": 5364, "loss": 0.008536288514733315, "lr": 1.3451780708996642e-05, "epoch": 1.347972027972028, "percentage": 44.93, "elapsed_time": "4:47:35", "remaining_time": "5:52:30"} +{"current_steps": 2411, "total_steps": 5364, "loss": 0.09074399620294571, "lr": 1.3445671626148218e-05, "epoch": 1.3485314685314687, "percentage": 44.95, "elapsed_time": "4:47:42", "remaining_time": "5:52:23"} +{"current_steps": 2412, "total_steps": 5364, "loss": 0.012437272816896439, "lr": 1.343956108374965e-05, "epoch": 1.3490909090909091, "percentage": 44.97, "elapsed_time": "4:47:50", "remaining_time": "5:52:16"} +{"current_steps": 2413, "total_steps": 5364, "loss": 0.013285376131534576, "lr": 1.3433449084389305e-05, "epoch": 1.3496503496503496, "percentage": 44.99, "elapsed_time": "4:47:57", "remaining_time": "5:52:09"} +{"current_steps": 2414, "total_steps": 5364, "loss": 0.07112790644168854, "lr": 1.3427335630656155e-05, "epoch": 1.3502097902097903, "percentage": 45.0, "elapsed_time": "4:48:05", "remaining_time": "5:52:02"} +{"current_steps": 2415, "total_steps": 5364, "loss": 0.06826774775981903, "lr": 1.3421220725139805e-05, "epoch": 1.3507692307692307, "percentage": 45.02, "elapsed_time": "4:48:12", "remaining_time": "5:51:55"} +{"current_steps": 2416, "total_steps": 5364, "loss": 0.1528787612915039, "lr": 1.3415104370430456e-05, "epoch": 1.3513286713286714, "percentage": 45.04, "elapsed_time": "4:48:18", "remaining_time": "5:51:48"} +{"current_steps": 2417, "total_steps": 5364, "loss": 0.07178325951099396, "lr": 1.3408986569118933e-05, "epoch": 1.3518881118881119, "percentage": 45.06, "elapsed_time": "4:48:25", "remaining_time": "5:51:40"} +{"current_steps": 2418, "total_steps": 5364, "loss": 0.09125544130802155, "lr": 1.3402867323796672e-05, "epoch": 1.3524475524475523, "percentage": 45.08, "elapsed_time": "4:48:32", "remaining_time": "5:51:33"} +{"current_steps": 2419, "total_steps": 5364, "loss": 0.10126252472400665, "lr": 1.339674663705572e-05, "epoch": 1.353006993006993, "percentage": 45.1, "elapsed_time": "4:48:39", "remaining_time": "5:51:25"} +{"current_steps": 2420, "total_steps": 5364, "loss": 0.04308270290493965, "lr": 1.3390624511488739e-05, "epoch": 1.3535664335664337, "percentage": 45.12, "elapsed_time": "4:48:46", "remaining_time": "5:51:18"} +{"current_steps": 2421, "total_steps": 5364, "loss": 0.16726504266262054, "lr": 1.3384500949688994e-05, "epoch": 1.3541258741258742, "percentage": 45.13, "elapsed_time": "4:48:54", "remaining_time": "5:51:11"} +{"current_steps": 2422, "total_steps": 5364, "loss": 0.0677989199757576, "lr": 1.3378375954250363e-05, "epoch": 1.3546853146853146, "percentage": 45.15, "elapsed_time": "4:49:01", "remaining_time": "5:51:04"} +{"current_steps": 2423, "total_steps": 5364, "loss": 0.12604407966136932, "lr": 1.3372249527767325e-05, "epoch": 1.3552447552447553, "percentage": 45.17, "elapsed_time": "4:49:08", "remaining_time": "5:50:56"} +{"current_steps": 2424, "total_steps": 5364, "loss": 0.12370515614748001, "lr": 1.3366121672834975e-05, "epoch": 1.3558041958041958, "percentage": 45.19, "elapsed_time": "4:49:14", "remaining_time": "5:50:49"} +{"current_steps": 2425, "total_steps": 5364, "loss": 0.17183727025985718, "lr": 1.3359992392049002e-05, "epoch": 1.3563636363636364, "percentage": 45.21, "elapsed_time": "4:49:21", "remaining_time": "5:50:41"} +{"current_steps": 2426, "total_steps": 5364, "loss": 0.20805403590202332, "lr": 1.335386168800571e-05, "epoch": 1.356923076923077, "percentage": 45.23, "elapsed_time": "4:49:29", "remaining_time": "5:50:35"} +{"current_steps": 2427, "total_steps": 5364, "loss": 0.23937296867370605, "lr": 1.3347729563302e-05, "epoch": 1.3574825174825174, "percentage": 45.25, "elapsed_time": "4:49:36", "remaining_time": "5:50:27"} +{"current_steps": 2428, "total_steps": 5364, "loss": 0.17234614491462708, "lr": 1.334159602053537e-05, "epoch": 1.358041958041958, "percentage": 45.26, "elapsed_time": "4:49:43", "remaining_time": "5:50:20"} +{"current_steps": 2429, "total_steps": 5364, "loss": 0.08037231117486954, "lr": 1.3335461062303932e-05, "epoch": 1.3586013986013987, "percentage": 45.28, "elapsed_time": "4:49:50", "remaining_time": "5:50:13"} +{"current_steps": 2430, "total_steps": 5364, "loss": 0.2533140182495117, "lr": 1.3329324691206382e-05, "epoch": 1.3591608391608392, "percentage": 45.3, "elapsed_time": "4:49:58", "remaining_time": "5:50:07"} +{"current_steps": 2431, "total_steps": 5364, "loss": 0.05418676137924194, "lr": 1.3323186909842027e-05, "epoch": 1.3597202797202796, "percentage": 45.32, "elapsed_time": "4:50:05", "remaining_time": "5:49:59"} +{"current_steps": 2432, "total_steps": 5364, "loss": 0.08776415884494781, "lr": 1.3317047720810766e-05, "epoch": 1.3602797202797203, "percentage": 45.34, "elapsed_time": "4:50:12", "remaining_time": "5:49:52"} +{"current_steps": 2433, "total_steps": 5364, "loss": 0.2018735557794571, "lr": 1.331090712671309e-05, "epoch": 1.3608391608391608, "percentage": 45.36, "elapsed_time": "4:50:19", "remaining_time": "5:49:45"} +{"current_steps": 2434, "total_steps": 5364, "loss": 0.052452098578214645, "lr": 1.3304765130150096e-05, "epoch": 1.3613986013986015, "percentage": 45.38, "elapsed_time": "4:50:26", "remaining_time": "5:49:37"} +{"current_steps": 2435, "total_steps": 5364, "loss": 0.1496383100748062, "lr": 1.3298621733723467e-05, "epoch": 1.361958041958042, "percentage": 45.4, "elapsed_time": "4:50:33", "remaining_time": "5:49:30"} +{"current_steps": 2436, "total_steps": 5364, "loss": 0.17631807923316956, "lr": 1.3292476940035477e-05, "epoch": 1.3625174825174824, "percentage": 45.41, "elapsed_time": "4:50:40", "remaining_time": "5:49:23"} +{"current_steps": 2437, "total_steps": 5364, "loss": 0.3062005341053009, "lr": 1.3286330751689e-05, "epoch": 1.363076923076923, "percentage": 45.43, "elapsed_time": "4:50:47", "remaining_time": "5:49:15"} +{"current_steps": 2438, "total_steps": 5364, "loss": 0.08563196659088135, "lr": 1.32801831712875e-05, "epoch": 1.3636363636363638, "percentage": 45.45, "elapsed_time": "4:50:54", "remaining_time": "5:49:08"} +{"current_steps": 2439, "total_steps": 5364, "loss": 0.27513307332992554, "lr": 1.3274034201435018e-05, "epoch": 1.3641958041958042, "percentage": 45.47, "elapsed_time": "4:51:01", "remaining_time": "5:49:01"} +{"current_steps": 2440, "total_steps": 5364, "loss": 0.22013990581035614, "lr": 1.3267883844736198e-05, "epoch": 1.3647552447552447, "percentage": 45.49, "elapsed_time": "4:51:08", "remaining_time": "5:48:53"} +{"current_steps": 2441, "total_steps": 5364, "loss": 0.15884727239608765, "lr": 1.3261732103796264e-05, "epoch": 1.3653146853146854, "percentage": 45.51, "elapsed_time": "4:51:16", "remaining_time": "5:48:47"} +{"current_steps": 2442, "total_steps": 5364, "loss": 0.24294617772102356, "lr": 1.3255578981221034e-05, "epoch": 1.3658741258741258, "percentage": 45.53, "elapsed_time": "4:51:23", "remaining_time": "5:48:39"} +{"current_steps": 2443, "total_steps": 5364, "loss": 0.20455977320671082, "lr": 1.3249424479616902e-05, "epoch": 1.3664335664335665, "percentage": 45.54, "elapsed_time": "4:51:30", "remaining_time": "5:48:32"} +{"current_steps": 2444, "total_steps": 5364, "loss": 0.0774412453174591, "lr": 1.3243268601590846e-05, "epoch": 1.366993006993007, "percentage": 45.56, "elapsed_time": "4:51:37", "remaining_time": "5:48:25"} +{"current_steps": 2445, "total_steps": 5364, "loss": 0.20527589321136475, "lr": 1.3237111349750435e-05, "epoch": 1.3675524475524474, "percentage": 45.58, "elapsed_time": "4:51:44", "remaining_time": "5:48:18"} +{"current_steps": 2446, "total_steps": 5364, "loss": 0.11029016971588135, "lr": 1.3230952726703819e-05, "epoch": 1.368111888111888, "percentage": 45.6, "elapsed_time": "4:51:51", "remaining_time": "5:48:10"} +{"current_steps": 2447, "total_steps": 5364, "loss": 0.1555417776107788, "lr": 1.3224792735059722e-05, "epoch": 1.3686713286713288, "percentage": 45.62, "elapsed_time": "4:51:58", "remaining_time": "5:48:03"} +{"current_steps": 2448, "total_steps": 5364, "loss": 0.11542768776416779, "lr": 1.3218631377427448e-05, "epoch": 1.3692307692307693, "percentage": 45.64, "elapsed_time": "4:52:05", "remaining_time": "5:47:55"} +{"current_steps": 2449, "total_steps": 5364, "loss": 0.08401789516210556, "lr": 1.3212468656416892e-05, "epoch": 1.3697902097902097, "percentage": 45.66, "elapsed_time": "4:52:12", "remaining_time": "5:47:48"} +{"current_steps": 2450, "total_steps": 5364, "loss": 0.17544862627983093, "lr": 1.3206304574638509e-05, "epoch": 1.3703496503496504, "percentage": 45.67, "elapsed_time": "4:52:19", "remaining_time": "5:47:41"} +{"current_steps": 2451, "total_steps": 5364, "loss": 0.17316821217536926, "lr": 1.3200139134703346e-05, "epoch": 1.3709090909090909, "percentage": 45.69, "elapsed_time": "4:52:26", "remaining_time": "5:47:34"} +{"current_steps": 2452, "total_steps": 5364, "loss": 0.16223135590553284, "lr": 1.3193972339223012e-05, "epoch": 1.3714685314685315, "percentage": 45.71, "elapsed_time": "4:52:34", "remaining_time": "5:47:27"} +{"current_steps": 2453, "total_steps": 5364, "loss": 0.22562295198440552, "lr": 1.31878041908097e-05, "epoch": 1.372027972027972, "percentage": 45.73, "elapsed_time": "4:52:41", "remaining_time": "5:47:19"} +{"current_steps": 2454, "total_steps": 5364, "loss": 0.138825923204422, "lr": 1.3181634692076173e-05, "epoch": 1.3725874125874125, "percentage": 45.75, "elapsed_time": "4:52:48", "remaining_time": "5:47:12"} +{"current_steps": 2455, "total_steps": 5364, "loss": 0.14975762367248535, "lr": 1.3175463845635763e-05, "epoch": 1.3731468531468531, "percentage": 45.77, "elapsed_time": "4:52:56", "remaining_time": "5:47:06"} +{"current_steps": 2456, "total_steps": 5364, "loss": 0.09186311811208725, "lr": 1.316929165410238e-05, "epoch": 1.3737062937062938, "percentage": 45.79, "elapsed_time": "4:53:03", "remaining_time": "5:46:59"} +{"current_steps": 2457, "total_steps": 5364, "loss": 0.1791735738515854, "lr": 1.3163118120090492e-05, "epoch": 1.3742657342657343, "percentage": 45.81, "elapsed_time": "4:53:10", "remaining_time": "5:46:51"} +{"current_steps": 2458, "total_steps": 5364, "loss": 0.01266934722661972, "lr": 1.3156943246215148e-05, "epoch": 1.3748251748251747, "percentage": 45.82, "elapsed_time": "4:53:17", "remaining_time": "5:46:44"} +{"current_steps": 2459, "total_steps": 5364, "loss": 0.03971713408827782, "lr": 1.3150767035091961e-05, "epoch": 1.3753846153846154, "percentage": 45.84, "elapsed_time": "4:53:24", "remaining_time": "5:46:37"} +{"current_steps": 2460, "total_steps": 5364, "loss": 0.26717349886894226, "lr": 1.3144589489337108e-05, "epoch": 1.3759440559440559, "percentage": 45.86, "elapsed_time": "4:53:31", "remaining_time": "5:46:29"} +{"current_steps": 2461, "total_steps": 5364, "loss": 0.028482381254434586, "lr": 1.3138410611567326e-05, "epoch": 1.3765034965034966, "percentage": 45.88, "elapsed_time": "4:53:38", "remaining_time": "5:46:22"} +{"current_steps": 2462, "total_steps": 5364, "loss": 0.20618808269500732, "lr": 1.3132230404399926e-05, "epoch": 1.377062937062937, "percentage": 45.9, "elapsed_time": "4:53:45", "remaining_time": "5:46:15"} +{"current_steps": 2463, "total_steps": 5364, "loss": 0.1066116988658905, "lr": 1.312604887045278e-05, "epoch": 1.3776223776223775, "percentage": 45.92, "elapsed_time": "4:53:52", "remaining_time": "5:46:08"} +{"current_steps": 2464, "total_steps": 5364, "loss": 0.2201676070690155, "lr": 1.3119866012344324e-05, "epoch": 1.3781818181818182, "percentage": 45.94, "elapsed_time": "4:53:59", "remaining_time": "5:46:00"} +{"current_steps": 2465, "total_steps": 5364, "loss": 0.07959502935409546, "lr": 1.3113681832693548e-05, "epoch": 1.3787412587412589, "percentage": 45.95, "elapsed_time": "4:54:06", "remaining_time": "5:45:53"} +{"current_steps": 2466, "total_steps": 5364, "loss": 0.1027555763721466, "lr": 1.310749633412e-05, "epoch": 1.3793006993006993, "percentage": 45.97, "elapsed_time": "4:54:14", "remaining_time": "5:45:46"} +{"current_steps": 2467, "total_steps": 5364, "loss": 0.11271212995052338, "lr": 1.31013095192438e-05, "epoch": 1.3798601398601398, "percentage": 45.99, "elapsed_time": "4:54:21", "remaining_time": "5:45:39"} +{"current_steps": 2468, "total_steps": 5364, "loss": 0.08231376111507416, "lr": 1.309512139068561e-05, "epoch": 1.3804195804195805, "percentage": 46.01, "elapsed_time": "4:54:28", "remaining_time": "5:45:32"} +{"current_steps": 2469, "total_steps": 5364, "loss": 0.12280185520648956, "lr": 1.3088931951066663e-05, "epoch": 1.380979020979021, "percentage": 46.03, "elapsed_time": "4:54:35", "remaining_time": "5:45:25"} +{"current_steps": 2470, "total_steps": 5364, "loss": 0.09480578452348709, "lr": 1.3082741203008737e-05, "epoch": 1.3815384615384616, "percentage": 46.05, "elapsed_time": "4:54:43", "remaining_time": "5:45:18"} +{"current_steps": 2471, "total_steps": 5364, "loss": 0.22818467020988464, "lr": 1.3076549149134167e-05, "epoch": 1.382097902097902, "percentage": 46.07, "elapsed_time": "4:54:50", "remaining_time": "5:45:11"} +{"current_steps": 2472, "total_steps": 5364, "loss": 0.1664549559354782, "lr": 1.3070355792065845e-05, "epoch": 1.3826573426573425, "percentage": 46.09, "elapsed_time": "4:54:57", "remaining_time": "5:45:03"} +{"current_steps": 2473, "total_steps": 5364, "loss": 0.17842406034469604, "lr": 1.306416113442721e-05, "epoch": 1.3832167832167832, "percentage": 46.1, "elapsed_time": "4:55:04", "remaining_time": "5:44:56"} +{"current_steps": 2474, "total_steps": 5364, "loss": 0.1323685199022293, "lr": 1.3057965178842249e-05, "epoch": 1.383776223776224, "percentage": 46.12, "elapsed_time": "4:55:11", "remaining_time": "5:44:49"} +{"current_steps": 2475, "total_steps": 5364, "loss": 0.09644083678722382, "lr": 1.3051767927935508e-05, "epoch": 1.3843356643356644, "percentage": 46.14, "elapsed_time": "4:55:18", "remaining_time": "5:44:41"} +{"current_steps": 2476, "total_steps": 5364, "loss": 0.3673185110092163, "lr": 1.3045569384332075e-05, "epoch": 1.3848951048951048, "percentage": 46.16, "elapsed_time": "4:55:25", "remaining_time": "5:44:34"} +{"current_steps": 2477, "total_steps": 5364, "loss": 0.16871073842048645, "lr": 1.3039369550657594e-05, "epoch": 1.3854545454545455, "percentage": 46.18, "elapsed_time": "4:55:32", "remaining_time": "5:44:27"} +{"current_steps": 2478, "total_steps": 5364, "loss": 0.22378090023994446, "lr": 1.3033168429538243e-05, "epoch": 1.386013986013986, "percentage": 46.2, "elapsed_time": "4:55:39", "remaining_time": "5:44:19"} +{"current_steps": 2479, "total_steps": 5364, "loss": 0.11769779771566391, "lr": 1.3026966023600751e-05, "epoch": 1.3865734265734266, "percentage": 46.22, "elapsed_time": "4:55:46", "remaining_time": "5:44:12"} +{"current_steps": 2480, "total_steps": 5364, "loss": 0.0978880524635315, "lr": 1.3020762335472393e-05, "epoch": 1.387132867132867, "percentage": 46.23, "elapsed_time": "4:55:52", "remaining_time": "5:44:04"} +{"current_steps": 2481, "total_steps": 5364, "loss": 0.229050874710083, "lr": 1.3014557367780984e-05, "epoch": 1.3876923076923076, "percentage": 46.25, "elapsed_time": "4:55:59", "remaining_time": "5:43:57"} +{"current_steps": 2482, "total_steps": 5364, "loss": 0.04241913557052612, "lr": 1.3008351123154889e-05, "epoch": 1.3882517482517482, "percentage": 46.27, "elapsed_time": "4:56:06", "remaining_time": "5:43:50"} +{"current_steps": 2483, "total_steps": 5364, "loss": 0.13000048696994781, "lr": 1.3002143604223006e-05, "epoch": 1.388811188811189, "percentage": 46.29, "elapsed_time": "4:56:13", "remaining_time": "5:43:42"} +{"current_steps": 2484, "total_steps": 5364, "loss": 0.09359068423509598, "lr": 1.2995934813614768e-05, "epoch": 1.3893706293706294, "percentage": 46.31, "elapsed_time": "4:56:21", "remaining_time": "5:43:35"} +{"current_steps": 2485, "total_steps": 5364, "loss": 0.3084734380245209, "lr": 1.2989724753960155e-05, "epoch": 1.3899300699300698, "percentage": 46.33, "elapsed_time": "4:56:28", "remaining_time": "5:43:28"} +{"current_steps": 2486, "total_steps": 5364, "loss": 0.19326458871364594, "lr": 1.2983513427889692e-05, "epoch": 1.3904895104895105, "percentage": 46.35, "elapsed_time": "4:56:35", "remaining_time": "5:43:21"} +{"current_steps": 2487, "total_steps": 5364, "loss": 0.26621443033218384, "lr": 1.2977300838034422e-05, "epoch": 1.391048951048951, "percentage": 46.36, "elapsed_time": "4:56:42", "remaining_time": "5:43:14"} +{"current_steps": 2488, "total_steps": 5364, "loss": 0.06607040762901306, "lr": 1.2971086987025933e-05, "epoch": 1.3916083916083917, "percentage": 46.38, "elapsed_time": "4:56:49", "remaining_time": "5:43:07"} +{"current_steps": 2489, "total_steps": 5364, "loss": 0.2056734710931778, "lr": 1.2964871877496347e-05, "epoch": 1.3921678321678321, "percentage": 46.4, "elapsed_time": "4:56:56", "remaining_time": "5:42:59"} +{"current_steps": 2490, "total_steps": 5364, "loss": 0.21388250589370728, "lr": 1.2958655512078324e-05, "epoch": 1.3927272727272726, "percentage": 46.42, "elapsed_time": "4:57:03", "remaining_time": "5:42:52"} +{"current_steps": 2491, "total_steps": 5364, "loss": 0.0713212862610817, "lr": 1.2952437893405048e-05, "epoch": 1.3932867132867133, "percentage": 46.44, "elapsed_time": "4:57:10", "remaining_time": "5:42:45"} +{"current_steps": 2492, "total_steps": 5364, "loss": 0.06287550926208496, "lr": 1.2946219024110232e-05, "epoch": 1.393846153846154, "percentage": 46.46, "elapsed_time": "4:57:18", "remaining_time": "5:42:38"} +{"current_steps": 2493, "total_steps": 5364, "loss": 0.1461336314678192, "lr": 1.2939998906828131e-05, "epoch": 1.3944055944055944, "percentage": 46.48, "elapsed_time": "4:57:25", "remaining_time": "5:42:31"} +{"current_steps": 2494, "total_steps": 5364, "loss": 0.16389258205890656, "lr": 1.2933777544193514e-05, "epoch": 1.3949650349650349, "percentage": 46.5, "elapsed_time": "4:57:32", "remaining_time": "5:42:24"} +{"current_steps": 2495, "total_steps": 5364, "loss": 0.07075624167919159, "lr": 1.2927554938841689e-05, "epoch": 1.3955244755244756, "percentage": 46.51, "elapsed_time": "4:57:39", "remaining_time": "5:42:16"} +{"current_steps": 2496, "total_steps": 5364, "loss": 0.2380932867527008, "lr": 1.2921331093408487e-05, "epoch": 1.396083916083916, "percentage": 46.53, "elapsed_time": "4:57:46", "remaining_time": "5:42:09"} +{"current_steps": 2497, "total_steps": 5364, "loss": 0.06349461525678635, "lr": 1.2915106010530259e-05, "epoch": 1.3966433566433567, "percentage": 46.55, "elapsed_time": "4:57:54", "remaining_time": "5:42:02"} +{"current_steps": 2498, "total_steps": 5364, "loss": 0.19395595788955688, "lr": 1.2908879692843883e-05, "epoch": 1.3972027972027972, "percentage": 46.57, "elapsed_time": "4:58:00", "remaining_time": "5:41:55"} +{"current_steps": 2499, "total_steps": 5364, "loss": 0.2335360199213028, "lr": 1.2902652142986771e-05, "epoch": 1.3977622377622376, "percentage": 46.59, "elapsed_time": "4:58:08", "remaining_time": "5:41:48"} +{"current_steps": 2500, "total_steps": 5364, "loss": 0.018601372838020325, "lr": 1.289642336359684e-05, "epoch": 1.3983216783216783, "percentage": 46.61, "elapsed_time": "4:58:15", "remaining_time": "5:41:40"} +{"current_steps": 2501, "total_steps": 5364, "loss": 0.1681356132030487, "lr": 1.2890193357312534e-05, "epoch": 1.398881118881119, "percentage": 46.63, "elapsed_time": "4:58:50", "remaining_time": "5:42:05"} +{"current_steps": 2502, "total_steps": 5364, "loss": 0.0960170328617096, "lr": 1.2883962126772823e-05, "epoch": 1.3994405594405595, "percentage": 46.64, "elapsed_time": "4:58:57", "remaining_time": "5:41:58"} +{"current_steps": 2503, "total_steps": 5364, "loss": 0.06612128019332886, "lr": 1.2877729674617185e-05, "epoch": 1.4, "percentage": 46.66, "elapsed_time": "4:59:04", "remaining_time": "5:41:51"} +{"current_steps": 2504, "total_steps": 5364, "loss": 0.1970897614955902, "lr": 1.2871496003485628e-05, "epoch": 1.4005594405594406, "percentage": 46.68, "elapsed_time": "4:59:11", "remaining_time": "5:41:43"} +{"current_steps": 2505, "total_steps": 5364, "loss": 0.29299888014793396, "lr": 1.2865261116018664e-05, "epoch": 1.401118881118881, "percentage": 46.7, "elapsed_time": "4:59:18", "remaining_time": "5:41:36"} +{"current_steps": 2506, "total_steps": 5364, "loss": 0.09906235337257385, "lr": 1.2859025014857327e-05, "epoch": 1.4016783216783217, "percentage": 46.72, "elapsed_time": "4:59:25", "remaining_time": "5:41:29"} +{"current_steps": 2507, "total_steps": 5364, "loss": 0.21030399203300476, "lr": 1.2852787702643162e-05, "epoch": 1.4022377622377622, "percentage": 46.74, "elapsed_time": "4:59:32", "remaining_time": "5:41:21"} +{"current_steps": 2508, "total_steps": 5364, "loss": 0.11501772701740265, "lr": 1.284654918201823e-05, "epoch": 1.4027972027972029, "percentage": 46.76, "elapsed_time": "4:59:40", "remaining_time": "5:41:15"} +{"current_steps": 2509, "total_steps": 5364, "loss": 0.15339097380638123, "lr": 1.2840309455625103e-05, "epoch": 1.4033566433566433, "percentage": 46.77, "elapsed_time": "4:59:47", "remaining_time": "5:41:07"} +{"current_steps": 2510, "total_steps": 5364, "loss": 0.25876539945602417, "lr": 1.2834068526106865e-05, "epoch": 1.403916083916084, "percentage": 46.79, "elapsed_time": "4:59:54", "remaining_time": "5:41:00"} +{"current_steps": 2511, "total_steps": 5364, "loss": 0.16346001625061035, "lr": 1.2827826396107103e-05, "epoch": 1.4044755244755245, "percentage": 46.81, "elapsed_time": "5:00:01", "remaining_time": "5:40:53"} +{"current_steps": 2512, "total_steps": 5364, "loss": 0.1334468573331833, "lr": 1.282158306826992e-05, "epoch": 1.405034965034965, "percentage": 46.83, "elapsed_time": "5:00:08", "remaining_time": "5:40:46"} +{"current_steps": 2513, "total_steps": 5364, "loss": 0.1394098550081253, "lr": 1.2815338545239926e-05, "epoch": 1.4055944055944056, "percentage": 46.85, "elapsed_time": "5:00:16", "remaining_time": "5:40:39"} +{"current_steps": 2514, "total_steps": 5364, "loss": 0.061817608773708344, "lr": 1.2809092829662233e-05, "epoch": 1.406153846153846, "percentage": 46.87, "elapsed_time": "5:00:23", "remaining_time": "5:40:32"} +{"current_steps": 2515, "total_steps": 5364, "loss": 0.15196193754673004, "lr": 1.2802845924182457e-05, "epoch": 1.4067132867132868, "percentage": 46.89, "elapsed_time": "5:00:30", "remaining_time": "5:40:25"} +{"current_steps": 2516, "total_steps": 5364, "loss": 0.07467684149742126, "lr": 1.2796597831446728e-05, "epoch": 1.4072727272727272, "percentage": 46.91, "elapsed_time": "5:00:37", "remaining_time": "5:40:18"} +{"current_steps": 2517, "total_steps": 5364, "loss": 0.16751444339752197, "lr": 1.2790348554101666e-05, "epoch": 1.407832167832168, "percentage": 46.92, "elapsed_time": "5:00:45", "remaining_time": "5:40:11"} +{"current_steps": 2518, "total_steps": 5364, "loss": 0.2549717128276825, "lr": 1.2784098094794406e-05, "epoch": 1.4083916083916084, "percentage": 46.94, "elapsed_time": "5:00:52", "remaining_time": "5:40:04"} +{"current_steps": 2519, "total_steps": 5364, "loss": 0.06868081539869308, "lr": 1.2777846456172568e-05, "epoch": 1.408951048951049, "percentage": 46.96, "elapsed_time": "5:01:00", "remaining_time": "5:39:57"} +{"current_steps": 2520, "total_steps": 5364, "loss": 0.28834298253059387, "lr": 1.2771593640884283e-05, "epoch": 1.4095104895104895, "percentage": 46.98, "elapsed_time": "5:01:07", "remaining_time": "5:39:50"} +{"current_steps": 2521, "total_steps": 5364, "loss": 0.09444493055343628, "lr": 1.2765339651578183e-05, "epoch": 1.41006993006993, "percentage": 47.0, "elapsed_time": "5:01:14", "remaining_time": "5:39:42"} +{"current_steps": 2522, "total_steps": 5364, "loss": 0.20549622178077698, "lr": 1.2759084490903388e-05, "epoch": 1.4106293706293707, "percentage": 47.02, "elapsed_time": "5:01:21", "remaining_time": "5:39:35"} +{"current_steps": 2523, "total_steps": 5364, "loss": 0.13213779032230377, "lr": 1.2752828161509514e-05, "epoch": 1.4111888111888111, "percentage": 47.04, "elapsed_time": "5:01:27", "remaining_time": "5:39:27"} +{"current_steps": 2524, "total_steps": 5364, "loss": 0.2385183572769165, "lr": 1.2746570666046681e-05, "epoch": 1.4117482517482518, "percentage": 47.05, "elapsed_time": "5:01:34", "remaining_time": "5:39:20"} +{"current_steps": 2525, "total_steps": 5364, "loss": 0.12792335450649261, "lr": 1.2740312007165496e-05, "epoch": 1.4123076923076923, "percentage": 47.07, "elapsed_time": "5:01:42", "remaining_time": "5:39:13"} +{"current_steps": 2526, "total_steps": 5364, "loss": 0.06595709919929504, "lr": 1.2734052187517065e-05, "epoch": 1.412867132867133, "percentage": 47.09, "elapsed_time": "5:01:49", "remaining_time": "5:39:06"} +{"current_steps": 2527, "total_steps": 5364, "loss": 0.39011693000793457, "lr": 1.2727791209752977e-05, "epoch": 1.4134265734265734, "percentage": 47.11, "elapsed_time": "5:01:56", "remaining_time": "5:38:58"} +{"current_steps": 2528, "total_steps": 5364, "loss": 0.27636608481407166, "lr": 1.2721529076525318e-05, "epoch": 1.413986013986014, "percentage": 47.13, "elapsed_time": "5:02:03", "remaining_time": "5:38:51"} +{"current_steps": 2529, "total_steps": 5364, "loss": 0.15842606127262115, "lr": 1.271526579048666e-05, "epoch": 1.4145454545454546, "percentage": 47.15, "elapsed_time": "5:02:10", "remaining_time": "5:38:44"} +{"current_steps": 2530, "total_steps": 5364, "loss": 0.0735919326543808, "lr": 1.2709001354290064e-05, "epoch": 1.415104895104895, "percentage": 47.17, "elapsed_time": "5:02:17", "remaining_time": "5:38:37"} +{"current_steps": 2531, "total_steps": 5364, "loss": 0.20895975828170776, "lr": 1.2702735770589085e-05, "epoch": 1.4156643356643357, "percentage": 47.18, "elapsed_time": "5:02:24", "remaining_time": "5:38:29"} +{"current_steps": 2532, "total_steps": 5364, "loss": 0.34306687116622925, "lr": 1.2696469042037752e-05, "epoch": 1.4162237762237762, "percentage": 47.2, "elapsed_time": "5:02:31", "remaining_time": "5:38:22"} +{"current_steps": 2533, "total_steps": 5364, "loss": 0.1085163950920105, "lr": 1.2690201171290588e-05, "epoch": 1.4167832167832168, "percentage": 47.22, "elapsed_time": "5:02:38", "remaining_time": "5:38:15"} +{"current_steps": 2534, "total_steps": 5364, "loss": 0.025592666119337082, "lr": 1.2683932161002596e-05, "epoch": 1.4173426573426573, "percentage": 47.24, "elapsed_time": "5:02:45", "remaining_time": "5:38:07"} +{"current_steps": 2535, "total_steps": 5364, "loss": 0.09960824251174927, "lr": 1.2677662013829265e-05, "epoch": 1.417902097902098, "percentage": 47.26, "elapsed_time": "5:02:52", "remaining_time": "5:37:59"} +{"current_steps": 2536, "total_steps": 5364, "loss": 0.10015767812728882, "lr": 1.2671390732426553e-05, "epoch": 1.4184615384615384, "percentage": 47.28, "elapsed_time": "5:02:59", "remaining_time": "5:37:52"} +{"current_steps": 2537, "total_steps": 5364, "loss": 0.11094857007265091, "lr": 1.2665118319450925e-05, "epoch": 1.4190209790209791, "percentage": 47.3, "elapsed_time": "5:03:06", "remaining_time": "5:37:44"} +{"current_steps": 2538, "total_steps": 5364, "loss": 0.1043037548661232, "lr": 1.2658844777559297e-05, "epoch": 1.4195804195804196, "percentage": 47.32, "elapsed_time": "5:03:13", "remaining_time": "5:37:38"} +{"current_steps": 2539, "total_steps": 5364, "loss": 0.26796773076057434, "lr": 1.2652570109409077e-05, "epoch": 1.42013986013986, "percentage": 47.33, "elapsed_time": "5:03:20", "remaining_time": "5:37:31"} +{"current_steps": 2540, "total_steps": 5364, "loss": 0.03737884759902954, "lr": 1.2646294317658151e-05, "epoch": 1.4206993006993007, "percentage": 47.35, "elapsed_time": "5:03:29", "remaining_time": "5:37:25"} +{"current_steps": 2541, "total_steps": 5364, "loss": 0.19634473323822021, "lr": 1.2640017404964877e-05, "epoch": 1.4212587412587412, "percentage": 47.37, "elapsed_time": "5:03:36", "remaining_time": "5:37:17"} +{"current_steps": 2542, "total_steps": 5364, "loss": 0.22753392159938812, "lr": 1.263373937398809e-05, "epoch": 1.4218181818181819, "percentage": 47.39, "elapsed_time": "5:03:44", "remaining_time": "5:37:11"} +{"current_steps": 2543, "total_steps": 5364, "loss": 0.19683019816875458, "lr": 1.2627460227387095e-05, "epoch": 1.4223776223776223, "percentage": 47.41, "elapsed_time": "5:03:51", "remaining_time": "5:37:04"} +{"current_steps": 2544, "total_steps": 5364, "loss": 0.11294884234666824, "lr": 1.2621179967821675e-05, "epoch": 1.422937062937063, "percentage": 47.43, "elapsed_time": "5:03:58", "remaining_time": "5:36:56"} +{"current_steps": 2545, "total_steps": 5364, "loss": 0.21752138435840607, "lr": 1.2614898597952084e-05, "epoch": 1.4234965034965035, "percentage": 47.45, "elapsed_time": "5:04:05", "remaining_time": "5:36:49"} +{"current_steps": 2546, "total_steps": 5364, "loss": 0.1418994665145874, "lr": 1.2608616120439034e-05, "epoch": 1.4240559440559442, "percentage": 47.46, "elapsed_time": "5:04:11", "remaining_time": "5:36:41"} +{"current_steps": 2547, "total_steps": 5364, "loss": 0.14981484413146973, "lr": 1.2602332537943727e-05, "epoch": 1.4246153846153846, "percentage": 47.48, "elapsed_time": "5:04:19", "remaining_time": "5:36:34"} +{"current_steps": 2548, "total_steps": 5364, "loss": 0.16420406103134155, "lr": 1.2596047853127818e-05, "epoch": 1.425174825174825, "percentage": 47.5, "elapsed_time": "5:04:26", "remaining_time": "5:36:27"} +{"current_steps": 2549, "total_steps": 5364, "loss": 0.180812269449234, "lr": 1.2589762068653434e-05, "epoch": 1.4257342657342658, "percentage": 47.52, "elapsed_time": "5:04:33", "remaining_time": "5:36:20"} +{"current_steps": 2550, "total_steps": 5364, "loss": 0.16212785243988037, "lr": 1.258347518718316e-05, "epoch": 1.4262937062937062, "percentage": 47.54, "elapsed_time": "5:04:40", "remaining_time": "5:36:13"} +{"current_steps": 2551, "total_steps": 5364, "loss": 0.11519194394350052, "lr": 1.2577187211380062e-05, "epoch": 1.426853146853147, "percentage": 47.56, "elapsed_time": "5:04:47", "remaining_time": "5:36:06"} +{"current_steps": 2552, "total_steps": 5364, "loss": 0.05389915406703949, "lr": 1.2570898143907657e-05, "epoch": 1.4274125874125874, "percentage": 47.58, "elapsed_time": "5:04:54", "remaining_time": "5:35:58"} +{"current_steps": 2553, "total_steps": 5364, "loss": 0.20748907327651978, "lr": 1.2564607987429928e-05, "epoch": 1.427972027972028, "percentage": 47.6, "elapsed_time": "5:05:01", "remaining_time": "5:35:51"} +{"current_steps": 2554, "total_steps": 5364, "loss": 0.07955549657344818, "lr": 1.255831674461132e-05, "epoch": 1.4285314685314685, "percentage": 47.61, "elapsed_time": "5:05:08", "remaining_time": "5:35:43"} +{"current_steps": 2555, "total_steps": 5364, "loss": 0.16949276626110077, "lr": 1.2552024418116735e-05, "epoch": 1.4290909090909092, "percentage": 47.63, "elapsed_time": "5:05:15", "remaining_time": "5:35:36"} +{"current_steps": 2556, "total_steps": 5364, "loss": 0.1735232174396515, "lr": 1.2545731010611535e-05, "epoch": 1.4296503496503496, "percentage": 47.65, "elapsed_time": "5:05:22", "remaining_time": "5:35:29"} +{"current_steps": 2557, "total_steps": 5364, "loss": 0.1384444534778595, "lr": 1.2539436524761542e-05, "epoch": 1.43020979020979, "percentage": 47.67, "elapsed_time": "5:05:30", "remaining_time": "5:35:22"} +{"current_steps": 2558, "total_steps": 5364, "loss": 0.08492761105298996, "lr": 1.2533140963233037e-05, "epoch": 1.4307692307692308, "percentage": 47.69, "elapsed_time": "5:05:37", "remaining_time": "5:35:15"} +{"current_steps": 2559, "total_steps": 5364, "loss": 0.1551174521446228, "lr": 1.2526844328692748e-05, "epoch": 1.4313286713286713, "percentage": 47.71, "elapsed_time": "5:05:44", "remaining_time": "5:35:07"} +{"current_steps": 2560, "total_steps": 5364, "loss": 0.2093212604522705, "lr": 1.252054662380787e-05, "epoch": 1.431888111888112, "percentage": 47.73, "elapsed_time": "5:05:51", "remaining_time": "5:35:00"} +{"current_steps": 2561, "total_steps": 5364, "loss": 0.17169030010700226, "lr": 1.2514247851246038e-05, "epoch": 1.4324475524475524, "percentage": 47.74, "elapsed_time": "5:05:58", "remaining_time": "5:34:52"} +{"current_steps": 2562, "total_steps": 5364, "loss": 0.015126602724194527, "lr": 1.2507948013675346e-05, "epoch": 1.433006993006993, "percentage": 47.76, "elapsed_time": "5:06:05", "remaining_time": "5:34:45"} +{"current_steps": 2563, "total_steps": 5364, "loss": 0.19893920421600342, "lr": 1.2501647113764342e-05, "epoch": 1.4335664335664335, "percentage": 47.78, "elapsed_time": "5:06:11", "remaining_time": "5:34:38"} +{"current_steps": 2564, "total_steps": 5364, "loss": 0.29627296328544617, "lr": 1.2495345154182018e-05, "epoch": 1.4341258741258742, "percentage": 47.8, "elapsed_time": "5:06:19", "remaining_time": "5:34:30"} +{"current_steps": 2565, "total_steps": 5364, "loss": 0.09909462928771973, "lr": 1.2489042137597817e-05, "epoch": 1.4346853146853147, "percentage": 47.82, "elapsed_time": "5:06:26", "remaining_time": "5:34:23"} +{"current_steps": 2566, "total_steps": 5364, "loss": 0.053327735513448715, "lr": 1.2482738066681636e-05, "epoch": 1.4352447552447551, "percentage": 47.84, "elapsed_time": "5:06:32", "remaining_time": "5:34:15"} +{"current_steps": 2567, "total_steps": 5364, "loss": 0.0766044333577156, "lr": 1.2476432944103811e-05, "epoch": 1.4358041958041958, "percentage": 47.86, "elapsed_time": "5:06:39", "remaining_time": "5:34:08"} +{"current_steps": 2568, "total_steps": 5364, "loss": 0.12152018398046494, "lr": 1.2470126772535118e-05, "epoch": 1.4363636363636363, "percentage": 47.87, "elapsed_time": "5:06:46", "remaining_time": "5:34:00"} +{"current_steps": 2569, "total_steps": 5364, "loss": 0.18582960963249207, "lr": 1.2463819554646792e-05, "epoch": 1.436923076923077, "percentage": 47.89, "elapsed_time": "5:06:53", "remaining_time": "5:33:53"} +{"current_steps": 2570, "total_steps": 5364, "loss": 0.08280140906572342, "lr": 1.24575112931105e-05, "epoch": 1.4374825174825174, "percentage": 47.91, "elapsed_time": "5:07:01", "remaining_time": "5:33:47"} +{"current_steps": 2571, "total_steps": 5364, "loss": 0.17425097525119781, "lr": 1.2451201990598362e-05, "epoch": 1.4380419580419581, "percentage": 47.93, "elapsed_time": "5:07:08", "remaining_time": "5:33:39"} +{"current_steps": 2572, "total_steps": 5364, "loss": 0.01326349750161171, "lr": 1.2444891649782925e-05, "epoch": 1.4386013986013986, "percentage": 47.95, "elapsed_time": "5:07:16", "remaining_time": "5:33:33"} +{"current_steps": 2573, "total_steps": 5364, "loss": 0.2997506856918335, "lr": 1.2438580273337185e-05, "epoch": 1.4391608391608393, "percentage": 47.97, "elapsed_time": "5:07:23", "remaining_time": "5:33:25"} +{"current_steps": 2574, "total_steps": 5364, "loss": 0.05882330983877182, "lr": 1.243226786393457e-05, "epoch": 1.4397202797202797, "percentage": 47.99, "elapsed_time": "5:07:30", "remaining_time": "5:33:18"} +{"current_steps": 2575, "total_steps": 5364, "loss": 0.05277079716324806, "lr": 1.2425954424248959e-05, "epoch": 1.4402797202797202, "percentage": 48.01, "elapsed_time": "5:07:37", "remaining_time": "5:33:11"} +{"current_steps": 2576, "total_steps": 5364, "loss": 0.15838435292243958, "lr": 1.2419639956954653e-05, "epoch": 1.4408391608391609, "percentage": 48.02, "elapsed_time": "5:07:44", "remaining_time": "5:33:04"} +{"current_steps": 2577, "total_steps": 5364, "loss": 0.12325419485569, "lr": 1.2413324464726393e-05, "epoch": 1.4413986013986013, "percentage": 48.04, "elapsed_time": "5:07:52", "remaining_time": "5:32:57"} +{"current_steps": 2578, "total_steps": 5364, "loss": 0.3183799982070923, "lr": 1.2407007950239356e-05, "epoch": 1.441958041958042, "percentage": 48.06, "elapsed_time": "5:07:59", "remaining_time": "5:32:50"} +{"current_steps": 2579, "total_steps": 5364, "loss": 0.07133820652961731, "lr": 1.2400690416169153e-05, "epoch": 1.4425174825174825, "percentage": 48.08, "elapsed_time": "5:08:06", "remaining_time": "5:32:42"} +{"current_steps": 2580, "total_steps": 5364, "loss": 0.17543654143810272, "lr": 1.2394371865191821e-05, "epoch": 1.4430769230769231, "percentage": 48.1, "elapsed_time": "5:08:13", "remaining_time": "5:32:35"} +{"current_steps": 2581, "total_steps": 5364, "loss": 0.1259579211473465, "lr": 1.2388052299983833e-05, "epoch": 1.4436363636363636, "percentage": 48.12, "elapsed_time": "5:08:20", "remaining_time": "5:32:28"} +{"current_steps": 2582, "total_steps": 5364, "loss": 0.19516217708587646, "lr": 1.2381731723222092e-05, "epoch": 1.4441958041958043, "percentage": 48.14, "elapsed_time": "5:08:27", "remaining_time": "5:32:21"} +{"current_steps": 2583, "total_steps": 5364, "loss": 0.3713221549987793, "lr": 1.2375410137583927e-05, "epoch": 1.4447552447552447, "percentage": 48.15, "elapsed_time": "5:08:34", "remaining_time": "5:32:13"} +{"current_steps": 2584, "total_steps": 5364, "loss": 0.12221759557723999, "lr": 1.2369087545747093e-05, "epoch": 1.4453146853146852, "percentage": 48.17, "elapsed_time": "5:08:41", "remaining_time": "5:32:06"} +{"current_steps": 2585, "total_steps": 5364, "loss": 0.24273799359798431, "lr": 1.2362763950389776e-05, "epoch": 1.445874125874126, "percentage": 48.19, "elapsed_time": "5:08:48", "remaining_time": "5:31:58"} +{"current_steps": 2586, "total_steps": 5364, "loss": 0.17039065062999725, "lr": 1.2356439354190578e-05, "epoch": 1.4464335664335664, "percentage": 48.21, "elapsed_time": "5:08:55", "remaining_time": "5:31:51"} +{"current_steps": 2587, "total_steps": 5364, "loss": 0.045549459755420685, "lr": 1.235011375982854e-05, "epoch": 1.446993006993007, "percentage": 48.23, "elapsed_time": "5:09:03", "remaining_time": "5:31:44"} +{"current_steps": 2588, "total_steps": 5364, "loss": 0.05645807832479477, "lr": 1.2343787169983112e-05, "epoch": 1.4475524475524475, "percentage": 48.25, "elapsed_time": "5:09:10", "remaining_time": "5:31:37"} +{"current_steps": 2589, "total_steps": 5364, "loss": 0.03421954810619354, "lr": 1.2337459587334173e-05, "epoch": 1.4481118881118882, "percentage": 48.27, "elapsed_time": "5:09:17", "remaining_time": "5:31:30"} +{"current_steps": 2590, "total_steps": 5364, "loss": 0.15909573435783386, "lr": 1.2331131014562017e-05, "epoch": 1.4486713286713286, "percentage": 48.28, "elapsed_time": "5:09:24", "remaining_time": "5:31:23"} +{"current_steps": 2591, "total_steps": 5364, "loss": 0.27141815423965454, "lr": 1.2324801454347362e-05, "epoch": 1.4492307692307693, "percentage": 48.3, "elapsed_time": "5:09:31", "remaining_time": "5:31:16"} +{"current_steps": 2592, "total_steps": 5364, "loss": 0.13169139623641968, "lr": 1.2318470909371344e-05, "epoch": 1.4497902097902098, "percentage": 48.32, "elapsed_time": "5:09:39", "remaining_time": "5:31:09"} +{"current_steps": 2593, "total_steps": 5364, "loss": 0.10707405209541321, "lr": 1.2312139382315514e-05, "epoch": 1.4503496503496502, "percentage": 48.34, "elapsed_time": "5:09:47", "remaining_time": "5:31:03"} +{"current_steps": 2594, "total_steps": 5364, "loss": 0.1734994798898697, "lr": 1.230580687586184e-05, "epoch": 1.450909090909091, "percentage": 48.36, "elapsed_time": "5:09:54", "remaining_time": "5:30:55"} +{"current_steps": 2595, "total_steps": 5364, "loss": 0.07255610823631287, "lr": 1.2299473392692707e-05, "epoch": 1.4514685314685314, "percentage": 48.38, "elapsed_time": "5:10:00", "remaining_time": "5:30:48"} +{"current_steps": 2596, "total_steps": 5364, "loss": 0.132124662399292, "lr": 1.2293138935490907e-05, "epoch": 1.452027972027972, "percentage": 48.4, "elapsed_time": "5:10:07", "remaining_time": "5:30:40"} +{"current_steps": 2597, "total_steps": 5364, "loss": 0.1521194577217102, "lr": 1.2286803506939655e-05, "epoch": 1.4525874125874125, "percentage": 48.42, "elapsed_time": "5:10:14", "remaining_time": "5:30:33"} +{"current_steps": 2598, "total_steps": 5364, "loss": 0.08124281466007233, "lr": 1.2280467109722566e-05, "epoch": 1.4531468531468532, "percentage": 48.43, "elapsed_time": "5:10:21", "remaining_time": "5:30:26"} +{"current_steps": 2599, "total_steps": 5364, "loss": 0.10194060206413269, "lr": 1.2274129746523678e-05, "epoch": 1.4537062937062937, "percentage": 48.45, "elapsed_time": "5:10:28", "remaining_time": "5:30:18"} +{"current_steps": 2600, "total_steps": 5364, "loss": 0.11040084064006805, "lr": 1.2267791420027426e-05, "epoch": 1.4542657342657344, "percentage": 48.47, "elapsed_time": "5:10:36", "remaining_time": "5:30:11"} +{"current_steps": 2601, "total_steps": 5364, "loss": 0.12690399587154388, "lr": 1.2261452132918656e-05, "epoch": 1.4548251748251748, "percentage": 48.49, "elapsed_time": "5:10:42", "remaining_time": "5:30:04"} +{"current_steps": 2602, "total_steps": 5364, "loss": 0.17024722695350647, "lr": 1.225511188788263e-05, "epoch": 1.4553846153846153, "percentage": 48.51, "elapsed_time": "5:10:50", "remaining_time": "5:29:57"} +{"current_steps": 2603, "total_steps": 5364, "loss": 0.07141393423080444, "lr": 1.2248770687605002e-05, "epoch": 1.455944055944056, "percentage": 48.53, "elapsed_time": "5:10:57", "remaining_time": "5:29:49"} +{"current_steps": 2604, "total_steps": 5364, "loss": 0.025211479514837265, "lr": 1.2242428534771844e-05, "epoch": 1.4565034965034966, "percentage": 48.55, "elapsed_time": "5:11:05", "remaining_time": "5:29:43"} +{"current_steps": 2605, "total_steps": 5364, "loss": 0.12128719687461853, "lr": 1.2236085432069623e-05, "epoch": 1.457062937062937, "percentage": 48.56, "elapsed_time": "5:11:12", "remaining_time": "5:29:36"} +{"current_steps": 2606, "total_steps": 5364, "loss": 0.2586372494697571, "lr": 1.2229741382185202e-05, "epoch": 1.4576223776223776, "percentage": 48.58, "elapsed_time": "5:11:20", "remaining_time": "5:29:29"} +{"current_steps": 2607, "total_steps": 5364, "loss": 0.11020811647176743, "lr": 1.2223396387805867e-05, "epoch": 1.4581818181818182, "percentage": 48.6, "elapsed_time": "5:11:27", "remaining_time": "5:29:22"} +{"current_steps": 2608, "total_steps": 5364, "loss": 0.08487781137228012, "lr": 1.2217050451619284e-05, "epoch": 1.4587412587412587, "percentage": 48.62, "elapsed_time": "5:11:34", "remaining_time": "5:29:14"} +{"current_steps": 2609, "total_steps": 5364, "loss": 0.057219114154577255, "lr": 1.2210703576313523e-05, "epoch": 1.4593006993006994, "percentage": 48.64, "elapsed_time": "5:11:41", "remaining_time": "5:29:07"} +{"current_steps": 2610, "total_steps": 5364, "loss": 0.12177891284227371, "lr": 1.2204355764577058e-05, "epoch": 1.4598601398601398, "percentage": 48.66, "elapsed_time": "5:11:48", "remaining_time": "5:29:00"} +{"current_steps": 2611, "total_steps": 5364, "loss": 0.23259848356246948, "lr": 1.2198007019098751e-05, "epoch": 1.4604195804195803, "percentage": 48.68, "elapsed_time": "5:11:56", "remaining_time": "5:28:53"} +{"current_steps": 2612, "total_steps": 5364, "loss": 0.3108658194541931, "lr": 1.2191657342567864e-05, "epoch": 1.460979020979021, "percentage": 48.7, "elapsed_time": "5:12:02", "remaining_time": "5:28:46"} +{"current_steps": 2613, "total_steps": 5364, "loss": 0.03010696731507778, "lr": 1.2185306737674055e-05, "epoch": 1.4615384615384617, "percentage": 48.71, "elapsed_time": "5:12:10", "remaining_time": "5:28:39"} +{"current_steps": 2614, "total_steps": 5364, "loss": 0.08955302834510803, "lr": 1.217895520710737e-05, "epoch": 1.4620979020979021, "percentage": 48.73, "elapsed_time": "5:12:17", "remaining_time": "5:28:32"} +{"current_steps": 2615, "total_steps": 5364, "loss": 0.18781223893165588, "lr": 1.2172602753558259e-05, "epoch": 1.4626573426573426, "percentage": 48.75, "elapsed_time": "5:12:24", "remaining_time": "5:28:24"} +{"current_steps": 2616, "total_steps": 5364, "loss": 0.10175687819719315, "lr": 1.2166249379717543e-05, "epoch": 1.4632167832167833, "percentage": 48.77, "elapsed_time": "5:12:31", "remaining_time": "5:28:17"} +{"current_steps": 2617, "total_steps": 5364, "loss": 0.23766371607780457, "lr": 1.215989508827645e-05, "epoch": 1.4637762237762237, "percentage": 48.79, "elapsed_time": "5:12:38", "remaining_time": "5:28:09"} +{"current_steps": 2618, "total_steps": 5364, "loss": 0.1668854057788849, "lr": 1.2153539881926588e-05, "epoch": 1.4643356643356644, "percentage": 48.81, "elapsed_time": "5:12:45", "remaining_time": "5:28:02"} +{"current_steps": 2619, "total_steps": 5364, "loss": 0.11260493844747543, "lr": 1.2147183763359958e-05, "epoch": 1.4648951048951049, "percentage": 48.83, "elapsed_time": "5:12:52", "remaining_time": "5:27:55"} +{"current_steps": 2620, "total_steps": 5364, "loss": 0.04278929531574249, "lr": 1.2140826735268947e-05, "epoch": 1.4654545454545453, "percentage": 48.84, "elapsed_time": "5:12:59", "remaining_time": "5:27:48"} +{"current_steps": 2621, "total_steps": 5364, "loss": 0.2554653584957123, "lr": 1.2134468800346317e-05, "epoch": 1.466013986013986, "percentage": 48.86, "elapsed_time": "5:13:07", "remaining_time": "5:27:41"} +{"current_steps": 2622, "total_steps": 5364, "loss": 0.1261654943227768, "lr": 1.2128109961285233e-05, "epoch": 1.4665734265734267, "percentage": 48.88, "elapsed_time": "5:13:14", "remaining_time": "5:27:34"} +{"current_steps": 2623, "total_steps": 5364, "loss": 0.19858568906784058, "lr": 1.2121750220779222e-05, "epoch": 1.4671328671328672, "percentage": 48.9, "elapsed_time": "5:13:21", "remaining_time": "5:27:26"} +{"current_steps": 2624, "total_steps": 5364, "loss": 0.1181764304637909, "lr": 1.211538958152221e-05, "epoch": 1.4676923076923076, "percentage": 48.92, "elapsed_time": "5:13:27", "remaining_time": "5:27:19"} +{"current_steps": 2625, "total_steps": 5364, "loss": 0.2142762839794159, "lr": 1.210902804620849e-05, "epoch": 1.4682517482517483, "percentage": 48.94, "elapsed_time": "5:13:34", "remaining_time": "5:27:11"} +{"current_steps": 2626, "total_steps": 5364, "loss": 0.0807376429438591, "lr": 1.2102665617532748e-05, "epoch": 1.4688111888111888, "percentage": 48.96, "elapsed_time": "5:13:41", "remaining_time": "5:27:04"} +{"current_steps": 2627, "total_steps": 5364, "loss": 0.05056123435497284, "lr": 1.2096302298190036e-05, "epoch": 1.4693706293706295, "percentage": 48.97, "elapsed_time": "5:13:48", "remaining_time": "5:26:57"} +{"current_steps": 2628, "total_steps": 5364, "loss": 0.24962349236011505, "lr": 1.2089938090875795e-05, "epoch": 1.46993006993007, "percentage": 48.99, "elapsed_time": "5:13:55", "remaining_time": "5:26:49"} +{"current_steps": 2629, "total_steps": 5364, "loss": 0.0875111073255539, "lr": 1.2083572998285832e-05, "epoch": 1.4704895104895104, "percentage": 49.01, "elapsed_time": "5:14:02", "remaining_time": "5:26:42"} +{"current_steps": 2630, "total_steps": 5364, "loss": 0.16640730202198029, "lr": 1.2077207023116333e-05, "epoch": 1.471048951048951, "percentage": 49.03, "elapsed_time": "5:14:09", "remaining_time": "5:26:35"} +{"current_steps": 2631, "total_steps": 5364, "loss": 0.17294815182685852, "lr": 1.2070840168063858e-05, "epoch": 1.4716083916083917, "percentage": 49.05, "elapsed_time": "5:14:16", "remaining_time": "5:26:27"} +{"current_steps": 2632, "total_steps": 5364, "loss": 0.06078283116221428, "lr": 1.2064472435825339e-05, "epoch": 1.4721678321678322, "percentage": 49.07, "elapsed_time": "5:14:24", "remaining_time": "5:26:21"} +{"current_steps": 2633, "total_steps": 5364, "loss": 0.329429566860199, "lr": 1.2058103829098087e-05, "epoch": 1.4727272727272727, "percentage": 49.09, "elapsed_time": "5:14:31", "remaining_time": "5:26:13"} +{"current_steps": 2634, "total_steps": 5364, "loss": 0.10931099206209183, "lr": 1.2051734350579767e-05, "epoch": 1.4732867132867133, "percentage": 49.11, "elapsed_time": "5:14:38", "remaining_time": "5:26:07"} +{"current_steps": 2635, "total_steps": 5364, "loss": 0.17708632349967957, "lr": 1.2045364002968425e-05, "epoch": 1.4738461538461538, "percentage": 49.12, "elapsed_time": "5:14:46", "remaining_time": "5:26:00"} +{"current_steps": 2636, "total_steps": 5364, "loss": 0.2560347318649292, "lr": 1.2038992788962478e-05, "epoch": 1.4744055944055945, "percentage": 49.14, "elapsed_time": "5:14:53", "remaining_time": "5:25:52"} +{"current_steps": 2637, "total_steps": 5364, "loss": 0.3001738488674164, "lr": 1.2032620711260702e-05, "epoch": 1.474965034965035, "percentage": 49.16, "elapsed_time": "5:15:00", "remaining_time": "5:25:45"} +{"current_steps": 2638, "total_steps": 5364, "loss": 0.28792041540145874, "lr": 1.2026247772562241e-05, "epoch": 1.4755244755244754, "percentage": 49.18, "elapsed_time": "5:15:07", "remaining_time": "5:25:38"} +{"current_steps": 2639, "total_steps": 5364, "loss": 0.1711592823266983, "lr": 1.2019873975566607e-05, "epoch": 1.476083916083916, "percentage": 49.2, "elapsed_time": "5:15:14", "remaining_time": "5:25:31"} +{"current_steps": 2640, "total_steps": 5364, "loss": 0.0962599664926529, "lr": 1.201349932297367e-05, "epoch": 1.4766433566433568, "percentage": 49.22, "elapsed_time": "5:15:22", "remaining_time": "5:25:24"} +{"current_steps": 2641, "total_steps": 5364, "loss": 0.059583887457847595, "lr": 1.2007123817483666e-05, "epoch": 1.4772027972027972, "percentage": 49.24, "elapsed_time": "5:15:30", "remaining_time": "5:25:17"} +{"current_steps": 2642, "total_steps": 5364, "loss": 0.11674322932958603, "lr": 1.2000747461797195e-05, "epoch": 1.4777622377622377, "percentage": 49.25, "elapsed_time": "5:15:37", "remaining_time": "5:25:10"} +{"current_steps": 2643, "total_steps": 5364, "loss": 0.277969628572464, "lr": 1.1994370258615211e-05, "epoch": 1.4783216783216784, "percentage": 49.27, "elapsed_time": "5:15:44", "remaining_time": "5:25:03"} +{"current_steps": 2644, "total_steps": 5364, "loss": 0.1078287810087204, "lr": 1.1987992210639033e-05, "epoch": 1.4788811188811188, "percentage": 49.29, "elapsed_time": "5:15:51", "remaining_time": "5:24:56"} +{"current_steps": 2645, "total_steps": 5364, "loss": 0.12364526093006134, "lr": 1.1981613320570328e-05, "epoch": 1.4794405594405595, "percentage": 49.31, "elapsed_time": "5:15:58", "remaining_time": "5:24:49"} +{"current_steps": 2646, "total_steps": 5364, "loss": 0.13781434297561646, "lr": 1.1975233591111138e-05, "epoch": 1.48, "percentage": 49.33, "elapsed_time": "5:16:05", "remaining_time": "5:24:41"} +{"current_steps": 2647, "total_steps": 5364, "loss": 0.11081355810165405, "lr": 1.1968853024963838e-05, "epoch": 1.4805594405594404, "percentage": 49.35, "elapsed_time": "5:16:12", "remaining_time": "5:24:34"} +{"current_steps": 2648, "total_steps": 5364, "loss": 0.0981362834572792, "lr": 1.1962471624831174e-05, "epoch": 1.4811188811188811, "percentage": 49.37, "elapsed_time": "5:16:19", "remaining_time": "5:24:27"} +{"current_steps": 2649, "total_steps": 5364, "loss": 0.17977339029312134, "lr": 1.1956089393416237e-05, "epoch": 1.4816783216783218, "percentage": 49.38, "elapsed_time": "5:16:27", "remaining_time": "5:24:20"} +{"current_steps": 2650, "total_steps": 5364, "loss": 0.032853513956069946, "lr": 1.1949706333422476e-05, "epoch": 1.4822377622377623, "percentage": 49.4, "elapsed_time": "5:16:34", "remaining_time": "5:24:13"} +{"current_steps": 2651, "total_steps": 5364, "loss": 0.01158855389803648, "lr": 1.1943322447553686e-05, "epoch": 1.4827972027972027, "percentage": 49.42, "elapsed_time": "5:16:42", "remaining_time": "5:24:06"} +{"current_steps": 2652, "total_steps": 5364, "loss": 0.03143032267689705, "lr": 1.1936937738514015e-05, "epoch": 1.4833566433566434, "percentage": 49.44, "elapsed_time": "5:16:50", "remaining_time": "5:24:00"} +{"current_steps": 2653, "total_steps": 5364, "loss": 0.23875242471694946, "lr": 1.1930552209007957e-05, "epoch": 1.4839160839160839, "percentage": 49.46, "elapsed_time": "5:16:56", "remaining_time": "5:23:52"} +{"current_steps": 2654, "total_steps": 5364, "loss": 0.08013374358415604, "lr": 1.1924165861740354e-05, "epoch": 1.4844755244755246, "percentage": 49.48, "elapsed_time": "5:17:03", "remaining_time": "5:23:45"} +{"current_steps": 2655, "total_steps": 5364, "loss": 0.10361069440841675, "lr": 1.1917778699416398e-05, "epoch": 1.485034965034965, "percentage": 49.5, "elapsed_time": "5:17:10", "remaining_time": "5:23:37"} +{"current_steps": 2656, "total_steps": 5364, "loss": 0.19031494855880737, "lr": 1.1911390724741625e-05, "epoch": 1.4855944055944055, "percentage": 49.52, "elapsed_time": "5:17:17", "remaining_time": "5:23:30"} +{"current_steps": 2657, "total_steps": 5364, "loss": 0.21351051330566406, "lr": 1.1905001940421906e-05, "epoch": 1.4861538461538462, "percentage": 49.53, "elapsed_time": "5:17:24", "remaining_time": "5:23:22"} +{"current_steps": 2658, "total_steps": 5364, "loss": 0.17911601066589355, "lr": 1.189861234916347e-05, "epoch": 1.4867132867132868, "percentage": 49.55, "elapsed_time": "5:17:31", "remaining_time": "5:23:15"} +{"current_steps": 2659, "total_steps": 5364, "loss": 0.0826939269900322, "lr": 1.1892221953672878e-05, "epoch": 1.4872727272727273, "percentage": 49.57, "elapsed_time": "5:17:38", "remaining_time": "5:23:08"} +{"current_steps": 2660, "total_steps": 5364, "loss": 0.037227433174848557, "lr": 1.1885830756657033e-05, "epoch": 1.4878321678321678, "percentage": 49.59, "elapsed_time": "5:17:46", "remaining_time": "5:23:01"} +{"current_steps": 2661, "total_steps": 5364, "loss": 0.06488566100597382, "lr": 1.1879438760823181e-05, "epoch": 1.4883916083916084, "percentage": 49.61, "elapsed_time": "5:17:53", "remaining_time": "5:22:54"} +{"current_steps": 2662, "total_steps": 5364, "loss": 0.04365688934922218, "lr": 1.18730459688789e-05, "epoch": 1.488951048951049, "percentage": 49.63, "elapsed_time": "5:17:59", "remaining_time": "5:22:46"} +{"current_steps": 2663, "total_steps": 5364, "loss": 0.1745668649673462, "lr": 1.1866652383532114e-05, "epoch": 1.4895104895104896, "percentage": 49.65, "elapsed_time": "5:18:07", "remaining_time": "5:22:39"} +{"current_steps": 2664, "total_steps": 5364, "loss": 0.08156780898571014, "lr": 1.1860258007491077e-05, "epoch": 1.49006993006993, "percentage": 49.66, "elapsed_time": "5:18:14", "remaining_time": "5:22:32"} +{"current_steps": 2665, "total_steps": 5364, "loss": 0.2273499071598053, "lr": 1.1853862843464375e-05, "epoch": 1.4906293706293705, "percentage": 49.68, "elapsed_time": "5:18:21", "remaining_time": "5:22:25"} +{"current_steps": 2666, "total_steps": 5364, "loss": 0.26180756092071533, "lr": 1.1847466894160937e-05, "epoch": 1.4911888111888112, "percentage": 49.7, "elapsed_time": "5:18:28", "remaining_time": "5:22:17"} +{"current_steps": 2667, "total_steps": 5364, "loss": 0.1497553139925003, "lr": 1.1841070162290019e-05, "epoch": 1.4917482517482519, "percentage": 49.72, "elapsed_time": "5:18:35", "remaining_time": "5:22:10"} +{"current_steps": 2668, "total_steps": 5364, "loss": 0.07229838520288467, "lr": 1.183467265056121e-05, "epoch": 1.4923076923076923, "percentage": 49.74, "elapsed_time": "5:18:42", "remaining_time": "5:22:03"} +{"current_steps": 2669, "total_steps": 5364, "loss": 0.36371690034866333, "lr": 1.1828274361684422e-05, "epoch": 1.4928671328671328, "percentage": 49.76, "elapsed_time": "5:18:50", "remaining_time": "5:21:56"} +{"current_steps": 2670, "total_steps": 5364, "loss": 0.1397406905889511, "lr": 1.1821875298369906e-05, "epoch": 1.4934265734265735, "percentage": 49.78, "elapsed_time": "5:18:57", "remaining_time": "5:21:49"} +{"current_steps": 2671, "total_steps": 5364, "loss": 0.17724046111106873, "lr": 1.1815475463328242e-05, "epoch": 1.493986013986014, "percentage": 49.79, "elapsed_time": "5:19:04", "remaining_time": "5:21:42"} +{"current_steps": 2672, "total_steps": 5364, "loss": 0.08694016933441162, "lr": 1.1809074859270333e-05, "epoch": 1.4945454545454546, "percentage": 49.81, "elapsed_time": "5:19:11", "remaining_time": "5:21:35"} +{"current_steps": 2673, "total_steps": 5364, "loss": 0.3452514111995697, "lr": 1.1802673488907401e-05, "epoch": 1.495104895104895, "percentage": 49.83, "elapsed_time": "5:19:19", "remaining_time": "5:21:28"} +{"current_steps": 2674, "total_steps": 5364, "loss": 0.03347887471318245, "lr": 1.1796271354951e-05, "epoch": 1.4956643356643355, "percentage": 49.85, "elapsed_time": "5:19:27", "remaining_time": "5:21:21"} +{"current_steps": 2675, "total_steps": 5364, "loss": 0.07179979979991913, "lr": 1.1789868460113008e-05, "epoch": 1.4962237762237762, "percentage": 49.87, "elapsed_time": "5:19:34", "remaining_time": "5:21:14"} +{"current_steps": 2676, "total_steps": 5364, "loss": 0.07434782385826111, "lr": 1.1783464807105625e-05, "epoch": 1.496783216783217, "percentage": 49.89, "elapsed_time": "5:19:41", "remaining_time": "5:21:07"} +{"current_steps": 2677, "total_steps": 5364, "loss": 0.17787566781044006, "lr": 1.177706039864137e-05, "epoch": 1.4973426573426574, "percentage": 49.91, "elapsed_time": "5:19:48", "remaining_time": "5:21:00"} +{"current_steps": 2678, "total_steps": 5364, "loss": 0.15541744232177734, "lr": 1.1770655237433084e-05, "epoch": 1.4979020979020978, "percentage": 49.93, "elapsed_time": "5:19:55", "remaining_time": "5:20:52"} +{"current_steps": 2679, "total_steps": 5364, "loss": 0.06536943465471268, "lr": 1.1764249326193922e-05, "epoch": 1.4984615384615385, "percentage": 49.94, "elapsed_time": "5:20:03", "remaining_time": "5:20:46"} +{"current_steps": 2680, "total_steps": 5364, "loss": 0.15037459135055542, "lr": 1.175784266763736e-05, "epoch": 1.499020979020979, "percentage": 49.96, "elapsed_time": "5:20:10", "remaining_time": "5:20:39"} +{"current_steps": 2681, "total_steps": 5364, "loss": 0.20892234146595, "lr": 1.17514352644772e-05, "epoch": 1.4995804195804197, "percentage": 49.98, "elapsed_time": "5:20:17", "remaining_time": "5:20:31"} +{"current_steps": 2682, "total_steps": 5364, "loss": 0.18815618753433228, "lr": 1.1745027119427538e-05, "epoch": 1.5001398601398601, "percentage": 50.0, "elapsed_time": "5:20:24", "remaining_time": "5:20:24"} +{"current_steps": 2683, "total_steps": 5364, "loss": 0.24392755329608917, "lr": 1.1738618235202805e-05, "epoch": 1.5006993006993006, "percentage": 50.02, "elapsed_time": "5:20:31", "remaining_time": "5:20:17"} +{"current_steps": 2684, "total_steps": 5364, "loss": 0.14816738665103912, "lr": 1.1732208614517736e-05, "epoch": 1.5012587412587413, "percentage": 50.04, "elapsed_time": "5:20:38", "remaining_time": "5:20:09"} +{"current_steps": 2685, "total_steps": 5364, "loss": 0.05145620182156563, "lr": 1.1725798260087374e-05, "epoch": 1.501818181818182, "percentage": 50.06, "elapsed_time": "5:20:45", "remaining_time": "5:20:02"} +{"current_steps": 2686, "total_steps": 5364, "loss": 0.14055439829826355, "lr": 1.1719387174627083e-05, "epoch": 1.5023776223776224, "percentage": 50.07, "elapsed_time": "5:20:52", "remaining_time": "5:19:55"} +{"current_steps": 2687, "total_steps": 5364, "loss": 0.06928224116563797, "lr": 1.1712975360852525e-05, "epoch": 1.5029370629370629, "percentage": 50.09, "elapsed_time": "5:20:59", "remaining_time": "5:19:47"} +{"current_steps": 2688, "total_steps": 5364, "loss": 0.18926560878753662, "lr": 1.1706562821479683e-05, "epoch": 1.5034965034965035, "percentage": 50.11, "elapsed_time": "5:21:06", "remaining_time": "5:19:40"} +{"current_steps": 2689, "total_steps": 5364, "loss": 0.211931049823761, "lr": 1.1700149559224839e-05, "epoch": 1.504055944055944, "percentage": 50.13, "elapsed_time": "5:21:13", "remaining_time": "5:19:32"} +{"current_steps": 2690, "total_steps": 5364, "loss": 0.08584072440862656, "lr": 1.1693735576804582e-05, "epoch": 1.5046153846153847, "percentage": 50.15, "elapsed_time": "5:21:20", "remaining_time": "5:19:25"} +{"current_steps": 2691, "total_steps": 5364, "loss": 0.1607426255941391, "lr": 1.1687320876935806e-05, "epoch": 1.5051748251748251, "percentage": 50.17, "elapsed_time": "5:21:27", "remaining_time": "5:19:18"} +{"current_steps": 2692, "total_steps": 5364, "loss": 0.08180231600999832, "lr": 1.1680905462335711e-05, "epoch": 1.5057342657342656, "percentage": 50.19, "elapsed_time": "5:21:35", "remaining_time": "5:19:12"} +{"current_steps": 2693, "total_steps": 5364, "loss": 0.25060927867889404, "lr": 1.1674489335721803e-05, "epoch": 1.5062937062937063, "percentage": 50.21, "elapsed_time": "5:21:43", "remaining_time": "5:19:05"} +{"current_steps": 2694, "total_steps": 5364, "loss": 0.08234649151563644, "lr": 1.1668072499811882e-05, "epoch": 1.506853146853147, "percentage": 50.22, "elapsed_time": "5:21:49", "remaining_time": "5:18:57"} +{"current_steps": 2695, "total_steps": 5364, "loss": 0.1301392912864685, "lr": 1.1661654957324053e-05, "epoch": 1.5074125874125874, "percentage": 50.24, "elapsed_time": "5:21:57", "remaining_time": "5:18:50"} +{"current_steps": 2696, "total_steps": 5364, "loss": 0.30901575088500977, "lr": 1.1655236710976716e-05, "epoch": 1.507972027972028, "percentage": 50.26, "elapsed_time": "5:22:04", "remaining_time": "5:18:43"} +{"current_steps": 2697, "total_steps": 5364, "loss": 0.12982720136642456, "lr": 1.1648817763488576e-05, "epoch": 1.5085314685314686, "percentage": 50.28, "elapsed_time": "5:22:11", "remaining_time": "5:18:36"} +{"current_steps": 2698, "total_steps": 5364, "loss": 0.24956434965133667, "lr": 1.164239811757863e-05, "epoch": 1.509090909090909, "percentage": 50.3, "elapsed_time": "5:22:18", "remaining_time": "5:18:29"} +{"current_steps": 2699, "total_steps": 5364, "loss": 0.20079776644706726, "lr": 1.1635977775966174e-05, "epoch": 1.5096503496503497, "percentage": 50.32, "elapsed_time": "5:22:25", "remaining_time": "5:18:21"} +{"current_steps": 2700, "total_steps": 5364, "loss": 0.06869709491729736, "lr": 1.1629556741370797e-05, "epoch": 1.5102097902097902, "percentage": 50.34, "elapsed_time": "5:22:32", "remaining_time": "5:18:14"} +{"current_steps": 2701, "total_steps": 5364, "loss": 0.20549604296684265, "lr": 1.1623135016512377e-05, "epoch": 1.5107692307692306, "percentage": 50.35, "elapsed_time": "5:22:39", "remaining_time": "5:18:07"} +{"current_steps": 2702, "total_steps": 5364, "loss": 0.18413308262825012, "lr": 1.1616712604111094e-05, "epoch": 1.5113286713286713, "percentage": 50.37, "elapsed_time": "5:22:46", "remaining_time": "5:17:59"} +{"current_steps": 2703, "total_steps": 5364, "loss": 0.1787792593240738, "lr": 1.1610289506887409e-05, "epoch": 1.511888111888112, "percentage": 50.39, "elapsed_time": "5:22:54", "remaining_time": "5:17:53"} +{"current_steps": 2704, "total_steps": 5364, "loss": 0.17872554063796997, "lr": 1.1603865727562088e-05, "epoch": 1.5124475524475525, "percentage": 50.41, "elapsed_time": "5:23:01", "remaining_time": "5:17:45"} +{"current_steps": 2705, "total_steps": 5364, "loss": 0.26431435346603394, "lr": 1.1597441268856163e-05, "epoch": 1.513006993006993, "percentage": 50.43, "elapsed_time": "5:23:07", "remaining_time": "5:17:38"} +{"current_steps": 2706, "total_steps": 5364, "loss": 0.10064855217933655, "lr": 1.1591016133490978e-05, "epoch": 1.5135664335664336, "percentage": 50.45, "elapsed_time": "5:23:14", "remaining_time": "5:17:30"} +{"current_steps": 2707, "total_steps": 5364, "loss": 0.1621956080198288, "lr": 1.1584590324188146e-05, "epoch": 1.514125874125874, "percentage": 50.47, "elapsed_time": "5:23:22", "remaining_time": "5:17:23"} +{"current_steps": 2708, "total_steps": 5364, "loss": 0.16317087411880493, "lr": 1.1578163843669575e-05, "epoch": 1.5146853146853148, "percentage": 50.48, "elapsed_time": "5:23:29", "remaining_time": "5:17:17"} +{"current_steps": 2709, "total_steps": 5364, "loss": 0.18540197610855103, "lr": 1.157173669465745e-05, "epoch": 1.5152447552447552, "percentage": 50.5, "elapsed_time": "5:23:36", "remaining_time": "5:17:09"} +{"current_steps": 2710, "total_steps": 5364, "loss": 0.08340314030647278, "lr": 1.1565308879874249e-05, "epoch": 1.5158041958041957, "percentage": 50.52, "elapsed_time": "5:23:44", "remaining_time": "5:17:02"} +{"current_steps": 2711, "total_steps": 5364, "loss": 0.1868424415588379, "lr": 1.1558880402042725e-05, "epoch": 1.5163636363636364, "percentage": 50.54, "elapsed_time": "5:23:50", "remaining_time": "5:16:55"} +{"current_steps": 2712, "total_steps": 5364, "loss": 0.0291266031563282, "lr": 1.155245126388591e-05, "epoch": 1.516923076923077, "percentage": 50.56, "elapsed_time": "5:23:58", "remaining_time": "5:16:48"} +{"current_steps": 2713, "total_steps": 5364, "loss": 0.1892201006412506, "lr": 1.1546021468127127e-05, "epoch": 1.5174825174825175, "percentage": 50.58, "elapsed_time": "5:24:05", "remaining_time": "5:16:41"} +{"current_steps": 2714, "total_steps": 5364, "loss": 0.07278981059789658, "lr": 1.1539591017489962e-05, "epoch": 1.518041958041958, "percentage": 50.6, "elapsed_time": "5:24:12", "remaining_time": "5:16:33"} +{"current_steps": 2715, "total_steps": 5364, "loss": 0.13239984214305878, "lr": 1.1533159914698287e-05, "epoch": 1.5186013986013986, "percentage": 50.62, "elapsed_time": "5:24:19", "remaining_time": "5:16:26"} +{"current_steps": 2716, "total_steps": 5364, "loss": 0.14451396465301514, "lr": 1.152672816247625e-05, "epoch": 1.5191608391608393, "percentage": 50.63, "elapsed_time": "5:24:26", "remaining_time": "5:16:19"} +{"current_steps": 2717, "total_steps": 5364, "loss": 0.12178383767604828, "lr": 1.1520295763548279e-05, "epoch": 1.5197202797202798, "percentage": 50.65, "elapsed_time": "5:24:34", "remaining_time": "5:16:12"} +{"current_steps": 2718, "total_steps": 5364, "loss": 0.1048208475112915, "lr": 1.1513862720639061e-05, "epoch": 1.5202797202797202, "percentage": 50.67, "elapsed_time": "5:24:41", "remaining_time": "5:16:05"} +{"current_steps": 2719, "total_steps": 5364, "loss": 0.05339743196964264, "lr": 1.1507429036473573e-05, "epoch": 1.5208391608391607, "percentage": 50.69, "elapsed_time": "5:24:48", "remaining_time": "5:15:57"} +{"current_steps": 2720, "total_steps": 5364, "loss": 0.14586740732192993, "lr": 1.1500994713777047e-05, "epoch": 1.5213986013986014, "percentage": 50.71, "elapsed_time": "5:24:55", "remaining_time": "5:15:50"} +{"current_steps": 2721, "total_steps": 5364, "loss": 0.10054674744606018, "lr": 1.1494559755275003e-05, "epoch": 1.521958041958042, "percentage": 50.73, "elapsed_time": "5:25:01", "remaining_time": "5:15:42"} +{"current_steps": 2722, "total_steps": 5364, "loss": 0.05900286138057709, "lr": 1.1488124163693217e-05, "epoch": 1.5225174825174825, "percentage": 50.75, "elapsed_time": "5:25:09", "remaining_time": "5:15:35"} +{"current_steps": 2723, "total_steps": 5364, "loss": 0.16780465841293335, "lr": 1.1481687941757743e-05, "epoch": 1.523076923076923, "percentage": 50.76, "elapsed_time": "5:25:16", "remaining_time": "5:15:28"} +{"current_steps": 2724, "total_steps": 5364, "loss": 0.15541806817054749, "lr": 1.1475251092194885e-05, "epoch": 1.5236363636363637, "percentage": 50.78, "elapsed_time": "5:25:23", "remaining_time": "5:15:21"} +{"current_steps": 2725, "total_steps": 5364, "loss": 0.2158806324005127, "lr": 1.1468813617731237e-05, "epoch": 1.5241958041958044, "percentage": 50.8, "elapsed_time": "5:25:30", "remaining_time": "5:15:13"} +{"current_steps": 2726, "total_steps": 5364, "loss": 0.11658114194869995, "lr": 1.1462375521093638e-05, "epoch": 1.5247552447552448, "percentage": 50.82, "elapsed_time": "5:25:37", "remaining_time": "5:15:06"} +{"current_steps": 2727, "total_steps": 5364, "loss": 0.0538557805120945, "lr": 1.14559368050092e-05, "epoch": 1.5253146853146853, "percentage": 50.84, "elapsed_time": "5:25:44", "remaining_time": "5:14:59"} +{"current_steps": 2728, "total_steps": 5364, "loss": 0.1075274795293808, "lr": 1.1449497472205296e-05, "epoch": 1.5258741258741257, "percentage": 50.86, "elapsed_time": "5:25:51", "remaining_time": "5:14:52"} +{"current_steps": 2729, "total_steps": 5364, "loss": 0.10640497505664825, "lr": 1.1443057525409558e-05, "epoch": 1.5264335664335664, "percentage": 50.88, "elapsed_time": "5:25:58", "remaining_time": "5:14:44"} +{"current_steps": 2730, "total_steps": 5364, "loss": 0.08769455552101135, "lr": 1.143661696734988e-05, "epoch": 1.526993006993007, "percentage": 50.89, "elapsed_time": "5:26:06", "remaining_time": "5:14:38"} +{"current_steps": 2731, "total_steps": 5364, "loss": 0.22475150227546692, "lr": 1.1430175800754416e-05, "epoch": 1.5275524475524476, "percentage": 50.91, "elapsed_time": "5:26:13", "remaining_time": "5:14:30"} +{"current_steps": 2732, "total_steps": 5364, "loss": 0.14381659030914307, "lr": 1.1423734028351572e-05, "epoch": 1.528111888111888, "percentage": 50.93, "elapsed_time": "5:26:20", "remaining_time": "5:14:23"} +{"current_steps": 2733, "total_steps": 5364, "loss": 0.17542880773544312, "lr": 1.1417291652870018e-05, "epoch": 1.5286713286713287, "percentage": 50.95, "elapsed_time": "5:26:27", "remaining_time": "5:14:16"} +{"current_steps": 2734, "total_steps": 5364, "loss": 0.07394293695688248, "lr": 1.1410848677038677e-05, "epoch": 1.5292307692307694, "percentage": 50.97, "elapsed_time": "5:26:34", "remaining_time": "5:14:09"} +{"current_steps": 2735, "total_steps": 5364, "loss": 0.15971913933753967, "lr": 1.1404405103586726e-05, "epoch": 1.5297902097902099, "percentage": 50.99, "elapsed_time": "5:26:41", "remaining_time": "5:14:02"} +{"current_steps": 2736, "total_steps": 5364, "loss": 0.23908200860023499, "lr": 1.139796093524359e-05, "epoch": 1.5303496503496503, "percentage": 51.01, "elapsed_time": "5:26:48", "remaining_time": "5:13:54"} +{"current_steps": 2737, "total_steps": 5364, "loss": 0.1496146023273468, "lr": 1.1391516174738953e-05, "epoch": 1.5309090909090908, "percentage": 51.03, "elapsed_time": "5:26:55", "remaining_time": "5:13:47"} +{"current_steps": 2738, "total_steps": 5364, "loss": 0.23367974162101746, "lr": 1.1385070824802749e-05, "epoch": 1.5314685314685315, "percentage": 51.04, "elapsed_time": "5:27:03", "remaining_time": "5:13:40"} +{"current_steps": 2739, "total_steps": 5364, "loss": 0.2965750992298126, "lr": 1.1378624888165162e-05, "epoch": 1.5320279720279721, "percentage": 51.06, "elapsed_time": "5:27:10", "remaining_time": "5:13:33"} +{"current_steps": 2740, "total_steps": 5364, "loss": 0.10041292011737823, "lr": 1.1372178367556621e-05, "epoch": 1.5325874125874126, "percentage": 51.08, "elapsed_time": "5:27:17", "remaining_time": "5:13:26"} +{"current_steps": 2741, "total_steps": 5364, "loss": 0.24014243483543396, "lr": 1.13657312657078e-05, "epoch": 1.533146853146853, "percentage": 51.1, "elapsed_time": "5:27:25", "remaining_time": "5:13:19"} +{"current_steps": 2742, "total_steps": 5364, "loss": 0.18213319778442383, "lr": 1.1359283585349628e-05, "epoch": 1.5337062937062937, "percentage": 51.12, "elapsed_time": "5:27:32", "remaining_time": "5:13:12"} +{"current_steps": 2743, "total_steps": 5364, "loss": 0.28087472915649414, "lr": 1.1352835329213272e-05, "epoch": 1.5342657342657344, "percentage": 51.14, "elapsed_time": "5:27:39", "remaining_time": "5:13:04"} +{"current_steps": 2744, "total_steps": 5364, "loss": 0.15751951932907104, "lr": 1.1346386500030146e-05, "epoch": 1.5348251748251749, "percentage": 51.16, "elapsed_time": "5:27:46", "remaining_time": "5:12:57"} +{"current_steps": 2745, "total_steps": 5364, "loss": 0.10551661252975464, "lr": 1.133993710053191e-05, "epoch": 1.5353846153846153, "percentage": 51.17, "elapsed_time": "5:27:53", "remaining_time": "5:12:50"} +{"current_steps": 2746, "total_steps": 5364, "loss": 0.07124066352844238, "lr": 1.1333487133450454e-05, "epoch": 1.5359440559440558, "percentage": 51.19, "elapsed_time": "5:28:01", "remaining_time": "5:12:43"} +{"current_steps": 2747, "total_steps": 5364, "loss": 0.05576277896761894, "lr": 1.132703660151792e-05, "epoch": 1.5365034965034965, "percentage": 51.21, "elapsed_time": "5:28:07", "remaining_time": "5:12:36"} +{"current_steps": 2748, "total_steps": 5364, "loss": 0.1267869472503662, "lr": 1.1320585507466689e-05, "epoch": 1.5370629370629372, "percentage": 51.23, "elapsed_time": "5:28:15", "remaining_time": "5:12:29"} +{"current_steps": 2749, "total_steps": 5364, "loss": 0.2700503170490265, "lr": 1.1314133854029367e-05, "epoch": 1.5376223776223776, "percentage": 51.25, "elapsed_time": "5:28:22", "remaining_time": "5:12:21"} +{"current_steps": 2750, "total_steps": 5364, "loss": 0.047837212681770325, "lr": 1.1307681643938816e-05, "epoch": 1.538181818181818, "percentage": 51.27, "elapsed_time": "5:28:29", "remaining_time": "5:12:14"} +{"current_steps": 2751, "total_steps": 5364, "loss": 0.2361563742160797, "lr": 1.1301228879928118e-05, "epoch": 1.5387412587412588, "percentage": 51.29, "elapsed_time": "5:28:36", "remaining_time": "5:12:07"} +{"current_steps": 2752, "total_steps": 5364, "loss": 0.11354345828294754, "lr": 1.1294775564730599e-05, "epoch": 1.5393006993006995, "percentage": 51.3, "elapsed_time": "5:28:42", "remaining_time": "5:11:59"} +{"current_steps": 2753, "total_steps": 5364, "loss": 0.10136795043945312, "lr": 1.1288321701079815e-05, "epoch": 1.53986013986014, "percentage": 51.32, "elapsed_time": "5:28:50", "remaining_time": "5:11:52"} +{"current_steps": 2754, "total_steps": 5364, "loss": 0.16819968819618225, "lr": 1.1281867291709549e-05, "epoch": 1.5404195804195804, "percentage": 51.34, "elapsed_time": "5:28:57", "remaining_time": "5:11:45"} +{"current_steps": 2755, "total_steps": 5364, "loss": 0.12270970642566681, "lr": 1.1275412339353826e-05, "epoch": 1.5409790209790208, "percentage": 51.36, "elapsed_time": "5:29:04", "remaining_time": "5:11:37"} +{"current_steps": 2756, "total_steps": 5364, "loss": 0.2946152687072754, "lr": 1.1268956846746897e-05, "epoch": 1.5415384615384615, "percentage": 51.38, "elapsed_time": "5:29:11", "remaining_time": "5:11:30"} +{"current_steps": 2757, "total_steps": 5364, "loss": 0.06950849294662476, "lr": 1.1262500816623239e-05, "epoch": 1.5420979020979022, "percentage": 51.4, "elapsed_time": "5:29:19", "remaining_time": "5:11:24"} +{"current_steps": 2758, "total_steps": 5364, "loss": 0.1372474581003189, "lr": 1.1256044251717555e-05, "epoch": 1.5426573426573427, "percentage": 51.42, "elapsed_time": "5:29:25", "remaining_time": "5:11:16"} +{"current_steps": 2759, "total_steps": 5364, "loss": 0.05484132841229439, "lr": 1.124958715476478e-05, "epoch": 1.5432167832167831, "percentage": 51.44, "elapsed_time": "5:29:33", "remaining_time": "5:11:09"} +{"current_steps": 2760, "total_steps": 5364, "loss": 0.09418675303459167, "lr": 1.1243129528500073e-05, "epoch": 1.5437762237762238, "percentage": 51.45, "elapsed_time": "5:29:40", "remaining_time": "5:11:02"} +{"current_steps": 2761, "total_steps": 5364, "loss": 0.15810918807983398, "lr": 1.123667137565882e-05, "epoch": 1.5443356643356645, "percentage": 51.47, "elapsed_time": "5:29:47", "remaining_time": "5:10:55"} +{"current_steps": 2762, "total_steps": 5364, "loss": 0.12960809469223022, "lr": 1.1230212698976619e-05, "epoch": 1.544895104895105, "percentage": 51.49, "elapsed_time": "5:29:54", "remaining_time": "5:10:47"} +{"current_steps": 2763, "total_steps": 5364, "loss": 0.17287057638168335, "lr": 1.12237535011893e-05, "epoch": 1.5454545454545454, "percentage": 51.51, "elapsed_time": "5:30:01", "remaining_time": "5:10:40"} +{"current_steps": 2764, "total_steps": 5364, "loss": 0.11621387302875519, "lr": 1.1217293785032911e-05, "epoch": 1.5460139860139859, "percentage": 51.53, "elapsed_time": "5:30:08", "remaining_time": "5:10:33"} +{"current_steps": 2765, "total_steps": 5364, "loss": 0.07183268666267395, "lr": 1.1210833553243721e-05, "epoch": 1.5465734265734266, "percentage": 51.55, "elapsed_time": "5:30:16", "remaining_time": "5:10:26"} +{"current_steps": 2766, "total_steps": 5364, "loss": 0.2578980624675751, "lr": 1.1204372808558218e-05, "epoch": 1.5471328671328672, "percentage": 51.57, "elapsed_time": "5:30:23", "remaining_time": "5:10:19"} +{"current_steps": 2767, "total_steps": 5364, "loss": 0.18708135187625885, "lr": 1.1197911553713098e-05, "epoch": 1.5476923076923077, "percentage": 51.58, "elapsed_time": "5:30:30", "remaining_time": "5:10:11"} +{"current_steps": 2768, "total_steps": 5364, "loss": 0.15511882305145264, "lr": 1.1191449791445285e-05, "epoch": 1.5482517482517482, "percentage": 51.6, "elapsed_time": "5:30:37", "remaining_time": "5:10:04"} +{"current_steps": 2769, "total_steps": 5364, "loss": 0.17547817528247833, "lr": 1.1184987524491915e-05, "epoch": 1.5488111888111888, "percentage": 51.62, "elapsed_time": "5:30:44", "remaining_time": "5:09:57"} +{"current_steps": 2770, "total_steps": 5364, "loss": 0.1429714411497116, "lr": 1.1178524755590334e-05, "epoch": 1.5493706293706295, "percentage": 51.64, "elapsed_time": "5:30:51", "remaining_time": "5:09:50"} +{"current_steps": 2771, "total_steps": 5364, "loss": 0.0978471115231514, "lr": 1.1172061487478103e-05, "epoch": 1.54993006993007, "percentage": 51.66, "elapsed_time": "5:30:58", "remaining_time": "5:09:43"} +{"current_steps": 2772, "total_steps": 5364, "loss": 0.1880197525024414, "lr": 1.1165597722892992e-05, "epoch": 1.5504895104895104, "percentage": 51.68, "elapsed_time": "5:31:05", "remaining_time": "5:09:35"} +{"current_steps": 2773, "total_steps": 5364, "loss": 0.1662192940711975, "lr": 1.1159133464572989e-05, "epoch": 1.551048951048951, "percentage": 51.7, "elapsed_time": "5:31:12", "remaining_time": "5:09:28"} +{"current_steps": 2774, "total_steps": 5364, "loss": 0.14673921465873718, "lr": 1.115266871525628e-05, "epoch": 1.5516083916083916, "percentage": 51.72, "elapsed_time": "5:31:19", "remaining_time": "5:09:21"} +{"current_steps": 2775, "total_steps": 5364, "loss": 0.33482736349105835, "lr": 1.1146203477681268e-05, "epoch": 1.5521678321678323, "percentage": 51.73, "elapsed_time": "5:31:26", "remaining_time": "5:09:13"} +{"current_steps": 2776, "total_steps": 5364, "loss": 0.20802396535873413, "lr": 1.1139737754586554e-05, "epoch": 1.5527272727272727, "percentage": 51.75, "elapsed_time": "5:31:33", "remaining_time": "5:09:06"} +{"current_steps": 2777, "total_steps": 5364, "loss": 0.2128729522228241, "lr": 1.1133271548710958e-05, "epoch": 1.5532867132867132, "percentage": 51.77, "elapsed_time": "5:31:40", "remaining_time": "5:08:59"} +{"current_steps": 2778, "total_steps": 5364, "loss": 0.2423178255558014, "lr": 1.112680486279349e-05, "epoch": 1.5538461538461539, "percentage": 51.79, "elapsed_time": "5:31:47", "remaining_time": "5:08:51"} +{"current_steps": 2779, "total_steps": 5364, "loss": 0.09590500593185425, "lr": 1.1120337699573378e-05, "epoch": 1.5544055944055946, "percentage": 51.81, "elapsed_time": "5:31:55", "remaining_time": "5:08:44"} +{"current_steps": 2780, "total_steps": 5364, "loss": 0.2182995080947876, "lr": 1.1113870061790035e-05, "epoch": 1.554965034965035, "percentage": 51.83, "elapsed_time": "5:32:02", "remaining_time": "5:08:37"} +{"current_steps": 2781, "total_steps": 5364, "loss": 0.09691740572452545, "lr": 1.1107401952183088e-05, "epoch": 1.5555244755244755, "percentage": 51.85, "elapsed_time": "5:32:09", "remaining_time": "5:08:30"} +{"current_steps": 2782, "total_steps": 5364, "loss": 0.24449270963668823, "lr": 1.110093337349236e-05, "epoch": 1.556083916083916, "percentage": 51.86, "elapsed_time": "5:32:16", "remaining_time": "5:08:23"} +{"current_steps": 2783, "total_steps": 5364, "loss": 0.14741364121437073, "lr": 1.1094464328457873e-05, "epoch": 1.5566433566433566, "percentage": 51.88, "elapsed_time": "5:32:24", "remaining_time": "5:08:16"} +{"current_steps": 2784, "total_steps": 5364, "loss": 0.17428237199783325, "lr": 1.1087994819819847e-05, "epoch": 1.5572027972027973, "percentage": 51.9, "elapsed_time": "5:32:31", "remaining_time": "5:08:09"} +{"current_steps": 2785, "total_steps": 5364, "loss": 0.19588162004947662, "lr": 1.1081524850318694e-05, "epoch": 1.5577622377622378, "percentage": 51.92, "elapsed_time": "5:32:39", "remaining_time": "5:08:03"} +{"current_steps": 2786, "total_steps": 5364, "loss": 0.09360744059085846, "lr": 1.1075054422695027e-05, "epoch": 1.5583216783216782, "percentage": 51.94, "elapsed_time": "5:32:47", "remaining_time": "5:07:56"} +{"current_steps": 2787, "total_steps": 5364, "loss": 0.258054256439209, "lr": 1.1068583539689652e-05, "epoch": 1.558881118881119, "percentage": 51.96, "elapsed_time": "5:32:53", "remaining_time": "5:07:48"} +{"current_steps": 2788, "total_steps": 5364, "loss": 0.05555944889783859, "lr": 1.1062112204043567e-05, "epoch": 1.5594405594405596, "percentage": 51.98, "elapsed_time": "5:33:01", "remaining_time": "5:07:41"} +{"current_steps": 2789, "total_steps": 5364, "loss": 0.10265368223190308, "lr": 1.1055640418497957e-05, "epoch": 1.56, "percentage": 51.99, "elapsed_time": "5:33:08", "remaining_time": "5:07:34"} +{"current_steps": 2790, "total_steps": 5364, "loss": 0.1817026436328888, "lr": 1.104916818579421e-05, "epoch": 1.5605594405594405, "percentage": 52.01, "elapsed_time": "5:33:15", "remaining_time": "5:07:27"} +{"current_steps": 2791, "total_steps": 5364, "loss": 0.22906853258609772, "lr": 1.1042695508673886e-05, "epoch": 1.561118881118881, "percentage": 52.03, "elapsed_time": "5:33:22", "remaining_time": "5:07:20"} +{"current_steps": 2792, "total_steps": 5364, "loss": 0.10952460765838623, "lr": 1.1036222389878752e-05, "epoch": 1.5616783216783217, "percentage": 52.05, "elapsed_time": "5:33:29", "remaining_time": "5:07:13"} +{"current_steps": 2793, "total_steps": 5364, "loss": 0.17642460763454437, "lr": 1.1029748832150746e-05, "epoch": 1.5622377622377623, "percentage": 52.07, "elapsed_time": "5:33:37", "remaining_time": "5:07:06"} +{"current_steps": 2794, "total_steps": 5364, "loss": 0.13231033086776733, "lr": 1.1023274838232e-05, "epoch": 1.5627972027972028, "percentage": 52.09, "elapsed_time": "5:33:44", "remaining_time": "5:06:59"} +{"current_steps": 2795, "total_steps": 5364, "loss": 0.13153162598609924, "lr": 1.1016800410864831e-05, "epoch": 1.5633566433566433, "percentage": 52.11, "elapsed_time": "5:33:51", "remaining_time": "5:06:52"} +{"current_steps": 2796, "total_steps": 5364, "loss": 0.014128871262073517, "lr": 1.1010325552791738e-05, "epoch": 1.563916083916084, "percentage": 52.13, "elapsed_time": "5:33:59", "remaining_time": "5:06:45"} +{"current_steps": 2797, "total_steps": 5364, "loss": 0.12916907668113708, "lr": 1.1003850266755405e-05, "epoch": 1.5644755244755246, "percentage": 52.14, "elapsed_time": "5:34:06", "remaining_time": "5:06:38"} +{"current_steps": 2798, "total_steps": 5364, "loss": 0.11345451325178146, "lr": 1.099737455549869e-05, "epoch": 1.565034965034965, "percentage": 52.16, "elapsed_time": "5:34:13", "remaining_time": "5:06:30"} +{"current_steps": 2799, "total_steps": 5364, "loss": 0.053648822009563446, "lr": 1.0990898421764638e-05, "epoch": 1.5655944055944055, "percentage": 52.18, "elapsed_time": "5:34:20", "remaining_time": "5:06:23"} +{"current_steps": 2800, "total_steps": 5364, "loss": 0.10897266864776611, "lr": 1.098442186829647e-05, "epoch": 1.566153846153846, "percentage": 52.2, "elapsed_time": "5:34:27", "remaining_time": "5:06:16"} +{"current_steps": 2801, "total_steps": 5364, "loss": 0.07249344885349274, "lr": 1.0977944897837586e-05, "epoch": 1.5667132867132867, "percentage": 52.22, "elapsed_time": "5:34:34", "remaining_time": "5:06:08"} +{"current_steps": 2802, "total_steps": 5364, "loss": 0.04046345129609108, "lr": 1.0971467513131568e-05, "epoch": 1.5672727272727274, "percentage": 52.24, "elapsed_time": "5:34:41", "remaining_time": "5:06:01"} +{"current_steps": 2803, "total_steps": 5364, "loss": 0.08685645461082458, "lr": 1.0964989716922158e-05, "epoch": 1.5678321678321678, "percentage": 52.26, "elapsed_time": "5:34:49", "remaining_time": "5:05:54"} +{"current_steps": 2804, "total_steps": 5364, "loss": 0.14584368467330933, "lr": 1.0958511511953287e-05, "epoch": 1.5683916083916083, "percentage": 52.27, "elapsed_time": "5:34:56", "remaining_time": "5:05:47"} +{"current_steps": 2805, "total_steps": 5364, "loss": 0.2007332146167755, "lr": 1.0952032900969056e-05, "epoch": 1.568951048951049, "percentage": 52.29, "elapsed_time": "5:35:03", "remaining_time": "5:05:40"} +{"current_steps": 2806, "total_steps": 5364, "loss": 0.31702375411987305, "lr": 1.0945553886713734e-05, "epoch": 1.5695104895104897, "percentage": 52.31, "elapsed_time": "5:35:10", "remaining_time": "5:05:32"} +{"current_steps": 2807, "total_steps": 5364, "loss": 0.11963526904582977, "lr": 1.0939074471931766e-05, "epoch": 1.5700699300699301, "percentage": 52.33, "elapsed_time": "5:35:17", "remaining_time": "5:05:25"} +{"current_steps": 2808, "total_steps": 5364, "loss": 0.0659194216132164, "lr": 1.0932594659367758e-05, "epoch": 1.5706293706293706, "percentage": 52.35, "elapsed_time": "5:35:24", "remaining_time": "5:05:18"} +{"current_steps": 2809, "total_steps": 5364, "loss": 0.32758593559265137, "lr": 1.0926114451766491e-05, "epoch": 1.571188811188811, "percentage": 52.37, "elapsed_time": "5:35:31", "remaining_time": "5:05:11"} +{"current_steps": 2810, "total_steps": 5364, "loss": 0.31249523162841797, "lr": 1.0919633851872922e-05, "epoch": 1.5717482517482517, "percentage": 52.39, "elapsed_time": "5:35:38", "remaining_time": "5:05:03"} +{"current_steps": 2811, "total_steps": 5364, "loss": 0.13023559749126434, "lr": 1.0913152862432153e-05, "epoch": 1.5723076923076924, "percentage": 52.4, "elapsed_time": "5:35:45", "remaining_time": "5:04:56"} +{"current_steps": 2812, "total_steps": 5364, "loss": 0.1404387652873993, "lr": 1.0906671486189469e-05, "epoch": 1.5728671328671329, "percentage": 52.42, "elapsed_time": "5:35:52", "remaining_time": "5:04:49"} +{"current_steps": 2813, "total_steps": 5364, "loss": 0.09957358986139297, "lr": 1.090018972589031e-05, "epoch": 1.5734265734265733, "percentage": 52.44, "elapsed_time": "5:36:00", "remaining_time": "5:04:42"} +{"current_steps": 2814, "total_steps": 5364, "loss": 0.2776985168457031, "lr": 1.0893707584280282e-05, "epoch": 1.573986013986014, "percentage": 52.46, "elapsed_time": "5:36:07", "remaining_time": "5:04:35"} +{"current_steps": 2815, "total_steps": 5364, "loss": 0.0713958889245987, "lr": 1.0887225064105154e-05, "epoch": 1.5745454545454547, "percentage": 52.48, "elapsed_time": "5:36:14", "remaining_time": "5:04:28"} +{"current_steps": 2816, "total_steps": 5364, "loss": 0.188124418258667, "lr": 1.088074216811085e-05, "epoch": 1.5751048951048952, "percentage": 52.5, "elapsed_time": "5:36:21", "remaining_time": "5:04:21"} +{"current_steps": 2817, "total_steps": 5364, "loss": 0.18631935119628906, "lr": 1.0874258899043456e-05, "epoch": 1.5756643356643356, "percentage": 52.52, "elapsed_time": "5:36:28", "remaining_time": "5:04:13"} +{"current_steps": 2818, "total_steps": 5364, "loss": 0.19625741243362427, "lr": 1.086777525964922e-05, "epoch": 1.576223776223776, "percentage": 52.54, "elapsed_time": "5:36:35", "remaining_time": "5:04:06"} +{"current_steps": 2819, "total_steps": 5364, "loss": 0.11994132399559021, "lr": 1.0861291252674544e-05, "epoch": 1.5767832167832168, "percentage": 52.55, "elapsed_time": "5:36:42", "remaining_time": "5:03:58"} +{"current_steps": 2820, "total_steps": 5364, "loss": 0.26735368371009827, "lr": 1.0854806880865977e-05, "epoch": 1.5773426573426574, "percentage": 52.57, "elapsed_time": "5:36:49", "remaining_time": "5:03:51"} +{"current_steps": 2821, "total_steps": 5364, "loss": 0.37379568815231323, "lr": 1.0848322146970238e-05, "epoch": 1.577902097902098, "percentage": 52.59, "elapsed_time": "5:36:56", "remaining_time": "5:03:44"} +{"current_steps": 2822, "total_steps": 5364, "loss": 0.2209223508834839, "lr": 1.0841837053734188e-05, "epoch": 1.5784615384615384, "percentage": 52.61, "elapsed_time": "5:37:03", "remaining_time": "5:03:37"} +{"current_steps": 2823, "total_steps": 5364, "loss": 0.05732685327529907, "lr": 1.0835351603904848e-05, "epoch": 1.579020979020979, "percentage": 52.63, "elapsed_time": "5:37:10", "remaining_time": "5:03:29"} +{"current_steps": 2824, "total_steps": 5364, "loss": 0.21869157254695892, "lr": 1.0828865800229383e-05, "epoch": 1.5795804195804197, "percentage": 52.65, "elapsed_time": "5:37:17", "remaining_time": "5:03:22"} +{"current_steps": 2825, "total_steps": 5364, "loss": 0.081295907497406, "lr": 1.082237964545511e-05, "epoch": 1.5801398601398602, "percentage": 52.67, "elapsed_time": "5:37:25", "remaining_time": "5:03:15"} +{"current_steps": 2826, "total_steps": 5364, "loss": 0.13505405187606812, "lr": 1.0815893142329496e-05, "epoch": 1.5806993006993006, "percentage": 52.68, "elapsed_time": "5:37:32", "remaining_time": "5:03:08"} +{"current_steps": 2827, "total_steps": 5364, "loss": 0.016825826838612556, "lr": 1.0809406293600158e-05, "epoch": 1.581258741258741, "percentage": 52.7, "elapsed_time": "5:37:39", "remaining_time": "5:03:01"} +{"current_steps": 2828, "total_steps": 5364, "loss": 0.08634943515062332, "lr": 1.0802919102014857e-05, "epoch": 1.5818181818181818, "percentage": 52.72, "elapsed_time": "5:37:46", "remaining_time": "5:02:53"} +{"current_steps": 2829, "total_steps": 5364, "loss": 0.023347143083810806, "lr": 1.07964315703215e-05, "epoch": 1.5823776223776225, "percentage": 52.74, "elapsed_time": "5:37:53", "remaining_time": "5:02:47"} +{"current_steps": 2830, "total_steps": 5364, "loss": 0.11973976343870163, "lr": 1.0789943701268129e-05, "epoch": 1.582937062937063, "percentage": 52.76, "elapsed_time": "5:38:01", "remaining_time": "5:02:39"} +{"current_steps": 2831, "total_steps": 5364, "loss": 0.08013211190700531, "lr": 1.0783455497602942e-05, "epoch": 1.5834965034965034, "percentage": 52.78, "elapsed_time": "5:38:08", "remaining_time": "5:02:32"} +{"current_steps": 2832, "total_steps": 5364, "loss": 0.10090242326259613, "lr": 1.0776966962074276e-05, "epoch": 1.584055944055944, "percentage": 52.8, "elapsed_time": "5:38:15", "remaining_time": "5:02:25"} +{"current_steps": 2833, "total_steps": 5364, "loss": 0.09611985832452774, "lr": 1.0770478097430601e-05, "epoch": 1.5846153846153848, "percentage": 52.82, "elapsed_time": "5:38:23", "remaining_time": "5:02:18"} +{"current_steps": 2834, "total_steps": 5364, "loss": 0.04323085770010948, "lr": 1.0763988906420536e-05, "epoch": 1.5851748251748252, "percentage": 52.83, "elapsed_time": "5:38:30", "remaining_time": "5:02:11"} +{"current_steps": 2835, "total_steps": 5364, "loss": 0.07391176372766495, "lr": 1.0757499391792832e-05, "epoch": 1.5857342657342657, "percentage": 52.85, "elapsed_time": "5:38:37", "remaining_time": "5:02:04"} +{"current_steps": 2836, "total_steps": 5364, "loss": 0.2970464527606964, "lr": 1.075100955629638e-05, "epoch": 1.5862937062937061, "percentage": 52.87, "elapsed_time": "5:38:44", "remaining_time": "5:01:57"} +{"current_steps": 2837, "total_steps": 5364, "loss": 0.3290688991546631, "lr": 1.0744519402680204e-05, "epoch": 1.5868531468531468, "percentage": 52.89, "elapsed_time": "5:38:51", "remaining_time": "5:01:50"} +{"current_steps": 2838, "total_steps": 5364, "loss": 0.13434967398643494, "lr": 1.0738028933693465e-05, "epoch": 1.5874125874125875, "percentage": 52.91, "elapsed_time": "5:38:59", "remaining_time": "5:01:43"} +{"current_steps": 2839, "total_steps": 5364, "loss": 0.23966121673583984, "lr": 1.073153815208546e-05, "epoch": 1.587972027972028, "percentage": 52.93, "elapsed_time": "5:39:06", "remaining_time": "5:01:35"} +{"current_steps": 2840, "total_steps": 5364, "loss": 0.34113994240760803, "lr": 1.072504706060561e-05, "epoch": 1.5885314685314684, "percentage": 52.95, "elapsed_time": "5:39:13", "remaining_time": "5:01:28"} +{"current_steps": 2841, "total_steps": 5364, "loss": 0.22892752289772034, "lr": 1.071855566200348e-05, "epoch": 1.589090909090909, "percentage": 52.96, "elapsed_time": "5:39:20", "remaining_time": "5:01:21"} +{"current_steps": 2842, "total_steps": 5364, "loss": 0.21259471774101257, "lr": 1.071206395902875e-05, "epoch": 1.5896503496503498, "percentage": 52.98, "elapsed_time": "5:39:27", "remaining_time": "5:01:14"} +{"current_steps": 2843, "total_steps": 5364, "loss": 0.1700703352689743, "lr": 1.0705571954431245e-05, "epoch": 1.5902097902097903, "percentage": 53.0, "elapsed_time": "5:39:34", "remaining_time": "5:01:07"} +{"current_steps": 2844, "total_steps": 5364, "loss": 0.12835678458213806, "lr": 1.0699079650960902e-05, "epoch": 1.5907692307692307, "percentage": 53.02, "elapsed_time": "5:39:42", "remaining_time": "5:01:00"} +{"current_steps": 2845, "total_steps": 5364, "loss": 0.136978417634964, "lr": 1.0692587051367799e-05, "epoch": 1.5913286713286712, "percentage": 53.04, "elapsed_time": "5:39:49", "remaining_time": "5:00:53"} +{"current_steps": 2846, "total_steps": 5364, "loss": 0.14044445753097534, "lr": 1.068609415840213e-05, "epoch": 1.5918881118881119, "percentage": 53.06, "elapsed_time": "5:39:56", "remaining_time": "5:00:45"} +{"current_steps": 2847, "total_steps": 5364, "loss": 0.013831639662384987, "lr": 1.0679600974814212e-05, "epoch": 1.5924475524475525, "percentage": 53.08, "elapsed_time": "5:40:03", "remaining_time": "5:00:38"} +{"current_steps": 2848, "total_steps": 5364, "loss": 0.1017160415649414, "lr": 1.0673107503354494e-05, "epoch": 1.593006993006993, "percentage": 53.09, "elapsed_time": "5:40:10", "remaining_time": "5:00:31"} +{"current_steps": 2849, "total_steps": 5364, "loss": 0.20781946182250977, "lr": 1.066661374677354e-05, "epoch": 1.5935664335664335, "percentage": 53.11, "elapsed_time": "5:40:17", "remaining_time": "5:00:24"} +{"current_steps": 2850, "total_steps": 5364, "loss": 0.006650528870522976, "lr": 1.0660119707822039e-05, "epoch": 1.5941258741258741, "percentage": 53.13, "elapsed_time": "5:40:25", "remaining_time": "5:00:17"} +{"current_steps": 2851, "total_steps": 5364, "loss": 0.17892834544181824, "lr": 1.0653625389250795e-05, "epoch": 1.5946853146853148, "percentage": 53.15, "elapsed_time": "5:40:31", "remaining_time": "5:00:09"} +{"current_steps": 2852, "total_steps": 5364, "loss": 0.030810000374913216, "lr": 1.0647130793810732e-05, "epoch": 1.5952447552447553, "percentage": 53.17, "elapsed_time": "5:40:39", "remaining_time": "5:00:02"} +{"current_steps": 2853, "total_steps": 5364, "loss": 0.14324603974819183, "lr": 1.0640635924252894e-05, "epoch": 1.5958041958041957, "percentage": 53.19, "elapsed_time": "5:40:46", "remaining_time": "4:59:55"} +{"current_steps": 2854, "total_steps": 5364, "loss": 0.12809765338897705, "lr": 1.0634140783328441e-05, "epoch": 1.5963636363636362, "percentage": 53.21, "elapsed_time": "5:40:53", "remaining_time": "4:59:48"} +{"current_steps": 2855, "total_steps": 5364, "loss": 0.27224209904670715, "lr": 1.0627645373788643e-05, "epoch": 1.596923076923077, "percentage": 53.23, "elapsed_time": "5:41:00", "remaining_time": "4:59:41"} +{"current_steps": 2856, "total_steps": 5364, "loss": 0.08574418723583221, "lr": 1.0621149698384885e-05, "epoch": 1.5974825174825176, "percentage": 53.24, "elapsed_time": "5:41:07", "remaining_time": "4:59:33"} +{"current_steps": 2857, "total_steps": 5364, "loss": 0.11137000471353531, "lr": 1.0614653759868674e-05, "epoch": 1.598041958041958, "percentage": 53.26, "elapsed_time": "5:41:14", "remaining_time": "4:59:26"} +{"current_steps": 2858, "total_steps": 5364, "loss": 0.14484480023384094, "lr": 1.0608157560991612e-05, "epoch": 1.5986013986013985, "percentage": 53.28, "elapsed_time": "5:41:21", "remaining_time": "4:59:19"} +{"current_steps": 2859, "total_steps": 5364, "loss": 0.14593040943145752, "lr": 1.0601661104505428e-05, "epoch": 1.5991608391608392, "percentage": 53.3, "elapsed_time": "5:41:29", "remaining_time": "4:59:12"} +{"current_steps": 2860, "total_steps": 5364, "loss": 0.28754502534866333, "lr": 1.0595164393161948e-05, "epoch": 1.5997202797202799, "percentage": 53.32, "elapsed_time": "5:41:36", "remaining_time": "4:59:05"} +{"current_steps": 2861, "total_steps": 5364, "loss": 0.47070586681365967, "lr": 1.0588667429713111e-05, "epoch": 1.6002797202797203, "percentage": 53.34, "elapsed_time": "5:41:43", "remaining_time": "4:58:57"} +{"current_steps": 2862, "total_steps": 5364, "loss": 0.2060789167881012, "lr": 1.0582170216910962e-05, "epoch": 1.6008391608391608, "percentage": 53.36, "elapsed_time": "5:41:50", "remaining_time": "4:58:50"} +{"current_steps": 2863, "total_steps": 5364, "loss": 0.12663966417312622, "lr": 1.0575672757507653e-05, "epoch": 1.6013986013986012, "percentage": 53.37, "elapsed_time": "5:41:57", "remaining_time": "4:58:43"} +{"current_steps": 2864, "total_steps": 5364, "loss": 0.0831083282828331, "lr": 1.0569175054255442e-05, "epoch": 1.601958041958042, "percentage": 53.39, "elapsed_time": "5:42:04", "remaining_time": "4:58:35"} +{"current_steps": 2865, "total_steps": 5364, "loss": 0.16874247789382935, "lr": 1.056267710990668e-05, "epoch": 1.6025174825174826, "percentage": 53.41, "elapsed_time": "5:42:11", "remaining_time": "4:58:29"} +{"current_steps": 2866, "total_steps": 5364, "loss": 0.1881970465183258, "lr": 1.0556178927213834e-05, "epoch": 1.603076923076923, "percentage": 53.43, "elapsed_time": "5:42:19", "remaining_time": "4:58:22"} +{"current_steps": 2867, "total_steps": 5364, "loss": 0.14702516794204712, "lr": 1.0549680508929464e-05, "epoch": 1.6036363636363635, "percentage": 53.45, "elapsed_time": "5:42:26", "remaining_time": "4:58:15"} +{"current_steps": 2868, "total_steps": 5364, "loss": 0.2145739644765854, "lr": 1.0543181857806231e-05, "epoch": 1.6041958041958042, "percentage": 53.47, "elapsed_time": "5:42:33", "remaining_time": "4:58:07"} +{"current_steps": 2869, "total_steps": 5364, "loss": 0.14220091700553894, "lr": 1.0536682976596892e-05, "epoch": 1.604755244755245, "percentage": 53.49, "elapsed_time": "5:42:40", "remaining_time": "4:58:00"} +{"current_steps": 2870, "total_steps": 5364, "loss": 0.14121350646018982, "lr": 1.0530183868054309e-05, "epoch": 1.6053146853146854, "percentage": 53.5, "elapsed_time": "5:42:47", "remaining_time": "4:57:53"} +{"current_steps": 2871, "total_steps": 5364, "loss": 0.21210911870002747, "lr": 1.0523684534931434e-05, "epoch": 1.6058741258741258, "percentage": 53.52, "elapsed_time": "5:42:54", "remaining_time": "4:57:45"} +{"current_steps": 2872, "total_steps": 5364, "loss": 0.04120757058262825, "lr": 1.0517184979981317e-05, "epoch": 1.6064335664335663, "percentage": 53.54, "elapsed_time": "5:43:01", "remaining_time": "4:57:38"} +{"current_steps": 2873, "total_steps": 5364, "loss": 0.21981248259544373, "lr": 1.0510685205957098e-05, "epoch": 1.606993006993007, "percentage": 53.56, "elapsed_time": "5:43:08", "remaining_time": "4:57:31"} +{"current_steps": 2874, "total_steps": 5364, "loss": 0.06539766490459442, "lr": 1.0504185215612017e-05, "epoch": 1.6075524475524476, "percentage": 53.58, "elapsed_time": "5:43:15", "remaining_time": "4:57:23"} +{"current_steps": 2875, "total_steps": 5364, "loss": 0.40811076760292053, "lr": 1.0497685011699399e-05, "epoch": 1.608111888111888, "percentage": 53.6, "elapsed_time": "5:43:22", "remaining_time": "4:57:16"} +{"current_steps": 2876, "total_steps": 5364, "loss": 0.14503905177116394, "lr": 1.0491184596972658e-05, "epoch": 1.6086713286713286, "percentage": 53.62, "elapsed_time": "5:43:29", "remaining_time": "4:57:08"} +{"current_steps": 2877, "total_steps": 5364, "loss": 0.23895788192749023, "lr": 1.0484683974185312e-05, "epoch": 1.6092307692307692, "percentage": 53.64, "elapsed_time": "5:43:36", "remaining_time": "4:57:01"} +{"current_steps": 2878, "total_steps": 5364, "loss": 0.06059478223323822, "lr": 1.0478183146090943e-05, "epoch": 1.60979020979021, "percentage": 53.65, "elapsed_time": "5:43:44", "remaining_time": "4:56:55"} +{"current_steps": 2879, "total_steps": 5364, "loss": 0.08490406721830368, "lr": 1.0471682115443244e-05, "epoch": 1.6103496503496504, "percentage": 53.67, "elapsed_time": "5:43:51", "remaining_time": "4:56:48"} +{"current_steps": 2880, "total_steps": 5364, "loss": 0.33345717191696167, "lr": 1.0465180884995976e-05, "epoch": 1.6109090909090908, "percentage": 53.69, "elapsed_time": "5:43:58", "remaining_time": "4:56:40"} +{"current_steps": 2881, "total_steps": 5364, "loss": 0.1357303261756897, "lr": 1.0458679457502994e-05, "epoch": 1.6114685314685313, "percentage": 53.71, "elapsed_time": "5:44:06", "remaining_time": "4:56:33"} +{"current_steps": 2882, "total_steps": 5364, "loss": 0.15024752914905548, "lr": 1.0452177835718234e-05, "epoch": 1.612027972027972, "percentage": 53.73, "elapsed_time": "5:44:13", "remaining_time": "4:56:26"} +{"current_steps": 2883, "total_steps": 5364, "loss": 0.12237966805696487, "lr": 1.0445676022395712e-05, "epoch": 1.6125874125874127, "percentage": 53.75, "elapsed_time": "5:44:20", "remaining_time": "4:56:19"} +{"current_steps": 2884, "total_steps": 5364, "loss": 0.0868656188249588, "lr": 1.0439174020289531e-05, "epoch": 1.6131468531468531, "percentage": 53.77, "elapsed_time": "5:44:27", "remaining_time": "4:56:12"} +{"current_steps": 2885, "total_steps": 5364, "loss": 0.09831293672323227, "lr": 1.0432671832153873e-05, "epoch": 1.6137062937062936, "percentage": 53.78, "elapsed_time": "5:44:34", "remaining_time": "4:56:05"} +{"current_steps": 2886, "total_steps": 5364, "loss": 0.08679448068141937, "lr": 1.0426169460742993e-05, "epoch": 1.6142657342657343, "percentage": 53.8, "elapsed_time": "5:44:42", "remaining_time": "4:55:58"} +{"current_steps": 2887, "total_steps": 5364, "loss": 0.09096908569335938, "lr": 1.041966690881122e-05, "epoch": 1.614825174825175, "percentage": 53.82, "elapsed_time": "5:44:48", "remaining_time": "4:55:50"} +{"current_steps": 2888, "total_steps": 5364, "loss": 0.21412667632102966, "lr": 1.0413164179112978e-05, "epoch": 1.6153846153846154, "percentage": 53.84, "elapsed_time": "5:44:55", "remaining_time": "4:55:43"} +{"current_steps": 2889, "total_steps": 5364, "loss": 0.1825488805770874, "lr": 1.0406661274402747e-05, "epoch": 1.6159440559440559, "percentage": 53.86, "elapsed_time": "5:45:02", "remaining_time": "4:55:36"} +{"current_steps": 2890, "total_steps": 5364, "loss": 0.02179224044084549, "lr": 1.0400158197435095e-05, "epoch": 1.6165034965034963, "percentage": 53.88, "elapsed_time": "5:45:10", "remaining_time": "4:55:29"} +{"current_steps": 2891, "total_steps": 5364, "loss": 0.16015592217445374, "lr": 1.0393654950964651e-05, "epoch": 1.617062937062937, "percentage": 53.9, "elapsed_time": "5:45:17", "remaining_time": "4:55:22"} +{"current_steps": 2892, "total_steps": 5364, "loss": 0.08906261622905731, "lr": 1.0387151537746123e-05, "epoch": 1.6176223776223777, "percentage": 53.91, "elapsed_time": "5:45:25", "remaining_time": "4:55:15"} +{"current_steps": 2893, "total_steps": 5364, "loss": 0.2815285921096802, "lr": 1.0380647960534289e-05, "epoch": 1.6181818181818182, "percentage": 53.93, "elapsed_time": "5:45:32", "remaining_time": "4:55:08"} +{"current_steps": 2894, "total_steps": 5364, "loss": 0.25456517934799194, "lr": 1.0374144222083997e-05, "epoch": 1.6187412587412586, "percentage": 53.95, "elapsed_time": "5:45:39", "remaining_time": "4:55:01"} +{"current_steps": 2895, "total_steps": 5364, "loss": 0.12548322975635529, "lr": 1.036764032515016e-05, "epoch": 1.6193006993006993, "percentage": 53.97, "elapsed_time": "5:45:46", "remaining_time": "4:54:53"} +{"current_steps": 2896, "total_steps": 5364, "loss": 0.07098907977342606, "lr": 1.0361136272487761e-05, "epoch": 1.61986013986014, "percentage": 53.99, "elapsed_time": "5:45:54", "remaining_time": "4:54:47"} +{"current_steps": 2897, "total_steps": 5364, "loss": 0.14188343286514282, "lr": 1.0354632066851848e-05, "epoch": 1.6204195804195805, "percentage": 54.01, "elapsed_time": "5:46:01", "remaining_time": "4:54:39"} +{"current_steps": 2898, "total_steps": 5364, "loss": 0.14035232365131378, "lr": 1.0348127710997534e-05, "epoch": 1.620979020979021, "percentage": 54.03, "elapsed_time": "5:46:08", "remaining_time": "4:54:32"} +{"current_steps": 2899, "total_steps": 5364, "loss": 0.24965976178646088, "lr": 1.0341623207679994e-05, "epoch": 1.6215384615384614, "percentage": 54.05, "elapsed_time": "5:46:15", "remaining_time": "4:54:25"} +{"current_steps": 2900, "total_steps": 5364, "loss": 0.22281786799430847, "lr": 1.033511855965447e-05, "epoch": 1.622097902097902, "percentage": 54.06, "elapsed_time": "5:46:22", "remaining_time": "4:54:17"} +{"current_steps": 2901, "total_steps": 5364, "loss": 0.11965949088335037, "lr": 1.0328613769676258e-05, "epoch": 1.6226573426573427, "percentage": 54.08, "elapsed_time": "5:46:29", "remaining_time": "4:54:10"} +{"current_steps": 2902, "total_steps": 5364, "loss": 0.04703614115715027, "lr": 1.0322108840500722e-05, "epoch": 1.6232167832167832, "percentage": 54.1, "elapsed_time": "5:46:36", "remaining_time": "4:54:03"} +{"current_steps": 2903, "total_steps": 5364, "loss": 0.07611198723316193, "lr": 1.0315603774883282e-05, "epoch": 1.6237762237762237, "percentage": 54.12, "elapsed_time": "5:46:43", "remaining_time": "4:53:56"} +{"current_steps": 2904, "total_steps": 5364, "loss": 0.18084417283535004, "lr": 1.0309098575579413e-05, "epoch": 1.6243356643356643, "percentage": 54.14, "elapsed_time": "5:46:50", "remaining_time": "4:53:48"} +{"current_steps": 2905, "total_steps": 5364, "loss": 0.195207878947258, "lr": 1.0302593245344646e-05, "epoch": 1.624895104895105, "percentage": 54.16, "elapsed_time": "5:46:57", "remaining_time": "4:53:41"} +{"current_steps": 2906, "total_steps": 5364, "loss": 0.14961335062980652, "lr": 1.0296087786934575e-05, "epoch": 1.6254545454545455, "percentage": 54.18, "elapsed_time": "5:47:04", "remaining_time": "4:53:34"} +{"current_steps": 2907, "total_steps": 5364, "loss": 0.27207764983177185, "lr": 1.0289582203104843e-05, "epoch": 1.626013986013986, "percentage": 54.19, "elapsed_time": "5:47:12", "remaining_time": "4:53:27"} +{"current_steps": 2908, "total_steps": 5364, "loss": 0.13873720169067383, "lr": 1.0283076496611145e-05, "epoch": 1.6265734265734266, "percentage": 54.21, "elapsed_time": "5:47:18", "remaining_time": "4:53:19"} +{"current_steps": 2909, "total_steps": 5364, "loss": 0.3017807602882385, "lr": 1.027657067020923e-05, "epoch": 1.627132867132867, "percentage": 54.23, "elapsed_time": "5:47:26", "remaining_time": "4:53:12"} +{"current_steps": 2910, "total_steps": 5364, "loss": 0.10161522030830383, "lr": 1.0270064726654897e-05, "epoch": 1.6276923076923078, "percentage": 54.25, "elapsed_time": "5:47:33", "remaining_time": "4:53:05"} +{"current_steps": 2911, "total_steps": 5364, "loss": 0.1412440687417984, "lr": 1.0263558668703995e-05, "epoch": 1.6282517482517482, "percentage": 54.27, "elapsed_time": "5:47:40", "remaining_time": "4:52:58"} +{"current_steps": 2912, "total_steps": 5364, "loss": 0.15437306463718414, "lr": 1.0257052499112425e-05, "epoch": 1.6288111888111887, "percentage": 54.29, "elapsed_time": "5:47:47", "remaining_time": "4:52:50"} +{"current_steps": 2913, "total_steps": 5364, "loss": 0.11204564571380615, "lr": 1.0250546220636128e-05, "epoch": 1.6293706293706294, "percentage": 54.31, "elapsed_time": "5:47:54", "remaining_time": "4:52:43"} +{"current_steps": 2914, "total_steps": 5364, "loss": 0.15478914976119995, "lr": 1.0244039836031094e-05, "epoch": 1.62993006993007, "percentage": 54.33, "elapsed_time": "5:48:01", "remaining_time": "4:52:36"} +{"current_steps": 2915, "total_steps": 5364, "loss": 0.13831430673599243, "lr": 1.0237533348053361e-05, "epoch": 1.6304895104895105, "percentage": 54.34, "elapsed_time": "5:48:08", "remaining_time": "4:52:29"} +{"current_steps": 2916, "total_steps": 5364, "loss": 0.03411205857992172, "lr": 1.023102675945901e-05, "epoch": 1.631048951048951, "percentage": 54.36, "elapsed_time": "5:48:15", "remaining_time": "4:52:21"} +{"current_steps": 2917, "total_steps": 5364, "loss": 0.08191650360822678, "lr": 1.022452007300416e-05, "epoch": 1.6316083916083917, "percentage": 54.38, "elapsed_time": "5:48:23", "remaining_time": "4:52:15"} +{"current_steps": 2918, "total_steps": 5364, "loss": 0.1504591703414917, "lr": 1.0218013291444979e-05, "epoch": 1.6321678321678321, "percentage": 54.4, "elapsed_time": "5:48:30", "remaining_time": "4:52:07"} +{"current_steps": 2919, "total_steps": 5364, "loss": 0.11800846457481384, "lr": 1.0211506417537666e-05, "epoch": 1.6327272727272728, "percentage": 54.42, "elapsed_time": "5:48:37", "remaining_time": "4:52:01"} +{"current_steps": 2920, "total_steps": 5364, "loss": 0.2851371765136719, "lr": 1.0204999454038466e-05, "epoch": 1.6332867132867133, "percentage": 54.44, "elapsed_time": "5:48:44", "remaining_time": "4:51:53"} +{"current_steps": 2921, "total_steps": 5364, "loss": 0.17838457226753235, "lr": 1.0198492403703662e-05, "epoch": 1.6338461538461537, "percentage": 54.46, "elapsed_time": "5:48:51", "remaining_time": "4:51:46"} +{"current_steps": 2922, "total_steps": 5364, "loss": 0.027160657569766045, "lr": 1.0191985269289569e-05, "epoch": 1.6344055944055944, "percentage": 54.47, "elapsed_time": "5:48:58", "remaining_time": "4:51:39"} +{"current_steps": 2923, "total_steps": 5364, "loss": 0.19782215356826782, "lr": 1.0185478053552543e-05, "epoch": 1.634965034965035, "percentage": 54.49, "elapsed_time": "5:49:05", "remaining_time": "4:51:31"} +{"current_steps": 2924, "total_steps": 5364, "loss": 0.09065507352352142, "lr": 1.017897075924897e-05, "epoch": 1.6355244755244756, "percentage": 54.51, "elapsed_time": "5:49:12", "remaining_time": "4:51:24"} +{"current_steps": 2925, "total_steps": 5364, "loss": 0.15299364924430847, "lr": 1.0172463389135273e-05, "epoch": 1.636083916083916, "percentage": 54.53, "elapsed_time": "5:49:20", "remaining_time": "4:51:17"} +{"current_steps": 2926, "total_steps": 5364, "loss": 0.10404511541128159, "lr": 1.0165955945967904e-05, "epoch": 1.6366433566433567, "percentage": 54.55, "elapsed_time": "5:49:27", "remaining_time": "4:51:10"} +{"current_steps": 2927, "total_steps": 5364, "loss": 0.07830087095499039, "lr": 1.0159448432503345e-05, "epoch": 1.6372027972027972, "percentage": 54.57, "elapsed_time": "5:49:34", "remaining_time": "4:51:03"} +{"current_steps": 2928, "total_steps": 5364, "loss": 0.19758430123329163, "lr": 1.0152940851498114e-05, "epoch": 1.6377622377622378, "percentage": 54.59, "elapsed_time": "5:49:42", "remaining_time": "4:50:56"} +{"current_steps": 2929, "total_steps": 5364, "loss": 0.12152351438999176, "lr": 1.0146433205708755e-05, "epoch": 1.6383216783216783, "percentage": 54.6, "elapsed_time": "5:49:49", "remaining_time": "4:50:49"} +{"current_steps": 2930, "total_steps": 5364, "loss": 0.12206335365772247, "lr": 1.0139925497891834e-05, "epoch": 1.6388811188811188, "percentage": 54.62, "elapsed_time": "5:49:56", "remaining_time": "4:50:41"} +{"current_steps": 2931, "total_steps": 5364, "loss": 0.03285663574934006, "lr": 1.0133417730803947e-05, "epoch": 1.6394405594405594, "percentage": 54.64, "elapsed_time": "5:50:03", "remaining_time": "4:50:34"} +{"current_steps": 2932, "total_steps": 5364, "loss": 0.060423556715250015, "lr": 1.0126909907201717e-05, "epoch": 1.6400000000000001, "percentage": 54.66, "elapsed_time": "5:50:10", "remaining_time": "4:50:27"} +{"current_steps": 2933, "total_steps": 5364, "loss": 0.24861504137516022, "lr": 1.012040202984179e-05, "epoch": 1.6405594405594406, "percentage": 54.68, "elapsed_time": "5:50:18", "remaining_time": "4:50:20"} +{"current_steps": 2934, "total_steps": 5364, "loss": 0.18891292810440063, "lr": 1.0113894101480834e-05, "epoch": 1.641118881118881, "percentage": 54.7, "elapsed_time": "5:50:25", "remaining_time": "4:50:13"} +{"current_steps": 2935, "total_steps": 5364, "loss": 0.08980001509189606, "lr": 1.0107386124875542e-05, "epoch": 1.6416783216783217, "percentage": 54.72, "elapsed_time": "5:50:32", "remaining_time": "4:50:06"} +{"current_steps": 2936, "total_steps": 5364, "loss": 0.08959965407848358, "lr": 1.0100878102782617e-05, "epoch": 1.6422377622377622, "percentage": 54.74, "elapsed_time": "5:50:39", "remaining_time": "4:49:59"} +{"current_steps": 2937, "total_steps": 5364, "loss": 0.14102260768413544, "lr": 1.009437003795879e-05, "epoch": 1.6427972027972029, "percentage": 54.75, "elapsed_time": "5:50:46", "remaining_time": "4:49:51"} +{"current_steps": 2938, "total_steps": 5364, "loss": 0.15976952016353607, "lr": 1.0087861933160811e-05, "epoch": 1.6433566433566433, "percentage": 54.77, "elapsed_time": "5:50:53", "remaining_time": "4:49:44"} +{"current_steps": 2939, "total_steps": 5364, "loss": 0.1374768167734146, "lr": 1.008135379114545e-05, "epoch": 1.6439160839160838, "percentage": 54.79, "elapsed_time": "5:51:00", "remaining_time": "4:49:37"} +{"current_steps": 2940, "total_steps": 5364, "loss": 0.12863121926784515, "lr": 1.0074845614669477e-05, "epoch": 1.6444755244755245, "percentage": 54.81, "elapsed_time": "5:51:07", "remaining_time": "4:49:29"} +{"current_steps": 2941, "total_steps": 5364, "loss": 0.08922328054904938, "lr": 1.0068337406489694e-05, "epoch": 1.6450349650349652, "percentage": 54.83, "elapsed_time": "5:51:14", "remaining_time": "4:49:22"} +{"current_steps": 2942, "total_steps": 5364, "loss": 0.2066095769405365, "lr": 1.0061829169362905e-05, "epoch": 1.6455944055944056, "percentage": 54.85, "elapsed_time": "5:51:21", "remaining_time": "4:49:15"} +{"current_steps": 2943, "total_steps": 5364, "loss": 0.11847516894340515, "lr": 1.0055320906045934e-05, "epoch": 1.646153846153846, "percentage": 54.87, "elapsed_time": "5:51:28", "remaining_time": "4:49:08"} +{"current_steps": 2944, "total_steps": 5364, "loss": 0.10722637176513672, "lr": 1.0048812619295607e-05, "epoch": 1.6467132867132868, "percentage": 54.88, "elapsed_time": "5:51:36", "remaining_time": "4:49:01"} +{"current_steps": 2945, "total_steps": 5364, "loss": 0.11080102622509003, "lr": 1.004230431186877e-05, "epoch": 1.6472727272727272, "percentage": 54.9, "elapsed_time": "5:51:43", "remaining_time": "4:48:54"} +{"current_steps": 2946, "total_steps": 5364, "loss": 0.27729618549346924, "lr": 1.0035795986522278e-05, "epoch": 1.647832167832168, "percentage": 54.92, "elapsed_time": "5:51:50", "remaining_time": "4:48:47"} +{"current_steps": 2947, "total_steps": 5364, "loss": 0.10443106293678284, "lr": 1.002928764601298e-05, "epoch": 1.6483916083916084, "percentage": 54.94, "elapsed_time": "5:51:57", "remaining_time": "4:48:39"} +{"current_steps": 2948, "total_steps": 5364, "loss": 0.23782573640346527, "lr": 1.0022779293097746e-05, "epoch": 1.6489510489510488, "percentage": 54.96, "elapsed_time": "5:52:04", "remaining_time": "4:48:32"} +{"current_steps": 2949, "total_steps": 5364, "loss": 0.1809787154197693, "lr": 1.0016270930533442e-05, "epoch": 1.6495104895104895, "percentage": 54.98, "elapsed_time": "5:52:11", "remaining_time": "4:48:24"} +{"current_steps": 2950, "total_steps": 5364, "loss": 0.37564533948898315, "lr": 1.0009762561076943e-05, "epoch": 1.6500699300699302, "percentage": 55.0, "elapsed_time": "5:52:18", "remaining_time": "4:48:17"} +{"current_steps": 2951, "total_steps": 5364, "loss": 0.19317422807216644, "lr": 1.0003254187485127e-05, "epoch": 1.6506293706293707, "percentage": 55.01, "elapsed_time": "5:52:25", "remaining_time": "4:48:10"} +{"current_steps": 2952, "total_steps": 5364, "loss": 0.2450426071882248, "lr": 9.996745812514876e-06, "epoch": 1.6511888111888111, "percentage": 55.03, "elapsed_time": "5:52:32", "remaining_time": "4:48:03"} +{"current_steps": 2953, "total_steps": 5364, "loss": 0.10384012758731842, "lr": 9.990237438923057e-06, "epoch": 1.6517482517482518, "percentage": 55.05, "elapsed_time": "5:52:40", "remaining_time": "4:47:56"} +{"current_steps": 2954, "total_steps": 5364, "loss": 0.1966351568698883, "lr": 9.983729069466561e-06, "epoch": 1.6523076923076923, "percentage": 55.07, "elapsed_time": "5:52:47", "remaining_time": "4:47:49"} +{"current_steps": 2955, "total_steps": 5364, "loss": 0.0763726457953453, "lr": 9.977220706902259e-06, "epoch": 1.652867132867133, "percentage": 55.09, "elapsed_time": "5:52:54", "remaining_time": "4:47:41"} +{"current_steps": 2956, "total_steps": 5364, "loss": 0.1695699691772461, "lr": 9.970712353987022e-06, "epoch": 1.6534265734265734, "percentage": 55.11, "elapsed_time": "5:53:01", "remaining_time": "4:47:35"} +{"current_steps": 2957, "total_steps": 5364, "loss": 0.15836942195892334, "lr": 9.964204013477727e-06, "epoch": 1.6539860139860139, "percentage": 55.13, "elapsed_time": "5:53:09", "remaining_time": "4:47:28"} +{"current_steps": 2958, "total_steps": 5364, "loss": 0.22079992294311523, "lr": 9.957695688131229e-06, "epoch": 1.6545454545454545, "percentage": 55.15, "elapsed_time": "5:53:16", "remaining_time": "4:47:20"} +{"current_steps": 2959, "total_steps": 5364, "loss": 0.013805646449327469, "lr": 9.951187380704396e-06, "epoch": 1.6551048951048952, "percentage": 55.16, "elapsed_time": "5:53:23", "remaining_time": "4:47:13"} +{"current_steps": 2960, "total_steps": 5364, "loss": 0.11412828415632248, "lr": 9.944679093954073e-06, "epoch": 1.6556643356643357, "percentage": 55.18, "elapsed_time": "5:53:30", "remaining_time": "4:47:06"} +{"current_steps": 2961, "total_steps": 5364, "loss": 0.4606684744358063, "lr": 9.938170830637099e-06, "epoch": 1.6562237762237761, "percentage": 55.2, "elapsed_time": "5:53:37", "remaining_time": "4:46:58"} +{"current_steps": 2962, "total_steps": 5364, "loss": 0.17266225814819336, "lr": 9.931662593510312e-06, "epoch": 1.6567832167832168, "percentage": 55.22, "elapsed_time": "5:53:44", "remaining_time": "4:46:51"} +{"current_steps": 2963, "total_steps": 5364, "loss": 0.11028154194355011, "lr": 9.925154385330524e-06, "epoch": 1.6573426573426573, "percentage": 55.24, "elapsed_time": "5:53:51", "remaining_time": "4:46:44"} +{"current_steps": 2964, "total_steps": 5364, "loss": 0.19698432087898254, "lr": 9.918646208854553e-06, "epoch": 1.657902097902098, "percentage": 55.26, "elapsed_time": "5:53:58", "remaining_time": "4:46:36"} +{"current_steps": 2965, "total_steps": 5364, "loss": 0.29060137271881104, "lr": 9.912138066839187e-06, "epoch": 1.6584615384615384, "percentage": 55.28, "elapsed_time": "5:54:05", "remaining_time": "4:46:29"} +{"current_steps": 2966, "total_steps": 5364, "loss": 0.23911085724830627, "lr": 9.905629962041212e-06, "epoch": 1.659020979020979, "percentage": 55.29, "elapsed_time": "5:54:12", "remaining_time": "4:46:22"} +{"current_steps": 2967, "total_steps": 5364, "loss": 0.11736980825662613, "lr": 9.89912189721739e-06, "epoch": 1.6595804195804196, "percentage": 55.31, "elapsed_time": "5:54:19", "remaining_time": "4:46:15"} +{"current_steps": 2968, "total_steps": 5364, "loss": 0.19721703231334686, "lr": 9.892613875124461e-06, "epoch": 1.6601398601398603, "percentage": 55.33, "elapsed_time": "5:54:27", "remaining_time": "4:46:08"} +{"current_steps": 2969, "total_steps": 5364, "loss": 0.09309659898281097, "lr": 9.886105898519168e-06, "epoch": 1.6606993006993007, "percentage": 55.35, "elapsed_time": "5:54:34", "remaining_time": "4:46:01"} +{"current_steps": 2970, "total_steps": 5364, "loss": 0.14063605666160583, "lr": 9.879597970158212e-06, "epoch": 1.6612587412587412, "percentage": 55.37, "elapsed_time": "5:54:41", "remaining_time": "4:45:54"} +{"current_steps": 2971, "total_steps": 5364, "loss": 0.0974130779504776, "lr": 9.873090092798285e-06, "epoch": 1.6618181818181819, "percentage": 55.39, "elapsed_time": "5:54:48", "remaining_time": "4:45:46"} +{"current_steps": 2972, "total_steps": 5364, "loss": 0.24445199966430664, "lr": 9.866582269196058e-06, "epoch": 1.6623776223776225, "percentage": 55.41, "elapsed_time": "5:54:55", "remaining_time": "4:45:39"} +{"current_steps": 2973, "total_steps": 5364, "loss": 0.21387150883674622, "lr": 9.86007450210817e-06, "epoch": 1.662937062937063, "percentage": 55.43, "elapsed_time": "5:55:02", "remaining_time": "4:45:32"} +{"current_steps": 2974, "total_steps": 5364, "loss": 0.04549172520637512, "lr": 9.853566794291248e-06, "epoch": 1.6634965034965035, "percentage": 55.44, "elapsed_time": "5:55:10", "remaining_time": "4:45:25"} +{"current_steps": 2975, "total_steps": 5364, "loss": 0.15380819141864777, "lr": 9.847059148501886e-06, "epoch": 1.664055944055944, "percentage": 55.46, "elapsed_time": "5:55:17", "remaining_time": "4:45:18"} +{"current_steps": 2976, "total_steps": 5364, "loss": 0.1858704388141632, "lr": 9.840551567496657e-06, "epoch": 1.6646153846153846, "percentage": 55.48, "elapsed_time": "5:55:24", "remaining_time": "4:45:11"} +{"current_steps": 2977, "total_steps": 5364, "loss": 0.2124638855457306, "lr": 9.834044054032101e-06, "epoch": 1.6651748251748253, "percentage": 55.5, "elapsed_time": "5:55:31", "remaining_time": "4:45:03"} +{"current_steps": 2978, "total_steps": 5364, "loss": 0.2959882318973541, "lr": 9.82753661086473e-06, "epoch": 1.6657342657342658, "percentage": 55.52, "elapsed_time": "5:55:38", "remaining_time": "4:44:56"} +{"current_steps": 2979, "total_steps": 5364, "loss": 0.15173383057117462, "lr": 9.821029240751035e-06, "epoch": 1.6662937062937062, "percentage": 55.54, "elapsed_time": "5:55:45", "remaining_time": "4:44:49"} +{"current_steps": 2980, "total_steps": 5364, "loss": 0.05073995888233185, "lr": 9.814521946447459e-06, "epoch": 1.666853146853147, "percentage": 55.56, "elapsed_time": "5:55:52", "remaining_time": "4:44:42"} +{"current_steps": 2981, "total_steps": 5364, "loss": 0.04940441995859146, "lr": 9.808014730710433e-06, "epoch": 1.6674125874125876, "percentage": 55.57, "elapsed_time": "5:55:59", "remaining_time": "4:44:34"} +{"current_steps": 2982, "total_steps": 5364, "loss": 0.10461293160915375, "lr": 9.80150759629634e-06, "epoch": 1.667972027972028, "percentage": 55.59, "elapsed_time": "5:56:07", "remaining_time": "4:44:27"} +{"current_steps": 2983, "total_steps": 5364, "loss": 0.3275340795516968, "lr": 9.795000545961537e-06, "epoch": 1.6685314685314685, "percentage": 55.61, "elapsed_time": "5:56:14", "remaining_time": "4:44:20"} +{"current_steps": 2984, "total_steps": 5364, "loss": 0.06173175200819969, "lr": 9.788493582462339e-06, "epoch": 1.669090909090909, "percentage": 55.63, "elapsed_time": "5:56:21", "remaining_time": "4:44:13"} +{"current_steps": 2985, "total_steps": 5364, "loss": 0.1987234354019165, "lr": 9.781986708555023e-06, "epoch": 1.6696503496503496, "percentage": 55.65, "elapsed_time": "5:56:28", "remaining_time": "4:44:06"} +{"current_steps": 2986, "total_steps": 5364, "loss": 0.19026236236095428, "lr": 9.775479926995841e-06, "epoch": 1.6702097902097903, "percentage": 55.67, "elapsed_time": "5:56:35", "remaining_time": "4:43:59"} +{"current_steps": 2987, "total_steps": 5364, "loss": 0.06463699042797089, "lr": 9.768973240540993e-06, "epoch": 1.6707692307692308, "percentage": 55.69, "elapsed_time": "5:56:42", "remaining_time": "4:43:52"} +{"current_steps": 2988, "total_steps": 5364, "loss": 0.133989155292511, "lr": 9.762466651946642e-06, "epoch": 1.6713286713286712, "percentage": 55.7, "elapsed_time": "5:56:49", "remaining_time": "4:43:44"} +{"current_steps": 2989, "total_steps": 5364, "loss": 0.3031919002532959, "lr": 9.75596016396891e-06, "epoch": 1.671888111888112, "percentage": 55.72, "elapsed_time": "5:56:56", "remaining_time": "4:43:37"} +{"current_steps": 2990, "total_steps": 5364, "loss": 0.10024263709783554, "lr": 9.749453779363875e-06, "epoch": 1.6724475524475526, "percentage": 55.74, "elapsed_time": "5:57:04", "remaining_time": "4:43:30"} +{"current_steps": 2991, "total_steps": 5364, "loss": 0.20258700847625732, "lr": 9.742947500887578e-06, "epoch": 1.673006993006993, "percentage": 55.76, "elapsed_time": "5:57:10", "remaining_time": "4:43:22"} +{"current_steps": 2992, "total_steps": 5364, "loss": 0.1786518394947052, "lr": 9.736441331296006e-06, "epoch": 1.6735664335664335, "percentage": 55.78, "elapsed_time": "5:57:17", "remaining_time": "4:43:15"} +{"current_steps": 2993, "total_steps": 5364, "loss": 0.2345372438430786, "lr": 9.729935273345105e-06, "epoch": 1.674125874125874, "percentage": 55.8, "elapsed_time": "5:57:24", "remaining_time": "4:43:08"} +{"current_steps": 2994, "total_steps": 5364, "loss": 0.22118642926216125, "lr": 9.723429329790774e-06, "epoch": 1.6746853146853147, "percentage": 55.82, "elapsed_time": "5:57:32", "remaining_time": "4:43:01"} +{"current_steps": 2995, "total_steps": 5364, "loss": 0.1314345896244049, "lr": 9.716923503388856e-06, "epoch": 1.6752447552447554, "percentage": 55.84, "elapsed_time": "5:57:38", "remaining_time": "4:42:53"} +{"current_steps": 2996, "total_steps": 5364, "loss": 0.08646439760923386, "lr": 9.710417796895159e-06, "epoch": 1.6758041958041958, "percentage": 55.85, "elapsed_time": "5:57:46", "remaining_time": "4:42:46"} +{"current_steps": 2997, "total_steps": 5364, "loss": 0.05085134878754616, "lr": 9.703912213065425e-06, "epoch": 1.6763636363636363, "percentage": 55.87, "elapsed_time": "5:57:53", "remaining_time": "4:42:39"} +{"current_steps": 2998, "total_steps": 5364, "loss": 0.1367836743593216, "lr": 9.697406754655357e-06, "epoch": 1.676923076923077, "percentage": 55.89, "elapsed_time": "5:58:00", "remaining_time": "4:42:32"} +{"current_steps": 2999, "total_steps": 5364, "loss": 0.13026833534240723, "lr": 9.690901424420588e-06, "epoch": 1.6774825174825176, "percentage": 55.91, "elapsed_time": "5:58:07", "remaining_time": "4:42:25"} +{"current_steps": 3000, "total_steps": 5364, "loss": 0.11952243000268936, "lr": 9.68439622511672e-06, "epoch": 1.678041958041958, "percentage": 55.93, "elapsed_time": "5:58:14", "remaining_time": "4:42:17"} +{"current_steps": 3001, "total_steps": 5364, "loss": 0.10476125776767731, "lr": 9.677891159499282e-06, "epoch": 1.6786013986013986, "percentage": 55.95, "elapsed_time": "5:58:49", "remaining_time": "4:42:32"} +{"current_steps": 3002, "total_steps": 5364, "loss": 0.17805519700050354, "lr": 9.671386230323744e-06, "epoch": 1.679160839160839, "percentage": 55.97, "elapsed_time": "5:58:56", "remaining_time": "4:42:24"} +{"current_steps": 3003, "total_steps": 5364, "loss": 0.12021731585264206, "lr": 9.664881440345534e-06, "epoch": 1.6797202797202797, "percentage": 55.98, "elapsed_time": "5:59:03", "remaining_time": "4:42:17"} +{"current_steps": 3004, "total_steps": 5364, "loss": 0.1792115569114685, "lr": 9.658376792320008e-06, "epoch": 1.6802797202797204, "percentage": 56.0, "elapsed_time": "5:59:10", "remaining_time": "4:42:10"} +{"current_steps": 3005, "total_steps": 5364, "loss": 0.20040813088417053, "lr": 9.651872289002471e-06, "epoch": 1.6808391608391609, "percentage": 56.02, "elapsed_time": "5:59:18", "remaining_time": "4:42:03"} +{"current_steps": 3006, "total_steps": 5364, "loss": 0.05844014883041382, "lr": 9.645367933148158e-06, "epoch": 1.6813986013986013, "percentage": 56.04, "elapsed_time": "5:59:25", "remaining_time": "4:41:56"} +{"current_steps": 3007, "total_steps": 5364, "loss": 0.06185554713010788, "lr": 9.63886372751224e-06, "epoch": 1.681958041958042, "percentage": 56.06, "elapsed_time": "5:59:32", "remaining_time": "4:41:49"} +{"current_steps": 3008, "total_steps": 5364, "loss": 0.08320507407188416, "lr": 9.632359674849842e-06, "epoch": 1.6825174825174827, "percentage": 56.08, "elapsed_time": "5:59:39", "remaining_time": "4:41:42"} +{"current_steps": 3009, "total_steps": 5364, "loss": 0.19655077159404755, "lr": 9.625855777916005e-06, "epoch": 1.6830769230769231, "percentage": 56.1, "elapsed_time": "5:59:46", "remaining_time": "4:41:34"} +{"current_steps": 3010, "total_steps": 5364, "loss": 0.1426973193883896, "lr": 9.619352039465715e-06, "epoch": 1.6836363636363636, "percentage": 56.11, "elapsed_time": "5:59:53", "remaining_time": "4:41:27"} +{"current_steps": 3011, "total_steps": 5364, "loss": 0.14301709830760956, "lr": 9.612848462253882e-06, "epoch": 1.684195804195804, "percentage": 56.13, "elapsed_time": "6:00:01", "remaining_time": "4:41:20"} +{"current_steps": 3012, "total_steps": 5364, "loss": 0.0847204178571701, "lr": 9.606345049035352e-06, "epoch": 1.6847552447552447, "percentage": 56.15, "elapsed_time": "6:00:08", "remaining_time": "4:41:13"} +{"current_steps": 3013, "total_steps": 5364, "loss": 0.09985585510730743, "lr": 9.59984180256491e-06, "epoch": 1.6853146853146854, "percentage": 56.17, "elapsed_time": "6:00:16", "remaining_time": "4:41:06"} +{"current_steps": 3014, "total_steps": 5364, "loss": 0.09818299114704132, "lr": 9.593338725597255e-06, "epoch": 1.6858741258741259, "percentage": 56.19, "elapsed_time": "6:00:23", "remaining_time": "4:40:59"} +{"current_steps": 3015, "total_steps": 5364, "loss": 0.15568727254867554, "lr": 9.586835820887026e-06, "epoch": 1.6864335664335663, "percentage": 56.21, "elapsed_time": "6:00:30", "remaining_time": "4:40:52"} +{"current_steps": 3016, "total_steps": 5364, "loss": 0.13128268718719482, "lr": 9.580333091188784e-06, "epoch": 1.686993006993007, "percentage": 56.23, "elapsed_time": "6:00:37", "remaining_time": "4:40:45"} +{"current_steps": 3017, "total_steps": 5364, "loss": 0.2241504043340683, "lr": 9.573830539257012e-06, "epoch": 1.6875524475524477, "percentage": 56.25, "elapsed_time": "6:00:44", "remaining_time": "4:40:37"} +{"current_steps": 3018, "total_steps": 5364, "loss": 0.16132117807865143, "lr": 9.56732816784613e-06, "epoch": 1.6881118881118882, "percentage": 56.26, "elapsed_time": "6:00:51", "remaining_time": "4:40:30"} +{"current_steps": 3019, "total_steps": 5364, "loss": 0.15984748303890228, "lr": 9.560825979710469e-06, "epoch": 1.6886713286713286, "percentage": 56.28, "elapsed_time": "6:00:58", "remaining_time": "4:40:22"} +{"current_steps": 3020, "total_steps": 5364, "loss": 0.25019705295562744, "lr": 9.554323977604291e-06, "epoch": 1.689230769230769, "percentage": 56.3, "elapsed_time": "6:01:05", "remaining_time": "4:40:15"} +{"current_steps": 3021, "total_steps": 5364, "loss": 0.007543501909822226, "lr": 9.547822164281766e-06, "epoch": 1.6897902097902098, "percentage": 56.32, "elapsed_time": "6:01:12", "remaining_time": "4:40:08"} +{"current_steps": 3022, "total_steps": 5364, "loss": 0.2000923752784729, "lr": 9.541320542497009e-06, "epoch": 1.6903496503496505, "percentage": 56.34, "elapsed_time": "6:01:19", "remaining_time": "4:40:01"} +{"current_steps": 3023, "total_steps": 5364, "loss": 0.33368125557899475, "lr": 9.534819115004027e-06, "epoch": 1.690909090909091, "percentage": 56.36, "elapsed_time": "6:01:26", "remaining_time": "4:39:54"} +{"current_steps": 3024, "total_steps": 5364, "loss": 0.13940201699733734, "lr": 9.528317884556759e-06, "epoch": 1.6914685314685314, "percentage": 56.38, "elapsed_time": "6:01:33", "remaining_time": "4:39:46"} +{"current_steps": 3025, "total_steps": 5364, "loss": 0.16801759600639343, "lr": 9.521816853909059e-06, "epoch": 1.692027972027972, "percentage": 56.39, "elapsed_time": "6:01:40", "remaining_time": "4:39:39"} +{"current_steps": 3026, "total_steps": 5364, "loss": 0.110100656747818, "lr": 9.51531602581469e-06, "epoch": 1.6925874125874127, "percentage": 56.41, "elapsed_time": "6:01:47", "remaining_time": "4:39:32"} +{"current_steps": 3027, "total_steps": 5364, "loss": 0.2312549203634262, "lr": 9.508815403027343e-06, "epoch": 1.6931468531468532, "percentage": 56.43, "elapsed_time": "6:01:55", "remaining_time": "4:39:25"} +{"current_steps": 3028, "total_steps": 5364, "loss": 0.2715122699737549, "lr": 9.502314988300606e-06, "epoch": 1.6937062937062937, "percentage": 56.45, "elapsed_time": "6:02:02", "remaining_time": "4:39:17"} +{"current_steps": 3029, "total_steps": 5364, "loss": 0.11637977510690689, "lr": 9.495814784387984e-06, "epoch": 1.6942657342657341, "percentage": 56.47, "elapsed_time": "6:02:09", "remaining_time": "4:39:11"} +{"current_steps": 3030, "total_steps": 5364, "loss": 0.24676674604415894, "lr": 9.489314794042904e-06, "epoch": 1.6948251748251748, "percentage": 56.49, "elapsed_time": "6:02:16", "remaining_time": "4:39:03"} +{"current_steps": 3031, "total_steps": 5364, "loss": 0.3834327161312103, "lr": 9.482815020018686e-06, "epoch": 1.6953846153846155, "percentage": 56.51, "elapsed_time": "6:02:23", "remaining_time": "4:38:56"} +{"current_steps": 3032, "total_steps": 5364, "loss": 0.21393917500972748, "lr": 9.47631546506857e-06, "epoch": 1.695944055944056, "percentage": 56.52, "elapsed_time": "6:02:30", "remaining_time": "4:38:49"} +{"current_steps": 3033, "total_steps": 5364, "loss": 0.10340915620326996, "lr": 9.469816131945698e-06, "epoch": 1.6965034965034964, "percentage": 56.54, "elapsed_time": "6:02:38", "remaining_time": "4:38:42"} +{"current_steps": 3034, "total_steps": 5364, "loss": 0.13875150680541992, "lr": 9.463317023403112e-06, "epoch": 1.697062937062937, "percentage": 56.56, "elapsed_time": "6:02:45", "remaining_time": "4:38:34"} +{"current_steps": 3035, "total_steps": 5364, "loss": 0.10419341921806335, "lr": 9.456818142193774e-06, "epoch": 1.6976223776223778, "percentage": 56.58, "elapsed_time": "6:02:51", "remaining_time": "4:38:27"} +{"current_steps": 3036, "total_steps": 5364, "loss": 0.12065944820642471, "lr": 9.45031949107054e-06, "epoch": 1.6981818181818182, "percentage": 56.6, "elapsed_time": "6:02:58", "remaining_time": "4:38:19"} +{"current_steps": 3037, "total_steps": 5364, "loss": 0.19922295212745667, "lr": 9.44382107278617e-06, "epoch": 1.6987412587412587, "percentage": 56.62, "elapsed_time": "6:03:05", "remaining_time": "4:38:12"} +{"current_steps": 3038, "total_steps": 5364, "loss": 0.12009718269109726, "lr": 9.43732289009332e-06, "epoch": 1.6993006993006992, "percentage": 56.64, "elapsed_time": "6:03:12", "remaining_time": "4:38:05"} +{"current_steps": 3039, "total_steps": 5364, "loss": 0.15200388431549072, "lr": 9.430824945744561e-06, "epoch": 1.6998601398601398, "percentage": 56.66, "elapsed_time": "6:03:20", "remaining_time": "4:37:58"} +{"current_steps": 3040, "total_steps": 5364, "loss": 0.08002086728811264, "lr": 9.424327242492349e-06, "epoch": 1.7004195804195805, "percentage": 56.67, "elapsed_time": "6:03:27", "remaining_time": "4:37:51"} +{"current_steps": 3041, "total_steps": 5364, "loss": 0.332578182220459, "lr": 9.41782978308904e-06, "epoch": 1.700979020979021, "percentage": 56.69, "elapsed_time": "6:03:34", "remaining_time": "4:37:44"} +{"current_steps": 3042, "total_steps": 5364, "loss": 0.13766759634017944, "lr": 9.411332570286894e-06, "epoch": 1.7015384615384614, "percentage": 56.71, "elapsed_time": "6:03:41", "remaining_time": "4:37:36"} +{"current_steps": 3043, "total_steps": 5364, "loss": 0.20101264119148254, "lr": 9.404835606838053e-06, "epoch": 1.7020979020979021, "percentage": 56.73, "elapsed_time": "6:03:48", "remaining_time": "4:37:29"} +{"current_steps": 3044, "total_steps": 5364, "loss": 0.33135032653808594, "lr": 9.398338895494574e-06, "epoch": 1.7026573426573428, "percentage": 56.75, "elapsed_time": "6:03:55", "remaining_time": "4:37:22"} +{"current_steps": 3045, "total_steps": 5364, "loss": 0.026278357952833176, "lr": 9.391842439008391e-06, "epoch": 1.7032167832167833, "percentage": 56.77, "elapsed_time": "6:04:02", "remaining_time": "4:37:14"} +{"current_steps": 3046, "total_steps": 5364, "loss": 0.13241004943847656, "lr": 9.38534624013133e-06, "epoch": 1.7037762237762237, "percentage": 56.79, "elapsed_time": "6:04:10", "remaining_time": "4:37:08"} +{"current_steps": 3047, "total_steps": 5364, "loss": 0.06253943592309952, "lr": 9.378850301615118e-06, "epoch": 1.7043356643356642, "percentage": 56.8, "elapsed_time": "6:04:17", "remaining_time": "4:37:00"} +{"current_steps": 3048, "total_steps": 5364, "loss": 0.16502872109413147, "lr": 9.372354626211359e-06, "epoch": 1.7048951048951049, "percentage": 56.82, "elapsed_time": "6:04:24", "remaining_time": "4:36:53"} +{"current_steps": 3049, "total_steps": 5364, "loss": 0.2116778939962387, "lr": 9.36585921667156e-06, "epoch": 1.7054545454545456, "percentage": 56.84, "elapsed_time": "6:04:31", "remaining_time": "4:36:46"} +{"current_steps": 3050, "total_steps": 5364, "loss": 0.3423939347267151, "lr": 9.359364075747111e-06, "epoch": 1.706013986013986, "percentage": 56.86, "elapsed_time": "6:04:38", "remaining_time": "4:36:38"} +{"current_steps": 3051, "total_steps": 5364, "loss": 0.03352992236614227, "lr": 9.35286920618927e-06, "epoch": 1.7065734265734265, "percentage": 56.88, "elapsed_time": "6:04:46", "remaining_time": "4:36:32"} +{"current_steps": 3052, "total_steps": 5364, "loss": 0.21770413219928741, "lr": 9.34637461074921e-06, "epoch": 1.7071328671328672, "percentage": 56.9, "elapsed_time": "6:04:53", "remaining_time": "4:36:24"} +{"current_steps": 3053, "total_steps": 5364, "loss": 0.04001019150018692, "lr": 9.339880292177964e-06, "epoch": 1.7076923076923078, "percentage": 56.92, "elapsed_time": "6:05:00", "remaining_time": "4:36:17"} +{"current_steps": 3054, "total_steps": 5364, "loss": 0.13102063536643982, "lr": 9.333386253226464e-06, "epoch": 1.7082517482517483, "percentage": 56.94, "elapsed_time": "6:05:07", "remaining_time": "4:36:10"} +{"current_steps": 3055, "total_steps": 5364, "loss": 0.14729930460453033, "lr": 9.326892496645512e-06, "epoch": 1.7088111888111888, "percentage": 56.95, "elapsed_time": "6:05:14", "remaining_time": "4:36:03"} +{"current_steps": 3056, "total_steps": 5364, "loss": 0.14406803250312805, "lr": 9.320399025185791e-06, "epoch": 1.7093706293706292, "percentage": 56.97, "elapsed_time": "6:05:21", "remaining_time": "4:35:55"} +{"current_steps": 3057, "total_steps": 5364, "loss": 0.1300947070121765, "lr": 9.313905841597875e-06, "epoch": 1.70993006993007, "percentage": 56.99, "elapsed_time": "6:05:28", "remaining_time": "4:35:48"} +{"current_steps": 3058, "total_steps": 5364, "loss": 0.23029902577400208, "lr": 9.307412948632204e-06, "epoch": 1.7104895104895106, "percentage": 57.01, "elapsed_time": "6:05:35", "remaining_time": "4:35:41"} +{"current_steps": 3059, "total_steps": 5364, "loss": 0.23121728003025055, "lr": 9.3009203490391e-06, "epoch": 1.711048951048951, "percentage": 57.03, "elapsed_time": "6:05:42", "remaining_time": "4:35:33"} +{"current_steps": 3060, "total_steps": 5364, "loss": 0.0074851601384580135, "lr": 9.294428045568757e-06, "epoch": 1.7116083916083915, "percentage": 57.05, "elapsed_time": "6:05:49", "remaining_time": "4:35:26"} +{"current_steps": 3061, "total_steps": 5364, "loss": 0.09484653174877167, "lr": 9.287936040971251e-06, "epoch": 1.7121678321678322, "percentage": 57.07, "elapsed_time": "6:05:56", "remaining_time": "4:35:19"} +{"current_steps": 3062, "total_steps": 5364, "loss": 0.13925182819366455, "lr": 9.281444337996525e-06, "epoch": 1.7127272727272729, "percentage": 57.08, "elapsed_time": "6:06:03", "remaining_time": "4:35:12"} +{"current_steps": 3063, "total_steps": 5364, "loss": 0.19127994775772095, "lr": 9.274952939394391e-06, "epoch": 1.7132867132867133, "percentage": 57.1, "elapsed_time": "6:06:10", "remaining_time": "4:35:04"} +{"current_steps": 3064, "total_steps": 5364, "loss": 0.1682354211807251, "lr": 9.268461847914545e-06, "epoch": 1.7138461538461538, "percentage": 57.12, "elapsed_time": "6:06:17", "remaining_time": "4:34:57"} +{"current_steps": 3065, "total_steps": 5364, "loss": 0.1610163301229477, "lr": 9.261971066306535e-06, "epoch": 1.7144055944055943, "percentage": 57.14, "elapsed_time": "6:06:24", "remaining_time": "4:34:49"} +{"current_steps": 3066, "total_steps": 5364, "loss": 0.19976727664470673, "lr": 9.255480597319798e-06, "epoch": 1.714965034965035, "percentage": 57.16, "elapsed_time": "6:06:31", "remaining_time": "4:34:43"} +{"current_steps": 3067, "total_steps": 5364, "loss": 0.15329812467098236, "lr": 9.248990443703624e-06, "epoch": 1.7155244755244756, "percentage": 57.18, "elapsed_time": "6:06:39", "remaining_time": "4:34:36"} +{"current_steps": 3068, "total_steps": 5364, "loss": 0.2606050670146942, "lr": 9.24250060820717e-06, "epoch": 1.716083916083916, "percentage": 57.2, "elapsed_time": "6:06:46", "remaining_time": "4:34:28"} +{"current_steps": 3069, "total_steps": 5364, "loss": 0.1810053437948227, "lr": 9.236011093579468e-06, "epoch": 1.7166433566433565, "percentage": 57.21, "elapsed_time": "6:06:53", "remaining_time": "4:34:21"} +{"current_steps": 3070, "total_steps": 5364, "loss": 0.13593192398548126, "lr": 9.229521902569399e-06, "epoch": 1.7172027972027972, "percentage": 57.23, "elapsed_time": "6:07:00", "remaining_time": "4:34:14"} +{"current_steps": 3071, "total_steps": 5364, "loss": 0.13370800018310547, "lr": 9.223033037925729e-06, "epoch": 1.717762237762238, "percentage": 57.25, "elapsed_time": "6:07:07", "remaining_time": "4:34:07"} +{"current_steps": 3072, "total_steps": 5364, "loss": 0.3885428309440613, "lr": 9.216544502397062e-06, "epoch": 1.7183216783216784, "percentage": 57.27, "elapsed_time": "6:07:15", "remaining_time": "4:34:00"} +{"current_steps": 3073, "total_steps": 5364, "loss": 0.1855706125497818, "lr": 9.210056298731876e-06, "epoch": 1.7188811188811188, "percentage": 57.29, "elapsed_time": "6:07:23", "remaining_time": "4:33:53"} +{"current_steps": 3074, "total_steps": 5364, "loss": 0.16089676320552826, "lr": 9.203568429678506e-06, "epoch": 1.7194405594405593, "percentage": 57.31, "elapsed_time": "6:07:29", "remaining_time": "4:33:46"} +{"current_steps": 3075, "total_steps": 5364, "loss": 0.2637489438056946, "lr": 9.197080897985143e-06, "epoch": 1.72, "percentage": 57.33, "elapsed_time": "6:07:37", "remaining_time": "4:33:39"} +{"current_steps": 3076, "total_steps": 5364, "loss": 0.039192814379930496, "lr": 9.190593706399844e-06, "epoch": 1.7205594405594407, "percentage": 57.35, "elapsed_time": "6:07:44", "remaining_time": "4:33:32"} +{"current_steps": 3077, "total_steps": 5364, "loss": 0.23223662376403809, "lr": 9.184106857670504e-06, "epoch": 1.7211188811188811, "percentage": 57.36, "elapsed_time": "6:07:51", "remaining_time": "4:33:24"} +{"current_steps": 3078, "total_steps": 5364, "loss": 0.12181822210550308, "lr": 9.177620354544895e-06, "epoch": 1.7216783216783216, "percentage": 57.38, "elapsed_time": "6:07:58", "remaining_time": "4:33:17"} +{"current_steps": 3079, "total_steps": 5364, "loss": 0.04619431123137474, "lr": 9.171134199770622e-06, "epoch": 1.7222377622377623, "percentage": 57.4, "elapsed_time": "6:08:06", "remaining_time": "4:33:10"} +{"current_steps": 3080, "total_steps": 5364, "loss": 0.1830262690782547, "lr": 9.164648396095156e-06, "epoch": 1.722797202797203, "percentage": 57.42, "elapsed_time": "6:08:13", "remaining_time": "4:33:03"} +{"current_steps": 3081, "total_steps": 5364, "loss": 0.18911191821098328, "lr": 9.158162946265816e-06, "epoch": 1.7233566433566434, "percentage": 57.44, "elapsed_time": "6:08:21", "remaining_time": "4:32:56"} +{"current_steps": 3082, "total_steps": 5364, "loss": 0.1590443104505539, "lr": 9.151677853029764e-06, "epoch": 1.7239160839160839, "percentage": 57.46, "elapsed_time": "6:08:28", "remaining_time": "4:32:49"} +{"current_steps": 3083, "total_steps": 5364, "loss": 0.1471109688282013, "lr": 9.145193119134026e-06, "epoch": 1.7244755244755243, "percentage": 57.48, "elapsed_time": "6:08:35", "remaining_time": "4:32:42"} +{"current_steps": 3084, "total_steps": 5364, "loss": 0.40186381340026855, "lr": 9.138708747325461e-06, "epoch": 1.725034965034965, "percentage": 57.49, "elapsed_time": "6:08:42", "remaining_time": "4:32:34"} +{"current_steps": 3085, "total_steps": 5364, "loss": 0.11406250298023224, "lr": 9.132224740350782e-06, "epoch": 1.7255944055944057, "percentage": 57.51, "elapsed_time": "6:08:49", "remaining_time": "4:32:27"} +{"current_steps": 3086, "total_steps": 5364, "loss": 0.23277802765369415, "lr": 9.125741100956546e-06, "epoch": 1.7261538461538461, "percentage": 57.53, "elapsed_time": "6:08:56", "remaining_time": "4:32:20"} +{"current_steps": 3087, "total_steps": 5364, "loss": 0.10958261787891388, "lr": 9.119257831889152e-06, "epoch": 1.7267132867132866, "percentage": 57.55, "elapsed_time": "6:09:03", "remaining_time": "4:32:13"} +{"current_steps": 3088, "total_steps": 5364, "loss": 0.14662399888038635, "lr": 9.11277493589485e-06, "epoch": 1.7272727272727273, "percentage": 57.57, "elapsed_time": "6:09:10", "remaining_time": "4:32:06"} +{"current_steps": 3089, "total_steps": 5364, "loss": 0.2085634171962738, "lr": 9.106292415719723e-06, "epoch": 1.727832167832168, "percentage": 57.59, "elapsed_time": "6:09:17", "remaining_time": "4:31:58"} +{"current_steps": 3090, "total_steps": 5364, "loss": 0.036556847393512726, "lr": 9.099810274109694e-06, "epoch": 1.7283916083916084, "percentage": 57.61, "elapsed_time": "6:09:25", "remaining_time": "4:31:51"} +{"current_steps": 3091, "total_steps": 5364, "loss": 0.06841983646154404, "lr": 9.093328513810536e-06, "epoch": 1.728951048951049, "percentage": 57.62, "elapsed_time": "6:09:32", "remaining_time": "4:31:44"} +{"current_steps": 3092, "total_steps": 5364, "loss": 0.007382201496511698, "lr": 9.086847137567849e-06, "epoch": 1.7295104895104894, "percentage": 57.64, "elapsed_time": "6:09:39", "remaining_time": "4:31:37"} +{"current_steps": 3093, "total_steps": 5364, "loss": 0.2771955132484436, "lr": 9.080366148127083e-06, "epoch": 1.73006993006993, "percentage": 57.66, "elapsed_time": "6:09:46", "remaining_time": "4:31:30"} +{"current_steps": 3094, "total_steps": 5364, "loss": 0.1524166762828827, "lr": 9.073885548233509e-06, "epoch": 1.7306293706293707, "percentage": 57.68, "elapsed_time": "6:09:53", "remaining_time": "4:31:23"} +{"current_steps": 3095, "total_steps": 5364, "loss": 0.07445754110813141, "lr": 9.067405340632246e-06, "epoch": 1.7311888111888112, "percentage": 57.7, "elapsed_time": "6:10:00", "remaining_time": "4:31:15"} +{"current_steps": 3096, "total_steps": 5364, "loss": 0.06812790036201477, "lr": 9.060925528068239e-06, "epoch": 1.7317482517482516, "percentage": 57.72, "elapsed_time": "6:10:08", "remaining_time": "4:31:08"} +{"current_steps": 3097, "total_steps": 5364, "loss": 0.31041720509529114, "lr": 9.054446113286266e-06, "epoch": 1.7323076923076923, "percentage": 57.74, "elapsed_time": "6:10:14", "remaining_time": "4:31:01"} +{"current_steps": 3098, "total_steps": 5364, "loss": 0.13497644662857056, "lr": 9.047967099030945e-06, "epoch": 1.732867132867133, "percentage": 57.76, "elapsed_time": "6:10:21", "remaining_time": "4:30:53"} +{"current_steps": 3099, "total_steps": 5364, "loss": 0.10116368532180786, "lr": 9.041488488046713e-06, "epoch": 1.7334265734265735, "percentage": 57.77, "elapsed_time": "6:10:28", "remaining_time": "4:30:46"} +{"current_steps": 3100, "total_steps": 5364, "loss": 0.028338227421045303, "lr": 9.035010283077845e-06, "epoch": 1.733986013986014, "percentage": 57.79, "elapsed_time": "6:10:36", "remaining_time": "4:30:39"} +{"current_steps": 3101, "total_steps": 5364, "loss": 0.10103368759155273, "lr": 9.028532486868437e-06, "epoch": 1.7345454545454544, "percentage": 57.81, "elapsed_time": "6:10:43", "remaining_time": "4:30:32"} +{"current_steps": 3102, "total_steps": 5364, "loss": 0.0964130088686943, "lr": 9.022055102162416e-06, "epoch": 1.735104895104895, "percentage": 57.83, "elapsed_time": "6:10:50", "remaining_time": "4:30:24"} +{"current_steps": 3103, "total_steps": 5364, "loss": 0.14891275763511658, "lr": 9.015578131703534e-06, "epoch": 1.7356643356643358, "percentage": 57.85, "elapsed_time": "6:10:56", "remaining_time": "4:30:17"} +{"current_steps": 3104, "total_steps": 5364, "loss": 0.20016345381736755, "lr": 9.009101578235364e-06, "epoch": 1.7362237762237762, "percentage": 57.87, "elapsed_time": "6:11:03", "remaining_time": "4:30:10"} +{"current_steps": 3105, "total_steps": 5364, "loss": 0.1195664256811142, "lr": 9.002625444501314e-06, "epoch": 1.7367832167832167, "percentage": 57.89, "elapsed_time": "6:11:11", "remaining_time": "4:30:03"} +{"current_steps": 3106, "total_steps": 5364, "loss": 0.0809250921010971, "lr": 8.996149733244599e-06, "epoch": 1.7373426573426574, "percentage": 57.9, "elapsed_time": "6:11:18", "remaining_time": "4:29:56"} +{"current_steps": 3107, "total_steps": 5364, "loss": 0.05960802361369133, "lr": 8.989674447208263e-06, "epoch": 1.737902097902098, "percentage": 57.92, "elapsed_time": "6:11:26", "remaining_time": "4:29:49"} +{"current_steps": 3108, "total_steps": 5364, "loss": 0.2069295346736908, "lr": 8.983199589135172e-06, "epoch": 1.7384615384615385, "percentage": 57.94, "elapsed_time": "6:11:33", "remaining_time": "4:29:42"} +{"current_steps": 3109, "total_steps": 5364, "loss": 0.2972172498703003, "lr": 8.976725161768e-06, "epoch": 1.739020979020979, "percentage": 57.96, "elapsed_time": "6:11:40", "remaining_time": "4:29:34"} +{"current_steps": 3110, "total_steps": 5364, "loss": 0.11996152251958847, "lr": 8.970251167849257e-06, "epoch": 1.7395804195804194, "percentage": 57.98, "elapsed_time": "6:11:47", "remaining_time": "4:29:27"} +{"current_steps": 3111, "total_steps": 5364, "loss": 0.08648248761892319, "lr": 8.963777610121253e-06, "epoch": 1.74013986013986, "percentage": 58.0, "elapsed_time": "6:11:54", "remaining_time": "4:29:20"} +{"current_steps": 3112, "total_steps": 5364, "loss": 0.20402149856090546, "lr": 8.957304491326115e-06, "epoch": 1.7406993006993008, "percentage": 58.02, "elapsed_time": "6:12:01", "remaining_time": "4:29:12"} +{"current_steps": 3113, "total_steps": 5364, "loss": 0.1489095240831375, "lr": 8.950831814205796e-06, "epoch": 1.7412587412587412, "percentage": 58.04, "elapsed_time": "6:12:08", "remaining_time": "4:29:05"} +{"current_steps": 3114, "total_steps": 5364, "loss": 0.13151924312114716, "lr": 8.944359581502042e-06, "epoch": 1.7418181818181817, "percentage": 58.05, "elapsed_time": "6:12:15", "remaining_time": "4:28:58"} +{"current_steps": 3115, "total_steps": 5364, "loss": 0.06777593493461609, "lr": 8.937887795956436e-06, "epoch": 1.7423776223776224, "percentage": 58.07, "elapsed_time": "6:12:23", "remaining_time": "4:28:51"} +{"current_steps": 3116, "total_steps": 5364, "loss": 0.2953047752380371, "lr": 8.93141646031035e-06, "epoch": 1.742937062937063, "percentage": 58.09, "elapsed_time": "6:12:30", "remaining_time": "4:28:44"} +{"current_steps": 3117, "total_steps": 5364, "loss": 0.19538140296936035, "lr": 8.924945577304975e-06, "epoch": 1.7434965034965035, "percentage": 58.11, "elapsed_time": "6:12:37", "remaining_time": "4:28:37"} +{"current_steps": 3118, "total_steps": 5364, "loss": 0.10583820939064026, "lr": 8.91847514968131e-06, "epoch": 1.744055944055944, "percentage": 58.13, "elapsed_time": "6:12:44", "remaining_time": "4:28:30"} +{"current_steps": 3119, "total_steps": 5364, "loss": 0.04031681641936302, "lr": 8.912005180180155e-06, "epoch": 1.7446153846153845, "percentage": 58.15, "elapsed_time": "6:12:51", "remaining_time": "4:28:22"} +{"current_steps": 3120, "total_steps": 5364, "loss": 0.0754489153623581, "lr": 8.905535671542129e-06, "epoch": 1.7451748251748251, "percentage": 58.17, "elapsed_time": "6:12:58", "remaining_time": "4:28:15"} +{"current_steps": 3121, "total_steps": 5364, "loss": 0.190330371260643, "lr": 8.899066626507641e-06, "epoch": 1.7457342657342658, "percentage": 58.18, "elapsed_time": "6:13:05", "remaining_time": "4:28:08"} +{"current_steps": 3122, "total_steps": 5364, "loss": 0.24131014943122864, "lr": 8.892598047816915e-06, "epoch": 1.7462937062937063, "percentage": 58.2, "elapsed_time": "6:13:14", "remaining_time": "4:28:01"} +{"current_steps": 3123, "total_steps": 5364, "loss": 0.15173746645450592, "lr": 8.88612993820997e-06, "epoch": 1.7468531468531467, "percentage": 58.22, "elapsed_time": "6:13:22", "remaining_time": "4:27:55"} +{"current_steps": 3124, "total_steps": 5364, "loss": 0.21992839872837067, "lr": 8.879662300426623e-06, "epoch": 1.7474125874125874, "percentage": 58.24, "elapsed_time": "6:13:28", "remaining_time": "4:27:47"} +{"current_steps": 3125, "total_steps": 5364, "loss": 0.09761320054531097, "lr": 8.873195137206511e-06, "epoch": 1.747972027972028, "percentage": 58.26, "elapsed_time": "6:13:35", "remaining_time": "4:27:40"} +{"current_steps": 3126, "total_steps": 5364, "loss": 0.2660660147666931, "lr": 8.866728451289043e-06, "epoch": 1.7485314685314686, "percentage": 58.28, "elapsed_time": "6:13:42", "remaining_time": "4:27:33"} +{"current_steps": 3127, "total_steps": 5364, "loss": 0.12606793642044067, "lr": 8.860262245413448e-06, "epoch": 1.749090909090909, "percentage": 58.3, "elapsed_time": "6:13:50", "remaining_time": "4:27:26"} +{"current_steps": 3128, "total_steps": 5364, "loss": 0.09767289459705353, "lr": 8.853796522318739e-06, "epoch": 1.7496503496503495, "percentage": 58.31, "elapsed_time": "6:13:57", "remaining_time": "4:27:19"} +{"current_steps": 3129, "total_steps": 5364, "loss": 0.05795377492904663, "lr": 8.847331284743724e-06, "epoch": 1.7502097902097902, "percentage": 58.33, "elapsed_time": "6:14:04", "remaining_time": "4:27:11"} +{"current_steps": 3130, "total_steps": 5364, "loss": 0.3228636682033539, "lr": 8.840866535427016e-06, "epoch": 1.7507692307692309, "percentage": 58.35, "elapsed_time": "6:14:11", "remaining_time": "4:27:04"} +{"current_steps": 3131, "total_steps": 5364, "loss": 0.1834675818681717, "lr": 8.834402277107008e-06, "epoch": 1.7513286713286713, "percentage": 58.37, "elapsed_time": "6:14:18", "remaining_time": "4:26:57"} +{"current_steps": 3132, "total_steps": 5364, "loss": 0.16292870044708252, "lr": 8.8279385125219e-06, "epoch": 1.7518881118881118, "percentage": 58.39, "elapsed_time": "6:14:25", "remaining_time": "4:26:49"} +{"current_steps": 3133, "total_steps": 5364, "loss": 0.3006657660007477, "lr": 8.821475244409668e-06, "epoch": 1.7524475524475525, "percentage": 58.41, "elapsed_time": "6:14:32", "remaining_time": "4:26:42"} +{"current_steps": 3134, "total_steps": 5364, "loss": 0.2342771738767624, "lr": 8.815012475508088e-06, "epoch": 1.7530069930069931, "percentage": 58.43, "elapsed_time": "6:14:39", "remaining_time": "4:26:35"} +{"current_steps": 3135, "total_steps": 5364, "loss": 0.0401204377412796, "lr": 8.808550208554718e-06, "epoch": 1.7535664335664336, "percentage": 58.45, "elapsed_time": "6:14:46", "remaining_time": "4:26:28"} +{"current_steps": 3136, "total_steps": 5364, "loss": 0.2918890118598938, "lr": 8.802088446286903e-06, "epoch": 1.754125874125874, "percentage": 58.46, "elapsed_time": "6:14:53", "remaining_time": "4:26:20"} +{"current_steps": 3137, "total_steps": 5364, "loss": 0.0670238733291626, "lr": 8.795627191441787e-06, "epoch": 1.7546853146853145, "percentage": 58.48, "elapsed_time": "6:15:01", "remaining_time": "4:26:13"} +{"current_steps": 3138, "total_steps": 5364, "loss": 0.11483792960643768, "lr": 8.78916644675628e-06, "epoch": 1.7552447552447552, "percentage": 58.5, "elapsed_time": "6:15:08", "remaining_time": "4:26:06"} +{"current_steps": 3139, "total_steps": 5364, "loss": 0.09549957513809204, "lr": 8.782706214967092e-06, "epoch": 1.7558041958041959, "percentage": 58.52, "elapsed_time": "6:15:15", "remaining_time": "4:25:59"} +{"current_steps": 3140, "total_steps": 5364, "loss": 0.14463545382022858, "lr": 8.776246498810704e-06, "epoch": 1.7563636363636363, "percentage": 58.54, "elapsed_time": "6:15:22", "remaining_time": "4:25:51"} +{"current_steps": 3141, "total_steps": 5364, "loss": 0.16351813077926636, "lr": 8.769787301023383e-06, "epoch": 1.7569230769230768, "percentage": 58.56, "elapsed_time": "6:15:29", "remaining_time": "4:25:44"} +{"current_steps": 3142, "total_steps": 5364, "loss": 0.25015830993652344, "lr": 8.763328624341183e-06, "epoch": 1.7574825174825175, "percentage": 58.58, "elapsed_time": "6:15:36", "remaining_time": "4:25:37"} +{"current_steps": 3143, "total_steps": 5364, "loss": 0.02947881817817688, "lr": 8.756870471499927e-06, "epoch": 1.7580419580419582, "percentage": 58.59, "elapsed_time": "6:15:43", "remaining_time": "4:25:30"} +{"current_steps": 3144, "total_steps": 5364, "loss": 0.1618751883506775, "lr": 8.750412845235221e-06, "epoch": 1.7586013986013986, "percentage": 58.61, "elapsed_time": "6:15:50", "remaining_time": "4:25:23"} +{"current_steps": 3145, "total_steps": 5364, "loss": 0.24477940797805786, "lr": 8.74395574828245e-06, "epoch": 1.759160839160839, "percentage": 58.63, "elapsed_time": "6:15:57", "remaining_time": "4:25:15"} +{"current_steps": 3146, "total_steps": 5364, "loss": 0.24421736598014832, "lr": 8.737499183376763e-06, "epoch": 1.7597202797202796, "percentage": 58.65, "elapsed_time": "6:16:05", "remaining_time": "4:25:09"} +{"current_steps": 3147, "total_steps": 5364, "loss": 0.10669678449630737, "lr": 8.731043153253105e-06, "epoch": 1.7602797202797202, "percentage": 58.67, "elapsed_time": "6:16:12", "remaining_time": "4:25:01"} +{"current_steps": 3148, "total_steps": 5364, "loss": 0.19479043781757355, "lr": 8.724587660646175e-06, "epoch": 1.760839160839161, "percentage": 58.69, "elapsed_time": "6:16:19", "remaining_time": "4:24:54"} +{"current_steps": 3149, "total_steps": 5364, "loss": 0.2415938377380371, "lr": 8.718132708290454e-06, "epoch": 1.7613986013986014, "percentage": 58.71, "elapsed_time": "6:16:26", "remaining_time": "4:24:47"} +{"current_steps": 3150, "total_steps": 5364, "loss": 0.268382728099823, "lr": 8.711678298920187e-06, "epoch": 1.7619580419580418, "percentage": 58.72, "elapsed_time": "6:16:32", "remaining_time": "4:24:39"} +{"current_steps": 3151, "total_steps": 5364, "loss": 0.15624383091926575, "lr": 8.705224435269405e-06, "epoch": 1.7625174825174825, "percentage": 58.74, "elapsed_time": "6:16:39", "remaining_time": "4:24:32"} +{"current_steps": 3152, "total_steps": 5364, "loss": 0.06342122703790665, "lr": 8.698771120071887e-06, "epoch": 1.7630769230769232, "percentage": 58.76, "elapsed_time": "6:16:46", "remaining_time": "4:24:24"} +{"current_steps": 3153, "total_steps": 5364, "loss": 0.18470287322998047, "lr": 8.692318356061186e-06, "epoch": 1.7636363636363637, "percentage": 58.78, "elapsed_time": "6:16:53", "remaining_time": "4:24:17"} +{"current_steps": 3154, "total_steps": 5364, "loss": 0.24003052711486816, "lr": 8.685866145970634e-06, "epoch": 1.7641958041958041, "percentage": 58.8, "elapsed_time": "6:17:00", "remaining_time": "4:24:10"} +{"current_steps": 3155, "total_steps": 5364, "loss": 0.27781444787979126, "lr": 8.679414492533315e-06, "epoch": 1.7647552447552446, "percentage": 58.82, "elapsed_time": "6:17:07", "remaining_time": "4:24:02"} +{"current_steps": 3156, "total_steps": 5364, "loss": 0.3203977346420288, "lr": 8.672963398482081e-06, "epoch": 1.7653146853146853, "percentage": 58.84, "elapsed_time": "6:17:15", "remaining_time": "4:23:55"} +{"current_steps": 3157, "total_steps": 5364, "loss": 0.14385612308979034, "lr": 8.66651286654955e-06, "epoch": 1.765874125874126, "percentage": 58.86, "elapsed_time": "6:17:21", "remaining_time": "4:23:48"} +{"current_steps": 3158, "total_steps": 5364, "loss": 0.06813342869281769, "lr": 8.660062899468093e-06, "epoch": 1.7664335664335664, "percentage": 58.87, "elapsed_time": "6:17:29", "remaining_time": "4:23:41"} +{"current_steps": 3159, "total_steps": 5364, "loss": 0.09013888239860535, "lr": 8.653613499969856e-06, "epoch": 1.7669930069930069, "percentage": 58.89, "elapsed_time": "6:17:36", "remaining_time": "4:23:34"} +{"current_steps": 3160, "total_steps": 5364, "loss": 0.2302234023809433, "lr": 8.64716467078673e-06, "epoch": 1.7675524475524476, "percentage": 58.91, "elapsed_time": "6:17:43", "remaining_time": "4:23:26"} +{"current_steps": 3161, "total_steps": 5364, "loss": 0.09541988372802734, "lr": 8.640716414650377e-06, "epoch": 1.7681118881118882, "percentage": 58.93, "elapsed_time": "6:17:50", "remaining_time": "4:23:19"} +{"current_steps": 3162, "total_steps": 5364, "loss": 0.12828890979290009, "lr": 8.634268734292205e-06, "epoch": 1.7686713286713287, "percentage": 58.95, "elapsed_time": "6:17:57", "remaining_time": "4:23:12"} +{"current_steps": 3163, "total_steps": 5364, "loss": 0.1581895500421524, "lr": 8.627821632443382e-06, "epoch": 1.7692307692307692, "percentage": 58.97, "elapsed_time": "6:18:04", "remaining_time": "4:23:05"} +{"current_steps": 3164, "total_steps": 5364, "loss": 0.08003498613834381, "lr": 8.62137511183484e-06, "epoch": 1.7697902097902098, "percentage": 58.99, "elapsed_time": "6:18:11", "remaining_time": "4:22:57"} +{"current_steps": 3165, "total_steps": 5364, "loss": 0.1973947286605835, "lr": 8.614929175197251e-06, "epoch": 1.7703496503496503, "percentage": 59.0, "elapsed_time": "6:18:18", "remaining_time": "4:22:50"} +{"current_steps": 3166, "total_steps": 5364, "loss": 0.10866017639636993, "lr": 8.608483825261049e-06, "epoch": 1.770909090909091, "percentage": 59.02, "elapsed_time": "6:18:26", "remaining_time": "4:22:43"} +{"current_steps": 3167, "total_steps": 5364, "loss": 0.1250346601009369, "lr": 8.602039064756416e-06, "epoch": 1.7714685314685314, "percentage": 59.04, "elapsed_time": "6:18:33", "remaining_time": "4:22:36"} +{"current_steps": 3168, "total_steps": 5364, "loss": 0.28763458132743835, "lr": 8.595594896413277e-06, "epoch": 1.772027972027972, "percentage": 59.06, "elapsed_time": "6:18:40", "remaining_time": "4:22:29"} +{"current_steps": 3169, "total_steps": 5364, "loss": 0.05523074418306351, "lr": 8.589151322961325e-06, "epoch": 1.7725874125874126, "percentage": 59.08, "elapsed_time": "6:18:47", "remaining_time": "4:22:22"} +{"current_steps": 3170, "total_steps": 5364, "loss": 0.2607904076576233, "lr": 8.582708347129984e-06, "epoch": 1.7731468531468533, "percentage": 59.1, "elapsed_time": "6:18:54", "remaining_time": "4:22:14"} +{"current_steps": 3171, "total_steps": 5364, "loss": 0.18346716463565826, "lr": 8.576265971648432e-06, "epoch": 1.7737062937062937, "percentage": 59.12, "elapsed_time": "6:19:01", "remaining_time": "4:22:07"} +{"current_steps": 3172, "total_steps": 5364, "loss": 0.09570343792438507, "lr": 8.569824199245586e-06, "epoch": 1.7742657342657342, "percentage": 59.13, "elapsed_time": "6:19:09", "remaining_time": "4:22:00"} +{"current_steps": 3173, "total_steps": 5364, "loss": 0.1438939869403839, "lr": 8.563383032650122e-06, "epoch": 1.7748251748251749, "percentage": 59.15, "elapsed_time": "6:19:16", "remaining_time": "4:21:53"} +{"current_steps": 3174, "total_steps": 5364, "loss": 0.06853242218494415, "lr": 8.556942474590447e-06, "epoch": 1.7753846153846153, "percentage": 59.17, "elapsed_time": "6:19:23", "remaining_time": "4:21:46"} +{"current_steps": 3175, "total_steps": 5364, "loss": 0.20595574378967285, "lr": 8.550502527794706e-06, "epoch": 1.775944055944056, "percentage": 59.19, "elapsed_time": "6:19:29", "remaining_time": "4:21:38"} +{"current_steps": 3176, "total_steps": 5364, "loss": 0.2571282684803009, "lr": 8.544063194990803e-06, "epoch": 1.7765034965034965, "percentage": 59.21, "elapsed_time": "6:19:37", "remaining_time": "4:21:31"} +{"current_steps": 3177, "total_steps": 5364, "loss": 0.10524804890155792, "lr": 8.537624478906364e-06, "epoch": 1.777062937062937, "percentage": 59.23, "elapsed_time": "6:19:44", "remaining_time": "4:21:24"} +{"current_steps": 3178, "total_steps": 5364, "loss": 0.0796380415558815, "lr": 8.531186382268768e-06, "epoch": 1.7776223776223776, "percentage": 59.25, "elapsed_time": "6:19:51", "remaining_time": "4:21:17"} +{"current_steps": 3179, "total_steps": 5364, "loss": 0.17871351540088654, "lr": 8.52474890780512e-06, "epoch": 1.7781818181818183, "percentage": 59.27, "elapsed_time": "6:19:58", "remaining_time": "4:21:09"} +{"current_steps": 3180, "total_steps": 5364, "loss": 0.19081605970859528, "lr": 8.518312058242262e-06, "epoch": 1.7787412587412588, "percentage": 59.28, "elapsed_time": "6:20:05", "remaining_time": "4:21:02"} +{"current_steps": 3181, "total_steps": 5364, "loss": 0.13260558247566223, "lr": 8.511875836306785e-06, "epoch": 1.7793006993006992, "percentage": 59.3, "elapsed_time": "6:20:12", "remaining_time": "4:20:55"} +{"current_steps": 3182, "total_steps": 5364, "loss": 0.12255693227052689, "lr": 8.505440244724999e-06, "epoch": 1.77986013986014, "percentage": 59.32, "elapsed_time": "6:20:19", "remaining_time": "4:20:48"} +{"current_steps": 3183, "total_steps": 5364, "loss": 0.1830892264842987, "lr": 8.499005286222956e-06, "epoch": 1.7804195804195804, "percentage": 59.34, "elapsed_time": "6:20:27", "remaining_time": "4:20:41"} +{"current_steps": 3184, "total_steps": 5364, "loss": 0.11568547785282135, "lr": 8.492570963526434e-06, "epoch": 1.780979020979021, "percentage": 59.36, "elapsed_time": "6:20:34", "remaining_time": "4:20:34"} +{"current_steps": 3185, "total_steps": 5364, "loss": 0.18830077350139618, "lr": 8.48613727936094e-06, "epoch": 1.7815384615384615, "percentage": 59.38, "elapsed_time": "6:20:41", "remaining_time": "4:20:26"} +{"current_steps": 3186, "total_steps": 5364, "loss": 0.11719164997339249, "lr": 8.479704236451726e-06, "epoch": 1.782097902097902, "percentage": 59.4, "elapsed_time": "6:20:48", "remaining_time": "4:20:19"} +{"current_steps": 3187, "total_steps": 5364, "loss": 0.1845027208328247, "lr": 8.473271837523752e-06, "epoch": 1.7826573426573427, "percentage": 59.41, "elapsed_time": "6:20:55", "remaining_time": "4:20:12"} +{"current_steps": 3188, "total_steps": 5364, "loss": 0.18442830443382263, "lr": 8.466840085301716e-06, "epoch": 1.7832167832167833, "percentage": 59.43, "elapsed_time": "6:21:02", "remaining_time": "4:20:04"} +{"current_steps": 3189, "total_steps": 5364, "loss": 0.13422854244709015, "lr": 8.46040898251004e-06, "epoch": 1.7837762237762238, "percentage": 59.45, "elapsed_time": "6:21:09", "remaining_time": "4:19:57"} +{"current_steps": 3190, "total_steps": 5364, "loss": 0.25387439131736755, "lr": 8.453978531872874e-06, "epoch": 1.7843356643356643, "percentage": 59.47, "elapsed_time": "6:21:16", "remaining_time": "4:19:50"} +{"current_steps": 3191, "total_steps": 5364, "loss": 0.1822403222322464, "lr": 8.447548736114091e-06, "epoch": 1.784895104895105, "percentage": 59.49, "elapsed_time": "6:21:23", "remaining_time": "4:19:42"} +{"current_steps": 3192, "total_steps": 5364, "loss": 0.15814903378486633, "lr": 8.441119597957276e-06, "epoch": 1.7854545454545454, "percentage": 59.51, "elapsed_time": "6:21:30", "remaining_time": "4:19:35"} +{"current_steps": 3193, "total_steps": 5364, "loss": 0.20812077820301056, "lr": 8.434691120125755e-06, "epoch": 1.786013986013986, "percentage": 59.53, "elapsed_time": "6:21:38", "remaining_time": "4:19:29"} +{"current_steps": 3194, "total_steps": 5364, "loss": 0.2235710322856903, "lr": 8.42826330534255e-06, "epoch": 1.7865734265734265, "percentage": 59.55, "elapsed_time": "6:21:45", "remaining_time": "4:19:21"} +{"current_steps": 3195, "total_steps": 5364, "loss": 0.06347090005874634, "lr": 8.42183615633043e-06, "epoch": 1.787132867132867, "percentage": 59.56, "elapsed_time": "6:21:52", "remaining_time": "4:19:14"} +{"current_steps": 3196, "total_steps": 5364, "loss": 0.21188542246818542, "lr": 8.41540967581186e-06, "epoch": 1.7876923076923077, "percentage": 59.58, "elapsed_time": "6:21:59", "remaining_time": "4:19:07"} +{"current_steps": 3197, "total_steps": 5364, "loss": 0.07785777747631073, "lr": 8.408983866509027e-06, "epoch": 1.7882517482517484, "percentage": 59.6, "elapsed_time": "6:22:06", "remaining_time": "4:19:00"} +{"current_steps": 3198, "total_steps": 5364, "loss": 0.10463696718215942, "lr": 8.40255873114384e-06, "epoch": 1.7888111888111888, "percentage": 59.62, "elapsed_time": "6:22:13", "remaining_time": "4:18:52"} +{"current_steps": 3199, "total_steps": 5364, "loss": 0.12862011790275574, "lr": 8.396134272437914e-06, "epoch": 1.7893706293706293, "percentage": 59.64, "elapsed_time": "6:22:20", "remaining_time": "4:18:45"} +{"current_steps": 3200, "total_steps": 5364, "loss": 0.21206022799015045, "lr": 8.389710493112593e-06, "epoch": 1.78993006993007, "percentage": 59.66, "elapsed_time": "6:22:27", "remaining_time": "4:18:38"} +{"current_steps": 3201, "total_steps": 5364, "loss": 0.10065078735351562, "lr": 8.383287395888913e-06, "epoch": 1.7904895104895104, "percentage": 59.68, "elapsed_time": "6:22:34", "remaining_time": "4:18:30"} +{"current_steps": 3202, "total_steps": 5364, "loss": 0.2102937549352646, "lr": 8.376864983487627e-06, "epoch": 1.7910489510489511, "percentage": 59.69, "elapsed_time": "6:22:41", "remaining_time": "4:18:23"} +{"current_steps": 3203, "total_steps": 5364, "loss": 0.12319180369377136, "lr": 8.370443258629208e-06, "epoch": 1.7916083916083916, "percentage": 59.71, "elapsed_time": "6:22:49", "remaining_time": "4:18:16"} +{"current_steps": 3204, "total_steps": 5364, "loss": 0.06279011070728302, "lr": 8.364022224033828e-06, "epoch": 1.792167832167832, "percentage": 59.73, "elapsed_time": "6:22:56", "remaining_time": "4:18:09"} +{"current_steps": 3205, "total_steps": 5364, "loss": 0.12921187281608582, "lr": 8.357601882421373e-06, "epoch": 1.7927272727272727, "percentage": 59.75, "elapsed_time": "6:23:03", "remaining_time": "4:18:02"} +{"current_steps": 3206, "total_steps": 5364, "loss": 0.30726486444473267, "lr": 8.35118223651143e-06, "epoch": 1.7932867132867134, "percentage": 59.77, "elapsed_time": "6:23:10", "remaining_time": "4:17:55"} +{"current_steps": 3207, "total_steps": 5364, "loss": 0.03763486444950104, "lr": 8.344763289023287e-06, "epoch": 1.7938461538461539, "percentage": 59.79, "elapsed_time": "6:23:17", "remaining_time": "4:17:48"} +{"current_steps": 3208, "total_steps": 5364, "loss": 0.14397838711738586, "lr": 8.338345042675953e-06, "epoch": 1.7944055944055943, "percentage": 59.81, "elapsed_time": "6:23:24", "remaining_time": "4:17:40"} +{"current_steps": 3209, "total_steps": 5364, "loss": 0.1817794144153595, "lr": 8.33192750018812e-06, "epoch": 1.794965034965035, "percentage": 59.82, "elapsed_time": "6:23:31", "remaining_time": "4:17:33"} +{"current_steps": 3210, "total_steps": 5364, "loss": 0.10801716148853302, "lr": 8.3255106642782e-06, "epoch": 1.7955244755244755, "percentage": 59.84, "elapsed_time": "6:23:38", "remaining_time": "4:17:26"} +{"current_steps": 3211, "total_steps": 5364, "loss": 0.05909350514411926, "lr": 8.319094537664289e-06, "epoch": 1.7960839160839162, "percentage": 59.86, "elapsed_time": "6:23:46", "remaining_time": "4:17:19"} +{"current_steps": 3212, "total_steps": 5364, "loss": 0.13181468844413757, "lr": 8.312679123064197e-06, "epoch": 1.7966433566433566, "percentage": 59.88, "elapsed_time": "6:23:53", "remaining_time": "4:17:12"} +{"current_steps": 3213, "total_steps": 5364, "loss": 0.12119457870721817, "lr": 8.306264423195423e-06, "epoch": 1.797202797202797, "percentage": 59.9, "elapsed_time": "6:24:00", "remaining_time": "4:17:04"} +{"current_steps": 3214, "total_steps": 5364, "loss": 0.09390321373939514, "lr": 8.299850440775163e-06, "epoch": 1.7977622377622378, "percentage": 59.92, "elapsed_time": "6:24:07", "remaining_time": "4:16:57"} +{"current_steps": 3215, "total_steps": 5364, "loss": 0.17034342885017395, "lr": 8.29343717852032e-06, "epoch": 1.7983216783216784, "percentage": 59.94, "elapsed_time": "6:24:14", "remaining_time": "4:16:50"} +{"current_steps": 3216, "total_steps": 5364, "loss": 0.10654303431510925, "lr": 8.287024639147475e-06, "epoch": 1.798881118881119, "percentage": 59.96, "elapsed_time": "6:24:21", "remaining_time": "4:16:42"} +{"current_steps": 3217, "total_steps": 5364, "loss": 0.06967205554246902, "lr": 8.28061282537292e-06, "epoch": 1.7994405594405594, "percentage": 59.97, "elapsed_time": "6:24:28", "remaining_time": "4:16:35"} +{"current_steps": 3218, "total_steps": 5364, "loss": 0.03199368715286255, "lr": 8.27420173991263e-06, "epoch": 1.8, "percentage": 59.99, "elapsed_time": "6:24:35", "remaining_time": "4:16:28"} +{"current_steps": 3219, "total_steps": 5364, "loss": 0.05807441845536232, "lr": 8.267791385482268e-06, "epoch": 1.8005594405594405, "percentage": 60.01, "elapsed_time": "6:24:42", "remaining_time": "4:16:20"} +{"current_steps": 3220, "total_steps": 5364, "loss": 0.15787997841835022, "lr": 8.261381764797198e-06, "epoch": 1.8011188811188812, "percentage": 60.03, "elapsed_time": "6:24:49", "remaining_time": "4:16:13"} +{"current_steps": 3221, "total_steps": 5364, "loss": 0.10627515614032745, "lr": 8.254972880572463e-06, "epoch": 1.8016783216783216, "percentage": 60.05, "elapsed_time": "6:24:56", "remaining_time": "4:16:06"} +{"current_steps": 3222, "total_steps": 5364, "loss": 0.0636792704463005, "lr": 8.248564735522804e-06, "epoch": 1.802237762237762, "percentage": 60.07, "elapsed_time": "6:25:02", "remaining_time": "4:15:58"} +{"current_steps": 3223, "total_steps": 5364, "loss": 0.05325791984796524, "lr": 8.242157332362643e-06, "epoch": 1.8027972027972028, "percentage": 60.09, "elapsed_time": "6:25:09", "remaining_time": "4:15:51"} +{"current_steps": 3224, "total_steps": 5364, "loss": 0.09092092514038086, "lr": 8.235750673806082e-06, "epoch": 1.8033566433566435, "percentage": 60.1, "elapsed_time": "6:25:17", "remaining_time": "4:15:44"} +{"current_steps": 3225, "total_steps": 5364, "loss": 0.18818290531635284, "lr": 8.229344762566921e-06, "epoch": 1.803916083916084, "percentage": 60.12, "elapsed_time": "6:25:24", "remaining_time": "4:15:37"} +{"current_steps": 3226, "total_steps": 5364, "loss": 0.21693797409534454, "lr": 8.222939601358633e-06, "epoch": 1.8044755244755244, "percentage": 60.14, "elapsed_time": "6:25:31", "remaining_time": "4:15:30"} +{"current_steps": 3227, "total_steps": 5364, "loss": 0.10395072400569916, "lr": 8.216535192894378e-06, "epoch": 1.805034965034965, "percentage": 60.16, "elapsed_time": "6:25:39", "remaining_time": "4:15:23"} +{"current_steps": 3228, "total_steps": 5364, "loss": 0.12642090022563934, "lr": 8.210131539886992e-06, "epoch": 1.8055944055944058, "percentage": 60.18, "elapsed_time": "6:25:47", "remaining_time": "4:15:16"} +{"current_steps": 3229, "total_steps": 5364, "loss": 0.11497074365615845, "lr": 8.203728645049002e-06, "epoch": 1.8061538461538462, "percentage": 60.2, "elapsed_time": "6:25:54", "remaining_time": "4:15:09"} +{"current_steps": 3230, "total_steps": 5364, "loss": 0.20979063212871552, "lr": 8.197326511092604e-06, "epoch": 1.8067132867132867, "percentage": 60.22, "elapsed_time": "6:26:01", "remaining_time": "4:15:02"} +{"current_steps": 3231, "total_steps": 5364, "loss": 0.044171929359436035, "lr": 8.19092514072967e-06, "epoch": 1.8072727272727271, "percentage": 60.23, "elapsed_time": "6:26:08", "remaining_time": "4:14:54"} +{"current_steps": 3232, "total_steps": 5364, "loss": 0.04106922075152397, "lr": 8.18452453667176e-06, "epoch": 1.8078321678321678, "percentage": 60.25, "elapsed_time": "6:26:15", "remaining_time": "4:14:47"} +{"current_steps": 3233, "total_steps": 5364, "loss": 0.22721415758132935, "lr": 8.178124701630092e-06, "epoch": 1.8083916083916085, "percentage": 60.27, "elapsed_time": "6:26:22", "remaining_time": "4:14:40"} +{"current_steps": 3234, "total_steps": 5364, "loss": 0.06986412405967712, "lr": 8.17172563831558e-06, "epoch": 1.808951048951049, "percentage": 60.29, "elapsed_time": "6:26:29", "remaining_time": "4:14:33"} +{"current_steps": 3235, "total_steps": 5364, "loss": 0.23333466053009033, "lr": 8.165327349438796e-06, "epoch": 1.8095104895104894, "percentage": 60.31, "elapsed_time": "6:26:36", "remaining_time": "4:14:26"} +{"current_steps": 3236, "total_steps": 5364, "loss": 0.0997273251414299, "lr": 8.158929837709983e-06, "epoch": 1.81006993006993, "percentage": 60.33, "elapsed_time": "6:26:43", "remaining_time": "4:14:18"} +{"current_steps": 3237, "total_steps": 5364, "loss": 0.3249717950820923, "lr": 8.152533105839066e-06, "epoch": 1.8106293706293708, "percentage": 60.35, "elapsed_time": "6:26:50", "remaining_time": "4:14:11"} +{"current_steps": 3238, "total_steps": 5364, "loss": 0.1833459585905075, "lr": 8.146137156535625e-06, "epoch": 1.8111888111888113, "percentage": 60.37, "elapsed_time": "6:26:57", "remaining_time": "4:14:04"} +{"current_steps": 3239, "total_steps": 5364, "loss": 0.2868287265300751, "lr": 8.139741992508928e-06, "epoch": 1.8117482517482517, "percentage": 60.38, "elapsed_time": "6:27:04", "remaining_time": "4:13:57"} +{"current_steps": 3240, "total_steps": 5364, "loss": 0.1585218608379364, "lr": 8.13334761646789e-06, "epoch": 1.8123076923076922, "percentage": 60.4, "elapsed_time": "6:27:11", "remaining_time": "4:13:49"} +{"current_steps": 3241, "total_steps": 5364, "loss": 0.12568265199661255, "lr": 8.126954031121103e-06, "epoch": 1.8128671328671329, "percentage": 60.42, "elapsed_time": "6:27:18", "remaining_time": "4:13:42"} +{"current_steps": 3242, "total_steps": 5364, "loss": 0.13871999084949493, "lr": 8.120561239176824e-06, "epoch": 1.8134265734265735, "percentage": 60.44, "elapsed_time": "6:27:25", "remaining_time": "4:13:35"} +{"current_steps": 3243, "total_steps": 5364, "loss": 0.019778726622462273, "lr": 8.114169243342968e-06, "epoch": 1.813986013986014, "percentage": 60.46, "elapsed_time": "6:27:33", "remaining_time": "4:13:28"} +{"current_steps": 3244, "total_steps": 5364, "loss": 0.07132961601018906, "lr": 8.107778046327126e-06, "epoch": 1.8145454545454545, "percentage": 60.48, "elapsed_time": "6:27:40", "remaining_time": "4:13:20"} +{"current_steps": 3245, "total_steps": 5364, "loss": 0.3750250041484833, "lr": 8.101387650836531e-06, "epoch": 1.8151048951048951, "percentage": 60.5, "elapsed_time": "6:27:47", "remaining_time": "4:13:13"} +{"current_steps": 3246, "total_steps": 5364, "loss": 0.18472930788993835, "lr": 8.094998059578096e-06, "epoch": 1.8156643356643358, "percentage": 60.51, "elapsed_time": "6:27:54", "remaining_time": "4:13:06"} +{"current_steps": 3247, "total_steps": 5364, "loss": 0.1066703274846077, "lr": 8.088609275258381e-06, "epoch": 1.8162237762237763, "percentage": 60.53, "elapsed_time": "6:28:01", "remaining_time": "4:12:59"} +{"current_steps": 3248, "total_steps": 5364, "loss": 0.2648044228553772, "lr": 8.082221300583603e-06, "epoch": 1.8167832167832167, "percentage": 60.55, "elapsed_time": "6:28:08", "remaining_time": "4:12:51"} +{"current_steps": 3249, "total_steps": 5364, "loss": 0.06476524472236633, "lr": 8.075834138259648e-06, "epoch": 1.8173426573426572, "percentage": 60.57, "elapsed_time": "6:28:15", "remaining_time": "4:12:44"} +{"current_steps": 3250, "total_steps": 5364, "loss": 0.2636910378932953, "lr": 8.069447790992044e-06, "epoch": 1.817902097902098, "percentage": 60.59, "elapsed_time": "6:28:22", "remaining_time": "4:12:37"} +{"current_steps": 3251, "total_steps": 5364, "loss": 0.15321972966194153, "lr": 8.063062261485988e-06, "epoch": 1.8184615384615386, "percentage": 60.61, "elapsed_time": "6:28:29", "remaining_time": "4:12:30"} +{"current_steps": 3252, "total_steps": 5364, "loss": 0.05630672723054886, "lr": 8.056677552446317e-06, "epoch": 1.819020979020979, "percentage": 60.63, "elapsed_time": "6:28:36", "remaining_time": "4:12:23"} +{"current_steps": 3253, "total_steps": 5364, "loss": 0.09161169826984406, "lr": 8.050293666577527e-06, "epoch": 1.8195804195804195, "percentage": 60.65, "elapsed_time": "6:28:44", "remaining_time": "4:12:15"} +{"current_steps": 3254, "total_steps": 5364, "loss": 0.0998765230178833, "lr": 8.043910606583767e-06, "epoch": 1.8201398601398602, "percentage": 60.66, "elapsed_time": "6:28:51", "remaining_time": "4:12:08"} +{"current_steps": 3255, "total_steps": 5364, "loss": 0.32714977860450745, "lr": 8.037528375168828e-06, "epoch": 1.8206993006993009, "percentage": 60.68, "elapsed_time": "6:28:58", "remaining_time": "4:12:01"} +{"current_steps": 3256, "total_steps": 5364, "loss": 0.13106168806552887, "lr": 8.031146975036164e-06, "epoch": 1.8212587412587413, "percentage": 60.7, "elapsed_time": "6:29:05", "remaining_time": "4:11:54"} +{"current_steps": 3257, "total_steps": 5364, "loss": 0.15179291367530823, "lr": 8.024766408888867e-06, "epoch": 1.8218181818181818, "percentage": 60.72, "elapsed_time": "6:29:12", "remaining_time": "4:11:47"} +{"current_steps": 3258, "total_steps": 5364, "loss": 0.023837648332118988, "lr": 8.018386679429673e-06, "epoch": 1.8223776223776222, "percentage": 60.74, "elapsed_time": "6:29:20", "remaining_time": "4:11:40"} +{"current_steps": 3259, "total_steps": 5364, "loss": 0.2113943099975586, "lr": 8.012007789360972e-06, "epoch": 1.822937062937063, "percentage": 60.76, "elapsed_time": "6:29:27", "remaining_time": "4:11:32"} +{"current_steps": 3260, "total_steps": 5364, "loss": 0.3009991943836212, "lr": 8.00562974138479e-06, "epoch": 1.8234965034965036, "percentage": 60.78, "elapsed_time": "6:29:34", "remaining_time": "4:11:25"} +{"current_steps": 3261, "total_steps": 5364, "loss": 0.08428531885147095, "lr": 7.999252538202807e-06, "epoch": 1.824055944055944, "percentage": 60.79, "elapsed_time": "6:29:41", "remaining_time": "4:11:18"} +{"current_steps": 3262, "total_steps": 5364, "loss": 0.11041277647018433, "lr": 7.992876182516339e-06, "epoch": 1.8246153846153845, "percentage": 60.81, "elapsed_time": "6:29:48", "remaining_time": "4:11:11"} +{"current_steps": 3263, "total_steps": 5364, "loss": 0.17227564752101898, "lr": 7.986500677026334e-06, "epoch": 1.8251748251748252, "percentage": 60.83, "elapsed_time": "6:29:55", "remaining_time": "4:11:04"} +{"current_steps": 3264, "total_steps": 5364, "loss": 0.19148889183998108, "lr": 7.980126024433398e-06, "epoch": 1.825734265734266, "percentage": 60.85, "elapsed_time": "6:30:02", "remaining_time": "4:10:56"} +{"current_steps": 3265, "total_steps": 5364, "loss": 0.1837417185306549, "lr": 7.97375222743776e-06, "epoch": 1.8262937062937064, "percentage": 60.87, "elapsed_time": "6:30:09", "remaining_time": "4:10:49"} +{"current_steps": 3266, "total_steps": 5364, "loss": 0.2764427065849304, "lr": 7.9673792887393e-06, "epoch": 1.8268531468531468, "percentage": 60.89, "elapsed_time": "6:30:16", "remaining_time": "4:10:41"} +{"current_steps": 3267, "total_steps": 5364, "loss": 0.1576852649450302, "lr": 7.961007211037523e-06, "epoch": 1.8274125874125873, "percentage": 60.91, "elapsed_time": "6:30:23", "remaining_time": "4:10:34"} +{"current_steps": 3268, "total_steps": 5364, "loss": 0.06918426603078842, "lr": 7.954635997031577e-06, "epoch": 1.827972027972028, "percentage": 60.92, "elapsed_time": "6:30:30", "remaining_time": "4:10:27"} +{"current_steps": 3269, "total_steps": 5364, "loss": 0.11541706323623657, "lr": 7.94826564942024e-06, "epoch": 1.8285314685314686, "percentage": 60.94, "elapsed_time": "6:30:37", "remaining_time": "4:10:20"} +{"current_steps": 3270, "total_steps": 5364, "loss": 0.22836562991142273, "lr": 7.941896170901916e-06, "epoch": 1.829090909090909, "percentage": 60.96, "elapsed_time": "6:30:44", "remaining_time": "4:10:12"} +{"current_steps": 3271, "total_steps": 5364, "loss": 0.1463472694158554, "lr": 7.935527564174664e-06, "epoch": 1.8296503496503496, "percentage": 60.98, "elapsed_time": "6:30:52", "remaining_time": "4:10:06"} +{"current_steps": 3272, "total_steps": 5364, "loss": 0.19594229757785797, "lr": 7.929159831936143e-06, "epoch": 1.8302097902097902, "percentage": 61.0, "elapsed_time": "6:30:59", "remaining_time": "4:09:58"} +{"current_steps": 3273, "total_steps": 5364, "loss": 0.08977712690830231, "lr": 7.92279297688367e-06, "epoch": 1.830769230769231, "percentage": 61.02, "elapsed_time": "6:31:05", "remaining_time": "4:09:51"} +{"current_steps": 3274, "total_steps": 5364, "loss": 0.10286234319210052, "lr": 7.916427001714173e-06, "epoch": 1.8313286713286714, "percentage": 61.04, "elapsed_time": "6:31:13", "remaining_time": "4:09:44"} +{"current_steps": 3275, "total_steps": 5364, "loss": 0.1839471161365509, "lr": 7.910061909124209e-06, "epoch": 1.8318881118881118, "percentage": 61.06, "elapsed_time": "6:31:20", "remaining_time": "4:09:37"} +{"current_steps": 3276, "total_steps": 5364, "loss": 0.132829949259758, "lr": 7.903697701809967e-06, "epoch": 1.8324475524475523, "percentage": 61.07, "elapsed_time": "6:31:27", "remaining_time": "4:09:29"} +{"current_steps": 3277, "total_steps": 5364, "loss": 0.13574588298797607, "lr": 7.897334382467254e-06, "epoch": 1.833006993006993, "percentage": 61.09, "elapsed_time": "6:31:34", "remaining_time": "4:09:22"} +{"current_steps": 3278, "total_steps": 5364, "loss": 0.07292518019676208, "lr": 7.890971953791513e-06, "epoch": 1.8335664335664337, "percentage": 61.11, "elapsed_time": "6:31:41", "remaining_time": "4:09:15"} +{"current_steps": 3279, "total_steps": 5364, "loss": 0.11348556727170944, "lr": 7.884610418477796e-06, "epoch": 1.8341258741258741, "percentage": 61.13, "elapsed_time": "6:31:48", "remaining_time": "4:09:08"} +{"current_steps": 3280, "total_steps": 5364, "loss": 0.1410672664642334, "lr": 7.878249779220781e-06, "epoch": 1.8346853146853146, "percentage": 61.15, "elapsed_time": "6:31:55", "remaining_time": "4:09:00"} +{"current_steps": 3281, "total_steps": 5364, "loss": 0.038923539221286774, "lr": 7.871890038714772e-06, "epoch": 1.8352447552447553, "percentage": 61.17, "elapsed_time": "6:32:02", "remaining_time": "4:08:53"} +{"current_steps": 3282, "total_steps": 5364, "loss": 0.11697768419981003, "lr": 7.865531199653681e-06, "epoch": 1.835804195804196, "percentage": 61.19, "elapsed_time": "6:32:09", "remaining_time": "4:08:46"} +{"current_steps": 3283, "total_steps": 5364, "loss": 0.08933316171169281, "lr": 7.859173264731058e-06, "epoch": 1.8363636363636364, "percentage": 61.2, "elapsed_time": "6:32:16", "remaining_time": "4:08:39"} +{"current_steps": 3284, "total_steps": 5364, "loss": 0.17128747701644897, "lr": 7.852816236640042e-06, "epoch": 1.8369230769230769, "percentage": 61.22, "elapsed_time": "6:32:23", "remaining_time": "4:08:31"} +{"current_steps": 3285, "total_steps": 5364, "loss": 0.23782578110694885, "lr": 7.846460118073414e-06, "epoch": 1.8374825174825173, "percentage": 61.24, "elapsed_time": "6:32:30", "remaining_time": "4:08:24"} +{"current_steps": 3286, "total_steps": 5364, "loss": 0.06162901595234871, "lr": 7.840104911723556e-06, "epoch": 1.838041958041958, "percentage": 61.26, "elapsed_time": "6:32:38", "remaining_time": "4:08:17"} +{"current_steps": 3287, "total_steps": 5364, "loss": 0.14769190549850464, "lr": 7.833750620282459e-06, "epoch": 1.8386013986013987, "percentage": 61.28, "elapsed_time": "6:32:45", "remaining_time": "4:08:10"} +{"current_steps": 3288, "total_steps": 5364, "loss": 0.09462253004312515, "lr": 7.827397246441747e-06, "epoch": 1.8391608391608392, "percentage": 61.3, "elapsed_time": "6:32:52", "remaining_time": "4:08:03"} +{"current_steps": 3289, "total_steps": 5364, "loss": 0.36728331446647644, "lr": 7.82104479289263e-06, "epoch": 1.8397202797202796, "percentage": 61.32, "elapsed_time": "6:32:59", "remaining_time": "4:07:56"} +{"current_steps": 3290, "total_steps": 5364, "loss": 0.13168132305145264, "lr": 7.814693262325949e-06, "epoch": 1.8402797202797203, "percentage": 61.33, "elapsed_time": "6:33:06", "remaining_time": "4:07:48"} +{"current_steps": 3291, "total_steps": 5364, "loss": 0.11074627935886383, "lr": 7.808342657432141e-06, "epoch": 1.840839160839161, "percentage": 61.35, "elapsed_time": "6:33:13", "remaining_time": "4:07:41"} +{"current_steps": 3292, "total_steps": 5364, "loss": 0.2530818581581116, "lr": 7.801992980901252e-06, "epoch": 1.8413986013986015, "percentage": 61.37, "elapsed_time": "6:33:20", "remaining_time": "4:07:34"} +{"current_steps": 3293, "total_steps": 5364, "loss": 0.38504558801651, "lr": 7.795644235422946e-06, "epoch": 1.841958041958042, "percentage": 61.39, "elapsed_time": "6:33:27", "remaining_time": "4:07:26"} +{"current_steps": 3294, "total_steps": 5364, "loss": 0.0960349515080452, "lr": 7.789296423686479e-06, "epoch": 1.8425174825174824, "percentage": 61.41, "elapsed_time": "6:33:34", "remaining_time": "4:07:19"} +{"current_steps": 3295, "total_steps": 5364, "loss": 0.04795685410499573, "lr": 7.78294954838072e-06, "epoch": 1.843076923076923, "percentage": 61.43, "elapsed_time": "6:33:41", "remaining_time": "4:07:12"} +{"current_steps": 3296, "total_steps": 5364, "loss": 0.17522096633911133, "lr": 7.776603612194138e-06, "epoch": 1.8436363636363637, "percentage": 61.45, "elapsed_time": "6:33:48", "remaining_time": "4:07:05"} +{"current_steps": 3297, "total_steps": 5364, "loss": 0.10706233978271484, "lr": 7.7702586178148e-06, "epoch": 1.8441958041958042, "percentage": 61.47, "elapsed_time": "6:33:55", "remaining_time": "4:06:57"} +{"current_steps": 3298, "total_steps": 5364, "loss": 0.20432394742965698, "lr": 7.763914567930384e-06, "epoch": 1.8447552447552447, "percentage": 61.48, "elapsed_time": "6:34:02", "remaining_time": "4:06:50"} +{"current_steps": 3299, "total_steps": 5364, "loss": 0.09755592048168182, "lr": 7.757571465228158e-06, "epoch": 1.8453146853146853, "percentage": 61.5, "elapsed_time": "6:34:09", "remaining_time": "4:06:43"} +{"current_steps": 3300, "total_steps": 5364, "loss": 0.06318570673465729, "lr": 7.751229312395e-06, "epoch": 1.845874125874126, "percentage": 61.52, "elapsed_time": "6:34:16", "remaining_time": "4:06:35"} +{"current_steps": 3301, "total_steps": 5364, "loss": 0.17142446339130402, "lr": 7.744888112117372e-06, "epoch": 1.8464335664335665, "percentage": 61.54, "elapsed_time": "6:34:23", "remaining_time": "4:06:28"} +{"current_steps": 3302, "total_steps": 5364, "loss": 0.04919543117284775, "lr": 7.738547867081347e-06, "epoch": 1.846993006993007, "percentage": 61.56, "elapsed_time": "6:34:30", "remaining_time": "4:06:21"} +{"current_steps": 3303, "total_steps": 5364, "loss": 0.05155760422348976, "lr": 7.73220857997258e-06, "epoch": 1.8475524475524474, "percentage": 61.58, "elapsed_time": "6:34:37", "remaining_time": "4:06:14"} +{"current_steps": 3304, "total_steps": 5364, "loss": 0.25951826572418213, "lr": 7.725870253476323e-06, "epoch": 1.848111888111888, "percentage": 61.6, "elapsed_time": "6:34:44", "remaining_time": "4:06:07"} +{"current_steps": 3305, "total_steps": 5364, "loss": 0.21192558109760284, "lr": 7.719532890277436e-06, "epoch": 1.8486713286713288, "percentage": 61.61, "elapsed_time": "6:34:51", "remaining_time": "4:05:59"} +{"current_steps": 3306, "total_steps": 5364, "loss": 0.05738449841737747, "lr": 7.713196493060347e-06, "epoch": 1.8492307692307692, "percentage": 61.63, "elapsed_time": "6:34:58", "remaining_time": "4:05:52"} +{"current_steps": 3307, "total_steps": 5364, "loss": 0.019613808020949364, "lr": 7.706861064509095e-06, "epoch": 1.8497902097902097, "percentage": 61.65, "elapsed_time": "6:35:05", "remaining_time": "4:05:45"} +{"current_steps": 3308, "total_steps": 5364, "loss": 0.08770640194416046, "lr": 7.700526607307298e-06, "epoch": 1.8503496503496504, "percentage": 61.67, "elapsed_time": "6:35:12", "remaining_time": "4:05:37"} +{"current_steps": 3309, "total_steps": 5364, "loss": 0.10374884307384491, "lr": 7.694193124138161e-06, "epoch": 1.850909090909091, "percentage": 61.69, "elapsed_time": "6:35:19", "remaining_time": "4:05:30"} +{"current_steps": 3310, "total_steps": 5364, "loss": 0.16725939512252808, "lr": 7.68786061768449e-06, "epoch": 1.8514685314685315, "percentage": 61.71, "elapsed_time": "6:35:27", "remaining_time": "4:05:23"} +{"current_steps": 3311, "total_steps": 5364, "loss": 0.12029227614402771, "lr": 7.681529090628658e-06, "epoch": 1.852027972027972, "percentage": 61.73, "elapsed_time": "6:35:34", "remaining_time": "4:05:16"} +{"current_steps": 3312, "total_steps": 5364, "loss": 0.14128750562667847, "lr": 7.675198545652643e-06, "epoch": 1.8525874125874124, "percentage": 61.74, "elapsed_time": "6:35:41", "remaining_time": "4:05:09"} +{"current_steps": 3313, "total_steps": 5364, "loss": 0.03516771271824837, "lr": 7.668868985437988e-06, "epoch": 1.8531468531468531, "percentage": 61.76, "elapsed_time": "6:35:48", "remaining_time": "4:05:02"} +{"current_steps": 3314, "total_steps": 5364, "loss": 0.08889401704072952, "lr": 7.66254041266583e-06, "epoch": 1.8537062937062938, "percentage": 61.78, "elapsed_time": "6:35:56", "remaining_time": "4:04:55"} +{"current_steps": 3315, "total_steps": 5364, "loss": 0.07184748351573944, "lr": 7.65621283001689e-06, "epoch": 1.8542657342657343, "percentage": 61.8, "elapsed_time": "6:36:03", "remaining_time": "4:04:48"} +{"current_steps": 3316, "total_steps": 5364, "loss": 0.18813163042068481, "lr": 7.649886240171462e-06, "epoch": 1.8548251748251747, "percentage": 61.82, "elapsed_time": "6:36:10", "remaining_time": "4:04:40"} +{"current_steps": 3317, "total_steps": 5364, "loss": 0.20797763764858246, "lr": 7.643560645809423e-06, "epoch": 1.8553846153846154, "percentage": 61.84, "elapsed_time": "6:36:17", "remaining_time": "4:04:33"} +{"current_steps": 3318, "total_steps": 5364, "loss": 0.0401025116443634, "lr": 7.637236049610229e-06, "epoch": 1.855944055944056, "percentage": 61.86, "elapsed_time": "6:36:25", "remaining_time": "4:04:26"} +{"current_steps": 3319, "total_steps": 5364, "loss": 0.04069935530424118, "lr": 7.630912454252909e-06, "epoch": 1.8565034965034966, "percentage": 61.88, "elapsed_time": "6:36:32", "remaining_time": "4:04:19"} +{"current_steps": 3320, "total_steps": 5364, "loss": 0.23037636280059814, "lr": 7.6245898624160775e-06, "epoch": 1.857062937062937, "percentage": 61.89, "elapsed_time": "6:36:39", "remaining_time": "4:04:12"} +{"current_steps": 3321, "total_steps": 5364, "loss": 0.29209572076797485, "lr": 7.618268276777909e-06, "epoch": 1.8576223776223775, "percentage": 61.91, "elapsed_time": "6:36:46", "remaining_time": "4:04:05"} +{"current_steps": 3322, "total_steps": 5364, "loss": 0.09035668522119522, "lr": 7.611947700016169e-06, "epoch": 1.8581818181818182, "percentage": 61.93, "elapsed_time": "6:36:53", "remaining_time": "4:03:57"} +{"current_steps": 3323, "total_steps": 5364, "loss": 0.4222186207771301, "lr": 7.6056281348081815e-06, "epoch": 1.8587412587412588, "percentage": 61.95, "elapsed_time": "6:37:00", "remaining_time": "4:03:50"} +{"current_steps": 3324, "total_steps": 5364, "loss": 0.05275161936879158, "lr": 7.599309583830852e-06, "epoch": 1.8593006993006993, "percentage": 61.97, "elapsed_time": "6:37:07", "remaining_time": "4:03:43"} +{"current_steps": 3325, "total_steps": 5364, "loss": 0.1441236287355423, "lr": 7.592992049760649e-06, "epoch": 1.8598601398601398, "percentage": 61.99, "elapsed_time": "6:37:14", "remaining_time": "4:03:35"} +{"current_steps": 3326, "total_steps": 5364, "loss": 0.02539297193288803, "lr": 7.58667553527361e-06, "epoch": 1.8604195804195804, "percentage": 62.01, "elapsed_time": "6:37:21", "remaining_time": "4:03:28"} +{"current_steps": 3327, "total_steps": 5364, "loss": 0.09021177887916565, "lr": 7.580360043045351e-06, "epoch": 1.8609790209790211, "percentage": 62.02, "elapsed_time": "6:37:28", "remaining_time": "4:03:21"} +{"current_steps": 3328, "total_steps": 5364, "loss": 0.2206195592880249, "lr": 7.5740455757510434e-06, "epoch": 1.8615384615384616, "percentage": 62.04, "elapsed_time": "6:37:35", "remaining_time": "4:03:14"} +{"current_steps": 3329, "total_steps": 5364, "loss": 0.11760076880455017, "lr": 7.567732136065432e-06, "epoch": 1.862097902097902, "percentage": 62.06, "elapsed_time": "6:37:42", "remaining_time": "4:03:06"} +{"current_steps": 3330, "total_steps": 5364, "loss": 0.06637430191040039, "lr": 7.561419726662822e-06, "epoch": 1.8626573426573425, "percentage": 62.08, "elapsed_time": "6:37:49", "remaining_time": "4:02:59"} +{"current_steps": 3331, "total_steps": 5364, "loss": 0.1258738934993744, "lr": 7.5551083502170775e-06, "epoch": 1.8632167832167832, "percentage": 62.1, "elapsed_time": "6:37:56", "remaining_time": "4:02:52"} +{"current_steps": 3332, "total_steps": 5364, "loss": 0.07161258161067963, "lr": 7.548798009401641e-06, "epoch": 1.8637762237762239, "percentage": 62.12, "elapsed_time": "6:38:03", "remaining_time": "4:02:45"} +{"current_steps": 3333, "total_steps": 5364, "loss": 0.08686239272356033, "lr": 7.5424887068894995e-06, "epoch": 1.8643356643356643, "percentage": 62.14, "elapsed_time": "6:38:10", "remaining_time": "4:02:37"} +{"current_steps": 3334, "total_steps": 5364, "loss": 0.10367372632026672, "lr": 7.53618044535321e-06, "epoch": 1.8648951048951048, "percentage": 62.16, "elapsed_time": "6:38:17", "remaining_time": "4:02:30"} +{"current_steps": 3335, "total_steps": 5364, "loss": 0.16587960720062256, "lr": 7.5298732274648865e-06, "epoch": 1.8654545454545455, "percentage": 62.17, "elapsed_time": "6:38:24", "remaining_time": "4:02:23"} +{"current_steps": 3336, "total_steps": 5364, "loss": 0.20577214658260345, "lr": 7.523567055896193e-06, "epoch": 1.8660139860139862, "percentage": 62.19, "elapsed_time": "6:38:30", "remaining_time": "4:02:15"} +{"current_steps": 3337, "total_steps": 5364, "loss": 0.019247375428676605, "lr": 7.517261933318365e-06, "epoch": 1.8665734265734266, "percentage": 62.21, "elapsed_time": "6:38:37", "remaining_time": "4:02:08"} +{"current_steps": 3338, "total_steps": 5364, "loss": 0.08310063928365707, "lr": 7.510957862402182e-06, "epoch": 1.867132867132867, "percentage": 62.23, "elapsed_time": "6:38:44", "remaining_time": "4:02:01"} +{"current_steps": 3339, "total_steps": 5364, "loss": 0.07640981674194336, "lr": 7.504654845817985e-06, "epoch": 1.8676923076923075, "percentage": 62.25, "elapsed_time": "6:38:51", "remaining_time": "4:01:53"} +{"current_steps": 3340, "total_steps": 5364, "loss": 0.12265245616436005, "lr": 7.498352886235658e-06, "epoch": 1.8682517482517482, "percentage": 62.27, "elapsed_time": "6:38:58", "remaining_time": "4:01:46"} +{"current_steps": 3341, "total_steps": 5364, "loss": 0.2052098512649536, "lr": 7.492051986324656e-06, "epoch": 1.868811188811189, "percentage": 62.29, "elapsed_time": "6:39:05", "remaining_time": "4:01:39"} +{"current_steps": 3342, "total_steps": 5364, "loss": 0.07031114399433136, "lr": 7.485752148753966e-06, "epoch": 1.8693706293706294, "percentage": 62.3, "elapsed_time": "6:39:13", "remaining_time": "4:01:32"} +{"current_steps": 3343, "total_steps": 5364, "loss": 0.09227945655584335, "lr": 7.479453376192134e-06, "epoch": 1.8699300699300698, "percentage": 62.32, "elapsed_time": "6:39:19", "remaining_time": "4:01:24"} +{"current_steps": 3344, "total_steps": 5364, "loss": 0.20535951852798462, "lr": 7.473155671307254e-06, "epoch": 1.8704895104895105, "percentage": 62.34, "elapsed_time": "6:39:27", "remaining_time": "4:01:17"} +{"current_steps": 3345, "total_steps": 5364, "loss": 0.02173934504389763, "lr": 7.4668590367669634e-06, "epoch": 1.8710489510489512, "percentage": 62.36, "elapsed_time": "6:39:34", "remaining_time": "4:01:10"} +{"current_steps": 3346, "total_steps": 5364, "loss": 0.1195792481303215, "lr": 7.46056347523846e-06, "epoch": 1.8716083916083917, "percentage": 62.38, "elapsed_time": "6:39:42", "remaining_time": "4:01:03"} +{"current_steps": 3347, "total_steps": 5364, "loss": 0.1426640897989273, "lr": 7.454268989388471e-06, "epoch": 1.8721678321678321, "percentage": 62.4, "elapsed_time": "6:39:49", "remaining_time": "4:00:56"} +{"current_steps": 3348, "total_steps": 5364, "loss": 0.08933590352535248, "lr": 7.447975581883269e-06, "epoch": 1.8727272727272726, "percentage": 62.42, "elapsed_time": "6:39:56", "remaining_time": "4:00:49"} +{"current_steps": 3349, "total_steps": 5364, "loss": 0.03875339776277542, "lr": 7.441683255388683e-06, "epoch": 1.8732867132867133, "percentage": 62.43, "elapsed_time": "6:40:03", "remaining_time": "4:00:42"} +{"current_steps": 3350, "total_steps": 5364, "loss": 0.0318184420466423, "lr": 7.435392012570073e-06, "epoch": 1.873846153846154, "percentage": 62.45, "elapsed_time": "6:40:11", "remaining_time": "4:00:35"} +{"current_steps": 3351, "total_steps": 5364, "loss": 0.05885808914899826, "lr": 7.429101856092346e-06, "epoch": 1.8744055944055944, "percentage": 62.47, "elapsed_time": "6:40:18", "remaining_time": "4:00:28"} +{"current_steps": 3352, "total_steps": 5364, "loss": 0.32151180505752563, "lr": 7.422812788619942e-06, "epoch": 1.8749650349650349, "percentage": 62.49, "elapsed_time": "6:40:25", "remaining_time": "4:00:21"} +{"current_steps": 3353, "total_steps": 5364, "loss": 0.14148828387260437, "lr": 7.416524812816842e-06, "epoch": 1.8755244755244755, "percentage": 62.51, "elapsed_time": "6:40:32", "remaining_time": "4:00:13"} +{"current_steps": 3354, "total_steps": 5364, "loss": 0.08353282511234283, "lr": 7.410237931346573e-06, "epoch": 1.8760839160839162, "percentage": 62.53, "elapsed_time": "6:40:39", "remaining_time": "4:00:06"} +{"current_steps": 3355, "total_steps": 5364, "loss": 0.11531280726194382, "lr": 7.403952146872186e-06, "epoch": 1.8766433566433567, "percentage": 62.55, "elapsed_time": "6:40:46", "remaining_time": "3:59:59"} +{"current_steps": 3356, "total_steps": 5364, "loss": 0.248196542263031, "lr": 7.397667462056277e-06, "epoch": 1.8772027972027971, "percentage": 62.57, "elapsed_time": "6:40:53", "remaining_time": "3:59:51"} +{"current_steps": 3357, "total_steps": 5364, "loss": 0.1805993616580963, "lr": 7.391383879560965e-06, "epoch": 1.8777622377622376, "percentage": 62.58, "elapsed_time": "6:41:00", "remaining_time": "3:59:44"} +{"current_steps": 3358, "total_steps": 5364, "loss": 0.12735983729362488, "lr": 7.3851014020479205e-06, "epoch": 1.8783216783216783, "percentage": 62.6, "elapsed_time": "6:41:07", "remaining_time": "3:59:37"} +{"current_steps": 3359, "total_steps": 5364, "loss": 0.04709480702877045, "lr": 7.378820032178327e-06, "epoch": 1.878881118881119, "percentage": 62.62, "elapsed_time": "6:41:14", "remaining_time": "3:59:30"} +{"current_steps": 3360, "total_steps": 5364, "loss": 0.1560848355293274, "lr": 7.372539772612905e-06, "epoch": 1.8794405594405594, "percentage": 62.64, "elapsed_time": "6:41:21", "remaining_time": "3:59:23"} +{"current_steps": 3361, "total_steps": 5364, "loss": 0.06077303737401962, "lr": 7.366260626011913e-06, "epoch": 1.88, "percentage": 62.66, "elapsed_time": "6:41:28", "remaining_time": "3:59:15"} +{"current_steps": 3362, "total_steps": 5364, "loss": 0.07205686718225479, "lr": 7.3599825950351226e-06, "epoch": 1.8805594405594406, "percentage": 62.68, "elapsed_time": "6:41:35", "remaining_time": "3:59:08"} +{"current_steps": 3363, "total_steps": 5364, "loss": 0.16174761950969696, "lr": 7.3537056823418494e-06, "epoch": 1.8811188811188813, "percentage": 62.7, "elapsed_time": "6:41:42", "remaining_time": "3:59:01"} +{"current_steps": 3364, "total_steps": 5364, "loss": 0.22913508117198944, "lr": 7.347429890590926e-06, "epoch": 1.8816783216783217, "percentage": 62.71, "elapsed_time": "6:41:50", "remaining_time": "3:58:54"} +{"current_steps": 3365, "total_steps": 5364, "loss": 0.22197626531124115, "lr": 7.341155222440707e-06, "epoch": 1.8822377622377622, "percentage": 62.73, "elapsed_time": "6:41:57", "remaining_time": "3:58:46"} +{"current_steps": 3366, "total_steps": 5364, "loss": 0.09171321988105774, "lr": 7.334881680549079e-06, "epoch": 1.8827972027972026, "percentage": 62.75, "elapsed_time": "6:42:04", "remaining_time": "3:58:39"} +{"current_steps": 3367, "total_steps": 5364, "loss": 0.23408979177474976, "lr": 7.328609267573446e-06, "epoch": 1.8833566433566433, "percentage": 62.77, "elapsed_time": "6:42:11", "remaining_time": "3:58:32"} +{"current_steps": 3368, "total_steps": 5364, "loss": 0.24631021916866302, "lr": 7.32233798617074e-06, "epoch": 1.883916083916084, "percentage": 62.79, "elapsed_time": "6:42:18", "remaining_time": "3:58:25"} +{"current_steps": 3369, "total_steps": 5364, "loss": 0.14442695677280426, "lr": 7.316067838997409e-06, "epoch": 1.8844755244755245, "percentage": 62.81, "elapsed_time": "6:42:24", "remaining_time": "3:58:17"} +{"current_steps": 3370, "total_steps": 5364, "loss": 0.08294790983200073, "lr": 7.309798828709413e-06, "epoch": 1.885034965034965, "percentage": 62.83, "elapsed_time": "6:42:31", "remaining_time": "3:58:10"} +{"current_steps": 3371, "total_steps": 5364, "loss": 0.06489971280097961, "lr": 7.303530957962251e-06, "epoch": 1.8855944055944056, "percentage": 62.84, "elapsed_time": "6:42:39", "remaining_time": "3:58:03"} +{"current_steps": 3372, "total_steps": 5364, "loss": 0.027363475412130356, "lr": 7.297264229410918e-06, "epoch": 1.8861538461538463, "percentage": 62.86, "elapsed_time": "6:42:46", "remaining_time": "3:57:56"} +{"current_steps": 3373, "total_steps": 5364, "loss": 0.09044063091278076, "lr": 7.290998645709938e-06, "epoch": 1.8867132867132868, "percentage": 62.88, "elapsed_time": "6:42:53", "remaining_time": "3:57:49"} +{"current_steps": 3374, "total_steps": 5364, "loss": 0.09621086716651917, "lr": 7.284734209513344e-06, "epoch": 1.8872727272727272, "percentage": 62.9, "elapsed_time": "6:43:00", "remaining_time": "3:57:41"} +{"current_steps": 3375, "total_steps": 5364, "loss": 0.23984023928642273, "lr": 7.278470923474685e-06, "epoch": 1.8878321678321677, "percentage": 62.92, "elapsed_time": "6:43:07", "remaining_time": "3:57:34"} +{"current_steps": 3376, "total_steps": 5364, "loss": 0.05971367657184601, "lr": 7.2722087902470265e-06, "epoch": 1.8883916083916084, "percentage": 62.94, "elapsed_time": "6:43:14", "remaining_time": "3:57:27"} +{"current_steps": 3377, "total_steps": 5364, "loss": 0.04140233248472214, "lr": 7.265947812482936e-06, "epoch": 1.888951048951049, "percentage": 62.96, "elapsed_time": "6:43:21", "remaining_time": "3:57:19"} +{"current_steps": 3378, "total_steps": 5364, "loss": 0.12734784185886383, "lr": 7.259687992834505e-06, "epoch": 1.8895104895104895, "percentage": 62.98, "elapsed_time": "6:43:28", "remaining_time": "3:57:12"} +{"current_steps": 3379, "total_steps": 5364, "loss": 0.14470745623111725, "lr": 7.253429333953319e-06, "epoch": 1.89006993006993, "percentage": 62.99, "elapsed_time": "6:43:35", "remaining_time": "3:57:05"} +{"current_steps": 3380, "total_steps": 5364, "loss": 0.1993153691291809, "lr": 7.2471718384904875e-06, "epoch": 1.8906293706293706, "percentage": 63.01, "elapsed_time": "6:43:43", "remaining_time": "3:56:58"} +{"current_steps": 3381, "total_steps": 5364, "loss": 0.236606627702713, "lr": 7.2409155090966175e-06, "epoch": 1.8911888111888113, "percentage": 63.03, "elapsed_time": "6:43:50", "remaining_time": "3:56:51"} +{"current_steps": 3382, "total_steps": 5364, "loss": 0.17082786560058594, "lr": 7.234660348421819e-06, "epoch": 1.8917482517482518, "percentage": 63.05, "elapsed_time": "6:43:57", "remaining_time": "3:56:44"} +{"current_steps": 3383, "total_steps": 5364, "loss": 0.14957362413406372, "lr": 7.228406359115718e-06, "epoch": 1.8923076923076922, "percentage": 63.07, "elapsed_time": "6:44:04", "remaining_time": "3:56:36"} +{"current_steps": 3384, "total_steps": 5364, "loss": 0.20060384273529053, "lr": 7.222153543827432e-06, "epoch": 1.8928671328671327, "percentage": 63.09, "elapsed_time": "6:44:11", "remaining_time": "3:56:29"} +{"current_steps": 3385, "total_steps": 5364, "loss": 0.2028248906135559, "lr": 7.215901905205598e-06, "epoch": 1.8934265734265734, "percentage": 63.11, "elapsed_time": "6:44:18", "remaining_time": "3:56:22"} +{"current_steps": 3386, "total_steps": 5364, "loss": 0.11954338848590851, "lr": 7.209651445898336e-06, "epoch": 1.893986013986014, "percentage": 63.12, "elapsed_time": "6:44:25", "remaining_time": "3:56:15"} +{"current_steps": 3387, "total_steps": 5364, "loss": 0.24602608382701874, "lr": 7.203402168553275e-06, "epoch": 1.8945454545454545, "percentage": 63.14, "elapsed_time": "6:44:32", "remaining_time": "3:56:07"} +{"current_steps": 3388, "total_steps": 5364, "loss": 0.1610192060470581, "lr": 7.197154075817545e-06, "epoch": 1.895104895104895, "percentage": 63.16, "elapsed_time": "6:44:39", "remaining_time": "3:56:00"} +{"current_steps": 3389, "total_steps": 5364, "loss": 0.0561186745762825, "lr": 7.190907170337768e-06, "epoch": 1.8956643356643357, "percentage": 63.18, "elapsed_time": "6:44:46", "remaining_time": "3:55:53"} +{"current_steps": 3390, "total_steps": 5364, "loss": 0.18576236069202423, "lr": 7.184661454760076e-06, "epoch": 1.8962237762237764, "percentage": 63.2, "elapsed_time": "6:44:53", "remaining_time": "3:55:45"} +{"current_steps": 3391, "total_steps": 5364, "loss": 0.22534814476966858, "lr": 7.1784169317300825e-06, "epoch": 1.8967832167832168, "percentage": 63.22, "elapsed_time": "6:45:00", "remaining_time": "3:55:38"} +{"current_steps": 3392, "total_steps": 5364, "loss": 0.11819382756948471, "lr": 7.1721736038929e-06, "epoch": 1.8973426573426573, "percentage": 63.24, "elapsed_time": "6:45:07", "remaining_time": "3:55:31"} +{"current_steps": 3393, "total_steps": 5364, "loss": 0.13060441613197327, "lr": 7.16593147389314e-06, "epoch": 1.8979020979020977, "percentage": 63.26, "elapsed_time": "6:45:14", "remaining_time": "3:55:24"} +{"current_steps": 3394, "total_steps": 5364, "loss": 0.1372489631175995, "lr": 7.1596905443748975e-06, "epoch": 1.8984615384615384, "percentage": 63.27, "elapsed_time": "6:45:21", "remaining_time": "3:55:17"} +{"current_steps": 3395, "total_steps": 5364, "loss": 0.1455768495798111, "lr": 7.153450817981774e-06, "epoch": 1.899020979020979, "percentage": 63.29, "elapsed_time": "6:45:28", "remaining_time": "3:55:09"} +{"current_steps": 3396, "total_steps": 5364, "loss": 0.1744433343410492, "lr": 7.1472122973568406e-06, "epoch": 1.8995804195804196, "percentage": 63.31, "elapsed_time": "6:45:35", "remaining_time": "3:55:02"} +{"current_steps": 3397, "total_steps": 5364, "loss": 0.2150195837020874, "lr": 7.140974985142677e-06, "epoch": 1.90013986013986, "percentage": 63.33, "elapsed_time": "6:45:42", "remaining_time": "3:54:55"} +{"current_steps": 3398, "total_steps": 5364, "loss": 0.18725216388702393, "lr": 7.13473888398134e-06, "epoch": 1.9006993006993007, "percentage": 63.35, "elapsed_time": "6:45:50", "remaining_time": "3:54:48"} +{"current_steps": 3399, "total_steps": 5364, "loss": 0.1420905441045761, "lr": 7.128503996514375e-06, "epoch": 1.9012587412587414, "percentage": 63.37, "elapsed_time": "6:45:57", "remaining_time": "3:54:41"} +{"current_steps": 3400, "total_steps": 5364, "loss": 0.06339014321565628, "lr": 7.122270325382818e-06, "epoch": 1.9018181818181819, "percentage": 63.39, "elapsed_time": "6:46:04", "remaining_time": "3:54:34"} +{"current_steps": 3401, "total_steps": 5364, "loss": 0.158181831240654, "lr": 7.116037873227178e-06, "epoch": 1.9023776223776223, "percentage": 63.4, "elapsed_time": "6:46:12", "remaining_time": "3:54:27"} +{"current_steps": 3402, "total_steps": 5364, "loss": 0.10833267867565155, "lr": 7.109806642687468e-06, "epoch": 1.9029370629370628, "percentage": 63.42, "elapsed_time": "6:46:19", "remaining_time": "3:54:20"} +{"current_steps": 3403, "total_steps": 5364, "loss": 0.09106975793838501, "lr": 7.103576636403164e-06, "epoch": 1.9034965034965035, "percentage": 63.44, "elapsed_time": "6:46:26", "remaining_time": "3:54:12"} +{"current_steps": 3404, "total_steps": 5364, "loss": 0.1995009183883667, "lr": 7.097347857013232e-06, "epoch": 1.9040559440559441, "percentage": 63.46, "elapsed_time": "6:46:33", "remaining_time": "3:54:05"} +{"current_steps": 3405, "total_steps": 5364, "loss": 0.1709326058626175, "lr": 7.091120307156118e-06, "epoch": 1.9046153846153846, "percentage": 63.48, "elapsed_time": "6:46:40", "remaining_time": "3:53:58"} +{"current_steps": 3406, "total_steps": 5364, "loss": 0.01894509792327881, "lr": 7.084893989469743e-06, "epoch": 1.905174825174825, "percentage": 63.5, "elapsed_time": "6:46:47", "remaining_time": "3:53:50"} +{"current_steps": 3407, "total_steps": 5364, "loss": 0.07084296643733978, "lr": 7.078668906591517e-06, "epoch": 1.9057342657342657, "percentage": 63.52, "elapsed_time": "6:46:54", "remaining_time": "3:53:43"} +{"current_steps": 3408, "total_steps": 5364, "loss": 0.03980420529842377, "lr": 7.072445061158315e-06, "epoch": 1.9062937062937064, "percentage": 63.53, "elapsed_time": "6:47:01", "remaining_time": "3:53:36"} +{"current_steps": 3409, "total_steps": 5364, "loss": 0.22033479809761047, "lr": 7.066222455806488e-06, "epoch": 1.9068531468531469, "percentage": 63.55, "elapsed_time": "6:47:08", "remaining_time": "3:53:29"} +{"current_steps": 3410, "total_steps": 5364, "loss": 0.07911434769630432, "lr": 7.0600010931718755e-06, "epoch": 1.9074125874125873, "percentage": 63.57, "elapsed_time": "6:47:14", "remaining_time": "3:53:21"} +{"current_steps": 3411, "total_steps": 5364, "loss": 0.059310268610715866, "lr": 7.053780975889769e-06, "epoch": 1.9079720279720278, "percentage": 63.59, "elapsed_time": "6:47:21", "remaining_time": "3:53:14"} +{"current_steps": 3412, "total_steps": 5364, "loss": 0.10286440700292587, "lr": 7.047562106594956e-06, "epoch": 1.9085314685314685, "percentage": 63.61, "elapsed_time": "6:47:29", "remaining_time": "3:53:07"} +{"current_steps": 3413, "total_steps": 5364, "loss": 0.18492405116558075, "lr": 7.041344487921679e-06, "epoch": 1.9090909090909092, "percentage": 63.63, "elapsed_time": "6:47:36", "remaining_time": "3:53:00"} +{"current_steps": 3414, "total_steps": 5364, "loss": 0.20573991537094116, "lr": 7.035128122503654e-06, "epoch": 1.9096503496503496, "percentage": 63.65, "elapsed_time": "6:47:43", "remaining_time": "3:52:52"} +{"current_steps": 3415, "total_steps": 5364, "loss": 0.08964357525110245, "lr": 7.0289130129740726e-06, "epoch": 1.91020979020979, "percentage": 63.67, "elapsed_time": "6:47:50", "remaining_time": "3:52:45"} +{"current_steps": 3416, "total_steps": 5364, "loss": 0.0956602692604065, "lr": 7.022699161965581e-06, "epoch": 1.9107692307692308, "percentage": 63.68, "elapsed_time": "6:47:56", "remaining_time": "3:52:38"} +{"current_steps": 3417, "total_steps": 5364, "loss": 0.11252351850271225, "lr": 7.0164865721103104e-06, "epoch": 1.9113286713286715, "percentage": 63.7, "elapsed_time": "6:48:04", "remaining_time": "3:52:31"} +{"current_steps": 3418, "total_steps": 5364, "loss": 0.01893317699432373, "lr": 7.010275246039843e-06, "epoch": 1.911888111888112, "percentage": 63.72, "elapsed_time": "6:48:11", "remaining_time": "3:52:23"} +{"current_steps": 3419, "total_steps": 5364, "loss": 0.18156349658966064, "lr": 7.004065186385236e-06, "epoch": 1.9124475524475524, "percentage": 63.74, "elapsed_time": "6:48:18", "remaining_time": "3:52:16"} +{"current_steps": 3420, "total_steps": 5364, "loss": 0.06971661001443863, "lr": 6.997856395776999e-06, "epoch": 1.913006993006993, "percentage": 63.76, "elapsed_time": "6:48:24", "remaining_time": "3:52:09"} +{"current_steps": 3421, "total_steps": 5364, "loss": 0.05821669101715088, "lr": 6.991648876845112e-06, "epoch": 1.9135664335664335, "percentage": 63.78, "elapsed_time": "6:48:31", "remaining_time": "3:52:01"} +{"current_steps": 3422, "total_steps": 5364, "loss": 0.12792979180812836, "lr": 6.985442632219017e-06, "epoch": 1.9141258741258742, "percentage": 63.8, "elapsed_time": "6:48:38", "remaining_time": "3:51:54"} +{"current_steps": 3423, "total_steps": 5364, "loss": 0.11260859668254852, "lr": 6.979237664527609e-06, "epoch": 1.9146853146853147, "percentage": 63.81, "elapsed_time": "6:48:46", "remaining_time": "3:51:47"} +{"current_steps": 3424, "total_steps": 5364, "loss": 0.22189411520957947, "lr": 6.973033976399253e-06, "epoch": 1.9152447552447551, "percentage": 63.83, "elapsed_time": "6:48:53", "remaining_time": "3:51:40"} +{"current_steps": 3425, "total_steps": 5364, "loss": 0.12042347341775894, "lr": 6.9668315704617625e-06, "epoch": 1.9158041958041958, "percentage": 63.85, "elapsed_time": "6:49:00", "remaining_time": "3:51:33"} +{"current_steps": 3426, "total_steps": 5364, "loss": 0.08907759934663773, "lr": 6.960630449342409e-06, "epoch": 1.9163636363636365, "percentage": 63.87, "elapsed_time": "6:49:07", "remaining_time": "3:51:26"} +{"current_steps": 3427, "total_steps": 5364, "loss": 0.16345861554145813, "lr": 6.954430615667927e-06, "epoch": 1.916923076923077, "percentage": 63.89, "elapsed_time": "6:49:14", "remaining_time": "3:51:18"} +{"current_steps": 3428, "total_steps": 5364, "loss": 0.1780029535293579, "lr": 6.948232072064493e-06, "epoch": 1.9174825174825174, "percentage": 63.91, "elapsed_time": "6:49:21", "remaining_time": "3:51:11"} +{"current_steps": 3429, "total_steps": 5364, "loss": 0.04524610564112663, "lr": 6.9420348211577546e-06, "epoch": 1.918041958041958, "percentage": 63.93, "elapsed_time": "6:49:29", "remaining_time": "3:51:04"} +{"current_steps": 3430, "total_steps": 5364, "loss": 0.22067275643348694, "lr": 6.935838865572797e-06, "epoch": 1.9186013986013986, "percentage": 63.94, "elapsed_time": "6:49:36", "remaining_time": "3:50:57"} +{"current_steps": 3431, "total_steps": 5364, "loss": 0.02736462652683258, "lr": 6.929644207934158e-06, "epoch": 1.9191608391608392, "percentage": 63.96, "elapsed_time": "6:49:43", "remaining_time": "3:50:50"} +{"current_steps": 3432, "total_steps": 5364, "loss": 0.060130778700113297, "lr": 6.923450850865836e-06, "epoch": 1.9197202797202797, "percentage": 63.98, "elapsed_time": "6:49:51", "remaining_time": "3:50:43"} +{"current_steps": 3433, "total_steps": 5364, "loss": 0.12045353651046753, "lr": 6.917258796991264e-06, "epoch": 1.9202797202797202, "percentage": 64.0, "elapsed_time": "6:49:58", "remaining_time": "3:50:36"} +{"current_steps": 3434, "total_steps": 5364, "loss": 0.2129320502281189, "lr": 6.91106804893334e-06, "epoch": 1.9208391608391608, "percentage": 64.02, "elapsed_time": "6:50:05", "remaining_time": "3:50:29"} +{"current_steps": 3435, "total_steps": 5364, "loss": 0.1366441249847412, "lr": 6.904878609314392e-06, "epoch": 1.9213986013986015, "percentage": 64.04, "elapsed_time": "6:50:12", "remaining_time": "3:50:21"} +{"current_steps": 3436, "total_steps": 5364, "loss": 0.10970372706651688, "lr": 6.898690480756205e-06, "epoch": 1.921958041958042, "percentage": 64.06, "elapsed_time": "6:50:19", "remaining_time": "3:50:14"} +{"current_steps": 3437, "total_steps": 5364, "loss": 0.18622155487537384, "lr": 6.892503665880005e-06, "epoch": 1.9225174825174824, "percentage": 64.08, "elapsed_time": "6:50:26", "remaining_time": "3:50:07"} +{"current_steps": 3438, "total_steps": 5364, "loss": 0.15056131780147552, "lr": 6.8863181673064565e-06, "epoch": 1.9230769230769231, "percentage": 64.09, "elapsed_time": "6:50:33", "remaining_time": "3:49:59"} +{"current_steps": 3439, "total_steps": 5364, "loss": 0.18443703651428223, "lr": 6.880133987655678e-06, "epoch": 1.9236363636363636, "percentage": 64.11, "elapsed_time": "6:50:40", "remaining_time": "3:49:52"} +{"current_steps": 3440, "total_steps": 5364, "loss": 0.25488942861557007, "lr": 6.873951129547219e-06, "epoch": 1.9241958041958043, "percentage": 64.13, "elapsed_time": "6:50:47", "remaining_time": "3:49:45"} +{"current_steps": 3441, "total_steps": 5364, "loss": 0.08821668475866318, "lr": 6.867769595600076e-06, "epoch": 1.9247552447552447, "percentage": 64.15, "elapsed_time": "6:50:54", "remaining_time": "3:49:38"} +{"current_steps": 3442, "total_steps": 5364, "loss": 0.08856074512004852, "lr": 6.861589388432679e-06, "epoch": 1.9253146853146852, "percentage": 64.17, "elapsed_time": "6:51:01", "remaining_time": "3:49:30"} +{"current_steps": 3443, "total_steps": 5364, "loss": 0.0803074985742569, "lr": 6.8554105106628954e-06, "epoch": 1.9258741258741259, "percentage": 64.19, "elapsed_time": "6:51:08", "remaining_time": "3:49:23"} +{"current_steps": 3444, "total_steps": 5364, "loss": 0.03420411795377731, "lr": 6.849232964908043e-06, "epoch": 1.9264335664335666, "percentage": 64.21, "elapsed_time": "6:51:16", "remaining_time": "3:49:17"} +{"current_steps": 3445, "total_steps": 5364, "loss": 0.16885244846343994, "lr": 6.843056753784851e-06, "epoch": 1.926993006993007, "percentage": 64.22, "elapsed_time": "6:51:23", "remaining_time": "3:49:09"} +{"current_steps": 3446, "total_steps": 5364, "loss": 0.2108049839735031, "lr": 6.836881879909509e-06, "epoch": 1.9275524475524475, "percentage": 64.24, "elapsed_time": "6:51:30", "remaining_time": "3:49:02"} +{"current_steps": 3447, "total_steps": 5364, "loss": 0.1319536566734314, "lr": 6.830708345897625e-06, "epoch": 1.9281118881118882, "percentage": 64.26, "elapsed_time": "6:51:37", "remaining_time": "3:48:55"} +{"current_steps": 3448, "total_steps": 5364, "loss": 0.09764501452445984, "lr": 6.824536154364239e-06, "epoch": 1.9286713286713286, "percentage": 64.28, "elapsed_time": "6:51:45", "remaining_time": "3:48:48"} +{"current_steps": 3449, "total_steps": 5364, "loss": 0.1413380354642868, "lr": 6.818365307923831e-06, "epoch": 1.9292307692307693, "percentage": 64.3, "elapsed_time": "6:51:53", "remaining_time": "3:48:41"} +{"current_steps": 3450, "total_steps": 5364, "loss": 0.05311393737792969, "lr": 6.8121958091903005e-06, "epoch": 1.9297902097902098, "percentage": 64.32, "elapsed_time": "6:52:00", "remaining_time": "3:48:34"} +{"current_steps": 3451, "total_steps": 5364, "loss": 0.19735713303089142, "lr": 6.806027660776991e-06, "epoch": 1.9303496503496502, "percentage": 64.34, "elapsed_time": "6:52:07", "remaining_time": "3:48:27"} +{"current_steps": 3452, "total_steps": 5364, "loss": 0.11381424218416214, "lr": 6.799860865296657e-06, "epoch": 1.930909090909091, "percentage": 64.35, "elapsed_time": "6:52:14", "remaining_time": "3:48:20"} +{"current_steps": 3453, "total_steps": 5364, "loss": 0.17068126797676086, "lr": 6.793695425361493e-06, "epoch": 1.9314685314685316, "percentage": 64.37, "elapsed_time": "6:52:21", "remaining_time": "3:48:12"} +{"current_steps": 3454, "total_steps": 5364, "loss": 0.017199130728840828, "lr": 6.787531343583113e-06, "epoch": 1.932027972027972, "percentage": 64.39, "elapsed_time": "6:52:29", "remaining_time": "3:48:05"} +{"current_steps": 3455, "total_steps": 5364, "loss": 0.08450612425804138, "lr": 6.781368622572553e-06, "epoch": 1.9325874125874125, "percentage": 64.41, "elapsed_time": "6:52:36", "remaining_time": "3:47:58"} +{"current_steps": 3456, "total_steps": 5364, "loss": 0.04806827753782272, "lr": 6.775207264940282e-06, "epoch": 1.9331468531468532, "percentage": 64.43, "elapsed_time": "6:52:43", "remaining_time": "3:47:51"} +{"current_steps": 3457, "total_steps": 5364, "loss": 0.07865659892559052, "lr": 6.7690472732961835e-06, "epoch": 1.9337062937062937, "percentage": 64.45, "elapsed_time": "6:52:50", "remaining_time": "3:47:44"} +{"current_steps": 3458, "total_steps": 5364, "loss": 0.31509312987327576, "lr": 6.7628886502495674e-06, "epoch": 1.9342657342657343, "percentage": 64.47, "elapsed_time": "6:52:57", "remaining_time": "3:47:37"} +{"current_steps": 3459, "total_steps": 5364, "loss": 0.16801634430885315, "lr": 6.75673139840916e-06, "epoch": 1.9348251748251748, "percentage": 64.49, "elapsed_time": "6:53:04", "remaining_time": "3:47:29"} +{"current_steps": 3460, "total_steps": 5364, "loss": 0.24640288949012756, "lr": 6.750575520383102e-06, "epoch": 1.9353846153846153, "percentage": 64.5, "elapsed_time": "6:53:11", "remaining_time": "3:47:22"} +{"current_steps": 3461, "total_steps": 5364, "loss": 0.20307449996471405, "lr": 6.74442101877897e-06, "epoch": 1.935944055944056, "percentage": 64.52, "elapsed_time": "6:53:18", "remaining_time": "3:47:15"} +{"current_steps": 3462, "total_steps": 5364, "loss": 0.15582285821437836, "lr": 6.738267896203736e-06, "epoch": 1.9365034965034966, "percentage": 64.54, "elapsed_time": "6:53:26", "remaining_time": "3:47:08"} +{"current_steps": 3463, "total_steps": 5364, "loss": 0.2522038519382477, "lr": 6.732116155263806e-06, "epoch": 1.937062937062937, "percentage": 64.56, "elapsed_time": "6:53:33", "remaining_time": "3:47:01"} +{"current_steps": 3464, "total_steps": 5364, "loss": 0.21848735213279724, "lr": 6.725965798564987e-06, "epoch": 1.9376223776223775, "percentage": 64.58, "elapsed_time": "6:53:39", "remaining_time": "3:46:53"} +{"current_steps": 3465, "total_steps": 5364, "loss": 0.09598897397518158, "lr": 6.719816828712505e-06, "epoch": 1.9381818181818182, "percentage": 64.6, "elapsed_time": "6:53:47", "remaining_time": "3:46:46"} +{"current_steps": 3466, "total_steps": 5364, "loss": 0.17435981333255768, "lr": 6.713669248311001e-06, "epoch": 1.9387412587412587, "percentage": 64.62, "elapsed_time": "6:53:54", "remaining_time": "3:46:39"} +{"current_steps": 3467, "total_steps": 5364, "loss": 0.19949239492416382, "lr": 6.7075230599645245e-06, "epoch": 1.9393006993006994, "percentage": 64.63, "elapsed_time": "6:54:01", "remaining_time": "3:46:32"} +{"current_steps": 3468, "total_steps": 5364, "loss": 0.12844239175319672, "lr": 6.701378266276537e-06, "epoch": 1.9398601398601398, "percentage": 64.65, "elapsed_time": "6:54:08", "remaining_time": "3:46:24"} +{"current_steps": 3469, "total_steps": 5364, "loss": 0.18290919065475464, "lr": 6.695234869849908e-06, "epoch": 1.9404195804195803, "percentage": 64.67, "elapsed_time": "6:54:15", "remaining_time": "3:46:17"} +{"current_steps": 3470, "total_steps": 5364, "loss": 0.22710451483726501, "lr": 6.689092873286913e-06, "epoch": 1.940979020979021, "percentage": 64.69, "elapsed_time": "6:54:22", "remaining_time": "3:46:10"} +{"current_steps": 3471, "total_steps": 5364, "loss": 0.2175981104373932, "lr": 6.6829522791892395e-06, "epoch": 1.9415384615384617, "percentage": 64.71, "elapsed_time": "6:54:29", "remaining_time": "3:46:03"} +{"current_steps": 3472, "total_steps": 5364, "loss": 0.18920888006687164, "lr": 6.676813090157974e-06, "epoch": 1.9420979020979021, "percentage": 64.73, "elapsed_time": "6:54:36", "remaining_time": "3:45:56"} +{"current_steps": 3473, "total_steps": 5364, "loss": 0.22502082586288452, "lr": 6.67067530879362e-06, "epoch": 1.9426573426573426, "percentage": 64.75, "elapsed_time": "6:54:43", "remaining_time": "3:45:48"} +{"current_steps": 3474, "total_steps": 5364, "loss": 0.23475733399391174, "lr": 6.664538937696071e-06, "epoch": 1.9432167832167833, "percentage": 64.77, "elapsed_time": "6:54:50", "remaining_time": "3:45:41"} +{"current_steps": 3475, "total_steps": 5364, "loss": 0.10718761384487152, "lr": 6.658403979464631e-06, "epoch": 1.9437762237762237, "percentage": 64.78, "elapsed_time": "6:54:57", "remaining_time": "3:45:34"} +{"current_steps": 3476, "total_steps": 5364, "loss": 0.20793825387954712, "lr": 6.652270436698005e-06, "epoch": 1.9443356643356644, "percentage": 64.8, "elapsed_time": "6:55:04", "remaining_time": "3:45:27"} +{"current_steps": 3477, "total_steps": 5364, "loss": 0.1298268437385559, "lr": 6.646138311994292e-06, "epoch": 1.9448951048951049, "percentage": 64.82, "elapsed_time": "6:55:11", "remaining_time": "3:45:19"} +{"current_steps": 3478, "total_steps": 5364, "loss": 0.21758905053138733, "lr": 6.6400076079510004e-06, "epoch": 1.9454545454545453, "percentage": 64.84, "elapsed_time": "6:55:19", "remaining_time": "3:45:12"} +{"current_steps": 3479, "total_steps": 5364, "loss": 0.1582922339439392, "lr": 6.633878327165028e-06, "epoch": 1.946013986013986, "percentage": 64.86, "elapsed_time": "6:55:25", "remaining_time": "3:45:05"} +{"current_steps": 3480, "total_steps": 5364, "loss": 0.3181908428668976, "lr": 6.627750472232679e-06, "epoch": 1.9465734265734267, "percentage": 64.88, "elapsed_time": "6:55:32", "remaining_time": "3:44:58"} +{"current_steps": 3481, "total_steps": 5364, "loss": 0.19448833167552948, "lr": 6.621624045749643e-06, "epoch": 1.9471328671328672, "percentage": 64.9, "elapsed_time": "6:55:39", "remaining_time": "3:44:50"} +{"current_steps": 3482, "total_steps": 5364, "loss": 0.21107375621795654, "lr": 6.615499050311007e-06, "epoch": 1.9476923076923076, "percentage": 64.91, "elapsed_time": "6:55:46", "remaining_time": "3:44:43"} +{"current_steps": 3483, "total_steps": 5364, "loss": 0.05676122009754181, "lr": 6.609375488511263e-06, "epoch": 1.9482517482517483, "percentage": 64.93, "elapsed_time": "6:55:53", "remaining_time": "3:44:36"} +{"current_steps": 3484, "total_steps": 5364, "loss": 0.21985793113708496, "lr": 6.603253362944281e-06, "epoch": 1.948811188811189, "percentage": 64.95, "elapsed_time": "6:56:00", "remaining_time": "3:44:28"} +{"current_steps": 3485, "total_steps": 5364, "loss": 0.36191850900650024, "lr": 6.597132676203331e-06, "epoch": 1.9493706293706294, "percentage": 64.97, "elapsed_time": "6:56:07", "remaining_time": "3:44:21"} +{"current_steps": 3486, "total_steps": 5364, "loss": 0.05745324864983559, "lr": 6.5910134308810725e-06, "epoch": 1.94993006993007, "percentage": 64.99, "elapsed_time": "6:56:14", "remaining_time": "3:44:14"} +{"current_steps": 3487, "total_steps": 5364, "loss": 0.17055457830429077, "lr": 6.584895629569545e-06, "epoch": 1.9504895104895104, "percentage": 65.01, "elapsed_time": "6:56:22", "remaining_time": "3:44:07"} +{"current_steps": 3488, "total_steps": 5364, "loss": 0.1341659277677536, "lr": 6.578779274860197e-06, "epoch": 1.951048951048951, "percentage": 65.03, "elapsed_time": "6:56:28", "remaining_time": "3:44:00"} +{"current_steps": 3489, "total_steps": 5364, "loss": 0.10700860619544983, "lr": 6.572664369343844e-06, "epoch": 1.9516083916083917, "percentage": 65.04, "elapsed_time": "6:56:36", "remaining_time": "3:43:53"} +{"current_steps": 3490, "total_steps": 5364, "loss": 0.14823457598686218, "lr": 6.566550915610699e-06, "epoch": 1.9521678321678322, "percentage": 65.06, "elapsed_time": "6:56:43", "remaining_time": "3:43:46"} +{"current_steps": 3491, "total_steps": 5364, "loss": 0.14534789323806763, "lr": 6.56043891625035e-06, "epoch": 1.9527272727272726, "percentage": 65.08, "elapsed_time": "6:56:50", "remaining_time": "3:43:38"} +{"current_steps": 3492, "total_steps": 5364, "loss": 0.15859714150428772, "lr": 6.554328373851785e-06, "epoch": 1.9532867132867133, "percentage": 65.1, "elapsed_time": "6:56:57", "remaining_time": "3:43:31"} +{"current_steps": 3493, "total_steps": 5364, "loss": 0.050215672701597214, "lr": 6.548219291003363e-06, "epoch": 1.953846153846154, "percentage": 65.12, "elapsed_time": "6:57:05", "remaining_time": "3:43:24"} +{"current_steps": 3494, "total_steps": 5364, "loss": 0.09473301470279694, "lr": 6.5421116702928165e-06, "epoch": 1.9544055944055945, "percentage": 65.14, "elapsed_time": "6:57:11", "remaining_time": "3:43:17"} +{"current_steps": 3495, "total_steps": 5364, "loss": 0.11327485740184784, "lr": 6.536005514307283e-06, "epoch": 1.954965034965035, "percentage": 65.16, "elapsed_time": "6:57:18", "remaining_time": "3:43:09"} +{"current_steps": 3496, "total_steps": 5364, "loss": 0.1411135047674179, "lr": 6.529900825633258e-06, "epoch": 1.9555244755244754, "percentage": 65.18, "elapsed_time": "6:57:25", "remaining_time": "3:43:02"} +{"current_steps": 3497, "total_steps": 5364, "loss": 0.07149860262870789, "lr": 6.5237976068566266e-06, "epoch": 1.956083916083916, "percentage": 65.19, "elapsed_time": "6:57:33", "remaining_time": "3:42:55"} +{"current_steps": 3498, "total_steps": 5364, "loss": 0.07713808119297028, "lr": 6.517695860562646e-06, "epoch": 1.9566433566433568, "percentage": 65.21, "elapsed_time": "6:57:40", "remaining_time": "3:42:48"} +{"current_steps": 3499, "total_steps": 5364, "loss": 0.17901864647865295, "lr": 6.511595589335948e-06, "epoch": 1.9572027972027972, "percentage": 65.23, "elapsed_time": "6:57:47", "remaining_time": "3:42:41"} +{"current_steps": 3500, "total_steps": 5364, "loss": 0.2368810474872589, "lr": 6.505496795760553e-06, "epoch": 1.9577622377622377, "percentage": 65.25, "elapsed_time": "6:57:54", "remaining_time": "3:42:34"} +{"current_steps": 3501, "total_steps": 5364, "loss": 0.07429467886686325, "lr": 6.499399482419835e-06, "epoch": 1.9583216783216784, "percentage": 65.27, "elapsed_time": "6:58:30", "remaining_time": "3:42:42"} +{"current_steps": 3502, "total_steps": 5364, "loss": 0.07490898668766022, "lr": 6.493303651896561e-06, "epoch": 1.958881118881119, "percentage": 65.29, "elapsed_time": "6:58:38", "remaining_time": "3:42:35"} +{"current_steps": 3503, "total_steps": 5364, "loss": 0.15737158060073853, "lr": 6.487209306772852e-06, "epoch": 1.9594405594405595, "percentage": 65.31, "elapsed_time": "6:58:45", "remaining_time": "3:42:28"} +{"current_steps": 3504, "total_steps": 5364, "loss": 0.2837570607662201, "lr": 6.4811164496302115e-06, "epoch": 1.96, "percentage": 65.32, "elapsed_time": "6:58:52", "remaining_time": "3:42:20"} +{"current_steps": 3505, "total_steps": 5364, "loss": 0.17096087336540222, "lr": 6.475025083049513e-06, "epoch": 1.9605594405594404, "percentage": 65.34, "elapsed_time": "6:58:59", "remaining_time": "3:42:13"} +{"current_steps": 3506, "total_steps": 5364, "loss": 0.08211810886859894, "lr": 6.46893520961099e-06, "epoch": 1.961118881118881, "percentage": 65.36, "elapsed_time": "6:59:06", "remaining_time": "3:42:06"} +{"current_steps": 3507, "total_steps": 5364, "loss": 0.02249753288924694, "lr": 6.462846831894253e-06, "epoch": 1.9616783216783218, "percentage": 65.38, "elapsed_time": "6:59:14", "remaining_time": "3:41:59"} +{"current_steps": 3508, "total_steps": 5364, "loss": 0.16572976112365723, "lr": 6.456759952478268e-06, "epoch": 1.9622377622377623, "percentage": 65.4, "elapsed_time": "6:59:21", "remaining_time": "3:41:52"} +{"current_steps": 3509, "total_steps": 5364, "loss": 0.11485010385513306, "lr": 6.450674573941381e-06, "epoch": 1.9627972027972027, "percentage": 65.42, "elapsed_time": "6:59:28", "remaining_time": "3:41:45"} +{"current_steps": 3510, "total_steps": 5364, "loss": 0.024327319115400314, "lr": 6.44459069886129e-06, "epoch": 1.9633566433566434, "percentage": 65.44, "elapsed_time": "6:59:36", "remaining_time": "3:41:38"} +{"current_steps": 3511, "total_steps": 5364, "loss": 0.12833207845687866, "lr": 6.438508329815061e-06, "epoch": 1.963916083916084, "percentage": 65.45, "elapsed_time": "6:59:43", "remaining_time": "3:41:31"} +{"current_steps": 3512, "total_steps": 5364, "loss": 0.09652186930179596, "lr": 6.432427469379121e-06, "epoch": 1.9644755244755245, "percentage": 65.47, "elapsed_time": "6:59:51", "remaining_time": "3:41:24"} +{"current_steps": 3513, "total_steps": 5364, "loss": 0.10218455642461777, "lr": 6.426348120129253e-06, "epoch": 1.965034965034965, "percentage": 65.49, "elapsed_time": "6:59:59", "remaining_time": "3:41:17"} +{"current_steps": 3514, "total_steps": 5364, "loss": 0.12388552725315094, "lr": 6.420270284640617e-06, "epoch": 1.9655944055944055, "percentage": 65.51, "elapsed_time": "7:00:06", "remaining_time": "3:41:10"} +{"current_steps": 3515, "total_steps": 5364, "loss": 0.06462287902832031, "lr": 6.414193965487712e-06, "epoch": 1.9661538461538461, "percentage": 65.53, "elapsed_time": "7:00:13", "remaining_time": "3:41:03"} +{"current_steps": 3516, "total_steps": 5364, "loss": 0.2515866160392761, "lr": 6.408119165244403e-06, "epoch": 1.9667132867132868, "percentage": 65.55, "elapsed_time": "7:00:20", "remaining_time": "3:40:55"} +{"current_steps": 3517, "total_steps": 5364, "loss": 0.14408966898918152, "lr": 6.402045886483914e-06, "epoch": 1.9672727272727273, "percentage": 65.57, "elapsed_time": "7:00:28", "remaining_time": "3:40:48"} +{"current_steps": 3518, "total_steps": 5364, "loss": 0.18547114729881287, "lr": 6.395974131778816e-06, "epoch": 1.9678321678321677, "percentage": 65.59, "elapsed_time": "7:00:35", "remaining_time": "3:40:41"} +{"current_steps": 3519, "total_steps": 5364, "loss": 0.134500190615654, "lr": 6.389903903701051e-06, "epoch": 1.9683916083916084, "percentage": 65.6, "elapsed_time": "7:00:42", "remaining_time": "3:40:34"} +{"current_steps": 3520, "total_steps": 5364, "loss": 0.11318524181842804, "lr": 6.383835204821895e-06, "epoch": 1.968951048951049, "percentage": 65.62, "elapsed_time": "7:00:49", "remaining_time": "3:40:27"} +{"current_steps": 3521, "total_steps": 5364, "loss": 0.07681688666343689, "lr": 6.377768037711982e-06, "epoch": 1.9695104895104896, "percentage": 65.64, "elapsed_time": "7:00:56", "remaining_time": "3:40:20"} +{"current_steps": 3522, "total_steps": 5364, "loss": 0.12052248418331146, "lr": 6.371702404941309e-06, "epoch": 1.97006993006993, "percentage": 65.66, "elapsed_time": "7:01:03", "remaining_time": "3:40:12"} +{"current_steps": 3523, "total_steps": 5364, "loss": 0.30993402004241943, "lr": 6.365638309079207e-06, "epoch": 1.9706293706293705, "percentage": 65.68, "elapsed_time": "7:01:10", "remaining_time": "3:40:05"} +{"current_steps": 3524, "total_steps": 5364, "loss": 0.07829742133617401, "lr": 6.359575752694369e-06, "epoch": 1.9711888111888112, "percentage": 65.7, "elapsed_time": "7:01:17", "remaining_time": "3:39:58"} +{"current_steps": 3525, "total_steps": 5364, "loss": 0.09763751924037933, "lr": 6.3535147383548265e-06, "epoch": 1.9717482517482519, "percentage": 65.72, "elapsed_time": "7:01:25", "remaining_time": "3:39:51"} +{"current_steps": 3526, "total_steps": 5364, "loss": 0.13628189265727997, "lr": 6.347455268627958e-06, "epoch": 1.9723076923076923, "percentage": 65.73, "elapsed_time": "7:01:31", "remaining_time": "3:39:43"} +{"current_steps": 3527, "total_steps": 5364, "loss": 0.19780652225017548, "lr": 6.341397346080498e-06, "epoch": 1.9728671328671328, "percentage": 65.75, "elapsed_time": "7:01:39", "remaining_time": "3:39:36"} +{"current_steps": 3528, "total_steps": 5364, "loss": 0.0873546376824379, "lr": 6.335340973278512e-06, "epoch": 1.9734265734265735, "percentage": 65.77, "elapsed_time": "7:01:46", "remaining_time": "3:39:29"} +{"current_steps": 3529, "total_steps": 5364, "loss": 0.12127527594566345, "lr": 6.329286152787422e-06, "epoch": 1.9739860139860141, "percentage": 65.79, "elapsed_time": "7:01:53", "remaining_time": "3:39:22"} +{"current_steps": 3530, "total_steps": 5364, "loss": 0.0987105518579483, "lr": 6.323232887171978e-06, "epoch": 1.9745454545454546, "percentage": 65.81, "elapsed_time": "7:02:00", "remaining_time": "3:39:14"} +{"current_steps": 3531, "total_steps": 5364, "loss": 0.10629743337631226, "lr": 6.317181178996286e-06, "epoch": 1.975104895104895, "percentage": 65.83, "elapsed_time": "7:02:07", "remaining_time": "3:39:07"} +{"current_steps": 3532, "total_steps": 5364, "loss": 0.0655193105340004, "lr": 6.311131030823786e-06, "epoch": 1.9756643356643355, "percentage": 65.85, "elapsed_time": "7:02:14", "remaining_time": "3:39:00"} +{"current_steps": 3533, "total_steps": 5364, "loss": 0.14093634486198425, "lr": 6.305082445217252e-06, "epoch": 1.9762237762237762, "percentage": 65.87, "elapsed_time": "7:02:21", "remaining_time": "3:38:53"} +{"current_steps": 3534, "total_steps": 5364, "loss": 0.10858261585235596, "lr": 6.299035424738807e-06, "epoch": 1.976783216783217, "percentage": 65.88, "elapsed_time": "7:02:28", "remaining_time": "3:38:46"} +{"current_steps": 3535, "total_steps": 5364, "loss": 0.21731534600257874, "lr": 6.292989971949896e-06, "epoch": 1.9773426573426574, "percentage": 65.9, "elapsed_time": "7:02:36", "remaining_time": "3:38:39"} +{"current_steps": 3536, "total_steps": 5364, "loss": 0.1250826120376587, "lr": 6.286946089411319e-06, "epoch": 1.9779020979020978, "percentage": 65.92, "elapsed_time": "7:02:43", "remaining_time": "3:38:31"} +{"current_steps": 3537, "total_steps": 5364, "loss": 0.20018091797828674, "lr": 6.280903779683197e-06, "epoch": 1.9784615384615385, "percentage": 65.94, "elapsed_time": "7:02:50", "remaining_time": "3:38:24"} +{"current_steps": 3538, "total_steps": 5364, "loss": 0.13562744855880737, "lr": 6.274863045324987e-06, "epoch": 1.9790209790209792, "percentage": 65.96, "elapsed_time": "7:02:57", "remaining_time": "3:38:17"} +{"current_steps": 3539, "total_steps": 5364, "loss": 0.21992667019367218, "lr": 6.2688238888954835e-06, "epoch": 1.9795804195804196, "percentage": 65.98, "elapsed_time": "7:03:04", "remaining_time": "3:38:10"} +{"current_steps": 3540, "total_steps": 5364, "loss": 0.09611614048480988, "lr": 6.262786312952802e-06, "epoch": 1.98013986013986, "percentage": 66.0, "elapsed_time": "7:03:11", "remaining_time": "3:38:03"} +{"current_steps": 3541, "total_steps": 5364, "loss": 0.17659834027290344, "lr": 6.256750320054408e-06, "epoch": 1.9806993006993006, "percentage": 66.01, "elapsed_time": "7:03:18", "remaining_time": "3:37:55"} +{"current_steps": 3542, "total_steps": 5364, "loss": 0.077400803565979, "lr": 6.2507159127570775e-06, "epoch": 1.9812587412587412, "percentage": 66.03, "elapsed_time": "7:03:25", "remaining_time": "3:37:48"} +{"current_steps": 3543, "total_steps": 5364, "loss": 0.07500603049993515, "lr": 6.244683093616917e-06, "epoch": 1.981818181818182, "percentage": 66.05, "elapsed_time": "7:03:32", "remaining_time": "3:37:41"} +{"current_steps": 3544, "total_steps": 5364, "loss": 0.07867264002561569, "lr": 6.238651865189375e-06, "epoch": 1.9823776223776224, "percentage": 66.07, "elapsed_time": "7:03:39", "remaining_time": "3:37:34"} +{"current_steps": 3545, "total_steps": 5364, "loss": 0.2818150520324707, "lr": 6.232622230029208e-06, "epoch": 1.9829370629370628, "percentage": 66.09, "elapsed_time": "7:03:46", "remaining_time": "3:37:26"} +{"current_steps": 3546, "total_steps": 5364, "loss": 0.2510114312171936, "lr": 6.226594190690512e-06, "epoch": 1.9834965034965035, "percentage": 66.11, "elapsed_time": "7:03:53", "remaining_time": "3:37:19"} +{"current_steps": 3547, "total_steps": 5364, "loss": 0.10725396871566772, "lr": 6.220567749726695e-06, "epoch": 1.9840559440559442, "percentage": 66.13, "elapsed_time": "7:04:00", "remaining_time": "3:37:12"} +{"current_steps": 3548, "total_steps": 5364, "loss": 0.056303471326828, "lr": 6.2145429096905e-06, "epoch": 1.9846153846153847, "percentage": 66.14, "elapsed_time": "7:04:07", "remaining_time": "3:37:04"} +{"current_steps": 3549, "total_steps": 5364, "loss": 0.10670888423919678, "lr": 6.208519673133982e-06, "epoch": 1.9851748251748251, "percentage": 66.16, "elapsed_time": "7:04:14", "remaining_time": "3:36:57"} +{"current_steps": 3550, "total_steps": 5364, "loss": 0.11498218774795532, "lr": 6.20249804260852e-06, "epoch": 1.9857342657342656, "percentage": 66.18, "elapsed_time": "7:04:21", "remaining_time": "3:36:50"} +{"current_steps": 3551, "total_steps": 5364, "loss": 0.12872734665870667, "lr": 6.196478020664816e-06, "epoch": 1.9862937062937063, "percentage": 66.2, "elapsed_time": "7:04:28", "remaining_time": "3:36:43"} +{"current_steps": 3552, "total_steps": 5364, "loss": 0.14444312453269958, "lr": 6.1904596098528815e-06, "epoch": 1.986853146853147, "percentage": 66.22, "elapsed_time": "7:04:35", "remaining_time": "3:36:35"} +{"current_steps": 3553, "total_steps": 5364, "loss": 0.10586313903331757, "lr": 6.18444281272206e-06, "epoch": 1.9874125874125874, "percentage": 66.24, "elapsed_time": "7:04:42", "remaining_time": "3:36:28"} +{"current_steps": 3554, "total_steps": 5364, "loss": 0.26201170682907104, "lr": 6.178427631821e-06, "epoch": 1.9879720279720279, "percentage": 66.26, "elapsed_time": "7:04:49", "remaining_time": "3:36:21"} +{"current_steps": 3555, "total_steps": 5364, "loss": 0.032081738114356995, "lr": 6.172414069697665e-06, "epoch": 1.9885314685314686, "percentage": 66.28, "elapsed_time": "7:04:56", "remaining_time": "3:36:14"} +{"current_steps": 3556, "total_steps": 5364, "loss": 0.0964776948094368, "lr": 6.166402128899342e-06, "epoch": 1.9890909090909092, "percentage": 66.29, "elapsed_time": "7:05:03", "remaining_time": "3:36:07"} +{"current_steps": 3557, "total_steps": 5364, "loss": 0.07122602313756943, "lr": 6.160391811972621e-06, "epoch": 1.9896503496503497, "percentage": 66.31, "elapsed_time": "7:05:11", "remaining_time": "3:35:59"} +{"current_steps": 3558, "total_steps": 5364, "loss": 0.12019982933998108, "lr": 6.154383121463414e-06, "epoch": 1.9902097902097902, "percentage": 66.33, "elapsed_time": "7:05:17", "remaining_time": "3:35:52"} +{"current_steps": 3559, "total_steps": 5364, "loss": 0.1056625172495842, "lr": 6.148376059916936e-06, "epoch": 1.9907692307692306, "percentage": 66.35, "elapsed_time": "7:05:25", "remaining_time": "3:35:45"} +{"current_steps": 3560, "total_steps": 5364, "loss": 0.09969577193260193, "lr": 6.142370629877713e-06, "epoch": 1.9913286713286713, "percentage": 66.37, "elapsed_time": "7:05:32", "remaining_time": "3:35:38"} +{"current_steps": 3561, "total_steps": 5364, "loss": 0.2296912670135498, "lr": 6.136366833889585e-06, "epoch": 1.991888111888112, "percentage": 66.39, "elapsed_time": "7:05:38", "remaining_time": "3:35:30"} +{"current_steps": 3562, "total_steps": 5364, "loss": 0.250143826007843, "lr": 6.130364674495692e-06, "epoch": 1.9924475524475525, "percentage": 66.41, "elapsed_time": "7:05:46", "remaining_time": "3:35:23"} +{"current_steps": 3563, "total_steps": 5364, "loss": 0.19993560016155243, "lr": 6.1243641542384935e-06, "epoch": 1.993006993006993, "percentage": 66.42, "elapsed_time": "7:05:52", "remaining_time": "3:35:16"} +{"current_steps": 3564, "total_steps": 5364, "loss": 0.07463625073432922, "lr": 6.11836527565974e-06, "epoch": 1.9935664335664336, "percentage": 66.44, "elapsed_time": "7:06:00", "remaining_time": "3:35:09"} +{"current_steps": 3565, "total_steps": 5364, "loss": 0.07276284694671631, "lr": 6.112368041300494e-06, "epoch": 1.9941258741258743, "percentage": 66.46, "elapsed_time": "7:06:07", "remaining_time": "3:35:01"} +{"current_steps": 3566, "total_steps": 5364, "loss": 0.1839003711938858, "lr": 6.1063724537011235e-06, "epoch": 1.9946853146853147, "percentage": 66.48, "elapsed_time": "7:06:14", "remaining_time": "3:34:54"} +{"current_steps": 3567, "total_steps": 5364, "loss": 0.20200061798095703, "lr": 6.10037851540129e-06, "epoch": 1.9952447552447552, "percentage": 66.5, "elapsed_time": "7:06:21", "remaining_time": "3:34:47"} +{"current_steps": 3568, "total_steps": 5364, "loss": 0.2476254552602768, "lr": 6.094386228939973e-06, "epoch": 1.9958041958041957, "percentage": 66.52, "elapsed_time": "7:06:27", "remaining_time": "3:34:40"} +{"current_steps": 3569, "total_steps": 5364, "loss": 0.14904817938804626, "lr": 6.08839559685543e-06, "epoch": 1.9963636363636363, "percentage": 66.54, "elapsed_time": "7:06:35", "remaining_time": "3:34:32"} +{"current_steps": 3570, "total_steps": 5364, "loss": 0.1679742932319641, "lr": 6.0824066216852376e-06, "epoch": 1.996923076923077, "percentage": 66.55, "elapsed_time": "7:06:41", "remaining_time": "3:34:25"} +{"current_steps": 3571, "total_steps": 5364, "loss": 0.1967964768409729, "lr": 6.076419305966261e-06, "epoch": 1.9974825174825175, "percentage": 66.57, "elapsed_time": "7:06:48", "remaining_time": "3:34:18"} +{"current_steps": 3572, "total_steps": 5364, "loss": 0.07619272172451019, "lr": 6.07043365223466e-06, "epoch": 1.998041958041958, "percentage": 66.59, "elapsed_time": "7:06:55", "remaining_time": "3:34:10"} +{"current_steps": 3573, "total_steps": 5364, "loss": 0.07677794247865677, "lr": 6.064449663025901e-06, "epoch": 1.9986013986013986, "percentage": 66.61, "elapsed_time": "7:07:03", "remaining_time": "3:34:03"} +{"current_steps": 3574, "total_steps": 5364, "loss": 0.1892307549715042, "lr": 6.0584673408747294e-06, "epoch": 1.9991608391608393, "percentage": 66.63, "elapsed_time": "7:07:10", "remaining_time": "3:33:56"} +{"current_steps": 3575, "total_steps": 5364, "loss": 0.26105064153671265, "lr": 6.052486688315205e-06, "epoch": 1.9997202797202798, "percentage": 66.65, "elapsed_time": "7:07:17", "remaining_time": "3:33:49"} +{"current_steps": 3576, "total_steps": 5364, "loss": 0.06920918077230453, "lr": 6.0465077078806624e-06, "epoch": 2.0, "percentage": 66.67, "elapsed_time": "7:07:22", "remaining_time": "3:33:41"} +{"current_steps": 3577, "total_steps": 5364, "loss": 0.07399743795394897, "lr": 6.040530402103733e-06, "epoch": 2.0005594405594405, "percentage": 66.69, "elapsed_time": "7:07:29", "remaining_time": "3:33:33"} +{"current_steps": 3578, "total_steps": 5364, "loss": 0.12395565211772919, "lr": 6.034554773516345e-06, "epoch": 2.001118881118881, "percentage": 66.7, "elapsed_time": "7:07:36", "remaining_time": "3:33:26"} +{"current_steps": 3579, "total_steps": 5364, "loss": 0.0989704430103302, "lr": 6.028580824649708e-06, "epoch": 2.001678321678322, "percentage": 66.72, "elapsed_time": "7:07:43", "remaining_time": "3:33:19"} +{"current_steps": 3580, "total_steps": 5364, "loss": 0.0751051977276802, "lr": 6.022608558034327e-06, "epoch": 2.0022377622377623, "percentage": 66.74, "elapsed_time": "7:07:50", "remaining_time": "3:33:12"} +{"current_steps": 3581, "total_steps": 5364, "loss": 0.04803916811943054, "lr": 6.01663797619999e-06, "epoch": 2.0027972027972027, "percentage": 66.76, "elapsed_time": "7:07:57", "remaining_time": "3:33:05"} +{"current_steps": 3582, "total_steps": 5364, "loss": 0.08659067004919052, "lr": 6.01066908167577e-06, "epoch": 2.003356643356643, "percentage": 66.78, "elapsed_time": "7:08:05", "remaining_time": "3:32:58"} +{"current_steps": 3583, "total_steps": 5364, "loss": 0.14988979697227478, "lr": 6.004701876990033e-06, "epoch": 2.003916083916084, "percentage": 66.8, "elapsed_time": "7:08:12", "remaining_time": "3:32:50"} +{"current_steps": 3584, "total_steps": 5364, "loss": 0.08567748963832855, "lr": 5.9987363646704165e-06, "epoch": 2.0044755244755246, "percentage": 66.82, "elapsed_time": "7:08:19", "remaining_time": "3:32:43"} +{"current_steps": 3585, "total_steps": 5364, "loss": 0.06175285577774048, "lr": 5.992772547243856e-06, "epoch": 2.005034965034965, "percentage": 66.83, "elapsed_time": "7:08:26", "remaining_time": "3:32:36"} +{"current_steps": 3586, "total_steps": 5364, "loss": 0.06493330746889114, "lr": 5.986810427236556e-06, "epoch": 2.0055944055944055, "percentage": 66.85, "elapsed_time": "7:08:33", "remaining_time": "3:32:29"} +{"current_steps": 3587, "total_steps": 5364, "loss": 0.15454116463661194, "lr": 5.980850007174013e-06, "epoch": 2.006153846153846, "percentage": 66.87, "elapsed_time": "7:08:40", "remaining_time": "3:32:21"} +{"current_steps": 3588, "total_steps": 5364, "loss": 0.0849255844950676, "lr": 5.974891289580994e-06, "epoch": 2.006713286713287, "percentage": 66.89, "elapsed_time": "7:08:47", "remaining_time": "3:32:14"} +{"current_steps": 3589, "total_steps": 5364, "loss": 0.09450843185186386, "lr": 5.968934276981546e-06, "epoch": 2.0072727272727273, "percentage": 66.91, "elapsed_time": "7:08:55", "remaining_time": "3:32:07"} +{"current_steps": 3590, "total_steps": 5364, "loss": 0.04279526323080063, "lr": 5.962978971899005e-06, "epoch": 2.007832167832168, "percentage": 66.93, "elapsed_time": "7:09:02", "remaining_time": "3:32:00"} +{"current_steps": 3591, "total_steps": 5364, "loss": 0.18739965558052063, "lr": 5.95702537685597e-06, "epoch": 2.0083916083916082, "percentage": 66.95, "elapsed_time": "7:09:09", "remaining_time": "3:31:53"} +{"current_steps": 3592, "total_steps": 5364, "loss": 0.0595402792096138, "lr": 5.951073494374322e-06, "epoch": 2.008951048951049, "percentage": 66.96, "elapsed_time": "7:09:17", "remaining_time": "3:31:46"} +{"current_steps": 3593, "total_steps": 5364, "loss": 0.15191105008125305, "lr": 5.945123326975217e-06, "epoch": 2.0095104895104896, "percentage": 66.98, "elapsed_time": "7:09:24", "remaining_time": "3:31:39"} +{"current_steps": 3594, "total_steps": 5364, "loss": 0.008990811184048653, "lr": 5.939174877179079e-06, "epoch": 2.01006993006993, "percentage": 67.0, "elapsed_time": "7:09:31", "remaining_time": "3:31:32"} +{"current_steps": 3595, "total_steps": 5364, "loss": 0.024509279057383537, "lr": 5.933228147505614e-06, "epoch": 2.0106293706293705, "percentage": 67.02, "elapsed_time": "7:09:38", "remaining_time": "3:31:25"} +{"current_steps": 3596, "total_steps": 5364, "loss": 0.06581870466470718, "lr": 5.927283140473786e-06, "epoch": 2.011188811188811, "percentage": 67.04, "elapsed_time": "7:09:46", "remaining_time": "3:31:17"} +{"current_steps": 3597, "total_steps": 5364, "loss": 0.01926508918404579, "lr": 5.921339858601848e-06, "epoch": 2.011748251748252, "percentage": 67.06, "elapsed_time": "7:09:53", "remaining_time": "3:31:10"} +{"current_steps": 3598, "total_steps": 5364, "loss": 0.06254900246858597, "lr": 5.915398304407305e-06, "epoch": 2.0123076923076924, "percentage": 67.08, "elapsed_time": "7:10:00", "remaining_time": "3:31:03"} +{"current_steps": 3599, "total_steps": 5364, "loss": 0.04075920581817627, "lr": 5.909458480406936e-06, "epoch": 2.012867132867133, "percentage": 67.1, "elapsed_time": "7:10:07", "remaining_time": "3:30:56"} +{"current_steps": 3600, "total_steps": 5364, "loss": 0.14295360445976257, "lr": 5.903520389116791e-06, "epoch": 2.0134265734265733, "percentage": 67.11, "elapsed_time": "7:10:14", "remaining_time": "3:30:49"} +{"current_steps": 3601, "total_steps": 5364, "loss": 0.09693768620491028, "lr": 5.897584033052177e-06, "epoch": 2.013986013986014, "percentage": 67.13, "elapsed_time": "7:10:22", "remaining_time": "3:30:42"} +{"current_steps": 3602, "total_steps": 5364, "loss": 0.07173395901918411, "lr": 5.891649414727681e-06, "epoch": 2.0145454545454546, "percentage": 67.15, "elapsed_time": "7:10:29", "remaining_time": "3:30:35"} +{"current_steps": 3603, "total_steps": 5364, "loss": 0.03856249898672104, "lr": 5.8857165366571376e-06, "epoch": 2.015104895104895, "percentage": 67.17, "elapsed_time": "7:10:36", "remaining_time": "3:30:28"} +{"current_steps": 3604, "total_steps": 5364, "loss": 0.1216660887002945, "lr": 5.879785401353655e-06, "epoch": 2.0156643356643356, "percentage": 67.19, "elapsed_time": "7:10:44", "remaining_time": "3:30:20"} +{"current_steps": 3605, "total_steps": 5364, "loss": 0.05517837777733803, "lr": 5.873856011329599e-06, "epoch": 2.016223776223776, "percentage": 67.21, "elapsed_time": "7:10:50", "remaining_time": "3:30:13"} +{"current_steps": 3606, "total_steps": 5364, "loss": 0.1694219708442688, "lr": 5.86792836909659e-06, "epoch": 2.016783216783217, "percentage": 67.23, "elapsed_time": "7:10:58", "remaining_time": "3:30:06"} +{"current_steps": 3607, "total_steps": 5364, "loss": 0.056889232248067856, "lr": 5.862002477165525e-06, "epoch": 2.0173426573426574, "percentage": 67.24, "elapsed_time": "7:11:05", "remaining_time": "3:29:59"} +{"current_steps": 3608, "total_steps": 5364, "loss": 0.07298022508621216, "lr": 5.8560783380465425e-06, "epoch": 2.017902097902098, "percentage": 67.26, "elapsed_time": "7:11:12", "remaining_time": "3:29:52"} +{"current_steps": 3609, "total_steps": 5364, "loss": 0.05521957203745842, "lr": 5.850155954249049e-06, "epoch": 2.0184615384615383, "percentage": 67.28, "elapsed_time": "7:11:19", "remaining_time": "3:29:44"} +{"current_steps": 3610, "total_steps": 5364, "loss": 0.01638389565050602, "lr": 5.844235328281703e-06, "epoch": 2.019020979020979, "percentage": 67.3, "elapsed_time": "7:11:26", "remaining_time": "3:29:37"} +{"current_steps": 3611, "total_steps": 5364, "loss": 0.06502893567085266, "lr": 5.838316462652414e-06, "epoch": 2.0195804195804197, "percentage": 67.32, "elapsed_time": "7:11:33", "remaining_time": "3:29:30"} +{"current_steps": 3612, "total_steps": 5364, "loss": 0.04117986932396889, "lr": 5.832399359868361e-06, "epoch": 2.02013986013986, "percentage": 67.34, "elapsed_time": "7:11:40", "remaining_time": "3:29:22"} +{"current_steps": 3613, "total_steps": 5364, "loss": 0.05694076046347618, "lr": 5.826484022435954e-06, "epoch": 2.0206993006993006, "percentage": 67.36, "elapsed_time": "7:11:47", "remaining_time": "3:29:15"} +{"current_steps": 3614, "total_steps": 5364, "loss": 0.07224252074956894, "lr": 5.820570452860881e-06, "epoch": 2.021258741258741, "percentage": 67.38, "elapsed_time": "7:11:54", "remaining_time": "3:29:08"} +{"current_steps": 3615, "total_steps": 5364, "loss": 0.09924677759408951, "lr": 5.814658653648062e-06, "epoch": 2.021818181818182, "percentage": 67.39, "elapsed_time": "7:12:02", "remaining_time": "3:29:01"} +{"current_steps": 3616, "total_steps": 5364, "loss": 0.015852684155106544, "lr": 5.808748627301663e-06, "epoch": 2.0223776223776224, "percentage": 67.41, "elapsed_time": "7:12:09", "remaining_time": "3:28:54"} +{"current_steps": 3617, "total_steps": 5364, "loss": 0.09133298695087433, "lr": 5.802840376325125e-06, "epoch": 2.022937062937063, "percentage": 67.43, "elapsed_time": "7:12:16", "remaining_time": "3:28:47"} +{"current_steps": 3618, "total_steps": 5364, "loss": 0.18458610773086548, "lr": 5.796933903221109e-06, "epoch": 2.0234965034965033, "percentage": 67.45, "elapsed_time": "7:12:23", "remaining_time": "3:28:39"} +{"current_steps": 3619, "total_steps": 5364, "loss": 0.049970563501119614, "lr": 5.791029210491539e-06, "epoch": 2.0240559440559442, "percentage": 67.47, "elapsed_time": "7:12:31", "remaining_time": "3:28:32"} +{"current_steps": 3620, "total_steps": 5364, "loss": 0.057555388659238815, "lr": 5.7851263006375846e-06, "epoch": 2.0246153846153847, "percentage": 67.49, "elapsed_time": "7:12:38", "remaining_time": "3:28:25"} +{"current_steps": 3621, "total_steps": 5364, "loss": 0.015221508219838142, "lr": 5.779225176159644e-06, "epoch": 2.025174825174825, "percentage": 67.51, "elapsed_time": "7:12:45", "remaining_time": "3:28:18"} +{"current_steps": 3622, "total_steps": 5364, "loss": 0.13432274758815765, "lr": 5.773325839557389e-06, "epoch": 2.0257342657342656, "percentage": 67.52, "elapsed_time": "7:12:52", "remaining_time": "3:28:11"} +{"current_steps": 3623, "total_steps": 5364, "loss": 0.12969036400318146, "lr": 5.7674282933297e-06, "epoch": 2.026293706293706, "percentage": 67.54, "elapsed_time": "7:12:59", "remaining_time": "3:28:04"} +{"current_steps": 3624, "total_steps": 5364, "loss": 0.07945172488689423, "lr": 5.761532539974723e-06, "epoch": 2.026853146853147, "percentage": 67.56, "elapsed_time": "7:13:06", "remaining_time": "3:27:56"} +{"current_steps": 3625, "total_steps": 5364, "loss": 0.04767220839858055, "lr": 5.755638581989836e-06, "epoch": 2.0274125874125875, "percentage": 67.58, "elapsed_time": "7:13:12", "remaining_time": "3:27:49"} +{"current_steps": 3626, "total_steps": 5364, "loss": 0.09730111062526703, "lr": 5.749746421871656e-06, "epoch": 2.027972027972028, "percentage": 67.6, "elapsed_time": "7:13:20", "remaining_time": "3:27:42"} +{"current_steps": 3627, "total_steps": 5364, "loss": 0.07119136303663254, "lr": 5.743856062116046e-06, "epoch": 2.0285314685314684, "percentage": 67.62, "elapsed_time": "7:13:28", "remaining_time": "3:27:35"} +{"current_steps": 3628, "total_steps": 5364, "loss": 0.017523549497127533, "lr": 5.737967505218092e-06, "epoch": 2.0290909090909093, "percentage": 67.64, "elapsed_time": "7:13:35", "remaining_time": "3:27:28"} +{"current_steps": 3629, "total_steps": 5364, "loss": 0.0720461830496788, "lr": 5.732080753672125e-06, "epoch": 2.0296503496503497, "percentage": 67.65, "elapsed_time": "7:13:42", "remaining_time": "3:27:21"} +{"current_steps": 3630, "total_steps": 5364, "loss": 0.06236068159341812, "lr": 5.726195809971716e-06, "epoch": 2.03020979020979, "percentage": 67.67, "elapsed_time": "7:13:49", "remaining_time": "3:27:13"} +{"current_steps": 3631, "total_steps": 5364, "loss": 0.08773636817932129, "lr": 5.720312676609663e-06, "epoch": 2.0307692307692307, "percentage": 67.69, "elapsed_time": "7:13:56", "remaining_time": "3:27:06"} +{"current_steps": 3632, "total_steps": 5364, "loss": 0.06945449858903885, "lr": 5.714431356078002e-06, "epoch": 2.031328671328671, "percentage": 67.71, "elapsed_time": "7:14:03", "remaining_time": "3:26:59"} +{"current_steps": 3633, "total_steps": 5364, "loss": 0.06037995219230652, "lr": 5.708551850867991e-06, "epoch": 2.031888111888112, "percentage": 67.73, "elapsed_time": "7:14:10", "remaining_time": "3:26:52"} +{"current_steps": 3634, "total_steps": 5364, "loss": 0.02973555400967598, "lr": 5.702674163470132e-06, "epoch": 2.0324475524475525, "percentage": 67.75, "elapsed_time": "7:14:17", "remaining_time": "3:26:44"} +{"current_steps": 3635, "total_steps": 5364, "loss": 0.16984383761882782, "lr": 5.696798296374149e-06, "epoch": 2.033006993006993, "percentage": 67.77, "elapsed_time": "7:14:24", "remaining_time": "3:26:37"} +{"current_steps": 3636, "total_steps": 5364, "loss": 0.03337867558002472, "lr": 5.690924252068998e-06, "epoch": 2.0335664335664334, "percentage": 67.79, "elapsed_time": "7:14:31", "remaining_time": "3:26:30"} +{"current_steps": 3637, "total_steps": 5364, "loss": 0.014196659438312054, "lr": 5.685052033042866e-06, "epoch": 2.0341258741258743, "percentage": 67.8, "elapsed_time": "7:14:38", "remaining_time": "3:26:23"} +{"current_steps": 3638, "total_steps": 5364, "loss": 0.1894918531179428, "lr": 5.679181641783157e-06, "epoch": 2.0346853146853148, "percentage": 67.82, "elapsed_time": "7:14:45", "remaining_time": "3:26:15"} +{"current_steps": 3639, "total_steps": 5364, "loss": 0.016828207299113274, "lr": 5.673313080776508e-06, "epoch": 2.0352447552447552, "percentage": 67.84, "elapsed_time": "7:14:52", "remaining_time": "3:26:08"} +{"current_steps": 3640, "total_steps": 5364, "loss": 0.011041881516575813, "lr": 5.667446352508781e-06, "epoch": 2.0358041958041957, "percentage": 67.86, "elapsed_time": "7:15:00", "remaining_time": "3:26:01"} +{"current_steps": 3641, "total_steps": 5364, "loss": 0.03983404487371445, "lr": 5.661581459465064e-06, "epoch": 2.036363636363636, "percentage": 67.88, "elapsed_time": "7:15:07", "remaining_time": "3:25:54"} +{"current_steps": 3642, "total_steps": 5364, "loss": 0.13667380809783936, "lr": 5.655718404129653e-06, "epoch": 2.036923076923077, "percentage": 67.9, "elapsed_time": "7:15:14", "remaining_time": "3:25:47"} +{"current_steps": 3643, "total_steps": 5364, "loss": 0.026292983442544937, "lr": 5.6498571889860895e-06, "epoch": 2.0374825174825175, "percentage": 67.92, "elapsed_time": "7:15:21", "remaining_time": "3:25:40"} +{"current_steps": 3644, "total_steps": 5364, "loss": 0.1653956025838852, "lr": 5.643997816517113e-06, "epoch": 2.038041958041958, "percentage": 67.93, "elapsed_time": "7:15:28", "remaining_time": "3:25:33"} +{"current_steps": 3645, "total_steps": 5364, "loss": 0.06516967713832855, "lr": 5.638140289204694e-06, "epoch": 2.0386013986013984, "percentage": 67.95, "elapsed_time": "7:15:35", "remaining_time": "3:25:25"} +{"current_steps": 3646, "total_steps": 5364, "loss": 0.15873503684997559, "lr": 5.632284609530026e-06, "epoch": 2.0391608391608393, "percentage": 67.97, "elapsed_time": "7:15:43", "remaining_time": "3:25:18"} +{"current_steps": 3647, "total_steps": 5364, "loss": 0.22914071381092072, "lr": 5.6264307799734975e-06, "epoch": 2.03972027972028, "percentage": 67.99, "elapsed_time": "7:15:50", "remaining_time": "3:25:11"} +{"current_steps": 3648, "total_steps": 5364, "loss": 0.0227402001619339, "lr": 5.620578803014747e-06, "epoch": 2.0402797202797203, "percentage": 68.01, "elapsed_time": "7:15:57", "remaining_time": "3:25:04"} +{"current_steps": 3649, "total_steps": 5364, "loss": 0.07469518482685089, "lr": 5.614728681132597e-06, "epoch": 2.0408391608391607, "percentage": 68.03, "elapsed_time": "7:16:04", "remaining_time": "3:24:57"} +{"current_steps": 3650, "total_steps": 5364, "loss": 0.024593062698841095, "lr": 5.608880416805104e-06, "epoch": 2.041398601398601, "percentage": 68.05, "elapsed_time": "7:16:11", "remaining_time": "3:24:49"} +{"current_steps": 3651, "total_steps": 5364, "loss": 0.012602729722857475, "lr": 5.6030340125095315e-06, "epoch": 2.041958041958042, "percentage": 68.06, "elapsed_time": "7:16:18", "remaining_time": "3:24:42"} +{"current_steps": 3652, "total_steps": 5364, "loss": 0.04418928921222687, "lr": 5.597189470722346e-06, "epoch": 2.0425174825174826, "percentage": 68.08, "elapsed_time": "7:16:25", "remaining_time": "3:24:35"} +{"current_steps": 3653, "total_steps": 5364, "loss": 0.0394764319062233, "lr": 5.5913467939192475e-06, "epoch": 2.043076923076923, "percentage": 68.1, "elapsed_time": "7:16:32", "remaining_time": "3:24:28"} +{"current_steps": 3654, "total_steps": 5364, "loss": 0.04966995120048523, "lr": 5.585505984575122e-06, "epoch": 2.0436363636363635, "percentage": 68.12, "elapsed_time": "7:16:40", "remaining_time": "3:24:21"} +{"current_steps": 3655, "total_steps": 5364, "loss": 0.056858088821172714, "lr": 5.579667045164075e-06, "epoch": 2.0441958041958044, "percentage": 68.14, "elapsed_time": "7:16:47", "remaining_time": "3:24:14"} +{"current_steps": 3656, "total_steps": 5364, "loss": 0.12896110117435455, "lr": 5.5738299781594286e-06, "epoch": 2.044755244755245, "percentage": 68.16, "elapsed_time": "7:16:55", "remaining_time": "3:24:07"} +{"current_steps": 3657, "total_steps": 5364, "loss": 0.07391506433486938, "lr": 5.5679947860336884e-06, "epoch": 2.0453146853146853, "percentage": 68.18, "elapsed_time": "7:17:02", "remaining_time": "3:24:00"} +{"current_steps": 3658, "total_steps": 5364, "loss": 0.061476100236177444, "lr": 5.5621614712585945e-06, "epoch": 2.0458741258741258, "percentage": 68.2, "elapsed_time": "7:17:09", "remaining_time": "3:23:52"} +{"current_steps": 3659, "total_steps": 5364, "loss": 0.09372629970312119, "lr": 5.556330036305069e-06, "epoch": 2.046433566433566, "percentage": 68.21, "elapsed_time": "7:17:16", "remaining_time": "3:23:45"} +{"current_steps": 3660, "total_steps": 5364, "loss": 0.17711451649665833, "lr": 5.550500483643248e-06, "epoch": 2.046993006993007, "percentage": 68.23, "elapsed_time": "7:17:23", "remaining_time": "3:23:38"} +{"current_steps": 3661, "total_steps": 5364, "loss": 0.10096335411071777, "lr": 5.544672815742472e-06, "epoch": 2.0475524475524476, "percentage": 68.25, "elapsed_time": "7:17:30", "remaining_time": "3:23:31"} +{"current_steps": 3662, "total_steps": 5364, "loss": 0.11925114691257477, "lr": 5.538847035071268e-06, "epoch": 2.048111888111888, "percentage": 68.27, "elapsed_time": "7:17:37", "remaining_time": "3:23:23"} +{"current_steps": 3663, "total_steps": 5364, "loss": 0.13466320931911469, "lr": 5.533023144097394e-06, "epoch": 2.0486713286713285, "percentage": 68.29, "elapsed_time": "7:17:44", "remaining_time": "3:23:16"} +{"current_steps": 3664, "total_steps": 5364, "loss": 0.1448664665222168, "lr": 5.527201145287774e-06, "epoch": 2.0492307692307694, "percentage": 68.31, "elapsed_time": "7:17:51", "remaining_time": "3:23:09"} +{"current_steps": 3665, "total_steps": 5364, "loss": 0.05142068862915039, "lr": 5.5213810411085485e-06, "epoch": 2.04979020979021, "percentage": 68.33, "elapsed_time": "7:17:58", "remaining_time": "3:23:02"} +{"current_steps": 3666, "total_steps": 5364, "loss": 0.11228204518556595, "lr": 5.515562834025059e-06, "epoch": 2.0503496503496503, "percentage": 68.34, "elapsed_time": "7:18:05", "remaining_time": "3:22:54"} +{"current_steps": 3667, "total_steps": 5364, "loss": 0.05745711177587509, "lr": 5.509746526501827e-06, "epoch": 2.050909090909091, "percentage": 68.36, "elapsed_time": "7:18:13", "remaining_time": "3:22:47"} +{"current_steps": 3668, "total_steps": 5364, "loss": 0.041490714997053146, "lr": 5.503932121002582e-06, "epoch": 2.0514685314685313, "percentage": 68.38, "elapsed_time": "7:18:20", "remaining_time": "3:22:40"} +{"current_steps": 3669, "total_steps": 5364, "loss": 0.08994293212890625, "lr": 5.498119619990248e-06, "epoch": 2.052027972027972, "percentage": 68.4, "elapsed_time": "7:18:27", "remaining_time": "3:22:33"} +{"current_steps": 3670, "total_steps": 5364, "loss": 0.11252099275588989, "lr": 5.492309025926934e-06, "epoch": 2.0525874125874126, "percentage": 68.42, "elapsed_time": "7:18:34", "remaining_time": "3:22:26"} +{"current_steps": 3671, "total_steps": 5364, "loss": 0.06122881546616554, "lr": 5.486500341273956e-06, "epoch": 2.053146853146853, "percentage": 68.44, "elapsed_time": "7:18:42", "remaining_time": "3:22:19"} +{"current_steps": 3672, "total_steps": 5364, "loss": 0.031060781329870224, "lr": 5.4806935684917995e-06, "epoch": 2.0537062937062935, "percentage": 68.46, "elapsed_time": "7:18:49", "remaining_time": "3:22:12"} +{"current_steps": 3673, "total_steps": 5364, "loss": 0.07920639216899872, "lr": 5.4748887100401585e-06, "epoch": 2.0542657342657344, "percentage": 68.48, "elapsed_time": "7:18:56", "remaining_time": "3:22:04"} +{"current_steps": 3674, "total_steps": 5364, "loss": 0.14494416117668152, "lr": 5.469085768377909e-06, "epoch": 2.054825174825175, "percentage": 68.49, "elapsed_time": "7:19:03", "remaining_time": "3:21:57"} +{"current_steps": 3675, "total_steps": 5364, "loss": 0.14469082653522491, "lr": 5.463284745963114e-06, "epoch": 2.0553846153846154, "percentage": 68.51, "elapsed_time": "7:19:11", "remaining_time": "3:21:50"} +{"current_steps": 3676, "total_steps": 5364, "loss": 0.15794336795806885, "lr": 5.457485645253031e-06, "epoch": 2.055944055944056, "percentage": 68.53, "elapsed_time": "7:19:18", "remaining_time": "3:21:43"} +{"current_steps": 3677, "total_steps": 5364, "loss": 0.06799627840518951, "lr": 5.451688468704093e-06, "epoch": 2.0565034965034963, "percentage": 68.55, "elapsed_time": "7:19:24", "remaining_time": "3:21:36"} +{"current_steps": 3678, "total_steps": 5364, "loss": 0.030850786715745926, "lr": 5.44589321877192e-06, "epoch": 2.057062937062937, "percentage": 68.57, "elapsed_time": "7:19:31", "remaining_time": "3:21:28"} +{"current_steps": 3679, "total_steps": 5364, "loss": 0.13444456458091736, "lr": 5.440099897911325e-06, "epoch": 2.0576223776223777, "percentage": 68.59, "elapsed_time": "7:19:38", "remaining_time": "3:21:21"} +{"current_steps": 3680, "total_steps": 5364, "loss": 0.027570869773626328, "lr": 5.434308508576294e-06, "epoch": 2.058181818181818, "percentage": 68.61, "elapsed_time": "7:19:47", "remaining_time": "3:21:14"} +{"current_steps": 3681, "total_steps": 5364, "loss": 0.032223835587501526, "lr": 5.428519053220001e-06, "epoch": 2.0587412587412586, "percentage": 68.62, "elapsed_time": "7:19:53", "remaining_time": "3:21:07"} +{"current_steps": 3682, "total_steps": 5364, "loss": 0.06694259494543076, "lr": 5.422731534294798e-06, "epoch": 2.0593006993006995, "percentage": 68.64, "elapsed_time": "7:20:01", "remaining_time": "3:21:00"} +{"current_steps": 3683, "total_steps": 5364, "loss": 0.031319841742515564, "lr": 5.416945954252213e-06, "epoch": 2.05986013986014, "percentage": 68.66, "elapsed_time": "7:20:08", "remaining_time": "3:20:53"} +{"current_steps": 3684, "total_steps": 5364, "loss": 0.0646713376045227, "lr": 5.41116231554296e-06, "epoch": 2.0604195804195804, "percentage": 68.68, "elapsed_time": "7:20:15", "remaining_time": "3:20:46"} +{"current_steps": 3685, "total_steps": 5364, "loss": 0.019697466865181923, "lr": 5.405380620616926e-06, "epoch": 2.060979020979021, "percentage": 68.7, "elapsed_time": "7:20:22", "remaining_time": "3:20:38"} +{"current_steps": 3686, "total_steps": 5364, "loss": 0.07502421736717224, "lr": 5.399600871923176e-06, "epoch": 2.0615384615384613, "percentage": 68.72, "elapsed_time": "7:20:29", "remaining_time": "3:20:31"} +{"current_steps": 3687, "total_steps": 5364, "loss": 0.12852756679058075, "lr": 5.393823071909956e-06, "epoch": 2.0620979020979022, "percentage": 68.74, "elapsed_time": "7:20:36", "remaining_time": "3:20:24"} +{"current_steps": 3688, "total_steps": 5364, "loss": 0.24184030294418335, "lr": 5.388047223024671e-06, "epoch": 2.0626573426573427, "percentage": 68.75, "elapsed_time": "7:20:43", "remaining_time": "3:20:17"} +{"current_steps": 3689, "total_steps": 5364, "loss": 0.1504954844713211, "lr": 5.382273327713916e-06, "epoch": 2.063216783216783, "percentage": 68.77, "elapsed_time": "7:20:50", "remaining_time": "3:20:09"} +{"current_steps": 3690, "total_steps": 5364, "loss": 0.012958584353327751, "lr": 5.376501388423454e-06, "epoch": 2.0637762237762236, "percentage": 68.79, "elapsed_time": "7:20:57", "remaining_time": "3:20:02"} +{"current_steps": 3691, "total_steps": 5364, "loss": 0.136464461684227, "lr": 5.370731407598209e-06, "epoch": 2.0643356643356645, "percentage": 68.81, "elapsed_time": "7:21:04", "remaining_time": "3:19:55"} +{"current_steps": 3692, "total_steps": 5364, "loss": 0.047736674547195435, "lr": 5.364963387682295e-06, "epoch": 2.064895104895105, "percentage": 68.83, "elapsed_time": "7:21:11", "remaining_time": "3:19:48"} +{"current_steps": 3693, "total_steps": 5364, "loss": 0.15696394443511963, "lr": 5.359197331118975e-06, "epoch": 2.0654545454545454, "percentage": 68.85, "elapsed_time": "7:21:18", "remaining_time": "3:19:41"} +{"current_steps": 3694, "total_steps": 5364, "loss": 0.025747347623109818, "lr": 5.353433240350694e-06, "epoch": 2.066013986013986, "percentage": 68.87, "elapsed_time": "7:21:25", "remaining_time": "3:19:33"} +{"current_steps": 3695, "total_steps": 5364, "loss": 0.06233817711472511, "lr": 5.3476711178190646e-06, "epoch": 2.0665734265734264, "percentage": 68.89, "elapsed_time": "7:21:32", "remaining_time": "3:19:26"} +{"current_steps": 3696, "total_steps": 5364, "loss": 0.07642102241516113, "lr": 5.341910965964847e-06, "epoch": 2.0671328671328673, "percentage": 68.9, "elapsed_time": "7:21:39", "remaining_time": "3:19:19"} +{"current_steps": 3697, "total_steps": 5364, "loss": 0.095681332051754, "lr": 5.336152787227998e-06, "epoch": 2.0676923076923077, "percentage": 68.92, "elapsed_time": "7:21:47", "remaining_time": "3:19:12"} +{"current_steps": 3698, "total_steps": 5364, "loss": 0.01283261552453041, "lr": 5.33039658404761e-06, "epoch": 2.068251748251748, "percentage": 68.94, "elapsed_time": "7:21:54", "remaining_time": "3:19:04"} +{"current_steps": 3699, "total_steps": 5364, "loss": 0.04589240625500679, "lr": 5.324642358861953e-06, "epoch": 2.0688111888111886, "percentage": 68.96, "elapsed_time": "7:22:01", "remaining_time": "3:18:57"} +{"current_steps": 3700, "total_steps": 5364, "loss": 0.1537456512451172, "lr": 5.3188901141084595e-06, "epoch": 2.0693706293706295, "percentage": 68.98, "elapsed_time": "7:22:08", "remaining_time": "3:18:50"} +{"current_steps": 3701, "total_steps": 5364, "loss": 0.08238714933395386, "lr": 5.31313985222371e-06, "epoch": 2.06993006993007, "percentage": 69.0, "elapsed_time": "7:22:14", "remaining_time": "3:18:43"} +{"current_steps": 3702, "total_steps": 5364, "loss": 0.159087136387825, "lr": 5.307391575643467e-06, "epoch": 2.0704895104895105, "percentage": 69.02, "elapsed_time": "7:22:22", "remaining_time": "3:18:35"} +{"current_steps": 3703, "total_steps": 5364, "loss": 0.02902410924434662, "lr": 5.301645286802631e-06, "epoch": 2.071048951048951, "percentage": 69.03, "elapsed_time": "7:22:29", "remaining_time": "3:18:28"} +{"current_steps": 3704, "total_steps": 5364, "loss": 0.08081980049610138, "lr": 5.295900988135272e-06, "epoch": 2.0716083916083914, "percentage": 69.05, "elapsed_time": "7:22:36", "remaining_time": "3:18:21"} +{"current_steps": 3705, "total_steps": 5364, "loss": 0.009823895990848541, "lr": 5.29015868207462e-06, "epoch": 2.0721678321678323, "percentage": 69.07, "elapsed_time": "7:22:44", "remaining_time": "3:18:14"} +{"current_steps": 3706, "total_steps": 5364, "loss": 0.1737496256828308, "lr": 5.284418371053039e-06, "epoch": 2.0727272727272728, "percentage": 69.09, "elapsed_time": "7:22:50", "remaining_time": "3:18:07"} +{"current_steps": 3707, "total_steps": 5364, "loss": 0.02741951495409012, "lr": 5.2786800575020855e-06, "epoch": 2.073286713286713, "percentage": 69.11, "elapsed_time": "7:22:58", "remaining_time": "3:18:00"} +{"current_steps": 3708, "total_steps": 5364, "loss": 0.03793314844369888, "lr": 5.2729437438524335e-06, "epoch": 2.0738461538461537, "percentage": 69.13, "elapsed_time": "7:23:05", "remaining_time": "3:17:53"} +{"current_steps": 3709, "total_steps": 5364, "loss": 0.15087255835533142, "lr": 5.267209432533929e-06, "epoch": 2.0744055944055946, "percentage": 69.15, "elapsed_time": "7:23:13", "remaining_time": "3:17:46"} +{"current_steps": 3710, "total_steps": 5364, "loss": 0.025123829022049904, "lr": 5.261477125975571e-06, "epoch": 2.074965034965035, "percentage": 69.16, "elapsed_time": "7:23:20", "remaining_time": "3:17:39"} +{"current_steps": 3711, "total_steps": 5364, "loss": 0.05077960714697838, "lr": 5.25574682660549e-06, "epoch": 2.0755244755244755, "percentage": 69.18, "elapsed_time": "7:23:27", "remaining_time": "3:17:32"} +{"current_steps": 3712, "total_steps": 5364, "loss": 0.17125150561332703, "lr": 5.250018536851e-06, "epoch": 2.076083916083916, "percentage": 69.2, "elapsed_time": "7:23:34", "remaining_time": "3:17:24"} +{"current_steps": 3713, "total_steps": 5364, "loss": 0.06280983239412308, "lr": 5.244292259138528e-06, "epoch": 2.0766433566433564, "percentage": 69.22, "elapsed_time": "7:23:41", "remaining_time": "3:17:17"} +{"current_steps": 3714, "total_steps": 5364, "loss": 0.030946964398026466, "lr": 5.238567995893671e-06, "epoch": 2.0772027972027973, "percentage": 69.24, "elapsed_time": "7:23:48", "remaining_time": "3:17:10"} +{"current_steps": 3715, "total_steps": 5364, "loss": 0.06303784251213074, "lr": 5.23284574954117e-06, "epoch": 2.077762237762238, "percentage": 69.26, "elapsed_time": "7:23:55", "remaining_time": "3:17:03"} +{"current_steps": 3716, "total_steps": 5364, "loss": 0.16512702405452728, "lr": 5.2271255225049e-06, "epoch": 2.0783216783216782, "percentage": 69.28, "elapsed_time": "7:24:02", "remaining_time": "3:16:55"} +{"current_steps": 3717, "total_steps": 5364, "loss": 0.15320153534412384, "lr": 5.221407317207894e-06, "epoch": 2.0788811188811187, "percentage": 69.3, "elapsed_time": "7:24:09", "remaining_time": "3:16:48"} +{"current_steps": 3718, "total_steps": 5364, "loss": 0.016733992844820023, "lr": 5.2156911360723204e-06, "epoch": 2.0794405594405596, "percentage": 69.31, "elapsed_time": "7:24:17", "remaining_time": "3:16:41"} +{"current_steps": 3719, "total_steps": 5364, "loss": 0.17772963643074036, "lr": 5.2099769815194975e-06, "epoch": 2.08, "percentage": 69.33, "elapsed_time": "7:24:24", "remaining_time": "3:16:34"} +{"current_steps": 3720, "total_steps": 5364, "loss": 0.059646524488925934, "lr": 5.2042648559698785e-06, "epoch": 2.0805594405594405, "percentage": 69.35, "elapsed_time": "7:24:31", "remaining_time": "3:16:27"} +{"current_steps": 3721, "total_steps": 5364, "loss": 0.10271695256233215, "lr": 5.198554761843062e-06, "epoch": 2.081118881118881, "percentage": 69.37, "elapsed_time": "7:24:38", "remaining_time": "3:16:19"} +{"current_steps": 3722, "total_steps": 5364, "loss": 0.053607027977705, "lr": 5.192846701557779e-06, "epoch": 2.0816783216783215, "percentage": 69.39, "elapsed_time": "7:24:45", "remaining_time": "3:16:12"} +{"current_steps": 3723, "total_steps": 5364, "loss": 0.09202791005373001, "lr": 5.187140677531905e-06, "epoch": 2.0822377622377624, "percentage": 69.41, "elapsed_time": "7:24:52", "remaining_time": "3:16:05"} +{"current_steps": 3724, "total_steps": 5364, "loss": 0.10471323132514954, "lr": 5.181436692182453e-06, "epoch": 2.082797202797203, "percentage": 69.43, "elapsed_time": "7:24:59", "remaining_time": "3:15:58"} +{"current_steps": 3725, "total_steps": 5364, "loss": 0.09282371401786804, "lr": 5.175734747925572e-06, "epoch": 2.0833566433566433, "percentage": 69.44, "elapsed_time": "7:25:06", "remaining_time": "3:15:50"} +{"current_steps": 3726, "total_steps": 5364, "loss": 0.035135261714458466, "lr": 5.170034847176545e-06, "epoch": 2.0839160839160837, "percentage": 69.46, "elapsed_time": "7:25:13", "remaining_time": "3:15:43"} +{"current_steps": 3727, "total_steps": 5364, "loss": 0.011200472712516785, "lr": 5.164336992349786e-06, "epoch": 2.0844755244755246, "percentage": 69.48, "elapsed_time": "7:25:20", "remaining_time": "3:15:36"} +{"current_steps": 3728, "total_steps": 5364, "loss": 0.04953043907880783, "lr": 5.158641185858848e-06, "epoch": 2.085034965034965, "percentage": 69.5, "elapsed_time": "7:25:27", "remaining_time": "3:15:29"} +{"current_steps": 3729, "total_steps": 5364, "loss": 0.07613462209701538, "lr": 5.152947430116418e-06, "epoch": 2.0855944055944056, "percentage": 69.52, "elapsed_time": "7:25:34", "remaining_time": "3:15:22"} +{"current_steps": 3730, "total_steps": 5364, "loss": 0.13478463888168335, "lr": 5.147255727534305e-06, "epoch": 2.086153846153846, "percentage": 69.54, "elapsed_time": "7:25:41", "remaining_time": "3:15:14"} +{"current_steps": 3731, "total_steps": 5364, "loss": 0.08629865944385529, "lr": 5.141566080523463e-06, "epoch": 2.0867132867132865, "percentage": 69.56, "elapsed_time": "7:25:49", "remaining_time": "3:15:07"} +{"current_steps": 3732, "total_steps": 5364, "loss": 0.06047341600060463, "lr": 5.135878491493956e-06, "epoch": 2.0872727272727274, "percentage": 69.57, "elapsed_time": "7:25:55", "remaining_time": "3:15:00"} +{"current_steps": 3733, "total_steps": 5364, "loss": 0.1401153951883316, "lr": 5.130192962854992e-06, "epoch": 2.087832167832168, "percentage": 69.59, "elapsed_time": "7:26:02", "remaining_time": "3:14:53"} +{"current_steps": 3734, "total_steps": 5364, "loss": 0.1497367024421692, "lr": 5.124509497014901e-06, "epoch": 2.0883916083916083, "percentage": 69.61, "elapsed_time": "7:26:09", "remaining_time": "3:14:45"} +{"current_steps": 3735, "total_steps": 5364, "loss": 0.12784309685230255, "lr": 5.118828096381139e-06, "epoch": 2.0889510489510488, "percentage": 69.63, "elapsed_time": "7:26:17", "remaining_time": "3:14:38"} +{"current_steps": 3736, "total_steps": 5364, "loss": 0.21873131394386292, "lr": 5.113148763360289e-06, "epoch": 2.0895104895104897, "percentage": 69.65, "elapsed_time": "7:26:24", "remaining_time": "3:14:31"} +{"current_steps": 3737, "total_steps": 5364, "loss": 0.09356757998466492, "lr": 5.107471500358048e-06, "epoch": 2.09006993006993, "percentage": 69.67, "elapsed_time": "7:26:31", "remaining_time": "3:14:24"} +{"current_steps": 3738, "total_steps": 5364, "loss": 0.14247947931289673, "lr": 5.1017963097792585e-06, "epoch": 2.0906293706293706, "percentage": 69.69, "elapsed_time": "7:26:38", "remaining_time": "3:14:16"} +{"current_steps": 3739, "total_steps": 5364, "loss": 0.14984065294265747, "lr": 5.096123194027864e-06, "epoch": 2.091188811188811, "percentage": 69.71, "elapsed_time": "7:26:45", "remaining_time": "3:14:09"} +{"current_steps": 3740, "total_steps": 5364, "loss": 0.09576083719730377, "lr": 5.0904521555069306e-06, "epoch": 2.091748251748252, "percentage": 69.72, "elapsed_time": "7:26:52", "remaining_time": "3:14:02"} +{"current_steps": 3741, "total_steps": 5364, "loss": 0.04242061451077461, "lr": 5.084783196618662e-06, "epoch": 2.0923076923076924, "percentage": 69.74, "elapsed_time": "7:26:59", "remaining_time": "3:13:55"} +{"current_steps": 3742, "total_steps": 5364, "loss": 0.14414675533771515, "lr": 5.0791163197643615e-06, "epoch": 2.092867132867133, "percentage": 69.76, "elapsed_time": "7:27:06", "remaining_time": "3:13:48"} +{"current_steps": 3743, "total_steps": 5364, "loss": 0.12326627224683762, "lr": 5.0734515273444585e-06, "epoch": 2.0934265734265733, "percentage": 69.78, "elapsed_time": "7:27:13", "remaining_time": "3:13:41"} +{"current_steps": 3744, "total_steps": 5364, "loss": 0.019266968593001366, "lr": 5.067788821758506e-06, "epoch": 2.093986013986014, "percentage": 69.8, "elapsed_time": "7:27:21", "remaining_time": "3:13:33"} +{"current_steps": 3745, "total_steps": 5364, "loss": 0.048575010150671005, "lr": 5.0621282054051536e-06, "epoch": 2.0945454545454547, "percentage": 69.82, "elapsed_time": "7:27:27", "remaining_time": "3:13:26"} +{"current_steps": 3746, "total_steps": 5364, "loss": 0.04107515886425972, "lr": 5.056469680682194e-06, "epoch": 2.095104895104895, "percentage": 69.84, "elapsed_time": "7:27:34", "remaining_time": "3:13:19"} +{"current_steps": 3747, "total_steps": 5364, "loss": 0.07658736407756805, "lr": 5.050813249986508e-06, "epoch": 2.0956643356643356, "percentage": 69.85, "elapsed_time": "7:27:42", "remaining_time": "3:13:12"} +{"current_steps": 3748, "total_steps": 5364, "loss": 0.10319835692644119, "lr": 5.0451589157141035e-06, "epoch": 2.096223776223776, "percentage": 69.87, "elapsed_time": "7:27:49", "remaining_time": "3:13:05"} +{"current_steps": 3749, "total_steps": 5364, "loss": 0.18548107147216797, "lr": 5.0395066802601e-06, "epoch": 2.096783216783217, "percentage": 69.89, "elapsed_time": "7:27:56", "remaining_time": "3:12:58"} +{"current_steps": 3750, "total_steps": 5364, "loss": 0.05447522923350334, "lr": 5.033856546018714e-06, "epoch": 2.0973426573426575, "percentage": 69.91, "elapsed_time": "7:28:04", "remaining_time": "3:12:50"} +{"current_steps": 3751, "total_steps": 5364, "loss": 0.1492200642824173, "lr": 5.028208515383297e-06, "epoch": 2.097902097902098, "percentage": 69.93, "elapsed_time": "7:28:10", "remaining_time": "3:12:43"} +{"current_steps": 3752, "total_steps": 5364, "loss": 0.02554817497730255, "lr": 5.022562590746285e-06, "epoch": 2.0984615384615384, "percentage": 69.95, "elapsed_time": "7:28:18", "remaining_time": "3:12:36"} +{"current_steps": 3753, "total_steps": 5364, "loss": 0.060051798820495605, "lr": 5.016918774499235e-06, "epoch": 2.099020979020979, "percentage": 69.97, "elapsed_time": "7:28:25", "remaining_time": "3:12:29"} +{"current_steps": 3754, "total_steps": 5364, "loss": 0.23002095520496368, "lr": 5.011277069032809e-06, "epoch": 2.0995804195804197, "percentage": 69.99, "elapsed_time": "7:28:32", "remaining_time": "3:12:22"} +{"current_steps": 3755, "total_steps": 5364, "loss": 0.13818466663360596, "lr": 5.005637476736773e-06, "epoch": 2.10013986013986, "percentage": 70.0, "elapsed_time": "7:28:39", "remaining_time": "3:12:15"} +{"current_steps": 3756, "total_steps": 5364, "loss": 0.12224321067333221, "lr": 5.000000000000003e-06, "epoch": 2.1006993006993007, "percentage": 70.02, "elapsed_time": "7:28:46", "remaining_time": "3:12:07"} +{"current_steps": 3757, "total_steps": 5364, "loss": 0.009026276879012585, "lr": 4.9943646412104665e-06, "epoch": 2.101258741258741, "percentage": 70.04, "elapsed_time": "7:28:53", "remaining_time": "3:12:00"} +{"current_steps": 3758, "total_steps": 5364, "loss": 0.042649008333683014, "lr": 4.9887314027552455e-06, "epoch": 2.101818181818182, "percentage": 70.06, "elapsed_time": "7:29:00", "remaining_time": "3:11:53"} +{"current_steps": 3759, "total_steps": 5364, "loss": 0.12513811886310577, "lr": 4.9831002870205216e-06, "epoch": 2.1023776223776225, "percentage": 70.08, "elapsed_time": "7:29:08", "remaining_time": "3:11:46"} +{"current_steps": 3760, "total_steps": 5364, "loss": 0.04833667725324631, "lr": 4.977471296391575e-06, "epoch": 2.102937062937063, "percentage": 70.1, "elapsed_time": "7:29:15", "remaining_time": "3:11:38"} +{"current_steps": 3761, "total_steps": 5364, "loss": 0.03420846164226532, "lr": 4.971844433252791e-06, "epoch": 2.1034965034965034, "percentage": 70.12, "elapsed_time": "7:29:21", "remaining_time": "3:11:31"} +{"current_steps": 3762, "total_steps": 5364, "loss": 0.08666019141674042, "lr": 4.966219699987643e-06, "epoch": 2.104055944055944, "percentage": 70.13, "elapsed_time": "7:29:28", "remaining_time": "3:11:24"} +{"current_steps": 3763, "total_steps": 5364, "loss": 0.07298225909471512, "lr": 4.960597098978711e-06, "epoch": 2.1046153846153848, "percentage": 70.15, "elapsed_time": "7:29:35", "remaining_time": "3:11:17"} +{"current_steps": 3764, "total_steps": 5364, "loss": 0.0925387293100357, "lr": 4.954976632607669e-06, "epoch": 2.1051748251748252, "percentage": 70.17, "elapsed_time": "7:29:43", "remaining_time": "3:11:09"} +{"current_steps": 3765, "total_steps": 5364, "loss": 0.10714008659124374, "lr": 4.949358303255293e-06, "epoch": 2.1057342657342657, "percentage": 70.19, "elapsed_time": "7:29:50", "remaining_time": "3:11:02"} +{"current_steps": 3766, "total_steps": 5364, "loss": 0.023769894614815712, "lr": 4.943742113301437e-06, "epoch": 2.106293706293706, "percentage": 70.21, "elapsed_time": "7:29:57", "remaining_time": "3:10:55"} +{"current_steps": 3767, "total_steps": 5364, "loss": 0.01928844302892685, "lr": 4.9381280651250685e-06, "epoch": 2.106853146853147, "percentage": 70.23, "elapsed_time": "7:30:04", "remaining_time": "3:10:48"} +{"current_steps": 3768, "total_steps": 5364, "loss": 0.06522198766469955, "lr": 4.932516161104235e-06, "epoch": 2.1074125874125875, "percentage": 70.25, "elapsed_time": "7:30:12", "remaining_time": "3:10:41"} +{"current_steps": 3769, "total_steps": 5364, "loss": 0.05866191163659096, "lr": 4.926906403616081e-06, "epoch": 2.107972027972028, "percentage": 70.26, "elapsed_time": "7:30:19", "remaining_time": "3:10:34"} +{"current_steps": 3770, "total_steps": 5364, "loss": 0.16815561056137085, "lr": 4.921298795036845e-06, "epoch": 2.1085314685314684, "percentage": 70.28, "elapsed_time": "7:30:26", "remaining_time": "3:10:27"} +{"current_steps": 3771, "total_steps": 5364, "loss": 0.08497156202793121, "lr": 4.9156933377418425e-06, "epoch": 2.109090909090909, "percentage": 70.3, "elapsed_time": "7:30:33", "remaining_time": "3:10:19"} +{"current_steps": 3772, "total_steps": 5364, "loss": 0.13801153004169464, "lr": 4.91009003410549e-06, "epoch": 2.10965034965035, "percentage": 70.32, "elapsed_time": "7:30:40", "remaining_time": "3:10:12"} +{"current_steps": 3773, "total_steps": 5364, "loss": 0.07784494012594223, "lr": 4.9044888865012884e-06, "epoch": 2.1102097902097903, "percentage": 70.34, "elapsed_time": "7:30:47", "remaining_time": "3:10:05"} +{"current_steps": 3774, "total_steps": 5364, "loss": 0.03854656592011452, "lr": 4.898889897301824e-06, "epoch": 2.1107692307692307, "percentage": 70.36, "elapsed_time": "7:30:54", "remaining_time": "3:09:58"} +{"current_steps": 3775, "total_steps": 5364, "loss": 0.117255300283432, "lr": 4.893293068878774e-06, "epoch": 2.111328671328671, "percentage": 70.38, "elapsed_time": "7:31:02", "remaining_time": "3:09:51"} +{"current_steps": 3776, "total_steps": 5364, "loss": 0.09304417669773102, "lr": 4.887698403602884e-06, "epoch": 2.111888111888112, "percentage": 70.4, "elapsed_time": "7:31:09", "remaining_time": "3:09:43"} +{"current_steps": 3777, "total_steps": 5364, "loss": 0.03790184110403061, "lr": 4.88210590384401e-06, "epoch": 2.1124475524475526, "percentage": 70.41, "elapsed_time": "7:31:17", "remaining_time": "3:09:37"} +{"current_steps": 3778, "total_steps": 5364, "loss": 0.15130586922168732, "lr": 4.876515571971065e-06, "epoch": 2.113006993006993, "percentage": 70.43, "elapsed_time": "7:31:24", "remaining_time": "3:09:30"} +{"current_steps": 3779, "total_steps": 5364, "loss": 0.0512462854385376, "lr": 4.870927410352059e-06, "epoch": 2.1135664335664335, "percentage": 70.45, "elapsed_time": "7:31:31", "remaining_time": "3:09:22"} +{"current_steps": 3780, "total_steps": 5364, "loss": 0.03197665885090828, "lr": 4.865341421354081e-06, "epoch": 2.114125874125874, "percentage": 70.47, "elapsed_time": "7:31:38", "remaining_time": "3:09:15"} +{"current_steps": 3781, "total_steps": 5364, "loss": 0.04053480923175812, "lr": 4.859757607343286e-06, "epoch": 2.114685314685315, "percentage": 70.49, "elapsed_time": "7:31:45", "remaining_time": "3:09:08"} +{"current_steps": 3782, "total_steps": 5364, "loss": 0.0027559478767216206, "lr": 4.854175970684934e-06, "epoch": 2.1152447552447553, "percentage": 70.51, "elapsed_time": "7:31:52", "remaining_time": "3:09:01"} +{"current_steps": 3783, "total_steps": 5364, "loss": 0.006786721758544445, "lr": 4.848596513743336e-06, "epoch": 2.1158041958041958, "percentage": 70.53, "elapsed_time": "7:32:00", "remaining_time": "3:08:54"} +{"current_steps": 3784, "total_steps": 5364, "loss": 0.10572367161512375, "lr": 4.843019238881894e-06, "epoch": 2.1163636363636362, "percentage": 70.54, "elapsed_time": "7:32:07", "remaining_time": "3:08:47"} +{"current_steps": 3785, "total_steps": 5364, "loss": 0.07152760773897171, "lr": 4.837444148463087e-06, "epoch": 2.116923076923077, "percentage": 70.56, "elapsed_time": "7:32:14", "remaining_time": "3:08:39"} +{"current_steps": 3786, "total_steps": 5364, "loss": 0.05206223577260971, "lr": 4.831871244848455e-06, "epoch": 2.1174825174825176, "percentage": 70.58, "elapsed_time": "7:32:21", "remaining_time": "3:08:32"} +{"current_steps": 3787, "total_steps": 5364, "loss": 0.09001053869724274, "lr": 4.826300530398635e-06, "epoch": 2.118041958041958, "percentage": 70.6, "elapsed_time": "7:32:28", "remaining_time": "3:08:25"} +{"current_steps": 3788, "total_steps": 5364, "loss": 0.008701407350599766, "lr": 4.8207320074733136e-06, "epoch": 2.1186013986013985, "percentage": 70.62, "elapsed_time": "7:32:36", "remaining_time": "3:08:18"} +{"current_steps": 3789, "total_steps": 5364, "loss": 0.03854415565729141, "lr": 4.815165678431253e-06, "epoch": 2.119160839160839, "percentage": 70.64, "elapsed_time": "7:32:43", "remaining_time": "3:08:11"} +{"current_steps": 3790, "total_steps": 5364, "loss": 0.009942376986145973, "lr": 4.809601545630305e-06, "epoch": 2.11972027972028, "percentage": 70.66, "elapsed_time": "7:32:50", "remaining_time": "3:08:04"} +{"current_steps": 3791, "total_steps": 5364, "loss": 0.13618825376033783, "lr": 4.804039611427367e-06, "epoch": 2.1202797202797203, "percentage": 70.67, "elapsed_time": "7:32:57", "remaining_time": "3:07:56"} +{"current_steps": 3792, "total_steps": 5364, "loss": 0.046507224440574646, "lr": 4.798479878178421e-06, "epoch": 2.120839160839161, "percentage": 70.69, "elapsed_time": "7:33:05", "remaining_time": "3:07:49"} +{"current_steps": 3793, "total_steps": 5364, "loss": 0.1054900586605072, "lr": 4.792922348238508e-06, "epoch": 2.1213986013986013, "percentage": 70.71, "elapsed_time": "7:33:12", "remaining_time": "3:07:42"} +{"current_steps": 3794, "total_steps": 5364, "loss": 0.02644207701086998, "lr": 4.787367023961741e-06, "epoch": 2.121958041958042, "percentage": 70.73, "elapsed_time": "7:33:18", "remaining_time": "3:07:35"} +{"current_steps": 3795, "total_steps": 5364, "loss": 0.13420142233371735, "lr": 4.781813907701301e-06, "epoch": 2.1225174825174826, "percentage": 70.75, "elapsed_time": "7:33:25", "remaining_time": "3:07:27"} +{"current_steps": 3796, "total_steps": 5364, "loss": 0.09387016296386719, "lr": 4.7762630018094216e-06, "epoch": 2.123076923076923, "percentage": 70.77, "elapsed_time": "7:33:32", "remaining_time": "3:07:20"} +{"current_steps": 3797, "total_steps": 5364, "loss": 0.04379844665527344, "lr": 4.77071430863741e-06, "epoch": 2.1236363636363635, "percentage": 70.79, "elapsed_time": "7:33:39", "remaining_time": "3:07:13"} +{"current_steps": 3798, "total_steps": 5364, "loss": 0.05403991416096687, "lr": 4.765167830535637e-06, "epoch": 2.124195804195804, "percentage": 70.81, "elapsed_time": "7:33:47", "remaining_time": "3:07:06"} +{"current_steps": 3799, "total_steps": 5364, "loss": 0.1604430228471756, "lr": 4.759623569853528e-06, "epoch": 2.124755244755245, "percentage": 70.82, "elapsed_time": "7:33:54", "remaining_time": "3:06:59"} +{"current_steps": 3800, "total_steps": 5364, "loss": 0.10386520624160767, "lr": 4.754081528939581e-06, "epoch": 2.1253146853146854, "percentage": 70.84, "elapsed_time": "7:34:01", "remaining_time": "3:06:52"} +{"current_steps": 3801, "total_steps": 5364, "loss": 0.11015531420707703, "lr": 4.748541710141339e-06, "epoch": 2.125874125874126, "percentage": 70.86, "elapsed_time": "7:34:08", "remaining_time": "3:06:44"} +{"current_steps": 3802, "total_steps": 5364, "loss": 0.12145122140645981, "lr": 4.7430041158054105e-06, "epoch": 2.1264335664335663, "percentage": 70.88, "elapsed_time": "7:34:15", "remaining_time": "3:06:37"} +{"current_steps": 3803, "total_steps": 5364, "loss": 0.00788156595081091, "lr": 4.737468748277465e-06, "epoch": 2.126993006993007, "percentage": 70.9, "elapsed_time": "7:34:23", "remaining_time": "3:06:30"} +{"current_steps": 3804, "total_steps": 5364, "loss": 0.04790748655796051, "lr": 4.731935609902225e-06, "epoch": 2.1275524475524477, "percentage": 70.92, "elapsed_time": "7:34:30", "remaining_time": "3:06:23"} +{"current_steps": 3805, "total_steps": 5364, "loss": 0.040815550833940506, "lr": 4.726404703023474e-06, "epoch": 2.128111888111888, "percentage": 70.94, "elapsed_time": "7:34:37", "remaining_time": "3:06:16"} +{"current_steps": 3806, "total_steps": 5364, "loss": 0.04240987449884415, "lr": 4.720876029984037e-06, "epoch": 2.1286713286713286, "percentage": 70.95, "elapsed_time": "7:34:44", "remaining_time": "3:06:09"} +{"current_steps": 3807, "total_steps": 5364, "loss": 0.013739352114498615, "lr": 4.715349593125805e-06, "epoch": 2.129230769230769, "percentage": 70.97, "elapsed_time": "7:34:51", "remaining_time": "3:06:01"} +{"current_steps": 3808, "total_steps": 5364, "loss": 0.09310592710971832, "lr": 4.709825394789721e-06, "epoch": 2.12979020979021, "percentage": 70.99, "elapsed_time": "7:34:58", "remaining_time": "3:05:54"} +{"current_steps": 3809, "total_steps": 5364, "loss": 0.04799744114279747, "lr": 4.704303437315774e-06, "epoch": 2.1303496503496504, "percentage": 71.01, "elapsed_time": "7:35:05", "remaining_time": "3:05:47"} +{"current_steps": 3810, "total_steps": 5364, "loss": 0.12010888010263443, "lr": 4.69878372304301e-06, "epoch": 2.130909090909091, "percentage": 71.03, "elapsed_time": "7:35:12", "remaining_time": "3:05:40"} +{"current_steps": 3811, "total_steps": 5364, "loss": 0.03949858993291855, "lr": 4.693266254309522e-06, "epoch": 2.1314685314685313, "percentage": 71.05, "elapsed_time": "7:35:20", "remaining_time": "3:05:33"} +{"current_steps": 3812, "total_steps": 5364, "loss": 0.1470131129026413, "lr": 4.687751033452447e-06, "epoch": 2.1320279720279722, "percentage": 71.07, "elapsed_time": "7:35:27", "remaining_time": "3:05:25"} +{"current_steps": 3813, "total_steps": 5364, "loss": 0.1138114482164383, "lr": 4.6822380628079755e-06, "epoch": 2.1325874125874127, "percentage": 71.09, "elapsed_time": "7:35:34", "remaining_time": "3:05:18"} +{"current_steps": 3814, "total_steps": 5364, "loss": 0.07835336029529572, "lr": 4.676727344711351e-06, "epoch": 2.133146853146853, "percentage": 71.1, "elapsed_time": "7:35:41", "remaining_time": "3:05:11"} +{"current_steps": 3815, "total_steps": 5364, "loss": 0.032216355204582214, "lr": 4.67121888149684e-06, "epoch": 2.1337062937062936, "percentage": 71.12, "elapsed_time": "7:35:48", "remaining_time": "3:05:04"} +{"current_steps": 3816, "total_steps": 5364, "loss": 0.09749799966812134, "lr": 4.665712675497788e-06, "epoch": 2.134265734265734, "percentage": 71.14, "elapsed_time": "7:35:55", "remaining_time": "3:04:56"} +{"current_steps": 3817, "total_steps": 5364, "loss": 0.07754640281200409, "lr": 4.660208729046552e-06, "epoch": 2.134825174825175, "percentage": 71.16, "elapsed_time": "7:36:02", "remaining_time": "3:04:49"} +{"current_steps": 3818, "total_steps": 5364, "loss": 0.1486811339855194, "lr": 4.654707044474551e-06, "epoch": 2.1353846153846154, "percentage": 71.18, "elapsed_time": "7:36:10", "remaining_time": "3:04:42"} +{"current_steps": 3819, "total_steps": 5364, "loss": 0.06307768821716309, "lr": 4.649207624112243e-06, "epoch": 2.135944055944056, "percentage": 71.2, "elapsed_time": "7:36:17", "remaining_time": "3:04:35"} +{"current_steps": 3820, "total_steps": 5364, "loss": 0.12122844159603119, "lr": 4.6437104702891146e-06, "epoch": 2.1365034965034964, "percentage": 71.22, "elapsed_time": "7:36:24", "remaining_time": "3:04:28"} +{"current_steps": 3821, "total_steps": 5364, "loss": 0.002600582782179117, "lr": 4.638215585333717e-06, "epoch": 2.1370629370629373, "percentage": 71.23, "elapsed_time": "7:36:31", "remaining_time": "3:04:21"} +{"current_steps": 3822, "total_steps": 5364, "loss": 0.09703333675861359, "lr": 4.6327229715736125e-06, "epoch": 2.1376223776223777, "percentage": 71.25, "elapsed_time": "7:36:39", "remaining_time": "3:04:14"} +{"current_steps": 3823, "total_steps": 5364, "loss": 0.11477223038673401, "lr": 4.627232631335419e-06, "epoch": 2.138181818181818, "percentage": 71.27, "elapsed_time": "7:36:45", "remaining_time": "3:04:06"} +{"current_steps": 3824, "total_steps": 5364, "loss": 0.137488454580307, "lr": 4.621744566944791e-06, "epoch": 2.1387412587412586, "percentage": 71.29, "elapsed_time": "7:36:53", "remaining_time": "3:03:59"} +{"current_steps": 3825, "total_steps": 5364, "loss": 0.06826088577508926, "lr": 4.616258780726404e-06, "epoch": 2.139300699300699, "percentage": 71.31, "elapsed_time": "7:37:00", "remaining_time": "3:03:52"} +{"current_steps": 3826, "total_steps": 5364, "loss": 0.036990582942962646, "lr": 4.610775275003992e-06, "epoch": 2.13986013986014, "percentage": 71.33, "elapsed_time": "7:37:07", "remaining_time": "3:03:45"} +{"current_steps": 3827, "total_steps": 5364, "loss": 0.23528940975666046, "lr": 4.605294052100301e-06, "epoch": 2.1404195804195805, "percentage": 71.35, "elapsed_time": "7:37:14", "remaining_time": "3:03:38"} +{"current_steps": 3828, "total_steps": 5364, "loss": 0.05900619924068451, "lr": 4.599815114337122e-06, "epoch": 2.140979020979021, "percentage": 71.36, "elapsed_time": "7:37:22", "remaining_time": "3:03:31"} +{"current_steps": 3829, "total_steps": 5364, "loss": 0.08169350028038025, "lr": 4.594338464035278e-06, "epoch": 2.1415384615384614, "percentage": 71.38, "elapsed_time": "7:37:29", "remaining_time": "3:03:24"} +{"current_steps": 3830, "total_steps": 5364, "loss": 0.16644254326820374, "lr": 4.588864103514612e-06, "epoch": 2.1420979020979023, "percentage": 71.4, "elapsed_time": "7:37:36", "remaining_time": "3:03:17"} +{"current_steps": 3831, "total_steps": 5364, "loss": 0.07631094753742218, "lr": 4.583392035094017e-06, "epoch": 2.1426573426573428, "percentage": 71.42, "elapsed_time": "7:37:44", "remaining_time": "3:03:09"} +{"current_steps": 3832, "total_steps": 5364, "loss": 0.08085634559392929, "lr": 4.577922261091394e-06, "epoch": 2.143216783216783, "percentage": 71.44, "elapsed_time": "7:37:51", "remaining_time": "3:03:02"} +{"current_steps": 3833, "total_steps": 5364, "loss": 0.08495327830314636, "lr": 4.572454783823685e-06, "epoch": 2.1437762237762237, "percentage": 71.46, "elapsed_time": "7:37:58", "remaining_time": "3:02:55"} +{"current_steps": 3834, "total_steps": 5364, "loss": 0.05321191996335983, "lr": 4.56698960560686e-06, "epoch": 2.144335664335664, "percentage": 71.48, "elapsed_time": "7:38:05", "remaining_time": "3:02:48"} +{"current_steps": 3835, "total_steps": 5364, "loss": 0.12114525586366653, "lr": 4.561526728755897e-06, "epoch": 2.144895104895105, "percentage": 71.5, "elapsed_time": "7:38:13", "remaining_time": "3:02:41"} +{"current_steps": 3836, "total_steps": 5364, "loss": 0.2179345190525055, "lr": 4.556066155584832e-06, "epoch": 2.1454545454545455, "percentage": 71.51, "elapsed_time": "7:38:20", "remaining_time": "3:02:34"} +{"current_steps": 3837, "total_steps": 5364, "loss": 0.11883588880300522, "lr": 4.550607888406692e-06, "epoch": 2.146013986013986, "percentage": 71.53, "elapsed_time": "7:38:28", "remaining_time": "3:02:27"} +{"current_steps": 3838, "total_steps": 5364, "loss": 0.10780727118253708, "lr": 4.545151929533546e-06, "epoch": 2.1465734265734264, "percentage": 71.55, "elapsed_time": "7:38:34", "remaining_time": "3:02:20"} +{"current_steps": 3839, "total_steps": 5364, "loss": 0.07032984495162964, "lr": 4.539698281276483e-06, "epoch": 2.1471328671328673, "percentage": 71.57, "elapsed_time": "7:38:41", "remaining_time": "3:02:12"} +{"current_steps": 3840, "total_steps": 5364, "loss": 0.05536240339279175, "lr": 4.5342469459456065e-06, "epoch": 2.147692307692308, "percentage": 71.59, "elapsed_time": "7:38:48", "remaining_time": "3:02:05"} +{"current_steps": 3841, "total_steps": 5364, "loss": 0.03774556517601013, "lr": 4.5287979258500456e-06, "epoch": 2.1482517482517482, "percentage": 71.61, "elapsed_time": "7:38:55", "remaining_time": "3:01:58"} +{"current_steps": 3842, "total_steps": 5364, "loss": 0.015268439427018166, "lr": 4.523351223297949e-06, "epoch": 2.1488111888111887, "percentage": 71.63, "elapsed_time": "7:39:03", "remaining_time": "3:01:51"} +{"current_steps": 3843, "total_steps": 5364, "loss": 0.12368149310350418, "lr": 4.517906840596482e-06, "epoch": 2.149370629370629, "percentage": 71.64, "elapsed_time": "7:39:10", "remaining_time": "3:01:44"} +{"current_steps": 3844, "total_steps": 5364, "loss": 0.004211365710943937, "lr": 4.512464780051831e-06, "epoch": 2.14993006993007, "percentage": 71.66, "elapsed_time": "7:39:17", "remaining_time": "3:01:37"} +{"current_steps": 3845, "total_steps": 5364, "loss": 0.04077071696519852, "lr": 4.507025043969188e-06, "epoch": 2.1504895104895105, "percentage": 71.68, "elapsed_time": "7:39:25", "remaining_time": "3:01:29"} +{"current_steps": 3846, "total_steps": 5364, "loss": 0.13250434398651123, "lr": 4.501587634652773e-06, "epoch": 2.151048951048951, "percentage": 71.7, "elapsed_time": "7:39:32", "remaining_time": "3:01:22"} +{"current_steps": 3847, "total_steps": 5364, "loss": 0.09368059039115906, "lr": 4.496152554405815e-06, "epoch": 2.1516083916083915, "percentage": 71.72, "elapsed_time": "7:39:39", "remaining_time": "3:01:15"} +{"current_steps": 3848, "total_steps": 5364, "loss": 0.05959971621632576, "lr": 4.490719805530554e-06, "epoch": 2.1521678321678324, "percentage": 71.74, "elapsed_time": "7:39:46", "remaining_time": "3:01:08"} +{"current_steps": 3849, "total_steps": 5364, "loss": 0.07026205211877823, "lr": 4.485289390328248e-06, "epoch": 2.152727272727273, "percentage": 71.76, "elapsed_time": "7:39:53", "remaining_time": "3:01:01"} +{"current_steps": 3850, "total_steps": 5364, "loss": 0.0821579173207283, "lr": 4.479861311099165e-06, "epoch": 2.1532867132867133, "percentage": 71.77, "elapsed_time": "7:40:00", "remaining_time": "3:00:53"} +{"current_steps": 3851, "total_steps": 5364, "loss": 0.24506758153438568, "lr": 4.474435570142577e-06, "epoch": 2.1538461538461537, "percentage": 71.79, "elapsed_time": "7:40:08", "remaining_time": "3:00:46"} +{"current_steps": 3852, "total_steps": 5364, "loss": 0.1311085820198059, "lr": 4.469012169756773e-06, "epoch": 2.154405594405594, "percentage": 71.81, "elapsed_time": "7:40:15", "remaining_time": "3:00:39"} +{"current_steps": 3853, "total_steps": 5364, "loss": 0.06196563318371773, "lr": 4.463591112239048e-06, "epoch": 2.154965034965035, "percentage": 71.83, "elapsed_time": "7:40:22", "remaining_time": "3:00:32"} +{"current_steps": 3854, "total_steps": 5364, "loss": 0.13947096467018127, "lr": 4.458172399885705e-06, "epoch": 2.1555244755244756, "percentage": 71.85, "elapsed_time": "7:40:29", "remaining_time": "3:00:25"} +{"current_steps": 3855, "total_steps": 5364, "loss": 0.014179615303874016, "lr": 4.452756034992057e-06, "epoch": 2.156083916083916, "percentage": 71.87, "elapsed_time": "7:40:36", "remaining_time": "3:00:17"} +{"current_steps": 3856, "total_steps": 5364, "loss": 0.10229015350341797, "lr": 4.447342019852412e-06, "epoch": 2.1566433566433565, "percentage": 71.89, "elapsed_time": "7:40:43", "remaining_time": "3:00:10"} +{"current_steps": 3857, "total_steps": 5364, "loss": 0.07068829238414764, "lr": 4.441930356760094e-06, "epoch": 2.1572027972027974, "percentage": 71.91, "elapsed_time": "7:40:50", "remaining_time": "3:00:03"} +{"current_steps": 3858, "total_steps": 5364, "loss": 0.13860081136226654, "lr": 4.436521048007424e-06, "epoch": 2.157762237762238, "percentage": 71.92, "elapsed_time": "7:40:57", "remaining_time": "2:59:56"} +{"current_steps": 3859, "total_steps": 5364, "loss": 0.08895541727542877, "lr": 4.431114095885731e-06, "epoch": 2.1583216783216783, "percentage": 71.94, "elapsed_time": "7:41:04", "remaining_time": "2:59:49"} +{"current_steps": 3860, "total_steps": 5364, "loss": 0.19275455176830292, "lr": 4.425709502685344e-06, "epoch": 2.1588811188811188, "percentage": 71.96, "elapsed_time": "7:41:11", "remaining_time": "2:59:41"} +{"current_steps": 3861, "total_steps": 5364, "loss": 0.1312236189842224, "lr": 4.420307270695586e-06, "epoch": 2.1594405594405592, "percentage": 71.98, "elapsed_time": "7:41:18", "remaining_time": "2:59:34"} +{"current_steps": 3862, "total_steps": 5364, "loss": 0.06387532502412796, "lr": 4.414907402204789e-06, "epoch": 2.16, "percentage": 72.0, "elapsed_time": "7:41:25", "remaining_time": "2:59:27"} +{"current_steps": 3863, "total_steps": 5364, "loss": 0.08187203854322433, "lr": 4.409509899500284e-06, "epoch": 2.1605594405594406, "percentage": 72.02, "elapsed_time": "7:41:32", "remaining_time": "2:59:20"} +{"current_steps": 3864, "total_steps": 5364, "loss": 0.06419216096401215, "lr": 4.404114764868384e-06, "epoch": 2.161118881118881, "percentage": 72.04, "elapsed_time": "7:41:39", "remaining_time": "2:59:12"} +{"current_steps": 3865, "total_steps": 5364, "loss": 0.06822659820318222, "lr": 4.398722000594427e-06, "epoch": 2.1616783216783215, "percentage": 72.05, "elapsed_time": "7:41:46", "remaining_time": "2:59:05"} +{"current_steps": 3866, "total_steps": 5364, "loss": 0.06794237345457077, "lr": 4.393331608962718e-06, "epoch": 2.1622377622377624, "percentage": 72.07, "elapsed_time": "7:41:53", "remaining_time": "2:58:58"} +{"current_steps": 3867, "total_steps": 5364, "loss": 0.13654986023902893, "lr": 4.387943592256577e-06, "epoch": 2.162797202797203, "percentage": 72.09, "elapsed_time": "7:42:00", "remaining_time": "2:58:51"} +{"current_steps": 3868, "total_steps": 5364, "loss": 0.029762111604213715, "lr": 4.382557952758313e-06, "epoch": 2.1633566433566433, "percentage": 72.11, "elapsed_time": "7:42:07", "remaining_time": "2:58:44"} +{"current_steps": 3869, "total_steps": 5364, "loss": 0.0476355217397213, "lr": 4.3771746927492145e-06, "epoch": 2.163916083916084, "percentage": 72.13, "elapsed_time": "7:42:14", "remaining_time": "2:58:36"} +{"current_steps": 3870, "total_steps": 5364, "loss": 0.07634837925434113, "lr": 4.371793814509589e-06, "epoch": 2.1644755244755243, "percentage": 72.15, "elapsed_time": "7:42:22", "remaining_time": "2:58:29"} +{"current_steps": 3871, "total_steps": 5364, "loss": 0.036532625555992126, "lr": 4.366415320318708e-06, "epoch": 2.165034965034965, "percentage": 72.17, "elapsed_time": "7:42:29", "remaining_time": "2:58:22"} +{"current_steps": 3872, "total_steps": 5364, "loss": 0.03828028589487076, "lr": 4.36103921245485e-06, "epoch": 2.1655944055944056, "percentage": 72.18, "elapsed_time": "7:42:37", "remaining_time": "2:58:15"} +{"current_steps": 3873, "total_steps": 5364, "loss": 0.044713299721479416, "lr": 4.355665493195279e-06, "epoch": 2.166153846153846, "percentage": 72.2, "elapsed_time": "7:42:44", "remaining_time": "2:58:08"} +{"current_steps": 3874, "total_steps": 5364, "loss": 0.06631101667881012, "lr": 4.350294164816237e-06, "epoch": 2.1667132867132866, "percentage": 72.22, "elapsed_time": "7:42:51", "remaining_time": "2:58:01"} +{"current_steps": 3875, "total_steps": 5364, "loss": 0.1317673623561859, "lr": 4.344925229592975e-06, "epoch": 2.1672727272727275, "percentage": 72.24, "elapsed_time": "7:42:58", "remaining_time": "2:57:53"} +{"current_steps": 3876, "total_steps": 5364, "loss": 0.06341883540153503, "lr": 4.339558689799708e-06, "epoch": 2.167832167832168, "percentage": 72.26, "elapsed_time": "7:43:04", "remaining_time": "2:57:46"} +{"current_steps": 3877, "total_steps": 5364, "loss": 0.10478334128856659, "lr": 4.334194547709648e-06, "epoch": 2.1683916083916084, "percentage": 72.28, "elapsed_time": "7:43:11", "remaining_time": "2:57:39"} +{"current_steps": 3878, "total_steps": 5364, "loss": 0.10518811643123627, "lr": 4.3288328055949935e-06, "epoch": 2.168951048951049, "percentage": 72.3, "elapsed_time": "7:43:18", "remaining_time": "2:57:32"} +{"current_steps": 3879, "total_steps": 5364, "loss": 0.11992601305246353, "lr": 4.323473465726911e-06, "epoch": 2.1695104895104893, "percentage": 72.32, "elapsed_time": "7:43:25", "remaining_time": "2:57:24"} +{"current_steps": 3880, "total_steps": 5364, "loss": 0.1269116997718811, "lr": 4.3181165303755725e-06, "epoch": 2.17006993006993, "percentage": 72.33, "elapsed_time": "7:43:32", "remaining_time": "2:57:17"} +{"current_steps": 3881, "total_steps": 5364, "loss": 0.11786731332540512, "lr": 4.312762001810111e-06, "epoch": 2.1706293706293707, "percentage": 72.35, "elapsed_time": "7:43:39", "remaining_time": "2:57:10"} +{"current_steps": 3882, "total_steps": 5364, "loss": 0.08864185214042664, "lr": 4.3074098822986525e-06, "epoch": 2.171188811188811, "percentage": 72.37, "elapsed_time": "7:43:47", "remaining_time": "2:57:03"} +{"current_steps": 3883, "total_steps": 5364, "loss": 0.056865446269512177, "lr": 4.302060174108299e-06, "epoch": 2.1717482517482516, "percentage": 72.39, "elapsed_time": "7:43:54", "remaining_time": "2:56:56"} +{"current_steps": 3884, "total_steps": 5364, "loss": 0.15149927139282227, "lr": 4.296712879505122e-06, "epoch": 2.1723076923076925, "percentage": 72.41, "elapsed_time": "7:44:01", "remaining_time": "2:56:48"} +{"current_steps": 3885, "total_steps": 5364, "loss": 0.05931822583079338, "lr": 4.2913680007541914e-06, "epoch": 2.172867132867133, "percentage": 72.43, "elapsed_time": "7:44:08", "remaining_time": "2:56:41"} +{"current_steps": 3886, "total_steps": 5364, "loss": 0.026843927800655365, "lr": 4.286025540119534e-06, "epoch": 2.1734265734265734, "percentage": 72.45, "elapsed_time": "7:44:15", "remaining_time": "2:56:34"} +{"current_steps": 3887, "total_steps": 5364, "loss": 0.06744428724050522, "lr": 4.28068549986416e-06, "epoch": 2.173986013986014, "percentage": 72.46, "elapsed_time": "7:44:22", "remaining_time": "2:56:27"} +{"current_steps": 3888, "total_steps": 5364, "loss": 0.1777801513671875, "lr": 4.2753478822500545e-06, "epoch": 2.174545454545455, "percentage": 72.48, "elapsed_time": "7:44:29", "remaining_time": "2:56:19"} +{"current_steps": 3889, "total_steps": 5364, "loss": 0.135552316904068, "lr": 4.270012689538182e-06, "epoch": 2.1751048951048952, "percentage": 72.5, "elapsed_time": "7:44:36", "remaining_time": "2:56:12"} +{"current_steps": 3890, "total_steps": 5364, "loss": 0.12418273091316223, "lr": 4.264679923988465e-06, "epoch": 2.1756643356643357, "percentage": 72.52, "elapsed_time": "7:44:43", "remaining_time": "2:56:05"} +{"current_steps": 3891, "total_steps": 5364, "loss": 0.028251193463802338, "lr": 4.25934958785981e-06, "epoch": 2.176223776223776, "percentage": 72.54, "elapsed_time": "7:44:51", "remaining_time": "2:55:58"} +{"current_steps": 3892, "total_steps": 5364, "loss": 0.15751181542873383, "lr": 4.254021683410092e-06, "epoch": 2.1767832167832166, "percentage": 72.56, "elapsed_time": "7:44:58", "remaining_time": "2:55:51"} +{"current_steps": 3893, "total_steps": 5364, "loss": 0.026731228455901146, "lr": 4.248696212896154e-06, "epoch": 2.1773426573426575, "percentage": 72.58, "elapsed_time": "7:45:05", "remaining_time": "2:55:44"} +{"current_steps": 3894, "total_steps": 5364, "loss": 0.008842997252941132, "lr": 4.243373178573814e-06, "epoch": 2.177902097902098, "percentage": 72.6, "elapsed_time": "7:45:13", "remaining_time": "2:55:37"} +{"current_steps": 3895, "total_steps": 5364, "loss": 0.07987696677446365, "lr": 4.2380525826978445e-06, "epoch": 2.1784615384615384, "percentage": 72.61, "elapsed_time": "7:45:20", "remaining_time": "2:55:30"} +{"current_steps": 3896, "total_steps": 5364, "loss": 0.13472937047481537, "lr": 4.232734427521999e-06, "epoch": 2.179020979020979, "percentage": 72.63, "elapsed_time": "7:45:27", "remaining_time": "2:55:23"} +{"current_steps": 3897, "total_steps": 5364, "loss": 0.05649025738239288, "lr": 4.2274187152989905e-06, "epoch": 2.17958041958042, "percentage": 72.65, "elapsed_time": "7:45:34", "remaining_time": "2:55:15"} +{"current_steps": 3898, "total_steps": 5364, "loss": 0.10967366397380829, "lr": 4.2221054482805e-06, "epoch": 2.1801398601398603, "percentage": 72.67, "elapsed_time": "7:45:41", "remaining_time": "2:55:08"} +{"current_steps": 3899, "total_steps": 5364, "loss": 0.13834531605243683, "lr": 4.216794628717175e-06, "epoch": 2.1806993006993007, "percentage": 72.69, "elapsed_time": "7:45:48", "remaining_time": "2:55:01"} +{"current_steps": 3900, "total_steps": 5364, "loss": 0.1697239875793457, "lr": 4.211486258858613e-06, "epoch": 2.181258741258741, "percentage": 72.71, "elapsed_time": "7:45:55", "remaining_time": "2:54:54"} +{"current_steps": 3901, "total_steps": 5364, "loss": 0.15156517922878265, "lr": 4.206180340953389e-06, "epoch": 2.1818181818181817, "percentage": 72.73, "elapsed_time": "7:46:02", "remaining_time": "2:54:46"} +{"current_steps": 3902, "total_steps": 5364, "loss": 0.10627919435501099, "lr": 4.200876877249034e-06, "epoch": 2.1823776223776226, "percentage": 72.74, "elapsed_time": "7:46:09", "remaining_time": "2:54:39"} +{"current_steps": 3903, "total_steps": 5364, "loss": 0.32856282591819763, "lr": 4.195575869992038e-06, "epoch": 2.182937062937063, "percentage": 72.76, "elapsed_time": "7:46:17", "remaining_time": "2:54:32"} +{"current_steps": 3904, "total_steps": 5364, "loss": 0.2708978056907654, "lr": 4.190277321427857e-06, "epoch": 2.1834965034965035, "percentage": 72.78, "elapsed_time": "7:46:24", "remaining_time": "2:54:25"} +{"current_steps": 3905, "total_steps": 5364, "loss": 0.06504981964826584, "lr": 4.184981233800888e-06, "epoch": 2.184055944055944, "percentage": 72.8, "elapsed_time": "7:46:31", "remaining_time": "2:54:18"} +{"current_steps": 3906, "total_steps": 5364, "loss": 0.028646428138017654, "lr": 4.179687609354512e-06, "epoch": 2.184615384615385, "percentage": 72.82, "elapsed_time": "7:46:38", "remaining_time": "2:54:11"} +{"current_steps": 3907, "total_steps": 5364, "loss": 0.12064141780138016, "lr": 4.174396450331043e-06, "epoch": 2.1851748251748253, "percentage": 72.84, "elapsed_time": "7:46:45", "remaining_time": "2:54:03"} +{"current_steps": 3908, "total_steps": 5364, "loss": 0.09545593708753586, "lr": 4.169107758971764e-06, "epoch": 2.1857342657342658, "percentage": 72.86, "elapsed_time": "7:46:53", "remaining_time": "2:53:56"} +{"current_steps": 3909, "total_steps": 5364, "loss": 0.023391593247652054, "lr": 4.163821537516911e-06, "epoch": 2.1862937062937062, "percentage": 72.87, "elapsed_time": "7:47:00", "remaining_time": "2:53:49"} +{"current_steps": 3910, "total_steps": 5364, "loss": 0.054874174296855927, "lr": 4.158537788205661e-06, "epoch": 2.1868531468531467, "percentage": 72.89, "elapsed_time": "7:47:07", "remaining_time": "2:53:42"} +{"current_steps": 3911, "total_steps": 5364, "loss": 0.06249412149190903, "lr": 4.153256513276169e-06, "epoch": 2.1874125874125876, "percentage": 72.91, "elapsed_time": "7:47:14", "remaining_time": "2:53:35"} +{"current_steps": 3912, "total_steps": 5364, "loss": 0.0644707977771759, "lr": 4.147977714965522e-06, "epoch": 2.187972027972028, "percentage": 72.93, "elapsed_time": "7:47:21", "remaining_time": "2:53:27"} +{"current_steps": 3913, "total_steps": 5364, "loss": 0.205059215426445, "lr": 4.142701395509754e-06, "epoch": 2.1885314685314685, "percentage": 72.95, "elapsed_time": "7:47:28", "remaining_time": "2:53:20"} +{"current_steps": 3914, "total_steps": 5364, "loss": 0.020121503621339798, "lr": 4.137427557143873e-06, "epoch": 2.189090909090909, "percentage": 72.97, "elapsed_time": "7:47:35", "remaining_time": "2:53:13"} +{"current_steps": 3915, "total_steps": 5364, "loss": 0.1439301073551178, "lr": 4.132156202101809e-06, "epoch": 2.18965034965035, "percentage": 72.99, "elapsed_time": "7:47:42", "remaining_time": "2:53:06"} +{"current_steps": 3916, "total_steps": 5364, "loss": 0.07918552309274673, "lr": 4.126887332616459e-06, "epoch": 2.1902097902097903, "percentage": 73.01, "elapsed_time": "7:47:49", "remaining_time": "2:52:59"} +{"current_steps": 3917, "total_steps": 5364, "loss": 0.10033179819583893, "lr": 4.121620950919662e-06, "epoch": 2.190769230769231, "percentage": 73.02, "elapsed_time": "7:47:56", "remaining_time": "2:52:51"} +{"current_steps": 3918, "total_steps": 5364, "loss": 0.11881642788648605, "lr": 4.116357059242193e-06, "epoch": 2.1913286713286713, "percentage": 73.04, "elapsed_time": "7:48:03", "remaining_time": "2:52:44"} +{"current_steps": 3919, "total_steps": 5364, "loss": 0.039310112595558167, "lr": 4.111095659813793e-06, "epoch": 2.1918881118881117, "percentage": 73.06, "elapsed_time": "7:48:10", "remaining_time": "2:52:37"} +{"current_steps": 3920, "total_steps": 5364, "loss": 0.08218531310558319, "lr": 4.105836754863126e-06, "epoch": 2.1924475524475526, "percentage": 73.08, "elapsed_time": "7:48:17", "remaining_time": "2:52:30"} +{"current_steps": 3921, "total_steps": 5364, "loss": 0.12492300570011139, "lr": 4.10058034661781e-06, "epoch": 2.193006993006993, "percentage": 73.1, "elapsed_time": "7:48:24", "remaining_time": "2:52:22"} +{"current_steps": 3922, "total_steps": 5364, "loss": 0.12985070049762726, "lr": 4.095326437304412e-06, "epoch": 2.1935664335664335, "percentage": 73.12, "elapsed_time": "7:48:31", "remaining_time": "2:52:15"} +{"current_steps": 3923, "total_steps": 5364, "loss": 0.058650773018598557, "lr": 4.090075029148417e-06, "epoch": 2.194125874125874, "percentage": 73.14, "elapsed_time": "7:48:37", "remaining_time": "2:52:08"} +{"current_steps": 3924, "total_steps": 5364, "loss": 0.05051948502659798, "lr": 4.084826124374284e-06, "epoch": 2.194685314685315, "percentage": 73.15, "elapsed_time": "7:48:45", "remaining_time": "2:52:01"} +{"current_steps": 3925, "total_steps": 5364, "loss": 0.05628602206707001, "lr": 4.07957972520538e-06, "epoch": 2.1952447552447554, "percentage": 73.17, "elapsed_time": "7:48:52", "remaining_time": "2:51:54"} +{"current_steps": 3926, "total_steps": 5364, "loss": 0.09606659412384033, "lr": 4.07433583386403e-06, "epoch": 2.195804195804196, "percentage": 73.19, "elapsed_time": "7:48:59", "remaining_time": "2:51:46"} +{"current_steps": 3927, "total_steps": 5364, "loss": 0.09657777845859528, "lr": 4.06909445257149e-06, "epoch": 2.1963636363636363, "percentage": 73.21, "elapsed_time": "7:49:06", "remaining_time": "2:51:39"} +{"current_steps": 3928, "total_steps": 5364, "loss": 0.0757531151175499, "lr": 4.0638555835479554e-06, "epoch": 2.1969230769230768, "percentage": 73.23, "elapsed_time": "7:49:13", "remaining_time": "2:51:32"} +{"current_steps": 3929, "total_steps": 5364, "loss": 0.08002664148807526, "lr": 4.058619229012558e-06, "epoch": 2.1974825174825177, "percentage": 73.25, "elapsed_time": "7:49:20", "remaining_time": "2:51:25"} +{"current_steps": 3930, "total_steps": 5364, "loss": 0.047871023416519165, "lr": 4.053385391183354e-06, "epoch": 2.198041958041958, "percentage": 73.27, "elapsed_time": "7:49:28", "remaining_time": "2:51:18"} +{"current_steps": 3931, "total_steps": 5364, "loss": 0.03471294417977333, "lr": 4.048154072277348e-06, "epoch": 2.1986013986013986, "percentage": 73.28, "elapsed_time": "7:49:35", "remaining_time": "2:51:10"} +{"current_steps": 3932, "total_steps": 5364, "loss": 0.037466805428266525, "lr": 4.042925274510471e-06, "epoch": 2.199160839160839, "percentage": 73.3, "elapsed_time": "7:49:42", "remaining_time": "2:51:03"} +{"current_steps": 3933, "total_steps": 5364, "loss": 0.1302366554737091, "lr": 4.037699000097585e-06, "epoch": 2.19972027972028, "percentage": 73.32, "elapsed_time": "7:49:49", "remaining_time": "2:50:56"} +{"current_steps": 3934, "total_steps": 5364, "loss": 0.042289648205041885, "lr": 4.032475251252489e-06, "epoch": 2.2002797202797204, "percentage": 73.34, "elapsed_time": "7:49:56", "remaining_time": "2:50:49"} +{"current_steps": 3935, "total_steps": 5364, "loss": 0.03058020770549774, "lr": 4.0272540301879006e-06, "epoch": 2.200839160839161, "percentage": 73.36, "elapsed_time": "7:50:03", "remaining_time": "2:50:42"} +{"current_steps": 3936, "total_steps": 5364, "loss": 0.16856679320335388, "lr": 4.022035339115479e-06, "epoch": 2.2013986013986013, "percentage": 73.38, "elapsed_time": "7:50:11", "remaining_time": "2:50:35"} +{"current_steps": 3937, "total_steps": 5364, "loss": 0.06180114671587944, "lr": 4.016819180245805e-06, "epoch": 2.201958041958042, "percentage": 73.4, "elapsed_time": "7:50:18", "remaining_time": "2:50:27"} +{"current_steps": 3938, "total_steps": 5364, "loss": 0.04430089890956879, "lr": 4.011605555788392e-06, "epoch": 2.2025174825174827, "percentage": 73.42, "elapsed_time": "7:50:25", "remaining_time": "2:50:20"} +{"current_steps": 3939, "total_steps": 5364, "loss": 0.15685822069644928, "lr": 4.00639446795167e-06, "epoch": 2.203076923076923, "percentage": 73.43, "elapsed_time": "7:50:32", "remaining_time": "2:50:13"} +{"current_steps": 3940, "total_steps": 5364, "loss": 0.0799047127366066, "lr": 4.001185918943003e-06, "epoch": 2.2036363636363636, "percentage": 73.45, "elapsed_time": "7:50:40", "remaining_time": "2:50:06"} +{"current_steps": 3941, "total_steps": 5364, "loss": 0.1177058070898056, "lr": 3.99597991096868e-06, "epoch": 2.204195804195804, "percentage": 73.47, "elapsed_time": "7:50:47", "remaining_time": "2:49:59"} +{"current_steps": 3942, "total_steps": 5364, "loss": 0.04188351333141327, "lr": 3.990776446233909e-06, "epoch": 2.204755244755245, "percentage": 73.49, "elapsed_time": "7:50:54", "remaining_time": "2:49:52"} +{"current_steps": 3943, "total_steps": 5364, "loss": 0.0670405924320221, "lr": 3.985575526942826e-06, "epoch": 2.2053146853146854, "percentage": 73.51, "elapsed_time": "7:51:02", "remaining_time": "2:49:45"} +{"current_steps": 3944, "total_steps": 5364, "loss": 0.07518081367015839, "lr": 3.980377155298477e-06, "epoch": 2.205874125874126, "percentage": 73.53, "elapsed_time": "7:51:09", "remaining_time": "2:49:38"} +{"current_steps": 3945, "total_steps": 5364, "loss": 0.05758193880319595, "lr": 3.97518133350285e-06, "epoch": 2.2064335664335664, "percentage": 73.55, "elapsed_time": "7:51:16", "remaining_time": "2:49:30"} +{"current_steps": 3946, "total_steps": 5364, "loss": 0.07331477105617523, "lr": 3.969988063756829e-06, "epoch": 2.206993006993007, "percentage": 73.56, "elapsed_time": "7:51:23", "remaining_time": "2:49:23"} +{"current_steps": 3947, "total_steps": 5364, "loss": 0.11568835377693176, "lr": 3.964797348260233e-06, "epoch": 2.2075524475524477, "percentage": 73.58, "elapsed_time": "7:51:30", "remaining_time": "2:49:16"} +{"current_steps": 3948, "total_steps": 5364, "loss": 0.07070145010948181, "lr": 3.959609189211798e-06, "epoch": 2.208111888111888, "percentage": 73.6, "elapsed_time": "7:51:37", "remaining_time": "2:49:09"} +{"current_steps": 3949, "total_steps": 5364, "loss": 0.07407890260219574, "lr": 3.954423588809163e-06, "epoch": 2.2086713286713286, "percentage": 73.62, "elapsed_time": "7:51:44", "remaining_time": "2:49:02"} +{"current_steps": 3950, "total_steps": 5364, "loss": 0.07815185189247131, "lr": 3.949240549248906e-06, "epoch": 2.209230769230769, "percentage": 73.64, "elapsed_time": "7:51:51", "remaining_time": "2:48:54"} +{"current_steps": 3951, "total_steps": 5364, "loss": 0.2690117359161377, "lr": 3.9440600727264985e-06, "epoch": 2.20979020979021, "percentage": 73.66, "elapsed_time": "7:51:59", "remaining_time": "2:48:47"} +{"current_steps": 3952, "total_steps": 5364, "loss": 0.0678568035364151, "lr": 3.938882161436339e-06, "epoch": 2.2103496503496505, "percentage": 73.68, "elapsed_time": "7:52:06", "remaining_time": "2:48:40"} +{"current_steps": 3953, "total_steps": 5364, "loss": 0.04408618062734604, "lr": 3.933706817571741e-06, "epoch": 2.210909090909091, "percentage": 73.7, "elapsed_time": "7:52:13", "remaining_time": "2:48:33"} +{"current_steps": 3954, "total_steps": 5364, "loss": 0.17852479219436646, "lr": 3.928534043324912e-06, "epoch": 2.2114685314685314, "percentage": 73.71, "elapsed_time": "7:52:20", "remaining_time": "2:48:26"} +{"current_steps": 3955, "total_steps": 5364, "loss": 0.07297449558973312, "lr": 3.923363840887002e-06, "epoch": 2.212027972027972, "percentage": 73.73, "elapsed_time": "7:52:27", "remaining_time": "2:48:19"} +{"current_steps": 3956, "total_steps": 5364, "loss": 0.10242324322462082, "lr": 3.91819621244804e-06, "epoch": 2.2125874125874128, "percentage": 73.75, "elapsed_time": "7:52:35", "remaining_time": "2:48:12"} +{"current_steps": 3957, "total_steps": 5364, "loss": 0.08546636253595352, "lr": 3.913031160196985e-06, "epoch": 2.213146853146853, "percentage": 73.77, "elapsed_time": "7:52:42", "remaining_time": "2:48:04"} +{"current_steps": 3958, "total_steps": 5364, "loss": 0.012173691764473915, "lr": 3.907868686321702e-06, "epoch": 2.2137062937062937, "percentage": 73.79, "elapsed_time": "7:52:49", "remaining_time": "2:47:57"} +{"current_steps": 3959, "total_steps": 5364, "loss": 0.04898948222398758, "lr": 3.9027087930089505e-06, "epoch": 2.214265734265734, "percentage": 73.81, "elapsed_time": "7:52:58", "remaining_time": "2:47:51"} +{"current_steps": 3960, "total_steps": 5364, "loss": 0.1368587762117386, "lr": 3.8975514824444184e-06, "epoch": 2.214825174825175, "percentage": 73.83, "elapsed_time": "7:53:04", "remaining_time": "2:47:43"} +{"current_steps": 3961, "total_steps": 5364, "loss": 0.05297800526022911, "lr": 3.892396756812679e-06, "epoch": 2.2153846153846155, "percentage": 73.84, "elapsed_time": "7:53:12", "remaining_time": "2:47:36"} +{"current_steps": 3962, "total_steps": 5364, "loss": 0.1151835173368454, "lr": 3.8872446182972235e-06, "epoch": 2.215944055944056, "percentage": 73.86, "elapsed_time": "7:53:19", "remaining_time": "2:47:29"} +{"current_steps": 3963, "total_steps": 5364, "loss": 0.0301944799721241, "lr": 3.882095069080446e-06, "epoch": 2.2165034965034964, "percentage": 73.88, "elapsed_time": "7:53:26", "remaining_time": "2:47:22"} +{"current_steps": 3964, "total_steps": 5364, "loss": 0.12432238459587097, "lr": 3.876948111343636e-06, "epoch": 2.217062937062937, "percentage": 73.9, "elapsed_time": "7:53:33", "remaining_time": "2:47:14"} +{"current_steps": 3965, "total_steps": 5364, "loss": 0.0917779952287674, "lr": 3.871803747266991e-06, "epoch": 2.217622377622378, "percentage": 73.92, "elapsed_time": "7:53:40", "remaining_time": "2:47:07"} +{"current_steps": 3966, "total_steps": 5364, "loss": 0.08906366676092148, "lr": 3.866661979029612e-06, "epoch": 2.2181818181818183, "percentage": 73.94, "elapsed_time": "7:53:47", "remaining_time": "2:47:00"} +{"current_steps": 3967, "total_steps": 5364, "loss": 0.08712785691022873, "lr": 3.861522808809497e-06, "epoch": 2.2187412587412587, "percentage": 73.96, "elapsed_time": "7:53:54", "remaining_time": "2:46:53"} +{"current_steps": 3968, "total_steps": 5364, "loss": 0.0677545890212059, "lr": 3.856386238783547e-06, "epoch": 2.219300699300699, "percentage": 73.97, "elapsed_time": "7:54:01", "remaining_time": "2:46:46"} +{"current_steps": 3969, "total_steps": 5364, "loss": 0.0493030920624733, "lr": 3.851252271127554e-06, "epoch": 2.21986013986014, "percentage": 73.99, "elapsed_time": "7:54:08", "remaining_time": "2:46:38"} +{"current_steps": 3970, "total_steps": 5364, "loss": 0.10384263098239899, "lr": 3.846120908016213e-06, "epoch": 2.2204195804195805, "percentage": 74.01, "elapsed_time": "7:54:15", "remaining_time": "2:46:31"} +{"current_steps": 3971, "total_steps": 5364, "loss": 0.09387702494859695, "lr": 3.840992151623117e-06, "epoch": 2.220979020979021, "percentage": 74.03, "elapsed_time": "7:54:22", "remaining_time": "2:46:24"} +{"current_steps": 3972, "total_steps": 5364, "loss": 0.09325749427080154, "lr": 3.835866004120753e-06, "epoch": 2.2215384615384615, "percentage": 74.05, "elapsed_time": "7:54:30", "remaining_time": "2:46:17"} +{"current_steps": 3973, "total_steps": 5364, "loss": 0.0520363375544548, "lr": 3.8307424676805076e-06, "epoch": 2.222097902097902, "percentage": 74.07, "elapsed_time": "7:54:37", "remaining_time": "2:46:10"} +{"current_steps": 3974, "total_steps": 5364, "loss": 0.026735279709100723, "lr": 3.8256215444726466e-06, "epoch": 2.222657342657343, "percentage": 74.09, "elapsed_time": "7:54:44", "remaining_time": "2:46:02"} +{"current_steps": 3975, "total_steps": 5364, "loss": 0.04936660826206207, "lr": 3.820503236666345e-06, "epoch": 2.2232167832167833, "percentage": 74.11, "elapsed_time": "7:54:51", "remaining_time": "2:45:55"} +{"current_steps": 3976, "total_steps": 5364, "loss": 0.05895145237445831, "lr": 3.815387546429663e-06, "epoch": 2.2237762237762237, "percentage": 74.12, "elapsed_time": "7:54:58", "remaining_time": "2:45:48"} +{"current_steps": 3977, "total_steps": 5364, "loss": 0.01264667883515358, "lr": 3.810274475929554e-06, "epoch": 2.224335664335664, "percentage": 74.14, "elapsed_time": "7:55:05", "remaining_time": "2:45:41"} +{"current_steps": 3978, "total_steps": 5364, "loss": 0.055085889995098114, "lr": 3.8051640273318646e-06, "epoch": 2.224895104895105, "percentage": 74.16, "elapsed_time": "7:55:12", "remaining_time": "2:45:34"} +{"current_steps": 3979, "total_steps": 5364, "loss": 0.12590879201889038, "lr": 3.8000562028013177e-06, "epoch": 2.2254545454545456, "percentage": 74.18, "elapsed_time": "7:55:19", "remaining_time": "2:45:27"} +{"current_steps": 3980, "total_steps": 5364, "loss": 0.13788959383964539, "lr": 3.794951004501539e-06, "epoch": 2.226013986013986, "percentage": 74.2, "elapsed_time": "7:55:26", "remaining_time": "2:45:19"} +{"current_steps": 3981, "total_steps": 5364, "loss": 0.05030418187379837, "lr": 3.789848434595036e-06, "epoch": 2.2265734265734265, "percentage": 74.22, "elapsed_time": "7:55:33", "remaining_time": "2:45:12"} +{"current_steps": 3982, "total_steps": 5364, "loss": 0.06063270568847656, "lr": 3.7847484952432033e-06, "epoch": 2.227132867132867, "percentage": 74.24, "elapsed_time": "7:55:40", "remaining_time": "2:45:05"} +{"current_steps": 3983, "total_steps": 5364, "loss": 0.0255948007106781, "lr": 3.7796511886063227e-06, "epoch": 2.227692307692308, "percentage": 74.25, "elapsed_time": "7:55:48", "remaining_time": "2:44:58"} +{"current_steps": 3984, "total_steps": 5364, "loss": 0.09658539295196533, "lr": 3.7745565168435614e-06, "epoch": 2.2282517482517483, "percentage": 74.27, "elapsed_time": "7:55:55", "remaining_time": "2:44:51"} +{"current_steps": 3985, "total_steps": 5364, "loss": 0.04903118312358856, "lr": 3.7694644821129635e-06, "epoch": 2.228811188811189, "percentage": 74.29, "elapsed_time": "7:56:02", "remaining_time": "2:44:44"} +{"current_steps": 3986, "total_steps": 5364, "loss": 0.03705323487520218, "lr": 3.764375086571462e-06, "epoch": 2.2293706293706292, "percentage": 74.31, "elapsed_time": "7:56:10", "remaining_time": "2:44:37"} +{"current_steps": 3987, "total_steps": 5364, "loss": 0.08233587443828583, "lr": 3.7592883323748762e-06, "epoch": 2.22993006993007, "percentage": 74.33, "elapsed_time": "7:56:17", "remaining_time": "2:44:29"} +{"current_steps": 3988, "total_steps": 5364, "loss": 0.14936336874961853, "lr": 3.754204221677892e-06, "epoch": 2.2304895104895106, "percentage": 74.35, "elapsed_time": "7:56:24", "remaining_time": "2:44:22"} +{"current_steps": 3989, "total_steps": 5364, "loss": 0.08057862520217896, "lr": 3.749122756634097e-06, "epoch": 2.231048951048951, "percentage": 74.37, "elapsed_time": "7:56:31", "remaining_time": "2:44:15"} +{"current_steps": 3990, "total_steps": 5364, "loss": 0.17346906661987305, "lr": 3.744043939395937e-06, "epoch": 2.2316083916083915, "percentage": 74.38, "elapsed_time": "7:56:38", "remaining_time": "2:44:08"} +{"current_steps": 3991, "total_steps": 5364, "loss": 0.16614481806755066, "lr": 3.738967772114749e-06, "epoch": 2.232167832167832, "percentage": 74.4, "elapsed_time": "7:56:45", "remaining_time": "2:44:00"} +{"current_steps": 3992, "total_steps": 5364, "loss": 0.05408097058534622, "lr": 3.7338942569407456e-06, "epoch": 2.232727272727273, "percentage": 74.42, "elapsed_time": "7:56:52", "remaining_time": "2:43:53"} +{"current_steps": 3993, "total_steps": 5364, "loss": 0.0801866427063942, "lr": 3.7288233960230058e-06, "epoch": 2.2332867132867134, "percentage": 74.44, "elapsed_time": "7:57:00", "remaining_time": "2:43:46"} +{"current_steps": 3994, "total_steps": 5364, "loss": 0.03332550823688507, "lr": 3.723755191509506e-06, "epoch": 2.233846153846154, "percentage": 74.46, "elapsed_time": "7:57:06", "remaining_time": "2:43:39"} +{"current_steps": 3995, "total_steps": 5364, "loss": 0.10848326981067657, "lr": 3.7186896455470745e-06, "epoch": 2.2344055944055943, "percentage": 74.48, "elapsed_time": "7:57:13", "remaining_time": "2:43:32"} +{"current_steps": 3996, "total_steps": 5364, "loss": 0.0342969186604023, "lr": 3.7136267602814246e-06, "epoch": 2.234965034965035, "percentage": 74.5, "elapsed_time": "7:57:21", "remaining_time": "2:43:25"} +{"current_steps": 3997, "total_steps": 5364, "loss": 0.17291346192359924, "lr": 3.7085665378571465e-06, "epoch": 2.2355244755244756, "percentage": 74.52, "elapsed_time": "7:57:29", "remaining_time": "2:43:18"} +{"current_steps": 3998, "total_steps": 5364, "loss": 0.082366943359375, "lr": 3.703508980417686e-06, "epoch": 2.236083916083916, "percentage": 74.53, "elapsed_time": "7:57:36", "remaining_time": "2:43:11"} +{"current_steps": 3999, "total_steps": 5364, "loss": 0.07363560795783997, "lr": 3.6984540901053844e-06, "epoch": 2.2366433566433566, "percentage": 74.55, "elapsed_time": "7:57:43", "remaining_time": "2:43:03"} +{"current_steps": 4000, "total_steps": 5364, "loss": 0.057901885360479355, "lr": 3.69340186906143e-06, "epoch": 2.237202797202797, "percentage": 74.57, "elapsed_time": "7:57:51", "remaining_time": "2:42:56"} +{"current_steps": 4001, "total_steps": 5364, "loss": 0.07826337963342667, "lr": 3.688352319425894e-06, "epoch": 2.237762237762238, "percentage": 74.59, "elapsed_time": "7:58:26", "remaining_time": "2:42:59"} +{"current_steps": 4002, "total_steps": 5364, "loss": 0.09077227115631104, "lr": 3.6833054433377145e-06, "epoch": 2.2383216783216784, "percentage": 74.61, "elapsed_time": "7:58:33", "remaining_time": "2:42:51"} +{"current_steps": 4003, "total_steps": 5364, "loss": 0.11784470081329346, "lr": 3.6782612429346863e-06, "epoch": 2.238881118881119, "percentage": 74.63, "elapsed_time": "7:58:39", "remaining_time": "2:42:44"} +{"current_steps": 4004, "total_steps": 5364, "loss": 0.09747405350208282, "lr": 3.6732197203534937e-06, "epoch": 2.2394405594405593, "percentage": 74.65, "elapsed_time": "7:58:46", "remaining_time": "2:42:37"} +{"current_steps": 4005, "total_steps": 5364, "loss": 0.10384588688611984, "lr": 3.668180877729659e-06, "epoch": 2.24, "percentage": 74.66, "elapsed_time": "7:58:53", "remaining_time": "2:42:30"} +{"current_steps": 4006, "total_steps": 5364, "loss": 0.09412307292222977, "lr": 3.66314471719759e-06, "epoch": 2.2405594405594407, "percentage": 74.68, "elapsed_time": "7:59:01", "remaining_time": "2:42:23"} +{"current_steps": 4007, "total_steps": 5364, "loss": 0.20190943777561188, "lr": 3.6581112408905517e-06, "epoch": 2.241118881118881, "percentage": 74.7, "elapsed_time": "7:59:08", "remaining_time": "2:42:15"} +{"current_steps": 4008, "total_steps": 5364, "loss": 0.056947026401758194, "lr": 3.6530804509406625e-06, "epoch": 2.2416783216783216, "percentage": 74.72, "elapsed_time": "7:59:15", "remaining_time": "2:42:08"} +{"current_steps": 4009, "total_steps": 5364, "loss": 0.049666061997413635, "lr": 3.6480523494789257e-06, "epoch": 2.242237762237762, "percentage": 74.74, "elapsed_time": "7:59:22", "remaining_time": "2:42:01"} +{"current_steps": 4010, "total_steps": 5364, "loss": 0.13487812876701355, "lr": 3.6430269386351813e-06, "epoch": 2.242797202797203, "percentage": 74.76, "elapsed_time": "7:59:29", "remaining_time": "2:41:54"} +{"current_steps": 4011, "total_steps": 5364, "loss": 0.1209891140460968, "lr": 3.638004220538144e-06, "epoch": 2.2433566433566434, "percentage": 74.78, "elapsed_time": "7:59:37", "remaining_time": "2:41:47"} +{"current_steps": 4012, "total_steps": 5364, "loss": 0.037573978304862976, "lr": 3.6329841973153867e-06, "epoch": 2.243916083916084, "percentage": 74.79, "elapsed_time": "7:59:44", "remaining_time": "2:41:40"} +{"current_steps": 4013, "total_steps": 5364, "loss": 0.0838402584195137, "lr": 3.6279668710933337e-06, "epoch": 2.2444755244755243, "percentage": 74.81, "elapsed_time": "7:59:52", "remaining_time": "2:41:33"} +{"current_steps": 4014, "total_steps": 5364, "loss": 0.08591905981302261, "lr": 3.6229522439972733e-06, "epoch": 2.2450349650349652, "percentage": 74.83, "elapsed_time": "7:59:59", "remaining_time": "2:41:25"} +{"current_steps": 4015, "total_steps": 5364, "loss": 0.0782245621085167, "lr": 3.6179403181513483e-06, "epoch": 2.2455944055944057, "percentage": 74.85, "elapsed_time": "8:00:06", "remaining_time": "2:41:18"} +{"current_steps": 4016, "total_steps": 5364, "loss": 0.17355681955814362, "lr": 3.612931095678558e-06, "epoch": 2.246153846153846, "percentage": 74.87, "elapsed_time": "8:00:13", "remaining_time": "2:41:11"} +{"current_steps": 4017, "total_steps": 5364, "loss": 0.06991080939769745, "lr": 3.6079245787007557e-06, "epoch": 2.2467132867132866, "percentage": 74.89, "elapsed_time": "8:00:20", "remaining_time": "2:41:04"} +{"current_steps": 4018, "total_steps": 5364, "loss": 0.05921034514904022, "lr": 3.6029207693386536e-06, "epoch": 2.247272727272727, "percentage": 74.91, "elapsed_time": "8:00:27", "remaining_time": "2:40:56"} +{"current_steps": 4019, "total_steps": 5364, "loss": 0.11045928299427032, "lr": 3.597919669711806e-06, "epoch": 2.247832167832168, "percentage": 74.93, "elapsed_time": "8:00:34", "remaining_time": "2:40:49"} +{"current_steps": 4020, "total_steps": 5364, "loss": 0.012293178588151932, "lr": 3.5929212819386285e-06, "epoch": 2.2483916083916085, "percentage": 74.94, "elapsed_time": "8:00:41", "remaining_time": "2:40:42"} +{"current_steps": 4021, "total_steps": 5364, "loss": 0.03238307312130928, "lr": 3.5879256081363843e-06, "epoch": 2.248951048951049, "percentage": 74.96, "elapsed_time": "8:00:49", "remaining_time": "2:40:35"} +{"current_steps": 4022, "total_steps": 5364, "loss": 0.051424071192741394, "lr": 3.58293265042119e-06, "epoch": 2.2495104895104894, "percentage": 74.98, "elapsed_time": "8:00:56", "remaining_time": "2:40:28"} +{"current_steps": 4023, "total_steps": 5364, "loss": 0.08836375176906586, "lr": 3.577942410908013e-06, "epoch": 2.2500699300699303, "percentage": 75.0, "elapsed_time": "8:01:03", "remaining_time": "2:40:21"} +{"current_steps": 4024, "total_steps": 5364, "loss": 0.06211835891008377, "lr": 3.5729548917106584e-06, "epoch": 2.2506293706293707, "percentage": 75.02, "elapsed_time": "8:01:10", "remaining_time": "2:40:13"} +{"current_steps": 4025, "total_steps": 5364, "loss": 0.015144189819693565, "lr": 3.5679700949417894e-06, "epoch": 2.251188811188811, "percentage": 75.04, "elapsed_time": "8:01:17", "remaining_time": "2:40:06"} +{"current_steps": 4026, "total_steps": 5364, "loss": 0.08175687491893768, "lr": 3.5629880227129154e-06, "epoch": 2.2517482517482517, "percentage": 75.06, "elapsed_time": "8:01:24", "remaining_time": "2:39:59"} +{"current_steps": 4027, "total_steps": 5364, "loss": 0.13278828561306, "lr": 3.558008677134389e-06, "epoch": 2.252307692307692, "percentage": 75.07, "elapsed_time": "8:01:31", "remaining_time": "2:39:52"} +{"current_steps": 4028, "total_steps": 5364, "loss": 0.04883341118693352, "lr": 3.5530320603154113e-06, "epoch": 2.252867132867133, "percentage": 75.09, "elapsed_time": "8:01:39", "remaining_time": "2:39:45"} +{"current_steps": 4029, "total_steps": 5364, "loss": 0.10329955071210861, "lr": 3.5480581743640176e-06, "epoch": 2.2534265734265735, "percentage": 75.11, "elapsed_time": "8:01:46", "remaining_time": "2:39:38"} +{"current_steps": 4030, "total_steps": 5364, "loss": 0.09205092489719391, "lr": 3.543087021387098e-06, "epoch": 2.253986013986014, "percentage": 75.13, "elapsed_time": "8:01:53", "remaining_time": "2:39:30"} +{"current_steps": 4031, "total_steps": 5364, "loss": 0.013555154204368591, "lr": 3.538118603490379e-06, "epoch": 2.2545454545454544, "percentage": 75.15, "elapsed_time": "8:02:00", "remaining_time": "2:39:23"} +{"current_steps": 4032, "total_steps": 5364, "loss": 0.09695370495319366, "lr": 3.5331529227784313e-06, "epoch": 2.2551048951048953, "percentage": 75.17, "elapsed_time": "8:02:07", "remaining_time": "2:39:16"} +{"current_steps": 4033, "total_steps": 5364, "loss": 0.03444574400782585, "lr": 3.528189981354667e-06, "epoch": 2.2556643356643358, "percentage": 75.19, "elapsed_time": "8:02:14", "remaining_time": "2:39:09"} +{"current_steps": 4034, "total_steps": 5364, "loss": 0.15827062726020813, "lr": 3.523229781321331e-06, "epoch": 2.2562237762237762, "percentage": 75.21, "elapsed_time": "8:02:21", "remaining_time": "2:39:01"} +{"current_steps": 4035, "total_steps": 5364, "loss": 0.06910944730043411, "lr": 3.5182723247795124e-06, "epoch": 2.2567832167832167, "percentage": 75.22, "elapsed_time": "8:02:28", "remaining_time": "2:38:54"} +{"current_steps": 4036, "total_steps": 5364, "loss": 0.12077774107456207, "lr": 3.5133176138291438e-06, "epoch": 2.257342657342657, "percentage": 75.24, "elapsed_time": "8:02:35", "remaining_time": "2:38:47"} +{"current_steps": 4037, "total_steps": 5364, "loss": 0.06353722512722015, "lr": 3.5083656505689756e-06, "epoch": 2.257902097902098, "percentage": 75.26, "elapsed_time": "8:02:43", "remaining_time": "2:38:40"} +{"current_steps": 4038, "total_steps": 5364, "loss": 0.16074544191360474, "lr": 3.503416437096622e-06, "epoch": 2.2584615384615385, "percentage": 75.28, "elapsed_time": "8:02:50", "remaining_time": "2:38:33"} +{"current_steps": 4039, "total_steps": 5364, "loss": 0.1852363497018814, "lr": 3.498469975508506e-06, "epoch": 2.259020979020979, "percentage": 75.3, "elapsed_time": "8:02:57", "remaining_time": "2:38:26"} +{"current_steps": 4040, "total_steps": 5364, "loss": 0.07301154732704163, "lr": 3.493526267899903e-06, "epoch": 2.2595804195804194, "percentage": 75.32, "elapsed_time": "8:03:04", "remaining_time": "2:38:18"} +{"current_steps": 4041, "total_steps": 5364, "loss": 0.06533936411142349, "lr": 3.4885853163649162e-06, "epoch": 2.2601398601398603, "percentage": 75.34, "elapsed_time": "8:03:11", "remaining_time": "2:38:11"} +{"current_steps": 4042, "total_steps": 5364, "loss": 0.0321870818734169, "lr": 3.483647122996472e-06, "epoch": 2.260699300699301, "percentage": 75.35, "elapsed_time": "8:03:19", "remaining_time": "2:38:04"} +{"current_steps": 4043, "total_steps": 5364, "loss": 0.08697166293859482, "lr": 3.478711689886348e-06, "epoch": 2.2612587412587413, "percentage": 75.37, "elapsed_time": "8:03:26", "remaining_time": "2:37:57"} +{"current_steps": 4044, "total_steps": 5364, "loss": 0.05689668655395508, "lr": 3.473779019125134e-06, "epoch": 2.2618181818181817, "percentage": 75.39, "elapsed_time": "8:03:34", "remaining_time": "2:37:50"} +{"current_steps": 4045, "total_steps": 5364, "loss": 0.1464952826499939, "lr": 3.4688491128022605e-06, "epoch": 2.262377622377622, "percentage": 75.41, "elapsed_time": "8:03:41", "remaining_time": "2:37:43"} +{"current_steps": 4046, "total_steps": 5364, "loss": 0.029885651543736458, "lr": 3.463921973005986e-06, "epoch": 2.262937062937063, "percentage": 75.43, "elapsed_time": "8:03:48", "remaining_time": "2:37:36"} +{"current_steps": 4047, "total_steps": 5364, "loss": 0.03507468104362488, "lr": 3.458997601823385e-06, "epoch": 2.2634965034965036, "percentage": 75.45, "elapsed_time": "8:03:55", "remaining_time": "2:37:28"} +{"current_steps": 4048, "total_steps": 5364, "loss": 0.011632539331912994, "lr": 3.454076001340383e-06, "epoch": 2.264055944055944, "percentage": 75.47, "elapsed_time": "8:04:02", "remaining_time": "2:37:21"} +{"current_steps": 4049, "total_steps": 5364, "loss": 0.07180996984243393, "lr": 3.4491571736417074e-06, "epoch": 2.2646153846153845, "percentage": 75.48, "elapsed_time": "8:04:09", "remaining_time": "2:37:14"} +{"current_steps": 4050, "total_steps": 5364, "loss": 0.11113995313644409, "lr": 3.444241120810926e-06, "epoch": 2.2651748251748254, "percentage": 75.5, "elapsed_time": "8:04:17", "remaining_time": "2:37:07"} +{"current_steps": 4051, "total_steps": 5364, "loss": 0.0048193675465881824, "lr": 3.43932784493043e-06, "epoch": 2.265734265734266, "percentage": 75.52, "elapsed_time": "8:04:24", "remaining_time": "2:37:00"} +{"current_steps": 4052, "total_steps": 5364, "loss": 0.0892617478966713, "lr": 3.43441734808142e-06, "epoch": 2.2662937062937063, "percentage": 75.54, "elapsed_time": "8:04:31", "remaining_time": "2:36:52"} +{"current_steps": 4053, "total_steps": 5364, "loss": 0.1479639708995819, "lr": 3.429509632343946e-06, "epoch": 2.2668531468531468, "percentage": 75.56, "elapsed_time": "8:04:38", "remaining_time": "2:36:45"} +{"current_steps": 4054, "total_steps": 5364, "loss": 0.09672702103853226, "lr": 3.424604699796853e-06, "epoch": 2.267412587412587, "percentage": 75.58, "elapsed_time": "8:04:45", "remaining_time": "2:36:38"} +{"current_steps": 4055, "total_steps": 5364, "loss": 0.04358110576868057, "lr": 3.419702552517823e-06, "epoch": 2.267972027972028, "percentage": 75.6, "elapsed_time": "8:04:52", "remaining_time": "2:36:31"} +{"current_steps": 4056, "total_steps": 5364, "loss": 0.08175045251846313, "lr": 3.4148031925833525e-06, "epoch": 2.2685314685314686, "percentage": 75.62, "elapsed_time": "8:04:59", "remaining_time": "2:36:24"} +{"current_steps": 4057, "total_steps": 5364, "loss": 0.14600539207458496, "lr": 3.4099066220687603e-06, "epoch": 2.269090909090909, "percentage": 75.63, "elapsed_time": "8:05:06", "remaining_time": "2:36:16"} +{"current_steps": 4058, "total_steps": 5364, "loss": 0.04659982770681381, "lr": 3.4050128430481844e-06, "epoch": 2.2696503496503495, "percentage": 75.65, "elapsed_time": "8:05:13", "remaining_time": "2:36:09"} +{"current_steps": 4059, "total_steps": 5364, "loss": 0.15947610139846802, "lr": 3.4001218575945716e-06, "epoch": 2.2702097902097904, "percentage": 75.67, "elapsed_time": "8:05:20", "remaining_time": "2:36:02"} +{"current_steps": 4060, "total_steps": 5364, "loss": 0.0396287702023983, "lr": 3.395233667779695e-06, "epoch": 2.270769230769231, "percentage": 75.69, "elapsed_time": "8:05:27", "remaining_time": "2:35:55"} +{"current_steps": 4061, "total_steps": 5364, "loss": 0.07789931446313858, "lr": 3.3903482756741402e-06, "epoch": 2.2713286713286713, "percentage": 75.71, "elapsed_time": "8:05:34", "remaining_time": "2:35:48"} +{"current_steps": 4062, "total_steps": 5364, "loss": 0.08802768588066101, "lr": 3.38546568334731e-06, "epoch": 2.271888111888112, "percentage": 75.73, "elapsed_time": "8:05:41", "remaining_time": "2:35:40"} +{"current_steps": 4063, "total_steps": 5364, "loss": 0.06346607953310013, "lr": 3.380585892867414e-06, "epoch": 2.2724475524475523, "percentage": 75.75, "elapsed_time": "8:05:49", "remaining_time": "2:35:33"} +{"current_steps": 4064, "total_steps": 5364, "loss": 0.1546359658241272, "lr": 3.3757089063014836e-06, "epoch": 2.273006993006993, "percentage": 75.76, "elapsed_time": "8:05:56", "remaining_time": "2:35:26"} +{"current_steps": 4065, "total_steps": 5364, "loss": 0.08639201521873474, "lr": 3.3708347257153562e-06, "epoch": 2.2735664335664336, "percentage": 75.78, "elapsed_time": "8:06:03", "remaining_time": "2:35:19"} +{"current_steps": 4066, "total_steps": 5364, "loss": 0.05767097324132919, "lr": 3.3659633531736858e-06, "epoch": 2.274125874125874, "percentage": 75.8, "elapsed_time": "8:06:10", "remaining_time": "2:35:12"} +{"current_steps": 4067, "total_steps": 5364, "loss": 0.10328216850757599, "lr": 3.3610947907399373e-06, "epoch": 2.2746853146853145, "percentage": 75.82, "elapsed_time": "8:06:17", "remaining_time": "2:35:05"} +{"current_steps": 4068, "total_steps": 5364, "loss": 0.06288590282201767, "lr": 3.356229040476375e-06, "epoch": 2.2752447552447554, "percentage": 75.84, "elapsed_time": "8:06:24", "remaining_time": "2:34:57"} +{"current_steps": 4069, "total_steps": 5364, "loss": 0.051631905138492584, "lr": 3.3513661044440825e-06, "epoch": 2.275804195804196, "percentage": 75.86, "elapsed_time": "8:06:32", "remaining_time": "2:34:50"} +{"current_steps": 4070, "total_steps": 5364, "loss": 0.12575900554656982, "lr": 3.346505984702949e-06, "epoch": 2.2763636363636364, "percentage": 75.88, "elapsed_time": "8:06:39", "remaining_time": "2:34:43"} +{"current_steps": 4071, "total_steps": 5364, "loss": 0.062405407428741455, "lr": 3.34164868331167e-06, "epoch": 2.276923076923077, "percentage": 75.89, "elapsed_time": "8:06:46", "remaining_time": "2:34:36"} +{"current_steps": 4072, "total_steps": 5364, "loss": 0.1576712429523468, "lr": 3.33679420232775e-06, "epoch": 2.2774825174825173, "percentage": 75.91, "elapsed_time": "8:06:54", "remaining_time": "2:34:29"} +{"current_steps": 4073, "total_steps": 5364, "loss": 0.15222272276878357, "lr": 3.331942543807488e-06, "epoch": 2.278041958041958, "percentage": 75.93, "elapsed_time": "8:07:00", "remaining_time": "2:34:22"} +{"current_steps": 4074, "total_steps": 5364, "loss": 0.026921415701508522, "lr": 3.327093709806002e-06, "epoch": 2.2786013986013987, "percentage": 75.95, "elapsed_time": "8:07:07", "remaining_time": "2:34:14"} +{"current_steps": 4075, "total_steps": 5364, "loss": 0.03489012271165848, "lr": 3.322247702377204e-06, "epoch": 2.279160839160839, "percentage": 75.97, "elapsed_time": "8:07:14", "remaining_time": "2:34:07"} +{"current_steps": 4076, "total_steps": 5364, "loss": 0.03215426206588745, "lr": 3.3174045235738117e-06, "epoch": 2.2797202797202796, "percentage": 75.99, "elapsed_time": "8:07:22", "remaining_time": "2:34:00"} +{"current_steps": 4077, "total_steps": 5364, "loss": 0.12308354675769806, "lr": 3.3125641754473503e-06, "epoch": 2.2802797202797205, "percentage": 76.01, "elapsed_time": "8:07:29", "remaining_time": "2:33:53"} +{"current_steps": 4078, "total_steps": 5364, "loss": 0.07809754461050034, "lr": 3.3077266600481274e-06, "epoch": 2.280839160839161, "percentage": 76.03, "elapsed_time": "8:07:36", "remaining_time": "2:33:46"} +{"current_steps": 4079, "total_steps": 5364, "loss": 0.06743837147951126, "lr": 3.302891979425278e-06, "epoch": 2.2813986013986014, "percentage": 76.04, "elapsed_time": "8:07:43", "remaining_time": "2:33:38"} +{"current_steps": 4080, "total_steps": 5364, "loss": 0.05035008117556572, "lr": 3.2980601356267126e-06, "epoch": 2.281958041958042, "percentage": 76.06, "elapsed_time": "8:07:50", "remaining_time": "2:33:31"} +{"current_steps": 4081, "total_steps": 5364, "loss": 0.17568081617355347, "lr": 3.2932311306991504e-06, "epoch": 2.2825174825174823, "percentage": 76.08, "elapsed_time": "8:07:57", "remaining_time": "2:33:24"} +{"current_steps": 4082, "total_steps": 5364, "loss": 0.15135107934474945, "lr": 3.2884049666881136e-06, "epoch": 2.2830769230769232, "percentage": 76.1, "elapsed_time": "8:08:04", "remaining_time": "2:33:17"} +{"current_steps": 4083, "total_steps": 5364, "loss": 0.08967706561088562, "lr": 3.283581645637901e-06, "epoch": 2.2836363636363637, "percentage": 76.12, "elapsed_time": "8:08:11", "remaining_time": "2:33:09"} +{"current_steps": 4084, "total_steps": 5364, "loss": 0.04252856224775314, "lr": 3.2787611695916355e-06, "epoch": 2.284195804195804, "percentage": 76.14, "elapsed_time": "8:08:18", "remaining_time": "2:33:02"} +{"current_steps": 4085, "total_steps": 5364, "loss": 0.05316941440105438, "lr": 3.273943540591212e-06, "epoch": 2.2847552447552446, "percentage": 76.16, "elapsed_time": "8:08:25", "remaining_time": "2:32:55"} +{"current_steps": 4086, "total_steps": 5364, "loss": 0.04093504324555397, "lr": 3.2691287606773226e-06, "epoch": 2.2853146853146855, "percentage": 76.17, "elapsed_time": "8:08:32", "remaining_time": "2:32:48"} +{"current_steps": 4087, "total_steps": 5364, "loss": 0.18728643655776978, "lr": 3.2643168318894693e-06, "epoch": 2.285874125874126, "percentage": 76.19, "elapsed_time": "8:08:39", "remaining_time": "2:32:40"} +{"current_steps": 4088, "total_steps": 5364, "loss": 0.1192634254693985, "lr": 3.2595077562659248e-06, "epoch": 2.2864335664335664, "percentage": 76.21, "elapsed_time": "8:08:46", "remaining_time": "2:32:33"} +{"current_steps": 4089, "total_steps": 5364, "loss": 0.03759636729955673, "lr": 3.254701535843766e-06, "epoch": 2.286993006993007, "percentage": 76.23, "elapsed_time": "8:08:53", "remaining_time": "2:32:26"} +{"current_steps": 4090, "total_steps": 5364, "loss": 0.03267242759466171, "lr": 3.249898172658861e-06, "epoch": 2.2875524475524474, "percentage": 76.25, "elapsed_time": "8:09:01", "remaining_time": "2:32:19"} +{"current_steps": 4091, "total_steps": 5364, "loss": 0.11120981723070145, "lr": 3.245097668745851e-06, "epoch": 2.2881118881118883, "percentage": 76.27, "elapsed_time": "8:09:08", "remaining_time": "2:32:12"} +{"current_steps": 4092, "total_steps": 5364, "loss": 0.08536656200885773, "lr": 3.240300026138197e-06, "epoch": 2.2886713286713287, "percentage": 76.29, "elapsed_time": "8:09:15", "remaining_time": "2:32:05"} +{"current_steps": 4093, "total_steps": 5364, "loss": 0.13874593377113342, "lr": 3.2355052468681147e-06, "epoch": 2.289230769230769, "percentage": 76.3, "elapsed_time": "8:09:22", "remaining_time": "2:31:57"} +{"current_steps": 4094, "total_steps": 5364, "loss": 0.05582427233457565, "lr": 3.2307133329666297e-06, "epoch": 2.2897902097902096, "percentage": 76.32, "elapsed_time": "8:09:29", "remaining_time": "2:31:50"} +{"current_steps": 4095, "total_steps": 5364, "loss": 0.08884363621473312, "lr": 3.225924286463543e-06, "epoch": 2.2903496503496505, "percentage": 76.34, "elapsed_time": "8:09:36", "remaining_time": "2:31:43"} +{"current_steps": 4096, "total_steps": 5364, "loss": 0.07232660055160522, "lr": 3.221138109387446e-06, "epoch": 2.290909090909091, "percentage": 76.36, "elapsed_time": "8:09:43", "remaining_time": "2:31:36"} +{"current_steps": 4097, "total_steps": 5364, "loss": 0.08202521502971649, "lr": 3.2163548037657143e-06, "epoch": 2.2914685314685315, "percentage": 76.38, "elapsed_time": "8:09:51", "remaining_time": "2:31:29"} +{"current_steps": 4098, "total_steps": 5364, "loss": 0.13254153728485107, "lr": 3.211574371624502e-06, "epoch": 2.292027972027972, "percentage": 76.4, "elapsed_time": "8:09:58", "remaining_time": "2:31:22"} +{"current_steps": 4099, "total_steps": 5364, "loss": 0.10033851116895676, "lr": 3.2067968149887496e-06, "epoch": 2.2925874125874124, "percentage": 76.42, "elapsed_time": "8:10:05", "remaining_time": "2:31:14"} +{"current_steps": 4100, "total_steps": 5364, "loss": 0.06426310539245605, "lr": 3.202022135882182e-06, "epoch": 2.2931468531468533, "percentage": 76.44, "elapsed_time": "8:10:12", "remaining_time": "2:31:07"} +{"current_steps": 4101, "total_steps": 5364, "loss": 0.19269779324531555, "lr": 3.1972503363273034e-06, "epoch": 2.2937062937062938, "percentage": 76.45, "elapsed_time": "8:10:19", "remaining_time": "2:31:00"} +{"current_steps": 4102, "total_steps": 5364, "loss": 0.12282806634902954, "lr": 3.1924814183453977e-06, "epoch": 2.294265734265734, "percentage": 76.47, "elapsed_time": "8:10:26", "remaining_time": "2:30:53"} +{"current_steps": 4103, "total_steps": 5364, "loss": 0.08900883048772812, "lr": 3.1877153839565255e-06, "epoch": 2.2948251748251747, "percentage": 76.49, "elapsed_time": "8:10:33", "remaining_time": "2:30:46"} +{"current_steps": 4104, "total_steps": 5364, "loss": 0.08213920891284943, "lr": 3.1829522351795294e-06, "epoch": 2.2953846153846156, "percentage": 76.51, "elapsed_time": "8:10:40", "remaining_time": "2:30:38"} +{"current_steps": 4105, "total_steps": 5364, "loss": 0.07684902846813202, "lr": 3.17819197403203e-06, "epoch": 2.295944055944056, "percentage": 76.53, "elapsed_time": "8:10:47", "remaining_time": "2:30:31"} +{"current_steps": 4106, "total_steps": 5364, "loss": 0.041487667709589005, "lr": 3.1734346025304242e-06, "epoch": 2.2965034965034965, "percentage": 76.55, "elapsed_time": "8:10:54", "remaining_time": "2:30:24"} +{"current_steps": 4107, "total_steps": 5364, "loss": 0.03580848127603531, "lr": 3.1686801226898854e-06, "epoch": 2.297062937062937, "percentage": 76.57, "elapsed_time": "8:11:01", "remaining_time": "2:30:17"} +{"current_steps": 4108, "total_steps": 5364, "loss": 0.09804193675518036, "lr": 3.1639285365243575e-06, "epoch": 2.2976223776223774, "percentage": 76.58, "elapsed_time": "8:11:08", "remaining_time": "2:30:09"} +{"current_steps": 4109, "total_steps": 5364, "loss": 0.19630542397499084, "lr": 3.159179846046563e-06, "epoch": 2.2981818181818183, "percentage": 76.6, "elapsed_time": "8:11:15", "remaining_time": "2:30:02"} +{"current_steps": 4110, "total_steps": 5364, "loss": 0.08246155083179474, "lr": 3.1544340532679983e-06, "epoch": 2.298741258741259, "percentage": 76.62, "elapsed_time": "8:11:22", "remaining_time": "2:29:55"} +{"current_steps": 4111, "total_steps": 5364, "loss": 0.09034265577793121, "lr": 3.1496911601989356e-06, "epoch": 2.2993006993006992, "percentage": 76.64, "elapsed_time": "8:11:29", "remaining_time": "2:29:48"} +{"current_steps": 4112, "total_steps": 5364, "loss": 0.2439144253730774, "lr": 3.1449511688484024e-06, "epoch": 2.2998601398601397, "percentage": 76.66, "elapsed_time": "8:11:36", "remaining_time": "2:29:40"} +{"current_steps": 4113, "total_steps": 5364, "loss": 0.053342171013355255, "lr": 3.1402140812242233e-06, "epoch": 2.3004195804195806, "percentage": 76.68, "elapsed_time": "8:11:43", "remaining_time": "2:29:33"} +{"current_steps": 4114, "total_steps": 5364, "loss": 0.04787937551736832, "lr": 3.1354798993329673e-06, "epoch": 2.300979020979021, "percentage": 76.7, "elapsed_time": "8:11:49", "remaining_time": "2:29:26"} +{"current_steps": 4115, "total_steps": 5364, "loss": 0.1571027636528015, "lr": 3.1307486251799903e-06, "epoch": 2.3015384615384615, "percentage": 76.72, "elapsed_time": "8:11:56", "remaining_time": "2:29:19"} +{"current_steps": 4116, "total_steps": 5364, "loss": 0.008515046909451485, "lr": 3.126020260769409e-06, "epoch": 2.302097902097902, "percentage": 76.73, "elapsed_time": "8:12:03", "remaining_time": "2:29:11"} +{"current_steps": 4117, "total_steps": 5364, "loss": 0.08391707390546799, "lr": 3.1212948081041028e-06, "epoch": 2.3026573426573425, "percentage": 76.75, "elapsed_time": "8:12:11", "remaining_time": "2:29:04"} +{"current_steps": 4118, "total_steps": 5364, "loss": 0.22323326766490936, "lr": 3.116572269185735e-06, "epoch": 2.3032167832167834, "percentage": 76.77, "elapsed_time": "8:12:18", "remaining_time": "2:28:57"} +{"current_steps": 4119, "total_steps": 5364, "loss": 0.06100432202219963, "lr": 3.1118526460147136e-06, "epoch": 2.303776223776224, "percentage": 76.79, "elapsed_time": "8:12:25", "remaining_time": "2:28:50"} +{"current_steps": 4120, "total_steps": 5364, "loss": 0.036345768719911575, "lr": 3.1071359405902246e-06, "epoch": 2.3043356643356643, "percentage": 76.81, "elapsed_time": "8:12:33", "remaining_time": "2:28:43"} +{"current_steps": 4121, "total_steps": 5364, "loss": 0.07189509272575378, "lr": 3.102422154910217e-06, "epoch": 2.3048951048951047, "percentage": 76.83, "elapsed_time": "8:12:40", "remaining_time": "2:28:36"} +{"current_steps": 4122, "total_steps": 5364, "loss": 0.0929940864443779, "lr": 3.0977112909713923e-06, "epoch": 2.3054545454545456, "percentage": 76.85, "elapsed_time": "8:12:47", "remaining_time": "2:28:28"} +{"current_steps": 4123, "total_steps": 5364, "loss": 0.1341400295495987, "lr": 3.0930033507692346e-06, "epoch": 2.306013986013986, "percentage": 76.86, "elapsed_time": "8:12:54", "remaining_time": "2:28:21"} +{"current_steps": 4124, "total_steps": 5364, "loss": 0.2187308520078659, "lr": 3.088298336297968e-06, "epoch": 2.3065734265734266, "percentage": 76.88, "elapsed_time": "8:13:01", "remaining_time": "2:28:14"} +{"current_steps": 4125, "total_steps": 5364, "loss": 0.027857935056090355, "lr": 3.083596249550591e-06, "epoch": 2.307132867132867, "percentage": 76.9, "elapsed_time": "8:13:08", "remaining_time": "2:28:07"} +{"current_steps": 4126, "total_steps": 5364, "loss": 0.06730443984270096, "lr": 3.0788970925188622e-06, "epoch": 2.3076923076923075, "percentage": 76.92, "elapsed_time": "8:13:15", "remaining_time": "2:27:59"} +{"current_steps": 4127, "total_steps": 5364, "loss": 0.061534080654382706, "lr": 3.0742008671932823e-06, "epoch": 2.3082517482517484, "percentage": 76.94, "elapsed_time": "8:13:22", "remaining_time": "2:27:52"} +{"current_steps": 4128, "total_steps": 5364, "loss": 0.09674179553985596, "lr": 3.0695075755631376e-06, "epoch": 2.308811188811189, "percentage": 76.96, "elapsed_time": "8:13:29", "remaining_time": "2:27:45"} +{"current_steps": 4129, "total_steps": 5364, "loss": 0.012089809402823448, "lr": 3.064817219616447e-06, "epoch": 2.3093706293706293, "percentage": 76.98, "elapsed_time": "8:13:36", "remaining_time": "2:27:38"} +{"current_steps": 4130, "total_steps": 5364, "loss": 0.023520154878497124, "lr": 3.0601298013399983e-06, "epoch": 2.3099300699300698, "percentage": 76.99, "elapsed_time": "8:13:43", "remaining_time": "2:27:31"} +{"current_steps": 4131, "total_steps": 5364, "loss": 0.061036668717861176, "lr": 3.055445322719337e-06, "epoch": 2.3104895104895107, "percentage": 77.01, "elapsed_time": "8:13:50", "remaining_time": "2:27:23"} +{"current_steps": 4132, "total_steps": 5364, "loss": 0.07786140590906143, "lr": 3.0507637857387486e-06, "epoch": 2.311048951048951, "percentage": 77.03, "elapsed_time": "8:13:57", "remaining_time": "2:27:16"} +{"current_steps": 4133, "total_steps": 5364, "loss": 0.16588738560676575, "lr": 3.0460851923812953e-06, "epoch": 2.3116083916083916, "percentage": 77.05, "elapsed_time": "8:14:04", "remaining_time": "2:27:09"} +{"current_steps": 4134, "total_steps": 5364, "loss": 0.06233539432287216, "lr": 3.0414095446287696e-06, "epoch": 2.312167832167832, "percentage": 77.07, "elapsed_time": "8:14:11", "remaining_time": "2:27:02"} +{"current_steps": 4135, "total_steps": 5364, "loss": 0.05115117132663727, "lr": 3.0367368444617307e-06, "epoch": 2.3127272727272725, "percentage": 77.09, "elapsed_time": "8:14:18", "remaining_time": "2:26:55"} +{"current_steps": 4136, "total_steps": 5364, "loss": 0.021471742540597916, "lr": 3.0320670938594887e-06, "epoch": 2.3132867132867134, "percentage": 77.11, "elapsed_time": "8:14:25", "remaining_time": "2:26:47"} +{"current_steps": 4137, "total_steps": 5364, "loss": 0.15726208686828613, "lr": 3.0274002948000925e-06, "epoch": 2.313846153846154, "percentage": 77.13, "elapsed_time": "8:14:33", "remaining_time": "2:26:40"} +{"current_steps": 4138, "total_steps": 5364, "loss": 0.029623623937368393, "lr": 3.022736449260354e-06, "epoch": 2.3144055944055943, "percentage": 77.14, "elapsed_time": "8:14:40", "remaining_time": "2:26:33"} +{"current_steps": 4139, "total_steps": 5364, "loss": 0.13494645059108734, "lr": 3.0180755592158286e-06, "epoch": 2.314965034965035, "percentage": 77.16, "elapsed_time": "8:14:47", "remaining_time": "2:26:26"} +{"current_steps": 4140, "total_steps": 5364, "loss": 0.07084748148918152, "lr": 3.01341762664082e-06, "epoch": 2.3155244755244757, "percentage": 77.18, "elapsed_time": "8:14:54", "remaining_time": "2:26:19"} +{"current_steps": 4141, "total_steps": 5364, "loss": 0.03671936318278313, "lr": 3.0087626535083814e-06, "epoch": 2.316083916083916, "percentage": 77.2, "elapsed_time": "8:15:01", "remaining_time": "2:26:12"} +{"current_steps": 4142, "total_steps": 5364, "loss": 0.17640714347362518, "lr": 3.0041106417903056e-06, "epoch": 2.3166433566433566, "percentage": 77.22, "elapsed_time": "8:15:09", "remaining_time": "2:26:04"} +{"current_steps": 4143, "total_steps": 5364, "loss": 0.035536833107471466, "lr": 2.999461593457137e-06, "epoch": 2.317202797202797, "percentage": 77.24, "elapsed_time": "8:15:16", "remaining_time": "2:25:57"} +{"current_steps": 4144, "total_steps": 5364, "loss": 0.041797198355197906, "lr": 2.994815510478165e-06, "epoch": 2.3177622377622376, "percentage": 77.26, "elapsed_time": "8:15:23", "remaining_time": "2:25:50"} +{"current_steps": 4145, "total_steps": 5364, "loss": 0.10188169777393341, "lr": 2.990172394821421e-06, "epoch": 2.3183216783216785, "percentage": 77.27, "elapsed_time": "8:15:30", "remaining_time": "2:25:43"} +{"current_steps": 4146, "total_steps": 5364, "loss": 0.06409934908151627, "lr": 2.9855322484536833e-06, "epoch": 2.318881118881119, "percentage": 77.29, "elapsed_time": "8:15:37", "remaining_time": "2:25:36"} +{"current_steps": 4147, "total_steps": 5364, "loss": 0.06261476874351501, "lr": 2.9808950733404606e-06, "epoch": 2.3194405594405594, "percentage": 77.31, "elapsed_time": "8:15:44", "remaining_time": "2:25:29"} +{"current_steps": 4148, "total_steps": 5364, "loss": 0.10893870145082474, "lr": 2.9762608714460173e-06, "epoch": 2.32, "percentage": 77.33, "elapsed_time": "8:15:51", "remaining_time": "2:25:21"} +{"current_steps": 4149, "total_steps": 5364, "loss": 0.24184440076351166, "lr": 2.9716296447333506e-06, "epoch": 2.3205594405594407, "percentage": 77.35, "elapsed_time": "8:15:58", "remaining_time": "2:25:14"} +{"current_steps": 4150, "total_steps": 5364, "loss": 0.09185001254081726, "lr": 2.9670013951641997e-06, "epoch": 2.321118881118881, "percentage": 77.37, "elapsed_time": "8:16:05", "remaining_time": "2:25:07"} +{"current_steps": 4151, "total_steps": 5364, "loss": 0.06915062665939331, "lr": 2.9623761246990423e-06, "epoch": 2.3216783216783217, "percentage": 77.39, "elapsed_time": "8:16:12", "remaining_time": "2:25:00"} +{"current_steps": 4152, "total_steps": 5364, "loss": 0.10790784657001495, "lr": 2.957753835297098e-06, "epoch": 2.322237762237762, "percentage": 77.4, "elapsed_time": "8:16:19", "remaining_time": "2:24:52"} +{"current_steps": 4153, "total_steps": 5364, "loss": 0.025478679686784744, "lr": 2.9531345289163114e-06, "epoch": 2.3227972027972026, "percentage": 77.42, "elapsed_time": "8:16:27", "remaining_time": "2:24:45"} +{"current_steps": 4154, "total_steps": 5364, "loss": 0.07190114259719849, "lr": 2.9485182075133777e-06, "epoch": 2.3233566433566435, "percentage": 77.44, "elapsed_time": "8:16:34", "remaining_time": "2:24:38"} +{"current_steps": 4155, "total_steps": 5364, "loss": 0.06151868775486946, "lr": 2.9439048730437214e-06, "epoch": 2.323916083916084, "percentage": 77.46, "elapsed_time": "8:16:41", "remaining_time": "2:24:31"} +{"current_steps": 4156, "total_steps": 5364, "loss": 0.24332216382026672, "lr": 2.9392945274615005e-06, "epoch": 2.3244755244755244, "percentage": 77.48, "elapsed_time": "8:16:47", "remaining_time": "2:24:24"} +{"current_steps": 4157, "total_steps": 5364, "loss": 0.030869830399751663, "lr": 2.9346871727196135e-06, "epoch": 2.325034965034965, "percentage": 77.5, "elapsed_time": "8:16:55", "remaining_time": "2:24:16"} +{"current_steps": 4158, "total_steps": 5364, "loss": 0.05284121632575989, "lr": 2.93008281076968e-06, "epoch": 2.325594405594406, "percentage": 77.52, "elapsed_time": "8:17:02", "remaining_time": "2:24:09"} +{"current_steps": 4159, "total_steps": 5364, "loss": 0.011515405960381031, "lr": 2.925481443562064e-06, "epoch": 2.3261538461538462, "percentage": 77.54, "elapsed_time": "8:17:09", "remaining_time": "2:24:02"} +{"current_steps": 4160, "total_steps": 5364, "loss": 0.06661350280046463, "lr": 2.920883073045856e-06, "epoch": 2.3267132867132867, "percentage": 77.55, "elapsed_time": "8:17:16", "remaining_time": "2:23:55"} +{"current_steps": 4161, "total_steps": 5364, "loss": 0.12680260837078094, "lr": 2.916287701168872e-06, "epoch": 2.327272727272727, "percentage": 77.57, "elapsed_time": "8:17:24", "remaining_time": "2:23:48"} +{"current_steps": 4162, "total_steps": 5364, "loss": 0.20449991524219513, "lr": 2.9116953298776717e-06, "epoch": 2.3278321678321676, "percentage": 77.59, "elapsed_time": "8:17:31", "remaining_time": "2:23:41"} +{"current_steps": 4163, "total_steps": 5364, "loss": 0.10279616713523865, "lr": 2.907105961117529e-06, "epoch": 2.3283916083916085, "percentage": 77.61, "elapsed_time": "8:17:37", "remaining_time": "2:23:33"} +{"current_steps": 4164, "total_steps": 5364, "loss": 0.055316224694252014, "lr": 2.9025195968324526e-06, "epoch": 2.328951048951049, "percentage": 77.63, "elapsed_time": "8:17:45", "remaining_time": "2:23:26"} +{"current_steps": 4165, "total_steps": 5364, "loss": 0.06609553098678589, "lr": 2.8979362389651824e-06, "epoch": 2.3295104895104894, "percentage": 77.65, "elapsed_time": "8:17:52", "remaining_time": "2:23:19"} +{"current_steps": 4166, "total_steps": 5364, "loss": 0.06627489626407623, "lr": 2.8933558894571713e-06, "epoch": 2.33006993006993, "percentage": 77.67, "elapsed_time": "8:17:59", "remaining_time": "2:23:12"} +{"current_steps": 4167, "total_steps": 5364, "loss": 0.015674112364649773, "lr": 2.888778550248618e-06, "epoch": 2.330629370629371, "percentage": 77.68, "elapsed_time": "8:18:07", "remaining_time": "2:23:05"} +{"current_steps": 4168, "total_steps": 5364, "loss": 0.052654676139354706, "lr": 2.8842042232784274e-06, "epoch": 2.3311888111888113, "percentage": 77.7, "elapsed_time": "8:18:13", "remaining_time": "2:22:58"} +{"current_steps": 4169, "total_steps": 5364, "loss": 0.15021787583827972, "lr": 2.879632910484239e-06, "epoch": 2.3317482517482517, "percentage": 77.72, "elapsed_time": "8:18:20", "remaining_time": "2:22:50"} +{"current_steps": 4170, "total_steps": 5364, "loss": 0.05389862507581711, "lr": 2.875064613802415e-06, "epoch": 2.332307692307692, "percentage": 77.74, "elapsed_time": "8:18:27", "remaining_time": "2:22:43"} +{"current_steps": 4171, "total_steps": 5364, "loss": 0.009877381846308708, "lr": 2.870499335168029e-06, "epoch": 2.3328671328671327, "percentage": 77.76, "elapsed_time": "8:18:34", "remaining_time": "2:22:36"} +{"current_steps": 4172, "total_steps": 5364, "loss": 0.09953927993774414, "lr": 2.8659370765148963e-06, "epoch": 2.3334265734265736, "percentage": 77.78, "elapsed_time": "8:18:41", "remaining_time": "2:22:29"} +{"current_steps": 4173, "total_steps": 5364, "loss": 0.0521346852183342, "lr": 2.8613778397755332e-06, "epoch": 2.333986013986014, "percentage": 77.8, "elapsed_time": "8:18:48", "remaining_time": "2:22:21"} +{"current_steps": 4174, "total_steps": 5364, "loss": 0.0974147617816925, "lr": 2.8568216268811876e-06, "epoch": 2.3345454545454545, "percentage": 77.82, "elapsed_time": "8:18:55", "remaining_time": "2:22:14"} +{"current_steps": 4175, "total_steps": 5364, "loss": 0.03616798669099808, "lr": 2.8522684397618263e-06, "epoch": 2.335104895104895, "percentage": 77.83, "elapsed_time": "8:19:03", "remaining_time": "2:22:07"} +{"current_steps": 4176, "total_steps": 5364, "loss": 0.04907721281051636, "lr": 2.84771828034612e-06, "epoch": 2.335664335664336, "percentage": 77.85, "elapsed_time": "8:19:10", "remaining_time": "2:22:00"} +{"current_steps": 4177, "total_steps": 5364, "loss": 0.18164277076721191, "lr": 2.8431711505614832e-06, "epoch": 2.3362237762237763, "percentage": 77.87, "elapsed_time": "8:19:17", "remaining_time": "2:21:53"} +{"current_steps": 4178, "total_steps": 5364, "loss": 0.11378076672554016, "lr": 2.838627052334022e-06, "epoch": 2.3367832167832168, "percentage": 77.89, "elapsed_time": "8:19:24", "remaining_time": "2:21:46"} +{"current_steps": 4179, "total_steps": 5364, "loss": 0.10286635160446167, "lr": 2.8340859875885708e-06, "epoch": 2.3373426573426572, "percentage": 77.91, "elapsed_time": "8:19:31", "remaining_time": "2:21:38"} +{"current_steps": 4180, "total_steps": 5364, "loss": 0.13022872805595398, "lr": 2.8295479582486794e-06, "epoch": 2.3379020979020977, "percentage": 77.93, "elapsed_time": "8:19:38", "remaining_time": "2:21:31"} +{"current_steps": 4181, "total_steps": 5364, "loss": 0.08087345957756042, "lr": 2.825012966236602e-06, "epoch": 2.3384615384615386, "percentage": 77.95, "elapsed_time": "8:19:45", "remaining_time": "2:21:24"} +{"current_steps": 4182, "total_steps": 5364, "loss": 0.05846138671040535, "lr": 2.820481013473324e-06, "epoch": 2.339020979020979, "percentage": 77.96, "elapsed_time": "8:19:52", "remaining_time": "2:21:17"} +{"current_steps": 4183, "total_steps": 5364, "loss": 0.11591841280460358, "lr": 2.815952101878524e-06, "epoch": 2.3395804195804195, "percentage": 77.98, "elapsed_time": "8:19:59", "remaining_time": "2:21:09"} +{"current_steps": 4184, "total_steps": 5364, "loss": 0.10986776649951935, "lr": 2.8114262333706045e-06, "epoch": 2.34013986013986, "percentage": 78.0, "elapsed_time": "8:20:06", "remaining_time": "2:21:02"} +{"current_steps": 4185, "total_steps": 5364, "loss": 0.1388150453567505, "lr": 2.806903409866678e-06, "epoch": 2.340699300699301, "percentage": 78.02, "elapsed_time": "8:20:13", "remaining_time": "2:20:55"} +{"current_steps": 4186, "total_steps": 5364, "loss": 0.06061611324548721, "lr": 2.8023836332825595e-06, "epoch": 2.3412587412587413, "percentage": 78.04, "elapsed_time": "8:20:20", "remaining_time": "2:20:48"} +{"current_steps": 4187, "total_steps": 5364, "loss": 0.06075330823659897, "lr": 2.7978669055327812e-06, "epoch": 2.341818181818182, "percentage": 78.06, "elapsed_time": "8:20:27", "remaining_time": "2:20:41"} +{"current_steps": 4188, "total_steps": 5364, "loss": 0.16876676678657532, "lr": 2.793353228530581e-06, "epoch": 2.3423776223776223, "percentage": 78.08, "elapsed_time": "8:20:34", "remaining_time": "2:20:33"} +{"current_steps": 4189, "total_steps": 5364, "loss": 0.09192757308483124, "lr": 2.7888426041879056e-06, "epoch": 2.3429370629370627, "percentage": 78.09, "elapsed_time": "8:20:41", "remaining_time": "2:20:26"} +{"current_steps": 4190, "total_steps": 5364, "loss": 0.0636255294084549, "lr": 2.7843350344154087e-06, "epoch": 2.3434965034965036, "percentage": 78.11, "elapsed_time": "8:20:49", "remaining_time": "2:20:19"} +{"current_steps": 4191, "total_steps": 5364, "loss": 0.05475534498691559, "lr": 2.7798305211224497e-06, "epoch": 2.344055944055944, "percentage": 78.13, "elapsed_time": "8:20:57", "remaining_time": "2:20:12"} +{"current_steps": 4192, "total_steps": 5364, "loss": 0.015556749887764454, "lr": 2.7753290662170896e-06, "epoch": 2.3446153846153845, "percentage": 78.15, "elapsed_time": "8:21:04", "remaining_time": "2:20:05"} +{"current_steps": 4193, "total_steps": 5364, "loss": 0.1287653148174286, "lr": 2.770830671606097e-06, "epoch": 2.345174825174825, "percentage": 78.17, "elapsed_time": "8:21:11", "remaining_time": "2:19:58"} +{"current_steps": 4194, "total_steps": 5364, "loss": 0.13449427485466003, "lr": 2.7663353391949477e-06, "epoch": 2.345734265734266, "percentage": 78.19, "elapsed_time": "8:21:18", "remaining_time": "2:19:50"} +{"current_steps": 4195, "total_steps": 5364, "loss": 0.12945351004600525, "lr": 2.7618430708878165e-06, "epoch": 2.3462937062937064, "percentage": 78.21, "elapsed_time": "8:21:25", "remaining_time": "2:19:43"} +{"current_steps": 4196, "total_steps": 5364, "loss": 0.050352297723293304, "lr": 2.757353868587582e-06, "epoch": 2.346853146853147, "percentage": 78.23, "elapsed_time": "8:21:33", "remaining_time": "2:19:36"} +{"current_steps": 4197, "total_steps": 5364, "loss": 0.10918182134628296, "lr": 2.7528677341958175e-06, "epoch": 2.3474125874125873, "percentage": 78.24, "elapsed_time": "8:21:40", "remaining_time": "2:19:29"} +{"current_steps": 4198, "total_steps": 5364, "loss": 0.0763801857829094, "lr": 2.7483846696128067e-06, "epoch": 2.3479720279720278, "percentage": 78.26, "elapsed_time": "8:21:47", "remaining_time": "2:19:22"} +{"current_steps": 4199, "total_steps": 5364, "loss": 0.05111249163746834, "lr": 2.7439046767375265e-06, "epoch": 2.3485314685314687, "percentage": 78.28, "elapsed_time": "8:21:54", "remaining_time": "2:19:15"} +{"current_steps": 4200, "total_steps": 5364, "loss": 0.018557261675596237, "lr": 2.7394277574676564e-06, "epoch": 2.349090909090909, "percentage": 78.3, "elapsed_time": "8:22:01", "remaining_time": "2:19:07"} +{"current_steps": 4201, "total_steps": 5364, "loss": 0.1314995288848877, "lr": 2.734953913699574e-06, "epoch": 2.3496503496503496, "percentage": 78.32, "elapsed_time": "8:22:08", "remaining_time": "2:19:00"} +{"current_steps": 4202, "total_steps": 5364, "loss": 0.11927692592144012, "lr": 2.7304831473283467e-06, "epoch": 2.35020979020979, "percentage": 78.34, "elapsed_time": "8:22:15", "remaining_time": "2:18:53"} +{"current_steps": 4203, "total_steps": 5364, "loss": 0.09752250462770462, "lr": 2.7260154602477474e-06, "epoch": 2.350769230769231, "percentage": 78.36, "elapsed_time": "8:22:22", "remaining_time": "2:18:46"} +{"current_steps": 4204, "total_steps": 5364, "loss": 0.06639213114976883, "lr": 2.7215508543502402e-06, "epoch": 2.3513286713286714, "percentage": 78.37, "elapsed_time": "8:22:29", "remaining_time": "2:18:39"} +{"current_steps": 4205, "total_steps": 5364, "loss": 0.10241188853979111, "lr": 2.7170893315269842e-06, "epoch": 2.351888111888112, "percentage": 78.39, "elapsed_time": "8:22:36", "remaining_time": "2:18:31"} +{"current_steps": 4206, "total_steps": 5364, "loss": 0.02258661389350891, "lr": 2.712630893667839e-06, "epoch": 2.3524475524475523, "percentage": 78.41, "elapsed_time": "8:22:43", "remaining_time": "2:18:24"} +{"current_steps": 4207, "total_steps": 5364, "loss": 0.09763824194669724, "lr": 2.708175542661342e-06, "epoch": 2.353006993006993, "percentage": 78.43, "elapsed_time": "8:22:50", "remaining_time": "2:18:17"} +{"current_steps": 4208, "total_steps": 5364, "loss": 0.0902264192700386, "lr": 2.7037232803947434e-06, "epoch": 2.3535664335664337, "percentage": 78.45, "elapsed_time": "8:22:57", "remaining_time": "2:18:10"} +{"current_steps": 4209, "total_steps": 5364, "loss": 0.0530712753534317, "lr": 2.6992741087539697e-06, "epoch": 2.354125874125874, "percentage": 78.47, "elapsed_time": "8:23:04", "remaining_time": "2:18:02"} +{"current_steps": 4210, "total_steps": 5364, "loss": 0.06394069641828537, "lr": 2.6948280296236373e-06, "epoch": 2.3546853146853146, "percentage": 78.49, "elapsed_time": "8:23:10", "remaining_time": "2:17:55"} +{"current_steps": 4211, "total_steps": 5364, "loss": 0.1131538450717926, "lr": 2.6903850448870693e-06, "epoch": 2.355244755244755, "percentage": 78.5, "elapsed_time": "8:23:17", "remaining_time": "2:17:48"} +{"current_steps": 4212, "total_steps": 5364, "loss": 0.062194716185331345, "lr": 2.6859451564262595e-06, "epoch": 2.355804195804196, "percentage": 78.52, "elapsed_time": "8:23:24", "remaining_time": "2:17:41"} +{"current_steps": 4213, "total_steps": 5364, "loss": 0.09972430020570755, "lr": 2.6815083661219e-06, "epoch": 2.3563636363636364, "percentage": 78.54, "elapsed_time": "8:23:31", "remaining_time": "2:17:33"} +{"current_steps": 4214, "total_steps": 5364, "loss": 0.1004643365740776, "lr": 2.6770746758533716e-06, "epoch": 2.356923076923077, "percentage": 78.56, "elapsed_time": "8:23:39", "remaining_time": "2:17:26"} +{"current_steps": 4215, "total_steps": 5364, "loss": 0.12935084104537964, "lr": 2.672644087498729e-06, "epoch": 2.3574825174825174, "percentage": 78.58, "elapsed_time": "8:23:46", "remaining_time": "2:17:19"} +{"current_steps": 4216, "total_steps": 5364, "loss": 0.16304895281791687, "lr": 2.6682166029347356e-06, "epoch": 2.358041958041958, "percentage": 78.6, "elapsed_time": "8:23:54", "remaining_time": "2:17:12"} +{"current_steps": 4217, "total_steps": 5364, "loss": 0.060444191098213196, "lr": 2.6637922240368184e-06, "epoch": 2.3586013986013987, "percentage": 78.62, "elapsed_time": "8:24:02", "remaining_time": "2:17:05"} +{"current_steps": 4218, "total_steps": 5364, "loss": 0.14504699409008026, "lr": 2.659370952679099e-06, "epoch": 2.359160839160839, "percentage": 78.64, "elapsed_time": "8:24:09", "remaining_time": "2:16:58"} +{"current_steps": 4219, "total_steps": 5364, "loss": 0.04341878741979599, "lr": 2.6549527907343865e-06, "epoch": 2.3597202797202796, "percentage": 78.65, "elapsed_time": "8:24:16", "remaining_time": "2:16:51"} +{"current_steps": 4220, "total_steps": 5364, "loss": 0.1116509884595871, "lr": 2.650537740074156e-06, "epoch": 2.36027972027972, "percentage": 78.67, "elapsed_time": "8:24:22", "remaining_time": "2:16:43"} +{"current_steps": 4221, "total_steps": 5364, "loss": 0.08082155138254166, "lr": 2.646125802568591e-06, "epoch": 2.360839160839161, "percentage": 78.69, "elapsed_time": "8:24:30", "remaining_time": "2:16:36"} +{"current_steps": 4222, "total_steps": 5364, "loss": 0.11466452479362488, "lr": 2.6417169800865306e-06, "epoch": 2.3613986013986015, "percentage": 78.71, "elapsed_time": "8:24:37", "remaining_time": "2:16:29"} +{"current_steps": 4223, "total_steps": 5364, "loss": 0.025832612067461014, "lr": 2.6373112744955098e-06, "epoch": 2.361958041958042, "percentage": 78.73, "elapsed_time": "8:24:44", "remaining_time": "2:16:22"} +{"current_steps": 4224, "total_steps": 5364, "loss": 0.1191239133477211, "lr": 2.632908687661743e-06, "epoch": 2.3625174825174824, "percentage": 78.75, "elapsed_time": "8:24:51", "remaining_time": "2:16:15"} +{"current_steps": 4225, "total_steps": 5364, "loss": 0.06522426009178162, "lr": 2.628509221450107e-06, "epoch": 2.363076923076923, "percentage": 78.77, "elapsed_time": "8:24:58", "remaining_time": "2:16:08"} +{"current_steps": 4226, "total_steps": 5364, "loss": 0.02990780584514141, "lr": 2.6241128777241844e-06, "epoch": 2.3636363636363638, "percentage": 78.78, "elapsed_time": "8:25:05", "remaining_time": "2:16:00"} +{"current_steps": 4227, "total_steps": 5364, "loss": 0.06367485970258713, "lr": 2.6197196583462104e-06, "epoch": 2.364195804195804, "percentage": 78.8, "elapsed_time": "8:25:12", "remaining_time": "2:15:53"} +{"current_steps": 4228, "total_steps": 5364, "loss": 0.019065149128437042, "lr": 2.6153295651771084e-06, "epoch": 2.3647552447552447, "percentage": 78.82, "elapsed_time": "8:25:19", "remaining_time": "2:15:46"} +{"current_steps": 4229, "total_steps": 5364, "loss": 0.0381324477493763, "lr": 2.610942600076476e-06, "epoch": 2.365314685314685, "percentage": 78.84, "elapsed_time": "8:25:26", "remaining_time": "2:15:39"} +{"current_steps": 4230, "total_steps": 5364, "loss": 0.24262671172618866, "lr": 2.606558764902585e-06, "epoch": 2.365874125874126, "percentage": 78.86, "elapsed_time": "8:25:33", "remaining_time": "2:15:31"} +{"current_steps": 4231, "total_steps": 5364, "loss": 0.133316770195961, "lr": 2.602178061512386e-06, "epoch": 2.3664335664335665, "percentage": 78.88, "elapsed_time": "8:25:41", "remaining_time": "2:15:24"} +{"current_steps": 4232, "total_steps": 5364, "loss": 0.14182843267917633, "lr": 2.597800491761492e-06, "epoch": 2.366993006993007, "percentage": 78.9, "elapsed_time": "8:25:47", "remaining_time": "2:15:17"} +{"current_steps": 4233, "total_steps": 5364, "loss": 0.18291951715946198, "lr": 2.5934260575041968e-06, "epoch": 2.3675524475524474, "percentage": 78.91, "elapsed_time": "8:25:54", "remaining_time": "2:15:10"} +{"current_steps": 4234, "total_steps": 5364, "loss": 0.06759610772132874, "lr": 2.589054760593468e-06, "epoch": 2.368111888111888, "percentage": 78.93, "elapsed_time": "8:26:01", "remaining_time": "2:15:03"} +{"current_steps": 4235, "total_steps": 5364, "loss": 0.1317383348941803, "lr": 2.5846866028809404e-06, "epoch": 2.368671328671329, "percentage": 78.95, "elapsed_time": "8:26:08", "remaining_time": "2:14:55"} +{"current_steps": 4236, "total_steps": 5364, "loss": 0.12118326127529144, "lr": 2.580321586216916e-06, "epoch": 2.3692307692307693, "percentage": 78.97, "elapsed_time": "8:26:15", "remaining_time": "2:14:48"} +{"current_steps": 4237, "total_steps": 5364, "loss": 0.11352255195379257, "lr": 2.575959712450371e-06, "epoch": 2.3697902097902097, "percentage": 78.99, "elapsed_time": "8:26:22", "remaining_time": "2:14:41"} +{"current_steps": 4238, "total_steps": 5364, "loss": 0.15802036225795746, "lr": 2.57160098342895e-06, "epoch": 2.37034965034965, "percentage": 79.01, "elapsed_time": "8:26:29", "remaining_time": "2:14:34"} +{"current_steps": 4239, "total_steps": 5364, "loss": 0.04245789349079132, "lr": 2.5672454009989645e-06, "epoch": 2.370909090909091, "percentage": 79.03, "elapsed_time": "8:26:37", "remaining_time": "2:14:27"} +{"current_steps": 4240, "total_steps": 5364, "loss": 0.14414645731449127, "lr": 2.562892967005397e-06, "epoch": 2.3714685314685315, "percentage": 79.05, "elapsed_time": "8:26:44", "remaining_time": "2:14:19"} +{"current_steps": 4241, "total_steps": 5364, "loss": 0.04403708502650261, "lr": 2.5585436832918853e-06, "epoch": 2.372027972027972, "percentage": 79.06, "elapsed_time": "8:26:51", "remaining_time": "2:14:12"} +{"current_steps": 4242, "total_steps": 5364, "loss": 0.11939035356044769, "lr": 2.5541975517007433e-06, "epoch": 2.3725874125874125, "percentage": 79.08, "elapsed_time": "8:26:58", "remaining_time": "2:14:05"} +{"current_steps": 4243, "total_steps": 5364, "loss": 0.06096184253692627, "lr": 2.5498545740729474e-06, "epoch": 2.373146853146853, "percentage": 79.1, "elapsed_time": "8:27:05", "remaining_time": "2:13:58"} +{"current_steps": 4244, "total_steps": 5364, "loss": 0.07828359305858612, "lr": 2.545514752248136e-06, "epoch": 2.373706293706294, "percentage": 79.12, "elapsed_time": "8:27:12", "remaining_time": "2:13:51"} +{"current_steps": 4245, "total_steps": 5364, "loss": 0.05178485065698624, "lr": 2.541178088064615e-06, "epoch": 2.3742657342657343, "percentage": 79.14, "elapsed_time": "8:27:18", "remaining_time": "2:13:43"} +{"current_steps": 4246, "total_steps": 5364, "loss": 0.09666630625724792, "lr": 2.5368445833593423e-06, "epoch": 2.3748251748251747, "percentage": 79.16, "elapsed_time": "8:27:25", "remaining_time": "2:13:36"} +{"current_steps": 4247, "total_steps": 5364, "loss": 0.1351543366909027, "lr": 2.532514239967955e-06, "epoch": 2.375384615384615, "percentage": 79.18, "elapsed_time": "8:27:32", "remaining_time": "2:13:29"} +{"current_steps": 4248, "total_steps": 5364, "loss": 0.07165377587080002, "lr": 2.5281870597247315e-06, "epoch": 2.375944055944056, "percentage": 79.19, "elapsed_time": "8:27:40", "remaining_time": "2:13:22"} +{"current_steps": 4249, "total_steps": 5364, "loss": 0.05010291934013367, "lr": 2.5238630444626243e-06, "epoch": 2.3765034965034966, "percentage": 79.21, "elapsed_time": "8:27:47", "remaining_time": "2:13:15"} +{"current_steps": 4250, "total_steps": 5364, "loss": 0.015423586592078209, "lr": 2.5195421960132416e-06, "epoch": 2.377062937062937, "percentage": 79.23, "elapsed_time": "8:27:54", "remaining_time": "2:13:07"} +{"current_steps": 4251, "total_steps": 5364, "loss": 0.05329650267958641, "lr": 2.515224516206842e-06, "epoch": 2.3776223776223775, "percentage": 79.25, "elapsed_time": "8:28:01", "remaining_time": "2:13:00"} +{"current_steps": 4252, "total_steps": 5364, "loss": 0.11685922741889954, "lr": 2.5109100068723614e-06, "epoch": 2.378181818181818, "percentage": 79.27, "elapsed_time": "8:28:08", "remaining_time": "2:12:53"} +{"current_steps": 4253, "total_steps": 5364, "loss": 0.01302420161664486, "lr": 2.506598669837369e-06, "epoch": 2.378741258741259, "percentage": 79.29, "elapsed_time": "8:28:16", "remaining_time": "2:12:46"} +{"current_steps": 4254, "total_steps": 5364, "loss": 0.12375587224960327, "lr": 2.502290506928108e-06, "epoch": 2.3793006993006993, "percentage": 79.31, "elapsed_time": "8:28:23", "remaining_time": "2:12:39"} +{"current_steps": 4255, "total_steps": 5364, "loss": 0.08629684150218964, "lr": 2.4979855199694704e-06, "epoch": 2.37986013986014, "percentage": 79.33, "elapsed_time": "8:28:30", "remaining_time": "2:12:32"} +{"current_steps": 4256, "total_steps": 5364, "loss": 0.13536661863327026, "lr": 2.4936837107849966e-06, "epoch": 2.3804195804195802, "percentage": 79.34, "elapsed_time": "8:28:38", "remaining_time": "2:12:25"} +{"current_steps": 4257, "total_steps": 5364, "loss": 0.09614861011505127, "lr": 2.4893850811968977e-06, "epoch": 2.380979020979021, "percentage": 79.36, "elapsed_time": "8:28:45", "remaining_time": "2:12:17"} +{"current_steps": 4258, "total_steps": 5364, "loss": 0.09313364326953888, "lr": 2.4850896330260244e-06, "epoch": 2.3815384615384616, "percentage": 79.38, "elapsed_time": "8:28:52", "remaining_time": "2:12:10"} +{"current_steps": 4259, "total_steps": 5364, "loss": 0.024739248678088188, "lr": 2.480797368091875e-06, "epoch": 2.382097902097902, "percentage": 79.4, "elapsed_time": "8:28:59", "remaining_time": "2:12:03"} +{"current_steps": 4260, "total_steps": 5364, "loss": 0.058190375566482544, "lr": 2.47650828821262e-06, "epoch": 2.3826573426573425, "percentage": 79.42, "elapsed_time": "8:29:06", "remaining_time": "2:11:56"} +{"current_steps": 4261, "total_steps": 5364, "loss": 0.13404443860054016, "lr": 2.4722223952050606e-06, "epoch": 2.383216783216783, "percentage": 79.44, "elapsed_time": "8:29:14", "remaining_time": "2:11:49"} +{"current_steps": 4262, "total_steps": 5364, "loss": 0.12624502182006836, "lr": 2.4679396908846563e-06, "epoch": 2.383776223776224, "percentage": 79.46, "elapsed_time": "8:29:21", "remaining_time": "2:11:41"} +{"current_steps": 4263, "total_steps": 5364, "loss": 0.05548630282282829, "lr": 2.463660177065519e-06, "epoch": 2.3843356643356644, "percentage": 79.47, "elapsed_time": "8:29:28", "remaining_time": "2:11:34"} +{"current_steps": 4264, "total_steps": 5364, "loss": 0.024162402376532555, "lr": 2.4593838555604035e-06, "epoch": 2.384895104895105, "percentage": 79.49, "elapsed_time": "8:29:35", "remaining_time": "2:11:27"} +{"current_steps": 4265, "total_steps": 5364, "loss": 0.04397023841738701, "lr": 2.455110728180716e-06, "epoch": 2.3854545454545453, "percentage": 79.51, "elapsed_time": "8:29:42", "remaining_time": "2:11:20"} +{"current_steps": 4266, "total_steps": 5364, "loss": 0.007988613098859787, "lr": 2.4508407967365054e-06, "epoch": 2.386013986013986, "percentage": 79.53, "elapsed_time": "8:29:50", "remaining_time": "2:11:13"} +{"current_steps": 4267, "total_steps": 5364, "loss": 0.04263989254832268, "lr": 2.4465740630364707e-06, "epoch": 2.3865734265734266, "percentage": 79.55, "elapsed_time": "8:29:57", "remaining_time": "2:11:06"} +{"current_steps": 4268, "total_steps": 5364, "loss": 0.13074183464050293, "lr": 2.4423105288879556e-06, "epoch": 2.387132867132867, "percentage": 79.57, "elapsed_time": "8:30:04", "remaining_time": "2:10:59"} +{"current_steps": 4269, "total_steps": 5364, "loss": 0.07761231809854507, "lr": 2.438050196096948e-06, "epoch": 2.3876923076923076, "percentage": 79.59, "elapsed_time": "8:30:11", "remaining_time": "2:10:51"} +{"current_steps": 4270, "total_steps": 5364, "loss": 0.155867338180542, "lr": 2.4337930664680843e-06, "epoch": 2.388251748251748, "percentage": 79.6, "elapsed_time": "8:30:18", "remaining_time": "2:10:44"} +{"current_steps": 4271, "total_steps": 5364, "loss": 0.22660008072853088, "lr": 2.429539141804631e-06, "epoch": 2.388811188811189, "percentage": 79.62, "elapsed_time": "8:30:25", "remaining_time": "2:10:37"} +{"current_steps": 4272, "total_steps": 5364, "loss": 0.2083190381526947, "lr": 2.4252884239085105e-06, "epoch": 2.3893706293706294, "percentage": 79.64, "elapsed_time": "8:30:32", "remaining_time": "2:10:30"} +{"current_steps": 4273, "total_steps": 5364, "loss": 0.073543980717659, "lr": 2.4210409145802825e-06, "epoch": 2.38993006993007, "percentage": 79.66, "elapsed_time": "8:30:39", "remaining_time": "2:10:23"} +{"current_steps": 4274, "total_steps": 5364, "loss": 0.006336988881230354, "lr": 2.416796615619146e-06, "epoch": 2.3904895104895103, "percentage": 79.68, "elapsed_time": "8:30:47", "remaining_time": "2:10:15"} +{"current_steps": 4275, "total_steps": 5364, "loss": 0.1102716252207756, "lr": 2.412555528822944e-06, "epoch": 2.391048951048951, "percentage": 79.7, "elapsed_time": "8:30:54", "remaining_time": "2:10:08"} +{"current_steps": 4276, "total_steps": 5364, "loss": 0.16808021068572998, "lr": 2.4083176559881517e-06, "epoch": 2.3916083916083917, "percentage": 79.72, "elapsed_time": "8:31:01", "remaining_time": "2:10:01"} +{"current_steps": 4277, "total_steps": 5364, "loss": 0.13183780014514923, "lr": 2.404082998909889e-06, "epoch": 2.392167832167832, "percentage": 79.74, "elapsed_time": "8:31:08", "remaining_time": "2:09:54"} +{"current_steps": 4278, "total_steps": 5364, "loss": 0.04239516705274582, "lr": 2.3998515593819126e-06, "epoch": 2.3927272727272726, "percentage": 79.75, "elapsed_time": "8:31:15", "remaining_time": "2:09:47"} +{"current_steps": 4279, "total_steps": 5364, "loss": 0.032993134111166, "lr": 2.395623339196614e-06, "epoch": 2.393286713286713, "percentage": 79.77, "elapsed_time": "8:31:22", "remaining_time": "2:09:39"} +{"current_steps": 4280, "total_steps": 5364, "loss": 0.20396141707897186, "lr": 2.391398340145028e-06, "epoch": 2.393846153846154, "percentage": 79.79, "elapsed_time": "8:31:29", "remaining_time": "2:09:32"} +{"current_steps": 4281, "total_steps": 5364, "loss": 0.008912870660424232, "lr": 2.387176564016813e-06, "epoch": 2.3944055944055944, "percentage": 79.81, "elapsed_time": "8:31:36", "remaining_time": "2:09:25"} +{"current_steps": 4282, "total_steps": 5364, "loss": 0.06428366899490356, "lr": 2.3829580126002715e-06, "epoch": 2.394965034965035, "percentage": 79.83, "elapsed_time": "8:31:43", "remaining_time": "2:09:18"} +{"current_steps": 4283, "total_steps": 5364, "loss": 0.16745463013648987, "lr": 2.378742687682336e-06, "epoch": 2.3955244755244753, "percentage": 79.85, "elapsed_time": "8:31:50", "remaining_time": "2:09:11"} +{"current_steps": 4284, "total_steps": 5364, "loss": 0.053865693509578705, "lr": 2.3745305910485795e-06, "epoch": 2.3960839160839162, "percentage": 79.87, "elapsed_time": "8:31:57", "remaining_time": "2:09:03"} +{"current_steps": 4285, "total_steps": 5364, "loss": 0.06182485073804855, "lr": 2.37032172448319e-06, "epoch": 2.3966433566433567, "percentage": 79.88, "elapsed_time": "8:32:04", "remaining_time": "2:08:56"} +{"current_steps": 4286, "total_steps": 5364, "loss": 0.09955702722072601, "lr": 2.3661160897690127e-06, "epoch": 2.397202797202797, "percentage": 79.9, "elapsed_time": "8:32:12", "remaining_time": "2:08:49"} +{"current_steps": 4287, "total_steps": 5364, "loss": 0.06909747421741486, "lr": 2.3619136886875005e-06, "epoch": 2.3977622377622376, "percentage": 79.92, "elapsed_time": "8:32:19", "remaining_time": "2:08:42"} +{"current_steps": 4288, "total_steps": 5364, "loss": 0.024588461965322495, "lr": 2.357714523018748e-06, "epoch": 2.398321678321678, "percentage": 79.94, "elapsed_time": "8:32:26", "remaining_time": "2:08:35"} +{"current_steps": 4289, "total_steps": 5364, "loss": 0.1883913278579712, "lr": 2.353518594541483e-06, "epoch": 2.398881118881119, "percentage": 79.96, "elapsed_time": "8:32:33", "remaining_time": "2:08:28"} +{"current_steps": 4290, "total_steps": 5364, "loss": 0.2154640108346939, "lr": 2.3493259050330452e-06, "epoch": 2.3994405594405595, "percentage": 79.98, "elapsed_time": "8:32:40", "remaining_time": "2:08:20"} +{"current_steps": 4291, "total_steps": 5364, "loss": 0.039584964513778687, "lr": 2.3451364562694267e-06, "epoch": 2.4, "percentage": 80.0, "elapsed_time": "8:32:47", "remaining_time": "2:08:13"} +{"current_steps": 4292, "total_steps": 5364, "loss": 0.1151922345161438, "lr": 2.3409502500252258e-06, "epoch": 2.4005594405594404, "percentage": 80.01, "elapsed_time": "8:32:54", "remaining_time": "2:08:06"} +{"current_steps": 4293, "total_steps": 5364, "loss": 0.0916643887758255, "lr": 2.336767288073676e-06, "epoch": 2.4011188811188813, "percentage": 80.03, "elapsed_time": "8:33:01", "remaining_time": "2:07:59"} +{"current_steps": 4294, "total_steps": 5364, "loss": 0.023895790800452232, "lr": 2.3325875721866396e-06, "epoch": 2.4016783216783217, "percentage": 80.05, "elapsed_time": "8:33:08", "remaining_time": "2:07:52"} +{"current_steps": 4295, "total_steps": 5364, "loss": 0.07810373604297638, "lr": 2.3284111041345923e-06, "epoch": 2.402237762237762, "percentage": 80.07, "elapsed_time": "8:33:15", "remaining_time": "2:07:44"} +{"current_steps": 4296, "total_steps": 5364, "loss": 0.04361053183674812, "lr": 2.3242378856866534e-06, "epoch": 2.4027972027972027, "percentage": 80.09, "elapsed_time": "8:33:22", "remaining_time": "2:07:37"} +{"current_steps": 4297, "total_steps": 5364, "loss": 0.0860578715801239, "lr": 2.3200679186105446e-06, "epoch": 2.403356643356643, "percentage": 80.11, "elapsed_time": "8:33:29", "remaining_time": "2:07:30"} +{"current_steps": 4298, "total_steps": 5364, "loss": 0.03552931547164917, "lr": 2.3159012046726228e-06, "epoch": 2.403916083916084, "percentage": 80.13, "elapsed_time": "8:33:36", "remaining_time": "2:07:23"} +{"current_steps": 4299, "total_steps": 5364, "loss": 0.1972390115261078, "lr": 2.3117377456378677e-06, "epoch": 2.4044755244755245, "percentage": 80.15, "elapsed_time": "8:33:43", "remaining_time": "2:07:16"} +{"current_steps": 4300, "total_steps": 5364, "loss": 0.14579619467258453, "lr": 2.3075775432698678e-06, "epoch": 2.405034965034965, "percentage": 80.16, "elapsed_time": "8:33:51", "remaining_time": "2:07:09"} +{"current_steps": 4301, "total_steps": 5364, "loss": 0.12425071001052856, "lr": 2.3034205993308525e-06, "epoch": 2.4055944055944054, "percentage": 80.18, "elapsed_time": "8:33:58", "remaining_time": "2:07:01"} +{"current_steps": 4302, "total_steps": 5364, "loss": 0.1435418426990509, "lr": 2.2992669155816496e-06, "epoch": 2.4061538461538463, "percentage": 80.2, "elapsed_time": "8:34:05", "remaining_time": "2:06:54"} +{"current_steps": 4303, "total_steps": 5364, "loss": 0.031560156494379044, "lr": 2.2951164937817206e-06, "epoch": 2.4067132867132868, "percentage": 80.22, "elapsed_time": "8:34:12", "remaining_time": "2:06:47"} +{"current_steps": 4304, "total_steps": 5364, "loss": 0.09113756567239761, "lr": 2.2909693356891418e-06, "epoch": 2.4072727272727272, "percentage": 80.24, "elapsed_time": "8:34:19", "remaining_time": "2:06:40"} +{"current_steps": 4305, "total_steps": 5364, "loss": 0.03356871008872986, "lr": 2.2868254430605975e-06, "epoch": 2.4078321678321677, "percentage": 80.26, "elapsed_time": "8:34:26", "remaining_time": "2:06:33"} +{"current_steps": 4306, "total_steps": 5364, "loss": 0.14119921624660492, "lr": 2.2826848176514074e-06, "epoch": 2.408391608391608, "percentage": 80.28, "elapsed_time": "8:34:34", "remaining_time": "2:06:25"} +{"current_steps": 4307, "total_steps": 5364, "loss": 0.11919906735420227, "lr": 2.2785474612154902e-06, "epoch": 2.408951048951049, "percentage": 80.29, "elapsed_time": "8:34:41", "remaining_time": "2:06:18"} +{"current_steps": 4308, "total_steps": 5364, "loss": 0.0956508070230484, "lr": 2.2744133755053866e-06, "epoch": 2.4095104895104895, "percentage": 80.31, "elapsed_time": "8:34:48", "remaining_time": "2:06:11"} +{"current_steps": 4309, "total_steps": 5364, "loss": 0.07001488655805588, "lr": 2.270282562272257e-06, "epoch": 2.41006993006993, "percentage": 80.33, "elapsed_time": "8:34:55", "remaining_time": "2:06:04"} +{"current_steps": 4310, "total_steps": 5364, "loss": 0.11895737797021866, "lr": 2.2661550232658637e-06, "epoch": 2.4106293706293704, "percentage": 80.35, "elapsed_time": "8:35:02", "remaining_time": "2:05:57"} +{"current_steps": 4311, "total_steps": 5364, "loss": 0.19488152861595154, "lr": 2.2620307602345916e-06, "epoch": 2.4111888111888113, "percentage": 80.37, "elapsed_time": "8:35:08", "remaining_time": "2:05:49"} +{"current_steps": 4312, "total_steps": 5364, "loss": 0.0734558254480362, "lr": 2.257909774925434e-06, "epoch": 2.411748251748252, "percentage": 80.39, "elapsed_time": "8:35:16", "remaining_time": "2:05:42"} +{"current_steps": 4313, "total_steps": 5364, "loss": 0.061678335070610046, "lr": 2.2537920690839976e-06, "epoch": 2.4123076923076923, "percentage": 80.41, "elapsed_time": "8:35:23", "remaining_time": "2:05:35"} +{"current_steps": 4314, "total_steps": 5364, "loss": 0.026166999712586403, "lr": 2.249677644454503e-06, "epoch": 2.4128671328671327, "percentage": 80.43, "elapsed_time": "8:35:30", "remaining_time": "2:05:28"} +{"current_steps": 4315, "total_steps": 5364, "loss": 0.005541364662349224, "lr": 2.24556650277977e-06, "epoch": 2.413426573426573, "percentage": 80.44, "elapsed_time": "8:35:38", "remaining_time": "2:05:21"} +{"current_steps": 4316, "total_steps": 5364, "loss": 0.12966686487197876, "lr": 2.2414586458012375e-06, "epoch": 2.413986013986014, "percentage": 80.46, "elapsed_time": "8:35:45", "remaining_time": "2:05:14"} +{"current_steps": 4317, "total_steps": 5364, "loss": 0.09022810310125351, "lr": 2.237354075258952e-06, "epoch": 2.4145454545454546, "percentage": 80.48, "elapsed_time": "8:35:53", "remaining_time": "2:05:07"} +{"current_steps": 4318, "total_steps": 5364, "loss": 0.038913995027542114, "lr": 2.2332527928915637e-06, "epoch": 2.415104895104895, "percentage": 80.5, "elapsed_time": "8:36:00", "remaining_time": "2:04:59"} +{"current_steps": 4319, "total_steps": 5364, "loss": 0.07941877841949463, "lr": 2.229154800436335e-06, "epoch": 2.4156643356643355, "percentage": 80.52, "elapsed_time": "8:36:07", "remaining_time": "2:04:52"} +{"current_steps": 4320, "total_steps": 5364, "loss": 0.11963687092065811, "lr": 2.225060099629134e-06, "epoch": 2.4162237762237764, "percentage": 80.54, "elapsed_time": "8:36:14", "remaining_time": "2:04:45"} +{"current_steps": 4321, "total_steps": 5364, "loss": 0.0468156561255455, "lr": 2.220968692204426e-06, "epoch": 2.416783216783217, "percentage": 80.56, "elapsed_time": "8:36:21", "remaining_time": "2:04:38"} +{"current_steps": 4322, "total_steps": 5364, "loss": 0.08945789188146591, "lr": 2.216880579895293e-06, "epoch": 2.4173426573426573, "percentage": 80.57, "elapsed_time": "8:36:29", "remaining_time": "2:04:31"} +{"current_steps": 4323, "total_steps": 5364, "loss": 0.07148704677820206, "lr": 2.2127957644334154e-06, "epoch": 2.4179020979020978, "percentage": 80.59, "elapsed_time": "8:36:36", "remaining_time": "2:04:24"} +{"current_steps": 4324, "total_steps": 5364, "loss": 0.12272177636623383, "lr": 2.2087142475490775e-06, "epoch": 2.418461538461538, "percentage": 80.61, "elapsed_time": "8:36:43", "remaining_time": "2:04:16"} +{"current_steps": 4325, "total_steps": 5364, "loss": 0.10421647131443024, "lr": 2.20463603097117e-06, "epoch": 2.419020979020979, "percentage": 80.63, "elapsed_time": "8:36:50", "remaining_time": "2:04:09"} +{"current_steps": 4326, "total_steps": 5364, "loss": 0.04226556420326233, "lr": 2.2005611164271766e-06, "epoch": 2.4195804195804196, "percentage": 80.65, "elapsed_time": "8:36:56", "remaining_time": "2:04:02"} +{"current_steps": 4327, "total_steps": 5364, "loss": 0.010642962530255318, "lr": 2.196489505643191e-06, "epoch": 2.42013986013986, "percentage": 80.67, "elapsed_time": "8:37:04", "remaining_time": "2:03:55"} +{"current_steps": 4328, "total_steps": 5364, "loss": 0.08660288900136948, "lr": 2.192421200343905e-06, "epoch": 2.420699300699301, "percentage": 80.69, "elapsed_time": "8:37:12", "remaining_time": "2:03:48"} +{"current_steps": 4329, "total_steps": 5364, "loss": 0.13848651945590973, "lr": 2.1883562022526083e-06, "epoch": 2.4212587412587414, "percentage": 80.7, "elapsed_time": "8:37:18", "remaining_time": "2:03:40"} +{"current_steps": 4330, "total_steps": 5364, "loss": 0.08055999875068665, "lr": 2.1842945130911965e-06, "epoch": 2.421818181818182, "percentage": 80.72, "elapsed_time": "8:37:26", "remaining_time": "2:03:33"} +{"current_steps": 4331, "total_steps": 5364, "loss": 0.03976232558488846, "lr": 2.1802361345801493e-06, "epoch": 2.4223776223776223, "percentage": 80.74, "elapsed_time": "8:37:33", "remaining_time": "2:03:26"} +{"current_steps": 4332, "total_steps": 5364, "loss": 0.04665861651301384, "lr": 2.176181068438559e-06, "epoch": 2.422937062937063, "percentage": 80.76, "elapsed_time": "8:37:40", "remaining_time": "2:03:19"} +{"current_steps": 4333, "total_steps": 5364, "loss": 0.021116310730576515, "lr": 2.1721293163841108e-06, "epoch": 2.4234965034965033, "percentage": 80.78, "elapsed_time": "8:37:47", "remaining_time": "2:03:12"} +{"current_steps": 4334, "total_steps": 5364, "loss": 0.054423797875642776, "lr": 2.1680808801330733e-06, "epoch": 2.424055944055944, "percentage": 80.8, "elapsed_time": "8:37:54", "remaining_time": "2:03:05"} +{"current_steps": 4335, "total_steps": 5364, "loss": 0.02370445430278778, "lr": 2.164035761400336e-06, "epoch": 2.4246153846153846, "percentage": 80.82, "elapsed_time": "8:38:01", "remaining_time": "2:02:57"} +{"current_steps": 4336, "total_steps": 5364, "loss": 0.06737499684095383, "lr": 2.159993961899358e-06, "epoch": 2.425174825174825, "percentage": 80.84, "elapsed_time": "8:38:08", "remaining_time": "2:02:50"} +{"current_steps": 4337, "total_steps": 5364, "loss": 0.05144071951508522, "lr": 2.155955483342207e-06, "epoch": 2.425734265734266, "percentage": 80.85, "elapsed_time": "8:38:15", "remaining_time": "2:02:43"} +{"current_steps": 4338, "total_steps": 5364, "loss": 0.12657935917377472, "lr": 2.1519203274395417e-06, "epoch": 2.4262937062937064, "percentage": 80.87, "elapsed_time": "8:38:22", "remaining_time": "2:02:36"} +{"current_steps": 4339, "total_steps": 5364, "loss": 0.13071051239967346, "lr": 2.147888495900604e-06, "epoch": 2.426853146853147, "percentage": 80.89, "elapsed_time": "8:38:29", "remaining_time": "2:02:29"} +{"current_steps": 4340, "total_steps": 5364, "loss": 0.07877501845359802, "lr": 2.1438599904332445e-06, "epoch": 2.4274125874125874, "percentage": 80.91, "elapsed_time": "8:38:36", "remaining_time": "2:02:21"} +{"current_steps": 4341, "total_steps": 5364, "loss": 0.08164171874523163, "lr": 2.1398348127438905e-06, "epoch": 2.427972027972028, "percentage": 80.93, "elapsed_time": "8:38:44", "remaining_time": "2:02:14"} +{"current_steps": 4342, "total_steps": 5364, "loss": 0.01728185825049877, "lr": 2.1358129645375648e-06, "epoch": 2.4285314685314683, "percentage": 80.95, "elapsed_time": "8:38:51", "remaining_time": "2:02:07"} +{"current_steps": 4343, "total_steps": 5364, "loss": 0.06505485624074936, "lr": 2.131794447517883e-06, "epoch": 2.429090909090909, "percentage": 80.97, "elapsed_time": "8:38:59", "remaining_time": "2:02:00"} +{"current_steps": 4344, "total_steps": 5364, "loss": 0.06129543483257294, "lr": 2.1277792633870386e-06, "epoch": 2.4296503496503496, "percentage": 80.98, "elapsed_time": "8:39:06", "remaining_time": "2:01:53"} +{"current_steps": 4345, "total_steps": 5364, "loss": 0.09214981645345688, "lr": 2.123767413845833e-06, "epoch": 2.43020979020979, "percentage": 81.0, "elapsed_time": "8:39:13", "remaining_time": "2:01:46"} +{"current_steps": 4346, "total_steps": 5364, "loss": 0.13617411255836487, "lr": 2.1197589005936346e-06, "epoch": 2.430769230769231, "percentage": 81.02, "elapsed_time": "8:39:20", "remaining_time": "2:01:38"} +{"current_steps": 4347, "total_steps": 5364, "loss": 0.0847213938832283, "lr": 2.11575372532841e-06, "epoch": 2.4313286713286715, "percentage": 81.04, "elapsed_time": "8:39:27", "remaining_time": "2:01:31"} +{"current_steps": 4348, "total_steps": 5364, "loss": 0.027891680598258972, "lr": 2.1117518897467113e-06, "epoch": 2.431888111888112, "percentage": 81.06, "elapsed_time": "8:39:34", "remaining_time": "2:01:24"} +{"current_steps": 4349, "total_steps": 5364, "loss": 0.08984841406345367, "lr": 2.107753395543667e-06, "epoch": 2.4324475524475524, "percentage": 81.08, "elapsed_time": "8:39:41", "remaining_time": "2:01:17"} +{"current_steps": 4350, "total_steps": 5364, "loss": 0.09435883164405823, "lr": 2.1037582444130067e-06, "epoch": 2.433006993006993, "percentage": 81.1, "elapsed_time": "8:39:49", "remaining_time": "2:01:10"} +{"current_steps": 4351, "total_steps": 5364, "loss": 0.11791590601205826, "lr": 2.0997664380470262e-06, "epoch": 2.4335664335664333, "percentage": 81.11, "elapsed_time": "8:39:55", "remaining_time": "2:01:03"} +{"current_steps": 4352, "total_steps": 5364, "loss": 0.1421177089214325, "lr": 2.095777978136616e-06, "epoch": 2.4341258741258742, "percentage": 81.13, "elapsed_time": "8:40:03", "remaining_time": "2:00:55"} +{"current_steps": 4353, "total_steps": 5364, "loss": 0.027294915169477463, "lr": 2.0917928663712484e-06, "epoch": 2.4346853146853147, "percentage": 81.15, "elapsed_time": "8:40:10", "remaining_time": "2:00:48"} +{"current_steps": 4354, "total_steps": 5364, "loss": 0.11855275183916092, "lr": 2.087811104438965e-06, "epoch": 2.435244755244755, "percentage": 81.17, "elapsed_time": "8:40:17", "remaining_time": "2:00:41"} +{"current_steps": 4355, "total_steps": 5364, "loss": 0.1294851154088974, "lr": 2.083832694026411e-06, "epoch": 2.435804195804196, "percentage": 81.19, "elapsed_time": "8:40:24", "remaining_time": "2:00:34"} +{"current_steps": 4356, "total_steps": 5364, "loss": 0.058321014046669006, "lr": 2.0798576368187893e-06, "epoch": 2.4363636363636365, "percentage": 81.21, "elapsed_time": "8:40:31", "remaining_time": "2:00:27"} +{"current_steps": 4357, "total_steps": 5364, "loss": 0.021740924566984177, "lr": 2.075885934499896e-06, "epoch": 2.436923076923077, "percentage": 81.23, "elapsed_time": "8:40:38", "remaining_time": "2:00:19"} +{"current_steps": 4358, "total_steps": 5364, "loss": 0.08060051500797272, "lr": 2.0719175887521016e-06, "epoch": 2.4374825174825174, "percentage": 81.25, "elapsed_time": "8:40:45", "remaining_time": "2:00:12"} +{"current_steps": 4359, "total_steps": 5364, "loss": 0.08996862918138504, "lr": 2.067952601256359e-06, "epoch": 2.438041958041958, "percentage": 81.26, "elapsed_time": "8:40:53", "remaining_time": "2:00:05"} +{"current_steps": 4360, "total_steps": 5364, "loss": 0.03524692356586456, "lr": 2.0639909736921892e-06, "epoch": 2.438601398601399, "percentage": 81.28, "elapsed_time": "8:40:59", "remaining_time": "1:59:58"} +{"current_steps": 4361, "total_steps": 5364, "loss": 0.03271853178739548, "lr": 2.060032707737698e-06, "epoch": 2.4391608391608393, "percentage": 81.3, "elapsed_time": "8:41:06", "remaining_time": "1:59:51"} +{"current_steps": 4362, "total_steps": 5364, "loss": 0.06356614083051682, "lr": 2.056077805069565e-06, "epoch": 2.4397202797202797, "percentage": 81.32, "elapsed_time": "8:41:13", "remaining_time": "1:59:43"} +{"current_steps": 4363, "total_steps": 5364, "loss": 0.032593294978141785, "lr": 2.052126267363047e-06, "epoch": 2.44027972027972, "percentage": 81.34, "elapsed_time": "8:41:21", "remaining_time": "1:59:36"} +{"current_steps": 4364, "total_steps": 5364, "loss": 0.07341725379228592, "lr": 2.048178096291975e-06, "epoch": 2.440839160839161, "percentage": 81.36, "elapsed_time": "8:41:28", "remaining_time": "1:59:29"} +{"current_steps": 4365, "total_steps": 5364, "loss": 0.1449374258518219, "lr": 2.0442332935287466e-06, "epoch": 2.4413986013986015, "percentage": 81.38, "elapsed_time": "8:41:35", "remaining_time": "1:59:22"} +{"current_steps": 4366, "total_steps": 5364, "loss": 0.053326014429330826, "lr": 2.040291860744343e-06, "epoch": 2.441958041958042, "percentage": 81.39, "elapsed_time": "8:41:42", "remaining_time": "1:59:15"} +{"current_steps": 4367, "total_steps": 5364, "loss": 0.17070254683494568, "lr": 2.0363537996083114e-06, "epoch": 2.4425174825174825, "percentage": 81.41, "elapsed_time": "8:41:50", "remaining_time": "1:59:08"} +{"current_steps": 4368, "total_steps": 5364, "loss": 0.09563805907964706, "lr": 2.032419111788776e-06, "epoch": 2.443076923076923, "percentage": 81.43, "elapsed_time": "8:41:56", "remaining_time": "1:59:00"} +{"current_steps": 4369, "total_steps": 5364, "loss": 0.04321911185979843, "lr": 2.0284877989524285e-06, "epoch": 2.443636363636364, "percentage": 81.45, "elapsed_time": "8:42:04", "remaining_time": "1:58:53"} +{"current_steps": 4370, "total_steps": 5364, "loss": 0.03923831507563591, "lr": 2.024559862764528e-06, "epoch": 2.4441958041958043, "percentage": 81.47, "elapsed_time": "8:42:11", "remaining_time": "1:58:46"} +{"current_steps": 4371, "total_steps": 5364, "loss": 0.057186614722013474, "lr": 2.0206353048889093e-06, "epoch": 2.4447552447552447, "percentage": 81.49, "elapsed_time": "8:42:18", "remaining_time": "1:58:39"} +{"current_steps": 4372, "total_steps": 5364, "loss": 0.04584155231714249, "lr": 2.016714126987973e-06, "epoch": 2.445314685314685, "percentage": 81.51, "elapsed_time": "8:42:25", "remaining_time": "1:58:32"} +{"current_steps": 4373, "total_steps": 5364, "loss": 0.02422763779759407, "lr": 2.0127963307226894e-06, "epoch": 2.445874125874126, "percentage": 81.52, "elapsed_time": "8:42:33", "remaining_time": "1:58:25"} +{"current_steps": 4374, "total_steps": 5364, "loss": 0.09444719552993774, "lr": 2.008881917752599e-06, "epoch": 2.4464335664335666, "percentage": 81.54, "elapsed_time": "8:42:40", "remaining_time": "1:58:18"} +{"current_steps": 4375, "total_steps": 5364, "loss": 0.09193938970565796, "lr": 2.0049708897357977e-06, "epoch": 2.446993006993007, "percentage": 81.56, "elapsed_time": "8:42:47", "remaining_time": "1:58:10"} +{"current_steps": 4376, "total_steps": 5364, "loss": 0.04639626666903496, "lr": 2.0010632483289616e-06, "epoch": 2.4475524475524475, "percentage": 81.58, "elapsed_time": "8:42:54", "remaining_time": "1:58:03"} +{"current_steps": 4377, "total_steps": 5364, "loss": 0.03390798345208168, "lr": 1.997158995187325e-06, "epoch": 2.448111888111888, "percentage": 81.6, "elapsed_time": "8:43:02", "remaining_time": "1:57:56"} +{"current_steps": 4378, "total_steps": 5364, "loss": 0.1308169662952423, "lr": 1.993258131964687e-06, "epoch": 2.448671328671329, "percentage": 81.62, "elapsed_time": "8:43:09", "remaining_time": "1:57:49"} +{"current_steps": 4379, "total_steps": 5364, "loss": 0.09462116658687592, "lr": 1.9893606603134174e-06, "epoch": 2.4492307692307693, "percentage": 81.64, "elapsed_time": "8:43:16", "remaining_time": "1:57:42"} +{"current_steps": 4380, "total_steps": 5364, "loss": 0.06645186990499496, "lr": 1.985466581884434e-06, "epoch": 2.44979020979021, "percentage": 81.66, "elapsed_time": "8:43:23", "remaining_time": "1:57:34"} +{"current_steps": 4381, "total_steps": 5364, "loss": 0.04732794687151909, "lr": 1.9815758983272404e-06, "epoch": 2.4503496503496502, "percentage": 81.67, "elapsed_time": "8:43:30", "remaining_time": "1:57:27"} +{"current_steps": 4382, "total_steps": 5364, "loss": 0.09362974017858505, "lr": 1.977688611289881e-06, "epoch": 2.450909090909091, "percentage": 81.69, "elapsed_time": "8:43:37", "remaining_time": "1:57:20"} +{"current_steps": 4383, "total_steps": 5364, "loss": 0.034056156873703, "lr": 1.973804722418965e-06, "epoch": 2.4514685314685316, "percentage": 81.71, "elapsed_time": "8:43:43", "remaining_time": "1:57:13"} +{"current_steps": 4384, "total_steps": 5364, "loss": 0.028239693492650986, "lr": 1.969924233359678e-06, "epoch": 2.452027972027972, "percentage": 81.73, "elapsed_time": "8:43:51", "remaining_time": "1:57:06"} +{"current_steps": 4385, "total_steps": 5364, "loss": 0.007724790833890438, "lr": 1.9660471457557462e-06, "epoch": 2.4525874125874125, "percentage": 81.75, "elapsed_time": "8:43:58", "remaining_time": "1:56:59"} +{"current_steps": 4386, "total_steps": 5364, "loss": 0.08283065259456635, "lr": 1.9621734612494646e-06, "epoch": 2.453146853146853, "percentage": 81.77, "elapsed_time": "8:44:05", "remaining_time": "1:56:51"} +{"current_steps": 4387, "total_steps": 5364, "loss": 0.08691463619470596, "lr": 1.95830318148169e-06, "epoch": 2.453706293706294, "percentage": 81.79, "elapsed_time": "8:44:12", "remaining_time": "1:56:44"} +{"current_steps": 4388, "total_steps": 5364, "loss": 0.03459593653678894, "lr": 1.9544363080918216e-06, "epoch": 2.4542657342657344, "percentage": 81.8, "elapsed_time": "8:44:20", "remaining_time": "1:56:37"} +{"current_steps": 4389, "total_steps": 5364, "loss": 0.11298046261072159, "lr": 1.9505728427178384e-06, "epoch": 2.454825174825175, "percentage": 81.82, "elapsed_time": "8:44:28", "remaining_time": "1:56:30"} +{"current_steps": 4390, "total_steps": 5364, "loss": 0.12835749983787537, "lr": 1.9467127869962543e-06, "epoch": 2.4553846153846153, "percentage": 81.84, "elapsed_time": "8:44:35", "remaining_time": "1:56:23"} +{"current_steps": 4391, "total_steps": 5364, "loss": 0.013115103356540203, "lr": 1.942856142562153e-06, "epoch": 2.455944055944056, "percentage": 81.86, "elapsed_time": "8:44:42", "remaining_time": "1:56:16"} +{"current_steps": 4392, "total_steps": 5364, "loss": 0.07676021009683609, "lr": 1.939002911049168e-06, "epoch": 2.4565034965034966, "percentage": 81.88, "elapsed_time": "8:44:50", "remaining_time": "1:56:09"} +{"current_steps": 4393, "total_steps": 5364, "loss": 0.07798892259597778, "lr": 1.9351530940894827e-06, "epoch": 2.457062937062937, "percentage": 81.9, "elapsed_time": "8:44:57", "remaining_time": "1:56:01"} +{"current_steps": 4394, "total_steps": 5364, "loss": 0.056541360914707184, "lr": 1.931306693313847e-06, "epoch": 2.4576223776223776, "percentage": 81.92, "elapsed_time": "8:45:04", "remaining_time": "1:55:54"} +{"current_steps": 4395, "total_steps": 5364, "loss": 0.08987961709499359, "lr": 1.927463710351549e-06, "epoch": 2.458181818181818, "percentage": 81.94, "elapsed_time": "8:45:11", "remaining_time": "1:55:47"} +{"current_steps": 4396, "total_steps": 5364, "loss": 0.15675915777683258, "lr": 1.9236241468304385e-06, "epoch": 2.458741258741259, "percentage": 81.95, "elapsed_time": "8:45:18", "remaining_time": "1:55:40"} +{"current_steps": 4397, "total_steps": 5364, "loss": 0.04454007372260094, "lr": 1.9197880043769122e-06, "epoch": 2.4593006993006994, "percentage": 81.97, "elapsed_time": "8:45:26", "remaining_time": "1:55:33"} +{"current_steps": 4398, "total_steps": 5364, "loss": 0.08004562556743622, "lr": 1.9159552846159233e-06, "epoch": 2.45986013986014, "percentage": 81.99, "elapsed_time": "8:45:34", "remaining_time": "1:55:26"} +{"current_steps": 4399, "total_steps": 5364, "loss": 0.061470042914152145, "lr": 1.912125989170971e-06, "epoch": 2.4604195804195803, "percentage": 82.01, "elapsed_time": "8:45:41", "remaining_time": "1:55:19"} +{"current_steps": 4400, "total_steps": 5364, "loss": 0.12167413532733917, "lr": 1.9083001196640994e-06, "epoch": 2.460979020979021, "percentage": 82.03, "elapsed_time": "8:45:48", "remaining_time": "1:55:11"} +{"current_steps": 4401, "total_steps": 5364, "loss": 0.023522404953837395, "lr": 1.90447767771591e-06, "epoch": 2.4615384615384617, "percentage": 82.05, "elapsed_time": "8:45:55", "remaining_time": "1:55:04"} +{"current_steps": 4402, "total_steps": 5364, "loss": 0.032641515135765076, "lr": 1.9006586649455495e-06, "epoch": 2.462097902097902, "percentage": 82.07, "elapsed_time": "8:46:02", "remaining_time": "1:54:57"} +{"current_steps": 4403, "total_steps": 5364, "loss": 0.04072485864162445, "lr": 1.8968430829707097e-06, "epoch": 2.4626573426573426, "percentage": 82.08, "elapsed_time": "8:46:10", "remaining_time": "1:54:50"} +{"current_steps": 4404, "total_steps": 5364, "loss": 0.10253074765205383, "lr": 1.8930309334076346e-06, "epoch": 2.463216783216783, "percentage": 82.1, "elapsed_time": "8:46:17", "remaining_time": "1:54:43"} +{"current_steps": 4405, "total_steps": 5364, "loss": 0.14560942351818085, "lr": 1.889222217871104e-06, "epoch": 2.463776223776224, "percentage": 82.12, "elapsed_time": "8:46:24", "remaining_time": "1:54:36"} +{"current_steps": 4406, "total_steps": 5364, "loss": 0.16155387461185455, "lr": 1.8854169379744548e-06, "epoch": 2.4643356643356644, "percentage": 82.14, "elapsed_time": "8:46:31", "remaining_time": "1:54:29"} +{"current_steps": 4407, "total_steps": 5364, "loss": 0.06448402255773544, "lr": 1.881615095329561e-06, "epoch": 2.464895104895105, "percentage": 82.16, "elapsed_time": "8:46:38", "remaining_time": "1:54:21"} +{"current_steps": 4408, "total_steps": 5364, "loss": 0.07830452173948288, "lr": 1.8778166915468465e-06, "epoch": 2.4654545454545453, "percentage": 82.18, "elapsed_time": "8:46:45", "remaining_time": "1:54:14"} +{"current_steps": 4409, "total_steps": 5364, "loss": 0.05782242864370346, "lr": 1.8740217282352691e-06, "epoch": 2.4660139860139862, "percentage": 82.2, "elapsed_time": "8:46:52", "remaining_time": "1:54:07"} +{"current_steps": 4410, "total_steps": 5364, "loss": 0.10452114045619965, "lr": 1.8702302070023392e-06, "epoch": 2.4665734265734267, "percentage": 82.21, "elapsed_time": "8:46:59", "remaining_time": "1:54:00"} +{"current_steps": 4411, "total_steps": 5364, "loss": 0.0969105213880539, "lr": 1.8664421294541046e-06, "epoch": 2.467132867132867, "percentage": 82.23, "elapsed_time": "8:47:06", "remaining_time": "1:53:52"} +{"current_steps": 4412, "total_steps": 5364, "loss": 0.101371169090271, "lr": 1.862657497195155e-06, "epoch": 2.4676923076923076, "percentage": 82.25, "elapsed_time": "8:47:14", "remaining_time": "1:53:45"} +{"current_steps": 4413, "total_steps": 5364, "loss": 0.1356501281261444, "lr": 1.8588763118286223e-06, "epoch": 2.468251748251748, "percentage": 82.27, "elapsed_time": "8:47:21", "remaining_time": "1:53:38"} +{"current_steps": 4414, "total_steps": 5364, "loss": 0.11275427043437958, "lr": 1.8550985749561712e-06, "epoch": 2.468811188811189, "percentage": 82.29, "elapsed_time": "8:47:28", "remaining_time": "1:53:31"} +{"current_steps": 4415, "total_steps": 5364, "loss": 0.13677522540092468, "lr": 1.8513242881780192e-06, "epoch": 2.4693706293706295, "percentage": 82.31, "elapsed_time": "8:47:35", "remaining_time": "1:53:24"} +{"current_steps": 4416, "total_steps": 5364, "loss": 0.16282282769680023, "lr": 1.8475534530929074e-06, "epoch": 2.46993006993007, "percentage": 82.33, "elapsed_time": "8:47:42", "remaining_time": "1:53:17"} +{"current_steps": 4417, "total_steps": 5364, "loss": 0.1123458594083786, "lr": 1.843786071298125e-06, "epoch": 2.4704895104895104, "percentage": 82.35, "elapsed_time": "8:47:49", "remaining_time": "1:53:09"} +{"current_steps": 4418, "total_steps": 5364, "loss": 0.1560671031475067, "lr": 1.8400221443894971e-06, "epoch": 2.4710489510489513, "percentage": 82.36, "elapsed_time": "8:47:56", "remaining_time": "1:53:02"} +{"current_steps": 4419, "total_steps": 5364, "loss": 0.1323964148759842, "lr": 1.836261673961376e-06, "epoch": 2.4716083916083917, "percentage": 82.38, "elapsed_time": "8:48:04", "remaining_time": "1:52:55"} +{"current_steps": 4420, "total_steps": 5364, "loss": 0.016651686280965805, "lr": 1.8325046616066667e-06, "epoch": 2.472167832167832, "percentage": 82.4, "elapsed_time": "8:48:11", "remaining_time": "1:52:48"} +{"current_steps": 4421, "total_steps": 5364, "loss": 0.1184217780828476, "lr": 1.8287511089167932e-06, "epoch": 2.4727272727272727, "percentage": 82.42, "elapsed_time": "8:48:18", "remaining_time": "1:52:41"} +{"current_steps": 4422, "total_steps": 5364, "loss": 0.17062152922153473, "lr": 1.8250010174817234e-06, "epoch": 2.473286713286713, "percentage": 82.44, "elapsed_time": "8:48:25", "remaining_time": "1:52:34"} +{"current_steps": 4423, "total_steps": 5364, "loss": 0.08395298570394516, "lr": 1.821254388889958e-06, "epoch": 2.473846153846154, "percentage": 82.46, "elapsed_time": "8:48:32", "remaining_time": "1:52:26"} +{"current_steps": 4424, "total_steps": 5364, "loss": 0.051518164575099945, "lr": 1.817511224728523e-06, "epoch": 2.4744055944055945, "percentage": 82.48, "elapsed_time": "8:48:40", "remaining_time": "1:52:19"} +{"current_steps": 4425, "total_steps": 5364, "loss": 0.06093098223209381, "lr": 1.8137715265829937e-06, "epoch": 2.474965034965035, "percentage": 82.49, "elapsed_time": "8:48:47", "remaining_time": "1:52:12"} +{"current_steps": 4426, "total_steps": 5364, "loss": 0.050407372415065765, "lr": 1.8100352960374568e-06, "epoch": 2.4755244755244754, "percentage": 82.51, "elapsed_time": "8:48:54", "remaining_time": "1:52:05"} +{"current_steps": 4427, "total_steps": 5364, "loss": 0.0873202234506607, "lr": 1.8063025346745455e-06, "epoch": 2.4760839160839163, "percentage": 82.53, "elapsed_time": "8:49:01", "remaining_time": "1:51:58"} +{"current_steps": 4428, "total_steps": 5364, "loss": 0.11311569809913635, "lr": 1.8025732440754184e-06, "epoch": 2.4766433566433568, "percentage": 82.55, "elapsed_time": "8:49:09", "remaining_time": "1:51:51"} +{"current_steps": 4429, "total_steps": 5364, "loss": 0.15003004670143127, "lr": 1.7988474258197574e-06, "epoch": 2.4772027972027972, "percentage": 82.57, "elapsed_time": "8:49:15", "remaining_time": "1:51:43"} +{"current_steps": 4430, "total_steps": 5364, "loss": 0.1436362862586975, "lr": 1.795125081485789e-06, "epoch": 2.4777622377622377, "percentage": 82.59, "elapsed_time": "8:49:23", "remaining_time": "1:51:36"} +{"current_steps": 4431, "total_steps": 5364, "loss": 0.12151581048965454, "lr": 1.7914062126502551e-06, "epoch": 2.478321678321678, "percentage": 82.61, "elapsed_time": "8:49:29", "remaining_time": "1:51:29"} +{"current_steps": 4432, "total_steps": 5364, "loss": 0.03245886415243149, "lr": 1.7876908208884236e-06, "epoch": 2.478881118881119, "percentage": 82.62, "elapsed_time": "8:49:36", "remaining_time": "1:51:22"} +{"current_steps": 4433, "total_steps": 5364, "loss": 0.021765537559986115, "lr": 1.7839789077741043e-06, "epoch": 2.4794405594405595, "percentage": 82.64, "elapsed_time": "8:49:44", "remaining_time": "1:51:15"} +{"current_steps": 4434, "total_steps": 5364, "loss": 0.04360999912023544, "lr": 1.7802704748796185e-06, "epoch": 2.48, "percentage": 82.66, "elapsed_time": "8:49:51", "remaining_time": "1:51:07"} +{"current_steps": 4435, "total_steps": 5364, "loss": 0.06616771966218948, "lr": 1.7765655237758217e-06, "epoch": 2.4805594405594404, "percentage": 82.68, "elapsed_time": "8:49:58", "remaining_time": "1:51:00"} +{"current_steps": 4436, "total_steps": 5364, "loss": 0.10865949094295502, "lr": 1.772864056032091e-06, "epoch": 2.4811188811188813, "percentage": 82.7, "elapsed_time": "8:50:05", "remaining_time": "1:50:53"} +{"current_steps": 4437, "total_steps": 5364, "loss": 0.1378260850906372, "lr": 1.7691660732163297e-06, "epoch": 2.481678321678322, "percentage": 82.72, "elapsed_time": "8:50:12", "remaining_time": "1:50:46"} +{"current_steps": 4438, "total_steps": 5364, "loss": 0.05870186537504196, "lr": 1.765471576894966e-06, "epoch": 2.4822377622377623, "percentage": 82.74, "elapsed_time": "8:50:19", "remaining_time": "1:50:39"} +{"current_steps": 4439, "total_steps": 5364, "loss": 0.11319805681705475, "lr": 1.7617805686329458e-06, "epoch": 2.4827972027972027, "percentage": 82.76, "elapsed_time": "8:50:26", "remaining_time": "1:50:32"} +{"current_steps": 4440, "total_steps": 5364, "loss": 0.0752384141087532, "lr": 1.7580930499937431e-06, "epoch": 2.483356643356643, "percentage": 82.77, "elapsed_time": "8:50:34", "remaining_time": "1:50:25"} +{"current_steps": 4441, "total_steps": 5364, "loss": 0.0386233776807785, "lr": 1.7544090225393529e-06, "epoch": 2.483916083916084, "percentage": 82.79, "elapsed_time": "8:50:41", "remaining_time": "1:50:17"} +{"current_steps": 4442, "total_steps": 5364, "loss": 0.07144824415445328, "lr": 1.7507284878302887e-06, "epoch": 2.4844755244755246, "percentage": 82.81, "elapsed_time": "8:50:48", "remaining_time": "1:50:10"} +{"current_steps": 4443, "total_steps": 5364, "loss": 0.10955846309661865, "lr": 1.7470514474255895e-06, "epoch": 2.485034965034965, "percentage": 82.83, "elapsed_time": "8:50:56", "remaining_time": "1:50:03"} +{"current_steps": 4444, "total_steps": 5364, "loss": 0.09911411255598068, "lr": 1.7433779028828046e-06, "epoch": 2.4855944055944055, "percentage": 82.85, "elapsed_time": "8:51:03", "remaining_time": "1:49:56"} +{"current_steps": 4445, "total_steps": 5364, "loss": 0.07427871227264404, "lr": 1.7397078557580139e-06, "epoch": 2.4861538461538464, "percentage": 82.87, "elapsed_time": "8:51:10", "remaining_time": "1:49:49"} +{"current_steps": 4446, "total_steps": 5364, "loss": 0.011145062744617462, "lr": 1.736041307605808e-06, "epoch": 2.486713286713287, "percentage": 82.89, "elapsed_time": "8:51:17", "remaining_time": "1:49:41"} +{"current_steps": 4447, "total_steps": 5364, "loss": 0.07623221725225449, "lr": 1.7323782599792994e-06, "epoch": 2.4872727272727273, "percentage": 82.9, "elapsed_time": "8:51:24", "remaining_time": "1:49:34"} +{"current_steps": 4448, "total_steps": 5364, "loss": 0.08683040738105774, "lr": 1.7287187144301187e-06, "epoch": 2.4878321678321678, "percentage": 82.92, "elapsed_time": "8:51:31", "remaining_time": "1:49:27"} +{"current_steps": 4449, "total_steps": 5364, "loss": 0.020749079063534737, "lr": 1.7250626725084062e-06, "epoch": 2.4883916083916082, "percentage": 82.94, "elapsed_time": "8:51:38", "remaining_time": "1:49:20"} +{"current_steps": 4450, "total_steps": 5364, "loss": 0.03496145084500313, "lr": 1.7214101357628243e-06, "epoch": 2.488951048951049, "percentage": 82.96, "elapsed_time": "8:51:45", "remaining_time": "1:49:13"} +{"current_steps": 4451, "total_steps": 5364, "loss": 0.026832567527890205, "lr": 1.7177611057405486e-06, "epoch": 2.4895104895104896, "percentage": 82.98, "elapsed_time": "8:51:52", "remaining_time": "1:49:06"} +{"current_steps": 4452, "total_steps": 5364, "loss": 0.036321818828582764, "lr": 1.7141155839872714e-06, "epoch": 2.49006993006993, "percentage": 83.0, "elapsed_time": "8:51:59", "remaining_time": "1:48:58"} +{"current_steps": 4453, "total_steps": 5364, "loss": 0.030157815665006638, "lr": 1.7104735720471954e-06, "epoch": 2.4906293706293705, "percentage": 83.02, "elapsed_time": "8:52:07", "remaining_time": "1:48:51"} +{"current_steps": 4454, "total_steps": 5364, "loss": 0.15856845676898956, "lr": 1.7068350714630422e-06, "epoch": 2.4911888111888114, "percentage": 83.04, "elapsed_time": "8:52:14", "remaining_time": "1:48:44"} +{"current_steps": 4455, "total_steps": 5364, "loss": 0.0874677449464798, "lr": 1.7032000837760377e-06, "epoch": 2.491748251748252, "percentage": 83.05, "elapsed_time": "8:52:20", "remaining_time": "1:48:37"} +{"current_steps": 4456, "total_steps": 5364, "loss": 0.05432897061109543, "lr": 1.6995686105259245e-06, "epoch": 2.4923076923076923, "percentage": 83.07, "elapsed_time": "8:52:28", "remaining_time": "1:48:30"} +{"current_steps": 4457, "total_steps": 5364, "loss": 0.15511249005794525, "lr": 1.6959406532509614e-06, "epoch": 2.492867132867133, "percentage": 83.09, "elapsed_time": "8:52:35", "remaining_time": "1:48:22"} +{"current_steps": 4458, "total_steps": 5364, "loss": 0.05809831619262695, "lr": 1.6923162134879045e-06, "epoch": 2.4934265734265733, "percentage": 83.11, "elapsed_time": "8:52:42", "remaining_time": "1:48:15"} +{"current_steps": 4459, "total_steps": 5364, "loss": 0.03205804526805878, "lr": 1.688695292772038e-06, "epoch": 2.493986013986014, "percentage": 83.13, "elapsed_time": "8:52:49", "remaining_time": "1:48:08"} +{"current_steps": 4460, "total_steps": 5364, "loss": 0.02270149625837803, "lr": 1.685077892637137e-06, "epoch": 2.4945454545454546, "percentage": 83.15, "elapsed_time": "8:52:55", "remaining_time": "1:48:01"} +{"current_steps": 4461, "total_steps": 5364, "loss": 0.037082523107528687, "lr": 1.6814640146154993e-06, "epoch": 2.495104895104895, "percentage": 83.17, "elapsed_time": "8:53:03", "remaining_time": "1:47:54"} +{"current_steps": 4462, "total_steps": 5364, "loss": 0.20298242568969727, "lr": 1.677853660237926e-06, "epoch": 2.4956643356643355, "percentage": 83.18, "elapsed_time": "8:53:10", "remaining_time": "1:47:46"} +{"current_steps": 4463, "total_steps": 5364, "loss": 0.1126210168004036, "lr": 1.6742468310337178e-06, "epoch": 2.4962237762237764, "percentage": 83.2, "elapsed_time": "8:53:17", "remaining_time": "1:47:39"} +{"current_steps": 4464, "total_steps": 5364, "loss": 0.021989360451698303, "lr": 1.6706435285306988e-06, "epoch": 2.496783216783217, "percentage": 83.22, "elapsed_time": "8:53:25", "remaining_time": "1:47:32"} +{"current_steps": 4465, "total_steps": 5364, "loss": 0.10087059438228607, "lr": 1.6670437542551843e-06, "epoch": 2.4973426573426574, "percentage": 83.24, "elapsed_time": "8:53:32", "remaining_time": "1:47:25"} +{"current_steps": 4466, "total_steps": 5364, "loss": 0.2375773936510086, "lr": 1.6634475097320024e-06, "epoch": 2.497902097902098, "percentage": 83.26, "elapsed_time": "8:53:39", "remaining_time": "1:47:18"} +{"current_steps": 4467, "total_steps": 5364, "loss": 0.012536962516605854, "lr": 1.6598547964844869e-06, "epoch": 2.4984615384615383, "percentage": 83.28, "elapsed_time": "8:53:46", "remaining_time": "1:47:11"} +{"current_steps": 4468, "total_steps": 5364, "loss": 0.053984515368938446, "lr": 1.6562656160344648e-06, "epoch": 2.499020979020979, "percentage": 83.3, "elapsed_time": "8:53:53", "remaining_time": "1:47:03"} +{"current_steps": 4469, "total_steps": 5364, "loss": 0.14984266459941864, "lr": 1.6526799699022865e-06, "epoch": 2.4995804195804197, "percentage": 83.31, "elapsed_time": "8:54:00", "remaining_time": "1:46:56"} +{"current_steps": 4470, "total_steps": 5364, "loss": 0.06923025846481323, "lr": 1.6490978596067852e-06, "epoch": 2.50013986013986, "percentage": 83.33, "elapsed_time": "8:54:07", "remaining_time": "1:46:49"} +{"current_steps": 4471, "total_steps": 5364, "loss": 0.15117859840393066, "lr": 1.645519286665308e-06, "epoch": 2.5006993006993006, "percentage": 83.35, "elapsed_time": "8:54:15", "remaining_time": "1:46:42"} +{"current_steps": 4472, "total_steps": 5364, "loss": 0.0656021311879158, "lr": 1.6419442525937035e-06, "epoch": 2.5012587412587415, "percentage": 83.37, "elapsed_time": "8:54:22", "remaining_time": "1:46:35"} +{"current_steps": 4473, "total_steps": 5364, "loss": 0.17433050274848938, "lr": 1.6383727589063103e-06, "epoch": 2.501818181818182, "percentage": 83.39, "elapsed_time": "8:54:29", "remaining_time": "1:46:28"} +{"current_steps": 4474, "total_steps": 5364, "loss": 0.04447856545448303, "lr": 1.634804807115985e-06, "epoch": 2.5023776223776224, "percentage": 83.41, "elapsed_time": "8:54:36", "remaining_time": "1:46:20"} +{"current_steps": 4475, "total_steps": 5364, "loss": 0.15348252654075623, "lr": 1.6312403987340686e-06, "epoch": 2.502937062937063, "percentage": 83.43, "elapsed_time": "8:54:43", "remaining_time": "1:46:13"} +{"current_steps": 4476, "total_steps": 5364, "loss": 0.22798123955726624, "lr": 1.6276795352704077e-06, "epoch": 2.5034965034965033, "percentage": 83.45, "elapsed_time": "8:54:50", "remaining_time": "1:46:06"} +{"current_steps": 4477, "total_steps": 5364, "loss": 0.06473525613546371, "lr": 1.6241222182333493e-06, "epoch": 2.504055944055944, "percentage": 83.46, "elapsed_time": "8:54:57", "remaining_time": "1:45:59"} +{"current_steps": 4478, "total_steps": 5364, "loss": 0.07639111578464508, "lr": 1.620568449129728e-06, "epoch": 2.5046153846153847, "percentage": 83.48, "elapsed_time": "8:55:04", "remaining_time": "1:45:52"} +{"current_steps": 4479, "total_steps": 5364, "loss": 0.0011923732236027718, "lr": 1.617018229464894e-06, "epoch": 2.505174825174825, "percentage": 83.5, "elapsed_time": "8:55:11", "remaining_time": "1:45:44"} +{"current_steps": 4480, "total_steps": 5364, "loss": 0.07638342678546906, "lr": 1.6134715607426732e-06, "epoch": 2.5057342657342656, "percentage": 83.52, "elapsed_time": "8:55:18", "remaining_time": "1:45:37"} +{"current_steps": 4481, "total_steps": 5364, "loss": 0.024365486577153206, "lr": 1.6099284444654006e-06, "epoch": 2.5062937062937065, "percentage": 83.54, "elapsed_time": "8:55:25", "remaining_time": "1:45:30"} +{"current_steps": 4482, "total_steps": 5364, "loss": 0.06167413666844368, "lr": 1.606388882133907e-06, "epoch": 2.506853146853147, "percentage": 83.56, "elapsed_time": "8:55:32", "remaining_time": "1:45:23"} +{"current_steps": 4483, "total_steps": 5364, "loss": 0.08631311357021332, "lr": 1.602852875247506e-06, "epoch": 2.5074125874125874, "percentage": 83.58, "elapsed_time": "8:55:39", "remaining_time": "1:45:16"} +{"current_steps": 4484, "total_steps": 5364, "loss": 0.10294310748577118, "lr": 1.5993204253040162e-06, "epoch": 2.507972027972028, "percentage": 83.59, "elapsed_time": "8:55:46", "remaining_time": "1:45:08"} +{"current_steps": 4485, "total_steps": 5364, "loss": 0.07146032899618149, "lr": 1.5957915337997464e-06, "epoch": 2.5085314685314684, "percentage": 83.61, "elapsed_time": "8:55:53", "remaining_time": "1:45:01"} +{"current_steps": 4486, "total_steps": 5364, "loss": 0.08983171731233597, "lr": 1.5922662022294987e-06, "epoch": 2.509090909090909, "percentage": 83.63, "elapsed_time": "8:56:00", "remaining_time": "1:44:54"} +{"current_steps": 4487, "total_steps": 5364, "loss": 0.10053353756666183, "lr": 1.5887444320865676e-06, "epoch": 2.5096503496503497, "percentage": 83.65, "elapsed_time": "8:56:08", "remaining_time": "1:44:47"} +{"current_steps": 4488, "total_steps": 5364, "loss": 0.2222030758857727, "lr": 1.5852262248627326e-06, "epoch": 2.51020979020979, "percentage": 83.67, "elapsed_time": "8:56:15", "remaining_time": "1:44:40"} +{"current_steps": 4489, "total_steps": 5364, "loss": 0.02491142228245735, "lr": 1.581711582048272e-06, "epoch": 2.5107692307692306, "percentage": 83.69, "elapsed_time": "8:56:22", "remaining_time": "1:44:32"} +{"current_steps": 4490, "total_steps": 5364, "loss": 0.24647054076194763, "lr": 1.5782005051319517e-06, "epoch": 2.5113286713286715, "percentage": 83.71, "elapsed_time": "8:56:28", "remaining_time": "1:44:25"} +{"current_steps": 4491, "total_steps": 5364, "loss": 0.1390518695116043, "lr": 1.574692995601027e-06, "epoch": 2.511888111888112, "percentage": 83.72, "elapsed_time": "8:56:36", "remaining_time": "1:44:18"} +{"current_steps": 4492, "total_steps": 5364, "loss": 0.04321552813053131, "lr": 1.5711890549412412e-06, "epoch": 2.5124475524475525, "percentage": 83.74, "elapsed_time": "8:56:43", "remaining_time": "1:44:11"} +{"current_steps": 4493, "total_steps": 5364, "loss": 0.08706653863191605, "lr": 1.56768868463683e-06, "epoch": 2.513006993006993, "percentage": 83.76, "elapsed_time": "8:56:50", "remaining_time": "1:44:04"} +{"current_steps": 4494, "total_steps": 5364, "loss": 0.044477079063653946, "lr": 1.5641918861705075e-06, "epoch": 2.5135664335664334, "percentage": 83.78, "elapsed_time": "8:56:57", "remaining_time": "1:43:57"} +{"current_steps": 4495, "total_steps": 5364, "loss": 0.09677645564079285, "lr": 1.5606986610234843e-06, "epoch": 2.514125874125874, "percentage": 83.8, "elapsed_time": "8:57:05", "remaining_time": "1:43:49"} +{"current_steps": 4496, "total_steps": 5364, "loss": 0.07995382696390152, "lr": 1.5572090106754511e-06, "epoch": 2.5146853146853148, "percentage": 83.82, "elapsed_time": "8:57:12", "remaining_time": "1:43:42"} +{"current_steps": 4497, "total_steps": 5364, "loss": 0.11393406242132187, "lr": 1.5537229366045903e-06, "epoch": 2.515244755244755, "percentage": 83.84, "elapsed_time": "8:57:19", "remaining_time": "1:43:35"} +{"current_steps": 4498, "total_steps": 5364, "loss": 0.04904080927371979, "lr": 1.550240440287566e-06, "epoch": 2.5158041958041957, "percentage": 83.86, "elapsed_time": "8:57:26", "remaining_time": "1:43:28"} +{"current_steps": 4499, "total_steps": 5364, "loss": 0.03350922837853432, "lr": 1.5467615231995237e-06, "epoch": 2.5163636363636366, "percentage": 83.87, "elapsed_time": "8:57:33", "remaining_time": "1:43:21"} +{"current_steps": 4500, "total_steps": 5364, "loss": 0.04065831005573273, "lr": 1.543286186814097e-06, "epoch": 2.516923076923077, "percentage": 83.89, "elapsed_time": "8:57:40", "remaining_time": "1:43:14"} +{"current_steps": 4501, "total_steps": 5364, "loss": 0.021185962483286858, "lr": 1.5398144326034027e-06, "epoch": 2.5174825174825175, "percentage": 83.91, "elapsed_time": "8:58:15", "remaining_time": "1:43:12"} +{"current_steps": 4502, "total_steps": 5364, "loss": 0.022791506722569466, "lr": 1.5363462620380387e-06, "epoch": 2.518041958041958, "percentage": 83.93, "elapsed_time": "8:58:23", "remaining_time": "1:43:05"} +{"current_steps": 4503, "total_steps": 5364, "loss": 0.013553442433476448, "lr": 1.5328816765870891e-06, "epoch": 2.5186013986013984, "percentage": 83.95, "elapsed_time": "8:58:30", "remaining_time": "1:42:57"} +{"current_steps": 4504, "total_steps": 5364, "loss": 0.04495007544755936, "lr": 1.5294206777181087e-06, "epoch": 2.5191608391608393, "percentage": 83.97, "elapsed_time": "8:58:37", "remaining_time": "1:42:50"} +{"current_steps": 4505, "total_steps": 5364, "loss": 0.014980919659137726, "lr": 1.5259632668971447e-06, "epoch": 2.51972027972028, "percentage": 83.99, "elapsed_time": "8:58:44", "remaining_time": "1:42:43"} +{"current_steps": 4506, "total_steps": 5364, "loss": 0.10430073738098145, "lr": 1.5225094455887202e-06, "epoch": 2.5202797202797202, "percentage": 84.0, "elapsed_time": "8:58:52", "remaining_time": "1:42:36"} +{"current_steps": 4507, "total_steps": 5364, "loss": 0.2103475034236908, "lr": 1.5190592152558314e-06, "epoch": 2.5208391608391607, "percentage": 84.02, "elapsed_time": "8:58:59", "remaining_time": "1:42:29"} +{"current_steps": 4508, "total_steps": 5364, "loss": 0.03322084993124008, "lr": 1.5156125773599694e-06, "epoch": 2.5213986013986016, "percentage": 84.04, "elapsed_time": "8:59:06", "remaining_time": "1:42:22"} +{"current_steps": 4509, "total_steps": 5364, "loss": 0.04260168969631195, "lr": 1.5121695333610864e-06, "epoch": 2.521958041958042, "percentage": 84.06, "elapsed_time": "8:59:13", "remaining_time": "1:42:14"} +{"current_steps": 4510, "total_steps": 5364, "loss": 0.1739736944437027, "lr": 1.5087300847176213e-06, "epoch": 2.5225174825174825, "percentage": 84.08, "elapsed_time": "8:59:20", "remaining_time": "1:42:07"} +{"current_steps": 4511, "total_steps": 5364, "loss": 0.16299282014369965, "lr": 1.5052942328864905e-06, "epoch": 2.523076923076923, "percentage": 84.1, "elapsed_time": "8:59:27", "remaining_time": "1:42:00"} +{"current_steps": 4512, "total_steps": 5364, "loss": 0.16750100255012512, "lr": 1.5018619793230783e-06, "epoch": 2.5236363636363635, "percentage": 84.12, "elapsed_time": "8:59:34", "remaining_time": "1:41:53"} +{"current_steps": 4513, "total_steps": 5364, "loss": 0.09584980458021164, "lr": 1.49843332548126e-06, "epoch": 2.5241958041958044, "percentage": 84.13, "elapsed_time": "8:59:41", "remaining_time": "1:41:46"} +{"current_steps": 4514, "total_steps": 5364, "loss": 0.06112157553434372, "lr": 1.4950082728133698e-06, "epoch": 2.524755244755245, "percentage": 84.15, "elapsed_time": "8:59:48", "remaining_time": "1:41:38"} +{"current_steps": 4515, "total_steps": 5364, "loss": 0.0299077108502388, "lr": 1.4915868227702268e-06, "epoch": 2.5253146853146853, "percentage": 84.17, "elapsed_time": "8:59:56", "remaining_time": "1:41:31"} +{"current_steps": 4516, "total_steps": 5364, "loss": 0.14305603504180908, "lr": 1.4881689768011232e-06, "epoch": 2.5258741258741257, "percentage": 84.19, "elapsed_time": "9:00:03", "remaining_time": "1:41:24"} +{"current_steps": 4517, "total_steps": 5364, "loss": 0.021605102345347404, "lr": 1.484754736353815e-06, "epoch": 2.5264335664335666, "percentage": 84.21, "elapsed_time": "9:00:10", "remaining_time": "1:41:17"} +{"current_steps": 4518, "total_steps": 5364, "loss": 0.02200378105044365, "lr": 1.481344102874549e-06, "epoch": 2.526993006993007, "percentage": 84.23, "elapsed_time": "9:00:18", "remaining_time": "1:41:10"} +{"current_steps": 4519, "total_steps": 5364, "loss": 0.1374070793390274, "lr": 1.477937077808026e-06, "epoch": 2.5275524475524476, "percentage": 84.25, "elapsed_time": "9:00:25", "remaining_time": "1:41:03"} +{"current_steps": 4520, "total_steps": 5364, "loss": 0.034848652780056, "lr": 1.4745336625974282e-06, "epoch": 2.528111888111888, "percentage": 84.27, "elapsed_time": "9:00:32", "remaining_time": "1:40:55"} +{"current_steps": 4521, "total_steps": 5364, "loss": 0.05923225358128548, "lr": 1.4711338586844081e-06, "epoch": 2.5286713286713285, "percentage": 84.28, "elapsed_time": "9:00:39", "remaining_time": "1:40:48"} +{"current_steps": 4522, "total_steps": 5364, "loss": 0.06846512109041214, "lr": 1.4677376675090805e-06, "epoch": 2.5292307692307694, "percentage": 84.3, "elapsed_time": "9:00:46", "remaining_time": "1:40:41"} +{"current_steps": 4523, "total_steps": 5364, "loss": 0.15426114201545715, "lr": 1.4643450905100453e-06, "epoch": 2.52979020979021, "percentage": 84.32, "elapsed_time": "9:00:53", "remaining_time": "1:40:34"} +{"current_steps": 4524, "total_steps": 5364, "loss": 0.04837312549352646, "lr": 1.4609561291243547e-06, "epoch": 2.5303496503496503, "percentage": 84.34, "elapsed_time": "9:01:00", "remaining_time": "1:40:27"} +{"current_steps": 4525, "total_steps": 5364, "loss": 0.0876699835062027, "lr": 1.4575707847875398e-06, "epoch": 2.5309090909090908, "percentage": 84.36, "elapsed_time": "9:01:08", "remaining_time": "1:40:20"} +{"current_steps": 4526, "total_steps": 5364, "loss": 0.055250100791454315, "lr": 1.4541890589335973e-06, "epoch": 2.5314685314685317, "percentage": 84.38, "elapsed_time": "9:01:15", "remaining_time": "1:40:12"} +{"current_steps": 4527, "total_steps": 5364, "loss": 0.17377258837223053, "lr": 1.4508109529949898e-06, "epoch": 2.532027972027972, "percentage": 84.4, "elapsed_time": "9:01:22", "remaining_time": "1:40:05"} +{"current_steps": 4528, "total_steps": 5364, "loss": 0.07808820903301239, "lr": 1.4474364684026499e-06, "epoch": 2.5325874125874126, "percentage": 84.41, "elapsed_time": "9:01:29", "remaining_time": "1:39:58"} +{"current_steps": 4529, "total_steps": 5364, "loss": 0.0693928599357605, "lr": 1.4440656065859694e-06, "epoch": 2.533146853146853, "percentage": 84.43, "elapsed_time": "9:01:36", "remaining_time": "1:39:51"} +{"current_steps": 4530, "total_steps": 5364, "loss": 0.14336681365966797, "lr": 1.4406983689728117e-06, "epoch": 2.5337062937062935, "percentage": 84.45, "elapsed_time": "9:01:43", "remaining_time": "1:39:44"} +{"current_steps": 4531, "total_steps": 5364, "loss": 0.08404998481273651, "lr": 1.4373347569895025e-06, "epoch": 2.5342657342657344, "percentage": 84.47, "elapsed_time": "9:01:50", "remaining_time": "1:39:36"} +{"current_steps": 4532, "total_steps": 5364, "loss": 0.09068721532821655, "lr": 1.4339747720608355e-06, "epoch": 2.534825174825175, "percentage": 84.49, "elapsed_time": "9:01:57", "remaining_time": "1:39:29"} +{"current_steps": 4533, "total_steps": 5364, "loss": 0.044285841286182404, "lr": 1.43061841561006e-06, "epoch": 2.5353846153846153, "percentage": 84.51, "elapsed_time": "9:02:04", "remaining_time": "1:39:22"} +{"current_steps": 4534, "total_steps": 5364, "loss": 0.004923766478896141, "lr": 1.4272656890588966e-06, "epoch": 2.535944055944056, "percentage": 84.53, "elapsed_time": "9:02:12", "remaining_time": "1:39:15"} +{"current_steps": 4535, "total_steps": 5364, "loss": 0.019866883754730225, "lr": 1.4239165938275224e-06, "epoch": 2.5365034965034967, "percentage": 84.55, "elapsed_time": "9:02:19", "remaining_time": "1:39:08"} +{"current_steps": 4536, "total_steps": 5364, "loss": 0.0585157610476017, "lr": 1.4205711313345816e-06, "epoch": 2.537062937062937, "percentage": 84.56, "elapsed_time": "9:02:26", "remaining_time": "1:39:01"} +{"current_steps": 4537, "total_steps": 5364, "loss": 0.018409237265586853, "lr": 1.4172293029971763e-06, "epoch": 2.5376223776223776, "percentage": 84.58, "elapsed_time": "9:02:33", "remaining_time": "1:38:53"} +{"current_steps": 4538, "total_steps": 5364, "loss": 0.09511364996433258, "lr": 1.4138911102308672e-06, "epoch": 2.538181818181818, "percentage": 84.6, "elapsed_time": "9:02:40", "remaining_time": "1:38:46"} +{"current_steps": 4539, "total_steps": 5364, "loss": 0.16395479440689087, "lr": 1.4105565544496803e-06, "epoch": 2.5387412587412586, "percentage": 84.62, "elapsed_time": "9:02:48", "remaining_time": "1:38:39"} +{"current_steps": 4540, "total_steps": 5364, "loss": 0.1138661801815033, "lr": 1.4072256370660952e-06, "epoch": 2.5393006993006995, "percentage": 84.64, "elapsed_time": "9:02:55", "remaining_time": "1:38:32"} +{"current_steps": 4541, "total_steps": 5364, "loss": 0.043411001563072205, "lr": 1.403898359491057e-06, "epoch": 2.53986013986014, "percentage": 84.66, "elapsed_time": "9:03:02", "remaining_time": "1:38:25"} +{"current_steps": 4542, "total_steps": 5364, "loss": 0.06683079898357391, "lr": 1.4005747231339651e-06, "epoch": 2.5404195804195804, "percentage": 84.68, "elapsed_time": "9:03:09", "remaining_time": "1:38:17"} +{"current_steps": 4543, "total_steps": 5364, "loss": 0.07385344058275223, "lr": 1.3972547294026728e-06, "epoch": 2.540979020979021, "percentage": 84.69, "elapsed_time": "9:03:16", "remaining_time": "1:38:10"} +{"current_steps": 4544, "total_steps": 5364, "loss": 0.16887669265270233, "lr": 1.3939383797034978e-06, "epoch": 2.5415384615384617, "percentage": 84.71, "elapsed_time": "9:03:24", "remaining_time": "1:38:03"} +{"current_steps": 4545, "total_steps": 5364, "loss": 0.1192508190870285, "lr": 1.3906256754412084e-06, "epoch": 2.542097902097902, "percentage": 84.73, "elapsed_time": "9:03:31", "remaining_time": "1:37:56"} +{"current_steps": 4546, "total_steps": 5364, "loss": 0.13044364750385284, "lr": 1.3873166180190345e-06, "epoch": 2.5426573426573427, "percentage": 84.75, "elapsed_time": "9:03:38", "remaining_time": "1:37:49"} +{"current_steps": 4547, "total_steps": 5364, "loss": 0.11944424360990524, "lr": 1.384011208838657e-06, "epoch": 2.543216783216783, "percentage": 84.77, "elapsed_time": "9:03:45", "remaining_time": "1:37:42"} +{"current_steps": 4548, "total_steps": 5364, "loss": 0.05917758494615555, "lr": 1.3807094493002082e-06, "epoch": 2.5437762237762236, "percentage": 84.79, "elapsed_time": "9:03:51", "remaining_time": "1:37:34"} +{"current_steps": 4549, "total_steps": 5364, "loss": 0.09083046764135361, "lr": 1.377411340802286e-06, "epoch": 2.5443356643356645, "percentage": 84.81, "elapsed_time": "9:03:58", "remaining_time": "1:37:27"} +{"current_steps": 4550, "total_steps": 5364, "loss": 0.05486498028039932, "lr": 1.3741168847419284e-06, "epoch": 2.544895104895105, "percentage": 84.82, "elapsed_time": "9:04:05", "remaining_time": "1:37:20"} +{"current_steps": 4551, "total_steps": 5364, "loss": 0.029036281630396843, "lr": 1.3708260825146335e-06, "epoch": 2.5454545454545454, "percentage": 84.84, "elapsed_time": "9:04:12", "remaining_time": "1:37:13"} +{"current_steps": 4552, "total_steps": 5364, "loss": 0.11576436460018158, "lr": 1.367538935514353e-06, "epoch": 2.546013986013986, "percentage": 84.86, "elapsed_time": "9:04:20", "remaining_time": "1:37:06"} +{"current_steps": 4553, "total_steps": 5364, "loss": 0.21533727645874023, "lr": 1.364255445133481e-06, "epoch": 2.546573426573427, "percentage": 84.88, "elapsed_time": "9:04:27", "remaining_time": "1:36:58"} +{"current_steps": 4554, "total_steps": 5364, "loss": 0.07367813587188721, "lr": 1.3609756127628792e-06, "epoch": 2.5471328671328672, "percentage": 84.9, "elapsed_time": "9:04:34", "remaining_time": "1:36:51"} +{"current_steps": 4555, "total_steps": 5364, "loss": 0.0598619319498539, "lr": 1.3576994397918453e-06, "epoch": 2.5476923076923077, "percentage": 84.92, "elapsed_time": "9:04:41", "remaining_time": "1:36:44"} +{"current_steps": 4556, "total_steps": 5364, "loss": 0.1191563829779625, "lr": 1.3544269276081257e-06, "epoch": 2.548251748251748, "percentage": 84.94, "elapsed_time": "9:04:48", "remaining_time": "1:36:37"} +{"current_steps": 4557, "total_steps": 5364, "loss": 0.0805964469909668, "lr": 1.3511580775979328e-06, "epoch": 2.5488111888111886, "percentage": 84.96, "elapsed_time": "9:04:55", "remaining_time": "1:36:30"} +{"current_steps": 4558, "total_steps": 5364, "loss": 0.06912267208099365, "lr": 1.3478928911459088e-06, "epoch": 2.5493706293706295, "percentage": 84.97, "elapsed_time": "9:05:02", "remaining_time": "1:36:22"} +{"current_steps": 4559, "total_steps": 5364, "loss": 0.06492966413497925, "lr": 1.3446313696351542e-06, "epoch": 2.54993006993007, "percentage": 84.99, "elapsed_time": "9:05:09", "remaining_time": "1:36:15"} +{"current_steps": 4560, "total_steps": 5364, "loss": 0.05315789580345154, "lr": 1.3413735144472195e-06, "epoch": 2.5504895104895104, "percentage": 85.01, "elapsed_time": "9:05:16", "remaining_time": "1:36:08"} +{"current_steps": 4561, "total_steps": 5364, "loss": 0.07461623102426529, "lr": 1.3381193269620896e-06, "epoch": 2.551048951048951, "percentage": 85.03, "elapsed_time": "9:05:23", "remaining_time": "1:36:01"} +{"current_steps": 4562, "total_steps": 5364, "loss": 0.16163872182369232, "lr": 1.334868808558213e-06, "epoch": 2.551608391608392, "percentage": 85.05, "elapsed_time": "9:05:30", "remaining_time": "1:35:54"} +{"current_steps": 4563, "total_steps": 5364, "loss": 0.013561589643359184, "lr": 1.331621960612468e-06, "epoch": 2.5521678321678323, "percentage": 85.07, "elapsed_time": "9:05:38", "remaining_time": "1:35:46"} +{"current_steps": 4564, "total_steps": 5364, "loss": 0.08386319130659103, "lr": 1.3283787845001884e-06, "epoch": 2.5527272727272727, "percentage": 85.09, "elapsed_time": "9:05:45", "remaining_time": "1:35:39"} +{"current_steps": 4565, "total_steps": 5364, "loss": 0.12511901557445526, "lr": 1.3251392815951481e-06, "epoch": 2.553286713286713, "percentage": 85.1, "elapsed_time": "9:05:52", "remaining_time": "1:35:32"} +{"current_steps": 4566, "total_steps": 5364, "loss": 0.1285950243473053, "lr": 1.3219034532695663e-06, "epoch": 2.5538461538461537, "percentage": 85.12, "elapsed_time": "9:05:59", "remaining_time": "1:35:25"} +{"current_steps": 4567, "total_steps": 5364, "loss": 0.008102228865027428, "lr": 1.3186713008941098e-06, "epoch": 2.5544055944055946, "percentage": 85.14, "elapsed_time": "9:06:06", "remaining_time": "1:35:18"} +{"current_steps": 4568, "total_steps": 5364, "loss": 0.058629631996154785, "lr": 1.3154428258378782e-06, "epoch": 2.554965034965035, "percentage": 85.16, "elapsed_time": "9:06:13", "remaining_time": "1:35:10"} +{"current_steps": 4569, "total_steps": 5364, "loss": 0.042920537292957306, "lr": 1.3122180294684217e-06, "epoch": 2.5555244755244755, "percentage": 85.18, "elapsed_time": "9:06:19", "remaining_time": "1:35:03"} +{"current_steps": 4570, "total_steps": 5364, "loss": 0.09765076637268066, "lr": 1.3089969131517311e-06, "epoch": 2.556083916083916, "percentage": 85.2, "elapsed_time": "9:06:26", "remaining_time": "1:34:56"} +{"current_steps": 4571, "total_steps": 5364, "loss": 0.10050079971551895, "lr": 1.305779478252236e-06, "epoch": 2.556643356643357, "percentage": 85.22, "elapsed_time": "9:06:33", "remaining_time": "1:34:49"} +{"current_steps": 4572, "total_steps": 5364, "loss": 0.17783184349536896, "lr": 1.3025657261328105e-06, "epoch": 2.5572027972027973, "percentage": 85.23, "elapsed_time": "9:06:41", "remaining_time": "1:34:42"} +{"current_steps": 4573, "total_steps": 5364, "loss": 0.0397367961704731, "lr": 1.2993556581547627e-06, "epoch": 2.5577622377622378, "percentage": 85.25, "elapsed_time": "9:06:48", "remaining_time": "1:34:34"} +{"current_steps": 4574, "total_steps": 5364, "loss": 0.18189845979213715, "lr": 1.2961492756778437e-06, "epoch": 2.5583216783216782, "percentage": 85.27, "elapsed_time": "9:06:55", "remaining_time": "1:34:27"} +{"current_steps": 4575, "total_steps": 5364, "loss": 0.11143840849399567, "lr": 1.2929465800602458e-06, "epoch": 2.5588811188811187, "percentage": 85.29, "elapsed_time": "9:07:03", "remaining_time": "1:34:20"} +{"current_steps": 4576, "total_steps": 5364, "loss": 0.10607460141181946, "lr": 1.2897475726585951e-06, "epoch": 2.5594405594405596, "percentage": 85.31, "elapsed_time": "9:07:10", "remaining_time": "1:34:13"} +{"current_steps": 4577, "total_steps": 5364, "loss": 0.21376782655715942, "lr": 1.2865522548279596e-06, "epoch": 2.56, "percentage": 85.33, "elapsed_time": "9:07:18", "remaining_time": "1:34:06"} +{"current_steps": 4578, "total_steps": 5364, "loss": 0.04318004474043846, "lr": 1.2833606279218392e-06, "epoch": 2.5605594405594405, "percentage": 85.35, "elapsed_time": "9:07:25", "remaining_time": "1:33:59"} +{"current_steps": 4579, "total_steps": 5364, "loss": 0.09065401554107666, "lr": 1.2801726932921743e-06, "epoch": 2.561118881118881, "percentage": 85.37, "elapsed_time": "9:07:32", "remaining_time": "1:33:52"} +{"current_steps": 4580, "total_steps": 5364, "loss": 0.06432337313890457, "lr": 1.2769884522893406e-06, "epoch": 2.561678321678322, "percentage": 85.38, "elapsed_time": "9:07:39", "remaining_time": "1:33:44"} +{"current_steps": 4581, "total_steps": 5364, "loss": 0.0363226979970932, "lr": 1.2738079062621523e-06, "epoch": 2.5622377622377623, "percentage": 85.4, "elapsed_time": "9:07:46", "remaining_time": "1:33:37"} +{"current_steps": 4582, "total_steps": 5364, "loss": 0.08044242858886719, "lr": 1.2706310565578495e-06, "epoch": 2.562797202797203, "percentage": 85.42, "elapsed_time": "9:07:53", "remaining_time": "1:33:30"} +{"current_steps": 4583, "total_steps": 5364, "loss": 0.06646215915679932, "lr": 1.2674579045221146e-06, "epoch": 2.5633566433566433, "percentage": 85.44, "elapsed_time": "9:08:00", "remaining_time": "1:33:23"} +{"current_steps": 4584, "total_steps": 5364, "loss": 0.09292811155319214, "lr": 1.2642884514990617e-06, "epoch": 2.5639160839160837, "percentage": 85.46, "elapsed_time": "9:08:08", "remaining_time": "1:33:16"} +{"current_steps": 4585, "total_steps": 5364, "loss": 0.03904930502176285, "lr": 1.2611226988312365e-06, "epoch": 2.5644755244755246, "percentage": 85.48, "elapsed_time": "9:08:15", "remaining_time": "1:33:08"} +{"current_steps": 4586, "total_steps": 5364, "loss": 0.0455571785569191, "lr": 1.257960647859623e-06, "epoch": 2.565034965034965, "percentage": 85.5, "elapsed_time": "9:08:22", "remaining_time": "1:33:01"} +{"current_steps": 4587, "total_steps": 5364, "loss": 0.11863440275192261, "lr": 1.2548022999236242e-06, "epoch": 2.5655944055944055, "percentage": 85.51, "elapsed_time": "9:08:29", "remaining_time": "1:32:54"} +{"current_steps": 4588, "total_steps": 5364, "loss": 0.09104831516742706, "lr": 1.2516476563610913e-06, "epoch": 2.566153846153846, "percentage": 85.53, "elapsed_time": "9:08:36", "remaining_time": "1:32:47"} +{"current_steps": 4589, "total_steps": 5364, "loss": 0.10867291688919067, "lr": 1.248496718508293e-06, "epoch": 2.566713286713287, "percentage": 85.55, "elapsed_time": "9:08:43", "remaining_time": "1:32:40"} +{"current_steps": 4590, "total_steps": 5364, "loss": 0.09799507260322571, "lr": 1.245349487699935e-06, "epoch": 2.5672727272727274, "percentage": 85.57, "elapsed_time": "9:08:50", "remaining_time": "1:32:32"} +{"current_steps": 4591, "total_steps": 5364, "loss": 0.08061123639345169, "lr": 1.242205965269152e-06, "epoch": 2.567832167832168, "percentage": 85.59, "elapsed_time": "9:08:57", "remaining_time": "1:32:25"} +{"current_steps": 4592, "total_steps": 5364, "loss": 0.0685778334736824, "lr": 1.2390661525475024e-06, "epoch": 2.5683916083916083, "percentage": 85.61, "elapsed_time": "9:09:04", "remaining_time": "1:32:18"} +{"current_steps": 4593, "total_steps": 5364, "loss": 0.03397190943360329, "lr": 1.2359300508649852e-06, "epoch": 2.5689510489510488, "percentage": 85.63, "elapsed_time": "9:09:11", "remaining_time": "1:32:11"} +{"current_steps": 4594, "total_steps": 5364, "loss": 0.0748581662774086, "lr": 1.2327976615500136e-06, "epoch": 2.5695104895104897, "percentage": 85.65, "elapsed_time": "9:09:18", "remaining_time": "1:32:04"} +{"current_steps": 4595, "total_steps": 5364, "loss": 0.08978810906410217, "lr": 1.2296689859294375e-06, "epoch": 2.57006993006993, "percentage": 85.66, "elapsed_time": "9:09:25", "remaining_time": "1:31:56"} +{"current_steps": 4596, "total_steps": 5364, "loss": 0.1846337616443634, "lr": 1.226544025328532e-06, "epoch": 2.5706293706293706, "percentage": 85.68, "elapsed_time": "9:09:32", "remaining_time": "1:31:49"} +{"current_steps": 4597, "total_steps": 5364, "loss": 0.17398908734321594, "lr": 1.223422781070992e-06, "epoch": 2.571188811188811, "percentage": 85.7, "elapsed_time": "9:09:39", "remaining_time": "1:31:42"} +{"current_steps": 4598, "total_steps": 5364, "loss": 0.09944969415664673, "lr": 1.2203052544789518e-06, "epoch": 2.571748251748252, "percentage": 85.72, "elapsed_time": "9:09:46", "remaining_time": "1:31:35"} +{"current_steps": 4599, "total_steps": 5364, "loss": 0.059454552829265594, "lr": 1.2171914468729552e-06, "epoch": 2.5723076923076924, "percentage": 85.74, "elapsed_time": "9:09:53", "remaining_time": "1:31:28"} +{"current_steps": 4600, "total_steps": 5364, "loss": 0.12345708906650543, "lr": 1.2140813595719825e-06, "epoch": 2.572867132867133, "percentage": 85.76, "elapsed_time": "9:10:00", "remaining_time": "1:31:21"} +{"current_steps": 4601, "total_steps": 5364, "loss": 0.06571106612682343, "lr": 1.2109749938934346e-06, "epoch": 2.5734265734265733, "percentage": 85.78, "elapsed_time": "9:10:07", "remaining_time": "1:31:13"} +{"current_steps": 4602, "total_steps": 5364, "loss": 0.17944902181625366, "lr": 1.2078723511531288e-06, "epoch": 2.573986013986014, "percentage": 85.79, "elapsed_time": "9:10:14", "remaining_time": "1:31:06"} +{"current_steps": 4603, "total_steps": 5364, "loss": 0.06039018556475639, "lr": 1.2047734326653205e-06, "epoch": 2.5745454545454547, "percentage": 85.81, "elapsed_time": "9:10:22", "remaining_time": "1:30:59"} +{"current_steps": 4604, "total_steps": 5364, "loss": 0.15344257652759552, "lr": 1.201678239742673e-06, "epoch": 2.575104895104895, "percentage": 85.83, "elapsed_time": "9:10:29", "remaining_time": "1:30:52"} +{"current_steps": 4605, "total_steps": 5364, "loss": 0.1656726747751236, "lr": 1.1985867736962798e-06, "epoch": 2.5756643356643356, "percentage": 85.85, "elapsed_time": "9:10:36", "remaining_time": "1:30:45"} +{"current_steps": 4606, "total_steps": 5364, "loss": 0.05471505597233772, "lr": 1.1954990358356533e-06, "epoch": 2.576223776223776, "percentage": 85.87, "elapsed_time": "9:10:43", "remaining_time": "1:30:37"} +{"current_steps": 4607, "total_steps": 5364, "loss": 0.008323697373270988, "lr": 1.1924150274687252e-06, "epoch": 2.576783216783217, "percentage": 85.89, "elapsed_time": "9:10:51", "remaining_time": "1:30:30"} +{"current_steps": 4608, "total_steps": 5364, "loss": 0.055867210030555725, "lr": 1.1893347499018493e-06, "epoch": 2.5773426573426574, "percentage": 85.91, "elapsed_time": "9:10:58", "remaining_time": "1:30:23"} +{"current_steps": 4609, "total_steps": 5364, "loss": 0.04744353145360947, "lr": 1.1862582044397985e-06, "epoch": 2.577902097902098, "percentage": 85.92, "elapsed_time": "9:11:05", "remaining_time": "1:30:16"} +{"current_steps": 4610, "total_steps": 5364, "loss": 0.03727944940328598, "lr": 1.183185392385766e-06, "epoch": 2.5784615384615384, "percentage": 85.94, "elapsed_time": "9:11:12", "remaining_time": "1:30:09"} +{"current_steps": 4611, "total_steps": 5364, "loss": 0.06568647921085358, "lr": 1.1801163150413652e-06, "epoch": 2.579020979020979, "percentage": 85.96, "elapsed_time": "9:11:19", "remaining_time": "1:30:02"} +{"current_steps": 4612, "total_steps": 5364, "loss": 0.056148070842027664, "lr": 1.1770509737066193e-06, "epoch": 2.5795804195804197, "percentage": 85.98, "elapsed_time": "9:11:26", "remaining_time": "1:29:54"} +{"current_steps": 4613, "total_steps": 5364, "loss": 0.08955402672290802, "lr": 1.1739893696799764e-06, "epoch": 2.58013986013986, "percentage": 86.0, "elapsed_time": "9:11:33", "remaining_time": "1:29:47"} +{"current_steps": 4614, "total_steps": 5364, "loss": 0.08713246881961823, "lr": 1.1709315042583002e-06, "epoch": 2.5806993006993006, "percentage": 86.02, "elapsed_time": "9:11:41", "remaining_time": "1:29:40"} +{"current_steps": 4615, "total_steps": 5364, "loss": 0.1412111520767212, "lr": 1.1678773787368714e-06, "epoch": 2.581258741258741, "percentage": 86.04, "elapsed_time": "9:11:48", "remaining_time": "1:29:33"} +{"current_steps": 4616, "total_steps": 5364, "loss": 0.16250455379486084, "lr": 1.164826994409386e-06, "epoch": 2.581818181818182, "percentage": 86.06, "elapsed_time": "9:11:55", "remaining_time": "1:29:26"} +{"current_steps": 4617, "total_steps": 5364, "loss": 0.13743548095226288, "lr": 1.1617803525679505e-06, "epoch": 2.5823776223776225, "percentage": 86.07, "elapsed_time": "9:12:02", "remaining_time": "1:29:18"} +{"current_steps": 4618, "total_steps": 5364, "loss": 0.08454421162605286, "lr": 1.1587374545030926e-06, "epoch": 2.582937062937063, "percentage": 86.09, "elapsed_time": "9:12:09", "remaining_time": "1:29:11"} +{"current_steps": 4619, "total_steps": 5364, "loss": 0.028263267129659653, "lr": 1.1556983015037515e-06, "epoch": 2.5834965034965034, "percentage": 86.11, "elapsed_time": "9:12:16", "remaining_time": "1:29:04"} +{"current_steps": 4620, "total_steps": 5364, "loss": 0.0640464574098587, "lr": 1.1526628948572815e-06, "epoch": 2.584055944055944, "percentage": 86.13, "elapsed_time": "9:12:23", "remaining_time": "1:28:57"} +{"current_steps": 4621, "total_steps": 5364, "loss": 0.09115159511566162, "lr": 1.1496312358494477e-06, "epoch": 2.5846153846153848, "percentage": 86.15, "elapsed_time": "9:12:30", "remaining_time": "1:28:50"} +{"current_steps": 4622, "total_steps": 5364, "loss": 0.08466363698244095, "lr": 1.1466033257644316e-06, "epoch": 2.585174825174825, "percentage": 86.17, "elapsed_time": "9:12:38", "remaining_time": "1:28:43"} +{"current_steps": 4623, "total_steps": 5364, "loss": 0.08374187350273132, "lr": 1.1435791658848194e-06, "epoch": 2.5857342657342657, "percentage": 86.19, "elapsed_time": "9:12:45", "remaining_time": "1:28:35"} +{"current_steps": 4624, "total_steps": 5364, "loss": 0.04049751162528992, "lr": 1.1405587574916145e-06, "epoch": 2.586293706293706, "percentage": 86.2, "elapsed_time": "9:12:52", "remaining_time": "1:28:28"} +{"current_steps": 4625, "total_steps": 5364, "loss": 0.07686944305896759, "lr": 1.1375421018642318e-06, "epoch": 2.586853146853147, "percentage": 86.22, "elapsed_time": "9:12:59", "remaining_time": "1:28:21"} +{"current_steps": 4626, "total_steps": 5364, "loss": 0.12538719177246094, "lr": 1.1345292002804943e-06, "epoch": 2.5874125874125875, "percentage": 86.24, "elapsed_time": "9:13:06", "remaining_time": "1:28:14"} +{"current_steps": 4627, "total_steps": 5364, "loss": 0.06467457860708237, "lr": 1.131520054016636e-06, "epoch": 2.587972027972028, "percentage": 86.26, "elapsed_time": "9:13:13", "remaining_time": "1:28:07"} +{"current_steps": 4628, "total_steps": 5364, "loss": 0.12892964482307434, "lr": 1.1285146643472978e-06, "epoch": 2.5885314685314684, "percentage": 86.28, "elapsed_time": "9:13:21", "remaining_time": "1:28:00"} +{"current_steps": 4629, "total_steps": 5364, "loss": 0.033108122646808624, "lr": 1.125513032545531e-06, "epoch": 2.589090909090909, "percentage": 86.3, "elapsed_time": "9:13:28", "remaining_time": "1:27:52"} +{"current_steps": 4630, "total_steps": 5364, "loss": 0.09453903138637543, "lr": 1.1225151598827977e-06, "epoch": 2.58965034965035, "percentage": 86.32, "elapsed_time": "9:13:35", "remaining_time": "1:27:45"} +{"current_steps": 4631, "total_steps": 5364, "loss": 0.10552264004945755, "lr": 1.1195210476289597e-06, "epoch": 2.5902097902097903, "percentage": 86.33, "elapsed_time": "9:13:42", "remaining_time": "1:27:38"} +{"current_steps": 4632, "total_steps": 5364, "loss": 0.12797769904136658, "lr": 1.1165306970522982e-06, "epoch": 2.5907692307692307, "percentage": 86.35, "elapsed_time": "9:13:49", "remaining_time": "1:27:31"} +{"current_steps": 4633, "total_steps": 5364, "loss": 0.0609578937292099, "lr": 1.11354410941949e-06, "epoch": 2.591328671328671, "percentage": 86.37, "elapsed_time": "9:13:57", "remaining_time": "1:27:24"} +{"current_steps": 4634, "total_steps": 5364, "loss": 0.10750634223222733, "lr": 1.1105612859956227e-06, "epoch": 2.591888111888112, "percentage": 86.39, "elapsed_time": "9:14:04", "remaining_time": "1:27:16"} +{"current_steps": 4635, "total_steps": 5364, "loss": 0.08289151638746262, "lr": 1.1075822280441907e-06, "epoch": 2.5924475524475525, "percentage": 86.41, "elapsed_time": "9:14:11", "remaining_time": "1:27:09"} +{"current_steps": 4636, "total_steps": 5364, "loss": 0.08348525315523148, "lr": 1.1046069368270861e-06, "epoch": 2.593006993006993, "percentage": 86.43, "elapsed_time": "9:14:18", "remaining_time": "1:27:02"} +{"current_steps": 4637, "total_steps": 5364, "loss": 0.006321026012301445, "lr": 1.1016354136046193e-06, "epoch": 2.5935664335664335, "percentage": 86.45, "elapsed_time": "9:14:25", "remaining_time": "1:26:55"} +{"current_steps": 4638, "total_steps": 5364, "loss": 0.03064294531941414, "lr": 1.0986676596354884e-06, "epoch": 2.594125874125874, "percentage": 86.47, "elapsed_time": "9:14:33", "remaining_time": "1:26:48"} +{"current_steps": 4639, "total_steps": 5364, "loss": 0.14625415205955505, "lr": 1.095703676176807e-06, "epoch": 2.594685314685315, "percentage": 86.48, "elapsed_time": "9:14:40", "remaining_time": "1:26:41"} +{"current_steps": 4640, "total_steps": 5364, "loss": 0.04113004729151726, "lr": 1.0927434644840885e-06, "epoch": 2.5952447552447553, "percentage": 86.5, "elapsed_time": "9:14:47", "remaining_time": "1:26:34"} +{"current_steps": 4641, "total_steps": 5364, "loss": 0.05982401221990585, "lr": 1.0897870258112396e-06, "epoch": 2.5958041958041957, "percentage": 86.52, "elapsed_time": "9:14:55", "remaining_time": "1:26:26"} +{"current_steps": 4642, "total_steps": 5364, "loss": 0.0202775988727808, "lr": 1.0868343614105869e-06, "epoch": 2.596363636363636, "percentage": 86.54, "elapsed_time": "9:15:02", "remaining_time": "1:26:19"} +{"current_steps": 4643, "total_steps": 5364, "loss": 0.10561646521091461, "lr": 1.0838854725328396e-06, "epoch": 2.596923076923077, "percentage": 86.56, "elapsed_time": "9:15:09", "remaining_time": "1:26:12"} +{"current_steps": 4644, "total_steps": 5364, "loss": 0.04724736139178276, "lr": 1.0809403604271208e-06, "epoch": 2.5974825174825176, "percentage": 86.58, "elapsed_time": "9:15:16", "remaining_time": "1:26:05"} +{"current_steps": 4645, "total_steps": 5364, "loss": 0.10748960822820663, "lr": 1.0779990263409479e-06, "epoch": 2.598041958041958, "percentage": 86.6, "elapsed_time": "9:15:23", "remaining_time": "1:25:58"} +{"current_steps": 4646, "total_steps": 5364, "loss": 0.10508938133716583, "lr": 1.0750614715202345e-06, "epoch": 2.5986013986013985, "percentage": 86.61, "elapsed_time": "9:15:30", "remaining_time": "1:25:51"} +{"current_steps": 4647, "total_steps": 5364, "loss": 0.04325022175908089, "lr": 1.0721276972093064e-06, "epoch": 2.599160839160839, "percentage": 86.63, "elapsed_time": "9:15:38", "remaining_time": "1:25:43"} +{"current_steps": 4648, "total_steps": 5364, "loss": 0.07308591157197952, "lr": 1.0691977046508718e-06, "epoch": 2.59972027972028, "percentage": 86.65, "elapsed_time": "9:15:46", "remaining_time": "1:25:36"} +{"current_steps": 4649, "total_steps": 5364, "loss": 0.036036331206560135, "lr": 1.0662714950860487e-06, "epoch": 2.6002797202797203, "percentage": 86.67, "elapsed_time": "9:15:53", "remaining_time": "1:25:29"} +{"current_steps": 4650, "total_steps": 5364, "loss": 0.03203625977039337, "lr": 1.0633490697543481e-06, "epoch": 2.600839160839161, "percentage": 86.69, "elapsed_time": "9:16:00", "remaining_time": "1:25:22"} +{"current_steps": 4651, "total_steps": 5364, "loss": 0.015751425176858902, "lr": 1.060430429893674e-06, "epoch": 2.6013986013986012, "percentage": 86.71, "elapsed_time": "9:16:07", "remaining_time": "1:25:15"} +{"current_steps": 4652, "total_steps": 5364, "loss": 0.24254080653190613, "lr": 1.05751557674034e-06, "epoch": 2.601958041958042, "percentage": 86.73, "elapsed_time": "9:16:14", "remaining_time": "1:25:08"} +{"current_steps": 4653, "total_steps": 5364, "loss": 0.03760292008519173, "lr": 1.0546045115290405e-06, "epoch": 2.6025174825174826, "percentage": 86.74, "elapsed_time": "9:16:22", "remaining_time": "1:25:01"} +{"current_steps": 4654, "total_steps": 5364, "loss": 0.029210545122623444, "lr": 1.0516972354928735e-06, "epoch": 2.603076923076923, "percentage": 86.76, "elapsed_time": "9:16:30", "remaining_time": "1:24:53"} +{"current_steps": 4655, "total_steps": 5364, "loss": 0.010981056839227676, "lr": 1.0487937498633327e-06, "epoch": 2.6036363636363635, "percentage": 86.78, "elapsed_time": "9:16:37", "remaining_time": "1:24:46"} +{"current_steps": 4656, "total_steps": 5364, "loss": 0.05680252984166145, "lr": 1.0458940558703013e-06, "epoch": 2.604195804195804, "percentage": 86.8, "elapsed_time": "9:16:44", "remaining_time": "1:24:39"} +{"current_steps": 4657, "total_steps": 5364, "loss": 0.058628883212804794, "lr": 1.0429981547420588e-06, "epoch": 2.604755244755245, "percentage": 86.82, "elapsed_time": "9:16:51", "remaining_time": "1:24:32"} +{"current_steps": 4658, "total_steps": 5364, "loss": 0.13952630758285522, "lr": 1.0401060477052793e-06, "epoch": 2.6053146853146854, "percentage": 86.84, "elapsed_time": "9:16:58", "remaining_time": "1:24:25"} +{"current_steps": 4659, "total_steps": 5364, "loss": 0.14513689279556274, "lr": 1.03721773598503e-06, "epoch": 2.605874125874126, "percentage": 86.86, "elapsed_time": "9:17:05", "remaining_time": "1:24:17"} +{"current_steps": 4660, "total_steps": 5364, "loss": 0.165986105799675, "lr": 1.0343332208047675e-06, "epoch": 2.6064335664335663, "percentage": 86.88, "elapsed_time": "9:17:12", "remaining_time": "1:24:10"} +{"current_steps": 4661, "total_steps": 5364, "loss": 0.0619468167424202, "lr": 1.0314525033863455e-06, "epoch": 2.606993006993007, "percentage": 86.89, "elapsed_time": "9:17:19", "remaining_time": "1:24:03"} +{"current_steps": 4662, "total_steps": 5364, "loss": 0.02004910632967949, "lr": 1.0285755849500001e-06, "epoch": 2.6075524475524476, "percentage": 86.91, "elapsed_time": "9:17:26", "remaining_time": "1:23:56"} +{"current_steps": 4663, "total_steps": 5364, "loss": 0.026728782802820206, "lr": 1.0257024667143667e-06, "epoch": 2.608111888111888, "percentage": 86.93, "elapsed_time": "9:17:34", "remaining_time": "1:23:49"} +{"current_steps": 4664, "total_steps": 5364, "loss": 0.08980909734964371, "lr": 1.0228331498964671e-06, "epoch": 2.6086713286713286, "percentage": 86.95, "elapsed_time": "9:17:41", "remaining_time": "1:23:42"} +{"current_steps": 4665, "total_steps": 5364, "loss": 0.0559174083173275, "lr": 1.019967635711714e-06, "epoch": 2.609230769230769, "percentage": 86.97, "elapsed_time": "9:17:48", "remaining_time": "1:23:34"} +{"current_steps": 4666, "total_steps": 5364, "loss": 0.05049540102481842, "lr": 1.0171059253739112e-06, "epoch": 2.60979020979021, "percentage": 86.99, "elapsed_time": "9:17:55", "remaining_time": "1:23:27"} +{"current_steps": 4667, "total_steps": 5364, "loss": 0.18063391745090485, "lr": 1.0142480200952453e-06, "epoch": 2.6103496503496504, "percentage": 87.01, "elapsed_time": "9:18:02", "remaining_time": "1:23:20"} +{"current_steps": 4668, "total_steps": 5364, "loss": 0.046123601496219635, "lr": 1.0113939210862955e-06, "epoch": 2.610909090909091, "percentage": 87.02, "elapsed_time": "9:18:09", "remaining_time": "1:23:13"} +{"current_steps": 4669, "total_steps": 5364, "loss": 0.12716144323349, "lr": 1.008543629556028e-06, "epoch": 2.6114685314685313, "percentage": 87.04, "elapsed_time": "9:18:16", "remaining_time": "1:23:06"} +{"current_steps": 4670, "total_steps": 5364, "loss": 0.1518527865409851, "lr": 1.0056971467117982e-06, "epoch": 2.612027972027972, "percentage": 87.06, "elapsed_time": "9:18:23", "remaining_time": "1:22:58"} +{"current_steps": 4671, "total_steps": 5364, "loss": 0.10051946341991425, "lr": 1.0028544737593471e-06, "epoch": 2.6125874125874127, "percentage": 87.08, "elapsed_time": "9:18:30", "remaining_time": "1:22:51"} +{"current_steps": 4672, "total_steps": 5364, "loss": 0.08212938159704208, "lr": 1.000015611902796e-06, "epoch": 2.613146853146853, "percentage": 87.1, "elapsed_time": "9:18:37", "remaining_time": "1:22:44"} +{"current_steps": 4673, "total_steps": 5364, "loss": 0.04050418362021446, "lr": 9.971805623446596e-07, "epoch": 2.6137062937062936, "percentage": 87.12, "elapsed_time": "9:18:45", "remaining_time": "1:22:37"} +{"current_steps": 4674, "total_steps": 5364, "loss": 0.07946804910898209, "lr": 9.94349326285836e-07, "epoch": 2.614265734265734, "percentage": 87.14, "elapsed_time": "9:18:51", "remaining_time": "1:22:30"} +{"current_steps": 4675, "total_steps": 5364, "loss": 0.10330444574356079, "lr": 9.915219049256052e-07, "epoch": 2.614825174825175, "percentage": 87.16, "elapsed_time": "9:18:58", "remaining_time": "1:22:22"} +{"current_steps": 4676, "total_steps": 5364, "loss": 0.134657084941864, "lr": 9.886982994616357e-07, "epoch": 2.6153846153846154, "percentage": 87.17, "elapsed_time": "9:19:05", "remaining_time": "1:22:15"} +{"current_steps": 4677, "total_steps": 5364, "loss": 0.01655227690935135, "lr": 9.858785110899717e-07, "epoch": 2.615944055944056, "percentage": 87.19, "elapsed_time": "9:19:12", "remaining_time": "1:22:08"} +{"current_steps": 4678, "total_steps": 5364, "loss": 0.1313818097114563, "lr": 9.83062541005053e-07, "epoch": 2.6165034965034963, "percentage": 87.21, "elapsed_time": "9:19:19", "remaining_time": "1:22:01"} +{"current_steps": 4679, "total_steps": 5364, "loss": 0.12005007266998291, "lr": 9.802503903996907e-07, "epoch": 2.6170629370629372, "percentage": 87.23, "elapsed_time": "9:19:26", "remaining_time": "1:21:54"} +{"current_steps": 4680, "total_steps": 5364, "loss": 0.040859755128622055, "lr": 9.774420604650792e-07, "epoch": 2.6176223776223777, "percentage": 87.25, "elapsed_time": "9:19:34", "remaining_time": "1:21:46"} +{"current_steps": 4681, "total_steps": 5364, "loss": 0.09826002269983292, "lr": 9.746375523908036e-07, "epoch": 2.618181818181818, "percentage": 87.27, "elapsed_time": "9:19:41", "remaining_time": "1:21:39"} +{"current_steps": 4682, "total_steps": 5364, "loss": 0.04227941110730171, "lr": 9.718368673648194e-07, "epoch": 2.6187412587412586, "percentage": 87.29, "elapsed_time": "9:19:48", "remaining_time": "1:21:32"} +{"current_steps": 4683, "total_steps": 5364, "loss": 0.18297724425792694, "lr": 9.690400065734695e-07, "epoch": 2.619300699300699, "percentage": 87.3, "elapsed_time": "9:19:55", "remaining_time": "1:21:25"} +{"current_steps": 4684, "total_steps": 5364, "loss": 0.23154008388519287, "lr": 9.662469712014744e-07, "epoch": 2.61986013986014, "percentage": 87.32, "elapsed_time": "9:20:02", "remaining_time": "1:21:18"} +{"current_steps": 4685, "total_steps": 5364, "loss": 0.036578088998794556, "lr": 9.634577624319318e-07, "epoch": 2.6204195804195805, "percentage": 87.34, "elapsed_time": "9:20:09", "remaining_time": "1:21:11"} +{"current_steps": 4686, "total_steps": 5364, "loss": 0.06452926248311996, "lr": 9.606723814463258e-07, "epoch": 2.620979020979021, "percentage": 87.36, "elapsed_time": "9:20:16", "remaining_time": "1:21:03"} +{"current_steps": 4687, "total_steps": 5364, "loss": 0.06394671648740768, "lr": 9.578908294245103e-07, "epoch": 2.6215384615384614, "percentage": 87.38, "elapsed_time": "9:20:24", "remaining_time": "1:20:56"} +{"current_steps": 4688, "total_steps": 5364, "loss": 0.12710806727409363, "lr": 9.551131075447228e-07, "epoch": 2.6220979020979023, "percentage": 87.4, "elapsed_time": "9:20:31", "remaining_time": "1:20:49"} +{"current_steps": 4689, "total_steps": 5364, "loss": 0.035390134900808334, "lr": 9.523392169835799e-07, "epoch": 2.6226573426573427, "percentage": 87.42, "elapsed_time": "9:20:39", "remaining_time": "1:20:42"} +{"current_steps": 4690, "total_steps": 5364, "loss": 0.19761255383491516, "lr": 9.49569158916065e-07, "epoch": 2.623216783216783, "percentage": 87.43, "elapsed_time": "9:20:46", "remaining_time": "1:20:35"} +{"current_steps": 4691, "total_steps": 5364, "loss": 0.06719936430454254, "lr": 9.468029345155527e-07, "epoch": 2.6237762237762237, "percentage": 87.45, "elapsed_time": "9:20:53", "remaining_time": "1:20:28"} +{"current_steps": 4692, "total_steps": 5364, "loss": 0.03917856886982918, "lr": 9.440405449537838e-07, "epoch": 2.624335664335664, "percentage": 87.47, "elapsed_time": "9:21:00", "remaining_time": "1:20:20"} +{"current_steps": 4693, "total_steps": 5364, "loss": 0.04212659224867821, "lr": 9.412819914008764e-07, "epoch": 2.624895104895105, "percentage": 87.49, "elapsed_time": "9:21:08", "remaining_time": "1:20:13"} +{"current_steps": 4694, "total_steps": 5364, "loss": 0.08476566523313522, "lr": 9.385272750253272e-07, "epoch": 2.6254545454545455, "percentage": 87.51, "elapsed_time": "9:21:15", "remaining_time": "1:20:06"} +{"current_steps": 4695, "total_steps": 5364, "loss": 0.16443303227424622, "lr": 9.357763969940004e-07, "epoch": 2.626013986013986, "percentage": 87.53, "elapsed_time": "9:21:22", "remaining_time": "1:19:59"} +{"current_steps": 4696, "total_steps": 5364, "loss": 0.07624956965446472, "lr": 9.33029358472145e-07, "epoch": 2.626573426573427, "percentage": 87.55, "elapsed_time": "9:21:29", "remaining_time": "1:19:52"} +{"current_steps": 4697, "total_steps": 5364, "loss": 0.06449459493160248, "lr": 9.302861606233738e-07, "epoch": 2.6271328671328673, "percentage": 87.57, "elapsed_time": "9:21:36", "remaining_time": "1:19:45"} +{"current_steps": 4698, "total_steps": 5364, "loss": 0.12648284435272217, "lr": 9.275468046096759e-07, "epoch": 2.6276923076923078, "percentage": 87.58, "elapsed_time": "9:21:43", "remaining_time": "1:19:37"} +{"current_steps": 4699, "total_steps": 5364, "loss": 0.15524229407310486, "lr": 9.248112915914153e-07, "epoch": 2.6282517482517482, "percentage": 87.6, "elapsed_time": "9:21:50", "remaining_time": "1:19:30"} +{"current_steps": 4700, "total_steps": 5364, "loss": 0.0979311466217041, "lr": 9.220796227273254e-07, "epoch": 2.6288111888111887, "percentage": 87.62, "elapsed_time": "9:21:57", "remaining_time": "1:19:23"} +{"current_steps": 4701, "total_steps": 5364, "loss": 0.11986128985881805, "lr": 9.19351799174516e-07, "epoch": 2.629370629370629, "percentage": 87.64, "elapsed_time": "9:22:04", "remaining_time": "1:19:16"} +{"current_steps": 4702, "total_steps": 5364, "loss": 0.11027545481920242, "lr": 9.166278220884584e-07, "epoch": 2.62993006993007, "percentage": 87.66, "elapsed_time": "9:22:11", "remaining_time": "1:19:09"} +{"current_steps": 4703, "total_steps": 5364, "loss": 0.027227971702814102, "lr": 9.13907692623004e-07, "epoch": 2.6304895104895105, "percentage": 87.68, "elapsed_time": "9:22:18", "remaining_time": "1:19:01"} +{"current_steps": 4704, "total_steps": 5364, "loss": 0.1036660224199295, "lr": 9.1119141193037e-07, "epoch": 2.631048951048951, "percentage": 87.7, "elapsed_time": "9:22:25", "remaining_time": "1:18:54"} +{"current_steps": 4705, "total_steps": 5364, "loss": 0.12865287065505981, "lr": 9.084789811611461e-07, "epoch": 2.631608391608392, "percentage": 87.71, "elapsed_time": "9:22:32", "remaining_time": "1:18:47"} +{"current_steps": 4706, "total_steps": 5364, "loss": 0.1411428302526474, "lr": 9.057704014642866e-07, "epoch": 2.6321678321678323, "percentage": 87.73, "elapsed_time": "9:22:39", "remaining_time": "1:18:40"} +{"current_steps": 4707, "total_steps": 5364, "loss": 0.014351305551826954, "lr": 9.03065673987118e-07, "epoch": 2.632727272727273, "percentage": 87.75, "elapsed_time": "9:22:46", "remaining_time": "1:18:33"} +{"current_steps": 4708, "total_steps": 5364, "loss": 0.09398384392261505, "lr": 9.00364799875334e-07, "epoch": 2.6332867132867133, "percentage": 87.77, "elapsed_time": "9:22:53", "remaining_time": "1:18:25"} +{"current_steps": 4709, "total_steps": 5364, "loss": 0.0376257486641407, "lr": 8.976677802729983e-07, "epoch": 2.6338461538461537, "percentage": 87.79, "elapsed_time": "9:23:01", "remaining_time": "1:18:18"} +{"current_steps": 4710, "total_steps": 5364, "loss": 0.021047677844762802, "lr": 8.949746163225403e-07, "epoch": 2.634405594405594, "percentage": 87.81, "elapsed_time": "9:23:08", "remaining_time": "1:18:11"} +{"current_steps": 4711, "total_steps": 5364, "loss": 0.106009840965271, "lr": 8.922853091647532e-07, "epoch": 2.634965034965035, "percentage": 87.83, "elapsed_time": "9:23:15", "remaining_time": "1:18:04"} +{"current_steps": 4712, "total_steps": 5364, "loss": 0.11076105386018753, "lr": 8.895998599388e-07, "epoch": 2.6355244755244756, "percentage": 87.84, "elapsed_time": "9:23:23", "remaining_time": "1:17:57"} +{"current_steps": 4713, "total_steps": 5364, "loss": 0.06869295239448547, "lr": 8.869182697822087e-07, "epoch": 2.636083916083916, "percentage": 87.86, "elapsed_time": "9:23:30", "remaining_time": "1:17:50"} +{"current_steps": 4714, "total_steps": 5364, "loss": 0.06974361091852188, "lr": 8.842405398308729e-07, "epoch": 2.636643356643357, "percentage": 87.88, "elapsed_time": "9:23:38", "remaining_time": "1:17:43"} +{"current_steps": 4715, "total_steps": 5364, "loss": 0.0003780899860430509, "lr": 8.815666712190518e-07, "epoch": 2.6372027972027974, "percentage": 87.9, "elapsed_time": "9:23:45", "remaining_time": "1:17:35"} +{"current_steps": 4716, "total_steps": 5364, "loss": 0.1295163333415985, "lr": 8.78896665079364e-07, "epoch": 2.637762237762238, "percentage": 87.92, "elapsed_time": "9:23:52", "remaining_time": "1:17:28"} +{"current_steps": 4717, "total_steps": 5364, "loss": 0.04817381128668785, "lr": 8.762305225428025e-07, "epoch": 2.6383216783216783, "percentage": 87.94, "elapsed_time": "9:23:59", "remaining_time": "1:17:21"} +{"current_steps": 4718, "total_steps": 5364, "loss": 0.03916630893945694, "lr": 8.735682447387117e-07, "epoch": 2.6388811188811188, "percentage": 87.96, "elapsed_time": "9:24:07", "remaining_time": "1:17:14"} +{"current_steps": 4719, "total_steps": 5364, "loss": 0.020356670022010803, "lr": 8.709098327948051e-07, "epoch": 2.639440559440559, "percentage": 87.98, "elapsed_time": "9:24:14", "remaining_time": "1:17:07"} +{"current_steps": 4720, "total_steps": 5364, "loss": 0.07135714590549469, "lr": 8.68255287837162e-07, "epoch": 2.64, "percentage": 87.99, "elapsed_time": "9:24:21", "remaining_time": "1:17:00"} +{"current_steps": 4721, "total_steps": 5364, "loss": 0.032696399837732315, "lr": 8.656046109902116e-07, "epoch": 2.6405594405594406, "percentage": 88.01, "elapsed_time": "9:24:28", "remaining_time": "1:16:52"} +{"current_steps": 4722, "total_steps": 5364, "loss": 0.04050498455762863, "lr": 8.629578033767616e-07, "epoch": 2.641118881118881, "percentage": 88.03, "elapsed_time": "9:24:35", "remaining_time": "1:16:45"} +{"current_steps": 4723, "total_steps": 5364, "loss": 0.010556847788393497, "lr": 8.603148661179672e-07, "epoch": 2.641678321678322, "percentage": 88.05, "elapsed_time": "9:24:42", "remaining_time": "1:16:38"} +{"current_steps": 4724, "total_steps": 5364, "loss": 0.12463138997554779, "lr": 8.576758003333474e-07, "epoch": 2.6422377622377624, "percentage": 88.07, "elapsed_time": "9:24:49", "remaining_time": "1:16:31"} +{"current_steps": 4725, "total_steps": 5364, "loss": 0.025165880098938942, "lr": 8.55040607140788e-07, "epoch": 2.642797202797203, "percentage": 88.09, "elapsed_time": "9:24:56", "remaining_time": "1:16:24"} +{"current_steps": 4726, "total_steps": 5364, "loss": 0.069175586104393, "lr": 8.524092876565204e-07, "epoch": 2.6433566433566433, "percentage": 88.11, "elapsed_time": "9:25:03", "remaining_time": "1:16:16"} +{"current_steps": 4727, "total_steps": 5364, "loss": 0.06919945776462555, "lr": 8.497818429951532e-07, "epoch": 2.643916083916084, "percentage": 88.12, "elapsed_time": "9:25:10", "remaining_time": "1:16:09"} +{"current_steps": 4728, "total_steps": 5364, "loss": 0.021942205727100372, "lr": 8.471582742696405e-07, "epoch": 2.6444755244755243, "percentage": 88.14, "elapsed_time": "9:25:18", "remaining_time": "1:16:02"} +{"current_steps": 4729, "total_steps": 5364, "loss": 0.10473151504993439, "lr": 8.445385825912933e-07, "epoch": 2.645034965034965, "percentage": 88.16, "elapsed_time": "9:25:24", "remaining_time": "1:15:55"} +{"current_steps": 4730, "total_steps": 5364, "loss": 0.04877055436372757, "lr": 8.41922769069794e-07, "epoch": 2.6455944055944056, "percentage": 88.18, "elapsed_time": "9:25:31", "remaining_time": "1:15:48"} +{"current_steps": 4731, "total_steps": 5364, "loss": 0.21006245911121368, "lr": 8.393108348131674e-07, "epoch": 2.646153846153846, "percentage": 88.2, "elapsed_time": "9:25:38", "remaining_time": "1:15:40"} +{"current_steps": 4732, "total_steps": 5364, "loss": 0.06726476550102234, "lr": 8.367027809278039e-07, "epoch": 2.646713286713287, "percentage": 88.22, "elapsed_time": "9:25:45", "remaining_time": "1:15:33"} +{"current_steps": 4733, "total_steps": 5364, "loss": 0.09445297718048096, "lr": 8.340986085184477e-07, "epoch": 2.6472727272727274, "percentage": 88.24, "elapsed_time": "9:25:53", "remaining_time": "1:15:26"} +{"current_steps": 4734, "total_steps": 5364, "loss": 0.07439548522233963, "lr": 8.314983186881954e-07, "epoch": 2.647832167832168, "percentage": 88.26, "elapsed_time": "9:26:01", "remaining_time": "1:15:19"} +{"current_steps": 4735, "total_steps": 5364, "loss": 0.06531370431184769, "lr": 8.289019125385089e-07, "epoch": 2.6483916083916084, "percentage": 88.27, "elapsed_time": "9:26:08", "remaining_time": "1:15:12"} +{"current_steps": 4736, "total_steps": 5364, "loss": 0.04804437607526779, "lr": 8.263093911691922e-07, "epoch": 2.648951048951049, "percentage": 88.29, "elapsed_time": "9:26:15", "remaining_time": "1:15:05"} +{"current_steps": 4737, "total_steps": 5364, "loss": 0.16451652348041534, "lr": 8.237207556784121e-07, "epoch": 2.6495104895104893, "percentage": 88.31, "elapsed_time": "9:26:22", "remaining_time": "1:14:57"} +{"current_steps": 4738, "total_steps": 5364, "loss": 0.06156668812036514, "lr": 8.211360071626873e-07, "epoch": 2.65006993006993, "percentage": 88.33, "elapsed_time": "9:26:29", "remaining_time": "1:14:50"} +{"current_steps": 4739, "total_steps": 5364, "loss": 0.14925451576709747, "lr": 8.185551467168906e-07, "epoch": 2.6506293706293707, "percentage": 88.35, "elapsed_time": "9:26:36", "remaining_time": "1:14:43"} +{"current_steps": 4740, "total_steps": 5364, "loss": 0.04738274961709976, "lr": 8.159781754342488e-07, "epoch": 2.651188811188811, "percentage": 88.37, "elapsed_time": "9:26:43", "remaining_time": "1:14:36"} +{"current_steps": 4741, "total_steps": 5364, "loss": 0.10660860687494278, "lr": 8.134050944063354e-07, "epoch": 2.651748251748252, "percentage": 88.39, "elapsed_time": "9:26:50", "remaining_time": "1:14:29"} +{"current_steps": 4742, "total_steps": 5364, "loss": 0.06804440915584564, "lr": 8.10835904723084e-07, "epoch": 2.6523076923076925, "percentage": 88.4, "elapsed_time": "9:26:58", "remaining_time": "1:14:22"} +{"current_steps": 4743, "total_steps": 5364, "loss": 0.16086620092391968, "lr": 8.082706074727741e-07, "epoch": 2.652867132867133, "percentage": 88.42, "elapsed_time": "9:27:05", "remaining_time": "1:14:14"} +{"current_steps": 4744, "total_steps": 5364, "loss": 0.03833664208650589, "lr": 8.057092037420399e-07, "epoch": 2.6534265734265734, "percentage": 88.44, "elapsed_time": "9:27:11", "remaining_time": "1:14:07"} +{"current_steps": 4745, "total_steps": 5364, "loss": 0.0356685072183609, "lr": 8.031516946158657e-07, "epoch": 2.653986013986014, "percentage": 88.46, "elapsed_time": "9:27:19", "remaining_time": "1:14:00"} +{"current_steps": 4746, "total_steps": 5364, "loss": 0.05908123403787613, "lr": 8.005980811775826e-07, "epoch": 2.6545454545454543, "percentage": 88.48, "elapsed_time": "9:27:25", "remaining_time": "1:13:53"} +{"current_steps": 4747, "total_steps": 5364, "loss": 0.11329571902751923, "lr": 7.980483645088766e-07, "epoch": 2.6551048951048952, "percentage": 88.5, "elapsed_time": "9:27:33", "remaining_time": "1:13:46"} +{"current_steps": 4748, "total_steps": 5364, "loss": 0.11430435627698898, "lr": 7.955025456897791e-07, "epoch": 2.6556643356643357, "percentage": 88.52, "elapsed_time": "9:27:40", "remaining_time": "1:13:38"} +{"current_steps": 4749, "total_steps": 5364, "loss": 0.02996731735765934, "lr": 7.929606257986744e-07, "epoch": 2.656223776223776, "percentage": 88.53, "elapsed_time": "9:27:47", "remaining_time": "1:13:31"} +{"current_steps": 4750, "total_steps": 5364, "loss": 0.18464013934135437, "lr": 7.904226059122921e-07, "epoch": 2.656783216783217, "percentage": 88.55, "elapsed_time": "9:27:54", "remaining_time": "1:13:24"} +{"current_steps": 4751, "total_steps": 5364, "loss": 0.13110829889774323, "lr": 7.87888487105709e-07, "epoch": 2.6573426573426575, "percentage": 88.57, "elapsed_time": "9:28:01", "remaining_time": "1:13:17"} +{"current_steps": 4752, "total_steps": 5364, "loss": 0.15432825684547424, "lr": 7.853582704523532e-07, "epoch": 2.657902097902098, "percentage": 88.59, "elapsed_time": "9:28:08", "remaining_time": "1:13:10"} +{"current_steps": 4753, "total_steps": 5364, "loss": 0.13284648954868317, "lr": 7.828319570239962e-07, "epoch": 2.6584615384615384, "percentage": 88.61, "elapsed_time": "9:28:14", "remaining_time": "1:13:02"} +{"current_steps": 4754, "total_steps": 5364, "loss": 0.13804754614830017, "lr": 7.803095478907596e-07, "epoch": 2.659020979020979, "percentage": 88.63, "elapsed_time": "9:28:22", "remaining_time": "1:12:55"} +{"current_steps": 4755, "total_steps": 5364, "loss": 0.004590172320604324, "lr": 7.777910441211045e-07, "epoch": 2.6595804195804194, "percentage": 88.65, "elapsed_time": "9:28:29", "remaining_time": "1:12:48"} +{"current_steps": 4756, "total_steps": 5364, "loss": 0.0848199650645256, "lr": 7.752764467818496e-07, "epoch": 2.6601398601398603, "percentage": 88.67, "elapsed_time": "9:28:36", "remaining_time": "1:12:41"} +{"current_steps": 4757, "total_steps": 5364, "loss": 0.08241723477840424, "lr": 7.727657569381474e-07, "epoch": 2.6606993006993007, "percentage": 88.68, "elapsed_time": "9:28:43", "remaining_time": "1:12:34"} +{"current_steps": 4758, "total_steps": 5364, "loss": 0.03735372796654701, "lr": 7.702589756534995e-07, "epoch": 2.661258741258741, "percentage": 88.7, "elapsed_time": "9:28:50", "remaining_time": "1:12:27"} +{"current_steps": 4759, "total_steps": 5364, "loss": 0.09620705991983414, "lr": 7.677561039897552e-07, "epoch": 2.661818181818182, "percentage": 88.72, "elapsed_time": "9:28:57", "remaining_time": "1:12:19"} +{"current_steps": 4760, "total_steps": 5364, "loss": 0.12508580088615417, "lr": 7.652571430070987e-07, "epoch": 2.6623776223776225, "percentage": 88.74, "elapsed_time": "9:29:05", "remaining_time": "1:12:12"} +{"current_steps": 4761, "total_steps": 5364, "loss": 0.10307709127664566, "lr": 7.627620937640712e-07, "epoch": 2.662937062937063, "percentage": 88.76, "elapsed_time": "9:29:13", "remaining_time": "1:12:05"} +{"current_steps": 4762, "total_steps": 5364, "loss": 0.075468048453331, "lr": 7.602709573175426e-07, "epoch": 2.6634965034965035, "percentage": 88.78, "elapsed_time": "9:29:19", "remaining_time": "1:11:58"} +{"current_steps": 4763, "total_steps": 5364, "loss": 0.09586608409881592, "lr": 7.577837347227346e-07, "epoch": 2.664055944055944, "percentage": 88.8, "elapsed_time": "9:29:26", "remaining_time": "1:11:51"} +{"current_steps": 4764, "total_steps": 5364, "loss": 0.10071629285812378, "lr": 7.55300427033211e-07, "epoch": 2.6646153846153844, "percentage": 88.81, "elapsed_time": "9:29:33", "remaining_time": "1:11:43"} +{"current_steps": 4765, "total_steps": 5364, "loss": 0.1285012662410736, "lr": 7.52821035300868e-07, "epoch": 2.6651748251748253, "percentage": 88.83, "elapsed_time": "9:29:40", "remaining_time": "1:11:36"} +{"current_steps": 4766, "total_steps": 5364, "loss": 0.06689862161874771, "lr": 7.503455605759568e-07, "epoch": 2.6657342657342658, "percentage": 88.85, "elapsed_time": "9:29:47", "remaining_time": "1:11:29"} +{"current_steps": 4767, "total_steps": 5364, "loss": 0.11492881178855896, "lr": 7.478740039070587e-07, "epoch": 2.666293706293706, "percentage": 88.87, "elapsed_time": "9:29:55", "remaining_time": "1:11:22"} +{"current_steps": 4768, "total_steps": 5364, "loss": 0.041836850345134735, "lr": 7.454063663410993e-07, "epoch": 2.666853146853147, "percentage": 88.89, "elapsed_time": "9:30:02", "remaining_time": "1:11:15"} +{"current_steps": 4769, "total_steps": 5364, "loss": 0.08689714223146439, "lr": 7.42942648923346e-07, "epoch": 2.6674125874125876, "percentage": 88.91, "elapsed_time": "9:30:09", "remaining_time": "1:11:08"} +{"current_steps": 4770, "total_steps": 5364, "loss": 0.0075935483910143375, "lr": 7.404828526973995e-07, "epoch": 2.667972027972028, "percentage": 88.93, "elapsed_time": "9:30:17", "remaining_time": "1:11:00"} +{"current_steps": 4771, "total_steps": 5364, "loss": 0.0607183612883091, "lr": 7.380269787052074e-07, "epoch": 2.6685314685314685, "percentage": 88.94, "elapsed_time": "9:30:23", "remaining_time": "1:10:53"} +{"current_steps": 4772, "total_steps": 5364, "loss": 0.057404614984989166, "lr": 7.3557502798705e-07, "epoch": 2.669090909090909, "percentage": 88.96, "elapsed_time": "9:30:30", "remaining_time": "1:10:46"} +{"current_steps": 4773, "total_steps": 5364, "loss": 0.050055570900440216, "lr": 7.331270015815484e-07, "epoch": 2.6696503496503494, "percentage": 88.98, "elapsed_time": "9:30:37", "remaining_time": "1:10:39"} +{"current_steps": 4774, "total_steps": 5364, "loss": 0.10699306428432465, "lr": 7.306829005256621e-07, "epoch": 2.6702097902097903, "percentage": 89.0, "elapsed_time": "9:30:44", "remaining_time": "1:10:32"} +{"current_steps": 4775, "total_steps": 5364, "loss": 0.11535626649856567, "lr": 7.282427258546832e-07, "epoch": 2.670769230769231, "percentage": 89.02, "elapsed_time": "9:30:51", "remaining_time": "1:10:24"} +{"current_steps": 4776, "total_steps": 5364, "loss": 0.050010714679956436, "lr": 7.258064786022456e-07, "epoch": 2.6713286713286712, "percentage": 89.04, "elapsed_time": "9:30:58", "remaining_time": "1:10:17"} +{"current_steps": 4777, "total_steps": 5364, "loss": 0.2015446275472641, "lr": 7.233741598003175e-07, "epoch": 2.671888111888112, "percentage": 89.06, "elapsed_time": "9:31:05", "remaining_time": "1:10:10"} +{"current_steps": 4778, "total_steps": 5364, "loss": 0.16903989017009735, "lr": 7.209457704792044e-07, "epoch": 2.6724475524475526, "percentage": 89.08, "elapsed_time": "9:31:12", "remaining_time": "1:10:03"} +{"current_steps": 4779, "total_steps": 5364, "loss": 0.07091642171144485, "lr": 7.185213116675482e-07, "epoch": 2.673006993006993, "percentage": 89.09, "elapsed_time": "9:31:19", "remaining_time": "1:09:56"} +{"current_steps": 4780, "total_steps": 5364, "loss": 0.03246210515499115, "lr": 7.161007843923195e-07, "epoch": 2.6735664335664335, "percentage": 89.11, "elapsed_time": "9:31:27", "remaining_time": "1:09:49"} +{"current_steps": 4781, "total_steps": 5364, "loss": 0.021352393552660942, "lr": 7.136841896788316e-07, "epoch": 2.674125874125874, "percentage": 89.13, "elapsed_time": "9:31:34", "remaining_time": "1:09:41"} +{"current_steps": 4782, "total_steps": 5364, "loss": 0.07430198788642883, "lr": 7.112715285507266e-07, "epoch": 2.6746853146853145, "percentage": 89.15, "elapsed_time": "9:31:41", "remaining_time": "1:09:34"} +{"current_steps": 4783, "total_steps": 5364, "loss": 0.14261582493782043, "lr": 7.088628020299826e-07, "epoch": 2.6752447552447554, "percentage": 89.17, "elapsed_time": "9:31:48", "remaining_time": "1:09:27"} +{"current_steps": 4784, "total_steps": 5364, "loss": 0.0869755819439888, "lr": 7.064580111369135e-07, "epoch": 2.675804195804196, "percentage": 89.19, "elapsed_time": "9:31:55", "remaining_time": "1:09:20"} +{"current_steps": 4785, "total_steps": 5364, "loss": 0.09919381886720657, "lr": 7.040571568901599e-07, "epoch": 2.6763636363636363, "percentage": 89.21, "elapsed_time": "9:32:02", "remaining_time": "1:09:13"} +{"current_steps": 4786, "total_steps": 5364, "loss": 0.06281782686710358, "lr": 7.016602403066997e-07, "epoch": 2.676923076923077, "percentage": 89.22, "elapsed_time": "9:32:09", "remaining_time": "1:09:05"} +{"current_steps": 4787, "total_steps": 5364, "loss": 0.050296515226364136, "lr": 6.992672624018404e-07, "epoch": 2.6774825174825176, "percentage": 89.24, "elapsed_time": "9:32:16", "remaining_time": "1:08:58"} +{"current_steps": 4788, "total_steps": 5364, "loss": 0.08135329186916351, "lr": 6.968782241892236e-07, "epoch": 2.678041958041958, "percentage": 89.26, "elapsed_time": "9:32:23", "remaining_time": "1:08:51"} +{"current_steps": 4789, "total_steps": 5364, "loss": 0.11414910107851028, "lr": 6.944931266808208e-07, "epoch": 2.6786013986013986, "percentage": 89.28, "elapsed_time": "9:32:30", "remaining_time": "1:08:44"} +{"current_steps": 4790, "total_steps": 5364, "loss": 0.08645497262477875, "lr": 6.921119708869329e-07, "epoch": 2.679160839160839, "percentage": 89.3, "elapsed_time": "9:32:37", "remaining_time": "1:08:37"} +{"current_steps": 4791, "total_steps": 5364, "loss": 0.09158705919981003, "lr": 6.897347578161928e-07, "epoch": 2.6797202797202795, "percentage": 89.32, "elapsed_time": "9:32:44", "remaining_time": "1:08:30"} +{"current_steps": 4792, "total_steps": 5364, "loss": 0.17973332107067108, "lr": 6.873614884755631e-07, "epoch": 2.6802797202797204, "percentage": 89.34, "elapsed_time": "9:32:51", "remaining_time": "1:08:22"} +{"current_steps": 4793, "total_steps": 5364, "loss": 0.059885382652282715, "lr": 6.84992163870335e-07, "epoch": 2.680839160839161, "percentage": 89.35, "elapsed_time": "9:32:58", "remaining_time": "1:08:15"} +{"current_steps": 4794, "total_steps": 5364, "loss": 0.13751119375228882, "lr": 6.826267850041313e-07, "epoch": 2.6813986013986013, "percentage": 89.37, "elapsed_time": "9:33:05", "remaining_time": "1:08:08"} +{"current_steps": 4795, "total_steps": 5364, "loss": 0.013737626373767853, "lr": 6.802653528789005e-07, "epoch": 2.681958041958042, "percentage": 89.39, "elapsed_time": "9:33:12", "remaining_time": "1:08:01"} +{"current_steps": 4796, "total_steps": 5364, "loss": 0.13206090033054352, "lr": 6.779078684949192e-07, "epoch": 2.6825174825174827, "percentage": 89.41, "elapsed_time": "9:33:19", "remaining_time": "1:07:54"} +{"current_steps": 4797, "total_steps": 5364, "loss": 0.07776590436697006, "lr": 6.75554332850793e-07, "epoch": 2.683076923076923, "percentage": 89.43, "elapsed_time": "9:33:26", "remaining_time": "1:07:46"} +{"current_steps": 4798, "total_steps": 5364, "loss": 0.038484424352645874, "lr": 6.732047469434577e-07, "epoch": 2.6836363636363636, "percentage": 89.45, "elapsed_time": "9:33:33", "remaining_time": "1:07:39"} +{"current_steps": 4799, "total_steps": 5364, "loss": 0.044652439653873444, "lr": 6.708591117681673e-07, "epoch": 2.684195804195804, "percentage": 89.47, "elapsed_time": "9:33:40", "remaining_time": "1:07:32"} +{"current_steps": 4800, "total_steps": 5364, "loss": 0.1714843213558197, "lr": 6.685174283185159e-07, "epoch": 2.6847552447552445, "percentage": 89.49, "elapsed_time": "9:33:47", "remaining_time": "1:07:25"} +{"current_steps": 4801, "total_steps": 5364, "loss": 0.07104168087244034, "lr": 6.661796975864088e-07, "epoch": 2.6853146853146854, "percentage": 89.5, "elapsed_time": "9:33:54", "remaining_time": "1:07:18"} +{"current_steps": 4802, "total_steps": 5364, "loss": 0.043704692274332047, "lr": 6.638459205620873e-07, "epoch": 2.685874125874126, "percentage": 89.52, "elapsed_time": "9:34:01", "remaining_time": "1:07:10"} +{"current_steps": 4803, "total_steps": 5364, "loss": 0.07925727218389511, "lr": 6.615160982341162e-07, "epoch": 2.6864335664335663, "percentage": 89.54, "elapsed_time": "9:34:08", "remaining_time": "1:07:03"} +{"current_steps": 4804, "total_steps": 5364, "loss": 0.1392529010772705, "lr": 6.591902315893794e-07, "epoch": 2.6869930069930072, "percentage": 89.56, "elapsed_time": "9:34:15", "remaining_time": "1:06:56"} +{"current_steps": 4805, "total_steps": 5364, "loss": 0.15204253792762756, "lr": 6.568683216130944e-07, "epoch": 2.6875524475524477, "percentage": 89.58, "elapsed_time": "9:34:22", "remaining_time": "1:06:49"} +{"current_steps": 4806, "total_steps": 5364, "loss": 0.06843969225883484, "lr": 6.545503692887934e-07, "epoch": 2.688111888111888, "percentage": 89.6, "elapsed_time": "9:34:29", "remaining_time": "1:06:42"} +{"current_steps": 4807, "total_steps": 5364, "loss": 0.1008538007736206, "lr": 6.522363755983386e-07, "epoch": 2.6886713286713286, "percentage": 89.62, "elapsed_time": "9:34:36", "remaining_time": "1:06:34"} +{"current_steps": 4808, "total_steps": 5364, "loss": 0.0961901918053627, "lr": 6.499263415219159e-07, "epoch": 2.689230769230769, "percentage": 89.63, "elapsed_time": "9:34:43", "remaining_time": "1:06:27"} +{"current_steps": 4809, "total_steps": 5364, "loss": 0.053754642605781555, "lr": 6.476202680380239e-07, "epoch": 2.6897902097902096, "percentage": 89.65, "elapsed_time": "9:34:50", "remaining_time": "1:06:20"} +{"current_steps": 4810, "total_steps": 5364, "loss": 0.032822247594594955, "lr": 6.453181561234989e-07, "epoch": 2.6903496503496505, "percentage": 89.67, "elapsed_time": "9:34:57", "remaining_time": "1:06:13"} +{"current_steps": 4811, "total_steps": 5364, "loss": 0.11335393041372299, "lr": 6.430200067534875e-07, "epoch": 2.690909090909091, "percentage": 89.69, "elapsed_time": "9:35:03", "remaining_time": "1:06:06"} +{"current_steps": 4812, "total_steps": 5364, "loss": 0.08931862562894821, "lr": 6.407258209014611e-07, "epoch": 2.6914685314685314, "percentage": 89.71, "elapsed_time": "9:35:10", "remaining_time": "1:05:58"} +{"current_steps": 4813, "total_steps": 5364, "loss": 0.15237212181091309, "lr": 6.384355995392155e-07, "epoch": 2.6920279720279723, "percentage": 89.73, "elapsed_time": "9:35:18", "remaining_time": "1:05:51"} +{"current_steps": 4814, "total_steps": 5364, "loss": 0.06374464929103851, "lr": 6.361493436368593e-07, "epoch": 2.6925874125874127, "percentage": 89.75, "elapsed_time": "9:35:25", "remaining_time": "1:05:44"} +{"current_steps": 4815, "total_steps": 5364, "loss": 0.02318044751882553, "lr": 6.338670541628322e-07, "epoch": 2.693146853146853, "percentage": 89.77, "elapsed_time": "9:35:32", "remaining_time": "1:05:37"} +{"current_steps": 4816, "total_steps": 5364, "loss": 0.07137596607208252, "lr": 6.315887320838842e-07, "epoch": 2.6937062937062937, "percentage": 89.78, "elapsed_time": "9:35:39", "remaining_time": "1:05:30"} +{"current_steps": 4817, "total_steps": 5364, "loss": 0.07192157208919525, "lr": 6.29314378365089e-07, "epoch": 2.694265734265734, "percentage": 89.8, "elapsed_time": "9:35:45", "remaining_time": "1:05:22"} +{"current_steps": 4818, "total_steps": 5364, "loss": 0.08717767149209976, "lr": 6.270439939698414e-07, "epoch": 2.6948251748251746, "percentage": 89.82, "elapsed_time": "9:35:52", "remaining_time": "1:05:15"} +{"current_steps": 4819, "total_steps": 5364, "loss": 0.010756965726613998, "lr": 6.247775798598466e-07, "epoch": 2.6953846153846155, "percentage": 89.84, "elapsed_time": "9:36:00", "remaining_time": "1:05:08"} +{"current_steps": 4820, "total_steps": 5364, "loss": 0.22568590939044952, "lr": 6.225151369951399e-07, "epoch": 2.695944055944056, "percentage": 89.86, "elapsed_time": "9:36:07", "remaining_time": "1:05:01"} +{"current_steps": 4821, "total_steps": 5364, "loss": 0.13338537514209747, "lr": 6.202566663340648e-07, "epoch": 2.6965034965034964, "percentage": 89.88, "elapsed_time": "9:36:14", "remaining_time": "1:04:54"} +{"current_steps": 4822, "total_steps": 5364, "loss": 0.21569012105464935, "lr": 6.18002168833286e-07, "epoch": 2.6970629370629373, "percentage": 89.9, "elapsed_time": "9:36:21", "remaining_time": "1:04:46"} +{"current_steps": 4823, "total_steps": 5364, "loss": 0.03815922141075134, "lr": 6.157516454477874e-07, "epoch": 2.6976223776223778, "percentage": 89.91, "elapsed_time": "9:36:28", "remaining_time": "1:04:39"} +{"current_steps": 4824, "total_steps": 5364, "loss": 0.2077419012784958, "lr": 6.13505097130862e-07, "epoch": 2.6981818181818182, "percentage": 89.93, "elapsed_time": "9:36:35", "remaining_time": "1:04:32"} +{"current_steps": 4825, "total_steps": 5364, "loss": 0.18102942407131195, "lr": 6.112625248341286e-07, "epoch": 2.6987412587412587, "percentage": 89.95, "elapsed_time": "9:36:42", "remaining_time": "1:04:25"} +{"current_steps": 4826, "total_steps": 5364, "loss": 0.17260771989822388, "lr": 6.09023929507514e-07, "epoch": 2.699300699300699, "percentage": 89.97, "elapsed_time": "9:36:49", "remaining_time": "1:04:18"} +{"current_steps": 4827, "total_steps": 5364, "loss": 0.26072627305984497, "lr": 6.067893120992652e-07, "epoch": 2.6998601398601396, "percentage": 89.99, "elapsed_time": "9:36:56", "remaining_time": "1:04:11"} +{"current_steps": 4828, "total_steps": 5364, "loss": 0.0822022408246994, "lr": 6.045586735559428e-07, "epoch": 2.7004195804195805, "percentage": 90.01, "elapsed_time": "9:37:03", "remaining_time": "1:04:03"} +{"current_steps": 4829, "total_steps": 5364, "loss": 0.10042086243629456, "lr": 6.023320148224221e-07, "epoch": 2.700979020979021, "percentage": 90.03, "elapsed_time": "9:37:10", "remaining_time": "1:03:56"} +{"current_steps": 4830, "total_steps": 5364, "loss": 0.1179453581571579, "lr": 6.001093368418897e-07, "epoch": 2.7015384615384614, "percentage": 90.04, "elapsed_time": "9:37:17", "remaining_time": "1:03:49"} +{"current_steps": 4831, "total_steps": 5364, "loss": 0.0413038544356823, "lr": 5.978906405558504e-07, "epoch": 2.7020979020979023, "percentage": 90.06, "elapsed_time": "9:37:24", "remaining_time": "1:03:42"} +{"current_steps": 4832, "total_steps": 5364, "loss": 0.13676247000694275, "lr": 5.956759269041201e-07, "epoch": 2.702657342657343, "percentage": 90.08, "elapsed_time": "9:37:31", "remaining_time": "1:03:35"} +{"current_steps": 4833, "total_steps": 5364, "loss": 0.05617497116327286, "lr": 5.934651968248295e-07, "epoch": 2.7032167832167833, "percentage": 90.1, "elapsed_time": "9:37:38", "remaining_time": "1:03:27"} +{"current_steps": 4834, "total_steps": 5364, "loss": 0.13416530191898346, "lr": 5.912584512544195e-07, "epoch": 2.7037762237762237, "percentage": 90.12, "elapsed_time": "9:37:45", "remaining_time": "1:03:20"} +{"current_steps": 4835, "total_steps": 5364, "loss": 0.1994042694568634, "lr": 5.890556911276435e-07, "epoch": 2.704335664335664, "percentage": 90.14, "elapsed_time": "9:37:53", "remaining_time": "1:03:13"} +{"current_steps": 4836, "total_steps": 5364, "loss": 0.05722580477595329, "lr": 5.868569173775673e-07, "epoch": 2.7048951048951047, "percentage": 90.16, "elapsed_time": "9:38:00", "remaining_time": "1:03:06"} +{"current_steps": 4837, "total_steps": 5364, "loss": 0.08654353022575378, "lr": 5.846621309355693e-07, "epoch": 2.7054545454545456, "percentage": 90.18, "elapsed_time": "9:38:07", "remaining_time": "1:02:59"} +{"current_steps": 4838, "total_steps": 5364, "loss": 0.15309712290763855, "lr": 5.82471332731338e-07, "epoch": 2.706013986013986, "percentage": 90.19, "elapsed_time": "9:38:14", "remaining_time": "1:02:52"} +{"current_steps": 4839, "total_steps": 5364, "loss": 0.06144693121314049, "lr": 5.802845236928733e-07, "epoch": 2.7065734265734265, "percentage": 90.21, "elapsed_time": "9:38:21", "remaining_time": "1:02:44"} +{"current_steps": 4840, "total_steps": 5364, "loss": 0.0667698010802269, "lr": 5.78101704746482e-07, "epoch": 2.7071328671328674, "percentage": 90.23, "elapsed_time": "9:38:29", "remaining_time": "1:02:37"} +{"current_steps": 4841, "total_steps": 5364, "loss": 0.24342583119869232, "lr": 5.759228768167835e-07, "epoch": 2.707692307692308, "percentage": 90.25, "elapsed_time": "9:38:36", "remaining_time": "1:02:30"} +{"current_steps": 4842, "total_steps": 5364, "loss": 0.04426802694797516, "lr": 5.737480408267082e-07, "epoch": 2.7082517482517483, "percentage": 90.27, "elapsed_time": "9:38:44", "remaining_time": "1:02:23"} +{"current_steps": 4843, "total_steps": 5364, "loss": 0.13820983469486237, "lr": 5.715771976974915e-07, "epoch": 2.7088111888111888, "percentage": 90.29, "elapsed_time": "9:38:51", "remaining_time": "1:02:16"} +{"current_steps": 4844, "total_steps": 5364, "loss": 0.028235316276550293, "lr": 5.694103483486835e-07, "epoch": 2.7093706293706292, "percentage": 90.31, "elapsed_time": "9:38:58", "remaining_time": "1:02:09"} +{"current_steps": 4845, "total_steps": 5364, "loss": 0.03292941302061081, "lr": 5.672474936981332e-07, "epoch": 2.7099300699300697, "percentage": 90.32, "elapsed_time": "9:39:05", "remaining_time": "1:02:01"} +{"current_steps": 4846, "total_steps": 5364, "loss": 0.04234527796506882, "lr": 5.650886346620066e-07, "epoch": 2.7104895104895106, "percentage": 90.34, "elapsed_time": "9:39:12", "remaining_time": "1:01:54"} +{"current_steps": 4847, "total_steps": 5364, "loss": 0.27317675948143005, "lr": 5.629337721547734e-07, "epoch": 2.711048951048951, "percentage": 90.36, "elapsed_time": "9:39:19", "remaining_time": "1:01:47"} +{"current_steps": 4848, "total_steps": 5364, "loss": 0.11615654826164246, "lr": 5.60782907089208e-07, "epoch": 2.7116083916083915, "percentage": 90.38, "elapsed_time": "9:39:26", "remaining_time": "1:01:40"} +{"current_steps": 4849, "total_steps": 5364, "loss": 0.085695780813694, "lr": 5.586360403763991e-07, "epoch": 2.7121678321678324, "percentage": 90.4, "elapsed_time": "9:39:33", "remaining_time": "1:01:33"} +{"current_steps": 4850, "total_steps": 5364, "loss": 0.04336924105882645, "lr": 5.564931729257317e-07, "epoch": 2.712727272727273, "percentage": 90.42, "elapsed_time": "9:39:40", "remaining_time": "1:01:26"} +{"current_steps": 4851, "total_steps": 5364, "loss": 0.10591655224561691, "lr": 5.543543056449041e-07, "epoch": 2.7132867132867133, "percentage": 90.44, "elapsed_time": "9:39:48", "remaining_time": "1:01:18"} +{"current_steps": 4852, "total_steps": 5364, "loss": 0.022871579974889755, "lr": 5.522194394399194e-07, "epoch": 2.713846153846154, "percentage": 90.45, "elapsed_time": "9:39:55", "remaining_time": "1:01:11"} +{"current_steps": 4853, "total_steps": 5364, "loss": 0.11475977301597595, "lr": 5.500885752150798e-07, "epoch": 2.7144055944055943, "percentage": 90.47, "elapsed_time": "9:40:03", "remaining_time": "1:01:04"} +{"current_steps": 4854, "total_steps": 5364, "loss": 0.03981172293424606, "lr": 5.479617138730031e-07, "epoch": 2.7149650349650347, "percentage": 90.49, "elapsed_time": "9:40:10", "remaining_time": "1:00:57"} +{"current_steps": 4855, "total_steps": 5364, "loss": 0.07584191113710403, "lr": 5.458388563146e-07, "epoch": 2.7155244755244756, "percentage": 90.51, "elapsed_time": "9:40:17", "remaining_time": "1:00:50"} +{"current_steps": 4856, "total_steps": 5364, "loss": 0.029487628489732742, "lr": 5.437200034390932e-07, "epoch": 2.716083916083916, "percentage": 90.53, "elapsed_time": "9:40:25", "remaining_time": "1:00:43"} +{"current_steps": 4857, "total_steps": 5364, "loss": 0.10549632459878922, "lr": 5.416051561440072e-07, "epoch": 2.7166433566433565, "percentage": 90.55, "elapsed_time": "9:40:32", "remaining_time": "1:00:35"} +{"current_steps": 4858, "total_steps": 5364, "loss": 0.12315791845321655, "lr": 5.394943153251642e-07, "epoch": 2.7172027972027974, "percentage": 90.57, "elapsed_time": "9:40:39", "remaining_time": "1:00:28"} +{"current_steps": 4859, "total_steps": 5364, "loss": 0.0706019476056099, "lr": 5.373874818767011e-07, "epoch": 2.717762237762238, "percentage": 90.59, "elapsed_time": "9:40:46", "remaining_time": "1:00:21"} +{"current_steps": 4860, "total_steps": 5364, "loss": 0.16092607378959656, "lr": 5.352846566910453e-07, "epoch": 2.7183216783216784, "percentage": 90.6, "elapsed_time": "9:40:53", "remaining_time": "1:00:14"} +{"current_steps": 4861, "total_steps": 5364, "loss": 0.01062640268355608, "lr": 5.33185840658933e-07, "epoch": 2.718881118881119, "percentage": 90.62, "elapsed_time": "9:41:00", "remaining_time": "1:00:07"} +{"current_steps": 4862, "total_steps": 5364, "loss": 0.10521841049194336, "lr": 5.310910346694009e-07, "epoch": 2.7194405594405593, "percentage": 90.64, "elapsed_time": "9:41:07", "remaining_time": "1:00:00"} +{"current_steps": 4863, "total_steps": 5364, "loss": 0.056228138506412506, "lr": 5.290002396097849e-07, "epoch": 2.7199999999999998, "percentage": 90.66, "elapsed_time": "9:41:14", "remaining_time": "0:59:52"} +{"current_steps": 4864, "total_steps": 5364, "loss": 0.09026574343442917, "lr": 5.269134563657264e-07, "epoch": 2.7205594405594407, "percentage": 90.68, "elapsed_time": "9:41:21", "remaining_time": "0:59:45"} +{"current_steps": 4865, "total_steps": 5364, "loss": 0.04187865927815437, "lr": 5.248306858211627e-07, "epoch": 2.721118881118881, "percentage": 90.7, "elapsed_time": "9:41:28", "remaining_time": "0:59:38"} +{"current_steps": 4866, "total_steps": 5364, "loss": 0.07991749793291092, "lr": 5.227519288583338e-07, "epoch": 2.7216783216783216, "percentage": 90.72, "elapsed_time": "9:41:35", "remaining_time": "0:59:31"} +{"current_steps": 4867, "total_steps": 5364, "loss": 0.10445399582386017, "lr": 5.206771863577797e-07, "epoch": 2.7222377622377625, "percentage": 90.73, "elapsed_time": "9:41:42", "remaining_time": "0:59:24"} +{"current_steps": 4868, "total_steps": 5364, "loss": 0.07234948128461838, "lr": 5.186064591983386e-07, "epoch": 2.722797202797203, "percentage": 90.75, "elapsed_time": "9:41:50", "remaining_time": "0:59:17"} +{"current_steps": 4869, "total_steps": 5364, "loss": 0.14555084705352783, "lr": 5.165397482571511e-07, "epoch": 2.7233566433566434, "percentage": 90.77, "elapsed_time": "9:41:57", "remaining_time": "0:59:09"} +{"current_steps": 4870, "total_steps": 5364, "loss": 0.11011561751365662, "lr": 5.144770544096512e-07, "epoch": 2.723916083916084, "percentage": 90.79, "elapsed_time": "9:42:05", "remaining_time": "0:59:02"} +{"current_steps": 4871, "total_steps": 5364, "loss": 0.10255992412567139, "lr": 5.124183785295755e-07, "epoch": 2.7244755244755243, "percentage": 90.81, "elapsed_time": "9:42:11", "remaining_time": "0:58:55"} +{"current_steps": 4872, "total_steps": 5364, "loss": 0.01201770268380642, "lr": 5.103637214889568e-07, "epoch": 2.725034965034965, "percentage": 90.83, "elapsed_time": "9:42:18", "remaining_time": "0:58:48"} +{"current_steps": 4873, "total_steps": 5364, "loss": 0.11315815150737762, "lr": 5.083130841581286e-07, "epoch": 2.7255944055944057, "percentage": 90.85, "elapsed_time": "9:42:25", "remaining_time": "0:58:41"} +{"current_steps": 4874, "total_steps": 5364, "loss": 0.05612741410732269, "lr": 5.06266467405715e-07, "epoch": 2.726153846153846, "percentage": 90.87, "elapsed_time": "9:42:32", "remaining_time": "0:58:33"} +{"current_steps": 4875, "total_steps": 5364, "loss": 0.2006445676088333, "lr": 5.042238720986436e-07, "epoch": 2.7267132867132866, "percentage": 90.88, "elapsed_time": "9:42:39", "remaining_time": "0:58:26"} +{"current_steps": 4876, "total_steps": 5364, "loss": 0.09439338743686676, "lr": 5.02185299102137e-07, "epoch": 2.7272727272727275, "percentage": 90.9, "elapsed_time": "9:42:46", "remaining_time": "0:58:19"} +{"current_steps": 4877, "total_steps": 5364, "loss": 0.08648508787155151, "lr": 5.001507492797119e-07, "epoch": 2.727832167832168, "percentage": 90.92, "elapsed_time": "9:42:53", "remaining_time": "0:58:12"} +{"current_steps": 4878, "total_steps": 5364, "loss": 0.09073715656995773, "lr": 4.981202234931837e-07, "epoch": 2.7283916083916084, "percentage": 90.94, "elapsed_time": "9:43:00", "remaining_time": "0:58:05"} +{"current_steps": 4879, "total_steps": 5364, "loss": 0.02977767214179039, "lr": 4.960937226026596e-07, "epoch": 2.728951048951049, "percentage": 90.96, "elapsed_time": "9:43:07", "remaining_time": "0:57:57"} +{"current_steps": 4880, "total_steps": 5364, "loss": 0.06906495243310928, "lr": 4.940712474665443e-07, "epoch": 2.7295104895104894, "percentage": 90.98, "elapsed_time": "9:43:14", "remaining_time": "0:57:50"} +{"current_steps": 4881, "total_steps": 5364, "loss": 0.14573818445205688, "lr": 4.92052798941538e-07, "epoch": 2.73006993006993, "percentage": 91.0, "elapsed_time": "9:43:21", "remaining_time": "0:57:43"} +{"current_steps": 4882, "total_steps": 5364, "loss": 0.12271745502948761, "lr": 4.900383778826345e-07, "epoch": 2.7306293706293707, "percentage": 91.01, "elapsed_time": "9:43:28", "remaining_time": "0:57:36"} +{"current_steps": 4883, "total_steps": 5364, "loss": 0.08478693664073944, "lr": 4.880279851431214e-07, "epoch": 2.731188811188811, "percentage": 91.03, "elapsed_time": "9:43:35", "remaining_time": "0:57:29"} +{"current_steps": 4884, "total_steps": 5364, "loss": 0.07550100237131119, "lr": 4.860216215745772e-07, "epoch": 2.7317482517482516, "percentage": 91.05, "elapsed_time": "9:43:42", "remaining_time": "0:57:22"} +{"current_steps": 4885, "total_steps": 5364, "loss": 0.12323139607906342, "lr": 4.84019288026879e-07, "epoch": 2.7323076923076925, "percentage": 91.07, "elapsed_time": "9:43:49", "remaining_time": "0:57:14"} +{"current_steps": 4886, "total_steps": 5364, "loss": 0.14971648156642914, "lr": 4.820209853481928e-07, "epoch": 2.732867132867133, "percentage": 91.09, "elapsed_time": "9:43:56", "remaining_time": "0:57:07"} +{"current_steps": 4887, "total_steps": 5364, "loss": 0.17323225736618042, "lr": 4.800267143849791e-07, "epoch": 2.7334265734265735, "percentage": 91.11, "elapsed_time": "9:44:03", "remaining_time": "0:57:00"} +{"current_steps": 4888, "total_steps": 5364, "loss": 0.09519673883914948, "lr": 4.780364759819911e-07, "epoch": 2.733986013986014, "percentage": 91.13, "elapsed_time": "9:44:10", "remaining_time": "0:56:53"} +{"current_steps": 4889, "total_steps": 5364, "loss": 0.044987231492996216, "lr": 4.7605027098226896e-07, "epoch": 2.7345454545454544, "percentage": 91.14, "elapsed_time": "9:44:17", "remaining_time": "0:56:46"} +{"current_steps": 4890, "total_steps": 5364, "loss": 0.139259934425354, "lr": 4.740681002271541e-07, "epoch": 2.735104895104895, "percentage": 91.16, "elapsed_time": "9:44:24", "remaining_time": "0:56:38"} +{"current_steps": 4891, "total_steps": 5364, "loss": 0.1729244440793991, "lr": 4.720899645562682e-07, "epoch": 2.7356643356643358, "percentage": 91.18, "elapsed_time": "9:44:31", "remaining_time": "0:56:31"} +{"current_steps": 4892, "total_steps": 5364, "loss": 0.009089294821023941, "lr": 4.7011586480753105e-07, "epoch": 2.736223776223776, "percentage": 91.2, "elapsed_time": "9:44:38", "remaining_time": "0:56:24"} +{"current_steps": 4893, "total_steps": 5364, "loss": 0.10830871760845184, "lr": 4.681458018171503e-07, "epoch": 2.7367832167832167, "percentage": 91.22, "elapsed_time": "9:44:45", "remaining_time": "0:56:17"} +{"current_steps": 4894, "total_steps": 5364, "loss": 0.12814119458198547, "lr": 4.6617977641962075e-07, "epoch": 2.7373426573426576, "percentage": 91.24, "elapsed_time": "9:44:52", "remaining_time": "0:56:10"} +{"current_steps": 4895, "total_steps": 5364, "loss": 0.0661759153008461, "lr": 4.642177894477351e-07, "epoch": 2.737902097902098, "percentage": 91.26, "elapsed_time": "9:44:59", "remaining_time": "0:56:02"} +{"current_steps": 4896, "total_steps": 5364, "loss": 0.20286014676094055, "lr": 4.6225984173256853e-07, "epoch": 2.7384615384615385, "percentage": 91.28, "elapsed_time": "9:45:05", "remaining_time": "0:55:55"} +{"current_steps": 4897, "total_steps": 5364, "loss": 0.06597593426704407, "lr": 4.603059341034832e-07, "epoch": 2.739020979020979, "percentage": 91.29, "elapsed_time": "9:45:12", "remaining_time": "0:55:48"} +{"current_steps": 4898, "total_steps": 5364, "loss": 0.09235873818397522, "lr": 4.583560673881404e-07, "epoch": 2.7395804195804194, "percentage": 91.31, "elapsed_time": "9:45:19", "remaining_time": "0:55:41"} +{"current_steps": 4899, "total_steps": 5364, "loss": 0.05744379758834839, "lr": 4.564102424124772e-07, "epoch": 2.74013986013986, "percentage": 91.33, "elapsed_time": "9:45:26", "remaining_time": "0:55:34"} +{"current_steps": 4900, "total_steps": 5364, "loss": 0.07656777650117874, "lr": 4.5446846000072765e-07, "epoch": 2.740699300699301, "percentage": 91.35, "elapsed_time": "9:45:33", "remaining_time": "0:55:26"} +{"current_steps": 4901, "total_steps": 5364, "loss": 0.10856030136346817, "lr": 4.525307209754104e-07, "epoch": 2.7412587412587412, "percentage": 91.37, "elapsed_time": "9:45:40", "remaining_time": "0:55:19"} +{"current_steps": 4902, "total_steps": 5364, "loss": 0.04348676651716232, "lr": 4.5059702615732783e-07, "epoch": 2.7418181818181817, "percentage": 91.39, "elapsed_time": "9:45:47", "remaining_time": "0:55:12"} +{"current_steps": 4903, "total_steps": 5364, "loss": 0.0500948429107666, "lr": 4.48667376365578e-07, "epoch": 2.7423776223776226, "percentage": 91.41, "elapsed_time": "9:45:55", "remaining_time": "0:55:05"} +{"current_steps": 4904, "total_steps": 5364, "loss": 0.1362314522266388, "lr": 4.46741772417536e-07, "epoch": 2.742937062937063, "percentage": 91.42, "elapsed_time": "9:46:02", "remaining_time": "0:54:58"} +{"current_steps": 4905, "total_steps": 5364, "loss": 0.015057351440191269, "lr": 4.448202151288683e-07, "epoch": 2.7434965034965035, "percentage": 91.44, "elapsed_time": "9:46:09", "remaining_time": "0:54:51"} +{"current_steps": 4906, "total_steps": 5364, "loss": 0.014581594616174698, "lr": 4.4290270531352707e-07, "epoch": 2.744055944055944, "percentage": 91.46, "elapsed_time": "9:46:16", "remaining_time": "0:54:43"} +{"current_steps": 4907, "total_steps": 5364, "loss": 0.04733207821846008, "lr": 4.409892437837493e-07, "epoch": 2.7446153846153845, "percentage": 91.48, "elapsed_time": "9:46:24", "remaining_time": "0:54:36"} +{"current_steps": 4908, "total_steps": 5364, "loss": 0.06507638096809387, "lr": 4.390798313500577e-07, "epoch": 2.745174825174825, "percentage": 91.5, "elapsed_time": "9:46:31", "remaining_time": "0:54:29"} +{"current_steps": 4909, "total_steps": 5364, "loss": 0.195505291223526, "lr": 4.371744688212576e-07, "epoch": 2.745734265734266, "percentage": 91.52, "elapsed_time": "9:46:38", "remaining_time": "0:54:22"} +{"current_steps": 4910, "total_steps": 5364, "loss": 0.061857469379901886, "lr": 4.3527315700443996e-07, "epoch": 2.7462937062937063, "percentage": 91.54, "elapsed_time": "9:46:45", "remaining_time": "0:54:15"} +{"current_steps": 4911, "total_steps": 5364, "loss": 0.16515888273715973, "lr": 4.333758967049828e-07, "epoch": 2.7468531468531467, "percentage": 91.55, "elapsed_time": "9:46:52", "remaining_time": "0:54:08"} +{"current_steps": 4912, "total_steps": 5364, "loss": 0.12153796851634979, "lr": 4.314826887265433e-07, "epoch": 2.7474125874125876, "percentage": 91.57, "elapsed_time": "9:46:59", "remaining_time": "0:54:00"} +{"current_steps": 4913, "total_steps": 5364, "loss": 0.13263708353042603, "lr": 4.2959353387106664e-07, "epoch": 2.747972027972028, "percentage": 91.59, "elapsed_time": "9:47:06", "remaining_time": "0:53:53"} +{"current_steps": 4914, "total_steps": 5364, "loss": 0.0773947685956955, "lr": 4.277084329387771e-07, "epoch": 2.7485314685314686, "percentage": 91.61, "elapsed_time": "9:47:13", "remaining_time": "0:53:46"} +{"current_steps": 4915, "total_steps": 5364, "loss": 0.012152200564742088, "lr": 4.258273867281837e-07, "epoch": 2.749090909090909, "percentage": 91.63, "elapsed_time": "9:47:20", "remaining_time": "0:53:39"} +{"current_steps": 4916, "total_steps": 5364, "loss": 0.06117810308933258, "lr": 4.2395039603607915e-07, "epoch": 2.7496503496503495, "percentage": 91.65, "elapsed_time": "9:47:27", "remaining_time": "0:53:32"} +{"current_steps": 4917, "total_steps": 5364, "loss": 0.16269856691360474, "lr": 4.2207746165753403e-07, "epoch": 2.75020979020979, "percentage": 91.67, "elapsed_time": "9:47:34", "remaining_time": "0:53:24"} +{"current_steps": 4918, "total_steps": 5364, "loss": 0.12468735873699188, "lr": 4.202085843859083e-07, "epoch": 2.750769230769231, "percentage": 91.69, "elapsed_time": "9:47:41", "remaining_time": "0:53:17"} +{"current_steps": 4919, "total_steps": 5364, "loss": 0.019279032945632935, "lr": 4.183437650128341e-07, "epoch": 2.7513286713286713, "percentage": 91.7, "elapsed_time": "9:47:48", "remaining_time": "0:53:10"} +{"current_steps": 4920, "total_steps": 5364, "loss": 0.030820688232779503, "lr": 4.1648300432823085e-07, "epoch": 2.7518881118881118, "percentage": 91.72, "elapsed_time": "9:47:55", "remaining_time": "0:53:03"} +{"current_steps": 4921, "total_steps": 5364, "loss": 0.1655111014842987, "lr": 4.146263031202968e-07, "epoch": 2.7524475524475527, "percentage": 91.74, "elapsed_time": "9:48:02", "remaining_time": "0:52:56"} +{"current_steps": 4922, "total_steps": 5364, "loss": 0.11915578693151474, "lr": 4.1277366217551296e-07, "epoch": 2.753006993006993, "percentage": 91.76, "elapsed_time": "9:48:09", "remaining_time": "0:52:49"} +{"current_steps": 4923, "total_steps": 5364, "loss": 0.07029218971729279, "lr": 4.109250822786348e-07, "epoch": 2.7535664335664336, "percentage": 91.78, "elapsed_time": "9:48:16", "remaining_time": "0:52:41"} +{"current_steps": 4924, "total_steps": 5364, "loss": 0.09724392741918564, "lr": 4.0908056421270606e-07, "epoch": 2.754125874125874, "percentage": 91.8, "elapsed_time": "9:48:23", "remaining_time": "0:52:34"} +{"current_steps": 4925, "total_steps": 5364, "loss": 0.05131128430366516, "lr": 4.072401087590405e-07, "epoch": 2.7546853146853145, "percentage": 91.82, "elapsed_time": "9:48:30", "remaining_time": "0:52:27"} +{"current_steps": 4926, "total_steps": 5364, "loss": 0.00523514486849308, "lr": 4.0540371669723797e-07, "epoch": 2.755244755244755, "percentage": 91.83, "elapsed_time": "9:48:38", "remaining_time": "0:52:20"} +{"current_steps": 4927, "total_steps": 5364, "loss": 0.06761936843395233, "lr": 4.035713888051762e-07, "epoch": 2.755804195804196, "percentage": 91.85, "elapsed_time": "9:48:45", "remaining_time": "0:52:13"} +{"current_steps": 4928, "total_steps": 5364, "loss": 0.060123756527900696, "lr": 4.0174312585900433e-07, "epoch": 2.7563636363636363, "percentage": 91.87, "elapsed_time": "9:48:52", "remaining_time": "0:52:05"} +{"current_steps": 4929, "total_steps": 5364, "loss": 0.07848545908927917, "lr": 3.999189286331628e-07, "epoch": 2.756923076923077, "percentage": 91.89, "elapsed_time": "9:48:59", "remaining_time": "0:51:58"} +{"current_steps": 4930, "total_steps": 5364, "loss": 0.016435006633400917, "lr": 3.98098797900357e-07, "epoch": 2.7574825174825177, "percentage": 91.91, "elapsed_time": "9:49:07", "remaining_time": "0:51:51"} +{"current_steps": 4931, "total_steps": 5364, "loss": 0.12187676876783371, "lr": 3.962827344315756e-07, "epoch": 2.758041958041958, "percentage": 91.93, "elapsed_time": "9:49:14", "remaining_time": "0:51:44"} +{"current_steps": 4932, "total_steps": 5364, "loss": 0.07257356494665146, "lr": 3.944707389960878e-07, "epoch": 2.7586013986013986, "percentage": 91.95, "elapsed_time": "9:49:21", "remaining_time": "0:51:37"} +{"current_steps": 4933, "total_steps": 5364, "loss": 0.15309825539588928, "lr": 3.9266281236142955e-07, "epoch": 2.759160839160839, "percentage": 91.96, "elapsed_time": "9:49:28", "remaining_time": "0:51:30"} +{"current_steps": 4934, "total_steps": 5364, "loss": 0.061837293207645416, "lr": 3.908589552934261e-07, "epoch": 2.7597202797202796, "percentage": 91.98, "elapsed_time": "9:49:35", "remaining_time": "0:51:22"} +{"current_steps": 4935, "total_steps": 5364, "loss": 0.015514417551457882, "lr": 3.890591685561673e-07, "epoch": 2.76027972027972, "percentage": 92.0, "elapsed_time": "9:49:42", "remaining_time": "0:51:15"} +{"current_steps": 4936, "total_steps": 5364, "loss": 0.17509452998638153, "lr": 3.872634529120256e-07, "epoch": 2.760839160839161, "percentage": 92.02, "elapsed_time": "9:49:49", "remaining_time": "0:51:08"} +{"current_steps": 4937, "total_steps": 5364, "loss": 0.1754770576953888, "lr": 3.8547180912164803e-07, "epoch": 2.7613986013986014, "percentage": 92.04, "elapsed_time": "9:49:56", "remaining_time": "0:51:01"} +{"current_steps": 4938, "total_steps": 5364, "loss": 0.028663069009780884, "lr": 3.836842379439531e-07, "epoch": 2.761958041958042, "percentage": 92.06, "elapsed_time": "9:50:04", "remaining_time": "0:50:54"} +{"current_steps": 4939, "total_steps": 5364, "loss": 0.06737642735242844, "lr": 3.8190074013614184e-07, "epoch": 2.7625174825174827, "percentage": 92.08, "elapsed_time": "9:50:11", "remaining_time": "0:50:47"} +{"current_steps": 4940, "total_steps": 5364, "loss": 0.07121893018484116, "lr": 3.80121316453681e-07, "epoch": 2.763076923076923, "percentage": 92.1, "elapsed_time": "9:50:18", "remaining_time": "0:50:39"} +{"current_steps": 4941, "total_steps": 5364, "loss": 0.16411644220352173, "lr": 3.783459676503165e-07, "epoch": 2.7636363636363637, "percentage": 92.11, "elapsed_time": "9:50:25", "remaining_time": "0:50:32"} +{"current_steps": 4942, "total_steps": 5364, "loss": 0.07366091012954712, "lr": 3.765746944780702e-07, "epoch": 2.764195804195804, "percentage": 92.13, "elapsed_time": "9:50:32", "remaining_time": "0:50:25"} +{"current_steps": 4943, "total_steps": 5364, "loss": 0.08414723724126816, "lr": 3.7480749768722957e-07, "epoch": 2.7647552447552446, "percentage": 92.15, "elapsed_time": "9:50:39", "remaining_time": "0:50:18"} +{"current_steps": 4944, "total_steps": 5364, "loss": 0.04255523160099983, "lr": 3.7304437802636486e-07, "epoch": 2.765314685314685, "percentage": 92.17, "elapsed_time": "9:50:46", "remaining_time": "0:50:11"} +{"current_steps": 4945, "total_steps": 5364, "loss": 0.051001179963350296, "lr": 3.71285336242313e-07, "epoch": 2.765874125874126, "percentage": 92.19, "elapsed_time": "9:50:54", "remaining_time": "0:50:04"} +{"current_steps": 4946, "total_steps": 5364, "loss": 0.01567598059773445, "lr": 3.695303730801869e-07, "epoch": 2.7664335664335664, "percentage": 92.21, "elapsed_time": "9:51:01", "remaining_time": "0:49:56"} +{"current_steps": 4947, "total_steps": 5364, "loss": 0.08082646876573563, "lr": 3.677794892833697e-07, "epoch": 2.766993006993007, "percentage": 92.23, "elapsed_time": "9:51:08", "remaining_time": "0:49:49"} +{"current_steps": 4948, "total_steps": 5364, "loss": 0.12716718018054962, "lr": 3.66032685593517e-07, "epoch": 2.767552447552448, "percentage": 92.24, "elapsed_time": "9:51:15", "remaining_time": "0:49:42"} +{"current_steps": 4949, "total_steps": 5364, "loss": 0.11594318598508835, "lr": 3.6428996275055585e-07, "epoch": 2.7681118881118882, "percentage": 92.26, "elapsed_time": "9:51:22", "remaining_time": "0:49:35"} +{"current_steps": 4950, "total_steps": 5364, "loss": 0.0710601657629013, "lr": 3.6255132149268566e-07, "epoch": 2.7686713286713287, "percentage": 92.28, "elapsed_time": "9:51:29", "remaining_time": "0:49:28"} +{"current_steps": 4951, "total_steps": 5364, "loss": 0.12579065561294556, "lr": 3.6081676255637743e-07, "epoch": 2.769230769230769, "percentage": 92.3, "elapsed_time": "9:51:36", "remaining_time": "0:49:21"} +{"current_steps": 4952, "total_steps": 5364, "loss": 0.03872643783688545, "lr": 3.5908628667637227e-07, "epoch": 2.7697902097902096, "percentage": 92.32, "elapsed_time": "9:51:43", "remaining_time": "0:49:13"} +{"current_steps": 4953, "total_steps": 5364, "loss": 0.04524435102939606, "lr": 3.5735989458567935e-07, "epoch": 2.77034965034965, "percentage": 92.34, "elapsed_time": "9:51:50", "remaining_time": "0:49:06"} +{"current_steps": 4954, "total_steps": 5364, "loss": 0.0766759067773819, "lr": 3.556375870155815e-07, "epoch": 2.770909090909091, "percentage": 92.36, "elapsed_time": "9:51:58", "remaining_time": "0:48:59"} +{"current_steps": 4955, "total_steps": 5364, "loss": 0.13944050669670105, "lr": 3.539193646956307e-07, "epoch": 2.7714685314685314, "percentage": 92.38, "elapsed_time": "9:52:05", "remaining_time": "0:48:52"} +{"current_steps": 4956, "total_steps": 5364, "loss": 0.0949394553899765, "lr": 3.522052283536459e-07, "epoch": 2.772027972027972, "percentage": 92.39, "elapsed_time": "9:52:12", "remaining_time": "0:48:45"} +{"current_steps": 4957, "total_steps": 5364, "loss": 0.022415973246097565, "lr": 3.504951787157207e-07, "epoch": 2.772587412587413, "percentage": 92.41, "elapsed_time": "9:52:20", "remaining_time": "0:48:38"} +{"current_steps": 4958, "total_steps": 5364, "loss": 0.16931122541427612, "lr": 3.4878921650621013e-07, "epoch": 2.7731468531468533, "percentage": 92.43, "elapsed_time": "9:52:26", "remaining_time": "0:48:30"} +{"current_steps": 4959, "total_steps": 5364, "loss": 0.08383970707654953, "lr": 3.470873424477428e-07, "epoch": 2.7737062937062937, "percentage": 92.45, "elapsed_time": "9:52:34", "remaining_time": "0:48:23"} +{"current_steps": 4960, "total_steps": 5364, "loss": 0.024366505444049835, "lr": 3.4538955726121536e-07, "epoch": 2.774265734265734, "percentage": 92.47, "elapsed_time": "9:52:40", "remaining_time": "0:48:16"} +{"current_steps": 4961, "total_steps": 5364, "loss": 0.060419823974370956, "lr": 3.4369586166579263e-07, "epoch": 2.7748251748251747, "percentage": 92.49, "elapsed_time": "9:52:47", "remaining_time": "0:48:09"} +{"current_steps": 4962, "total_steps": 5364, "loss": 0.3101741075515747, "lr": 3.4200625637890394e-07, "epoch": 2.775384615384615, "percentage": 92.51, "elapsed_time": "9:52:54", "remaining_time": "0:48:02"} +{"current_steps": 4963, "total_steps": 5364, "loss": 0.05042243376374245, "lr": 3.403207421162513e-07, "epoch": 2.775944055944056, "percentage": 92.52, "elapsed_time": "9:53:02", "remaining_time": "0:47:54"} +{"current_steps": 4964, "total_steps": 5364, "loss": 0.11928140372037888, "lr": 3.3863931959179697e-07, "epoch": 2.7765034965034965, "percentage": 92.54, "elapsed_time": "9:53:09", "remaining_time": "0:47:47"} +{"current_steps": 4965, "total_steps": 5364, "loss": 0.06877477467060089, "lr": 3.369619895177756e-07, "epoch": 2.777062937062937, "percentage": 92.56, "elapsed_time": "9:53:15", "remaining_time": "0:47:40"} +{"current_steps": 4966, "total_steps": 5364, "loss": 0.11583748459815979, "lr": 3.352887526046877e-07, "epoch": 2.777622377622378, "percentage": 92.58, "elapsed_time": "9:53:22", "remaining_time": "0:47:33"} +{"current_steps": 4967, "total_steps": 5364, "loss": 0.030591977760195732, "lr": 3.3361960956129644e-07, "epoch": 2.7781818181818183, "percentage": 92.6, "elapsed_time": "9:53:29", "remaining_time": "0:47:26"} +{"current_steps": 4968, "total_steps": 5364, "loss": 0.0904340073466301, "lr": 3.319545610946362e-07, "epoch": 2.7787412587412588, "percentage": 92.62, "elapsed_time": "9:53:37", "remaining_time": "0:47:19"} +{"current_steps": 4969, "total_steps": 5364, "loss": 0.2623438239097595, "lr": 3.302936079100005e-07, "epoch": 2.7793006993006992, "percentage": 92.64, "elapsed_time": "9:53:44", "remaining_time": "0:47:11"} +{"current_steps": 4970, "total_steps": 5364, "loss": 0.04227359592914581, "lr": 3.286367507109545e-07, "epoch": 2.7798601398601397, "percentage": 92.65, "elapsed_time": "9:53:51", "remaining_time": "0:47:04"} +{"current_steps": 4971, "total_steps": 5364, "loss": 0.08644302934408188, "lr": 3.269839901993244e-07, "epoch": 2.78041958041958, "percentage": 92.67, "elapsed_time": "9:53:59", "remaining_time": "0:46:57"} +{"current_steps": 4972, "total_steps": 5364, "loss": 0.03157906234264374, "lr": 3.253353270751991e-07, "epoch": 2.780979020979021, "percentage": 92.69, "elapsed_time": "9:54:06", "remaining_time": "0:46:50"} +{"current_steps": 4973, "total_steps": 5364, "loss": 0.07002183794975281, "lr": 3.2369076203694114e-07, "epoch": 2.7815384615384615, "percentage": 92.71, "elapsed_time": "9:54:13", "remaining_time": "0:46:43"} +{"current_steps": 4974, "total_steps": 5364, "loss": 0.04069294035434723, "lr": 3.220502957811655e-07, "epoch": 2.782097902097902, "percentage": 92.73, "elapsed_time": "9:54:20", "remaining_time": "0:46:36"} +{"current_steps": 4975, "total_steps": 5364, "loss": 0.03844890370965004, "lr": 3.204139290027586e-07, "epoch": 2.782657342657343, "percentage": 92.75, "elapsed_time": "9:54:27", "remaining_time": "0:46:28"} +{"current_steps": 4976, "total_steps": 5364, "loss": 0.1076214462518692, "lr": 3.187816623948692e-07, "epoch": 2.7832167832167833, "percentage": 92.77, "elapsed_time": "9:54:34", "remaining_time": "0:46:21"} +{"current_steps": 4977, "total_steps": 5364, "loss": 0.057912588119506836, "lr": 3.1715349664890425e-07, "epoch": 2.783776223776224, "percentage": 92.79, "elapsed_time": "9:54:42", "remaining_time": "0:46:14"} +{"current_steps": 4978, "total_steps": 5364, "loss": 0.14520801603794098, "lr": 3.1552943245454215e-07, "epoch": 2.7843356643356643, "percentage": 92.8, "elapsed_time": "9:54:49", "remaining_time": "0:46:07"} +{"current_steps": 4979, "total_steps": 5364, "loss": 0.05710513889789581, "lr": 3.13909470499717e-07, "epoch": 2.7848951048951047, "percentage": 92.82, "elapsed_time": "9:54:55", "remaining_time": "0:46:00"} +{"current_steps": 4980, "total_steps": 5364, "loss": 0.017719509080052376, "lr": 3.122936114706276e-07, "epoch": 2.785454545454545, "percentage": 92.84, "elapsed_time": "9:55:03", "remaining_time": "0:45:53"} +{"current_steps": 4981, "total_steps": 5364, "loss": 0.027054881677031517, "lr": 3.106818560517366e-07, "epoch": 2.786013986013986, "percentage": 92.86, "elapsed_time": "9:55:10", "remaining_time": "0:45:45"} +{"current_steps": 4982, "total_steps": 5364, "loss": 0.02746594324707985, "lr": 3.0907420492576334e-07, "epoch": 2.7865734265734265, "percentage": 92.88, "elapsed_time": "9:55:17", "remaining_time": "0:45:38"} +{"current_steps": 4983, "total_steps": 5364, "loss": 0.1467277556657791, "lr": 3.074706587736953e-07, "epoch": 2.787132867132867, "percentage": 92.9, "elapsed_time": "9:55:25", "remaining_time": "0:45:31"} +{"current_steps": 4984, "total_steps": 5364, "loss": 0.08682317286729813, "lr": 3.0587121827477473e-07, "epoch": 2.787692307692308, "percentage": 92.92, "elapsed_time": "9:55:32", "remaining_time": "0:45:24"} +{"current_steps": 4985, "total_steps": 5364, "loss": 0.050984062254428864, "lr": 3.0427588410650965e-07, "epoch": 2.7882517482517484, "percentage": 92.93, "elapsed_time": "9:55:40", "remaining_time": "0:45:17"} +{"current_steps": 4986, "total_steps": 5364, "loss": 0.06781367212533951, "lr": 3.026846569446673e-07, "epoch": 2.788811188811189, "percentage": 92.95, "elapsed_time": "9:55:47", "remaining_time": "0:45:10"} +{"current_steps": 4987, "total_steps": 5364, "loss": 0.16698311269283295, "lr": 3.0109753746327183e-07, "epoch": 2.7893706293706293, "percentage": 92.97, "elapsed_time": "9:55:54", "remaining_time": "0:45:02"} +{"current_steps": 4988, "total_steps": 5364, "loss": 0.13171671330928802, "lr": 2.995145263346133e-07, "epoch": 2.7899300699300698, "percentage": 92.99, "elapsed_time": "9:56:02", "remaining_time": "0:44:55"} +{"current_steps": 4989, "total_steps": 5364, "loss": 0.1539926677942276, "lr": 2.979356242292375e-07, "epoch": 2.79048951048951, "percentage": 93.01, "elapsed_time": "9:56:09", "remaining_time": "0:44:48"} +{"current_steps": 4990, "total_steps": 5364, "loss": 0.027205422520637512, "lr": 2.9636083181595053e-07, "epoch": 2.791048951048951, "percentage": 93.03, "elapsed_time": "9:56:16", "remaining_time": "0:44:41"} +{"current_steps": 4991, "total_steps": 5364, "loss": 0.04700343683362007, "lr": 2.9479014976181775e-07, "epoch": 2.7916083916083916, "percentage": 93.05, "elapsed_time": "9:56:23", "remaining_time": "0:44:34"} +{"current_steps": 4992, "total_steps": 5364, "loss": 0.05692625418305397, "lr": 2.9322357873216245e-07, "epoch": 2.792167832167832, "percentage": 93.06, "elapsed_time": "9:56:30", "remaining_time": "0:44:27"} +{"current_steps": 4993, "total_steps": 5364, "loss": 0.12826746702194214, "lr": 2.9166111939057053e-07, "epoch": 2.792727272727273, "percentage": 93.08, "elapsed_time": "9:56:37", "remaining_time": "0:44:19"} +{"current_steps": 4994, "total_steps": 5364, "loss": 0.11037755757570267, "lr": 2.901027723988803e-07, "epoch": 2.7932867132867134, "percentage": 93.1, "elapsed_time": "9:56:44", "remaining_time": "0:44:12"} +{"current_steps": 4995, "total_steps": 5364, "loss": 0.06750046461820602, "lr": 2.8854853841719265e-07, "epoch": 2.793846153846154, "percentage": 93.12, "elapsed_time": "9:56:51", "remaining_time": "0:44:05"} +{"current_steps": 4996, "total_steps": 5364, "loss": 0.12424492835998535, "lr": 2.869984181038654e-07, "epoch": 2.7944055944055943, "percentage": 93.14, "elapsed_time": "9:56:58", "remaining_time": "0:43:58"} +{"current_steps": 4997, "total_steps": 5364, "loss": 0.11639942973852158, "lr": 2.854524121155111e-07, "epoch": 2.794965034965035, "percentage": 93.16, "elapsed_time": "9:57:05", "remaining_time": "0:43:51"} +{"current_steps": 4998, "total_steps": 5364, "loss": 0.13956589996814728, "lr": 2.839105211070026e-07, "epoch": 2.7955244755244752, "percentage": 93.18, "elapsed_time": "9:57:12", "remaining_time": "0:43:43"} +{"current_steps": 4999, "total_steps": 5364, "loss": 0.05117888003587723, "lr": 2.8237274573146955e-07, "epoch": 2.796083916083916, "percentage": 93.2, "elapsed_time": "9:57:19", "remaining_time": "0:43:36"} +{"current_steps": 5000, "total_steps": 5364, "loss": 0.07312138378620148, "lr": 2.808390866402955e-07, "epoch": 2.7966433566433566, "percentage": 93.21, "elapsed_time": "9:57:27", "remaining_time": "0:43:29"} +{"current_steps": 5001, "total_steps": 5364, "loss": 0.10789519548416138, "lr": 2.793095444831229e-07, "epoch": 2.797202797202797, "percentage": 93.23, "elapsed_time": "9:58:02", "remaining_time": "0:43:24"} +{"current_steps": 5002, "total_steps": 5364, "loss": 0.10860700905323029, "lr": 2.7778411990785124e-07, "epoch": 2.797762237762238, "percentage": 93.25, "elapsed_time": "9:58:09", "remaining_time": "0:43:17"} +{"current_steps": 5003, "total_steps": 5364, "loss": 0.15171238780021667, "lr": 2.762628135606327e-07, "epoch": 2.7983216783216784, "percentage": 93.27, "elapsed_time": "9:58:16", "remaining_time": "0:43:10"} +{"current_steps": 5004, "total_steps": 5364, "loss": 0.019601235166192055, "lr": 2.747456260858761e-07, "epoch": 2.798881118881119, "percentage": 93.29, "elapsed_time": "9:58:23", "remaining_time": "0:43:02"} +{"current_steps": 5005, "total_steps": 5364, "loss": 0.06370805203914642, "lr": 2.732325581262474e-07, "epoch": 2.7994405594405594, "percentage": 93.31, "elapsed_time": "9:58:30", "remaining_time": "0:42:55"} +{"current_steps": 5006, "total_steps": 5364, "loss": 0.14868727326393127, "lr": 2.7172361032266504e-07, "epoch": 2.8, "percentage": 93.33, "elapsed_time": "9:58:37", "remaining_time": "0:42:48"} +{"current_steps": 5007, "total_steps": 5364, "loss": 0.012390978634357452, "lr": 2.702187833143055e-07, "epoch": 2.8005594405594403, "percentage": 93.34, "elapsed_time": "9:58:44", "remaining_time": "0:42:41"} +{"current_steps": 5008, "total_steps": 5364, "loss": 0.09371516853570938, "lr": 2.687180777385956e-07, "epoch": 2.801118881118881, "percentage": 93.36, "elapsed_time": "9:58:51", "remaining_time": "0:42:34"} +{"current_steps": 5009, "total_steps": 5364, "loss": 0.10727708041667938, "lr": 2.6722149423121903e-07, "epoch": 2.8016783216783216, "percentage": 93.38, "elapsed_time": "9:58:58", "remaining_time": "0:42:27"} +{"current_steps": 5010, "total_steps": 5364, "loss": 0.08972786366939545, "lr": 2.65729033426112e-07, "epoch": 2.802237762237762, "percentage": 93.4, "elapsed_time": "9:59:05", "remaining_time": "0:42:19"} +{"current_steps": 5011, "total_steps": 5364, "loss": 0.049296434968709946, "lr": 2.642406959554655e-07, "epoch": 2.802797202797203, "percentage": 93.42, "elapsed_time": "9:59:12", "remaining_time": "0:42:12"} +{"current_steps": 5012, "total_steps": 5364, "loss": 0.16677223145961761, "lr": 2.627564824497253e-07, "epoch": 2.8033566433566435, "percentage": 93.44, "elapsed_time": "9:59:19", "remaining_time": "0:42:05"} +{"current_steps": 5013, "total_steps": 5364, "loss": 0.043198127299547195, "lr": 2.6127639353758726e-07, "epoch": 2.803916083916084, "percentage": 93.46, "elapsed_time": "9:59:26", "remaining_time": "0:41:58"} +{"current_steps": 5014, "total_steps": 5364, "loss": 0.01802062802016735, "lr": 2.5980042984599994e-07, "epoch": 2.8044755244755244, "percentage": 93.48, "elapsed_time": "9:59:34", "remaining_time": "0:41:51"} +{"current_steps": 5015, "total_steps": 5364, "loss": 0.07416543364524841, "lr": 2.583285920001677e-07, "epoch": 2.805034965034965, "percentage": 93.49, "elapsed_time": "9:59:41", "remaining_time": "0:41:43"} +{"current_steps": 5016, "total_steps": 5364, "loss": 0.08955828845500946, "lr": 2.568608806235462e-07, "epoch": 2.8055944055944058, "percentage": 93.51, "elapsed_time": "9:59:48", "remaining_time": "0:41:36"} +{"current_steps": 5017, "total_steps": 5364, "loss": 0.05864696949720383, "lr": 2.553972963378415e-07, "epoch": 2.806153846153846, "percentage": 93.53, "elapsed_time": "9:59:56", "remaining_time": "0:41:29"} +{"current_steps": 5018, "total_steps": 5364, "loss": 0.12506571412086487, "lr": 2.53937839763011e-07, "epoch": 2.8067132867132867, "percentage": 93.55, "elapsed_time": "10:00:03", "remaining_time": "0:41:22"} +{"current_steps": 5019, "total_steps": 5364, "loss": 0.09943557530641556, "lr": 2.52482511517268e-07, "epoch": 2.807272727272727, "percentage": 93.57, "elapsed_time": "10:00:10", "remaining_time": "0:41:15"} +{"current_steps": 5020, "total_steps": 5364, "loss": 0.19397728145122528, "lr": 2.510313122170738e-07, "epoch": 2.807832167832168, "percentage": 93.59, "elapsed_time": "10:00:17", "remaining_time": "0:41:08"} +{"current_steps": 5021, "total_steps": 5364, "loss": 0.11500547081232071, "lr": 2.4958424247713776e-07, "epoch": 2.8083916083916085, "percentage": 93.61, "elapsed_time": "10:00:24", "remaining_time": "0:41:00"} +{"current_steps": 5022, "total_steps": 5364, "loss": 0.08655523508787155, "lr": 2.481413029104274e-07, "epoch": 2.808951048951049, "percentage": 93.62, "elapsed_time": "10:00:31", "remaining_time": "0:40:53"} +{"current_steps": 5023, "total_steps": 5364, "loss": 0.025496497750282288, "lr": 2.467024941281537e-07, "epoch": 2.8095104895104894, "percentage": 93.64, "elapsed_time": "10:00:38", "remaining_time": "0:40:46"} +{"current_steps": 5024, "total_steps": 5364, "loss": 0.0329664871096611, "lr": 2.4526781673978263e-07, "epoch": 2.81006993006993, "percentage": 93.66, "elapsed_time": "10:00:44", "remaining_time": "0:40:39"} +{"current_steps": 5025, "total_steps": 5364, "loss": 0.06955914199352264, "lr": 2.438372713530279e-07, "epoch": 2.810629370629371, "percentage": 93.68, "elapsed_time": "10:00:53", "remaining_time": "0:40:32"} +{"current_steps": 5026, "total_steps": 5364, "loss": 0.051360834389925, "lr": 2.424108585738527e-07, "epoch": 2.8111888111888113, "percentage": 93.7, "elapsed_time": "10:01:00", "remaining_time": "0:40:25"} +{"current_steps": 5027, "total_steps": 5364, "loss": 0.03358414024114609, "lr": 2.409885790064725e-07, "epoch": 2.8117482517482517, "percentage": 93.72, "elapsed_time": "10:01:06", "remaining_time": "0:40:17"} +{"current_steps": 5028, "total_steps": 5364, "loss": 0.06066404655575752, "lr": 2.3957043325334883e-07, "epoch": 2.812307692307692, "percentage": 93.74, "elapsed_time": "10:01:13", "remaining_time": "0:40:10"} +{"current_steps": 5029, "total_steps": 5364, "loss": 0.07924894988536835, "lr": 2.3815642191519107e-07, "epoch": 2.812867132867133, "percentage": 93.75, "elapsed_time": "10:01:20", "remaining_time": "0:40:03"} +{"current_steps": 5030, "total_steps": 5364, "loss": 0.10304653644561768, "lr": 2.3674654559096344e-07, "epoch": 2.8134265734265735, "percentage": 93.77, "elapsed_time": "10:01:27", "remaining_time": "0:39:56"} +{"current_steps": 5031, "total_steps": 5364, "loss": 0.09917331486940384, "lr": 2.3534080487787158e-07, "epoch": 2.813986013986014, "percentage": 93.79, "elapsed_time": "10:01:34", "remaining_time": "0:39:49"} +{"current_steps": 5032, "total_steps": 5364, "loss": 0.12496791034936905, "lr": 2.3393920037137474e-07, "epoch": 2.8145454545454545, "percentage": 93.81, "elapsed_time": "10:01:41", "remaining_time": "0:39:41"} +{"current_steps": 5033, "total_steps": 5364, "loss": 0.13458216190338135, "lr": 2.3254173266517466e-07, "epoch": 2.815104895104895, "percentage": 93.83, "elapsed_time": "10:01:48", "remaining_time": "0:39:34"} +{"current_steps": 5034, "total_steps": 5364, "loss": 0.08030325174331665, "lr": 2.3114840235122783e-07, "epoch": 2.815664335664336, "percentage": 93.85, "elapsed_time": "10:01:55", "remaining_time": "0:39:27"} +{"current_steps": 5035, "total_steps": 5364, "loss": 0.08696933090686798, "lr": 2.2975921001973212e-07, "epoch": 2.8162237762237763, "percentage": 93.87, "elapsed_time": "10:02:01", "remaining_time": "0:39:20"} +{"current_steps": 5036, "total_steps": 5364, "loss": 0.18368548154830933, "lr": 2.283741562591324e-07, "epoch": 2.8167832167832167, "percentage": 93.89, "elapsed_time": "10:02:09", "remaining_time": "0:39:13"} +{"current_steps": 5037, "total_steps": 5364, "loss": 0.04156312346458435, "lr": 2.2699324165612824e-07, "epoch": 2.817342657342657, "percentage": 93.9, "elapsed_time": "10:02:16", "remaining_time": "0:39:05"} +{"current_steps": 5038, "total_steps": 5364, "loss": 0.06862244755029678, "lr": 2.2561646679565619e-07, "epoch": 2.817902097902098, "percentage": 93.92, "elapsed_time": "10:02:23", "remaining_time": "0:38:58"} +{"current_steps": 5039, "total_steps": 5364, "loss": 0.045548420399427414, "lr": 2.242438322609053e-07, "epoch": 2.8184615384615386, "percentage": 93.94, "elapsed_time": "10:02:30", "remaining_time": "0:38:51"} +{"current_steps": 5040, "total_steps": 5364, "loss": 0.05208326876163483, "lr": 2.2287533863330824e-07, "epoch": 2.819020979020979, "percentage": 93.96, "elapsed_time": "10:02:38", "remaining_time": "0:38:44"} +{"current_steps": 5041, "total_steps": 5364, "loss": 0.04804254695773125, "lr": 2.2151098649254577e-07, "epoch": 2.8195804195804195, "percentage": 93.98, "elapsed_time": "10:02:45", "remaining_time": "0:38:37"} +{"current_steps": 5042, "total_steps": 5364, "loss": 0.06865841895341873, "lr": 2.2015077641654225e-07, "epoch": 2.82013986013986, "percentage": 94.0, "elapsed_time": "10:02:53", "remaining_time": "0:38:30"} +{"current_steps": 5043, "total_steps": 5364, "loss": 0.081064872443676, "lr": 2.1879470898146792e-07, "epoch": 2.820699300699301, "percentage": 94.02, "elapsed_time": "10:03:00", "remaining_time": "0:38:22"} +{"current_steps": 5044, "total_steps": 5364, "loss": 0.04247250407934189, "lr": 2.1744278476173885e-07, "epoch": 2.8212587412587413, "percentage": 94.03, "elapsed_time": "10:03:07", "remaining_time": "0:38:15"} +{"current_steps": 5045, "total_steps": 5364, "loss": 0.1341962069272995, "lr": 2.1609500433001583e-07, "epoch": 2.821818181818182, "percentage": 94.05, "elapsed_time": "10:03:14", "remaining_time": "0:38:08"} +{"current_steps": 5046, "total_steps": 5364, "loss": 0.056738875806331635, "lr": 2.1475136825720666e-07, "epoch": 2.8223776223776222, "percentage": 94.07, "elapsed_time": "10:03:20", "remaining_time": "0:38:01"} +{"current_steps": 5047, "total_steps": 5364, "loss": 0.0701540932059288, "lr": 2.1341187711245713e-07, "epoch": 2.822937062937063, "percentage": 94.09, "elapsed_time": "10:03:27", "remaining_time": "0:37:54"} +{"current_steps": 5048, "total_steps": 5364, "loss": 0.18930019438266754, "lr": 2.1207653146316454e-07, "epoch": 2.8234965034965036, "percentage": 94.11, "elapsed_time": "10:03:34", "remaining_time": "0:37:47"} +{"current_steps": 5049, "total_steps": 5364, "loss": 0.06333623826503754, "lr": 2.1074533187496638e-07, "epoch": 2.824055944055944, "percentage": 94.13, "elapsed_time": "10:03:42", "remaining_time": "0:37:39"} +{"current_steps": 5050, "total_steps": 5364, "loss": 0.05690986290574074, "lr": 2.0941827891174492e-07, "epoch": 2.8246153846153845, "percentage": 94.15, "elapsed_time": "10:03:48", "remaining_time": "0:37:32"} +{"current_steps": 5051, "total_steps": 5364, "loss": 0.10526102781295776, "lr": 2.080953731356261e-07, "epoch": 2.825174825174825, "percentage": 94.16, "elapsed_time": "10:03:55", "remaining_time": "0:37:25"} +{"current_steps": 5052, "total_steps": 5364, "loss": 0.0016192694893106818, "lr": 2.0677661510697833e-07, "epoch": 2.825734265734266, "percentage": 94.18, "elapsed_time": "10:04:03", "remaining_time": "0:37:18"} +{"current_steps": 5053, "total_steps": 5364, "loss": 0.17949801683425903, "lr": 2.0546200538441363e-07, "epoch": 2.8262937062937064, "percentage": 94.2, "elapsed_time": "10:04:10", "remaining_time": "0:37:11"} +{"current_steps": 5054, "total_steps": 5364, "loss": 0.03927082195878029, "lr": 2.041515445247866e-07, "epoch": 2.826853146853147, "percentage": 94.22, "elapsed_time": "10:04:17", "remaining_time": "0:37:03"} +{"current_steps": 5055, "total_steps": 5364, "loss": 0.027906186878681183, "lr": 2.0284523308319538e-07, "epoch": 2.8274125874125873, "percentage": 94.24, "elapsed_time": "10:04:25", "remaining_time": "0:36:56"} +{"current_steps": 5056, "total_steps": 5364, "loss": 0.10151061415672302, "lr": 2.015430716129796e-07, "epoch": 2.827972027972028, "percentage": 94.26, "elapsed_time": "10:04:32", "remaining_time": "0:36:49"} +{"current_steps": 5057, "total_steps": 5364, "loss": 0.08257726579904556, "lr": 2.0024506066571915e-07, "epoch": 2.8285314685314686, "percentage": 94.28, "elapsed_time": "10:04:39", "remaining_time": "0:36:42"} +{"current_steps": 5058, "total_steps": 5364, "loss": 0.049539223313331604, "lr": 1.9895120079124085e-07, "epoch": 2.829090909090909, "percentage": 94.3, "elapsed_time": "10:04:46", "remaining_time": "0:36:35"} +{"current_steps": 5059, "total_steps": 5364, "loss": 0.09938731789588928, "lr": 1.9766149253760858e-07, "epoch": 2.8296503496503496, "percentage": 94.31, "elapsed_time": "10:04:53", "remaining_time": "0:36:28"} +{"current_steps": 5060, "total_steps": 5364, "loss": 0.060706980526447296, "lr": 1.9637593645112974e-07, "epoch": 2.83020979020979, "percentage": 94.33, "elapsed_time": "10:05:01", "remaining_time": "0:36:20"} +{"current_steps": 5061, "total_steps": 5364, "loss": 0.0917670726776123, "lr": 1.9509453307635096e-07, "epoch": 2.830769230769231, "percentage": 94.35, "elapsed_time": "10:05:08", "remaining_time": "0:36:13"} +{"current_steps": 5062, "total_steps": 5364, "loss": 0.1297025978565216, "lr": 1.938172829560614e-07, "epoch": 2.8313286713286714, "percentage": 94.37, "elapsed_time": "10:05:15", "remaining_time": "0:36:06"} +{"current_steps": 5063, "total_steps": 5364, "loss": 0.07143554091453552, "lr": 1.925441866312927e-07, "epoch": 2.831888111888112, "percentage": 94.39, "elapsed_time": "10:05:22", "remaining_time": "0:35:59"} +{"current_steps": 5064, "total_steps": 5364, "loss": 0.019956976175308228, "lr": 1.9127524464131353e-07, "epoch": 2.8324475524475523, "percentage": 94.41, "elapsed_time": "10:05:29", "remaining_time": "0:35:52"} +{"current_steps": 5065, "total_steps": 5364, "loss": 0.06862765550613403, "lr": 1.900104575236339e-07, "epoch": 2.833006993006993, "percentage": 94.43, "elapsed_time": "10:05:37", "remaining_time": "0:35:45"} +{"current_steps": 5066, "total_steps": 5364, "loss": 0.009724807925522327, "lr": 1.8874982581400636e-07, "epoch": 2.8335664335664337, "percentage": 94.44, "elapsed_time": "10:05:44", "remaining_time": "0:35:37"} +{"current_steps": 5067, "total_steps": 5364, "loss": 0.268549382686615, "lr": 1.874933500464171e-07, "epoch": 2.834125874125874, "percentage": 94.46, "elapsed_time": "10:05:52", "remaining_time": "0:35:30"} +{"current_steps": 5068, "total_steps": 5364, "loss": 0.10982885956764221, "lr": 1.862410307531015e-07, "epoch": 2.8346853146853146, "percentage": 94.48, "elapsed_time": "10:05:59", "remaining_time": "0:35:23"} +{"current_steps": 5069, "total_steps": 5364, "loss": 0.08165832608938217, "lr": 1.8499286846452414e-07, "epoch": 2.835244755244755, "percentage": 94.5, "elapsed_time": "10:06:06", "remaining_time": "0:35:16"} +{"current_steps": 5070, "total_steps": 5364, "loss": 0.0976344496011734, "lr": 1.8374886370939538e-07, "epoch": 2.835804195804196, "percentage": 94.52, "elapsed_time": "10:06:13", "remaining_time": "0:35:09"} +{"current_steps": 5071, "total_steps": 5364, "loss": 0.13575875759124756, "lr": 1.825090170146626e-07, "epoch": 2.8363636363636364, "percentage": 94.54, "elapsed_time": "10:06:21", "remaining_time": "0:35:02"} +{"current_steps": 5072, "total_steps": 5364, "loss": 0.048558276146650314, "lr": 1.812733289055113e-07, "epoch": 2.836923076923077, "percentage": 94.56, "elapsed_time": "10:06:28", "remaining_time": "0:34:54"} +{"current_steps": 5073, "total_steps": 5364, "loss": 0.31836220622062683, "lr": 1.8004179990536607e-07, "epoch": 2.8374825174825173, "percentage": 94.57, "elapsed_time": "10:06:35", "remaining_time": "0:34:47"} +{"current_steps": 5074, "total_steps": 5364, "loss": 0.06775633245706558, "lr": 1.7881443053588966e-07, "epoch": 2.8380419580419582, "percentage": 94.59, "elapsed_time": "10:06:42", "remaining_time": "0:34:40"} +{"current_steps": 5075, "total_steps": 5364, "loss": 0.03139274939894676, "lr": 1.775912213169828e-07, "epoch": 2.8386013986013987, "percentage": 94.61, "elapsed_time": "10:06:49", "remaining_time": "0:34:33"} +{"current_steps": 5076, "total_steps": 5364, "loss": 0.09580185264348984, "lr": 1.763721727667844e-07, "epoch": 2.839160839160839, "percentage": 94.63, "elapsed_time": "10:06:56", "remaining_time": "0:34:26"} +{"current_steps": 5077, "total_steps": 5364, "loss": 0.07367917895317078, "lr": 1.751572854016692e-07, "epoch": 2.8397202797202796, "percentage": 94.65, "elapsed_time": "10:07:03", "remaining_time": "0:34:18"} +{"current_steps": 5078, "total_steps": 5364, "loss": 0.06942617893218994, "lr": 1.7394655973625219e-07, "epoch": 2.84027972027972, "percentage": 94.67, "elapsed_time": "10:07:10", "remaining_time": "0:34:11"} +{"current_steps": 5079, "total_steps": 5364, "loss": 0.018502244725823402, "lr": 1.7273999628338311e-07, "epoch": 2.840839160839161, "percentage": 94.69, "elapsed_time": "10:07:17", "remaining_time": "0:34:04"} +{"current_steps": 5080, "total_steps": 5364, "loss": 0.035788435488939285, "lr": 1.7153759555414984e-07, "epoch": 2.8413986013986015, "percentage": 94.71, "elapsed_time": "10:07:25", "remaining_time": "0:33:57"} +{"current_steps": 5081, "total_steps": 5364, "loss": 0.12827466428279877, "lr": 1.7033935805787715e-07, "epoch": 2.841958041958042, "percentage": 94.72, "elapsed_time": "10:07:32", "remaining_time": "0:33:50"} +{"current_steps": 5082, "total_steps": 5364, "loss": 0.09196305274963379, "lr": 1.6914528430212463e-07, "epoch": 2.8425174825174824, "percentage": 94.74, "elapsed_time": "10:07:39", "remaining_time": "0:33:43"} +{"current_steps": 5083, "total_steps": 5364, "loss": 0.06387708336114883, "lr": 1.6795537479268986e-07, "epoch": 2.8430769230769233, "percentage": 94.76, "elapsed_time": "10:07:46", "remaining_time": "0:33:35"} +{"current_steps": 5084, "total_steps": 5364, "loss": 0.1405351161956787, "lr": 1.6676963003360636e-07, "epoch": 2.8436363636363637, "percentage": 94.78, "elapsed_time": "10:07:53", "remaining_time": "0:33:28"} +{"current_steps": 5085, "total_steps": 5364, "loss": 0.03291900083422661, "lr": 1.6558805052714234e-07, "epoch": 2.844195804195804, "percentage": 94.8, "elapsed_time": "10:08:00", "remaining_time": "0:33:21"} +{"current_steps": 5086, "total_steps": 5364, "loss": 0.06944390386343002, "lr": 1.644106367738041e-07, "epoch": 2.8447552447552447, "percentage": 94.82, "elapsed_time": "10:08:07", "remaining_time": "0:33:14"} +{"current_steps": 5087, "total_steps": 5364, "loss": 0.06560652703046799, "lr": 1.6323738927232824e-07, "epoch": 2.845314685314685, "percentage": 94.84, "elapsed_time": "10:08:14", "remaining_time": "0:33:07"} +{"current_steps": 5088, "total_steps": 5364, "loss": 0.09246764332056046, "lr": 1.620683085196939e-07, "epoch": 2.845874125874126, "percentage": 94.85, "elapsed_time": "10:08:21", "remaining_time": "0:33:00"} +{"current_steps": 5089, "total_steps": 5364, "loss": 0.08510936796665192, "lr": 1.609033950111083e-07, "epoch": 2.8464335664335665, "percentage": 94.87, "elapsed_time": "10:08:29", "remaining_time": "0:32:52"} +{"current_steps": 5090, "total_steps": 5364, "loss": 0.13489529490470886, "lr": 1.597426492400189e-07, "epoch": 2.846993006993007, "percentage": 94.89, "elapsed_time": "10:08:36", "remaining_time": "0:32:45"} +{"current_steps": 5091, "total_steps": 5364, "loss": 0.0592874176800251, "lr": 1.5858607169810468e-07, "epoch": 2.8475524475524474, "percentage": 94.91, "elapsed_time": "10:08:43", "remaining_time": "0:32:38"} +{"current_steps": 5092, "total_steps": 5364, "loss": 0.049627311527729034, "lr": 1.5743366287527816e-07, "epoch": 2.8481118881118883, "percentage": 94.93, "elapsed_time": "10:08:50", "remaining_time": "0:32:31"} +{"current_steps": 5093, "total_steps": 5364, "loss": 0.11759379506111145, "lr": 1.5628542325968887e-07, "epoch": 2.8486713286713288, "percentage": 94.95, "elapsed_time": "10:08:57", "remaining_time": "0:32:24"} +{"current_steps": 5094, "total_steps": 5364, "loss": 0.049187518656253815, "lr": 1.5514135333771996e-07, "epoch": 2.8492307692307692, "percentage": 94.97, "elapsed_time": "10:09:05", "remaining_time": "0:32:17"} +{"current_steps": 5095, "total_steps": 5364, "loss": 0.03734589368104935, "lr": 1.54001453593986e-07, "epoch": 2.8497902097902097, "percentage": 94.99, "elapsed_time": "10:09:12", "remaining_time": "0:32:09"} +{"current_steps": 5096, "total_steps": 5364, "loss": 0.07761257886886597, "lr": 1.528657245113352e-07, "epoch": 2.85034965034965, "percentage": 95.0, "elapsed_time": "10:09:18", "remaining_time": "0:32:02"} +{"current_steps": 5097, "total_steps": 5364, "loss": 0.16581541299819946, "lr": 1.5173416657085271e-07, "epoch": 2.850909090909091, "percentage": 95.02, "elapsed_time": "10:09:26", "remaining_time": "0:31:55"} +{"current_steps": 5098, "total_steps": 5364, "loss": 0.09356231242418289, "lr": 1.5060678025185404e-07, "epoch": 2.8514685314685315, "percentage": 95.04, "elapsed_time": "10:09:34", "remaining_time": "0:31:48"} +{"current_steps": 5099, "total_steps": 5364, "loss": 0.040314048528671265, "lr": 1.494835660318883e-07, "epoch": 2.852027972027972, "percentage": 95.06, "elapsed_time": "10:09:41", "remaining_time": "0:31:41"} +{"current_steps": 5100, "total_steps": 5364, "loss": 0.0820893794298172, "lr": 1.4836452438673599e-07, "epoch": 2.8525874125874124, "percentage": 95.08, "elapsed_time": "10:09:48", "remaining_time": "0:31:33"} +{"current_steps": 5101, "total_steps": 5364, "loss": 0.05582234635949135, "lr": 1.4724965579041239e-07, "epoch": 2.8531468531468533, "percentage": 95.1, "elapsed_time": "10:09:55", "remaining_time": "0:31:26"} +{"current_steps": 5102, "total_steps": 5364, "loss": 0.011546442285180092, "lr": 1.461389607151642e-07, "epoch": 2.853706293706294, "percentage": 95.12, "elapsed_time": "10:10:02", "remaining_time": "0:31:19"} +{"current_steps": 5103, "total_steps": 5364, "loss": 0.15707063674926758, "lr": 1.450324396314684e-07, "epoch": 2.8542657342657343, "percentage": 95.13, "elapsed_time": "10:10:09", "remaining_time": "0:31:12"} +{"current_steps": 5104, "total_steps": 5364, "loss": 0.03148286044597626, "lr": 1.4393009300803785e-07, "epoch": 2.8548251748251747, "percentage": 95.15, "elapsed_time": "10:10:16", "remaining_time": "0:31:05"} +{"current_steps": 5105, "total_steps": 5364, "loss": 0.06608149409294128, "lr": 1.4283192131181456e-07, "epoch": 2.855384615384615, "percentage": 95.17, "elapsed_time": "10:10:23", "remaining_time": "0:30:58"} +{"current_steps": 5106, "total_steps": 5364, "loss": 0.05685903877019882, "lr": 1.417379250079709e-07, "epoch": 2.855944055944056, "percentage": 95.19, "elapsed_time": "10:10:30", "remaining_time": "0:30:50"} +{"current_steps": 5107, "total_steps": 5364, "loss": 0.08568979799747467, "lr": 1.4064810455991507e-07, "epoch": 2.8565034965034966, "percentage": 95.21, "elapsed_time": "10:10:37", "remaining_time": "0:30:43"} +{"current_steps": 5108, "total_steps": 5364, "loss": 0.010590219870209694, "lr": 1.3956246042928002e-07, "epoch": 2.857062937062937, "percentage": 95.23, "elapsed_time": "10:10:45", "remaining_time": "0:30:36"} +{"current_steps": 5109, "total_steps": 5364, "loss": 0.04023498296737671, "lr": 1.3848099307593455e-07, "epoch": 2.8576223776223775, "percentage": 95.25, "elapsed_time": "10:10:52", "remaining_time": "0:30:29"} +{"current_steps": 5110, "total_steps": 5364, "loss": 0.0387575663626194, "lr": 1.374037029579789e-07, "epoch": 2.8581818181818184, "percentage": 95.26, "elapsed_time": "10:10:59", "remaining_time": "0:30:22"} +{"current_steps": 5111, "total_steps": 5364, "loss": 0.06853128969669342, "lr": 1.3633059053173804e-07, "epoch": 2.858741258741259, "percentage": 95.28, "elapsed_time": "10:11:06", "remaining_time": "0:30:15"} +{"current_steps": 5112, "total_steps": 5364, "loss": 0.07619741559028625, "lr": 1.352616562517739e-07, "epoch": 2.8593006993006993, "percentage": 95.3, "elapsed_time": "10:11:13", "remaining_time": "0:30:07"} +{"current_steps": 5113, "total_steps": 5364, "loss": 0.08275170624256134, "lr": 1.3419690057087432e-07, "epoch": 2.8598601398601398, "percentage": 95.32, "elapsed_time": "10:11:20", "remaining_time": "0:30:00"} +{"current_steps": 5114, "total_steps": 5364, "loss": 0.20167924463748932, "lr": 1.331363239400596e-07, "epoch": 2.86041958041958, "percentage": 95.34, "elapsed_time": "10:11:27", "remaining_time": "0:29:53"} +{"current_steps": 5115, "total_steps": 5364, "loss": 0.18678340315818787, "lr": 1.320799268085793e-07, "epoch": 2.860979020979021, "percentage": 95.36, "elapsed_time": "10:11:34", "remaining_time": "0:29:46"} +{"current_steps": 5116, "total_steps": 5364, "loss": 0.0465087890625, "lr": 1.3102770962390875e-07, "epoch": 2.8615384615384616, "percentage": 95.38, "elapsed_time": "10:11:41", "remaining_time": "0:29:39"} +{"current_steps": 5117, "total_steps": 5364, "loss": 0.059813205152750015, "lr": 1.299796728317604e-07, "epoch": 2.862097902097902, "percentage": 95.4, "elapsed_time": "10:11:48", "remaining_time": "0:29:31"} +{"current_steps": 5118, "total_steps": 5364, "loss": 0.24804198741912842, "lr": 1.2893581687606903e-07, "epoch": 2.8626573426573425, "percentage": 95.41, "elapsed_time": "10:11:56", "remaining_time": "0:29:24"} +{"current_steps": 5119, "total_steps": 5364, "loss": 0.0936436876654625, "lr": 1.2789614219900103e-07, "epoch": 2.8632167832167834, "percentage": 95.43, "elapsed_time": "10:12:03", "remaining_time": "0:29:17"} +{"current_steps": 5120, "total_steps": 5364, "loss": 0.08598095178604126, "lr": 1.2686064924095299e-07, "epoch": 2.863776223776224, "percentage": 95.45, "elapsed_time": "10:12:10", "remaining_time": "0:29:10"} +{"current_steps": 5121, "total_steps": 5364, "loss": 0.13691863417625427, "lr": 1.2582933844054733e-07, "epoch": 2.8643356643356643, "percentage": 95.47, "elapsed_time": "10:12:17", "remaining_time": "0:29:03"} +{"current_steps": 5122, "total_steps": 5364, "loss": 0.05046981945633888, "lr": 1.24802210234638e-07, "epoch": 2.864895104895105, "percentage": 95.49, "elapsed_time": "10:12:24", "remaining_time": "0:28:56"} +{"current_steps": 5123, "total_steps": 5364, "loss": 0.05974654480814934, "lr": 1.2377926505830363e-07, "epoch": 2.8654545454545453, "percentage": 95.51, "elapsed_time": "10:12:31", "remaining_time": "0:28:48"} +{"current_steps": 5124, "total_steps": 5364, "loss": 0.09566514194011688, "lr": 1.2276050334485424e-07, "epoch": 2.866013986013986, "percentage": 95.53, "elapsed_time": "10:12:38", "remaining_time": "0:28:41"} +{"current_steps": 5125, "total_steps": 5364, "loss": 0.07634823769330978, "lr": 1.2174592552582686e-07, "epoch": 2.8665734265734266, "percentage": 95.54, "elapsed_time": "10:12:45", "remaining_time": "0:28:34"} +{"current_steps": 5126, "total_steps": 5364, "loss": 0.099985271692276, "lr": 1.2073553203098552e-07, "epoch": 2.867132867132867, "percentage": 95.56, "elapsed_time": "10:12:52", "remaining_time": "0:28:27"} +{"current_steps": 5127, "total_steps": 5364, "loss": 0.09185055643320084, "lr": 1.197293232883212e-07, "epoch": 2.8676923076923075, "percentage": 95.58, "elapsed_time": "10:12:59", "remaining_time": "0:28:20"} +{"current_steps": 5128, "total_steps": 5364, "loss": 0.071977898478508, "lr": 1.1872729972405517e-07, "epoch": 2.8682517482517484, "percentage": 95.6, "elapsed_time": "10:13:06", "remaining_time": "0:28:13"} +{"current_steps": 5129, "total_steps": 5364, "loss": 0.08545494079589844, "lr": 1.1772946176263345e-07, "epoch": 2.868811188811189, "percentage": 95.62, "elapsed_time": "10:13:13", "remaining_time": "0:28:05"} +{"current_steps": 5130, "total_steps": 5364, "loss": 0.04457031190395355, "lr": 1.1673580982672905e-07, "epoch": 2.8693706293706294, "percentage": 95.64, "elapsed_time": "10:13:20", "remaining_time": "0:27:58"} +{"current_steps": 5131, "total_steps": 5364, "loss": 0.09940159320831299, "lr": 1.1574634433724307e-07, "epoch": 2.86993006993007, "percentage": 95.66, "elapsed_time": "10:13:28", "remaining_time": "0:27:51"} +{"current_steps": 5132, "total_steps": 5364, "loss": 0.12142953276634216, "lr": 1.1476106571330247e-07, "epoch": 2.8704895104895103, "percentage": 95.67, "elapsed_time": "10:13:35", "remaining_time": "0:27:44"} +{"current_steps": 5133, "total_steps": 5364, "loss": 0.09178759157657623, "lr": 1.1377997437226008e-07, "epoch": 2.871048951048951, "percentage": 95.69, "elapsed_time": "10:13:42", "remaining_time": "0:27:37"} +{"current_steps": 5134, "total_steps": 5364, "loss": 0.045527152717113495, "lr": 1.1280307072969677e-07, "epoch": 2.8716083916083917, "percentage": 95.71, "elapsed_time": "10:13:49", "remaining_time": "0:27:29"} +{"current_steps": 5135, "total_steps": 5364, "loss": 0.09320954978466034, "lr": 1.118303551994182e-07, "epoch": 2.872167832167832, "percentage": 95.73, "elapsed_time": "10:13:56", "remaining_time": "0:27:22"} +{"current_steps": 5136, "total_steps": 5364, "loss": 0.027723446488380432, "lr": 1.1086182819345703e-07, "epoch": 2.8727272727272726, "percentage": 95.75, "elapsed_time": "10:14:02", "remaining_time": "0:27:15"} +{"current_steps": 5137, "total_steps": 5364, "loss": 0.19189725816249847, "lr": 1.0989749012207063e-07, "epoch": 2.8732867132867135, "percentage": 95.77, "elapsed_time": "10:14:10", "remaining_time": "0:27:08"} +{"current_steps": 5138, "total_steps": 5364, "loss": 0.041648589074611664, "lr": 1.0893734139374223e-07, "epoch": 2.873846153846154, "percentage": 95.79, "elapsed_time": "10:14:17", "remaining_time": "0:27:01"} +{"current_steps": 5139, "total_steps": 5364, "loss": 0.05024919658899307, "lr": 1.0798138241518097e-07, "epoch": 2.8744055944055944, "percentage": 95.81, "elapsed_time": "10:14:25", "remaining_time": "0:26:54"} +{"current_steps": 5140, "total_steps": 5364, "loss": 0.05095798522233963, "lr": 1.0702961359132069e-07, "epoch": 2.874965034965035, "percentage": 95.82, "elapsed_time": "10:14:32", "remaining_time": "0:26:46"} +{"current_steps": 5141, "total_steps": 5364, "loss": 0.03731197863817215, "lr": 1.0608203532532224e-07, "epoch": 2.8755244755244753, "percentage": 95.84, "elapsed_time": "10:14:39", "remaining_time": "0:26:39"} +{"current_steps": 5142, "total_steps": 5364, "loss": 0.24399888515472412, "lr": 1.0513864801856677e-07, "epoch": 2.8760839160839162, "percentage": 95.86, "elapsed_time": "10:14:46", "remaining_time": "0:26:32"} +{"current_steps": 5143, "total_steps": 5364, "loss": 0.04520170018076897, "lr": 1.0419945207066573e-07, "epoch": 2.8766433566433567, "percentage": 95.88, "elapsed_time": "10:14:54", "remaining_time": "0:26:25"} +{"current_steps": 5144, "total_steps": 5364, "loss": 0.03667888790369034, "lr": 1.03264447879452e-07, "epoch": 2.877202797202797, "percentage": 95.9, "elapsed_time": "10:15:01", "remaining_time": "0:26:18"} +{"current_steps": 5145, "total_steps": 5364, "loss": 0.05186565965414047, "lr": 1.0233363584098211e-07, "epoch": 2.8777622377622376, "percentage": 95.92, "elapsed_time": "10:15:08", "remaining_time": "0:26:11"} +{"current_steps": 5146, "total_steps": 5364, "loss": 0.07279538363218307, "lr": 1.0140701634953954e-07, "epoch": 2.8783216783216785, "percentage": 95.94, "elapsed_time": "10:15:15", "remaining_time": "0:26:03"} +{"current_steps": 5147, "total_steps": 5364, "loss": 0.05483987182378769, "lr": 1.0048458979763031e-07, "epoch": 2.878881118881119, "percentage": 95.95, "elapsed_time": "10:15:23", "remaining_time": "0:25:56"} +{"current_steps": 5148, "total_steps": 5364, "loss": 0.08227822184562683, "lr": 9.956635657598524e-08, "epoch": 2.8794405594405594, "percentage": 95.97, "elapsed_time": "10:15:29", "remaining_time": "0:25:49"} +{"current_steps": 5149, "total_steps": 5364, "loss": 0.03304259851574898, "lr": 9.86523170735576e-08, "epoch": 2.88, "percentage": 95.99, "elapsed_time": "10:15:36", "remaining_time": "0:25:42"} +{"current_steps": 5150, "total_steps": 5364, "loss": 0.10360705852508545, "lr": 9.774247167752326e-08, "epoch": 2.8805594405594404, "percentage": 96.01, "elapsed_time": "10:15:43", "remaining_time": "0:25:35"} +{"current_steps": 5151, "total_steps": 5364, "loss": 0.06976587325334549, "lr": 9.683682077328616e-08, "epoch": 2.8811188811188813, "percentage": 96.03, "elapsed_time": "10:15:51", "remaining_time": "0:25:27"} +{"current_steps": 5152, "total_steps": 5364, "loss": 0.18481668829917908, "lr": 9.59353647444683e-08, "epoch": 2.8816783216783217, "percentage": 96.05, "elapsed_time": "10:15:57", "remaining_time": "0:25:20"} +{"current_steps": 5153, "total_steps": 5364, "loss": 0.11155405640602112, "lr": 9.503810397291758e-08, "epoch": 2.882237762237762, "percentage": 96.07, "elapsed_time": "10:16:05", "remaining_time": "0:25:13"} +{"current_steps": 5154, "total_steps": 5364, "loss": 0.16135969758033752, "lr": 9.414503883870552e-08, "epoch": 2.8827972027972026, "percentage": 96.09, "elapsed_time": "10:16:12", "remaining_time": "0:25:06"} +{"current_steps": 5155, "total_steps": 5364, "loss": 0.13592611253261566, "lr": 9.325616972012175e-08, "epoch": 2.8833566433566435, "percentage": 96.1, "elapsed_time": "10:16:19", "remaining_time": "0:24:59"} +{"current_steps": 5156, "total_steps": 5364, "loss": 0.08500874787569046, "lr": 9.237149699368508e-08, "epoch": 2.883916083916084, "percentage": 96.12, "elapsed_time": "10:16:26", "remaining_time": "0:24:52"} +{"current_steps": 5157, "total_steps": 5364, "loss": 0.1580979824066162, "lr": 9.149102103413243e-08, "epoch": 2.8844755244755245, "percentage": 96.14, "elapsed_time": "10:16:32", "remaining_time": "0:24:44"} +{"current_steps": 5158, "total_steps": 5364, "loss": 0.12007184326648712, "lr": 9.061474221442324e-08, "epoch": 2.885034965034965, "percentage": 96.16, "elapsed_time": "10:16:39", "remaining_time": "0:24:37"} +{"current_steps": 5159, "total_steps": 5364, "loss": 0.018520301207900047, "lr": 8.97426609057428e-08, "epoch": 2.8855944055944054, "percentage": 96.18, "elapsed_time": "10:16:46", "remaining_time": "0:24:30"} +{"current_steps": 5160, "total_steps": 5364, "loss": 0.060542263090610504, "lr": 8.887477747749118e-08, "epoch": 2.8861538461538463, "percentage": 96.2, "elapsed_time": "10:16:53", "remaining_time": "0:24:23"} +{"current_steps": 5161, "total_steps": 5364, "loss": 0.08037333935499191, "lr": 8.801109229729877e-08, "epoch": 2.8867132867132868, "percentage": 96.22, "elapsed_time": "10:17:01", "remaining_time": "0:24:16"} +{"current_steps": 5162, "total_steps": 5364, "loss": 0.059672582894563675, "lr": 8.715160573101067e-08, "epoch": 2.887272727272727, "percentage": 96.23, "elapsed_time": "10:17:08", "remaining_time": "0:24:09"} +{"current_steps": 5163, "total_steps": 5364, "loss": 0.0336616113781929, "lr": 8.62963181426979e-08, "epoch": 2.8878321678321677, "percentage": 96.25, "elapsed_time": "10:17:16", "remaining_time": "0:24:01"} +{"current_steps": 5164, "total_steps": 5364, "loss": 0.0680554062128067, "lr": 8.544522989465064e-08, "epoch": 2.8883916083916086, "percentage": 96.27, "elapsed_time": "10:17:23", "remaining_time": "0:23:54"} +{"current_steps": 5165, "total_steps": 5364, "loss": 0.12276100367307663, "lr": 8.459834134738054e-08, "epoch": 2.888951048951049, "percentage": 96.29, "elapsed_time": "10:17:30", "remaining_time": "0:23:47"} +{"current_steps": 5166, "total_steps": 5364, "loss": 0.10284769535064697, "lr": 8.375565285962173e-08, "epoch": 2.8895104895104895, "percentage": 96.31, "elapsed_time": "10:17:37", "remaining_time": "0:23:40"} +{"current_steps": 5167, "total_steps": 5364, "loss": 0.16166862845420837, "lr": 8.291716478832757e-08, "epoch": 2.89006993006993, "percentage": 96.33, "elapsed_time": "10:17:44", "remaining_time": "0:23:33"} +{"current_steps": 5168, "total_steps": 5364, "loss": 0.07191438972949982, "lr": 8.208287748867172e-08, "epoch": 2.8906293706293704, "percentage": 96.35, "elapsed_time": "10:17:51", "remaining_time": "0:23:25"} +{"current_steps": 5169, "total_steps": 5364, "loss": 0.2645817697048187, "lr": 8.125279131405151e-08, "epoch": 2.8911888111888113, "percentage": 96.36, "elapsed_time": "10:17:59", "remaining_time": "0:23:18"} +{"current_steps": 5170, "total_steps": 5364, "loss": 0.07099726796150208, "lr": 8.042690661608121e-08, "epoch": 2.891748251748252, "percentage": 96.38, "elapsed_time": "10:18:06", "remaining_time": "0:23:11"} +{"current_steps": 5171, "total_steps": 5364, "loss": 0.014326678588986397, "lr": 7.960522374459767e-08, "epoch": 2.8923076923076922, "percentage": 96.4, "elapsed_time": "10:18:13", "remaining_time": "0:23:04"} +{"current_steps": 5172, "total_steps": 5364, "loss": 0.07095473259687424, "lr": 7.87877430476569e-08, "epoch": 2.8928671328671327, "percentage": 96.42, "elapsed_time": "10:18:20", "remaining_time": "0:22:57"} +{"current_steps": 5173, "total_steps": 5364, "loss": 0.07701880484819412, "lr": 7.797446487153415e-08, "epoch": 2.8934265734265736, "percentage": 96.44, "elapsed_time": "10:18:28", "remaining_time": "0:22:50"} +{"current_steps": 5174, "total_steps": 5364, "loss": 0.042147181928157806, "lr": 7.716538956072606e-08, "epoch": 2.893986013986014, "percentage": 96.46, "elapsed_time": "10:18:35", "remaining_time": "0:22:42"} +{"current_steps": 5175, "total_steps": 5364, "loss": 0.010980570688843727, "lr": 7.636051745794959e-08, "epoch": 2.8945454545454545, "percentage": 96.48, "elapsed_time": "10:18:42", "remaining_time": "0:22:35"} +{"current_steps": 5176, "total_steps": 5364, "loss": 0.10655295848846436, "lr": 7.55598489041387e-08, "epoch": 2.895104895104895, "percentage": 96.5, "elapsed_time": "10:18:49", "remaining_time": "0:22:28"} +{"current_steps": 5177, "total_steps": 5364, "loss": 0.033465802669525146, "lr": 7.476338423844876e-08, "epoch": 2.8956643356643355, "percentage": 96.51, "elapsed_time": "10:18:56", "remaining_time": "0:22:21"} +{"current_steps": 5178, "total_steps": 5364, "loss": 0.050919309258461, "lr": 7.397112379825322e-08, "epoch": 2.8962237762237764, "percentage": 96.53, "elapsed_time": "10:19:03", "remaining_time": "0:22:14"} +{"current_steps": 5179, "total_steps": 5364, "loss": 0.1203484758734703, "lr": 7.318306791914587e-08, "epoch": 2.896783216783217, "percentage": 96.55, "elapsed_time": "10:19:10", "remaining_time": "0:22:07"} +{"current_steps": 5180, "total_steps": 5364, "loss": 0.0619923397898674, "lr": 7.239921693493857e-08, "epoch": 2.8973426573426573, "percentage": 96.57, "elapsed_time": "10:19:17", "remaining_time": "0:21:59"} +{"current_steps": 5181, "total_steps": 5364, "loss": 0.10916341096162796, "lr": 7.161957117766128e-08, "epoch": 2.8979020979020977, "percentage": 96.59, "elapsed_time": "10:19:24", "remaining_time": "0:21:52"} +{"current_steps": 5182, "total_steps": 5364, "loss": 0.1473524123430252, "lr": 7.084413097756537e-08, "epoch": 2.8984615384615386, "percentage": 96.61, "elapsed_time": "10:19:31", "remaining_time": "0:21:45"} +{"current_steps": 5183, "total_steps": 5364, "loss": 0.011255375109612942, "lr": 7.007289666311923e-08, "epoch": 2.899020979020979, "percentage": 96.63, "elapsed_time": "10:19:38", "remaining_time": "0:21:38"} +{"current_steps": 5184, "total_steps": 5364, "loss": 0.09858214110136032, "lr": 6.930586856100818e-08, "epoch": 2.8995804195804196, "percentage": 96.64, "elapsed_time": "10:19:45", "remaining_time": "0:21:31"} +{"current_steps": 5185, "total_steps": 5364, "loss": 0.1014171689748764, "lr": 6.854304699613789e-08, "epoch": 2.90013986013986, "percentage": 96.66, "elapsed_time": "10:19:52", "remaining_time": "0:21:23"} +{"current_steps": 5186, "total_steps": 5364, "loss": 0.10433556139469147, "lr": 6.778443229163101e-08, "epoch": 2.9006993006993005, "percentage": 96.68, "elapsed_time": "10:19:59", "remaining_time": "0:21:16"} +{"current_steps": 5187, "total_steps": 5364, "loss": 0.03780497610569, "lr": 6.703002476883046e-08, "epoch": 2.9012587412587414, "percentage": 96.7, "elapsed_time": "10:20:06", "remaining_time": "0:21:09"} +{"current_steps": 5188, "total_steps": 5364, "loss": 0.04920152202248573, "lr": 6.627982474729289e-08, "epoch": 2.901818181818182, "percentage": 96.72, "elapsed_time": "10:20:13", "remaining_time": "0:21:02"} +{"current_steps": 5189, "total_steps": 5364, "loss": 0.0343848317861557, "lr": 6.55338325447974e-08, "epoch": 2.9023776223776223, "percentage": 96.74, "elapsed_time": "10:20:20", "remaining_time": "0:20:55"} +{"current_steps": 5190, "total_steps": 5364, "loss": 0.20481495559215546, "lr": 6.479204847733678e-08, "epoch": 2.9029370629370628, "percentage": 96.76, "elapsed_time": "10:20:27", "remaining_time": "0:20:48"} +{"current_steps": 5191, "total_steps": 5364, "loss": 0.10618521273136139, "lr": 6.405447285912302e-08, "epoch": 2.9034965034965037, "percentage": 96.77, "elapsed_time": "10:20:34", "remaining_time": "0:20:40"} +{"current_steps": 5192, "total_steps": 5364, "loss": 0.14831744134426117, "lr": 6.332110600258623e-08, "epoch": 2.904055944055944, "percentage": 96.79, "elapsed_time": "10:20:42", "remaining_time": "0:20:33"} +{"current_steps": 5193, "total_steps": 5364, "loss": 0.13791592419147491, "lr": 6.259194821837345e-08, "epoch": 2.9046153846153846, "percentage": 96.81, "elapsed_time": "10:20:48", "remaining_time": "0:20:26"} +{"current_steps": 5194, "total_steps": 5364, "loss": 0.21965214610099792, "lr": 6.186699981534538e-08, "epoch": 2.905174825174825, "percentage": 96.83, "elapsed_time": "10:20:55", "remaining_time": "0:20:19"} +{"current_steps": 5195, "total_steps": 5364, "loss": 0.02646435797214508, "lr": 6.114626110058641e-08, "epoch": 2.9057342657342655, "percentage": 96.85, "elapsed_time": "10:21:02", "remaining_time": "0:20:12"} +{"current_steps": 5196, "total_steps": 5364, "loss": 0.06938068568706512, "lr": 6.04297323793901e-08, "epoch": 2.9062937062937064, "percentage": 96.87, "elapsed_time": "10:21:10", "remaining_time": "0:20:05"} +{"current_steps": 5197, "total_steps": 5364, "loss": 0.019158296287059784, "lr": 5.971741395527364e-08, "epoch": 2.906853146853147, "percentage": 96.89, "elapsed_time": "10:21:17", "remaining_time": "0:19:57"} +{"current_steps": 5198, "total_steps": 5364, "loss": 0.031087584793567657, "lr": 5.900930612996569e-08, "epoch": 2.9074125874125873, "percentage": 96.91, "elapsed_time": "10:21:24", "remaining_time": "0:19:50"} +{"current_steps": 5199, "total_steps": 5364, "loss": 0.22102338075637817, "lr": 5.830540920341299e-08, "epoch": 2.907972027972028, "percentage": 96.92, "elapsed_time": "10:21:31", "remaining_time": "0:19:43"} +{"current_steps": 5200, "total_steps": 5364, "loss": 0.12027538567781448, "lr": 5.760572347378035e-08, "epoch": 2.9085314685314687, "percentage": 96.94, "elapsed_time": "10:21:38", "remaining_time": "0:19:36"} +{"current_steps": 5201, "total_steps": 5364, "loss": 0.042640320956707, "lr": 5.691024923744515e-08, "epoch": 2.909090909090909, "percentage": 96.96, "elapsed_time": "10:21:45", "remaining_time": "0:19:29"} +{"current_steps": 5202, "total_steps": 5364, "loss": 0.02074226550757885, "lr": 5.621898678900506e-08, "epoch": 2.9096503496503496, "percentage": 96.98, "elapsed_time": "10:21:51", "remaining_time": "0:19:21"} +{"current_steps": 5203, "total_steps": 5364, "loss": 0.13586930930614471, "lr": 5.5531936421271416e-08, "epoch": 2.91020979020979, "percentage": 97.0, "elapsed_time": "10:21:59", "remaining_time": "0:19:14"} +{"current_steps": 5204, "total_steps": 5364, "loss": 0.11441251635551453, "lr": 5.484909842527031e-08, "epoch": 2.9107692307692306, "percentage": 97.02, "elapsed_time": "10:22:07", "remaining_time": "0:19:07"} +{"current_steps": 5205, "total_steps": 5364, "loss": 0.001350384671241045, "lr": 5.417047309024481e-08, "epoch": 2.9113286713286715, "percentage": 97.04, "elapsed_time": "10:22:13", "remaining_time": "0:19:00"} +{"current_steps": 5206, "total_steps": 5364, "loss": 0.1591944396495819, "lr": 5.349606070365498e-08, "epoch": 2.911888111888112, "percentage": 97.05, "elapsed_time": "10:22:20", "remaining_time": "0:18:53"} +{"current_steps": 5207, "total_steps": 5364, "loss": 0.13842764496803284, "lr": 5.282586155117231e-08, "epoch": 2.9124475524475524, "percentage": 97.07, "elapsed_time": "10:22:28", "remaining_time": "0:18:46"} +{"current_steps": 5208, "total_steps": 5364, "loss": 0.06949430704116821, "lr": 5.2159875916688584e-08, "epoch": 2.9130069930069933, "percentage": 97.09, "elapsed_time": "10:22:35", "remaining_time": "0:18:38"} +{"current_steps": 5209, "total_steps": 5364, "loss": 0.05434008687734604, "lr": 5.149810408230704e-08, "epoch": 2.9135664335664337, "percentage": 97.11, "elapsed_time": "10:22:42", "remaining_time": "0:18:31"} +{"current_steps": 5210, "total_steps": 5364, "loss": 0.2237141728401184, "lr": 5.084054632834678e-08, "epoch": 2.914125874125874, "percentage": 97.13, "elapsed_time": "10:22:49", "remaining_time": "0:18:24"} +{"current_steps": 5211, "total_steps": 5364, "loss": 0.013265509158372879, "lr": 5.0187202933343894e-08, "epoch": 2.9146853146853147, "percentage": 97.15, "elapsed_time": "10:22:56", "remaining_time": "0:18:17"} +{"current_steps": 5212, "total_steps": 5364, "loss": 0.10479851067066193, "lr": 4.953807417404588e-08, "epoch": 2.915244755244755, "percentage": 97.17, "elapsed_time": "10:23:03", "remaining_time": "0:18:10"} +{"current_steps": 5213, "total_steps": 5364, "loss": 0.07117408514022827, "lr": 4.8893160325418357e-08, "epoch": 2.9158041958041956, "percentage": 97.18, "elapsed_time": "10:23:10", "remaining_time": "0:18:03"} +{"current_steps": 5214, "total_steps": 5364, "loss": 0.0591374896466732, "lr": 4.825246166063835e-08, "epoch": 2.9163636363636365, "percentage": 97.2, "elapsed_time": "10:23:18", "remaining_time": "0:17:55"} +{"current_steps": 5215, "total_steps": 5364, "loss": 0.07768068462610245, "lr": 4.761597845110211e-08, "epoch": 2.916923076923077, "percentage": 97.22, "elapsed_time": "10:23:25", "remaining_time": "0:17:48"} +{"current_steps": 5216, "total_steps": 5364, "loss": 0.005802792031317949, "lr": 4.698371096641286e-08, "epoch": 2.9174825174825174, "percentage": 97.24, "elapsed_time": "10:23:33", "remaining_time": "0:17:41"} +{"current_steps": 5217, "total_steps": 5364, "loss": 0.011472095735371113, "lr": 4.6355659474396355e-08, "epoch": 2.9180419580419583, "percentage": 97.26, "elapsed_time": "10:23:39", "remaining_time": "0:17:34"} +{"current_steps": 5218, "total_steps": 5364, "loss": 0.05923731252551079, "lr": 4.573182424108647e-08, "epoch": 2.9186013986013988, "percentage": 97.28, "elapsed_time": "10:23:46", "remaining_time": "0:17:27"} +{"current_steps": 5219, "total_steps": 5364, "loss": 0.02489723637700081, "lr": 4.511220553073403e-08, "epoch": 2.9191608391608392, "percentage": 97.3, "elapsed_time": "10:23:53", "remaining_time": "0:17:20"} +{"current_steps": 5220, "total_steps": 5364, "loss": 0.05929826572537422, "lr": 4.449680360580244e-08, "epoch": 2.9197202797202797, "percentage": 97.32, "elapsed_time": "10:24:00", "remaining_time": "0:17:12"} +{"current_steps": 5221, "total_steps": 5364, "loss": 0.10521118342876434, "lr": 4.3885618726969836e-08, "epoch": 2.92027972027972, "percentage": 97.33, "elapsed_time": "10:24:07", "remaining_time": "0:17:05"} +{"current_steps": 5222, "total_steps": 5364, "loss": 0.08411519229412079, "lr": 4.327865115312802e-08, "epoch": 2.9208391608391606, "percentage": 97.35, "elapsed_time": "10:24:14", "remaining_time": "0:16:58"} +{"current_steps": 5223, "total_steps": 5364, "loss": 0.09228894114494324, "lr": 4.2675901141380204e-08, "epoch": 2.9213986013986015, "percentage": 97.37, "elapsed_time": "10:24:21", "remaining_time": "0:16:51"} +{"current_steps": 5224, "total_steps": 5364, "loss": 0.04566708207130432, "lr": 4.2077368947047724e-08, "epoch": 2.921958041958042, "percentage": 97.39, "elapsed_time": "10:24:29", "remaining_time": "0:16:44"} +{"current_steps": 5225, "total_steps": 5364, "loss": 0.14249449968338013, "lr": 4.1483054823661104e-08, "epoch": 2.9225174825174824, "percentage": 97.41, "elapsed_time": "10:24:36", "remaining_time": "0:16:36"} +{"current_steps": 5226, "total_steps": 5364, "loss": 0.11930012702941895, "lr": 4.089295902296453e-08, "epoch": 2.9230769230769234, "percentage": 97.43, "elapsed_time": "10:24:44", "remaining_time": "0:16:29"} +{"current_steps": 5227, "total_steps": 5364, "loss": 0.15889689326286316, "lr": 4.030708179491805e-08, "epoch": 2.923636363636364, "percentage": 97.45, "elapsed_time": "10:24:51", "remaining_time": "0:16:22"} +{"current_steps": 5228, "total_steps": 5364, "loss": 0.14439088106155396, "lr": 3.972542338769314e-08, "epoch": 2.9241958041958043, "percentage": 97.46, "elapsed_time": "10:24:58", "remaining_time": "0:16:15"} +{"current_steps": 5229, "total_steps": 5364, "loss": 0.10586054623126984, "lr": 3.9147984047672724e-08, "epoch": 2.9247552447552447, "percentage": 97.48, "elapsed_time": "10:25:04", "remaining_time": "0:16:08"} +{"current_steps": 5230, "total_steps": 5364, "loss": 0.05668766051530838, "lr": 3.857476401945337e-08, "epoch": 2.925314685314685, "percentage": 97.5, "elapsed_time": "10:25:11", "remaining_time": "0:16:01"} +{"current_steps": 5231, "total_steps": 5364, "loss": 0.057426631450653076, "lr": 3.8005763545848615e-08, "epoch": 2.9258741258741257, "percentage": 97.52, "elapsed_time": "10:25:18", "remaining_time": "0:15:53"} +{"current_steps": 5232, "total_steps": 5364, "loss": 0.2849140465259552, "lr": 3.74409828678779e-08, "epoch": 2.9264335664335666, "percentage": 97.54, "elapsed_time": "10:25:26", "remaining_time": "0:15:46"} +{"current_steps": 5233, "total_steps": 5364, "loss": 0.3134947121143341, "lr": 3.6880422224776503e-08, "epoch": 2.926993006993007, "percentage": 97.56, "elapsed_time": "10:25:33", "remaining_time": "0:15:39"} +{"current_steps": 5234, "total_steps": 5364, "loss": 0.031861674040555954, "lr": 3.63240818539945e-08, "epoch": 2.9275524475524475, "percentage": 97.58, "elapsed_time": "10:25:40", "remaining_time": "0:15:32"} +{"current_steps": 5235, "total_steps": 5364, "loss": 0.058092888444662094, "lr": 3.577196199118893e-08, "epoch": 2.9281118881118884, "percentage": 97.6, "elapsed_time": "10:25:47", "remaining_time": "0:15:25"} +{"current_steps": 5236, "total_steps": 5364, "loss": 0.20722651481628418, "lr": 3.522406287023272e-08, "epoch": 2.928671328671329, "percentage": 97.61, "elapsed_time": "10:25:54", "remaining_time": "0:15:18"} +{"current_steps": 5237, "total_steps": 5364, "loss": 0.11399199068546295, "lr": 3.4680384723211335e-08, "epoch": 2.9292307692307693, "percentage": 97.63, "elapsed_time": "10:26:01", "remaining_time": "0:15:10"} +{"current_steps": 5238, "total_steps": 5364, "loss": 0.1043810248374939, "lr": 3.414092778041944e-08, "epoch": 2.9297902097902098, "percentage": 97.65, "elapsed_time": "10:26:08", "remaining_time": "0:15:03"} +{"current_steps": 5239, "total_steps": 5364, "loss": 0.14572879672050476, "lr": 3.360569227036759e-08, "epoch": 2.9303496503496502, "percentage": 97.67, "elapsed_time": "10:26:15", "remaining_time": "0:14:56"} +{"current_steps": 5240, "total_steps": 5364, "loss": 0.052745770663022995, "lr": 3.3074678419773296e-08, "epoch": 2.9309090909090907, "percentage": 97.69, "elapsed_time": "10:26:23", "remaining_time": "0:14:49"} +{"current_steps": 5241, "total_steps": 5364, "loss": 0.1724082976579666, "lr": 3.254788645356999e-08, "epoch": 2.9314685314685316, "percentage": 97.71, "elapsed_time": "10:26:30", "remaining_time": "0:14:42"} +{"current_steps": 5242, "total_steps": 5364, "loss": 0.10703390091657639, "lr": 3.202531659490138e-08, "epoch": 2.932027972027972, "percentage": 97.73, "elapsed_time": "10:26:37", "remaining_time": "0:14:35"} +{"current_steps": 5243, "total_steps": 5364, "loss": 0.17900709807872772, "lr": 3.150696906512041e-08, "epoch": 2.9325874125874125, "percentage": 97.74, "elapsed_time": "10:26:43", "remaining_time": "0:14:27"} +{"current_steps": 5244, "total_steps": 5364, "loss": 0.08107936382293701, "lr": 3.099284408379588e-08, "epoch": 2.9331468531468534, "percentage": 97.76, "elapsed_time": "10:26:50", "remaining_time": "0:14:20"} +{"current_steps": 5245, "total_steps": 5364, "loss": 0.02277417853474617, "lr": 3.048294186870582e-08, "epoch": 2.933706293706294, "percentage": 97.78, "elapsed_time": "10:26:58", "remaining_time": "0:14:13"} +{"current_steps": 5246, "total_steps": 5364, "loss": 0.16745372116565704, "lr": 2.9977262635837447e-08, "epoch": 2.9342657342657343, "percentage": 97.8, "elapsed_time": "10:27:05", "remaining_time": "0:14:06"} +{"current_steps": 5247, "total_steps": 5364, "loss": 0.025397373363375664, "lr": 2.9475806599392753e-08, "epoch": 2.934825174825175, "percentage": 97.82, "elapsed_time": "10:27:12", "remaining_time": "0:13:59"} +{"current_steps": 5248, "total_steps": 5364, "loss": 0.07618074864149094, "lr": 2.897857397178294e-08, "epoch": 2.9353846153846153, "percentage": 97.84, "elapsed_time": "10:27:19", "remaining_time": "0:13:51"} +{"current_steps": 5249, "total_steps": 5364, "loss": 0.09367871284484863, "lr": 2.848556496362953e-08, "epoch": 2.9359440559440557, "percentage": 97.86, "elapsed_time": "10:27:26", "remaining_time": "0:13:44"} +{"current_steps": 5250, "total_steps": 5364, "loss": 0.10746271163225174, "lr": 2.799677978376769e-08, "epoch": 2.9365034965034966, "percentage": 97.87, "elapsed_time": "10:27:33", "remaining_time": "0:13:37"} +{"current_steps": 5251, "total_steps": 5364, "loss": 0.09788582473993301, "lr": 2.7512218639240696e-08, "epoch": 2.937062937062937, "percentage": 97.89, "elapsed_time": "10:27:40", "remaining_time": "0:13:30"} +{"current_steps": 5252, "total_steps": 5364, "loss": 0.05372265353798866, "lr": 2.7031881735303245e-08, "epoch": 2.9376223776223775, "percentage": 97.91, "elapsed_time": "10:27:47", "remaining_time": "0:13:23"} +{"current_steps": 5253, "total_steps": 5364, "loss": 0.14616882801055908, "lr": 2.6555769275420362e-08, "epoch": 2.9381818181818184, "percentage": 97.93, "elapsed_time": "10:27:54", "remaining_time": "0:13:16"} +{"current_steps": 5254, "total_steps": 5364, "loss": 0.3291112184524536, "lr": 2.6083881461270723e-08, "epoch": 2.938741258741259, "percentage": 97.95, "elapsed_time": "10:28:02", "remaining_time": "0:13:08"} +{"current_steps": 5255, "total_steps": 5364, "loss": 0.2190902978181839, "lr": 2.5616218492738877e-08, "epoch": 2.9393006993006994, "percentage": 97.97, "elapsed_time": "10:28:09", "remaining_time": "0:13:01"} +{"current_steps": 5256, "total_steps": 5364, "loss": 0.0288071446120739, "lr": 2.515278056792303e-08, "epoch": 2.93986013986014, "percentage": 97.99, "elapsed_time": "10:28:16", "remaining_time": "0:12:54"} +{"current_steps": 5257, "total_steps": 5364, "loss": 0.07421399652957916, "lr": 2.4693567883129488e-08, "epoch": 2.9404195804195803, "percentage": 98.01, "elapsed_time": "10:28:23", "remaining_time": "0:12:47"} +{"current_steps": 5258, "total_steps": 5364, "loss": 0.07199421525001526, "lr": 2.4238580632875984e-08, "epoch": 2.9409790209790208, "percentage": 98.02, "elapsed_time": "10:28:30", "remaining_time": "0:12:40"} +{"current_steps": 5259, "total_steps": 5364, "loss": 0.1259925663471222, "lr": 2.378781900989169e-08, "epoch": 2.9415384615384617, "percentage": 98.04, "elapsed_time": "10:28:37", "remaining_time": "0:12:33"} +{"current_steps": 5260, "total_steps": 5364, "loss": 0.03300304710865021, "lr": 2.3341283205113863e-08, "epoch": 2.942097902097902, "percentage": 98.06, "elapsed_time": "10:28:44", "remaining_time": "0:12:25"} +{"current_steps": 5261, "total_steps": 5364, "loss": 0.062602199614048, "lr": 2.2898973407688985e-08, "epoch": 2.9426573426573426, "percentage": 98.08, "elapsed_time": "10:28:52", "remaining_time": "0:12:18"} +{"current_steps": 5262, "total_steps": 5364, "loss": 0.072819784283638, "lr": 2.2460889804977182e-08, "epoch": 2.9432167832167835, "percentage": 98.1, "elapsed_time": "10:28:58", "remaining_time": "0:12:11"} +{"current_steps": 5263, "total_steps": 5364, "loss": 0.062079332768917084, "lr": 2.202703258254446e-08, "epoch": 2.943776223776224, "percentage": 98.12, "elapsed_time": "10:29:05", "remaining_time": "0:12:04"} +{"current_steps": 5264, "total_steps": 5364, "loss": 0.017300274223089218, "lr": 2.159740192416826e-08, "epoch": 2.9443356643356644, "percentage": 98.14, "elapsed_time": "10:29:12", "remaining_time": "0:11:57"} +{"current_steps": 5265, "total_steps": 5364, "loss": 0.01414365042001009, "lr": 2.1171998011836338e-08, "epoch": 2.944895104895105, "percentage": 98.15, "elapsed_time": "10:29:19", "remaining_time": "0:11:50"} +{"current_steps": 5266, "total_steps": 5364, "loss": 0.1340750902891159, "lr": 2.075082102574344e-08, "epoch": 2.9454545454545453, "percentage": 98.17, "elapsed_time": "10:29:26", "remaining_time": "0:11:42"} +{"current_steps": 5267, "total_steps": 5364, "loss": 0.020536599680781364, "lr": 2.0333871144297966e-08, "epoch": 2.946013986013986, "percentage": 98.19, "elapsed_time": "10:29:33", "remaining_time": "0:11:35"} +{"current_steps": 5268, "total_steps": 5364, "loss": 0.08401452004909515, "lr": 1.992114854411531e-08, "epoch": 2.9465734265734267, "percentage": 98.21, "elapsed_time": "10:29:40", "remaining_time": "0:11:28"} +{"current_steps": 5269, "total_steps": 5364, "loss": 0.0364210270345211, "lr": 1.9512653400018955e-08, "epoch": 2.947132867132867, "percentage": 98.23, "elapsed_time": "10:29:47", "remaining_time": "0:11:21"} +{"current_steps": 5270, "total_steps": 5364, "loss": 0.013481775298714638, "lr": 1.9108385885043824e-08, "epoch": 2.9476923076923076, "percentage": 98.25, "elapsed_time": "10:29:55", "remaining_time": "0:11:14"} +{"current_steps": 5271, "total_steps": 5364, "loss": 0.023128246888518333, "lr": 1.8708346170434043e-08, "epoch": 2.9482517482517485, "percentage": 98.27, "elapsed_time": "10:30:02", "remaining_time": "0:11:06"} +{"current_steps": 5272, "total_steps": 5364, "loss": 0.008567575365304947, "lr": 1.8312534425640737e-08, "epoch": 2.948811188811189, "percentage": 98.28, "elapsed_time": "10:30:09", "remaining_time": "0:10:59"} +{"current_steps": 5273, "total_steps": 5364, "loss": 0.07446427643299103, "lr": 1.7920950818327566e-08, "epoch": 2.9493706293706294, "percentage": 98.3, "elapsed_time": "10:30:16", "remaining_time": "0:10:52"} +{"current_steps": 5274, "total_steps": 5364, "loss": 0.10595938563346863, "lr": 1.7533595514362954e-08, "epoch": 2.94993006993007, "percentage": 98.32, "elapsed_time": "10:30:23", "remaining_time": "0:10:45"} +{"current_steps": 5275, "total_steps": 5364, "loss": 0.013983756303787231, "lr": 1.715046867782788e-08, "epoch": 2.9504895104895104, "percentage": 98.34, "elapsed_time": "10:30:30", "remaining_time": "0:10:38"} +{"current_steps": 5276, "total_steps": 5364, "loss": 0.07022291421890259, "lr": 1.67715704710103e-08, "epoch": 2.951048951048951, "percentage": 98.36, "elapsed_time": "10:30:37", "remaining_time": "0:10:31"} +{"current_steps": 5277, "total_steps": 5364, "loss": 0.09447485208511353, "lr": 1.63969010544085e-08, "epoch": 2.9516083916083917, "percentage": 98.38, "elapsed_time": "10:30:44", "remaining_time": "0:10:23"} +{"current_steps": 5278, "total_steps": 5364, "loss": 0.027280420064926147, "lr": 1.6026460586727743e-08, "epoch": 2.952167832167832, "percentage": 98.4, "elapsed_time": "10:30:51", "remaining_time": "0:10:16"} +{"current_steps": 5279, "total_steps": 5364, "loss": 0.030660949647426605, "lr": 1.566024922488252e-08, "epoch": 2.9527272727272726, "percentage": 98.42, "elapsed_time": "10:30:58", "remaining_time": "0:10:09"} +{"current_steps": 5280, "total_steps": 5364, "loss": 0.07480526715517044, "lr": 1.5298267123996513e-08, "epoch": 2.9532867132867135, "percentage": 98.43, "elapsed_time": "10:31:05", "remaining_time": "0:10:02"} +{"current_steps": 5281, "total_steps": 5364, "loss": 0.05191662162542343, "lr": 1.4940514437401523e-08, "epoch": 2.953846153846154, "percentage": 98.45, "elapsed_time": "10:31:12", "remaining_time": "0:09:55"} +{"current_steps": 5282, "total_steps": 5364, "loss": 0.0955943763256073, "lr": 1.4586991316636323e-08, "epoch": 2.9544055944055945, "percentage": 98.47, "elapsed_time": "10:31:19", "remaining_time": "0:09:48"} +{"current_steps": 5283, "total_steps": 5364, "loss": 0.07685510814189911, "lr": 1.4237697911452242e-08, "epoch": 2.954965034965035, "percentage": 98.49, "elapsed_time": "10:31:26", "remaining_time": "0:09:40"} +{"current_steps": 5284, "total_steps": 5364, "loss": 0.06673769652843475, "lr": 1.3892634369804258e-08, "epoch": 2.9555244755244754, "percentage": 98.51, "elapsed_time": "10:31:33", "remaining_time": "0:09:33"} +{"current_steps": 5285, "total_steps": 5364, "loss": 0.023648586124181747, "lr": 1.3551800837858787e-08, "epoch": 2.956083916083916, "percentage": 98.53, "elapsed_time": "10:31:40", "remaining_time": "0:09:26"} +{"current_steps": 5286, "total_steps": 5364, "loss": 0.14275985956192017, "lr": 1.3215197459988115e-08, "epoch": 2.9566433566433568, "percentage": 98.55, "elapsed_time": "10:31:47", "remaining_time": "0:09:19"} +{"current_steps": 5287, "total_steps": 5364, "loss": 0.05834471806883812, "lr": 1.288282437877597e-08, "epoch": 2.957202797202797, "percentage": 98.56, "elapsed_time": "10:31:54", "remaining_time": "0:09:12"} +{"current_steps": 5288, "total_steps": 5364, "loss": 0.08282680064439774, "lr": 1.2554681735009733e-08, "epoch": 2.9577622377622377, "percentage": 98.58, "elapsed_time": "10:32:01", "remaining_time": "0:09:05"} +{"current_steps": 5289, "total_steps": 5364, "loss": 0.042859189212322235, "lr": 1.2230769667687104e-08, "epoch": 2.9583216783216786, "percentage": 98.6, "elapsed_time": "10:32:09", "remaining_time": "0:08:57"} +{"current_steps": 5290, "total_steps": 5364, "loss": 0.19588994979858398, "lr": 1.1911088314016106e-08, "epoch": 2.958881118881119, "percentage": 98.62, "elapsed_time": "10:32:16", "remaining_time": "0:08:50"} +{"current_steps": 5291, "total_steps": 5364, "loss": 0.03237784653902054, "lr": 1.1595637809409532e-08, "epoch": 2.9594405594405595, "percentage": 98.64, "elapsed_time": "10:32:24", "remaining_time": "0:08:43"} +{"current_steps": 5292, "total_steps": 5364, "loss": 0.09009547531604767, "lr": 1.1284418287488274e-08, "epoch": 2.96, "percentage": 98.66, "elapsed_time": "10:32:31", "remaining_time": "0:08:36"} +{"current_steps": 5293, "total_steps": 5364, "loss": 0.0711212158203125, "lr": 1.0977429880081325e-08, "epoch": 2.9605594405594404, "percentage": 98.68, "elapsed_time": "10:32:39", "remaining_time": "0:08:29"} +{"current_steps": 5294, "total_steps": 5364, "loss": 0.11750383675098419, "lr": 1.0674672717225776e-08, "epoch": 2.961118881118881, "percentage": 98.7, "elapsed_time": "10:32:45", "remaining_time": "0:08:22"} +{"current_steps": 5295, "total_steps": 5364, "loss": 0.021160634234547615, "lr": 1.0376146927167929e-08, "epoch": 2.961678321678322, "percentage": 98.71, "elapsed_time": "10:32:53", "remaining_time": "0:08:14"} +{"current_steps": 5296, "total_steps": 5364, "loss": 0.05732947215437889, "lr": 1.0081852636358858e-08, "epoch": 2.9622377622377623, "percentage": 98.73, "elapsed_time": "10:33:00", "remaining_time": "0:08:07"} +{"current_steps": 5297, "total_steps": 5364, "loss": 0.2054559290409088, "lr": 9.791789969458843e-09, "epoch": 2.9627972027972027, "percentage": 98.75, "elapsed_time": "10:33:07", "remaining_time": "0:08:00"} +{"current_steps": 5298, "total_steps": 5364, "loss": 0.06776751577854156, "lr": 9.505959049335155e-09, "epoch": 2.9633566433566436, "percentage": 98.77, "elapsed_time": "10:33:15", "remaining_time": "0:07:53"} +{"current_steps": 5299, "total_steps": 5364, "loss": 0.11562247574329376, "lr": 9.22435999706206e-09, "epoch": 2.963916083916084, "percentage": 98.79, "elapsed_time": "10:33:22", "remaining_time": "0:07:46"} +{"current_steps": 5300, "total_steps": 5364, "loss": 0.025352679193019867, "lr": 8.946992931923026e-09, "epoch": 2.9644755244755245, "percentage": 98.81, "elapsed_time": "10:33:29", "remaining_time": "0:07:38"} +{"current_steps": 5301, "total_steps": 5364, "loss": 0.04324430599808693, "lr": 8.673857971408517e-09, "epoch": 2.965034965034965, "percentage": 98.83, "elapsed_time": "10:33:36", "remaining_time": "0:07:31"} +{"current_steps": 5302, "total_steps": 5364, "loss": 0.02898053079843521, "lr": 8.404955231213763e-09, "epoch": 2.9655944055944055, "percentage": 98.84, "elapsed_time": "10:33:43", "remaining_time": "0:07:24"} +{"current_steps": 5303, "total_steps": 5364, "loss": 0.10131726413965225, "lr": 8.140284825244315e-09, "epoch": 2.966153846153846, "percentage": 98.86, "elapsed_time": "10:33:50", "remaining_time": "0:07:17"} +{"current_steps": 5304, "total_steps": 5364, "loss": 0.19311170279979706, "lr": 7.879846865611606e-09, "epoch": 2.966713286713287, "percentage": 98.88, "elapsed_time": "10:33:57", "remaining_time": "0:07:10"} +{"current_steps": 5305, "total_steps": 5364, "loss": 0.06657704710960388, "lr": 7.623641462634057e-09, "epoch": 2.9672727272727273, "percentage": 98.9, "elapsed_time": "10:34:04", "remaining_time": "0:07:03"} +{"current_steps": 5306, "total_steps": 5364, "loss": 0.12123794853687286, "lr": 7.371668724837078e-09, "epoch": 2.9678321678321677, "percentage": 98.92, "elapsed_time": "10:34:11", "remaining_time": "0:06:55"} +{"current_steps": 5307, "total_steps": 5364, "loss": 0.005168756935745478, "lr": 7.12392875895529e-09, "epoch": 2.9683916083916086, "percentage": 98.94, "elapsed_time": "10:34:19", "remaining_time": "0:06:48"} +{"current_steps": 5308, "total_steps": 5364, "loss": 0.068355992436409, "lr": 6.8804216699280836e-09, "epoch": 2.968951048951049, "percentage": 98.96, "elapsed_time": "10:34:26", "remaining_time": "0:06:41"} +{"current_steps": 5309, "total_steps": 5364, "loss": 0.13130581378936768, "lr": 6.64114756090184e-09, "epoch": 2.9695104895104896, "percentage": 98.97, "elapsed_time": "10:34:33", "remaining_time": "0:06:34"} +{"current_steps": 5310, "total_steps": 5364, "loss": 0.05100453272461891, "lr": 6.406106533231038e-09, "epoch": 2.97006993006993, "percentage": 98.99, "elapsed_time": "10:34:40", "remaining_time": "0:06:27"} +{"current_steps": 5311, "total_steps": 5364, "loss": 0.05833175405859947, "lr": 6.175298686476039e-09, "epoch": 2.9706293706293705, "percentage": 99.01, "elapsed_time": "10:34:47", "remaining_time": "0:06:20"} +{"current_steps": 5312, "total_steps": 5364, "loss": 0.06594595313072205, "lr": 5.948724118405303e-09, "epoch": 2.971188811188811, "percentage": 99.03, "elapsed_time": "10:34:55", "remaining_time": "0:06:12"} +{"current_steps": 5313, "total_steps": 5364, "loss": 0.21014057099819183, "lr": 5.726382924992058e-09, "epoch": 2.971748251748252, "percentage": 99.05, "elapsed_time": "10:35:01", "remaining_time": "0:06:05"} +{"current_steps": 5314, "total_steps": 5364, "loss": 0.0358211025595665, "lr": 5.508275200419855e-09, "epoch": 2.9723076923076923, "percentage": 99.07, "elapsed_time": "10:35:09", "remaining_time": "0:05:58"} +{"current_steps": 5315, "total_steps": 5364, "loss": 0.00839526578783989, "lr": 5.294401037075902e-09, "epoch": 2.9728671328671328, "percentage": 99.09, "elapsed_time": "10:35:16", "remaining_time": "0:05:51"} +{"current_steps": 5316, "total_steps": 5364, "loss": 0.039361946284770966, "lr": 5.0847605255543995e-09, "epoch": 2.9734265734265737, "percentage": 99.11, "elapsed_time": "10:35:23", "remaining_time": "0:05:44"} +{"current_steps": 5317, "total_steps": 5364, "loss": 0.08253898471593857, "lr": 4.879353754657645e-09, "epoch": 2.973986013986014, "percentage": 99.12, "elapsed_time": "10:35:30", "remaining_time": "0:05:37"} +{"current_steps": 5318, "total_steps": 5364, "loss": 0.04213719069957733, "lr": 4.678180811393818e-09, "epoch": 2.9745454545454546, "percentage": 99.14, "elapsed_time": "10:35:37", "remaining_time": "0:05:29"} +{"current_steps": 5319, "total_steps": 5364, "loss": 0.136572003364563, "lr": 4.481241780976975e-09, "epoch": 2.975104895104895, "percentage": 99.16, "elapsed_time": "10:35:45", "remaining_time": "0:05:22"} +{"current_steps": 5320, "total_steps": 5364, "loss": 0.13080723583698273, "lr": 4.288536746828164e-09, "epoch": 2.9756643356643355, "percentage": 99.18, "elapsed_time": "10:35:52", "remaining_time": "0:05:15"} +{"current_steps": 5321, "total_steps": 5364, "loss": 0.2413359433412552, "lr": 4.100065790576535e-09, "epoch": 2.976223776223776, "percentage": 99.2, "elapsed_time": "10:35:59", "remaining_time": "0:05:08"} +{"current_steps": 5322, "total_steps": 5364, "loss": 0.04885898903012276, "lr": 3.915828992056003e-09, "epoch": 2.976783216783217, "percentage": 99.22, "elapsed_time": "10:36:06", "remaining_time": "0:05:01"} +{"current_steps": 5323, "total_steps": 5364, "loss": 0.04643958434462547, "lr": 3.735826429305256e-09, "epoch": 2.9773426573426574, "percentage": 99.24, "elapsed_time": "10:36:13", "remaining_time": "0:04:54"} +{"current_steps": 5324, "total_steps": 5364, "loss": 0.128621444106102, "lr": 3.5600581785744104e-09, "epoch": 2.977902097902098, "percentage": 99.25, "elapsed_time": "10:36:20", "remaining_time": "0:04:46"} +{"current_steps": 5325, "total_steps": 5364, "loss": 0.14379192888736725, "lr": 3.3885243143161327e-09, "epoch": 2.9784615384615387, "percentage": 99.27, "elapsed_time": "10:36:27", "remaining_time": "0:04:39"} +{"current_steps": 5326, "total_steps": 5364, "loss": 0.1060786321759224, "lr": 3.2212249091900793e-09, "epoch": 2.979020979020979, "percentage": 99.29, "elapsed_time": "10:36:34", "remaining_time": "0:04:32"} +{"current_steps": 5327, "total_steps": 5364, "loss": 0.13857942819595337, "lr": 3.0581600340617857e-09, "epoch": 2.9795804195804196, "percentage": 99.31, "elapsed_time": "10:36:41", "remaining_time": "0:04:25"} +{"current_steps": 5328, "total_steps": 5364, "loss": 0.08162636309862137, "lr": 2.8993297580048874e-09, "epoch": 2.98013986013986, "percentage": 99.33, "elapsed_time": "10:36:48", "remaining_time": "0:04:18"} +{"current_steps": 5329, "total_steps": 5364, "loss": 0.05172056332230568, "lr": 2.7447341482977897e-09, "epoch": 2.9806993006993006, "percentage": 99.35, "elapsed_time": "10:36:55", "remaining_time": "0:04:10"} +{"current_steps": 5330, "total_steps": 5364, "loss": 0.028990596532821655, "lr": 2.594373270424777e-09, "epoch": 2.981258741258741, "percentage": 99.37, "elapsed_time": "10:37:02", "remaining_time": "0:04:03"} +{"current_steps": 5331, "total_steps": 5364, "loss": 0.046703826636075974, "lr": 2.4482471880793447e-09, "epoch": 2.981818181818182, "percentage": 99.38, "elapsed_time": "10:37:09", "remaining_time": "0:03:56"} +{"current_steps": 5332, "total_steps": 5364, "loss": 0.039994753897190094, "lr": 2.3063559631575363e-09, "epoch": 2.9823776223776224, "percentage": 99.4, "elapsed_time": "10:37:16", "remaining_time": "0:03:49"} +{"current_steps": 5333, "total_steps": 5364, "loss": 0.04648361727595329, "lr": 2.168699655762385e-09, "epoch": 2.982937062937063, "percentage": 99.42, "elapsed_time": "10:37:23", "remaining_time": "0:03:42"} +{"current_steps": 5334, "total_steps": 5364, "loss": 0.10541056096553802, "lr": 2.0352783242039155e-09, "epoch": 2.9834965034965037, "percentage": 99.44, "elapsed_time": "10:37:30", "remaining_time": "0:03:35"} +{"current_steps": 5335, "total_steps": 5364, "loss": 0.10288608074188232, "lr": 1.90609202499914e-09, "epoch": 2.984055944055944, "percentage": 99.46, "elapsed_time": "10:37:37", "remaining_time": "0:03:27"} +{"current_steps": 5336, "total_steps": 5364, "loss": 0.06302841007709503, "lr": 1.7811408128687312e-09, "epoch": 2.9846153846153847, "percentage": 99.48, "elapsed_time": "10:37:44", "remaining_time": "0:03:20"} +{"current_steps": 5337, "total_steps": 5364, "loss": 0.0783616304397583, "lr": 1.660424740741462e-09, "epoch": 2.985174825174825, "percentage": 99.5, "elapsed_time": "10:37:51", "remaining_time": "0:03:13"} +{"current_steps": 5338, "total_steps": 5364, "loss": 0.11086538434028625, "lr": 1.5439438597508739e-09, "epoch": 2.9857342657342656, "percentage": 99.52, "elapsed_time": "10:37:58", "remaining_time": "0:03:06"} +{"current_steps": 5339, "total_steps": 5364, "loss": 0.06495922058820724, "lr": 1.4316982192374984e-09, "epoch": 2.986293706293706, "percentage": 99.53, "elapsed_time": "10:38:05", "remaining_time": "0:02:59"} +{"current_steps": 5340, "total_steps": 5364, "loss": 0.060855019837617874, "lr": 1.3236878667477471e-09, "epoch": 2.986853146853147, "percentage": 99.55, "elapsed_time": "10:38:12", "remaining_time": "0:02:52"} +{"current_steps": 5341, "total_steps": 5364, "loss": 0.10884314775466919, "lr": 1.2199128480328004e-09, "epoch": 2.9874125874125874, "percentage": 99.57, "elapsed_time": "10:38:19", "remaining_time": "0:02:44"} +{"current_steps": 5342, "total_steps": 5364, "loss": 0.09586424380540848, "lr": 1.120373207050829e-09, "epoch": 2.987972027972028, "percentage": 99.59, "elapsed_time": "10:38:26", "remaining_time": "0:02:37"} +{"current_steps": 5343, "total_steps": 5364, "loss": 0.16469401121139526, "lr": 1.0250689859658824e-09, "epoch": 2.988531468531469, "percentage": 99.61, "elapsed_time": "10:38:33", "remaining_time": "0:02:30"} +{"current_steps": 5344, "total_steps": 5364, "loss": 0.06607284396886826, "lr": 9.340002251478907e-10, "epoch": 2.9890909090909092, "percentage": 99.63, "elapsed_time": "10:38:40", "remaining_time": "0:02:23"} +{"current_steps": 5345, "total_steps": 5364, "loss": 0.06861595809459686, "lr": 8.471669631726631e-10, "epoch": 2.9896503496503497, "percentage": 99.65, "elapsed_time": "10:38:47", "remaining_time": "0:02:16"} +{"current_steps": 5346, "total_steps": 5364, "loss": 0.05425921082496643, "lr": 7.64569236820778e-10, "epoch": 2.99020979020979, "percentage": 99.66, "elapsed_time": "10:38:54", "remaining_time": "0:02:09"} +{"current_steps": 5347, "total_steps": 5364, "loss": 0.07142813503742218, "lr": 6.862070810820243e-10, "epoch": 2.9907692307692306, "percentage": 99.68, "elapsed_time": "10:39:01", "remaining_time": "0:02:01"} +{"current_steps": 5348, "total_steps": 5364, "loss": 0.022150443866848946, "lr": 6.120805291487397e-10, "epoch": 2.991328671328671, "percentage": 99.7, "elapsed_time": "10:39:09", "remaining_time": "0:01:54"} +{"current_steps": 5349, "total_steps": 5364, "loss": 0.022666743025183678, "lr": 5.421896124191417e-10, "epoch": 2.991888111888112, "percentage": 99.72, "elapsed_time": "10:39:16", "remaining_time": "0:01:47"} +{"current_steps": 5350, "total_steps": 5364, "loss": 0.005687844939529896, "lr": 4.765343604995476e-10, "epoch": 2.9924475524475525, "percentage": 99.74, "elapsed_time": "10:39:23", "remaining_time": "0:01:40"} +{"current_steps": 5351, "total_steps": 5364, "loss": 0.1533994823694229, "lr": 4.1511480119993395e-10, "epoch": 2.993006993006993, "percentage": 99.76, "elapsed_time": "10:39:31", "remaining_time": "0:01:33"} +{"current_steps": 5352, "total_steps": 5364, "loss": 0.1376902461051941, "lr": 3.579309605383774e-10, "epoch": 2.993566433566434, "percentage": 99.78, "elapsed_time": "10:39:38", "remaining_time": "0:01:26"} +{"current_steps": 5353, "total_steps": 5364, "loss": 0.017431624233722687, "lr": 3.049828627355034e-10, "epoch": 2.9941258741258743, "percentage": 99.79, "elapsed_time": "10:39:45", "remaining_time": "0:01:18"} +{"current_steps": 5354, "total_steps": 5364, "loss": 0.056222110986709595, "lr": 2.562705302211477e-10, "epoch": 2.9946853146853147, "percentage": 99.81, "elapsed_time": "10:39:52", "remaining_time": "0:01:11"} +{"current_steps": 5355, "total_steps": 5364, "loss": 0.18948790431022644, "lr": 2.11793983627695e-10, "epoch": 2.995244755244755, "percentage": 99.83, "elapsed_time": "10:39:59", "remaining_time": "0:01:04"} +{"current_steps": 5356, "total_steps": 5364, "loss": 0.03144899383187294, "lr": 1.7155324179674026e-10, "epoch": 2.9958041958041957, "percentage": 99.85, "elapsed_time": "10:40:06", "remaining_time": "0:00:57"} +{"current_steps": 5357, "total_steps": 5364, "loss": 0.12154553830623627, "lr": 1.3554832177242738e-10, "epoch": 2.996363636363636, "percentage": 99.87, "elapsed_time": "10:40:13", "remaining_time": "0:00:50"} +{"current_steps": 5358, "total_steps": 5364, "loss": 0.08542687445878983, "lr": 1.0377923880700025e-10, "epoch": 2.996923076923077, "percentage": 99.89, "elapsed_time": "10:40:20", "remaining_time": "0:00:43"} +{"current_steps": 5359, "total_steps": 5364, "loss": 0.08679330348968506, "lr": 7.624600635636192e-11, "epoch": 2.9974825174825175, "percentage": 99.91, "elapsed_time": "10:40:27", "remaining_time": "0:00:35"} +{"current_steps": 5360, "total_steps": 5364, "loss": 0.11148659139871597, "lr": 5.29486360845155e-11, "epoch": 2.998041958041958, "percentage": 99.93, "elapsed_time": "10:40:34", "remaining_time": "0:00:28"} +{"current_steps": 5361, "total_steps": 5364, "loss": 0.12423374503850937, "lr": 3.388713786023345e-11, "epoch": 2.998601398601399, "percentage": 99.94, "elapsed_time": "10:40:41", "remaining_time": "0:00:21"} +{"current_steps": 5362, "total_steps": 5364, "loss": 0.17181716859340668, "lr": 1.906151975594739e-11, "epoch": 2.9991608391608393, "percentage": 99.96, "elapsed_time": "10:40:48", "remaining_time": "0:00:14"} +{"current_steps": 5363, "total_steps": 5364, "loss": 0.04787980765104294, "lr": 8.47178805329918e-12, "epoch": 2.9997202797202798, "percentage": 99.98, "elapsed_time": "10:40:55", "remaining_time": "0:00:07"} +{"current_steps": 5364, "total_steps": 5364, "loss": 0.05316243693232536, "lr": 2.117947237589846e-12, "epoch": 3.0, "percentage": 100.0, "elapsed_time": "10:40:58", "remaining_time": "0:00:00"} +{"current_steps": 5364, "total_steps": 5364, "epoch": 3.0, "percentage": 100.0, "elapsed_time": "10:41:26", "remaining_time": "0:00:00"} diff --git a/trainer_state.json b/trainer_state.json new file mode 100644 index 0000000..c255d8d --- /dev/null +++ b/trainer_state.json @@ -0,0 +1,37591 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 3.0, + "eval_steps": 500, + "global_step": 5364, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0005594405594405594, + "grad_norm": 2.9197237342419418, + "learning_rate": 0.0, + "loss": 0.1708594113588333, + "step": 1 + }, + { + "epoch": 0.0011188811188811189, + "grad_norm": 3.692259504608811, + "learning_rate": 3.7243947858473e-08, + "loss": 0.2891339659690857, + "step": 2 + }, + { + "epoch": 0.0016783216783216783, + "grad_norm": 3.588784216894788, + "learning_rate": 7.4487895716946e-08, + "loss": 0.4187517762184143, + "step": 3 + }, + { + "epoch": 0.0022377622377622378, + "grad_norm": 3.0611463918085438, + "learning_rate": 1.11731843575419e-07, + "loss": 0.13779963552951813, + "step": 4 + }, + { + "epoch": 0.002797202797202797, + "grad_norm": 3.9153777159460965, + "learning_rate": 1.48975791433892e-07, + "loss": 0.31178635358810425, + "step": 5 + }, + { + "epoch": 0.0033566433566433566, + "grad_norm": 3.7279731841468204, + "learning_rate": 1.8621973929236503e-07, + "loss": 0.31994569301605225, + "step": 6 + }, + { + "epoch": 0.003916083916083916, + "grad_norm": 3.3571514963940925, + "learning_rate": 2.23463687150838e-07, + "loss": 0.3239685893058777, + "step": 7 + }, + { + "epoch": 0.0044755244755244755, + "grad_norm": 3.3099162885998954, + "learning_rate": 2.60707635009311e-07, + "loss": 0.40604960918426514, + "step": 8 + }, + { + "epoch": 0.005034965034965035, + "grad_norm": 3.5315656997526506, + "learning_rate": 2.97951582867784e-07, + "loss": 0.26543134450912476, + "step": 9 + }, + { + "epoch": 0.005594405594405594, + "grad_norm": 1.8287563595515066, + "learning_rate": 3.3519553072625703e-07, + "loss": 0.15973007678985596, + "step": 10 + }, + { + "epoch": 0.006153846153846154, + "grad_norm": 2.7884648301474355, + "learning_rate": 3.7243947858473006e-07, + "loss": 0.24497920274734497, + "step": 11 + }, + { + "epoch": 0.006713286713286713, + "grad_norm": 3.675078973859732, + "learning_rate": 4.09683426443203e-07, + "loss": 0.507112443447113, + "step": 12 + }, + { + "epoch": 0.007272727272727273, + "grad_norm": 2.801101359612238, + "learning_rate": 4.46927374301676e-07, + "loss": 0.36340224742889404, + "step": 13 + }, + { + "epoch": 0.007832167832167832, + "grad_norm": 2.92804959046496, + "learning_rate": 4.84171322160149e-07, + "loss": 0.29563137888908386, + "step": 14 + }, + { + "epoch": 0.008391608391608392, + "grad_norm": 3.0553389737633108, + "learning_rate": 5.21415270018622e-07, + "loss": 0.3542381227016449, + "step": 15 + }, + { + "epoch": 0.008951048951048951, + "grad_norm": 3.0913084045820396, + "learning_rate": 5.58659217877095e-07, + "loss": 0.2901057004928589, + "step": 16 + }, + { + "epoch": 0.00951048951048951, + "grad_norm": 2.781126386732038, + "learning_rate": 5.95903165735568e-07, + "loss": 0.3724617660045624, + "step": 17 + }, + { + "epoch": 0.01006993006993007, + "grad_norm": 3.8120063163243834, + "learning_rate": 6.331471135940411e-07, + "loss": 0.20881980657577515, + "step": 18 + }, + { + "epoch": 0.01062937062937063, + "grad_norm": 2.1397219812314323, + "learning_rate": 6.703910614525141e-07, + "loss": 0.1372430920600891, + "step": 19 + }, + { + "epoch": 0.011188811188811189, + "grad_norm": 3.428443868081563, + "learning_rate": 7.076350093109869e-07, + "loss": 0.4805028736591339, + "step": 20 + }, + { + "epoch": 0.011748251748251748, + "grad_norm": 2.0573352828798015, + "learning_rate": 7.448789571694601e-07, + "loss": 0.28077012300491333, + "step": 21 + }, + { + "epoch": 0.012307692307692308, + "grad_norm": 4.292463683403444, + "learning_rate": 7.82122905027933e-07, + "loss": 0.3277529776096344, + "step": 22 + }, + { + "epoch": 0.012867132867132867, + "grad_norm": 3.5564715300362892, + "learning_rate": 8.19366852886406e-07, + "loss": 0.4050637483596802, + "step": 23 + }, + { + "epoch": 0.013426573426573427, + "grad_norm": 3.1410478357983376, + "learning_rate": 8.566108007448791e-07, + "loss": 0.22474241256713867, + "step": 24 + }, + { + "epoch": 0.013986013986013986, + "grad_norm": 1.7274917087451054, + "learning_rate": 8.93854748603352e-07, + "loss": 0.20507575571537018, + "step": 25 + }, + { + "epoch": 0.014545454545454545, + "grad_norm": 1.9484738908238035, + "learning_rate": 9.31098696461825e-07, + "loss": 0.26187247037887573, + "step": 26 + }, + { + "epoch": 0.015104895104895105, + "grad_norm": 4.809682614887802, + "learning_rate": 9.68342644320298e-07, + "loss": 0.28988659381866455, + "step": 27 + }, + { + "epoch": 0.015664335664335664, + "grad_norm": 2.5254200393873254, + "learning_rate": 1.005586592178771e-06, + "loss": 0.5565254092216492, + "step": 28 + }, + { + "epoch": 0.016223776223776225, + "grad_norm": 1.9383050746117274, + "learning_rate": 1.042830540037244e-06, + "loss": 0.3537827432155609, + "step": 29 + }, + { + "epoch": 0.016783216783216783, + "grad_norm": 1.6934717825748415, + "learning_rate": 1.080074487895717e-06, + "loss": 0.33376017212867737, + "step": 30 + }, + { + "epoch": 0.017342657342657344, + "grad_norm": 1.3582419497104339, + "learning_rate": 1.11731843575419e-06, + "loss": 0.3507142663002014, + "step": 31 + }, + { + "epoch": 0.017902097902097902, + "grad_norm": 1.8906402038292858, + "learning_rate": 1.1545623836126631e-06, + "loss": 0.11607243120670319, + "step": 32 + }, + { + "epoch": 0.018461538461538463, + "grad_norm": 3.0242806938623517, + "learning_rate": 1.191806331471136e-06, + "loss": 0.24526657164096832, + "step": 33 + }, + { + "epoch": 0.01902097902097902, + "grad_norm": 1.4024266308931217, + "learning_rate": 1.229050279329609e-06, + "loss": 0.35029327869415283, + "step": 34 + }, + { + "epoch": 0.019580419580419582, + "grad_norm": 1.401467334119554, + "learning_rate": 1.2662942271880822e-06, + "loss": 0.2289755493402481, + "step": 35 + }, + { + "epoch": 0.02013986013986014, + "grad_norm": 1.3564416933674919, + "learning_rate": 1.303538175046555e-06, + "loss": 0.3760473132133484, + "step": 36 + }, + { + "epoch": 0.0206993006993007, + "grad_norm": 1.294432477144678, + "learning_rate": 1.3407821229050281e-06, + "loss": 0.3920930325984955, + "step": 37 + }, + { + "epoch": 0.02125874125874126, + "grad_norm": 1.0096710495922756, + "learning_rate": 1.378026070763501e-06, + "loss": 0.22879764437675476, + "step": 38 + }, + { + "epoch": 0.02181818181818182, + "grad_norm": 0.9629840452949139, + "learning_rate": 1.4152700186219739e-06, + "loss": 0.09033755958080292, + "step": 39 + }, + { + "epoch": 0.022377622377622378, + "grad_norm": 1.6428987833387223, + "learning_rate": 1.4525139664804472e-06, + "loss": 0.46498849987983704, + "step": 40 + }, + { + "epoch": 0.02293706293706294, + "grad_norm": 0.9105565219035306, + "learning_rate": 1.4897579143389202e-06, + "loss": 0.3834105134010315, + "step": 41 + }, + { + "epoch": 0.023496503496503496, + "grad_norm": 1.3129837873981682, + "learning_rate": 1.5270018621973931e-06, + "loss": 0.45146605372428894, + "step": 42 + }, + { + "epoch": 0.024055944055944058, + "grad_norm": 0.9193204661048321, + "learning_rate": 1.564245810055866e-06, + "loss": 0.3595430850982666, + "step": 43 + }, + { + "epoch": 0.024615384615384615, + "grad_norm": 0.6726275014588311, + "learning_rate": 1.601489757914339e-06, + "loss": 0.14026623964309692, + "step": 44 + }, + { + "epoch": 0.025174825174825177, + "grad_norm": 1.0628982951093027, + "learning_rate": 1.638733705772812e-06, + "loss": 0.2923401892185211, + "step": 45 + }, + { + "epoch": 0.025734265734265734, + "grad_norm": 1.7802002849381497, + "learning_rate": 1.675977653631285e-06, + "loss": 0.263121098279953, + "step": 46 + }, + { + "epoch": 0.026293706293706295, + "grad_norm": 0.8996470813538586, + "learning_rate": 1.7132216014897581e-06, + "loss": 0.29029422998428345, + "step": 47 + }, + { + "epoch": 0.026853146853146853, + "grad_norm": 0.6738239886288222, + "learning_rate": 1.7504655493482312e-06, + "loss": 0.07528115808963776, + "step": 48 + }, + { + "epoch": 0.027412587412587414, + "grad_norm": 0.8331027057671616, + "learning_rate": 1.787709497206704e-06, + "loss": 0.3030151128768921, + "step": 49 + }, + { + "epoch": 0.027972027972027972, + "grad_norm": 0.7983489583280228, + "learning_rate": 1.8249534450651772e-06, + "loss": 0.21776533126831055, + "step": 50 + }, + { + "epoch": 0.028531468531468533, + "grad_norm": 0.7384019670458639, + "learning_rate": 1.86219739292365e-06, + "loss": 0.1647198349237442, + "step": 51 + }, + { + "epoch": 0.02909090909090909, + "grad_norm": 0.7265619412282439, + "learning_rate": 1.899441340782123e-06, + "loss": 0.39184725284576416, + "step": 52 + }, + { + "epoch": 0.029650349650349652, + "grad_norm": 0.6959462139953257, + "learning_rate": 1.936685288640596e-06, + "loss": 0.22568494081497192, + "step": 53 + }, + { + "epoch": 0.03020979020979021, + "grad_norm": 0.9719097388554532, + "learning_rate": 1.9739292364990693e-06, + "loss": 0.3494471311569214, + "step": 54 + }, + { + "epoch": 0.03076923076923077, + "grad_norm": 0.6015601297966352, + "learning_rate": 2.011173184357542e-06, + "loss": 0.15478362143039703, + "step": 55 + }, + { + "epoch": 0.03132867132867133, + "grad_norm": 0.8840319317297511, + "learning_rate": 2.048417132216015e-06, + "loss": 0.31515175104141235, + "step": 56 + }, + { + "epoch": 0.031888111888111886, + "grad_norm": 0.8259148240902539, + "learning_rate": 2.085661080074488e-06, + "loss": 0.2727990746498108, + "step": 57 + }, + { + "epoch": 0.03244755244755245, + "grad_norm": 0.4441071477383887, + "learning_rate": 2.1229050279329612e-06, + "loss": 0.08602505922317505, + "step": 58 + }, + { + "epoch": 0.03300699300699301, + "grad_norm": 1.2503698356444217, + "learning_rate": 2.160148975791434e-06, + "loss": 0.3949652910232544, + "step": 59 + }, + { + "epoch": 0.033566433566433566, + "grad_norm": 0.6982818555426696, + "learning_rate": 2.197392923649907e-06, + "loss": 0.2999691367149353, + "step": 60 + }, + { + "epoch": 0.034125874125874124, + "grad_norm": 0.6569794581446365, + "learning_rate": 2.23463687150838e-06, + "loss": 0.33670473098754883, + "step": 61 + }, + { + "epoch": 0.03468531468531469, + "grad_norm": 0.7384811293643592, + "learning_rate": 2.271880819366853e-06, + "loss": 0.10987913608551025, + "step": 62 + }, + { + "epoch": 0.035244755244755246, + "grad_norm": 0.6506821879413079, + "learning_rate": 2.3091247672253262e-06, + "loss": 0.170758455991745, + "step": 63 + }, + { + "epoch": 0.035804195804195804, + "grad_norm": 0.6151568842931352, + "learning_rate": 2.3463687150837993e-06, + "loss": 0.27042537927627563, + "step": 64 + }, + { + "epoch": 0.03636363636363636, + "grad_norm": 0.6543688581964249, + "learning_rate": 2.383612662942272e-06, + "loss": 0.18736714124679565, + "step": 65 + }, + { + "epoch": 0.036923076923076927, + "grad_norm": 0.72637545237279, + "learning_rate": 2.420856610800745e-06, + "loss": 0.24357211589813232, + "step": 66 + }, + { + "epoch": 0.037482517482517484, + "grad_norm": 0.8714533553447702, + "learning_rate": 2.458100558659218e-06, + "loss": 0.5307877659797668, + "step": 67 + }, + { + "epoch": 0.03804195804195804, + "grad_norm": 0.7921884340726971, + "learning_rate": 2.495344506517691e-06, + "loss": 0.33759641647338867, + "step": 68 + }, + { + "epoch": 0.0386013986013986, + "grad_norm": 0.40505449408212435, + "learning_rate": 2.5325884543761643e-06, + "loss": 0.06529378145933151, + "step": 69 + }, + { + "epoch": 0.039160839160839164, + "grad_norm": 0.4629263927520877, + "learning_rate": 2.569832402234637e-06, + "loss": 0.09254449605941772, + "step": 70 + }, + { + "epoch": 0.03972027972027972, + "grad_norm": 0.6416897941283076, + "learning_rate": 2.60707635009311e-06, + "loss": 0.17956393957138062, + "step": 71 + }, + { + "epoch": 0.04027972027972028, + "grad_norm": 0.5718003038552566, + "learning_rate": 2.644320297951583e-06, + "loss": 0.17691554129123688, + "step": 72 + }, + { + "epoch": 0.04083916083916084, + "grad_norm": 1.2134969417382564, + "learning_rate": 2.6815642458100562e-06, + "loss": 0.4376865327358246, + "step": 73 + }, + { + "epoch": 0.0413986013986014, + "grad_norm": 0.606211215884481, + "learning_rate": 2.718808193668529e-06, + "loss": 0.09415476024150848, + "step": 74 + }, + { + "epoch": 0.04195804195804196, + "grad_norm": 0.5090575681848114, + "learning_rate": 2.756052141527002e-06, + "loss": 0.16237586736679077, + "step": 75 + }, + { + "epoch": 0.04251748251748252, + "grad_norm": 0.4381130926174227, + "learning_rate": 2.793296089385475e-06, + "loss": 0.0929182767868042, + "step": 76 + }, + { + "epoch": 0.043076923076923075, + "grad_norm": 0.8280137457187345, + "learning_rate": 2.8305400372439477e-06, + "loss": 0.2737734913825989, + "step": 77 + }, + { + "epoch": 0.04363636363636364, + "grad_norm": 0.5908008153788887, + "learning_rate": 2.867783985102421e-06, + "loss": 0.1599085032939911, + "step": 78 + }, + { + "epoch": 0.0441958041958042, + "grad_norm": 0.8980271254892994, + "learning_rate": 2.9050279329608943e-06, + "loss": 0.3695093095302582, + "step": 79 + }, + { + "epoch": 0.044755244755244755, + "grad_norm": 0.7196845925114118, + "learning_rate": 2.9422718808193674e-06, + "loss": 0.34134459495544434, + "step": 80 + }, + { + "epoch": 0.04531468531468531, + "grad_norm": 0.7659357199418624, + "learning_rate": 2.9795158286778405e-06, + "loss": 0.28330808877944946, + "step": 81 + }, + { + "epoch": 0.04587412587412588, + "grad_norm": 0.502470779748972, + "learning_rate": 3.016759776536313e-06, + "loss": 0.15364162623882294, + "step": 82 + }, + { + "epoch": 0.046433566433566435, + "grad_norm": 0.8119283555870879, + "learning_rate": 3.0540037243947862e-06, + "loss": 0.07387331873178482, + "step": 83 + }, + { + "epoch": 0.04699300699300699, + "grad_norm": 0.8981366558425712, + "learning_rate": 3.0912476722532593e-06, + "loss": 0.14365491271018982, + "step": 84 + }, + { + "epoch": 0.04755244755244755, + "grad_norm": 0.5338763764560029, + "learning_rate": 3.128491620111732e-06, + "loss": 0.11993853747844696, + "step": 85 + }, + { + "epoch": 0.048111888111888115, + "grad_norm": 0.7807660679959038, + "learning_rate": 3.165735567970205e-06, + "loss": 0.3427037000656128, + "step": 86 + }, + { + "epoch": 0.04867132867132867, + "grad_norm": 0.54335908168607, + "learning_rate": 3.202979515828678e-06, + "loss": 0.17792752385139465, + "step": 87 + }, + { + "epoch": 0.04923076923076923, + "grad_norm": 0.5905123910841287, + "learning_rate": 3.240223463687151e-06, + "loss": 0.17775408923625946, + "step": 88 + }, + { + "epoch": 0.04979020979020979, + "grad_norm": 0.638117991047077, + "learning_rate": 3.277467411545624e-06, + "loss": 0.3558940887451172, + "step": 89 + }, + { + "epoch": 0.05034965034965035, + "grad_norm": 0.7045800475107825, + "learning_rate": 3.314711359404097e-06, + "loss": 0.1892475187778473, + "step": 90 + }, + { + "epoch": 0.05090909090909091, + "grad_norm": 0.6190974215469384, + "learning_rate": 3.35195530726257e-06, + "loss": 0.2643950581550598, + "step": 91 + }, + { + "epoch": 0.05146853146853147, + "grad_norm": 0.7275274037985527, + "learning_rate": 3.3891992551210427e-06, + "loss": 0.34897392988204956, + "step": 92 + }, + { + "epoch": 0.052027972027972026, + "grad_norm": 0.8126089646326822, + "learning_rate": 3.4264432029795162e-06, + "loss": 0.3010078966617584, + "step": 93 + }, + { + "epoch": 0.05258741258741259, + "grad_norm": 0.7600274932939938, + "learning_rate": 3.4636871508379893e-06, + "loss": 0.47209683060646057, + "step": 94 + }, + { + "epoch": 0.05314685314685315, + "grad_norm": 0.527331570802039, + "learning_rate": 3.5009310986964624e-06, + "loss": 0.12591159343719482, + "step": 95 + }, + { + "epoch": 0.053706293706293706, + "grad_norm": 0.5733700875114613, + "learning_rate": 3.538175046554935e-06, + "loss": 0.2532845139503479, + "step": 96 + }, + { + "epoch": 0.054265734265734264, + "grad_norm": 0.4858623048880027, + "learning_rate": 3.575418994413408e-06, + "loss": 0.10538607090711594, + "step": 97 + }, + { + "epoch": 0.05482517482517483, + "grad_norm": 0.5165949651008546, + "learning_rate": 3.6126629422718813e-06, + "loss": 0.07029303908348083, + "step": 98 + }, + { + "epoch": 0.055384615384615386, + "grad_norm": 0.5958158310527621, + "learning_rate": 3.6499068901303543e-06, + "loss": 0.2196342945098877, + "step": 99 + }, + { + "epoch": 0.055944055944055944, + "grad_norm": 0.5776884976982972, + "learning_rate": 3.687150837988827e-06, + "loss": 0.3075711131095886, + "step": 100 + }, + { + "epoch": 0.0565034965034965, + "grad_norm": 0.5503479054389154, + "learning_rate": 3.7243947858473e-06, + "loss": 0.16532987356185913, + "step": 101 + }, + { + "epoch": 0.057062937062937066, + "grad_norm": 0.4998159158577328, + "learning_rate": 3.761638733705773e-06, + "loss": 0.10717526078224182, + "step": 102 + }, + { + "epoch": 0.057622377622377624, + "grad_norm": 1.270904417426402, + "learning_rate": 3.798882681564246e-06, + "loss": 0.2941814661026001, + "step": 103 + }, + { + "epoch": 0.05818181818181818, + "grad_norm": 0.5913659005618204, + "learning_rate": 3.836126629422719e-06, + "loss": 0.31177815794944763, + "step": 104 + }, + { + "epoch": 0.05874125874125874, + "grad_norm": 0.4449543927644539, + "learning_rate": 3.873370577281192e-06, + "loss": 0.05996173247694969, + "step": 105 + }, + { + "epoch": 0.059300699300699304, + "grad_norm": 0.7329912266946541, + "learning_rate": 3.910614525139665e-06, + "loss": 0.3034001290798187, + "step": 106 + }, + { + "epoch": 0.05986013986013986, + "grad_norm": 0.639090137637412, + "learning_rate": 3.947858472998139e-06, + "loss": 0.18240708112716675, + "step": 107 + }, + { + "epoch": 0.06041958041958042, + "grad_norm": 0.5475555641006102, + "learning_rate": 3.985102420856611e-06, + "loss": 0.17454060912132263, + "step": 108 + }, + { + "epoch": 0.06097902097902098, + "grad_norm": 0.945533685766672, + "learning_rate": 4.022346368715084e-06, + "loss": 0.4517461955547333, + "step": 109 + }, + { + "epoch": 0.06153846153846154, + "grad_norm": 0.6886106889895997, + "learning_rate": 4.0595903165735574e-06, + "loss": 0.36985060572624207, + "step": 110 + }, + { + "epoch": 0.0620979020979021, + "grad_norm": 0.5739703656279221, + "learning_rate": 4.09683426443203e-06, + "loss": 0.13078774511814117, + "step": 111 + }, + { + "epoch": 0.06265734265734266, + "grad_norm": 0.6657485473545149, + "learning_rate": 4.134078212290504e-06, + "loss": 0.23888853192329407, + "step": 112 + }, + { + "epoch": 0.06321678321678321, + "grad_norm": 0.7905017500019588, + "learning_rate": 4.171322160148976e-06, + "loss": 0.2135712057352066, + "step": 113 + }, + { + "epoch": 0.06377622377622377, + "grad_norm": 0.8405982351750321, + "learning_rate": 4.208566108007449e-06, + "loss": 0.223187655210495, + "step": 114 + }, + { + "epoch": 0.06433566433566433, + "grad_norm": 0.4702123311943065, + "learning_rate": 4.2458100558659224e-06, + "loss": 0.11557766050100327, + "step": 115 + }, + { + "epoch": 0.0648951048951049, + "grad_norm": 0.8033092433188308, + "learning_rate": 4.283054003724395e-06, + "loss": 0.37016069889068604, + "step": 116 + }, + { + "epoch": 0.06545454545454546, + "grad_norm": 0.6153047459323562, + "learning_rate": 4.320297951582868e-06, + "loss": 0.21168112754821777, + "step": 117 + }, + { + "epoch": 0.06601398601398602, + "grad_norm": 0.6121917469353054, + "learning_rate": 4.357541899441341e-06, + "loss": 0.2518339455127716, + "step": 118 + }, + { + "epoch": 0.06657342657342658, + "grad_norm": 0.8905564920868401, + "learning_rate": 4.394785847299814e-06, + "loss": 0.3767576217651367, + "step": 119 + }, + { + "epoch": 0.06713286713286713, + "grad_norm": 0.6503419659938909, + "learning_rate": 4.432029795158287e-06, + "loss": 0.21407866477966309, + "step": 120 + }, + { + "epoch": 0.06769230769230769, + "grad_norm": 0.6058080802949822, + "learning_rate": 4.46927374301676e-06, + "loss": 0.1839624047279358, + "step": 121 + }, + { + "epoch": 0.06825174825174825, + "grad_norm": 8.8710596951778, + "learning_rate": 4.506517690875234e-06, + "loss": 0.4413369297981262, + "step": 122 + }, + { + "epoch": 0.0688111888111888, + "grad_norm": 0.6497696170161037, + "learning_rate": 4.543761638733706e-06, + "loss": 0.21440665423870087, + "step": 123 + }, + { + "epoch": 0.06937062937062938, + "grad_norm": 0.6817621830454861, + "learning_rate": 4.581005586592179e-06, + "loss": 0.259691059589386, + "step": 124 + }, + { + "epoch": 0.06993006993006994, + "grad_norm": 0.9224058372304568, + "learning_rate": 4.6182495344506524e-06, + "loss": 0.4421902298927307, + "step": 125 + }, + { + "epoch": 0.07048951048951049, + "grad_norm": 0.7130019347111556, + "learning_rate": 4.655493482309125e-06, + "loss": 0.30450770258903503, + "step": 126 + }, + { + "epoch": 0.07104895104895105, + "grad_norm": 0.583735550576196, + "learning_rate": 4.692737430167599e-06, + "loss": 0.14847812056541443, + "step": 127 + }, + { + "epoch": 0.07160839160839161, + "grad_norm": 0.6715650308500007, + "learning_rate": 4.729981378026071e-06, + "loss": 0.3133907914161682, + "step": 128 + }, + { + "epoch": 0.07216783216783217, + "grad_norm": 0.7592919764691967, + "learning_rate": 4.767225325884544e-06, + "loss": 0.24020129442214966, + "step": 129 + }, + { + "epoch": 0.07272727272727272, + "grad_norm": 0.8186755664763226, + "learning_rate": 4.8044692737430175e-06, + "loss": 0.4030919075012207, + "step": 130 + }, + { + "epoch": 0.07328671328671328, + "grad_norm": 0.885964542190149, + "learning_rate": 4.84171322160149e-06, + "loss": 0.44573646783828735, + "step": 131 + }, + { + "epoch": 0.07384615384615385, + "grad_norm": 0.6565037655115212, + "learning_rate": 4.878957169459963e-06, + "loss": 0.09619377553462982, + "step": 132 + }, + { + "epoch": 0.07440559440559441, + "grad_norm": 0.6831390778312841, + "learning_rate": 4.916201117318436e-06, + "loss": 0.23731550574302673, + "step": 133 + }, + { + "epoch": 0.07496503496503497, + "grad_norm": 0.7247100123084085, + "learning_rate": 4.953445065176909e-06, + "loss": 0.24186912178993225, + "step": 134 + }, + { + "epoch": 0.07552447552447553, + "grad_norm": 0.757864091309888, + "learning_rate": 4.990689013035382e-06, + "loss": 0.2194354236125946, + "step": 135 + }, + { + "epoch": 0.07608391608391608, + "grad_norm": 0.694324586158812, + "learning_rate": 5.027932960893855e-06, + "loss": 0.2353523075580597, + "step": 136 + }, + { + "epoch": 0.07664335664335664, + "grad_norm": 0.6560897096323426, + "learning_rate": 5.065176908752329e-06, + "loss": 0.19667869806289673, + "step": 137 + }, + { + "epoch": 0.0772027972027972, + "grad_norm": 0.5847133730127764, + "learning_rate": 5.102420856610801e-06, + "loss": 0.19589757919311523, + "step": 138 + }, + { + "epoch": 0.07776223776223776, + "grad_norm": 0.4846647084657547, + "learning_rate": 5.139664804469274e-06, + "loss": 0.11321631073951721, + "step": 139 + }, + { + "epoch": 0.07832167832167833, + "grad_norm": 0.582059911542778, + "learning_rate": 5.1769087523277475e-06, + "loss": 0.1956067979335785, + "step": 140 + }, + { + "epoch": 0.07888111888111889, + "grad_norm": 0.47042382715691583, + "learning_rate": 5.21415270018622e-06, + "loss": 0.1621169149875641, + "step": 141 + }, + { + "epoch": 0.07944055944055944, + "grad_norm": 0.4607685218453164, + "learning_rate": 5.251396648044693e-06, + "loss": 0.07715841382741928, + "step": 142 + }, + { + "epoch": 0.08, + "grad_norm": 0.6783000234331666, + "learning_rate": 5.288640595903166e-06, + "loss": 0.24442407488822937, + "step": 143 + }, + { + "epoch": 0.08055944055944056, + "grad_norm": 0.7034180982213968, + "learning_rate": 5.325884543761639e-06, + "loss": 0.16002245247364044, + "step": 144 + }, + { + "epoch": 0.08111888111888112, + "grad_norm": 0.7118097876875871, + "learning_rate": 5.3631284916201125e-06, + "loss": 0.24429768323898315, + "step": 145 + }, + { + "epoch": 0.08167832167832167, + "grad_norm": 0.8643574658122292, + "learning_rate": 5.400372439478585e-06, + "loss": 0.35507357120513916, + "step": 146 + }, + { + "epoch": 0.08223776223776223, + "grad_norm": 0.7175581750907782, + "learning_rate": 5.437616387337058e-06, + "loss": 0.3768399953842163, + "step": 147 + }, + { + "epoch": 0.0827972027972028, + "grad_norm": 0.36909902269733885, + "learning_rate": 5.474860335195531e-06, + "loss": 0.04639776796102524, + "step": 148 + }, + { + "epoch": 0.08335664335664336, + "grad_norm": 0.8322514962096716, + "learning_rate": 5.512104283054004e-06, + "loss": 0.22011643648147583, + "step": 149 + }, + { + "epoch": 0.08391608391608392, + "grad_norm": 0.5594175101751949, + "learning_rate": 5.549348230912477e-06, + "loss": 0.153517946600914, + "step": 150 + }, + { + "epoch": 0.08447552447552448, + "grad_norm": 0.6388083664111138, + "learning_rate": 5.58659217877095e-06, + "loss": 0.19159212708473206, + "step": 151 + }, + { + "epoch": 0.08503496503496503, + "grad_norm": 0.44773624077762875, + "learning_rate": 5.623836126629423e-06, + "loss": 0.12672311067581177, + "step": 152 + }, + { + "epoch": 0.08559440559440559, + "grad_norm": 0.49953783616619285, + "learning_rate": 5.6610800744878955e-06, + "loss": 0.09087824821472168, + "step": 153 + }, + { + "epoch": 0.08615384615384615, + "grad_norm": 0.49466150958248234, + "learning_rate": 5.698324022346369e-06, + "loss": 0.06624268740415573, + "step": 154 + }, + { + "epoch": 0.08671328671328671, + "grad_norm": 0.7111757656075016, + "learning_rate": 5.735567970204842e-06, + "loss": 0.2487543672323227, + "step": 155 + }, + { + "epoch": 0.08727272727272728, + "grad_norm": 0.7081734323220977, + "learning_rate": 5.772811918063314e-06, + "loss": 0.3196878731250763, + "step": 156 + }, + { + "epoch": 0.08783216783216784, + "grad_norm": 0.6938996843980109, + "learning_rate": 5.810055865921789e-06, + "loss": 0.19816963374614716, + "step": 157 + }, + { + "epoch": 0.0883916083916084, + "grad_norm": 0.6860056383605823, + "learning_rate": 5.847299813780261e-06, + "loss": 0.20571818947792053, + "step": 158 + }, + { + "epoch": 0.08895104895104895, + "grad_norm": 0.7047534617938089, + "learning_rate": 5.884543761638735e-06, + "loss": 0.24458524584770203, + "step": 159 + }, + { + "epoch": 0.08951048951048951, + "grad_norm": 1.026348343952594, + "learning_rate": 5.9217877094972075e-06, + "loss": 0.11320653557777405, + "step": 160 + }, + { + "epoch": 0.09006993006993007, + "grad_norm": 0.6474429296354924, + "learning_rate": 5.959031657355681e-06, + "loss": 0.22428768873214722, + "step": 161 + }, + { + "epoch": 0.09062937062937063, + "grad_norm": 1.2689770377025829, + "learning_rate": 5.996275605214154e-06, + "loss": 0.669741690158844, + "step": 162 + }, + { + "epoch": 0.09118881118881118, + "grad_norm": 1.1645097459549545, + "learning_rate": 6.033519553072626e-06, + "loss": 0.27598151564598083, + "step": 163 + }, + { + "epoch": 0.09174825174825176, + "grad_norm": 0.681518466489075, + "learning_rate": 6.0707635009311e-06, + "loss": 0.22851765155792236, + "step": 164 + }, + { + "epoch": 0.09230769230769231, + "grad_norm": 0.7446721937215875, + "learning_rate": 6.1080074487895725e-06, + "loss": 0.20998801290988922, + "step": 165 + }, + { + "epoch": 0.09286713286713287, + "grad_norm": 0.5718099623274117, + "learning_rate": 6.145251396648045e-06, + "loss": 0.1636764258146286, + "step": 166 + }, + { + "epoch": 0.09342657342657343, + "grad_norm": 0.6094891354585585, + "learning_rate": 6.182495344506519e-06, + "loss": 0.1978091597557068, + "step": 167 + }, + { + "epoch": 0.09398601398601399, + "grad_norm": 0.9104460511712239, + "learning_rate": 6.219739292364991e-06, + "loss": 0.381095826625824, + "step": 168 + }, + { + "epoch": 0.09454545454545454, + "grad_norm": 0.7077874331611348, + "learning_rate": 6.256983240223464e-06, + "loss": 0.19355298578739166, + "step": 169 + }, + { + "epoch": 0.0951048951048951, + "grad_norm": 1.119262369176454, + "learning_rate": 6.2942271880819375e-06, + "loss": 0.46214985847473145, + "step": 170 + }, + { + "epoch": 0.09566433566433566, + "grad_norm": 0.8250333786238949, + "learning_rate": 6.33147113594041e-06, + "loss": 0.3472660183906555, + "step": 171 + }, + { + "epoch": 0.09622377622377623, + "grad_norm": 0.7173444520665829, + "learning_rate": 6.368715083798883e-06, + "loss": 0.26900821924209595, + "step": 172 + }, + { + "epoch": 0.09678321678321679, + "grad_norm": 0.6753721654936515, + "learning_rate": 6.405959031657356e-06, + "loss": 0.17796164751052856, + "step": 173 + }, + { + "epoch": 0.09734265734265735, + "grad_norm": 0.9799125467422652, + "learning_rate": 6.443202979515829e-06, + "loss": 0.47886112332344055, + "step": 174 + }, + { + "epoch": 0.0979020979020979, + "grad_norm": 0.6056188575855868, + "learning_rate": 6.480446927374302e-06, + "loss": 0.1489739865064621, + "step": 175 + }, + { + "epoch": 0.09846153846153846, + "grad_norm": 0.8019321342713106, + "learning_rate": 6.517690875232775e-06, + "loss": 0.30581769347190857, + "step": 176 + }, + { + "epoch": 0.09902097902097902, + "grad_norm": 0.7043230004953681, + "learning_rate": 6.554934823091248e-06, + "loss": 0.2701963484287262, + "step": 177 + }, + { + "epoch": 0.09958041958041958, + "grad_norm": 0.4909494086565752, + "learning_rate": 6.592178770949721e-06, + "loss": 0.0773768424987793, + "step": 178 + }, + { + "epoch": 0.10013986013986013, + "grad_norm": 0.8049075618610132, + "learning_rate": 6.629422718808194e-06, + "loss": 0.24532026052474976, + "step": 179 + }, + { + "epoch": 0.1006993006993007, + "grad_norm": 0.6218370694951026, + "learning_rate": 6.666666666666667e-06, + "loss": 0.2108195722103119, + "step": 180 + }, + { + "epoch": 0.10125874125874126, + "grad_norm": 0.7568792379009885, + "learning_rate": 6.70391061452514e-06, + "loss": 0.20850294828414917, + "step": 181 + }, + { + "epoch": 0.10181818181818182, + "grad_norm": 2.204867625788177, + "learning_rate": 6.741154562383613e-06, + "loss": 0.18542012572288513, + "step": 182 + }, + { + "epoch": 0.10237762237762238, + "grad_norm": 0.6483641443651852, + "learning_rate": 6.7783985102420855e-06, + "loss": 0.1984855681657791, + "step": 183 + }, + { + "epoch": 0.10293706293706294, + "grad_norm": 0.6234801968940772, + "learning_rate": 6.815642458100559e-06, + "loss": 0.142118439078331, + "step": 184 + }, + { + "epoch": 0.1034965034965035, + "grad_norm": 0.7744322970934345, + "learning_rate": 6.8528864059590325e-06, + "loss": 0.1488180309534073, + "step": 185 + }, + { + "epoch": 0.10405594405594405, + "grad_norm": 0.6083686543030203, + "learning_rate": 6.890130353817506e-06, + "loss": 0.10070636868476868, + "step": 186 + }, + { + "epoch": 0.10461538461538461, + "grad_norm": 0.7330034373183372, + "learning_rate": 6.927374301675979e-06, + "loss": 0.21599172055721283, + "step": 187 + }, + { + "epoch": 0.10517482517482518, + "grad_norm": 0.7343378088448371, + "learning_rate": 6.964618249534451e-06, + "loss": 0.2776916027069092, + "step": 188 + }, + { + "epoch": 0.10573426573426574, + "grad_norm": 0.7617219749911406, + "learning_rate": 7.001862197392925e-06, + "loss": 0.26190677285194397, + "step": 189 + }, + { + "epoch": 0.1062937062937063, + "grad_norm": 1.1932928066136026, + "learning_rate": 7.0391061452513975e-06, + "loss": 0.45778489112854004, + "step": 190 + }, + { + "epoch": 0.10685314685314685, + "grad_norm": 0.7483610746019292, + "learning_rate": 7.07635009310987e-06, + "loss": 0.23035813868045807, + "step": 191 + }, + { + "epoch": 0.10741258741258741, + "grad_norm": 0.6166255085053276, + "learning_rate": 7.113594040968344e-06, + "loss": 0.13400602340698242, + "step": 192 + }, + { + "epoch": 0.10797202797202797, + "grad_norm": 0.580188450711339, + "learning_rate": 7.150837988826816e-06, + "loss": 0.107833132147789, + "step": 193 + }, + { + "epoch": 0.10853146853146853, + "grad_norm": 0.4881676977373535, + "learning_rate": 7.188081936685289e-06, + "loss": 0.10887670516967773, + "step": 194 + }, + { + "epoch": 0.10909090909090909, + "grad_norm": 0.7860904598096525, + "learning_rate": 7.2253258845437625e-06, + "loss": 0.45634034276008606, + "step": 195 + }, + { + "epoch": 0.10965034965034966, + "grad_norm": 0.5974929815779407, + "learning_rate": 7.262569832402235e-06, + "loss": 0.13844041526317596, + "step": 196 + }, + { + "epoch": 0.11020979020979021, + "grad_norm": 0.7139678405174972, + "learning_rate": 7.299813780260709e-06, + "loss": 0.24760495126247406, + "step": 197 + }, + { + "epoch": 0.11076923076923077, + "grad_norm": 0.6501311661677297, + "learning_rate": 7.337057728119181e-06, + "loss": 0.2501208186149597, + "step": 198 + }, + { + "epoch": 0.11132867132867133, + "grad_norm": 0.6776768358247515, + "learning_rate": 7.374301675977654e-06, + "loss": 0.3180907368659973, + "step": 199 + }, + { + "epoch": 0.11188811188811189, + "grad_norm": 0.9020591775708353, + "learning_rate": 7.4115456238361275e-06, + "loss": 0.4095803499221802, + "step": 200 + }, + { + "epoch": 0.11244755244755245, + "grad_norm": 0.43716084713645487, + "learning_rate": 7.4487895716946e-06, + "loss": 0.06699497997760773, + "step": 201 + }, + { + "epoch": 0.113006993006993, + "grad_norm": 0.4482862892261059, + "learning_rate": 7.486033519553073e-06, + "loss": 0.14195895195007324, + "step": 202 + }, + { + "epoch": 0.11356643356643356, + "grad_norm": 0.7593271017603378, + "learning_rate": 7.523277467411546e-06, + "loss": 0.3649834990501404, + "step": 203 + }, + { + "epoch": 0.11412587412587413, + "grad_norm": 0.5301921670950808, + "learning_rate": 7.560521415270019e-06, + "loss": 0.15401607751846313, + "step": 204 + }, + { + "epoch": 0.11468531468531469, + "grad_norm": 0.914412000243228, + "learning_rate": 7.597765363128492e-06, + "loss": 0.4673020839691162, + "step": 205 + }, + { + "epoch": 0.11524475524475525, + "grad_norm": 0.6620340333852319, + "learning_rate": 7.635009310986964e-06, + "loss": 0.19694717228412628, + "step": 206 + }, + { + "epoch": 0.1158041958041958, + "grad_norm": 0.786473941726886, + "learning_rate": 7.672253258845438e-06, + "loss": 0.38036859035491943, + "step": 207 + }, + { + "epoch": 0.11636363636363636, + "grad_norm": 0.7599008935211434, + "learning_rate": 7.709497206703911e-06, + "loss": 0.20669475197792053, + "step": 208 + }, + { + "epoch": 0.11692307692307692, + "grad_norm": 0.7023290027371979, + "learning_rate": 7.746741154562383e-06, + "loss": 0.30299532413482666, + "step": 209 + }, + { + "epoch": 0.11748251748251748, + "grad_norm": 0.5049162605559804, + "learning_rate": 7.783985102420857e-06, + "loss": 0.13257357478141785, + "step": 210 + }, + { + "epoch": 0.11804195804195804, + "grad_norm": 0.5371544680925446, + "learning_rate": 7.82122905027933e-06, + "loss": 0.17287683486938477, + "step": 211 + }, + { + "epoch": 0.11860139860139861, + "grad_norm": 0.6833679985413331, + "learning_rate": 7.858472998137802e-06, + "loss": 0.4957885444164276, + "step": 212 + }, + { + "epoch": 0.11916083916083917, + "grad_norm": 0.5479819451451922, + "learning_rate": 7.895716945996277e-06, + "loss": 0.14581745862960815, + "step": 213 + }, + { + "epoch": 0.11972027972027972, + "grad_norm": 0.6132135877587116, + "learning_rate": 7.932960893854749e-06, + "loss": 0.21463730931282043, + "step": 214 + }, + { + "epoch": 0.12027972027972028, + "grad_norm": 0.5875640788548351, + "learning_rate": 7.970204841713223e-06, + "loss": 0.20221847295761108, + "step": 215 + }, + { + "epoch": 0.12083916083916084, + "grad_norm": 0.8244118090315607, + "learning_rate": 8.007448789571696e-06, + "loss": 0.258322149515152, + "step": 216 + }, + { + "epoch": 0.1213986013986014, + "grad_norm": 0.656892569974264, + "learning_rate": 8.044692737430168e-06, + "loss": 0.22753967344760895, + "step": 217 + }, + { + "epoch": 0.12195804195804195, + "grad_norm": 0.7009149117720636, + "learning_rate": 8.081936685288641e-06, + "loss": 0.2552938759326935, + "step": 218 + }, + { + "epoch": 0.12251748251748251, + "grad_norm": 0.5243076999525044, + "learning_rate": 8.119180633147115e-06, + "loss": 0.19879496097564697, + "step": 219 + }, + { + "epoch": 0.12307692307692308, + "grad_norm": 0.6604310637848751, + "learning_rate": 8.156424581005588e-06, + "loss": 0.167031928896904, + "step": 220 + }, + { + "epoch": 0.12363636363636364, + "grad_norm": 0.44833817705326096, + "learning_rate": 8.19366852886406e-06, + "loss": 0.09976990520954132, + "step": 221 + }, + { + "epoch": 0.1241958041958042, + "grad_norm": 0.652691392869524, + "learning_rate": 8.230912476722534e-06, + "loss": 0.25700443983078003, + "step": 222 + }, + { + "epoch": 0.12475524475524476, + "grad_norm": 0.34752286192290194, + "learning_rate": 8.268156424581007e-06, + "loss": 0.06720789521932602, + "step": 223 + }, + { + "epoch": 0.12531468531468531, + "grad_norm": 0.48828261818390517, + "learning_rate": 8.305400372439479e-06, + "loss": 0.1633196324110031, + "step": 224 + }, + { + "epoch": 0.1258741258741259, + "grad_norm": 0.38606848095574003, + "learning_rate": 8.342644320297953e-06, + "loss": 0.07755030691623688, + "step": 225 + }, + { + "epoch": 0.12643356643356643, + "grad_norm": 0.5566136173662624, + "learning_rate": 8.379888268156426e-06, + "loss": 0.22566697001457214, + "step": 226 + }, + { + "epoch": 0.126993006993007, + "grad_norm": 0.49229785274045423, + "learning_rate": 8.417132216014898e-06, + "loss": 0.14982138574123383, + "step": 227 + }, + { + "epoch": 0.12755244755244755, + "grad_norm": 0.686301101100574, + "learning_rate": 8.454376163873371e-06, + "loss": 0.2582307457923889, + "step": 228 + }, + { + "epoch": 0.12811188811188812, + "grad_norm": 0.8091035555451601, + "learning_rate": 8.491620111731845e-06, + "loss": 0.32020890712738037, + "step": 229 + }, + { + "epoch": 0.12867132867132866, + "grad_norm": 1.020101901153558, + "learning_rate": 8.528864059590317e-06, + "loss": 0.20898734033107758, + "step": 230 + }, + { + "epoch": 0.12923076923076923, + "grad_norm": 0.48127630900755053, + "learning_rate": 8.56610800744879e-06, + "loss": 0.07202384620904922, + "step": 231 + }, + { + "epoch": 0.1297902097902098, + "grad_norm": 0.39127525829929455, + "learning_rate": 8.603351955307264e-06, + "loss": 0.08093154430389404, + "step": 232 + }, + { + "epoch": 0.13034965034965035, + "grad_norm": 0.5769393195836693, + "learning_rate": 8.640595903165736e-06, + "loss": 0.13062073290348053, + "step": 233 + }, + { + "epoch": 0.13090909090909092, + "grad_norm": 0.5447646757220693, + "learning_rate": 8.677839851024209e-06, + "loss": 0.09924593567848206, + "step": 234 + }, + { + "epoch": 0.13146853146853146, + "grad_norm": 0.7044458800997087, + "learning_rate": 8.715083798882683e-06, + "loss": 0.23715560138225555, + "step": 235 + }, + { + "epoch": 0.13202797202797203, + "grad_norm": 0.7873483770941294, + "learning_rate": 8.752327746741154e-06, + "loss": 0.20219574868679047, + "step": 236 + }, + { + "epoch": 0.13258741258741258, + "grad_norm": 0.5588693597608008, + "learning_rate": 8.789571694599628e-06, + "loss": 0.17427216470241547, + "step": 237 + }, + { + "epoch": 0.13314685314685315, + "grad_norm": 0.9391007771290981, + "learning_rate": 8.826815642458101e-06, + "loss": 0.38720574975013733, + "step": 238 + }, + { + "epoch": 0.1337062937062937, + "grad_norm": 0.7545104473413679, + "learning_rate": 8.864059590316573e-06, + "loss": 0.32538652420043945, + "step": 239 + }, + { + "epoch": 0.13426573426573427, + "grad_norm": 0.6993839939827596, + "learning_rate": 8.901303538175047e-06, + "loss": 0.23696181178092957, + "step": 240 + }, + { + "epoch": 0.13482517482517484, + "grad_norm": 0.8211256629741615, + "learning_rate": 8.93854748603352e-06, + "loss": 0.25718289613723755, + "step": 241 + }, + { + "epoch": 0.13538461538461538, + "grad_norm": 0.5491962251930167, + "learning_rate": 8.975791433891994e-06, + "loss": 0.14556273818016052, + "step": 242 + }, + { + "epoch": 0.13594405594405595, + "grad_norm": 0.6022405212712698, + "learning_rate": 9.013035381750467e-06, + "loss": 0.16508439183235168, + "step": 243 + }, + { + "epoch": 0.1365034965034965, + "grad_norm": 0.45024317974568145, + "learning_rate": 9.050279329608939e-06, + "loss": 0.1431969553232193, + "step": 244 + }, + { + "epoch": 0.13706293706293707, + "grad_norm": 0.5531474223624964, + "learning_rate": 9.087523277467413e-06, + "loss": 0.12685249745845795, + "step": 245 + }, + { + "epoch": 0.1376223776223776, + "grad_norm": 1.0315305325352482, + "learning_rate": 9.124767225325886e-06, + "loss": 0.24326680600643158, + "step": 246 + }, + { + "epoch": 0.13818181818181818, + "grad_norm": 0.5537570189614346, + "learning_rate": 9.162011173184358e-06, + "loss": 0.18389342725276947, + "step": 247 + }, + { + "epoch": 0.13874125874125876, + "grad_norm": 0.7418339921835205, + "learning_rate": 9.199255121042831e-06, + "loss": 0.387996643781662, + "step": 248 + }, + { + "epoch": 0.1393006993006993, + "grad_norm": 0.5020799590111263, + "learning_rate": 9.236499068901305e-06, + "loss": 0.1683967113494873, + "step": 249 + }, + { + "epoch": 0.13986013986013987, + "grad_norm": 0.7823388148211736, + "learning_rate": 9.273743016759777e-06, + "loss": 0.38224977254867554, + "step": 250 + }, + { + "epoch": 0.14041958041958041, + "grad_norm": 1.104000008614978, + "learning_rate": 9.31098696461825e-06, + "loss": 0.2202875018119812, + "step": 251 + }, + { + "epoch": 0.14097902097902099, + "grad_norm": 0.5829755241773205, + "learning_rate": 9.348230912476724e-06, + "loss": 0.0943504124879837, + "step": 252 + }, + { + "epoch": 0.14153846153846153, + "grad_norm": 0.36780804544026413, + "learning_rate": 9.385474860335197e-06, + "loss": 0.051278285682201385, + "step": 253 + }, + { + "epoch": 0.1420979020979021, + "grad_norm": 0.4997772674252771, + "learning_rate": 9.422718808193669e-06, + "loss": 0.16502758860588074, + "step": 254 + }, + { + "epoch": 0.14265734265734265, + "grad_norm": 0.5570026379144322, + "learning_rate": 9.459962756052143e-06, + "loss": 0.2247346043586731, + "step": 255 + }, + { + "epoch": 0.14321678321678322, + "grad_norm": 0.6020269132258821, + "learning_rate": 9.497206703910616e-06, + "loss": 0.18743066489696503, + "step": 256 + }, + { + "epoch": 0.1437762237762238, + "grad_norm": 0.4406667394127413, + "learning_rate": 9.534450651769088e-06, + "loss": 0.13635236024856567, + "step": 257 + }, + { + "epoch": 0.14433566433566433, + "grad_norm": 0.8084545170696706, + "learning_rate": 9.571694599627561e-06, + "loss": 0.29340460896492004, + "step": 258 + }, + { + "epoch": 0.1448951048951049, + "grad_norm": 0.5693011114707053, + "learning_rate": 9.608938547486035e-06, + "loss": 0.2556872069835663, + "step": 259 + }, + { + "epoch": 0.14545454545454545, + "grad_norm": 0.7165905688517574, + "learning_rate": 9.646182495344507e-06, + "loss": 0.27386772632598877, + "step": 260 + }, + { + "epoch": 0.14601398601398602, + "grad_norm": 0.3491997953007715, + "learning_rate": 9.68342644320298e-06, + "loss": 0.03233690559864044, + "step": 261 + }, + { + "epoch": 0.14657342657342656, + "grad_norm": 0.572730431249581, + "learning_rate": 9.720670391061454e-06, + "loss": 0.26094627380371094, + "step": 262 + }, + { + "epoch": 0.14713286713286713, + "grad_norm": 0.5503332873080298, + "learning_rate": 9.757914338919926e-06, + "loss": 0.1413864940404892, + "step": 263 + }, + { + "epoch": 0.1476923076923077, + "grad_norm": 0.5698314472134977, + "learning_rate": 9.795158286778399e-06, + "loss": 0.1462213695049286, + "step": 264 + }, + { + "epoch": 0.14825174825174825, + "grad_norm": 0.7348138832125692, + "learning_rate": 9.832402234636873e-06, + "loss": 0.44167637825012207, + "step": 265 + }, + { + "epoch": 0.14881118881118882, + "grad_norm": 0.8414470492306425, + "learning_rate": 9.869646182495344e-06, + "loss": 0.4117547869682312, + "step": 266 + }, + { + "epoch": 0.14937062937062937, + "grad_norm": 0.9217289679034341, + "learning_rate": 9.906890130353818e-06, + "loss": 0.3823855519294739, + "step": 267 + }, + { + "epoch": 0.14993006993006994, + "grad_norm": 0.5426616220941106, + "learning_rate": 9.944134078212291e-06, + "loss": 0.14567437767982483, + "step": 268 + }, + { + "epoch": 0.15048951048951048, + "grad_norm": 0.7326722101157161, + "learning_rate": 9.981378026070763e-06, + "loss": 0.32547974586486816, + "step": 269 + }, + { + "epoch": 0.15104895104895105, + "grad_norm": 0.5071425100066824, + "learning_rate": 1.0018621973929237e-05, + "loss": 0.21595175564289093, + "step": 270 + }, + { + "epoch": 0.1516083916083916, + "grad_norm": 0.7891324374802596, + "learning_rate": 1.005586592178771e-05, + "loss": 0.2322026938199997, + "step": 271 + }, + { + "epoch": 0.15216783216783217, + "grad_norm": 0.5151565705162546, + "learning_rate": 1.0093109869646182e-05, + "loss": 0.14051656424999237, + "step": 272 + }, + { + "epoch": 0.15272727272727274, + "grad_norm": 0.8444891131495129, + "learning_rate": 1.0130353817504657e-05, + "loss": 0.30927422642707825, + "step": 273 + }, + { + "epoch": 0.15328671328671328, + "grad_norm": 0.451737703401404, + "learning_rate": 1.0167597765363129e-05, + "loss": 0.14183266460895538, + "step": 274 + }, + { + "epoch": 0.15384615384615385, + "grad_norm": 0.4963734172486707, + "learning_rate": 1.0204841713221603e-05, + "loss": 0.11704777181148529, + "step": 275 + }, + { + "epoch": 0.1544055944055944, + "grad_norm": 0.5860846651164688, + "learning_rate": 1.0242085661080074e-05, + "loss": 0.17365817725658417, + "step": 276 + }, + { + "epoch": 0.15496503496503497, + "grad_norm": 0.5934129833354985, + "learning_rate": 1.0279329608938548e-05, + "loss": 0.14538846909999847, + "step": 277 + }, + { + "epoch": 0.15552447552447551, + "grad_norm": 0.7809911740781067, + "learning_rate": 1.031657355679702e-05, + "loss": 0.2588534653186798, + "step": 278 + }, + { + "epoch": 0.15608391608391609, + "grad_norm": 0.8151145241388195, + "learning_rate": 1.0353817504655495e-05, + "loss": 0.2705457806587219, + "step": 279 + }, + { + "epoch": 0.15664335664335666, + "grad_norm": 0.6679457584682315, + "learning_rate": 1.0391061452513967e-05, + "loss": 0.24899393320083618, + "step": 280 + }, + { + "epoch": 0.1572027972027972, + "grad_norm": 0.8128782344125784, + "learning_rate": 1.042830540037244e-05, + "loss": 0.20160573720932007, + "step": 281 + }, + { + "epoch": 0.15776223776223777, + "grad_norm": 0.720068785915854, + "learning_rate": 1.0465549348230912e-05, + "loss": 0.25672706961631775, + "step": 282 + }, + { + "epoch": 0.15832167832167832, + "grad_norm": 0.844626732741477, + "learning_rate": 1.0502793296089386e-05, + "loss": 0.16140006482601166, + "step": 283 + }, + { + "epoch": 0.1588811188811189, + "grad_norm": 0.7767357327917033, + "learning_rate": 1.054003724394786e-05, + "loss": 0.30654963850975037, + "step": 284 + }, + { + "epoch": 0.15944055944055943, + "grad_norm": 0.5477205926769453, + "learning_rate": 1.0577281191806333e-05, + "loss": 0.15222902595996857, + "step": 285 + }, + { + "epoch": 0.16, + "grad_norm": 0.724389162541715, + "learning_rate": 1.0614525139664806e-05, + "loss": 0.2799169719219208, + "step": 286 + }, + { + "epoch": 0.16055944055944055, + "grad_norm": 0.8121678192116165, + "learning_rate": 1.0651769087523278e-05, + "loss": 0.2820596396923065, + "step": 287 + }, + { + "epoch": 0.16111888111888112, + "grad_norm": 0.580923466388924, + "learning_rate": 1.0689013035381753e-05, + "loss": 0.26518529653549194, + "step": 288 + }, + { + "epoch": 0.1616783216783217, + "grad_norm": 0.6241217378714216, + "learning_rate": 1.0726256983240225e-05, + "loss": 0.25247806310653687, + "step": 289 + }, + { + "epoch": 0.16223776223776223, + "grad_norm": 0.5898111507534417, + "learning_rate": 1.0763500931098698e-05, + "loss": 0.3577049970626831, + "step": 290 + }, + { + "epoch": 0.1627972027972028, + "grad_norm": 0.5785102991589665, + "learning_rate": 1.080074487895717e-05, + "loss": 0.31357356905937195, + "step": 291 + }, + { + "epoch": 0.16335664335664335, + "grad_norm": 0.5884151504940158, + "learning_rate": 1.0837988826815644e-05, + "loss": 0.17181327939033508, + "step": 292 + }, + { + "epoch": 0.16391608391608392, + "grad_norm": 0.729802094704648, + "learning_rate": 1.0875232774674116e-05, + "loss": 0.38565880060195923, + "step": 293 + }, + { + "epoch": 0.16447552447552446, + "grad_norm": 0.9339336377124192, + "learning_rate": 1.091247672253259e-05, + "loss": 0.4313648045063019, + "step": 294 + }, + { + "epoch": 0.16503496503496504, + "grad_norm": 0.6654965048495728, + "learning_rate": 1.0949720670391063e-05, + "loss": 0.2349538952112198, + "step": 295 + }, + { + "epoch": 0.1655944055944056, + "grad_norm": 0.36061205789408024, + "learning_rate": 1.0986964618249536e-05, + "loss": 0.08374534547328949, + "step": 296 + }, + { + "epoch": 0.16615384615384615, + "grad_norm": 0.6616721156623862, + "learning_rate": 1.1024208566108008e-05, + "loss": 0.237091526389122, + "step": 297 + }, + { + "epoch": 0.16671328671328672, + "grad_norm": 0.5522819645702962, + "learning_rate": 1.1061452513966481e-05, + "loss": 0.16771084070205688, + "step": 298 + }, + { + "epoch": 0.16727272727272727, + "grad_norm": 0.6768703185116618, + "learning_rate": 1.1098696461824953e-05, + "loss": 0.2835752069950104, + "step": 299 + }, + { + "epoch": 0.16783216783216784, + "grad_norm": 0.30161296674586735, + "learning_rate": 1.1135940409683428e-05, + "loss": 0.06887897849082947, + "step": 300 + }, + { + "epoch": 0.16839160839160838, + "grad_norm": 1.1320605649984188, + "learning_rate": 1.11731843575419e-05, + "loss": 0.3984430730342865, + "step": 301 + }, + { + "epoch": 0.16895104895104895, + "grad_norm": 0.6688688093737563, + "learning_rate": 1.1210428305400374e-05, + "loss": 0.35076266527175903, + "step": 302 + }, + { + "epoch": 0.1695104895104895, + "grad_norm": 0.4834406884990006, + "learning_rate": 1.1247672253258846e-05, + "loss": 0.21225865185260773, + "step": 303 + }, + { + "epoch": 0.17006993006993007, + "grad_norm": 0.5373191463488853, + "learning_rate": 1.1284916201117319e-05, + "loss": 0.15718327462673187, + "step": 304 + }, + { + "epoch": 0.17062937062937064, + "grad_norm": 0.617765733043953, + "learning_rate": 1.1322160148975791e-05, + "loss": 0.22824551165103912, + "step": 305 + }, + { + "epoch": 0.17118881118881119, + "grad_norm": 0.9155653156598743, + "learning_rate": 1.1359404096834266e-05, + "loss": 0.6170090436935425, + "step": 306 + }, + { + "epoch": 0.17174825174825176, + "grad_norm": 0.4225456898265531, + "learning_rate": 1.1396648044692738e-05, + "loss": 0.11314629763364792, + "step": 307 + }, + { + "epoch": 0.1723076923076923, + "grad_norm": 0.6489848691475049, + "learning_rate": 1.1433891992551211e-05, + "loss": 0.29253992438316345, + "step": 308 + }, + { + "epoch": 0.17286713286713287, + "grad_norm": 0.49203961402107055, + "learning_rate": 1.1471135940409683e-05, + "loss": 0.16079992055892944, + "step": 309 + }, + { + "epoch": 0.17342657342657342, + "grad_norm": 0.8466214129744567, + "learning_rate": 1.1508379888268157e-05, + "loss": 0.23877716064453125, + "step": 310 + }, + { + "epoch": 0.173986013986014, + "grad_norm": 0.8340626157208982, + "learning_rate": 1.1545623836126629e-05, + "loss": 0.09725035727024078, + "step": 311 + }, + { + "epoch": 0.17454545454545456, + "grad_norm": 0.8373905513302456, + "learning_rate": 1.1582867783985104e-05, + "loss": 0.3041437268257141, + "step": 312 + }, + { + "epoch": 0.1751048951048951, + "grad_norm": 0.4688700315829674, + "learning_rate": 1.1620111731843577e-05, + "loss": 0.14455457031726837, + "step": 313 + }, + { + "epoch": 0.17566433566433567, + "grad_norm": 0.9255041450216211, + "learning_rate": 1.1657355679702049e-05, + "loss": 0.13114410638809204, + "step": 314 + }, + { + "epoch": 0.17622377622377622, + "grad_norm": 0.581358122540598, + "learning_rate": 1.1694599627560523e-05, + "loss": 0.1356728971004486, + "step": 315 + }, + { + "epoch": 0.1767832167832168, + "grad_norm": 0.7533779803118279, + "learning_rate": 1.1731843575418994e-05, + "loss": 0.2697957754135132, + "step": 316 + }, + { + "epoch": 0.17734265734265733, + "grad_norm": 0.8276650267845913, + "learning_rate": 1.176908752327747e-05, + "loss": 0.41352230310440063, + "step": 317 + }, + { + "epoch": 0.1779020979020979, + "grad_norm": 0.6459565399358903, + "learning_rate": 1.1806331471135941e-05, + "loss": 0.1763618290424347, + "step": 318 + }, + { + "epoch": 0.17846153846153845, + "grad_norm": 0.4045700901878682, + "learning_rate": 1.1843575418994415e-05, + "loss": 0.06649969518184662, + "step": 319 + }, + { + "epoch": 0.17902097902097902, + "grad_norm": 0.6208241360902059, + "learning_rate": 1.1880819366852887e-05, + "loss": 0.2024984359741211, + "step": 320 + }, + { + "epoch": 0.1795804195804196, + "grad_norm": 0.6910430202726695, + "learning_rate": 1.1918063314711362e-05, + "loss": 0.18501436710357666, + "step": 321 + }, + { + "epoch": 0.18013986013986014, + "grad_norm": 0.5600455237540202, + "learning_rate": 1.1955307262569834e-05, + "loss": 0.28460073471069336, + "step": 322 + }, + { + "epoch": 0.1806993006993007, + "grad_norm": 0.6176537982609474, + "learning_rate": 1.1992551210428307e-05, + "loss": 0.18380779027938843, + "step": 323 + }, + { + "epoch": 0.18125874125874125, + "grad_norm": 0.8298147371967786, + "learning_rate": 1.2029795158286779e-05, + "loss": 0.4374958574771881, + "step": 324 + }, + { + "epoch": 0.18181818181818182, + "grad_norm": 0.7660793117835693, + "learning_rate": 1.2067039106145253e-05, + "loss": 0.21650944650173187, + "step": 325 + }, + { + "epoch": 0.18237762237762237, + "grad_norm": 0.5471908137135216, + "learning_rate": 1.2104283054003724e-05, + "loss": 0.21292313933372498, + "step": 326 + }, + { + "epoch": 0.18293706293706294, + "grad_norm": 0.7463851077752695, + "learning_rate": 1.21415270018622e-05, + "loss": 0.3288453221321106, + "step": 327 + }, + { + "epoch": 0.1834965034965035, + "grad_norm": 0.6146480887854827, + "learning_rate": 1.2178770949720671e-05, + "loss": 0.1771489679813385, + "step": 328 + }, + { + "epoch": 0.18405594405594405, + "grad_norm": 0.7005570898138087, + "learning_rate": 1.2216014897579145e-05, + "loss": 0.11406119167804718, + "step": 329 + }, + { + "epoch": 0.18461538461538463, + "grad_norm": 0.8046540508718654, + "learning_rate": 1.2253258845437617e-05, + "loss": 0.24953454732894897, + "step": 330 + }, + { + "epoch": 0.18517482517482517, + "grad_norm": 0.3391938253523961, + "learning_rate": 1.229050279329609e-05, + "loss": 0.07181106507778168, + "step": 331 + }, + { + "epoch": 0.18573426573426574, + "grad_norm": 0.6496290104028324, + "learning_rate": 1.2327746741154562e-05, + "loss": 0.2392260581254959, + "step": 332 + }, + { + "epoch": 0.18629370629370628, + "grad_norm": 0.3999115873146793, + "learning_rate": 1.2364990689013037e-05, + "loss": 0.10333205759525299, + "step": 333 + }, + { + "epoch": 0.18685314685314686, + "grad_norm": 0.9250737498105759, + "learning_rate": 1.2402234636871509e-05, + "loss": 0.1868247389793396, + "step": 334 + }, + { + "epoch": 0.1874125874125874, + "grad_norm": 0.33885575750053587, + "learning_rate": 1.2439478584729983e-05, + "loss": 0.0466235987842083, + "step": 335 + }, + { + "epoch": 0.18797202797202797, + "grad_norm": 0.6584923743315388, + "learning_rate": 1.2476722532588454e-05, + "loss": 0.2161976397037506, + "step": 336 + }, + { + "epoch": 0.18853146853146854, + "grad_norm": 0.5482362805515104, + "learning_rate": 1.2513966480446928e-05, + "loss": 0.2055448591709137, + "step": 337 + }, + { + "epoch": 0.1890909090909091, + "grad_norm": 0.6136816950527892, + "learning_rate": 1.25512104283054e-05, + "loss": 0.25013861060142517, + "step": 338 + }, + { + "epoch": 0.18965034965034966, + "grad_norm": 0.4736262453651948, + "learning_rate": 1.2588454376163875e-05, + "loss": 0.1826154589653015, + "step": 339 + }, + { + "epoch": 0.1902097902097902, + "grad_norm": 0.8362505973081411, + "learning_rate": 1.2625698324022347e-05, + "loss": 0.4009782671928406, + "step": 340 + }, + { + "epoch": 0.19076923076923077, + "grad_norm": 0.6793250721140438, + "learning_rate": 1.266294227188082e-05, + "loss": 0.2914050221443176, + "step": 341 + }, + { + "epoch": 0.19132867132867132, + "grad_norm": 0.532885036819345, + "learning_rate": 1.2700186219739294e-05, + "loss": 0.11122561991214752, + "step": 342 + }, + { + "epoch": 0.1918881118881119, + "grad_norm": 0.5371928148513151, + "learning_rate": 1.2737430167597766e-05, + "loss": 0.18652702867984772, + "step": 343 + }, + { + "epoch": 0.19244755244755246, + "grad_norm": 0.6431117838712368, + "learning_rate": 1.277467411545624e-05, + "loss": 0.13408324122428894, + "step": 344 + }, + { + "epoch": 0.193006993006993, + "grad_norm": 0.6304694836569884, + "learning_rate": 1.2811918063314713e-05, + "loss": 0.13975383341312408, + "step": 345 + }, + { + "epoch": 0.19356643356643358, + "grad_norm": 0.5224234308517259, + "learning_rate": 1.2849162011173186e-05, + "loss": 0.15843935310840607, + "step": 346 + }, + { + "epoch": 0.19412587412587412, + "grad_norm": 0.7489998702696299, + "learning_rate": 1.2886405959031658e-05, + "loss": 0.37123599648475647, + "step": 347 + }, + { + "epoch": 0.1946853146853147, + "grad_norm": 0.6083028860273543, + "learning_rate": 1.2923649906890131e-05, + "loss": 0.17427772283554077, + "step": 348 + }, + { + "epoch": 0.19524475524475524, + "grad_norm": 0.6662261495695245, + "learning_rate": 1.2960893854748603e-05, + "loss": 0.4043179452419281, + "step": 349 + }, + { + "epoch": 0.1958041958041958, + "grad_norm": 0.5069117407293394, + "learning_rate": 1.2998137802607078e-05, + "loss": 0.22299587726593018, + "step": 350 + }, + { + "epoch": 0.19636363636363635, + "grad_norm": 0.9067926677299918, + "learning_rate": 1.303538175046555e-05, + "loss": 0.30284884572029114, + "step": 351 + }, + { + "epoch": 0.19692307692307692, + "grad_norm": 0.7228059068415302, + "learning_rate": 1.3072625698324024e-05, + "loss": 0.3166622221469879, + "step": 352 + }, + { + "epoch": 0.1974825174825175, + "grad_norm": 0.8820815856470334, + "learning_rate": 1.3109869646182496e-05, + "loss": 0.395719051361084, + "step": 353 + }, + { + "epoch": 0.19804195804195804, + "grad_norm": 0.590081201250214, + "learning_rate": 1.314711359404097e-05, + "loss": 0.16517296433448792, + "step": 354 + }, + { + "epoch": 0.1986013986013986, + "grad_norm": 0.44773378507921335, + "learning_rate": 1.3184357541899443e-05, + "loss": 0.14390204846858978, + "step": 355 + }, + { + "epoch": 0.19916083916083915, + "grad_norm": 0.5968814648601309, + "learning_rate": 1.3221601489757916e-05, + "loss": 0.15885721147060394, + "step": 356 + }, + { + "epoch": 0.19972027972027973, + "grad_norm": 0.6547487385393078, + "learning_rate": 1.3258845437616388e-05, + "loss": 0.18029023706912994, + "step": 357 + }, + { + "epoch": 0.20027972027972027, + "grad_norm": 0.5983331830663939, + "learning_rate": 1.3296089385474861e-05, + "loss": 0.24055664241313934, + "step": 358 + }, + { + "epoch": 0.20083916083916084, + "grad_norm": 0.8391345751537961, + "learning_rate": 1.3333333333333333e-05, + "loss": 0.40840989351272583, + "step": 359 + }, + { + "epoch": 0.2013986013986014, + "grad_norm": 1.042804442915725, + "learning_rate": 1.3370577281191808e-05, + "loss": 0.3558077812194824, + "step": 360 + }, + { + "epoch": 0.20195804195804196, + "grad_norm": 0.5443590371759988, + "learning_rate": 1.340782122905028e-05, + "loss": 0.10893086344003677, + "step": 361 + }, + { + "epoch": 0.20251748251748253, + "grad_norm": 0.6785519863268369, + "learning_rate": 1.3445065176908754e-05, + "loss": 0.26309388875961304, + "step": 362 + }, + { + "epoch": 0.20307692307692307, + "grad_norm": 0.7242176721708261, + "learning_rate": 1.3482309124767226e-05, + "loss": 0.1265127807855606, + "step": 363 + }, + { + "epoch": 0.20363636363636364, + "grad_norm": 0.7270758840973456, + "learning_rate": 1.3519553072625699e-05, + "loss": 0.23097705841064453, + "step": 364 + }, + { + "epoch": 0.2041958041958042, + "grad_norm": 0.33653832301156905, + "learning_rate": 1.3556797020484171e-05, + "loss": 0.06925821304321289, + "step": 365 + }, + { + "epoch": 0.20475524475524476, + "grad_norm": 0.5773415837717109, + "learning_rate": 1.3594040968342646e-05, + "loss": 0.16943222284317017, + "step": 366 + }, + { + "epoch": 0.2053146853146853, + "grad_norm": 0.45532913479820497, + "learning_rate": 1.3631284916201118e-05, + "loss": 0.0987367033958435, + "step": 367 + }, + { + "epoch": 0.20587412587412587, + "grad_norm": 0.8448104478398881, + "learning_rate": 1.3668528864059591e-05, + "loss": 0.17131972312927246, + "step": 368 + }, + { + "epoch": 0.20643356643356645, + "grad_norm": 0.6397931486285741, + "learning_rate": 1.3705772811918065e-05, + "loss": 0.16666218638420105, + "step": 369 + }, + { + "epoch": 0.206993006993007, + "grad_norm": 0.48299565904306213, + "learning_rate": 1.3743016759776537e-05, + "loss": 0.1290956735610962, + "step": 370 + }, + { + "epoch": 0.20755244755244756, + "grad_norm": 0.7737542585655884, + "learning_rate": 1.3780260707635012e-05, + "loss": 0.4316938519477844, + "step": 371 + }, + { + "epoch": 0.2081118881118881, + "grad_norm": 0.7073621532672992, + "learning_rate": 1.3817504655493484e-05, + "loss": 0.31795668601989746, + "step": 372 + }, + { + "epoch": 0.20867132867132868, + "grad_norm": 0.5214326908275703, + "learning_rate": 1.3854748603351957e-05, + "loss": 0.10897572338581085, + "step": 373 + }, + { + "epoch": 0.20923076923076922, + "grad_norm": 0.7037145715748028, + "learning_rate": 1.389199255121043e-05, + "loss": 0.22178541123867035, + "step": 374 + }, + { + "epoch": 0.2097902097902098, + "grad_norm": 0.5055661435612474, + "learning_rate": 1.3929236499068903e-05, + "loss": 0.15575313568115234, + "step": 375 + }, + { + "epoch": 0.21034965034965036, + "grad_norm": 0.7900454887121559, + "learning_rate": 1.3966480446927374e-05, + "loss": 0.3155873119831085, + "step": 376 + }, + { + "epoch": 0.2109090909090909, + "grad_norm": 0.5858936491106438, + "learning_rate": 1.400372439478585e-05, + "loss": 0.25676119327545166, + "step": 377 + }, + { + "epoch": 0.21146853146853148, + "grad_norm": 0.5402137158941014, + "learning_rate": 1.4040968342644321e-05, + "loss": 0.18600091338157654, + "step": 378 + }, + { + "epoch": 0.21202797202797202, + "grad_norm": 0.5463613671447719, + "learning_rate": 1.4078212290502795e-05, + "loss": 0.1625310480594635, + "step": 379 + }, + { + "epoch": 0.2125874125874126, + "grad_norm": 0.7270768862588064, + "learning_rate": 1.4115456238361267e-05, + "loss": 0.42895030975341797, + "step": 380 + }, + { + "epoch": 0.21314685314685314, + "grad_norm": 0.5746645461098919, + "learning_rate": 1.415270018621974e-05, + "loss": 0.29572248458862305, + "step": 381 + }, + { + "epoch": 0.2137062937062937, + "grad_norm": 0.5669696588701618, + "learning_rate": 1.4189944134078212e-05, + "loss": 0.21049325168132782, + "step": 382 + }, + { + "epoch": 0.21426573426573425, + "grad_norm": 0.8517199773761903, + "learning_rate": 1.4227188081936687e-05, + "loss": 0.4587564468383789, + "step": 383 + }, + { + "epoch": 0.21482517482517482, + "grad_norm": 0.5153907063980031, + "learning_rate": 1.426443202979516e-05, + "loss": 0.19551566243171692, + "step": 384 + }, + { + "epoch": 0.2153846153846154, + "grad_norm": 0.6933943213439778, + "learning_rate": 1.4301675977653633e-05, + "loss": 0.30103886127471924, + "step": 385 + }, + { + "epoch": 0.21594405594405594, + "grad_norm": 0.7319981936288447, + "learning_rate": 1.4338919925512104e-05, + "loss": 0.3774051070213318, + "step": 386 + }, + { + "epoch": 0.2165034965034965, + "grad_norm": 0.3399386793705644, + "learning_rate": 1.4376163873370578e-05, + "loss": 0.12075591832399368, + "step": 387 + }, + { + "epoch": 0.21706293706293706, + "grad_norm": 0.6631360073973969, + "learning_rate": 1.4413407821229052e-05, + "loss": 0.2473677098751068, + "step": 388 + }, + { + "epoch": 0.21762237762237763, + "grad_norm": 0.6052552693807103, + "learning_rate": 1.4450651769087525e-05, + "loss": 0.26568669080734253, + "step": 389 + }, + { + "epoch": 0.21818181818181817, + "grad_norm": 0.6734663670184445, + "learning_rate": 1.4487895716945997e-05, + "loss": 0.22834305465221405, + "step": 390 + }, + { + "epoch": 0.21874125874125874, + "grad_norm": 0.5324636484634115, + "learning_rate": 1.452513966480447e-05, + "loss": 0.19060564041137695, + "step": 391 + }, + { + "epoch": 0.21930069930069931, + "grad_norm": 0.6713721883676342, + "learning_rate": 1.4562383612662942e-05, + "loss": 0.3759705424308777, + "step": 392 + }, + { + "epoch": 0.21986013986013986, + "grad_norm": 0.6058998702582413, + "learning_rate": 1.4599627560521417e-05, + "loss": 0.1936790645122528, + "step": 393 + }, + { + "epoch": 0.22041958041958043, + "grad_norm": 0.7249544475309077, + "learning_rate": 1.463687150837989e-05, + "loss": 0.21975558996200562, + "step": 394 + }, + { + "epoch": 0.22097902097902097, + "grad_norm": 0.64769916677911, + "learning_rate": 1.4674115456238363e-05, + "loss": 0.2661827504634857, + "step": 395 + }, + { + "epoch": 0.22153846153846155, + "grad_norm": 0.7758228576770663, + "learning_rate": 1.4711359404096834e-05, + "loss": 0.41784244775772095, + "step": 396 + }, + { + "epoch": 0.2220979020979021, + "grad_norm": 0.6572200097723351, + "learning_rate": 1.4748603351955308e-05, + "loss": 0.375749409198761, + "step": 397 + }, + { + "epoch": 0.22265734265734266, + "grad_norm": 0.6505631598724294, + "learning_rate": 1.4785847299813783e-05, + "loss": 0.05348823219537735, + "step": 398 + }, + { + "epoch": 0.2232167832167832, + "grad_norm": 0.5333885112813971, + "learning_rate": 1.4823091247672255e-05, + "loss": 0.1662866175174713, + "step": 399 + }, + { + "epoch": 0.22377622377622378, + "grad_norm": 0.7098493944046977, + "learning_rate": 1.4860335195530729e-05, + "loss": 0.1660555601119995, + "step": 400 + }, + { + "epoch": 0.22433566433566435, + "grad_norm": 0.559503415532578, + "learning_rate": 1.48975791433892e-05, + "loss": 0.18232253193855286, + "step": 401 + }, + { + "epoch": 0.2248951048951049, + "grad_norm": 0.5405522358407641, + "learning_rate": 1.4934823091247674e-05, + "loss": 0.2067871391773224, + "step": 402 + }, + { + "epoch": 0.22545454545454546, + "grad_norm": 0.5603521012233286, + "learning_rate": 1.4972067039106146e-05, + "loss": 0.2378668189048767, + "step": 403 + }, + { + "epoch": 0.226013986013986, + "grad_norm": 0.434453316509424, + "learning_rate": 1.5009310986964621e-05, + "loss": 0.10804593563079834, + "step": 404 + }, + { + "epoch": 0.22657342657342658, + "grad_norm": 0.5600776629829546, + "learning_rate": 1.5046554934823093e-05, + "loss": 0.21366216242313385, + "step": 405 + }, + { + "epoch": 0.22713286713286712, + "grad_norm": 0.6871361326838401, + "learning_rate": 1.5083798882681566e-05, + "loss": 0.3075847327709198, + "step": 406 + }, + { + "epoch": 0.2276923076923077, + "grad_norm": 0.5798771611165667, + "learning_rate": 1.5121042830540038e-05, + "loss": 0.21630480885505676, + "step": 407 + }, + { + "epoch": 0.22825174825174827, + "grad_norm": 0.3940285611968662, + "learning_rate": 1.5158286778398512e-05, + "loss": 0.0948122888803482, + "step": 408 + }, + { + "epoch": 0.2288111888111888, + "grad_norm": 0.5615871678940072, + "learning_rate": 1.5195530726256983e-05, + "loss": 0.0394311286509037, + "step": 409 + }, + { + "epoch": 0.22937062937062938, + "grad_norm": 0.1196596682742178, + "learning_rate": 1.5232774674115459e-05, + "loss": 0.004848596174269915, + "step": 410 + }, + { + "epoch": 0.22993006993006992, + "grad_norm": 0.5617628463733623, + "learning_rate": 1.527001862197393e-05, + "loss": 0.16655394434928894, + "step": 411 + }, + { + "epoch": 0.2304895104895105, + "grad_norm": 0.8215200423147032, + "learning_rate": 1.5307262569832404e-05, + "loss": 0.4191613793373108, + "step": 412 + }, + { + "epoch": 0.23104895104895104, + "grad_norm": 0.5950080419515392, + "learning_rate": 1.5344506517690876e-05, + "loss": 0.22198426723480225, + "step": 413 + }, + { + "epoch": 0.2316083916083916, + "grad_norm": 0.6591846890228801, + "learning_rate": 1.538175046554935e-05, + "loss": 0.11012548953294754, + "step": 414 + }, + { + "epoch": 0.23216783216783216, + "grad_norm": 0.5707269042242166, + "learning_rate": 1.5418994413407823e-05, + "loss": 0.18286389112472534, + "step": 415 + }, + { + "epoch": 0.23272727272727273, + "grad_norm": 0.5174886839599465, + "learning_rate": 1.5456238361266295e-05, + "loss": 0.11632704734802246, + "step": 416 + }, + { + "epoch": 0.2332867132867133, + "grad_norm": 0.5002932874391136, + "learning_rate": 1.5493482309124766e-05, + "loss": 0.18329191207885742, + "step": 417 + }, + { + "epoch": 0.23384615384615384, + "grad_norm": 0.49876466339961706, + "learning_rate": 1.553072625698324e-05, + "loss": 0.1778550148010254, + "step": 418 + }, + { + "epoch": 0.23440559440559441, + "grad_norm": 1.8065313740156388, + "learning_rate": 1.5567970204841713e-05, + "loss": 0.1682322919368744, + "step": 419 + }, + { + "epoch": 0.23496503496503496, + "grad_norm": 0.6233677174602369, + "learning_rate": 1.560521415270019e-05, + "loss": 0.20202751457691193, + "step": 420 + }, + { + "epoch": 0.23552447552447553, + "grad_norm": 0.566214426151676, + "learning_rate": 1.564245810055866e-05, + "loss": 0.1698126345872879, + "step": 421 + }, + { + "epoch": 0.23608391608391607, + "grad_norm": 0.5378919951025284, + "learning_rate": 1.5679702048417132e-05, + "loss": 0.22727744281291962, + "step": 422 + }, + { + "epoch": 0.23664335664335664, + "grad_norm": 0.4771062572089051, + "learning_rate": 1.5716945996275604e-05, + "loss": 0.20175057649612427, + "step": 423 + }, + { + "epoch": 0.23720279720279722, + "grad_norm": 0.526243659600649, + "learning_rate": 1.575418994413408e-05, + "loss": 0.1681423783302307, + "step": 424 + }, + { + "epoch": 0.23776223776223776, + "grad_norm": 0.776312463037181, + "learning_rate": 1.5791433891992554e-05, + "loss": 0.36445632576942444, + "step": 425 + }, + { + "epoch": 0.23832167832167833, + "grad_norm": 0.6792696636583263, + "learning_rate": 1.5828677839851026e-05, + "loss": 0.26764899492263794, + "step": 426 + }, + { + "epoch": 0.23888111888111888, + "grad_norm": 0.7275042726945699, + "learning_rate": 1.5865921787709498e-05, + "loss": 0.1031881645321846, + "step": 427 + }, + { + "epoch": 0.23944055944055945, + "grad_norm": 0.27181675532980026, + "learning_rate": 1.590316573556797e-05, + "loss": 0.0615655779838562, + "step": 428 + }, + { + "epoch": 0.24, + "grad_norm": 0.6079960346073131, + "learning_rate": 1.5940409683426445e-05, + "loss": 0.1426311582326889, + "step": 429 + }, + { + "epoch": 0.24055944055944056, + "grad_norm": 41.89293191792446, + "learning_rate": 1.5977653631284917e-05, + "loss": 0.5700697898864746, + "step": 430 + }, + { + "epoch": 0.2411188811188811, + "grad_norm": 0.9264642813398324, + "learning_rate": 1.6014897579143392e-05, + "loss": 0.19980737566947937, + "step": 431 + }, + { + "epoch": 0.24167832167832168, + "grad_norm": 0.6151522058908536, + "learning_rate": 1.6052141527001864e-05, + "loss": 0.22451502084732056, + "step": 432 + }, + { + "epoch": 0.24223776223776225, + "grad_norm": 0.6635939784302595, + "learning_rate": 1.6089385474860336e-05, + "loss": 0.33274221420288086, + "step": 433 + }, + { + "epoch": 0.2427972027972028, + "grad_norm": 0.7888676273825777, + "learning_rate": 1.6126629422718808e-05, + "loss": 0.19224220514297485, + "step": 434 + }, + { + "epoch": 0.24335664335664337, + "grad_norm": 0.47275065485095646, + "learning_rate": 1.6163873370577283e-05, + "loss": 0.14506609737873077, + "step": 435 + }, + { + "epoch": 0.2439160839160839, + "grad_norm": 0.40307357313494613, + "learning_rate": 1.6201117318435755e-05, + "loss": 0.049820952117443085, + "step": 436 + }, + { + "epoch": 0.24447552447552448, + "grad_norm": 0.6574586441533499, + "learning_rate": 1.623836126629423e-05, + "loss": 0.19768625497817993, + "step": 437 + }, + { + "epoch": 0.24503496503496502, + "grad_norm": 0.605836282112127, + "learning_rate": 1.62756052141527e-05, + "loss": 0.23779600858688354, + "step": 438 + }, + { + "epoch": 0.2455944055944056, + "grad_norm": 1.085776164220034, + "learning_rate": 1.6312849162011177e-05, + "loss": 0.4842148423194885, + "step": 439 + }, + { + "epoch": 0.24615384615384617, + "grad_norm": 0.4674846929300058, + "learning_rate": 1.635009310986965e-05, + "loss": 0.11237426102161407, + "step": 440 + }, + { + "epoch": 0.2467132867132867, + "grad_norm": 0.4995866935973659, + "learning_rate": 1.638733705772812e-05, + "loss": 0.14319780468940735, + "step": 441 + }, + { + "epoch": 0.24727272727272728, + "grad_norm": 0.8199876065669118, + "learning_rate": 1.6424581005586592e-05, + "loss": 0.35914701223373413, + "step": 442 + }, + { + "epoch": 0.24783216783216783, + "grad_norm": 0.40863672473508844, + "learning_rate": 1.6461824953445067e-05, + "loss": 0.12121884524822235, + "step": 443 + }, + { + "epoch": 0.2483916083916084, + "grad_norm": 1.0191890496398024, + "learning_rate": 1.649906890130354e-05, + "loss": 0.34015312790870667, + "step": 444 + }, + { + "epoch": 0.24895104895104894, + "grad_norm": 0.6469761868081476, + "learning_rate": 1.6536312849162014e-05, + "loss": 0.2301608771085739, + "step": 445 + }, + { + "epoch": 0.2495104895104895, + "grad_norm": 0.6597950166422247, + "learning_rate": 1.6573556797020486e-05, + "loss": 0.3519575893878937, + "step": 446 + }, + { + "epoch": 0.25006993006993006, + "grad_norm": 0.4176194496482552, + "learning_rate": 1.6610800744878958e-05, + "loss": 0.08999969810247421, + "step": 447 + }, + { + "epoch": 0.25062937062937063, + "grad_norm": 0.7351345010852466, + "learning_rate": 1.664804469273743e-05, + "loss": 0.28024739027023315, + "step": 448 + }, + { + "epoch": 0.2511888111888112, + "grad_norm": 0.7456672034944458, + "learning_rate": 1.6685288640595905e-05, + "loss": 0.43207600712776184, + "step": 449 + }, + { + "epoch": 0.2517482517482518, + "grad_norm": 0.6322157012614065, + "learning_rate": 1.6722532588454377e-05, + "loss": 0.2294520139694214, + "step": 450 + }, + { + "epoch": 0.2523076923076923, + "grad_norm": 0.6114315634156927, + "learning_rate": 1.6759776536312852e-05, + "loss": 0.1383638232946396, + "step": 451 + }, + { + "epoch": 0.25286713286713286, + "grad_norm": 0.5528728499840396, + "learning_rate": 1.6797020484171324e-05, + "loss": 0.25563934445381165, + "step": 452 + }, + { + "epoch": 0.25342657342657343, + "grad_norm": 0.5659791655464828, + "learning_rate": 1.6834264432029796e-05, + "loss": 0.21258139610290527, + "step": 453 + }, + { + "epoch": 0.253986013986014, + "grad_norm": 0.6540833072161577, + "learning_rate": 1.687150837988827e-05, + "loss": 0.24583888053894043, + "step": 454 + }, + { + "epoch": 0.2545454545454545, + "grad_norm": 0.8746136685972271, + "learning_rate": 1.6908752327746743e-05, + "loss": 0.49477094411849976, + "step": 455 + }, + { + "epoch": 0.2551048951048951, + "grad_norm": 0.727924344577947, + "learning_rate": 1.6945996275605218e-05, + "loss": 0.19425439834594727, + "step": 456 + }, + { + "epoch": 0.25566433566433566, + "grad_norm": 0.5336764280569197, + "learning_rate": 1.698324022346369e-05, + "loss": 0.1849464774131775, + "step": 457 + }, + { + "epoch": 0.25622377622377623, + "grad_norm": 0.5276431779414066, + "learning_rate": 1.702048417132216e-05, + "loss": 0.2405935525894165, + "step": 458 + }, + { + "epoch": 0.2567832167832168, + "grad_norm": 0.5653302308897664, + "learning_rate": 1.7057728119180633e-05, + "loss": 0.21965207159519196, + "step": 459 + }, + { + "epoch": 0.2573426573426573, + "grad_norm": 0.6152373551084098, + "learning_rate": 1.709497206703911e-05, + "loss": 0.25860822200775146, + "step": 460 + }, + { + "epoch": 0.2579020979020979, + "grad_norm": 0.4172440645079593, + "learning_rate": 1.713221601489758e-05, + "loss": 0.08967152237892151, + "step": 461 + }, + { + "epoch": 0.25846153846153846, + "grad_norm": 0.6877316368593945, + "learning_rate": 1.7169459962756056e-05, + "loss": 0.21769946813583374, + "step": 462 + }, + { + "epoch": 0.25902097902097904, + "grad_norm": 0.4850616473572567, + "learning_rate": 1.7206703910614527e-05, + "loss": 0.1710788607597351, + "step": 463 + }, + { + "epoch": 0.2595804195804196, + "grad_norm": 0.7839504810754372, + "learning_rate": 1.7243947858473e-05, + "loss": 0.3484511077404022, + "step": 464 + }, + { + "epoch": 0.2601398601398601, + "grad_norm": 0.5375158253656618, + "learning_rate": 1.728119180633147e-05, + "loss": 0.2597725987434387, + "step": 465 + }, + { + "epoch": 0.2606993006993007, + "grad_norm": 0.4955328453110693, + "learning_rate": 1.7318435754189946e-05, + "loss": 0.2551834285259247, + "step": 466 + }, + { + "epoch": 0.26125874125874127, + "grad_norm": 0.44875277352740534, + "learning_rate": 1.7355679702048418e-05, + "loss": 0.0673862025141716, + "step": 467 + }, + { + "epoch": 0.26181818181818184, + "grad_norm": 0.4417890075453276, + "learning_rate": 1.7392923649906893e-05, + "loss": 0.12457968294620514, + "step": 468 + }, + { + "epoch": 0.26237762237762235, + "grad_norm": 0.6355724689625216, + "learning_rate": 1.7430167597765365e-05, + "loss": 0.20733694732189178, + "step": 469 + }, + { + "epoch": 0.2629370629370629, + "grad_norm": 0.611966868034867, + "learning_rate": 1.7467411545623837e-05, + "loss": 0.2562675178050995, + "step": 470 + }, + { + "epoch": 0.2634965034965035, + "grad_norm": 0.7956913912431819, + "learning_rate": 1.750465549348231e-05, + "loss": 0.39981701970100403, + "step": 471 + }, + { + "epoch": 0.26405594405594407, + "grad_norm": 0.5753312350255094, + "learning_rate": 1.7541899441340784e-05, + "loss": 0.1550302356481552, + "step": 472 + }, + { + "epoch": 0.26461538461538464, + "grad_norm": 0.6095764243603401, + "learning_rate": 1.7579143389199256e-05, + "loss": 0.3247944712638855, + "step": 473 + }, + { + "epoch": 0.26517482517482516, + "grad_norm": 0.49705512838475074, + "learning_rate": 1.761638733705773e-05, + "loss": 0.14899270236492157, + "step": 474 + }, + { + "epoch": 0.26573426573426573, + "grad_norm": 0.41914814874098993, + "learning_rate": 1.7653631284916203e-05, + "loss": 0.10161678493022919, + "step": 475 + }, + { + "epoch": 0.2662937062937063, + "grad_norm": 0.7743766363107085, + "learning_rate": 1.7690875232774675e-05, + "loss": 0.21037091314792633, + "step": 476 + }, + { + "epoch": 0.26685314685314687, + "grad_norm": 0.43632575816667735, + "learning_rate": 1.7728119180633146e-05, + "loss": 0.12281668931245804, + "step": 477 + }, + { + "epoch": 0.2674125874125874, + "grad_norm": 0.7288749778106602, + "learning_rate": 1.776536312849162e-05, + "loss": 0.3243931531906128, + "step": 478 + }, + { + "epoch": 0.26797202797202796, + "grad_norm": 0.3055253946877526, + "learning_rate": 1.7802607076350093e-05, + "loss": 0.07544611394405365, + "step": 479 + }, + { + "epoch": 0.26853146853146853, + "grad_norm": 0.6481252003297353, + "learning_rate": 1.783985102420857e-05, + "loss": 0.13213147222995758, + "step": 480 + }, + { + "epoch": 0.2690909090909091, + "grad_norm": 0.44763680375702813, + "learning_rate": 1.787709497206704e-05, + "loss": 0.14075420796871185, + "step": 481 + }, + { + "epoch": 0.2696503496503497, + "grad_norm": 1.2826085609235796, + "learning_rate": 1.7914338919925512e-05, + "loss": 0.3751899003982544, + "step": 482 + }, + { + "epoch": 0.2702097902097902, + "grad_norm": 0.585768182357368, + "learning_rate": 1.7951582867783987e-05, + "loss": 0.21884986758232117, + "step": 483 + }, + { + "epoch": 0.27076923076923076, + "grad_norm": 0.7603779317757269, + "learning_rate": 1.798882681564246e-05, + "loss": 0.4682178497314453, + "step": 484 + }, + { + "epoch": 0.27132867132867133, + "grad_norm": 0.6592506507534223, + "learning_rate": 1.8026070763500934e-05, + "loss": 0.2147725522518158, + "step": 485 + }, + { + "epoch": 0.2718881118881119, + "grad_norm": 0.38307263282617676, + "learning_rate": 1.8063314711359406e-05, + "loss": 0.10786425322294235, + "step": 486 + }, + { + "epoch": 0.2724475524475524, + "grad_norm": 0.7147883048443288, + "learning_rate": 1.8100558659217878e-05, + "loss": 0.23253406584262848, + "step": 487 + }, + { + "epoch": 0.273006993006993, + "grad_norm": 0.6447664161631911, + "learning_rate": 1.813780260707635e-05, + "loss": 0.2840860188007355, + "step": 488 + }, + { + "epoch": 0.27356643356643356, + "grad_norm": 0.6184937002120896, + "learning_rate": 1.8175046554934825e-05, + "loss": 0.20700715482234955, + "step": 489 + }, + { + "epoch": 0.27412587412587414, + "grad_norm": 0.5593968872267058, + "learning_rate": 1.8212290502793297e-05, + "loss": 0.2722780406475067, + "step": 490 + }, + { + "epoch": 0.2746853146853147, + "grad_norm": 0.5528668175545255, + "learning_rate": 1.8249534450651772e-05, + "loss": 0.26255929470062256, + "step": 491 + }, + { + "epoch": 0.2752447552447552, + "grad_norm": 0.6770594227758875, + "learning_rate": 1.8286778398510244e-05, + "loss": 0.3739148676395416, + "step": 492 + }, + { + "epoch": 0.2758041958041958, + "grad_norm": 0.681337399231226, + "learning_rate": 1.8324022346368716e-05, + "loss": 0.3016756474971771, + "step": 493 + }, + { + "epoch": 0.27636363636363637, + "grad_norm": 0.7182715627002348, + "learning_rate": 1.8361266294227188e-05, + "loss": 0.3016898036003113, + "step": 494 + }, + { + "epoch": 0.27692307692307694, + "grad_norm": 0.3022978003036269, + "learning_rate": 1.8398510242085663e-05, + "loss": 0.036744993180036545, + "step": 495 + }, + { + "epoch": 0.2774825174825175, + "grad_norm": 0.6518551231800501, + "learning_rate": 1.8435754189944135e-05, + "loss": 0.20957455039024353, + "step": 496 + }, + { + "epoch": 0.278041958041958, + "grad_norm": 0.692074614484784, + "learning_rate": 1.847299813780261e-05, + "loss": 0.4254513382911682, + "step": 497 + }, + { + "epoch": 0.2786013986013986, + "grad_norm": 0.20381704655229868, + "learning_rate": 1.851024208566108e-05, + "loss": 0.020748628303408623, + "step": 498 + }, + { + "epoch": 0.27916083916083917, + "grad_norm": 0.6016771095606499, + "learning_rate": 1.8547486033519553e-05, + "loss": 0.36569589376449585, + "step": 499 + }, + { + "epoch": 0.27972027972027974, + "grad_norm": 0.7029184835201229, + "learning_rate": 1.8584729981378025e-05, + "loss": 0.3316744863986969, + "step": 500 + }, + { + "epoch": 0.28027972027972026, + "grad_norm": 0.6939501420361155, + "learning_rate": 1.86219739292365e-05, + "loss": 0.23616641759872437, + "step": 501 + }, + { + "epoch": 0.28083916083916083, + "grad_norm": 0.3096678893982763, + "learning_rate": 1.8659217877094972e-05, + "loss": 0.03917497396469116, + "step": 502 + }, + { + "epoch": 0.2813986013986014, + "grad_norm": 0.5750893064505095, + "learning_rate": 1.8696461824953447e-05, + "loss": 0.22358393669128418, + "step": 503 + }, + { + "epoch": 0.28195804195804197, + "grad_norm": 0.7594505249291624, + "learning_rate": 1.873370577281192e-05, + "loss": 0.3953176736831665, + "step": 504 + }, + { + "epoch": 0.28251748251748254, + "grad_norm": 0.5428562647320696, + "learning_rate": 1.8770949720670394e-05, + "loss": 0.26574385166168213, + "step": 505 + }, + { + "epoch": 0.28307692307692306, + "grad_norm": 0.6608673457299766, + "learning_rate": 1.8808193668528866e-05, + "loss": 0.24285589158535004, + "step": 506 + }, + { + "epoch": 0.28363636363636363, + "grad_norm": 0.420800948808082, + "learning_rate": 1.8845437616387338e-05, + "loss": 0.1264358013868332, + "step": 507 + }, + { + "epoch": 0.2841958041958042, + "grad_norm": 0.5687644365002809, + "learning_rate": 1.888268156424581e-05, + "loss": 0.20133408904075623, + "step": 508 + }, + { + "epoch": 0.2847552447552448, + "grad_norm": 0.7043262378346077, + "learning_rate": 1.8919925512104285e-05, + "loss": 0.26330941915512085, + "step": 509 + }, + { + "epoch": 0.2853146853146853, + "grad_norm": 0.567501159254064, + "learning_rate": 1.895716945996276e-05, + "loss": 0.20988181233406067, + "step": 510 + }, + { + "epoch": 0.28587412587412586, + "grad_norm": 0.5896122791395227, + "learning_rate": 1.8994413407821232e-05, + "loss": 0.18932560086250305, + "step": 511 + }, + { + "epoch": 0.28643356643356643, + "grad_norm": 0.6708499627708345, + "learning_rate": 1.9031657355679704e-05, + "loss": 0.2920285165309906, + "step": 512 + }, + { + "epoch": 0.286993006993007, + "grad_norm": 0.5434781998029931, + "learning_rate": 1.9068901303538176e-05, + "loss": 0.20274771749973297, + "step": 513 + }, + { + "epoch": 0.2875524475524476, + "grad_norm": 0.4958234621789348, + "learning_rate": 1.910614525139665e-05, + "loss": 0.11821774393320084, + "step": 514 + }, + { + "epoch": 0.2881118881118881, + "grad_norm": 0.6558416238989833, + "learning_rate": 1.9143389199255123e-05, + "loss": 0.17717942595481873, + "step": 515 + }, + { + "epoch": 0.28867132867132866, + "grad_norm": 0.47282147404221175, + "learning_rate": 1.9180633147113598e-05, + "loss": 0.12990251183509827, + "step": 516 + }, + { + "epoch": 0.28923076923076924, + "grad_norm": 0.3077501281660363, + "learning_rate": 1.921787709497207e-05, + "loss": 0.08428557962179184, + "step": 517 + }, + { + "epoch": 0.2897902097902098, + "grad_norm": 0.5882078711459552, + "learning_rate": 1.925512104283054e-05, + "loss": 0.16661573946475983, + "step": 518 + }, + { + "epoch": 0.2903496503496503, + "grad_norm": 0.37770036258646267, + "learning_rate": 1.9292364990689013e-05, + "loss": 0.07727187126874924, + "step": 519 + }, + { + "epoch": 0.2909090909090909, + "grad_norm": 0.2992926286258636, + "learning_rate": 1.932960893854749e-05, + "loss": 0.08176373690366745, + "step": 520 + }, + { + "epoch": 0.29146853146853147, + "grad_norm": 0.5597464714987342, + "learning_rate": 1.936685288640596e-05, + "loss": 0.29716911911964417, + "step": 521 + }, + { + "epoch": 0.29202797202797204, + "grad_norm": 0.4963646089431364, + "learning_rate": 1.9404096834264436e-05, + "loss": 0.18725723028182983, + "step": 522 + }, + { + "epoch": 0.2925874125874126, + "grad_norm": 0.5710552045050755, + "learning_rate": 1.9441340782122907e-05, + "loss": 0.2251569777727127, + "step": 523 + }, + { + "epoch": 0.2931468531468531, + "grad_norm": 0.4763120762131157, + "learning_rate": 1.947858472998138e-05, + "loss": 0.17864307761192322, + "step": 524 + }, + { + "epoch": 0.2937062937062937, + "grad_norm": 0.558697619106673, + "learning_rate": 1.951582867783985e-05, + "loss": 0.21538743376731873, + "step": 525 + }, + { + "epoch": 0.29426573426573427, + "grad_norm": 0.7065200316468357, + "learning_rate": 1.9553072625698326e-05, + "loss": 0.30963966250419617, + "step": 526 + }, + { + "epoch": 0.29482517482517484, + "grad_norm": 0.6928415993286607, + "learning_rate": 1.9590316573556798e-05, + "loss": 0.3740716278553009, + "step": 527 + }, + { + "epoch": 0.2953846153846154, + "grad_norm": 0.7293190568852718, + "learning_rate": 1.9627560521415273e-05, + "loss": 0.29379579424858093, + "step": 528 + }, + { + "epoch": 0.29594405594405593, + "grad_norm": 0.42409398822019045, + "learning_rate": 1.9664804469273745e-05, + "loss": 0.1428535431623459, + "step": 529 + }, + { + "epoch": 0.2965034965034965, + "grad_norm": 0.3386649913365023, + "learning_rate": 1.9702048417132217e-05, + "loss": 0.09543010592460632, + "step": 530 + }, + { + "epoch": 0.29706293706293707, + "grad_norm": 0.5998899992503403, + "learning_rate": 1.973929236499069e-05, + "loss": 0.2161623239517212, + "step": 531 + }, + { + "epoch": 0.29762237762237764, + "grad_norm": 0.5492553888264003, + "learning_rate": 1.9776536312849164e-05, + "loss": 0.23159924149513245, + "step": 532 + }, + { + "epoch": 0.29818181818181816, + "grad_norm": 0.5080739003555875, + "learning_rate": 1.9813780260707636e-05, + "loss": 0.17287680506706238, + "step": 533 + }, + { + "epoch": 0.29874125874125873, + "grad_norm": 0.6576904849980666, + "learning_rate": 1.985102420856611e-05, + "loss": 0.4449688792228699, + "step": 534 + }, + { + "epoch": 0.2993006993006993, + "grad_norm": 0.7459827209061667, + "learning_rate": 1.9888268156424583e-05, + "loss": 0.2960590422153473, + "step": 535 + }, + { + "epoch": 0.2998601398601399, + "grad_norm": 0.6236773529487676, + "learning_rate": 1.9925512104283055e-05, + "loss": 0.3170793652534485, + "step": 536 + }, + { + "epoch": 0.30041958041958045, + "grad_norm": 0.5112926275461217, + "learning_rate": 1.9962756052141526e-05, + "loss": 0.357006311416626, + "step": 537 + }, + { + "epoch": 0.30097902097902096, + "grad_norm": 0.2588522979236738, + "learning_rate": 2e-05, + "loss": 0.05219828709959984, + "step": 538 + }, + { + "epoch": 0.30153846153846153, + "grad_norm": 0.09955963817138244, + "learning_rate": 1.9999997882052766e-05, + "loss": 0.005741588771343231, + "step": 539 + }, + { + "epoch": 0.3020979020979021, + "grad_norm": 0.5485706991728175, + "learning_rate": 1.999999152821195e-05, + "loss": 0.2549275755882263, + "step": 540 + }, + { + "epoch": 0.3026573426573427, + "grad_norm": 0.5710367006995997, + "learning_rate": 1.9999980938480244e-05, + "loss": 0.28893324732780457, + "step": 541 + }, + { + "epoch": 0.3032167832167832, + "grad_norm": 0.6119601077208713, + "learning_rate": 1.999996611286214e-05, + "loss": 0.3215692639350891, + "step": 542 + }, + { + "epoch": 0.30377622377622376, + "grad_norm": 0.5664171612006691, + "learning_rate": 1.999994705136392e-05, + "loss": 0.21357734501361847, + "step": 543 + }, + { + "epoch": 0.30433566433566434, + "grad_norm": 0.4008329649093053, + "learning_rate": 1.9999923753993643e-05, + "loss": 0.08483588695526123, + "step": 544 + }, + { + "epoch": 0.3048951048951049, + "grad_norm": 0.5697256155189999, + "learning_rate": 1.9999896220761195e-05, + "loss": 0.20176199078559875, + "step": 545 + }, + { + "epoch": 0.3054545454545455, + "grad_norm": 0.5422570391214891, + "learning_rate": 1.999986445167823e-05, + "loss": 0.2352019101381302, + "step": 546 + }, + { + "epoch": 0.306013986013986, + "grad_norm": 0.4210705839051027, + "learning_rate": 1.9999828446758205e-05, + "loss": 0.16775330901145935, + "step": 547 + }, + { + "epoch": 0.30657342657342657, + "grad_norm": 0.44406457932991666, + "learning_rate": 1.9999788206016375e-05, + "loss": 0.21414825320243835, + "step": 548 + }, + { + "epoch": 0.30713286713286714, + "grad_norm": 0.4563610211058553, + "learning_rate": 1.9999743729469784e-05, + "loss": 0.18333928287029266, + "step": 549 + }, + { + "epoch": 0.3076923076923077, + "grad_norm": 0.6553625124257928, + "learning_rate": 1.9999695017137266e-05, + "loss": 0.3981580436229706, + "step": 550 + }, + { + "epoch": 0.3082517482517482, + "grad_norm": 0.646883216245377, + "learning_rate": 1.9999642069039466e-05, + "loss": 0.23577269911766052, + "step": 551 + }, + { + "epoch": 0.3088111888111888, + "grad_norm": 0.6114732438673908, + "learning_rate": 1.9999584885198802e-05, + "loss": 0.2632318437099457, + "step": 552 + }, + { + "epoch": 0.30937062937062937, + "grad_norm": 0.6432116586470514, + "learning_rate": 1.99995234656395e-05, + "loss": 0.2891254425048828, + "step": 553 + }, + { + "epoch": 0.30993006993006994, + "grad_norm": 0.3831579024810142, + "learning_rate": 1.999945781038758e-05, + "loss": 0.08629722893238068, + "step": 554 + }, + { + "epoch": 0.3104895104895105, + "grad_norm": 0.40540314151162465, + "learning_rate": 1.999938791947085e-05, + "loss": 0.07846999913454056, + "step": 555 + }, + { + "epoch": 0.31104895104895103, + "grad_norm": 0.36428502733438844, + "learning_rate": 1.9999313792918918e-05, + "loss": 0.11268086731433868, + "step": 556 + }, + { + "epoch": 0.3116083916083916, + "grad_norm": 0.5505826942738354, + "learning_rate": 1.999923543076318e-05, + "loss": 0.16018052399158478, + "step": 557 + }, + { + "epoch": 0.31216783216783217, + "grad_norm": 0.43853893649388564, + "learning_rate": 1.999915283303683e-05, + "loss": 0.08534029126167297, + "step": 558 + }, + { + "epoch": 0.31272727272727274, + "grad_norm": 0.5932480302976744, + "learning_rate": 1.9999065999774855e-05, + "loss": 0.2227429896593094, + "step": 559 + }, + { + "epoch": 0.3132867132867133, + "grad_norm": 0.32041143251940857, + "learning_rate": 1.9998974931014036e-05, + "loss": 0.07833819836378098, + "step": 560 + }, + { + "epoch": 0.31384615384615383, + "grad_norm": 0.6510010504132044, + "learning_rate": 1.999887962679295e-05, + "loss": 0.21360847353935242, + "step": 561 + }, + { + "epoch": 0.3144055944055944, + "grad_norm": 0.6073419035703598, + "learning_rate": 1.999878008715197e-05, + "loss": 0.26231253147125244, + "step": 562 + }, + { + "epoch": 0.314965034965035, + "grad_norm": 0.5008766222197276, + "learning_rate": 1.9998676312133254e-05, + "loss": 0.18054431676864624, + "step": 563 + }, + { + "epoch": 0.31552447552447555, + "grad_norm": 0.5313966374422028, + "learning_rate": 1.9998568301780764e-05, + "loss": 0.17667227983474731, + "step": 564 + }, + { + "epoch": 0.31608391608391606, + "grad_norm": 0.5260101386623623, + "learning_rate": 1.999845605614025e-05, + "loss": 0.09886716306209564, + "step": 565 + }, + { + "epoch": 0.31664335664335663, + "grad_norm": 0.5700829684459591, + "learning_rate": 1.999833957525926e-05, + "loss": 0.0950867235660553, + "step": 566 + }, + { + "epoch": 0.3172027972027972, + "grad_norm": 0.4673359300650603, + "learning_rate": 1.9998218859187132e-05, + "loss": 0.10726246982812881, + "step": 567 + }, + { + "epoch": 0.3177622377622378, + "grad_norm": 0.585155182256199, + "learning_rate": 1.9998093907975e-05, + "loss": 0.22885611653327942, + "step": 568 + }, + { + "epoch": 0.31832167832167835, + "grad_norm": 0.47707338934606974, + "learning_rate": 1.9997964721675797e-05, + "loss": 0.16578838229179382, + "step": 569 + }, + { + "epoch": 0.31888111888111886, + "grad_norm": 0.6646568787855933, + "learning_rate": 1.999783130034424e-05, + "loss": 0.25660938024520874, + "step": 570 + }, + { + "epoch": 0.31944055944055944, + "grad_norm": 0.5293395568069118, + "learning_rate": 1.9997693644036844e-05, + "loss": 0.20382758975028992, + "step": 571 + }, + { + "epoch": 0.32, + "grad_norm": 0.4048621765202779, + "learning_rate": 1.9997551752811922e-05, + "loss": 0.15027384459972382, + "step": 572 + }, + { + "epoch": 0.3205594405594406, + "grad_norm": 0.5944357025523712, + "learning_rate": 1.9997405626729577e-05, + "loss": 0.30086737871170044, + "step": 573 + }, + { + "epoch": 0.3211188811188811, + "grad_norm": 0.5724302320203283, + "learning_rate": 1.9997255265851705e-05, + "loss": 0.22597002983093262, + "step": 574 + }, + { + "epoch": 0.32167832167832167, + "grad_norm": 0.6473390477312067, + "learning_rate": 1.9997100670241994e-05, + "loss": 0.3176121115684509, + "step": 575 + }, + { + "epoch": 0.32223776223776224, + "grad_norm": 0.5568416614176801, + "learning_rate": 1.999694183996594e-05, + "loss": 0.20588722825050354, + "step": 576 + }, + { + "epoch": 0.3227972027972028, + "grad_norm": 0.5291519265603902, + "learning_rate": 1.999677877509081e-05, + "loss": 0.23704226315021515, + "step": 577 + }, + { + "epoch": 0.3233566433566434, + "grad_norm": 0.7866952189870885, + "learning_rate": 1.9996611475685685e-05, + "loss": 0.3305433392524719, + "step": 578 + }, + { + "epoch": 0.3239160839160839, + "grad_norm": 0.6395620712693595, + "learning_rate": 1.9996439941821427e-05, + "loss": 0.2543655037879944, + "step": 579 + }, + { + "epoch": 0.32447552447552447, + "grad_norm": 0.4141940474174266, + "learning_rate": 1.9996264173570697e-05, + "loss": 0.1492166519165039, + "step": 580 + }, + { + "epoch": 0.32503496503496504, + "grad_norm": 0.5107851325719412, + "learning_rate": 1.999608417100795e-05, + "loss": 0.2557552754878998, + "step": 581 + }, + { + "epoch": 0.3255944055944056, + "grad_norm": 0.6849110791058359, + "learning_rate": 1.9995899934209422e-05, + "loss": 0.29245346784591675, + "step": 582 + }, + { + "epoch": 0.3261538461538461, + "grad_norm": 0.5554705780747584, + "learning_rate": 1.9995711463253174e-05, + "loss": 0.2192707657814026, + "step": 583 + }, + { + "epoch": 0.3267132867132867, + "grad_norm": 0.6357616146251244, + "learning_rate": 1.9995518758219027e-05, + "loss": 0.34399017691612244, + "step": 584 + }, + { + "epoch": 0.32727272727272727, + "grad_norm": 0.6609144531210807, + "learning_rate": 1.9995321819188608e-05, + "loss": 0.23436295986175537, + "step": 585 + }, + { + "epoch": 0.32783216783216784, + "grad_norm": 0.5005412797974282, + "learning_rate": 1.9995120646245345e-05, + "loss": 0.19554248452186584, + "step": 586 + }, + { + "epoch": 0.3283916083916084, + "grad_norm": 0.37522320734685793, + "learning_rate": 1.9994915239474447e-05, + "loss": 0.10892355442047119, + "step": 587 + }, + { + "epoch": 0.32895104895104893, + "grad_norm": 0.4602752860494205, + "learning_rate": 1.9994705598962927e-05, + "loss": 0.022315168753266335, + "step": 588 + }, + { + "epoch": 0.3295104895104895, + "grad_norm": 0.698452982941737, + "learning_rate": 1.9994491724799583e-05, + "loss": 0.212662011384964, + "step": 589 + }, + { + "epoch": 0.3300699300699301, + "grad_norm": 0.6618516131465417, + "learning_rate": 1.9994273617075007e-05, + "loss": 0.23674353957176208, + "step": 590 + }, + { + "epoch": 0.33062937062937064, + "grad_norm": 0.6374675468701437, + "learning_rate": 1.9994051275881597e-05, + "loss": 0.23225592076778412, + "step": 591 + }, + { + "epoch": 0.3311888111888112, + "grad_norm": 0.5255129970539337, + "learning_rate": 1.9993824701313523e-05, + "loss": 0.21238228678703308, + "step": 592 + }, + { + "epoch": 0.33174825174825173, + "grad_norm": 0.5244237443067354, + "learning_rate": 1.999359389346677e-05, + "loss": 0.20713210105895996, + "step": 593 + }, + { + "epoch": 0.3323076923076923, + "grad_norm": 0.6752276252614303, + "learning_rate": 1.99933588524391e-05, + "loss": 0.23903657495975494, + "step": 594 + }, + { + "epoch": 0.3328671328671329, + "grad_norm": 0.599665204549331, + "learning_rate": 1.9993119578330073e-05, + "loss": 0.33758509159088135, + "step": 595 + }, + { + "epoch": 0.33342657342657345, + "grad_norm": 0.5367396710049526, + "learning_rate": 1.999287607124105e-05, + "loss": 0.2791229784488678, + "step": 596 + }, + { + "epoch": 0.33398601398601396, + "grad_norm": 0.49491096875128937, + "learning_rate": 1.9992628331275164e-05, + "loss": 0.22226542234420776, + "step": 597 + }, + { + "epoch": 0.33454545454545453, + "grad_norm": 0.6268942799224394, + "learning_rate": 1.9992376358537367e-05, + "loss": 0.28454047441482544, + "step": 598 + }, + { + "epoch": 0.3351048951048951, + "grad_norm": 0.37198578287673423, + "learning_rate": 1.999212015313439e-05, + "loss": 0.10577382892370224, + "step": 599 + }, + { + "epoch": 0.3356643356643357, + "grad_norm": 0.5323492507915036, + "learning_rate": 1.9991859715174758e-05, + "loss": 0.27060481905937195, + "step": 600 + }, + { + "epoch": 0.33622377622377625, + "grad_norm": 0.46800069136351324, + "learning_rate": 1.9991595044768788e-05, + "loss": 0.13188296556472778, + "step": 601 + }, + { + "epoch": 0.33678321678321677, + "grad_norm": 0.5610023902162545, + "learning_rate": 1.9991326142028595e-05, + "loss": 0.18403789401054382, + "step": 602 + }, + { + "epoch": 0.33734265734265734, + "grad_norm": 0.39117750631530884, + "learning_rate": 1.9991053007068077e-05, + "loss": 0.11701376736164093, + "step": 603 + }, + { + "epoch": 0.3379020979020979, + "grad_norm": 0.41972442009994665, + "learning_rate": 1.999077564000294e-05, + "loss": 0.16472938656806946, + "step": 604 + }, + { + "epoch": 0.3384615384615385, + "grad_norm": 0.5227779824123012, + "learning_rate": 1.9990494040950666e-05, + "loss": 0.24946929514408112, + "step": 605 + }, + { + "epoch": 0.339020979020979, + "grad_norm": 0.16511984982470232, + "learning_rate": 1.9990208210030543e-05, + "loss": 0.01795024238526821, + "step": 606 + }, + { + "epoch": 0.33958041958041957, + "grad_norm": 0.35113925212067854, + "learning_rate": 1.9989918147363644e-05, + "loss": 0.058757804334163666, + "step": 607 + }, + { + "epoch": 0.34013986013986014, + "grad_norm": 0.7389416743824018, + "learning_rate": 1.9989623853072833e-05, + "loss": 0.40756332874298096, + "step": 608 + }, + { + "epoch": 0.3406993006993007, + "grad_norm": 0.5236891416028604, + "learning_rate": 1.9989325327282777e-05, + "loss": 0.2548201382160187, + "step": 609 + }, + { + "epoch": 0.3412587412587413, + "grad_norm": 0.7544125300209171, + "learning_rate": 1.998902257011992e-05, + "loss": 0.2671986520290375, + "step": 610 + }, + { + "epoch": 0.3418181818181818, + "grad_norm": 0.8138720258819945, + "learning_rate": 1.9988715581712516e-05, + "loss": 0.5934276580810547, + "step": 611 + }, + { + "epoch": 0.34237762237762237, + "grad_norm": 0.48447131096028223, + "learning_rate": 1.998840436219059e-05, + "loss": 0.17480790615081787, + "step": 612 + }, + { + "epoch": 0.34293706293706294, + "grad_norm": 0.34966616403070105, + "learning_rate": 1.9988088911685986e-05, + "loss": 0.09544789046049118, + "step": 613 + }, + { + "epoch": 0.3434965034965035, + "grad_norm": 0.7874092173798246, + "learning_rate": 1.998776923033231e-05, + "loss": 0.32014167308807373, + "step": 614 + }, + { + "epoch": 0.34405594405594403, + "grad_norm": 0.4827978865754322, + "learning_rate": 1.9987445318264994e-05, + "loss": 0.16702398657798767, + "step": 615 + }, + { + "epoch": 0.3446153846153846, + "grad_norm": 4.291875642528854, + "learning_rate": 1.9987117175621223e-05, + "loss": 0.4500359892845154, + "step": 616 + }, + { + "epoch": 0.3451748251748252, + "grad_norm": 0.45672563439395375, + "learning_rate": 1.9986784802540014e-05, + "loss": 0.18182170391082764, + "step": 617 + }, + { + "epoch": 0.34573426573426574, + "grad_norm": 0.4480369555149996, + "learning_rate": 1.9986448199162142e-05, + "loss": 0.2514946758747101, + "step": 618 + }, + { + "epoch": 0.3462937062937063, + "grad_norm": 0.5753866865351692, + "learning_rate": 1.99861073656302e-05, + "loss": 0.21140369772911072, + "step": 619 + }, + { + "epoch": 0.34685314685314683, + "grad_norm": 0.5676228469419385, + "learning_rate": 1.9985762302088548e-05, + "loss": 0.22062942385673523, + "step": 620 + }, + { + "epoch": 0.3474125874125874, + "grad_norm": 0.5137054571960226, + "learning_rate": 1.9985413008683364e-05, + "loss": 0.18166667222976685, + "step": 621 + }, + { + "epoch": 0.347972027972028, + "grad_norm": 0.6073448272319864, + "learning_rate": 1.99850594855626e-05, + "loss": 0.21900129318237305, + "step": 622 + }, + { + "epoch": 0.34853146853146855, + "grad_norm": 0.14415284736002523, + "learning_rate": 1.9984701732876008e-05, + "loss": 0.008729547262191772, + "step": 623 + }, + { + "epoch": 0.3490909090909091, + "grad_norm": 0.655384999149106, + "learning_rate": 1.9984339750775118e-05, + "loss": 0.19723348319530487, + "step": 624 + }, + { + "epoch": 0.34965034965034963, + "grad_norm": 0.3927389643917055, + "learning_rate": 1.9983973539413274e-05, + "loss": 0.14206211268901825, + "step": 625 + }, + { + "epoch": 0.3502097902097902, + "grad_norm": 0.3756041596070185, + "learning_rate": 1.9983603098945596e-05, + "loss": 0.11825959384441376, + "step": 626 + }, + { + "epoch": 0.3507692307692308, + "grad_norm": 0.47436842958627345, + "learning_rate": 1.998322842952899e-05, + "loss": 0.15087342262268066, + "step": 627 + }, + { + "epoch": 0.35132867132867135, + "grad_norm": 0.42409311898000623, + "learning_rate": 1.9982849531322174e-05, + "loss": 0.12859410047531128, + "step": 628 + }, + { + "epoch": 0.35188811188811187, + "grad_norm": 0.5259871737926831, + "learning_rate": 1.998246640448564e-05, + "loss": 0.2304256111383438, + "step": 629 + }, + { + "epoch": 0.35244755244755244, + "grad_norm": 0.5004586353979921, + "learning_rate": 1.9982079049181676e-05, + "loss": 0.1558963656425476, + "step": 630 + }, + { + "epoch": 0.353006993006993, + "grad_norm": 0.6160194022966561, + "learning_rate": 1.9981687465574363e-05, + "loss": 0.24547037482261658, + "step": 631 + }, + { + "epoch": 0.3535664335664336, + "grad_norm": 0.40251846067034, + "learning_rate": 1.998129165382957e-05, + "loss": 0.0876728892326355, + "step": 632 + }, + { + "epoch": 0.35412587412587415, + "grad_norm": 0.32391983810499114, + "learning_rate": 1.9980891614114956e-05, + "loss": 0.09364181756973267, + "step": 633 + }, + { + "epoch": 0.35468531468531467, + "grad_norm": 0.4265858227059463, + "learning_rate": 1.998048734659998e-05, + "loss": 0.15264952182769775, + "step": 634 + }, + { + "epoch": 0.35524475524475524, + "grad_norm": 0.2737114096355145, + "learning_rate": 1.9980078851455888e-05, + "loss": 0.024990465492010117, + "step": 635 + }, + { + "epoch": 0.3558041958041958, + "grad_norm": 0.6589929392346666, + "learning_rate": 1.9979666128855705e-05, + "loss": 0.30937856435775757, + "step": 636 + }, + { + "epoch": 0.3563636363636364, + "grad_norm": 0.4880443349385666, + "learning_rate": 1.9979249178974258e-05, + "loss": 0.25523439049720764, + "step": 637 + }, + { + "epoch": 0.3569230769230769, + "grad_norm": 0.5197023459030673, + "learning_rate": 1.9978828001988165e-05, + "loss": 0.2529276907444, + "step": 638 + }, + { + "epoch": 0.35748251748251747, + "grad_norm": 0.47763389355830943, + "learning_rate": 1.9978402598075833e-05, + "loss": 0.12054543197154999, + "step": 639 + }, + { + "epoch": 0.35804195804195804, + "grad_norm": 0.499974171136444, + "learning_rate": 1.997797296741746e-05, + "loss": 0.22388938069343567, + "step": 640 + }, + { + "epoch": 0.3586013986013986, + "grad_norm": 0.4276692814117555, + "learning_rate": 1.9977539110195025e-05, + "loss": 0.13563752174377441, + "step": 641 + }, + { + "epoch": 0.3591608391608392, + "grad_norm": 0.4221449578093158, + "learning_rate": 1.9977101026592313e-05, + "loss": 0.13494934141635895, + "step": 642 + }, + { + "epoch": 0.3597202797202797, + "grad_norm": 0.27722174289693124, + "learning_rate": 1.997665871679489e-05, + "loss": 0.05932402238249779, + "step": 643 + }, + { + "epoch": 0.3602797202797203, + "grad_norm": 0.5682658826635307, + "learning_rate": 1.997621218099011e-05, + "loss": 0.18919488787651062, + "step": 644 + }, + { + "epoch": 0.36083916083916084, + "grad_norm": 0.5462707119631437, + "learning_rate": 1.9975761419367126e-05, + "loss": 0.2577393352985382, + "step": 645 + }, + { + "epoch": 0.3613986013986014, + "grad_norm": 0.6431590074512274, + "learning_rate": 1.997530643211687e-05, + "loss": 0.2787085771560669, + "step": 646 + }, + { + "epoch": 0.36195804195804193, + "grad_norm": 0.6497118821287129, + "learning_rate": 1.997484721943208e-05, + "loss": 0.43514931201934814, + "step": 647 + }, + { + "epoch": 0.3625174825174825, + "grad_norm": 0.6202593301110316, + "learning_rate": 1.9974383781507264e-05, + "loss": 0.2881569266319275, + "step": 648 + }, + { + "epoch": 0.3630769230769231, + "grad_norm": 0.3554244221689389, + "learning_rate": 1.997391611853873e-05, + "loss": 0.11132467538118362, + "step": 649 + }, + { + "epoch": 0.36363636363636365, + "grad_norm": 0.831606329153677, + "learning_rate": 1.997344423072458e-05, + "loss": 0.29948365688323975, + "step": 650 + }, + { + "epoch": 0.3641958041958042, + "grad_norm": 0.681254549895207, + "learning_rate": 1.9972968118264698e-05, + "loss": 0.44342485070228577, + "step": 651 + }, + { + "epoch": 0.36475524475524473, + "grad_norm": 0.8058394586169605, + "learning_rate": 1.9972487781360762e-05, + "loss": 0.3593648076057434, + "step": 652 + }, + { + "epoch": 0.3653146853146853, + "grad_norm": 0.7192797842612415, + "learning_rate": 1.9972003220216233e-05, + "loss": 0.4287704825401306, + "step": 653 + }, + { + "epoch": 0.3658741258741259, + "grad_norm": 0.5830499635776264, + "learning_rate": 1.997151443503637e-05, + "loss": 0.2837563455104828, + "step": 654 + }, + { + "epoch": 0.36643356643356645, + "grad_norm": 0.8497749179803338, + "learning_rate": 1.9971021426028218e-05, + "loss": 0.5405474305152893, + "step": 655 + }, + { + "epoch": 0.366993006993007, + "grad_norm": 0.43607113298681255, + "learning_rate": 1.9970524193400607e-05, + "loss": 0.09462281316518784, + "step": 656 + }, + { + "epoch": 0.36755244755244754, + "grad_norm": 0.527828270527393, + "learning_rate": 1.9970022737364167e-05, + "loss": 0.32398372888565063, + "step": 657 + }, + { + "epoch": 0.3681118881118881, + "grad_norm": 0.5550404441658452, + "learning_rate": 1.9969517058131294e-05, + "loss": 0.1537763923406601, + "step": 658 + }, + { + "epoch": 0.3686713286713287, + "grad_norm": 0.5865224057735395, + "learning_rate": 1.9969007155916205e-05, + "loss": 0.22327053546905518, + "step": 659 + }, + { + "epoch": 0.36923076923076925, + "grad_norm": 0.5503921334716682, + "learning_rate": 1.996849303093488e-05, + "loss": 0.21873745322227478, + "step": 660 + }, + { + "epoch": 0.36979020979020977, + "grad_norm": 0.4433978440880688, + "learning_rate": 1.99679746834051e-05, + "loss": 0.17564524710178375, + "step": 661 + }, + { + "epoch": 0.37034965034965034, + "grad_norm": 0.09685384200788073, + "learning_rate": 1.996745211354643e-05, + "loss": 0.00918425340205431, + "step": 662 + }, + { + "epoch": 0.3709090909090909, + "grad_norm": 0.6120149578807678, + "learning_rate": 1.9966925321580228e-05, + "loss": 0.30388692021369934, + "step": 663 + }, + { + "epoch": 0.3714685314685315, + "grad_norm": 0.5100542710105482, + "learning_rate": 1.9966394307729634e-05, + "loss": 0.20366382598876953, + "step": 664 + }, + { + "epoch": 0.37202797202797205, + "grad_norm": 1.6527809248970362, + "learning_rate": 1.9965859072219583e-05, + "loss": 0.16192631423473358, + "step": 665 + }, + { + "epoch": 0.37258741258741257, + "grad_norm": 0.7317784092811546, + "learning_rate": 1.996531961527679e-05, + "loss": 0.38440436124801636, + "step": 666 + }, + { + "epoch": 0.37314685314685314, + "grad_norm": 0.47828368542810146, + "learning_rate": 1.996477593712977e-05, + "loss": 0.17998406291007996, + "step": 667 + }, + { + "epoch": 0.3737062937062937, + "grad_norm": 0.23675744457115422, + "learning_rate": 1.9964228038008812e-05, + "loss": 0.05032619461417198, + "step": 668 + }, + { + "epoch": 0.3742657342657343, + "grad_norm": 0.7785706670595571, + "learning_rate": 1.996367591814601e-05, + "loss": 0.36880049109458923, + "step": 669 + }, + { + "epoch": 0.3748251748251748, + "grad_norm": 0.4375935776213113, + "learning_rate": 1.9963119577775225e-05, + "loss": 0.23061102628707886, + "step": 670 + }, + { + "epoch": 0.37538461538461537, + "grad_norm": 0.47496055214711447, + "learning_rate": 1.9962559017132124e-05, + "loss": 0.19010961055755615, + "step": 671 + }, + { + "epoch": 0.37594405594405594, + "grad_norm": 0.38270637111218603, + "learning_rate": 1.9961994236454153e-05, + "loss": 0.13563646376132965, + "step": 672 + }, + { + "epoch": 0.3765034965034965, + "grad_norm": 0.8328745126509943, + "learning_rate": 1.9961425235980547e-05, + "loss": 0.23564599454402924, + "step": 673 + }, + { + "epoch": 0.3770629370629371, + "grad_norm": 0.4631518740089339, + "learning_rate": 1.996085201595233e-05, + "loss": 0.1854414939880371, + "step": 674 + }, + { + "epoch": 0.3776223776223776, + "grad_norm": 0.3583143119867986, + "learning_rate": 1.9960274576612307e-05, + "loss": 0.11217008531093597, + "step": 675 + }, + { + "epoch": 0.3781818181818182, + "grad_norm": 0.7253931327544664, + "learning_rate": 1.9959692918205083e-05, + "loss": 0.353920042514801, + "step": 676 + }, + { + "epoch": 0.37874125874125875, + "grad_norm": 0.604362966959615, + "learning_rate": 1.9959107040977037e-05, + "loss": 0.35348647832870483, + "step": 677 + }, + { + "epoch": 0.3793006993006993, + "grad_norm": 0.5669056452344923, + "learning_rate": 1.995851694517634e-05, + "loss": 0.1647416651248932, + "step": 678 + }, + { + "epoch": 0.37986013986013983, + "grad_norm": 0.403240829679156, + "learning_rate": 1.9957922631052954e-05, + "loss": 0.1425113081932068, + "step": 679 + }, + { + "epoch": 0.3804195804195804, + "grad_norm": 0.5693463689714278, + "learning_rate": 1.995732409885862e-05, + "loss": 0.3788657784461975, + "step": 680 + }, + { + "epoch": 0.380979020979021, + "grad_norm": 0.552539216256409, + "learning_rate": 1.9956721348846876e-05, + "loss": 0.14348703622817993, + "step": 681 + }, + { + "epoch": 0.38153846153846155, + "grad_norm": 0.5019643909730159, + "learning_rate": 1.995611438127303e-05, + "loss": 0.18510159850120544, + "step": 682 + }, + { + "epoch": 0.3820979020979021, + "grad_norm": 0.25848464122017484, + "learning_rate": 1.9955503196394198e-05, + "loss": 0.03736811503767967, + "step": 683 + }, + { + "epoch": 0.38265734265734264, + "grad_norm": 0.657567773075943, + "learning_rate": 1.9954887794469266e-05, + "loss": 0.3070823550224304, + "step": 684 + }, + { + "epoch": 0.3832167832167832, + "grad_norm": 0.4509756019837993, + "learning_rate": 1.9954268175758915e-05, + "loss": 0.11352913826704025, + "step": 685 + }, + { + "epoch": 0.3837762237762238, + "grad_norm": 0.5964336248869283, + "learning_rate": 1.9953644340525607e-05, + "loss": 0.12527580559253693, + "step": 686 + }, + { + "epoch": 0.38433566433566435, + "grad_norm": 0.7242495719600796, + "learning_rate": 1.995301628903359e-05, + "loss": 0.4368479549884796, + "step": 687 + }, + { + "epoch": 0.3848951048951049, + "grad_norm": 0.6621858175207591, + "learning_rate": 1.99523840215489e-05, + "loss": 0.32364341616630554, + "step": 688 + }, + { + "epoch": 0.38545454545454544, + "grad_norm": 0.5646559024483815, + "learning_rate": 1.9951747538339365e-05, + "loss": 0.25932592153549194, + "step": 689 + }, + { + "epoch": 0.386013986013986, + "grad_norm": 0.49665798414527007, + "learning_rate": 1.9951106839674585e-05, + "loss": 0.19032754004001617, + "step": 690 + }, + { + "epoch": 0.3865734265734266, + "grad_norm": 0.5007183635798262, + "learning_rate": 1.9950461925825956e-05, + "loss": 0.21685388684272766, + "step": 691 + }, + { + "epoch": 0.38713286713286715, + "grad_norm": 0.5140994426846135, + "learning_rate": 1.9949812797066656e-05, + "loss": 0.19696876406669617, + "step": 692 + }, + { + "epoch": 0.38769230769230767, + "grad_norm": 0.6119888230436024, + "learning_rate": 1.9949159453671653e-05, + "loss": 0.26852601766586304, + "step": 693 + }, + { + "epoch": 0.38825174825174824, + "grad_norm": 0.3387358122930784, + "learning_rate": 1.9948501895917695e-05, + "loss": 0.08862755447626114, + "step": 694 + }, + { + "epoch": 0.3888111888111888, + "grad_norm": 0.6763226743712464, + "learning_rate": 1.9947840124083312e-05, + "loss": 0.020724620670080185, + "step": 695 + }, + { + "epoch": 0.3893706293706294, + "grad_norm": 0.22534541969542238, + "learning_rate": 1.994717413844883e-05, + "loss": 0.03882191702723503, + "step": 696 + }, + { + "epoch": 0.38993006993006996, + "grad_norm": 0.5951297892955677, + "learning_rate": 1.9946503939296348e-05, + "loss": 0.32231131196022034, + "step": 697 + }, + { + "epoch": 0.39048951048951047, + "grad_norm": 0.42905680822844083, + "learning_rate": 1.9945829526909758e-05, + "loss": 0.10666343569755554, + "step": 698 + }, + { + "epoch": 0.39104895104895104, + "grad_norm": 0.45372316643040594, + "learning_rate": 1.994515090157473e-05, + "loss": 0.1684054136276245, + "step": 699 + }, + { + "epoch": 0.3916083916083916, + "grad_norm": 0.42445864130484595, + "learning_rate": 1.994446806357873e-05, + "loss": 0.13398505747318268, + "step": 700 + }, + { + "epoch": 0.3921678321678322, + "grad_norm": 0.4581756205243075, + "learning_rate": 1.9943781013210995e-05, + "loss": 0.3183130621910095, + "step": 701 + }, + { + "epoch": 0.3927272727272727, + "grad_norm": 0.4588990256368521, + "learning_rate": 1.9943089750762556e-05, + "loss": 0.15352673828601837, + "step": 702 + }, + { + "epoch": 0.3932867132867133, + "grad_norm": 0.551633970730141, + "learning_rate": 1.9942394276526222e-05, + "loss": 0.245853990316391, + "step": 703 + }, + { + "epoch": 0.39384615384615385, + "grad_norm": 0.34586336129214834, + "learning_rate": 1.9941694590796586e-05, + "loss": 0.0638028085231781, + "step": 704 + }, + { + "epoch": 0.3944055944055944, + "grad_norm": 0.5097779912600945, + "learning_rate": 1.9940990693870038e-05, + "loss": 0.14579376578330994, + "step": 705 + }, + { + "epoch": 0.394965034965035, + "grad_norm": 0.6453902823504081, + "learning_rate": 1.994028258604473e-05, + "loss": 0.17308682203292847, + "step": 706 + }, + { + "epoch": 0.3955244755244755, + "grad_norm": 0.6528203136801187, + "learning_rate": 1.993957026762061e-05, + "loss": 0.3714531660079956, + "step": 707 + }, + { + "epoch": 0.3960839160839161, + "grad_norm": 0.6707242703082154, + "learning_rate": 1.9938853738899415e-05, + "loss": 0.2819252908229828, + "step": 708 + }, + { + "epoch": 0.39664335664335665, + "grad_norm": 0.508144592867942, + "learning_rate": 1.9938133000184655e-05, + "loss": 0.29829293489456177, + "step": 709 + }, + { + "epoch": 0.3972027972027972, + "grad_norm": 0.3679335752706279, + "learning_rate": 1.993740805178163e-05, + "loss": 0.09194601327180862, + "step": 710 + }, + { + "epoch": 0.39776223776223774, + "grad_norm": 0.6290366367439326, + "learning_rate": 1.9936678893997416e-05, + "loss": 0.4935469627380371, + "step": 711 + }, + { + "epoch": 0.3983216783216783, + "grad_norm": 0.6086198980717892, + "learning_rate": 1.9935945527140878e-05, + "loss": 0.3046504855155945, + "step": 712 + }, + { + "epoch": 0.3988811188811189, + "grad_norm": 0.37361997671932556, + "learning_rate": 1.9935207951522665e-05, + "loss": 0.208089217543602, + "step": 713 + }, + { + "epoch": 0.39944055944055945, + "grad_norm": 0.7063320509029593, + "learning_rate": 1.9934466167455206e-05, + "loss": 0.2735172510147095, + "step": 714 + }, + { + "epoch": 0.4, + "grad_norm": 0.31582418966416786, + "learning_rate": 1.993372017525271e-05, + "loss": 0.07762259989976883, + "step": 715 + }, + { + "epoch": 0.40055944055944054, + "grad_norm": 0.4241187839849317, + "learning_rate": 1.9932969975231173e-05, + "loss": 0.14039283990859985, + "step": 716 + }, + { + "epoch": 0.4011188811188811, + "grad_norm": 0.2710232546386063, + "learning_rate": 1.993221556770837e-05, + "loss": 0.046993188560009, + "step": 717 + }, + { + "epoch": 0.4016783216783217, + "grad_norm": 0.3418106489355633, + "learning_rate": 1.9931456953003864e-05, + "loss": 0.12122257053852081, + "step": 718 + }, + { + "epoch": 0.40223776223776225, + "grad_norm": 0.23441697520225802, + "learning_rate": 1.9930694131438994e-05, + "loss": 0.022285472601652145, + "step": 719 + }, + { + "epoch": 0.4027972027972028, + "grad_norm": 0.5411088359375686, + "learning_rate": 1.9929927103336882e-05, + "loss": 0.24089959263801575, + "step": 720 + }, + { + "epoch": 0.40335664335664334, + "grad_norm": 0.2631132238892778, + "learning_rate": 1.9929155869022437e-05, + "loss": 0.06587815284729004, + "step": 721 + }, + { + "epoch": 0.4039160839160839, + "grad_norm": 0.34990334746142054, + "learning_rate": 1.992838042882234e-05, + "loss": 0.048804886639118195, + "step": 722 + }, + { + "epoch": 0.4044755244755245, + "grad_norm": 0.7656381954667435, + "learning_rate": 1.9927600783065065e-05, + "loss": 0.5738763809204102, + "step": 723 + }, + { + "epoch": 0.40503496503496506, + "grad_norm": 0.44286218414526785, + "learning_rate": 1.9926816932080853e-05, + "loss": 0.29001617431640625, + "step": 724 + }, + { + "epoch": 0.40559440559440557, + "grad_norm": 0.42266075572065687, + "learning_rate": 1.992602887620175e-05, + "loss": 0.1561538130044937, + "step": 725 + }, + { + "epoch": 0.40615384615384614, + "grad_norm": 0.5393564857162861, + "learning_rate": 1.992523661576155e-05, + "loss": 0.24576209485530853, + "step": 726 + }, + { + "epoch": 0.4067132867132867, + "grad_norm": 0.379727330450497, + "learning_rate": 1.9924440151095863e-05, + "loss": 0.1025722399353981, + "step": 727 + }, + { + "epoch": 0.4072727272727273, + "grad_norm": 0.5771833519028271, + "learning_rate": 1.9923639482542053e-05, + "loss": 0.21115200221538544, + "step": 728 + }, + { + "epoch": 0.40783216783216786, + "grad_norm": 0.5142647579527585, + "learning_rate": 1.9922834610439276e-05, + "loss": 0.20292046666145325, + "step": 729 + }, + { + "epoch": 0.4083916083916084, + "grad_norm": 0.5113244464547777, + "learning_rate": 1.9922025535128465e-05, + "loss": 0.2252373993396759, + "step": 730 + }, + { + "epoch": 0.40895104895104895, + "grad_norm": 0.36094979768228547, + "learning_rate": 1.9921212256952345e-05, + "loss": 0.15693120658397675, + "step": 731 + }, + { + "epoch": 0.4095104895104895, + "grad_norm": 0.4645617302744365, + "learning_rate": 1.9920394776255403e-05, + "loss": 0.1647944301366806, + "step": 732 + }, + { + "epoch": 0.4100699300699301, + "grad_norm": 0.5736578559538483, + "learning_rate": 1.9919573093383923e-05, + "loss": 0.26393818855285645, + "step": 733 + }, + { + "epoch": 0.4106293706293706, + "grad_norm": 0.6816576117632496, + "learning_rate": 1.9918747208685948e-05, + "loss": 0.32554638385772705, + "step": 734 + }, + { + "epoch": 0.4111888111888112, + "grad_norm": 0.6078357543501821, + "learning_rate": 1.991791712251133e-05, + "loss": 0.3616198003292084, + "step": 735 + }, + { + "epoch": 0.41174825174825175, + "grad_norm": 0.5559268440280477, + "learning_rate": 1.9917082835211678e-05, + "loss": 0.2198535054922104, + "step": 736 + }, + { + "epoch": 0.4123076923076923, + "grad_norm": 0.5088211547826885, + "learning_rate": 1.991624434714038e-05, + "loss": 0.31540998816490173, + "step": 737 + }, + { + "epoch": 0.4128671328671329, + "grad_norm": 0.5520112254652779, + "learning_rate": 1.9915401658652622e-05, + "loss": 0.2003219723701477, + "step": 738 + }, + { + "epoch": 0.4134265734265734, + "grad_norm": 0.506099926768105, + "learning_rate": 1.9914554770105352e-05, + "loss": 0.19488200545310974, + "step": 739 + }, + { + "epoch": 0.413986013986014, + "grad_norm": 0.3057529839937815, + "learning_rate": 1.9913703681857305e-05, + "loss": 0.09745018184185028, + "step": 740 + }, + { + "epoch": 0.41454545454545455, + "grad_norm": 0.2571773164324853, + "learning_rate": 1.991284839426899e-05, + "loss": 0.059455666691064835, + "step": 741 + }, + { + "epoch": 0.4151048951048951, + "grad_norm": 0.41871968698341877, + "learning_rate": 1.9911988907702704e-05, + "loss": 0.341095507144928, + "step": 742 + }, + { + "epoch": 0.41566433566433564, + "grad_norm": 0.6336316033479839, + "learning_rate": 1.9911125222522513e-05, + "loss": 0.31905606389045715, + "step": 743 + }, + { + "epoch": 0.4162237762237762, + "grad_norm": 0.8205560271997195, + "learning_rate": 1.991025733909426e-05, + "loss": 0.3375856280326843, + "step": 744 + }, + { + "epoch": 0.4167832167832168, + "grad_norm": 0.46065724364425803, + "learning_rate": 1.990938525778558e-05, + "loss": 0.24612030386924744, + "step": 745 + }, + { + "epoch": 0.41734265734265735, + "grad_norm": 0.47567612192363523, + "learning_rate": 1.990850897896587e-05, + "loss": 0.20706714689731598, + "step": 746 + }, + { + "epoch": 0.4179020979020979, + "grad_norm": 0.5916661290227369, + "learning_rate": 1.9907628503006317e-05, + "loss": 0.24323835968971252, + "step": 747 + }, + { + "epoch": 0.41846153846153844, + "grad_norm": 0.529800253205611, + "learning_rate": 1.990674383027988e-05, + "loss": 0.26747220754623413, + "step": 748 + }, + { + "epoch": 0.419020979020979, + "grad_norm": 0.44234336159258025, + "learning_rate": 1.99058549611613e-05, + "loss": 0.24660389125347137, + "step": 749 + }, + { + "epoch": 0.4195804195804196, + "grad_norm": 0.5126829686618091, + "learning_rate": 1.9904961896027085e-05, + "loss": 0.19418266415596008, + "step": 750 + }, + { + "epoch": 0.42013986013986016, + "grad_norm": 0.5698267669133597, + "learning_rate": 1.9904064635255533e-05, + "loss": 0.33401572704315186, + "step": 751 + }, + { + "epoch": 0.4206993006993007, + "grad_norm": 0.5963308712669986, + "learning_rate": 1.9903163179226715e-05, + "loss": 0.3395441770553589, + "step": 752 + }, + { + "epoch": 0.42125874125874124, + "grad_norm": 0.49042687161314796, + "learning_rate": 1.990225752832248e-05, + "loss": 0.37687867879867554, + "step": 753 + }, + { + "epoch": 0.4218181818181818, + "grad_norm": 0.4276891860988093, + "learning_rate": 1.9901347682926447e-05, + "loss": 0.1782049685716629, + "step": 754 + }, + { + "epoch": 0.4223776223776224, + "grad_norm": 0.49921942632013777, + "learning_rate": 1.9900433643424018e-05, + "loss": 0.18066397309303284, + "step": 755 + }, + { + "epoch": 0.42293706293706296, + "grad_norm": 0.5789972509969482, + "learning_rate": 1.9899515410202372e-05, + "loss": 0.33274298906326294, + "step": 756 + }, + { + "epoch": 0.4234965034965035, + "grad_norm": 0.5828897644792967, + "learning_rate": 1.9898592983650462e-05, + "loss": 0.2913612723350525, + "step": 757 + }, + { + "epoch": 0.42405594405594405, + "grad_norm": 0.615935250814808, + "learning_rate": 1.9897666364159018e-05, + "loss": 0.2726729214191437, + "step": 758 + }, + { + "epoch": 0.4246153846153846, + "grad_norm": 0.4553698220108172, + "learning_rate": 1.989673555212055e-05, + "loss": 0.12482152134180069, + "step": 759 + }, + { + "epoch": 0.4251748251748252, + "grad_norm": 0.6287885690345092, + "learning_rate": 1.9895800547929337e-05, + "loss": 0.2711922526359558, + "step": 760 + }, + { + "epoch": 0.42573426573426576, + "grad_norm": 0.7467349801357285, + "learning_rate": 1.9894861351981432e-05, + "loss": 0.36687546968460083, + "step": 761 + }, + { + "epoch": 0.4262937062937063, + "grad_norm": 0.381251208713884, + "learning_rate": 1.989391796467468e-05, + "loss": 0.12306220829486847, + "step": 762 + }, + { + "epoch": 0.42685314685314685, + "grad_norm": 0.47852236250446417, + "learning_rate": 1.989297038640868e-05, + "loss": 0.21053528785705566, + "step": 763 + }, + { + "epoch": 0.4274125874125874, + "grad_norm": 0.5446213742935146, + "learning_rate": 1.989201861758482e-05, + "loss": 0.2965262532234192, + "step": 764 + }, + { + "epoch": 0.427972027972028, + "grad_norm": 0.5106437402104947, + "learning_rate": 1.989106265860626e-05, + "loss": 0.17599058151245117, + "step": 765 + }, + { + "epoch": 0.4285314685314685, + "grad_norm": 0.5310780273921185, + "learning_rate": 1.989010250987793e-05, + "loss": 0.2206752896308899, + "step": 766 + }, + { + "epoch": 0.4290909090909091, + "grad_norm": 0.5653189160688856, + "learning_rate": 1.9889138171806545e-05, + "loss": 0.31108683347702026, + "step": 767 + }, + { + "epoch": 0.42965034965034965, + "grad_norm": 0.7942610990211003, + "learning_rate": 1.9888169644800584e-05, + "loss": 0.4014092683792114, + "step": 768 + }, + { + "epoch": 0.4302097902097902, + "grad_norm": 0.5846377841924643, + "learning_rate": 1.9887196929270306e-05, + "loss": 0.40258675813674927, + "step": 769 + }, + { + "epoch": 0.4307692307692308, + "grad_norm": 0.32791958422439554, + "learning_rate": 1.988622002562774e-05, + "loss": 0.046959392726421356, + "step": 770 + }, + { + "epoch": 0.4313286713286713, + "grad_norm": 0.4464499086430305, + "learning_rate": 1.98852389342867e-05, + "loss": 0.12582331895828247, + "step": 771 + }, + { + "epoch": 0.4318881118881119, + "grad_norm": 0.5191503161190516, + "learning_rate": 1.988425365566276e-05, + "loss": 0.18430662155151367, + "step": 772 + }, + { + "epoch": 0.43244755244755245, + "grad_norm": 0.5486039985154055, + "learning_rate": 1.9883264190173274e-05, + "loss": 0.25388067960739136, + "step": 773 + }, + { + "epoch": 0.433006993006993, + "grad_norm": 0.4652591628258421, + "learning_rate": 1.9882270538237368e-05, + "loss": 0.17330634593963623, + "step": 774 + }, + { + "epoch": 0.43356643356643354, + "grad_norm": 0.4389510047260821, + "learning_rate": 1.9881272700275947e-05, + "loss": 0.11140122264623642, + "step": 775 + }, + { + "epoch": 0.4341258741258741, + "grad_norm": 0.42837244651687484, + "learning_rate": 1.9880270676711682e-05, + "loss": 0.17463473975658417, + "step": 776 + }, + { + "epoch": 0.4346853146853147, + "grad_norm": 0.4162819228876, + "learning_rate": 1.9879264467969016e-05, + "loss": 0.1907040774822235, + "step": 777 + }, + { + "epoch": 0.43524475524475525, + "grad_norm": 0.659965980390764, + "learning_rate": 1.9878254074474176e-05, + "loss": 0.35833752155303955, + "step": 778 + }, + { + "epoch": 0.4358041958041958, + "grad_norm": 0.3900694461739226, + "learning_rate": 1.9877239496655146e-05, + "loss": 0.14887432754039764, + "step": 779 + }, + { + "epoch": 0.43636363636363634, + "grad_norm": 0.5308300003338371, + "learning_rate": 1.98762207349417e-05, + "loss": 0.3146654963493347, + "step": 780 + }, + { + "epoch": 0.4369230769230769, + "grad_norm": 0.5523426738693246, + "learning_rate": 1.9875197789765366e-05, + "loss": 0.06458631157875061, + "step": 781 + }, + { + "epoch": 0.4374825174825175, + "grad_norm": 0.5573847283032164, + "learning_rate": 1.9874170661559456e-05, + "loss": 0.23434588313102722, + "step": 782 + }, + { + "epoch": 0.43804195804195806, + "grad_norm": 0.44944776650208423, + "learning_rate": 1.987313935075905e-05, + "loss": 0.23638221621513367, + "step": 783 + }, + { + "epoch": 0.43860139860139863, + "grad_norm": 0.6323408983727832, + "learning_rate": 1.9872103857801003e-05, + "loss": 0.2840083837509155, + "step": 784 + }, + { + "epoch": 0.43916083916083914, + "grad_norm": 0.4789266540335849, + "learning_rate": 1.987106418312393e-05, + "loss": 0.1863371878862381, + "step": 785 + }, + { + "epoch": 0.4397202797202797, + "grad_norm": 0.4203985325378458, + "learning_rate": 1.987002032716824e-05, + "loss": 0.144439235329628, + "step": 786 + }, + { + "epoch": 0.4402797202797203, + "grad_norm": 0.49878007627335297, + "learning_rate": 1.9868972290376094e-05, + "loss": 0.20231716334819794, + "step": 787 + }, + { + "epoch": 0.44083916083916086, + "grad_norm": 0.5249349372421775, + "learning_rate": 1.9867920073191424e-05, + "loss": 0.32276082038879395, + "step": 788 + }, + { + "epoch": 0.4413986013986014, + "grad_norm": 0.39259353682442255, + "learning_rate": 1.9866863676059943e-05, + "loss": 0.133274644613266, + "step": 789 + }, + { + "epoch": 0.44195804195804195, + "grad_norm": 0.6133007592406627, + "learning_rate": 1.986580309942913e-05, + "loss": 0.28316041827201843, + "step": 790 + }, + { + "epoch": 0.4425174825174825, + "grad_norm": 0.5944072232737287, + "learning_rate": 1.986473834374823e-05, + "loss": 0.35311996936798096, + "step": 791 + }, + { + "epoch": 0.4430769230769231, + "grad_norm": 0.6051688377655196, + "learning_rate": 1.9863669409468262e-05, + "loss": 0.3782188296318054, + "step": 792 + }, + { + "epoch": 0.44363636363636366, + "grad_norm": 0.5216230049719182, + "learning_rate": 1.986259629704202e-05, + "loss": 0.31462928652763367, + "step": 793 + }, + { + "epoch": 0.4441958041958042, + "grad_norm": 0.5378312298235806, + "learning_rate": 1.9861519006924066e-05, + "loss": 0.2784261703491211, + "step": 794 + }, + { + "epoch": 0.44475524475524475, + "grad_norm": 0.5797484173537752, + "learning_rate": 1.986043753957072e-05, + "loss": 0.2810211479663849, + "step": 795 + }, + { + "epoch": 0.4453146853146853, + "grad_norm": 0.8928316481185816, + "learning_rate": 1.9859351895440085e-05, + "loss": 0.49437206983566284, + "step": 796 + }, + { + "epoch": 0.4458741258741259, + "grad_norm": 0.5815193139669039, + "learning_rate": 1.9858262074992034e-05, + "loss": 0.37909960746765137, + "step": 797 + }, + { + "epoch": 0.4464335664335664, + "grad_norm": 0.6988781272956055, + "learning_rate": 1.9857168078688188e-05, + "loss": 0.18024909496307373, + "step": 798 + }, + { + "epoch": 0.446993006993007, + "grad_norm": 0.5968973495525609, + "learning_rate": 1.9856069906991963e-05, + "loss": 0.32987910509109497, + "step": 799 + }, + { + "epoch": 0.44755244755244755, + "grad_norm": 0.3951241354071373, + "learning_rate": 1.9854967560368533e-05, + "loss": 0.22431644797325134, + "step": 800 + }, + { + "epoch": 0.4481118881118881, + "grad_norm": 0.6123555429118513, + "learning_rate": 1.985386103928484e-05, + "loss": 0.27222999930381775, + "step": 801 + }, + { + "epoch": 0.4486713286713287, + "grad_norm": 0.5618687498300933, + "learning_rate": 1.985275034420959e-05, + "loss": 0.161219984292984, + "step": 802 + }, + { + "epoch": 0.4492307692307692, + "grad_norm": 0.49517158279335816, + "learning_rate": 1.9851635475613266e-05, + "loss": 0.22425110638141632, + "step": 803 + }, + { + "epoch": 0.4497902097902098, + "grad_norm": 0.35761448181309125, + "learning_rate": 1.9850516433968114e-05, + "loss": 0.0884673148393631, + "step": 804 + }, + { + "epoch": 0.45034965034965035, + "grad_norm": 0.18514925241411598, + "learning_rate": 1.9849393219748147e-05, + "loss": 0.025272320955991745, + "step": 805 + }, + { + "epoch": 0.4509090909090909, + "grad_norm": 0.5318601845976434, + "learning_rate": 1.984826583342915e-05, + "loss": 0.23998430371284485, + "step": 806 + }, + { + "epoch": 0.45146853146853144, + "grad_norm": 0.6291357990892752, + "learning_rate": 1.9847134275488665e-05, + "loss": 0.3951619863510132, + "step": 807 + }, + { + "epoch": 0.452027972027972, + "grad_norm": 0.5080909307475014, + "learning_rate": 1.9845998546406017e-05, + "loss": 0.3042892515659332, + "step": 808 + }, + { + "epoch": 0.4525874125874126, + "grad_norm": 0.6216362711993472, + "learning_rate": 1.9844858646662285e-05, + "loss": 0.3718934655189514, + "step": 809 + }, + { + "epoch": 0.45314685314685316, + "grad_norm": 0.647738411776936, + "learning_rate": 1.984371457674031e-05, + "loss": 0.25356805324554443, + "step": 810 + }, + { + "epoch": 0.45370629370629373, + "grad_norm": 0.41822632867193543, + "learning_rate": 1.9842566337124723e-05, + "loss": 0.12882554531097412, + "step": 811 + }, + { + "epoch": 0.45426573426573424, + "grad_norm": 0.3857308508875446, + "learning_rate": 1.9841413928301898e-05, + "loss": 0.11873381584882736, + "step": 812 + }, + { + "epoch": 0.4548251748251748, + "grad_norm": 0.7363894189728506, + "learning_rate": 1.9840257350759985e-05, + "loss": 0.45210009813308716, + "step": 813 + }, + { + "epoch": 0.4553846153846154, + "grad_norm": 0.5960275901412995, + "learning_rate": 1.9839096604988892e-05, + "loss": 0.37297913432121277, + "step": 814 + }, + { + "epoch": 0.45594405594405596, + "grad_norm": 0.44764461112704546, + "learning_rate": 1.9837931691480307e-05, + "loss": 0.2697679400444031, + "step": 815 + }, + { + "epoch": 0.45650349650349653, + "grad_norm": 0.2868254358915563, + "learning_rate": 1.9836762610727674e-05, + "loss": 0.06176023930311203, + "step": 816 + }, + { + "epoch": 0.45706293706293705, + "grad_norm": 0.5153217068896744, + "learning_rate": 1.98355893632262e-05, + "loss": 0.19568443298339844, + "step": 817 + }, + { + "epoch": 0.4576223776223776, + "grad_norm": 0.4214432798723545, + "learning_rate": 1.983441194947286e-05, + "loss": 0.11257348209619522, + "step": 818 + }, + { + "epoch": 0.4581818181818182, + "grad_norm": 0.39387927611687096, + "learning_rate": 1.9833230369966397e-05, + "loss": 0.07968845963478088, + "step": 819 + }, + { + "epoch": 0.45874125874125876, + "grad_norm": 1.0126890302003269, + "learning_rate": 1.9832044625207313e-05, + "loss": 0.16019681096076965, + "step": 820 + }, + { + "epoch": 0.4593006993006993, + "grad_norm": 0.4622446813760569, + "learning_rate": 1.9830854715697878e-05, + "loss": 0.20029059052467346, + "step": 821 + }, + { + "epoch": 0.45986013986013985, + "grad_norm": 0.6249287445435043, + "learning_rate": 1.9829660641942125e-05, + "loss": 0.22813692688941956, + "step": 822 + }, + { + "epoch": 0.4604195804195804, + "grad_norm": 0.7030509171614748, + "learning_rate": 1.9828462404445853e-05, + "loss": 0.10862059891223907, + "step": 823 + }, + { + "epoch": 0.460979020979021, + "grad_norm": 0.6899995814640736, + "learning_rate": 1.982726000371662e-05, + "loss": 0.32558006048202515, + "step": 824 + }, + { + "epoch": 0.46153846153846156, + "grad_norm": 0.3810230087763866, + "learning_rate": 1.982605344026375e-05, + "loss": 0.21844086050987244, + "step": 825 + }, + { + "epoch": 0.4620979020979021, + "grad_norm": 0.443596677002616, + "learning_rate": 1.9824842714598335e-05, + "loss": 0.18136177957057953, + "step": 826 + }, + { + "epoch": 0.46265734265734265, + "grad_norm": 0.31533150014796646, + "learning_rate": 1.982362782723322e-05, + "loss": 0.07972593605518341, + "step": 827 + }, + { + "epoch": 0.4632167832167832, + "grad_norm": 0.2765624643844938, + "learning_rate": 1.9822408778683016e-05, + "loss": 0.04100705683231354, + "step": 828 + }, + { + "epoch": 0.4637762237762238, + "grad_norm": 0.39483665747391433, + "learning_rate": 1.982118556946411e-05, + "loss": 0.12353743612766266, + "step": 829 + }, + { + "epoch": 0.4643356643356643, + "grad_norm": 0.517865142641403, + "learning_rate": 1.9819958200094636e-05, + "loss": 0.2682718336582184, + "step": 830 + }, + { + "epoch": 0.4648951048951049, + "grad_norm": 0.49197607992533404, + "learning_rate": 1.9818726671094492e-05, + "loss": 0.2304733246564865, + "step": 831 + }, + { + "epoch": 0.46545454545454545, + "grad_norm": 0.36919880805219946, + "learning_rate": 1.9817490982985337e-05, + "loss": 0.1034315824508667, + "step": 832 + }, + { + "epoch": 0.466013986013986, + "grad_norm": 0.5428095207438932, + "learning_rate": 1.9816251136290608e-05, + "loss": 0.32902252674102783, + "step": 833 + }, + { + "epoch": 0.4665734265734266, + "grad_norm": 0.4442382680248568, + "learning_rate": 1.9815007131535477e-05, + "loss": 0.1359844207763672, + "step": 834 + }, + { + "epoch": 0.4671328671328671, + "grad_norm": 0.462166541058912, + "learning_rate": 1.98137589692469e-05, + "loss": 0.15279315412044525, + "step": 835 + }, + { + "epoch": 0.4676923076923077, + "grad_norm": 0.4318145154579856, + "learning_rate": 1.981250664995358e-05, + "loss": 0.212006613612175, + "step": 836 + }, + { + "epoch": 0.46825174825174826, + "grad_norm": 0.5682220240892286, + "learning_rate": 1.9811250174185998e-05, + "loss": 0.07839135825634003, + "step": 837 + }, + { + "epoch": 0.46881118881118883, + "grad_norm": 0.2584445047770419, + "learning_rate": 1.9809989542476368e-05, + "loss": 0.04277295991778374, + "step": 838 + }, + { + "epoch": 0.46937062937062934, + "grad_norm": 0.4120649778772615, + "learning_rate": 1.9808724755358688e-05, + "loss": 0.09534847736358643, + "step": 839 + }, + { + "epoch": 0.4699300699300699, + "grad_norm": 0.5866064344732744, + "learning_rate": 1.980745581336871e-05, + "loss": 0.3481186628341675, + "step": 840 + }, + { + "epoch": 0.4704895104895105, + "grad_norm": 1.0731537857733242, + "learning_rate": 1.980618271704394e-05, + "loss": 0.2360180914402008, + "step": 841 + }, + { + "epoch": 0.47104895104895106, + "grad_norm": 0.6994679809381047, + "learning_rate": 1.980490546692365e-05, + "loss": 0.4861489534378052, + "step": 842 + }, + { + "epoch": 0.47160839160839163, + "grad_norm": 0.8056469584128081, + "learning_rate": 1.980362406354887e-05, + "loss": 0.237271249294281, + "step": 843 + }, + { + "epoch": 0.47216783216783215, + "grad_norm": 0.5185506209650115, + "learning_rate": 1.9802338507462395e-05, + "loss": 0.27513962984085083, + "step": 844 + }, + { + "epoch": 0.4727272727272727, + "grad_norm": 0.46499396636974344, + "learning_rate": 1.9801048799208763e-05, + "loss": 0.07421436160802841, + "step": 845 + }, + { + "epoch": 0.4732867132867133, + "grad_norm": 0.48464529850701543, + "learning_rate": 1.979975493933428e-05, + "loss": 0.1646387130022049, + "step": 846 + }, + { + "epoch": 0.47384615384615386, + "grad_norm": 0.3097587362779047, + "learning_rate": 1.9798456928387023e-05, + "loss": 0.10142280906438828, + "step": 847 + }, + { + "epoch": 0.47440559440559443, + "grad_norm": 0.6687306338465702, + "learning_rate": 1.9797154766916805e-05, + "loss": 0.2740744352340698, + "step": 848 + }, + { + "epoch": 0.47496503496503495, + "grad_norm": 0.3077133683329721, + "learning_rate": 1.9795848455475216e-05, + "loss": 0.07237780839204788, + "step": 849 + }, + { + "epoch": 0.4755244755244755, + "grad_norm": 0.8315278354248303, + "learning_rate": 1.9794537994615587e-05, + "loss": 0.05178440734744072, + "step": 850 + }, + { + "epoch": 0.4760839160839161, + "grad_norm": 0.5739827464320343, + "learning_rate": 1.9793223384893023e-05, + "loss": 0.191196471452713, + "step": 851 + }, + { + "epoch": 0.47664335664335666, + "grad_norm": 0.6527484286964517, + "learning_rate": 1.9791904626864378e-05, + "loss": 0.34000539779663086, + "step": 852 + }, + { + "epoch": 0.4772027972027972, + "grad_norm": 0.31226192101049427, + "learning_rate": 1.9790581721088257e-05, + "loss": 0.0698869526386261, + "step": 853 + }, + { + "epoch": 0.47776223776223775, + "grad_norm": 0.4240978944009363, + "learning_rate": 1.9789254668125035e-05, + "loss": 0.141836479306221, + "step": 854 + }, + { + "epoch": 0.4783216783216783, + "grad_norm": 0.42657642203785007, + "learning_rate": 1.9787923468536838e-05, + "loss": 0.1997612714767456, + "step": 855 + }, + { + "epoch": 0.4788811188811189, + "grad_norm": 0.5914016782954545, + "learning_rate": 1.9786588122887545e-05, + "loss": 0.2586309015750885, + "step": 856 + }, + { + "epoch": 0.47944055944055947, + "grad_norm": 0.7250030036048524, + "learning_rate": 1.9785248631742796e-05, + "loss": 0.37913331389427185, + "step": 857 + }, + { + "epoch": 0.48, + "grad_norm": 0.6759882890315124, + "learning_rate": 1.9783904995669987e-05, + "loss": 0.249455526471138, + "step": 858 + }, + { + "epoch": 0.48055944055944055, + "grad_norm": 0.41152608873617197, + "learning_rate": 1.9782557215238263e-05, + "loss": 0.20057329535484314, + "step": 859 + }, + { + "epoch": 0.4811188811188811, + "grad_norm": 0.40469878668483145, + "learning_rate": 1.978120529101853e-05, + "loss": 0.14626863598823547, + "step": 860 + }, + { + "epoch": 0.4816783216783217, + "grad_norm": 0.5384614358937473, + "learning_rate": 1.977984922358346e-05, + "loss": 0.2326146960258484, + "step": 861 + }, + { + "epoch": 0.4822377622377622, + "grad_norm": 0.5764788446071701, + "learning_rate": 1.9778489013507455e-05, + "loss": 0.2547037899494171, + "step": 862 + }, + { + "epoch": 0.4827972027972028, + "grad_norm": 0.564314149990606, + "learning_rate": 1.9777124661366693e-05, + "loss": 0.25627368688583374, + "step": 863 + }, + { + "epoch": 0.48335664335664336, + "grad_norm": 0.6038921441154158, + "learning_rate": 1.9775756167739098e-05, + "loss": 0.42010629177093506, + "step": 864 + }, + { + "epoch": 0.48391608391608393, + "grad_norm": 0.41211789709900554, + "learning_rate": 1.9774383533204345e-05, + "loss": 0.0896216481924057, + "step": 865 + }, + { + "epoch": 0.4844755244755245, + "grad_norm": 0.601172181547031, + "learning_rate": 1.9773006758343873e-05, + "loss": 0.13138100504875183, + "step": 866 + }, + { + "epoch": 0.485034965034965, + "grad_norm": 0.4171215979566442, + "learning_rate": 1.977162584374087e-05, + "loss": 0.12429220229387283, + "step": 867 + }, + { + "epoch": 0.4855944055944056, + "grad_norm": 0.39315584281535243, + "learning_rate": 1.977024078998027e-05, + "loss": 0.16589125990867615, + "step": 868 + }, + { + "epoch": 0.48615384615384616, + "grad_norm": 0.6102601962815352, + "learning_rate": 1.9768851597648774e-05, + "loss": 0.2439456731081009, + "step": 869 + }, + { + "epoch": 0.48671328671328673, + "grad_norm": 0.5999659441026637, + "learning_rate": 1.9767458267334828e-05, + "loss": 0.3458921015262604, + "step": 870 + }, + { + "epoch": 0.48727272727272725, + "grad_norm": 0.5483005486993848, + "learning_rate": 1.9766060799628627e-05, + "loss": 0.25526830554008484, + "step": 871 + }, + { + "epoch": 0.4878321678321678, + "grad_norm": 1.5145604646487338, + "learning_rate": 1.976465919512213e-05, + "loss": 0.2836688756942749, + "step": 872 + }, + { + "epoch": 0.4883916083916084, + "grad_norm": 0.310475987534435, + "learning_rate": 1.9763253454409038e-05, + "loss": 0.049724068492650986, + "step": 873 + }, + { + "epoch": 0.48895104895104896, + "grad_norm": 0.5646519490235608, + "learning_rate": 1.9761843578084813e-05, + "loss": 0.19707101583480835, + "step": 874 + }, + { + "epoch": 0.48951048951048953, + "grad_norm": 0.5059572107663716, + "learning_rate": 1.9760429566746656e-05, + "loss": 0.29132080078125, + "step": 875 + }, + { + "epoch": 0.49006993006993005, + "grad_norm": 0.48647471238238577, + "learning_rate": 1.9759011420993527e-05, + "loss": 0.19825905561447144, + "step": 876 + }, + { + "epoch": 0.4906293706293706, + "grad_norm": 0.5640670334899685, + "learning_rate": 1.975758914142615e-05, + "loss": 0.2920805513858795, + "step": 877 + }, + { + "epoch": 0.4911888111888112, + "grad_norm": 0.6544021541951976, + "learning_rate": 1.9756162728646972e-05, + "loss": 0.41261839866638184, + "step": 878 + }, + { + "epoch": 0.49174825174825176, + "grad_norm": 0.3425387148369511, + "learning_rate": 1.975473218326022e-05, + "loss": 0.14197523891925812, + "step": 879 + }, + { + "epoch": 0.49230769230769234, + "grad_norm": 0.44099064833143653, + "learning_rate": 1.975329750587185e-05, + "loss": 0.16157715022563934, + "step": 880 + }, + { + "epoch": 0.49286713286713285, + "grad_norm": 0.6165183479803659, + "learning_rate": 1.9751858697089577e-05, + "loss": 0.3838087320327759, + "step": 881 + }, + { + "epoch": 0.4934265734265734, + "grad_norm": 0.45230863266771576, + "learning_rate": 1.9750415757522863e-05, + "loss": 0.19504405558109283, + "step": 882 + }, + { + "epoch": 0.493986013986014, + "grad_norm": 0.4549333072748831, + "learning_rate": 1.974896868778293e-05, + "loss": 0.2133583277463913, + "step": 883 + }, + { + "epoch": 0.49454545454545457, + "grad_norm": 0.436649696885772, + "learning_rate": 1.9747517488482732e-05, + "loss": 0.2611194849014282, + "step": 884 + }, + { + "epoch": 0.4951048951048951, + "grad_norm": 0.2996273818742287, + "learning_rate": 1.974606216023699e-05, + "loss": 0.04857818782329559, + "step": 885 + }, + { + "epoch": 0.49566433566433565, + "grad_norm": 0.5171797435543154, + "learning_rate": 1.9744602703662158e-05, + "loss": 0.3626279830932617, + "step": 886 + }, + { + "epoch": 0.4962237762237762, + "grad_norm": 0.6377280949195813, + "learning_rate": 1.9743139119376457e-05, + "loss": 0.4035492539405823, + "step": 887 + }, + { + "epoch": 0.4967832167832168, + "grad_norm": 0.4822711252513988, + "learning_rate": 1.9741671407999835e-05, + "loss": 0.3036898076534271, + "step": 888 + }, + { + "epoch": 0.49734265734265737, + "grad_norm": 0.5002837832008178, + "learning_rate": 1.9740199570154003e-05, + "loss": 0.12967118620872498, + "step": 889 + }, + { + "epoch": 0.4979020979020979, + "grad_norm": 0.47596180781071107, + "learning_rate": 1.9738723606462415e-05, + "loss": 0.23828613758087158, + "step": 890 + }, + { + "epoch": 0.49846153846153846, + "grad_norm": 0.4837432989414203, + "learning_rate": 1.973724351755028e-05, + "loss": 0.2823325991630554, + "step": 891 + }, + { + "epoch": 0.499020979020979, + "grad_norm": 0.6515387094660723, + "learning_rate": 1.9735759304044536e-05, + "loss": 0.26219749450683594, + "step": 892 + }, + { + "epoch": 0.4995804195804196, + "grad_norm": 0.543535720164126, + "learning_rate": 1.973427096657389e-05, + "loss": 0.25803399085998535, + "step": 893 + }, + { + "epoch": 0.5001398601398601, + "grad_norm": 0.6141049767761076, + "learning_rate": 1.9732778505768785e-05, + "loss": 0.2967890799045563, + "step": 894 + }, + { + "epoch": 0.5006993006993007, + "grad_norm": 0.5795144968703164, + "learning_rate": 1.9731281922261406e-05, + "loss": 0.2366620898246765, + "step": 895 + }, + { + "epoch": 0.5012587412587413, + "grad_norm": 0.6153304021224246, + "learning_rate": 1.9729781216685695e-05, + "loss": 0.21914657950401306, + "step": 896 + }, + { + "epoch": 0.5018181818181818, + "grad_norm": 0.4045201809541528, + "learning_rate": 1.9728276389677338e-05, + "loss": 0.11656968295574188, + "step": 897 + }, + { + "epoch": 0.5023776223776224, + "grad_norm": 0.41747171186203724, + "learning_rate": 1.9726767441873753e-05, + "loss": 0.16053169965744019, + "step": 898 + }, + { + "epoch": 0.502937062937063, + "grad_norm": 0.2814733100435221, + "learning_rate": 1.9725254373914126e-05, + "loss": 0.06676869839429855, + "step": 899 + }, + { + "epoch": 0.5034965034965035, + "grad_norm": 0.5605315128660955, + "learning_rate": 1.9723737186439372e-05, + "loss": 0.3024398982524872, + "step": 900 + }, + { + "epoch": 0.504055944055944, + "grad_norm": 0.4668411523485862, + "learning_rate": 1.9722215880092152e-05, + "loss": 0.2132788747549057, + "step": 901 + }, + { + "epoch": 0.5046153846153846, + "grad_norm": 0.5051303691377054, + "learning_rate": 1.972069045551688e-05, + "loss": 0.24151788651943207, + "step": 902 + }, + { + "epoch": 0.5051748251748251, + "grad_norm": 0.12536307558556387, + "learning_rate": 1.9719160913359706e-05, + "loss": 0.01287270151078701, + "step": 903 + }, + { + "epoch": 0.5057342657342657, + "grad_norm": 0.42439748014320644, + "learning_rate": 1.9717627254268533e-05, + "loss": 0.12392720580101013, + "step": 904 + }, + { + "epoch": 0.5062937062937063, + "grad_norm": 0.4961340025800344, + "learning_rate": 1.9716089478893e-05, + "loss": 0.26476046442985535, + "step": 905 + }, + { + "epoch": 0.5068531468531469, + "grad_norm": 0.42627170310758233, + "learning_rate": 1.9714547587884492e-05, + "loss": 0.15284022688865662, + "step": 906 + }, + { + "epoch": 0.5074125874125874, + "grad_norm": 0.6151848264914244, + "learning_rate": 1.9713001581896136e-05, + "loss": 0.2212950438261032, + "step": 907 + }, + { + "epoch": 0.507972027972028, + "grad_norm": 0.283165775448647, + "learning_rate": 1.971145146158281e-05, + "loss": 0.033081285655498505, + "step": 908 + }, + { + "epoch": 0.5085314685314686, + "grad_norm": 0.5721708434338769, + "learning_rate": 1.970989722760112e-05, + "loss": 0.37525105476379395, + "step": 909 + }, + { + "epoch": 0.509090909090909, + "grad_norm": 0.3012461067261731, + "learning_rate": 1.970833888060943e-05, + "loss": 0.1389291137456894, + "step": 910 + }, + { + "epoch": 0.5096503496503496, + "grad_norm": 5.110438363836095, + "learning_rate": 1.970677642126784e-05, + "loss": 0.1689014732837677, + "step": 911 + }, + { + "epoch": 0.5102097902097902, + "grad_norm": 0.8161205677181159, + "learning_rate": 1.9705209850238185e-05, + "loss": 0.3219829201698303, + "step": 912 + }, + { + "epoch": 0.5107692307692308, + "grad_norm": 0.22717706224643375, + "learning_rate": 1.9703639168184052e-05, + "loss": 0.030895326286554337, + "step": 913 + }, + { + "epoch": 0.5113286713286713, + "grad_norm": 0.4295499639467397, + "learning_rate": 1.9702064375770765e-05, + "loss": 0.11441054195165634, + "step": 914 + }, + { + "epoch": 0.5118881118881119, + "grad_norm": 0.5531439676726262, + "learning_rate": 1.9700485473665386e-05, + "loss": 0.20935362577438354, + "step": 915 + }, + { + "epoch": 0.5124475524475525, + "grad_norm": 0.7349369399204299, + "learning_rate": 1.969890246253673e-05, + "loss": 0.31683462858200073, + "step": 916 + }, + { + "epoch": 0.513006993006993, + "grad_norm": 0.5628460773684217, + "learning_rate": 1.9697315343055336e-05, + "loss": 0.2438500076532364, + "step": 917 + }, + { + "epoch": 0.5135664335664336, + "grad_norm": 0.27480912961350173, + "learning_rate": 1.969572411589349e-05, + "loss": 0.10532652586698532, + "step": 918 + }, + { + "epoch": 0.5141258741258741, + "grad_norm": 0.7688583462518426, + "learning_rate": 1.9694128781725225e-05, + "loss": 0.48565253615379333, + "step": 919 + }, + { + "epoch": 0.5146853146853146, + "grad_norm": 0.8595865073704696, + "learning_rate": 1.969252934122631e-05, + "loss": 0.3974907696247101, + "step": 920 + }, + { + "epoch": 0.5152447552447552, + "grad_norm": 0.6076720290273178, + "learning_rate": 1.969092579507424e-05, + "loss": 0.2859296202659607, + "step": 921 + }, + { + "epoch": 0.5158041958041958, + "grad_norm": 0.5698522170016044, + "learning_rate": 1.9689318143948266e-05, + "loss": 0.22826267778873444, + "step": 922 + }, + { + "epoch": 0.5163636363636364, + "grad_norm": 0.5558407413040856, + "learning_rate": 1.9687706388529376e-05, + "loss": 0.26515689492225647, + "step": 923 + }, + { + "epoch": 0.5169230769230769, + "grad_norm": 0.5266228251699516, + "learning_rate": 1.9686090529500282e-05, + "loss": 0.22250650823116302, + "step": 924 + }, + { + "epoch": 0.5174825174825175, + "grad_norm": 0.4445912561912751, + "learning_rate": 1.968447056754546e-05, + "loss": 0.17744609713554382, + "step": 925 + }, + { + "epoch": 0.5180419580419581, + "grad_norm": 0.3545768482367548, + "learning_rate": 1.9682846503351097e-05, + "loss": 0.1091318279504776, + "step": 926 + }, + { + "epoch": 0.5186013986013986, + "grad_norm": 0.5634047975591421, + "learning_rate": 1.9681218337605134e-05, + "loss": 0.22607238590717316, + "step": 927 + }, + { + "epoch": 0.5191608391608392, + "grad_norm": 0.6205696761707331, + "learning_rate": 1.9679586070997243e-05, + "loss": 0.3530200719833374, + "step": 928 + }, + { + "epoch": 0.5197202797202797, + "grad_norm": 0.43389162923158475, + "learning_rate": 1.9677949704218836e-05, + "loss": 0.2328314483165741, + "step": 929 + }, + { + "epoch": 0.5202797202797202, + "grad_norm": 0.4320974681056739, + "learning_rate": 1.967630923796306e-05, + "loss": 0.10878923535346985, + "step": 930 + }, + { + "epoch": 0.5208391608391608, + "grad_norm": 0.4720667255990928, + "learning_rate": 1.96746646729248e-05, + "loss": 0.1710374653339386, + "step": 931 + }, + { + "epoch": 0.5213986013986014, + "grad_norm": 0.4754940172841316, + "learning_rate": 1.967301600980068e-05, + "loss": 0.36729955673217773, + "step": 932 + }, + { + "epoch": 0.521958041958042, + "grad_norm": 0.39031288679885306, + "learning_rate": 1.967136324928905e-05, + "loss": 0.06783082336187363, + "step": 933 + }, + { + "epoch": 0.5225174825174825, + "grad_norm": 0.5719994105861139, + "learning_rate": 1.966970639209e-05, + "loss": 0.2012096643447876, + "step": 934 + }, + { + "epoch": 0.5230769230769231, + "grad_norm": 0.2481609759046759, + "learning_rate": 1.9668045438905367e-05, + "loss": 0.05832952633500099, + "step": 935 + }, + { + "epoch": 0.5236363636363637, + "grad_norm": 0.6097414310459729, + "learning_rate": 1.9666380390438706e-05, + "loss": 0.37049317359924316, + "step": 936 + }, + { + "epoch": 0.5241958041958042, + "grad_norm": 0.39625108024467665, + "learning_rate": 1.9664711247395313e-05, + "loss": 0.10587126016616821, + "step": 937 + }, + { + "epoch": 0.5247552447552447, + "grad_norm": 0.6283130146930502, + "learning_rate": 1.9663038010482224e-05, + "loss": 0.39503955841064453, + "step": 938 + }, + { + "epoch": 0.5253146853146853, + "grad_norm": 0.43447366303818813, + "learning_rate": 1.9661360680408202e-05, + "loss": 0.20081260800361633, + "step": 939 + }, + { + "epoch": 0.5258741258741259, + "grad_norm": 0.6893715745283596, + "learning_rate": 1.965967925788375e-05, + "loss": 0.4869721829891205, + "step": 940 + }, + { + "epoch": 0.5264335664335664, + "grad_norm": 0.4482880008306642, + "learning_rate": 1.9657993743621098e-05, + "loss": 0.17658472061157227, + "step": 941 + }, + { + "epoch": 0.526993006993007, + "grad_norm": 0.59875030409008, + "learning_rate": 1.965630413833421e-05, + "loss": 0.3039664626121521, + "step": 942 + }, + { + "epoch": 0.5275524475524476, + "grad_norm": 0.6476748141420604, + "learning_rate": 1.9654610442738786e-05, + "loss": 0.43305009603500366, + "step": 943 + }, + { + "epoch": 0.5281118881118881, + "grad_norm": 0.35257660331865576, + "learning_rate": 1.965291265755226e-05, + "loss": 0.14372022449970245, + "step": 944 + }, + { + "epoch": 0.5286713286713287, + "grad_norm": 0.6777446415949456, + "learning_rate": 1.9651210783493794e-05, + "loss": 0.21755558252334595, + "step": 945 + }, + { + "epoch": 0.5292307692307693, + "grad_norm": 0.6549430934942426, + "learning_rate": 1.964950482128428e-05, + "loss": 0.2760806083679199, + "step": 946 + }, + { + "epoch": 0.5297902097902097, + "grad_norm": 0.47597942430351514, + "learning_rate": 1.9647794771646356e-05, + "loss": 0.13141363859176636, + "step": 947 + }, + { + "epoch": 0.5303496503496503, + "grad_norm": 0.5006709867682277, + "learning_rate": 1.9646080635304374e-05, + "loss": 0.2564341127872467, + "step": 948 + }, + { + "epoch": 0.5309090909090909, + "grad_norm": 0.7819516816394958, + "learning_rate": 1.964436241298442e-05, + "loss": 0.37522318959236145, + "step": 949 + }, + { + "epoch": 0.5314685314685315, + "grad_norm": 0.5672639105601728, + "learning_rate": 1.9642640105414324e-05, + "loss": 0.4672318696975708, + "step": 950 + }, + { + "epoch": 0.532027972027972, + "grad_norm": 0.29877546805183625, + "learning_rate": 1.9640913713323632e-05, + "loss": 0.08840466290712357, + "step": 951 + }, + { + "epoch": 0.5325874125874126, + "grad_norm": 0.7821611762531359, + "learning_rate": 1.9639183237443623e-05, + "loss": 0.3974948525428772, + "step": 952 + }, + { + "epoch": 0.5331468531468532, + "grad_norm": 0.3560620441956737, + "learning_rate": 1.9637448678507317e-05, + "loss": 0.08519744873046875, + "step": 953 + }, + { + "epoch": 0.5337062937062937, + "grad_norm": 0.500652340635663, + "learning_rate": 1.9635710037249446e-05, + "loss": 0.22603653371334076, + "step": 954 + }, + { + "epoch": 0.5342657342657343, + "grad_norm": 0.5255656617487605, + "learning_rate": 1.9633967314406485e-05, + "loss": 0.11233668029308319, + "step": 955 + }, + { + "epoch": 0.5348251748251748, + "grad_norm": 0.5241596258308964, + "learning_rate": 1.9632220510716635e-05, + "loss": 0.30219048261642456, + "step": 956 + }, + { + "epoch": 0.5353846153846153, + "grad_norm": 0.7022155161652487, + "learning_rate": 1.9630469626919814e-05, + "loss": 0.3165651559829712, + "step": 957 + }, + { + "epoch": 0.5359440559440559, + "grad_norm": 0.4267553375781151, + "learning_rate": 1.962871466375769e-05, + "loss": 0.14552021026611328, + "step": 958 + }, + { + "epoch": 0.5365034965034965, + "grad_norm": 0.5082559110189423, + "learning_rate": 1.9626955621973634e-05, + "loss": 0.31409138441085815, + "step": 959 + }, + { + "epoch": 0.5370629370629371, + "grad_norm": 0.4995175045666699, + "learning_rate": 1.9625192502312772e-05, + "loss": 0.13729244470596313, + "step": 960 + }, + { + "epoch": 0.5376223776223776, + "grad_norm": 0.6062229503461546, + "learning_rate": 1.9623425305521932e-05, + "loss": 0.21573972702026367, + "step": 961 + }, + { + "epoch": 0.5381818181818182, + "grad_norm": 0.45313964266842344, + "learning_rate": 1.9621654032349685e-05, + "loss": 0.2081308215856552, + "step": 962 + }, + { + "epoch": 0.5387412587412588, + "grad_norm": 0.5515128470140086, + "learning_rate": 1.961987868354632e-05, + "loss": 0.2961956262588501, + "step": 963 + }, + { + "epoch": 0.5393006993006993, + "grad_norm": 0.444805831802089, + "learning_rate": 1.9618099259863862e-05, + "loss": 0.12377064675092697, + "step": 964 + }, + { + "epoch": 0.5398601398601398, + "grad_norm": 0.22797573246984912, + "learning_rate": 1.961631576205605e-05, + "loss": 0.043612588196992874, + "step": 965 + }, + { + "epoch": 0.5404195804195804, + "grad_norm": 0.49656514425739123, + "learning_rate": 1.9614528190878356e-05, + "loss": 0.24353444576263428, + "step": 966 + }, + { + "epoch": 0.540979020979021, + "grad_norm": 0.49230409796535074, + "learning_rate": 1.9612736547087977e-05, + "loss": 0.21760807931423187, + "step": 967 + }, + { + "epoch": 0.5415384615384615, + "grad_norm": 0.46824959547489425, + "learning_rate": 1.9610940831443836e-05, + "loss": 0.26910704374313354, + "step": 968 + }, + { + "epoch": 0.5420979020979021, + "grad_norm": 0.4960645823109151, + "learning_rate": 1.9609141044706576e-05, + "loss": 0.2254650592803955, + "step": 969 + }, + { + "epoch": 0.5426573426573427, + "grad_norm": 0.36026213255278344, + "learning_rate": 1.9607337187638573e-05, + "loss": 0.10507339239120483, + "step": 970 + }, + { + "epoch": 0.5432167832167832, + "grad_norm": 0.5727732894064373, + "learning_rate": 1.9605529261003916e-05, + "loss": 0.19645673036575317, + "step": 971 + }, + { + "epoch": 0.5437762237762238, + "grad_norm": 0.5196004582165811, + "learning_rate": 1.9603717265568427e-05, + "loss": 0.32230716943740845, + "step": 972 + }, + { + "epoch": 0.5443356643356644, + "grad_norm": 0.41136808276788234, + "learning_rate": 1.9601901202099648e-05, + "loss": 0.15786448121070862, + "step": 973 + }, + { + "epoch": 0.5448951048951048, + "grad_norm": 0.40865665283884256, + "learning_rate": 1.960008107136684e-05, + "loss": 0.144511878490448, + "step": 974 + }, + { + "epoch": 0.5454545454545454, + "grad_norm": 0.6528014945401623, + "learning_rate": 1.9598256874140995e-05, + "loss": 0.19467437267303467, + "step": 975 + }, + { + "epoch": 0.546013986013986, + "grad_norm": 0.39729490257773314, + "learning_rate": 1.9596428611194825e-05, + "loss": 0.16548840701580048, + "step": 976 + }, + { + "epoch": 0.5465734265734266, + "grad_norm": 0.45982070622105214, + "learning_rate": 1.9594596283302762e-05, + "loss": 0.14353393018245697, + "step": 977 + }, + { + "epoch": 0.5471328671328671, + "grad_norm": 0.3828075121032531, + "learning_rate": 1.959275989124096e-05, + "loss": 0.15446056425571442, + "step": 978 + }, + { + "epoch": 0.5476923076923077, + "grad_norm": 0.49340173315971, + "learning_rate": 1.9590919435787295e-05, + "loss": 0.1718764752149582, + "step": 979 + }, + { + "epoch": 0.5482517482517483, + "grad_norm": 1.3817348735712065, + "learning_rate": 1.9589074917721364e-05, + "loss": 0.16891226172447205, + "step": 980 + }, + { + "epoch": 0.5488111888111888, + "grad_norm": 0.5417255553237206, + "learning_rate": 1.958722633782449e-05, + "loss": 0.18720705807209015, + "step": 981 + }, + { + "epoch": 0.5493706293706294, + "grad_norm": 0.9004317232982706, + "learning_rate": 1.9585373696879705e-05, + "loss": 0.29135066270828247, + "step": 982 + }, + { + "epoch": 0.5499300699300699, + "grad_norm": 0.5574359057438082, + "learning_rate": 1.9583516995671773e-05, + "loss": 0.3132430911064148, + "step": 983 + }, + { + "epoch": 0.5504895104895104, + "grad_norm": 0.441012953481346, + "learning_rate": 1.958165623498717e-05, + "loss": 0.21503329277038574, + "step": 984 + }, + { + "epoch": 0.551048951048951, + "grad_norm": 0.309753427076812, + "learning_rate": 1.9579791415614094e-05, + "loss": 0.04449233412742615, + "step": 985 + }, + { + "epoch": 0.5516083916083916, + "grad_norm": 0.526219391645922, + "learning_rate": 1.9577922538342467e-05, + "loss": 0.28126204013824463, + "step": 986 + }, + { + "epoch": 0.5521678321678322, + "grad_norm": 0.41903380244220095, + "learning_rate": 1.9576049603963922e-05, + "loss": 0.14415481686592102, + "step": 987 + }, + { + "epoch": 0.5527272727272727, + "grad_norm": 0.5039864199035275, + "learning_rate": 1.9574172613271816e-05, + "loss": 0.2581061124801636, + "step": 988 + }, + { + "epoch": 0.5532867132867133, + "grad_norm": 0.4351920068402927, + "learning_rate": 1.9572291567061224e-05, + "loss": 0.17576652765274048, + "step": 989 + }, + { + "epoch": 0.5538461538461539, + "grad_norm": 0.325476175629847, + "learning_rate": 1.9570406466128934e-05, + "loss": 0.0734764039516449, + "step": 990 + }, + { + "epoch": 0.5544055944055944, + "grad_norm": 0.5252044943901742, + "learning_rate": 1.956851731127346e-05, + "loss": 0.2866566777229309, + "step": 991 + }, + { + "epoch": 0.554965034965035, + "grad_norm": 0.6078809179490325, + "learning_rate": 1.956662410329502e-05, + "loss": 0.3186303973197937, + "step": 992 + }, + { + "epoch": 0.5555244755244755, + "grad_norm": 0.34570269920909485, + "learning_rate": 1.956472684299556e-05, + "loss": 0.11688470840454102, + "step": 993 + }, + { + "epoch": 0.556083916083916, + "grad_norm": 0.5629758342405143, + "learning_rate": 1.9562825531178746e-05, + "loss": 0.17378711700439453, + "step": 994 + }, + { + "epoch": 0.5566433566433566, + "grad_norm": 0.6635879202336314, + "learning_rate": 1.9560920168649944e-05, + "loss": 0.3526480793952942, + "step": 995 + }, + { + "epoch": 0.5572027972027972, + "grad_norm": 0.5014038844955042, + "learning_rate": 1.955901075621625e-05, + "loss": 0.18774738907814026, + "step": 996 + }, + { + "epoch": 0.5577622377622378, + "grad_norm": 0.40728311865404654, + "learning_rate": 1.9557097294686477e-05, + "loss": 0.12678533792495728, + "step": 997 + }, + { + "epoch": 0.5583216783216783, + "grad_norm": 0.4892753139052474, + "learning_rate": 1.9555179784871132e-05, + "loss": 0.20538268983364105, + "step": 998 + }, + { + "epoch": 0.5588811188811189, + "grad_norm": 0.5784879572330683, + "learning_rate": 1.955325822758247e-05, + "loss": 0.2257968932390213, + "step": 999 + }, + { + "epoch": 0.5594405594405595, + "grad_norm": 0.642045308432166, + "learning_rate": 1.9551332623634426e-05, + "loss": 0.3862190246582031, + "step": 1000 + }, + { + "epoch": 0.56, + "grad_norm": 0.3592242289994624, + "learning_rate": 1.954940297384267e-05, + "loss": 0.07053911685943604, + "step": 1001 + }, + { + "epoch": 0.5605594405594405, + "grad_norm": 0.5981824093166823, + "learning_rate": 1.9547469279024593e-05, + "loss": 0.32261836528778076, + "step": 1002 + }, + { + "epoch": 0.5611188811188811, + "grad_norm": 0.45066016971879785, + "learning_rate": 1.9545531539999275e-05, + "loss": 0.15613330900669098, + "step": 1003 + }, + { + "epoch": 0.5616783216783217, + "grad_norm": 0.5092913985497579, + "learning_rate": 1.9543589757587525e-05, + "loss": 0.17701421678066254, + "step": 1004 + }, + { + "epoch": 0.5622377622377622, + "grad_norm": 0.5551288629406811, + "learning_rate": 1.954164393261186e-05, + "loss": 0.3297882676124573, + "step": 1005 + }, + { + "epoch": 0.5627972027972028, + "grad_norm": 0.2286687005704506, + "learning_rate": 1.9539694065896516e-05, + "loss": 0.06182924285531044, + "step": 1006 + }, + { + "epoch": 0.5633566433566434, + "grad_norm": 0.4175514106755521, + "learning_rate": 1.9537740158267432e-05, + "loss": 0.1331361085176468, + "step": 1007 + }, + { + "epoch": 0.5639160839160839, + "grad_norm": 0.38977531620508193, + "learning_rate": 1.9535782210552266e-05, + "loss": 0.13544823229312897, + "step": 1008 + }, + { + "epoch": 0.5644755244755245, + "grad_norm": 0.4412139189264216, + "learning_rate": 1.953382022358038e-05, + "loss": 0.16716547310352325, + "step": 1009 + }, + { + "epoch": 0.5650349650349651, + "grad_norm": 0.5409888857776122, + "learning_rate": 1.9531854198182853e-05, + "loss": 0.33848369121551514, + "step": 1010 + }, + { + "epoch": 0.5655944055944055, + "grad_norm": 0.6572011048948926, + "learning_rate": 1.952988413519247e-05, + "loss": 0.4353000819683075, + "step": 1011 + }, + { + "epoch": 0.5661538461538461, + "grad_norm": 0.31144488089812505, + "learning_rate": 1.9527910035443735e-05, + "loss": 0.05286436527967453, + "step": 1012 + }, + { + "epoch": 0.5667132867132867, + "grad_norm": 0.4889108903525951, + "learning_rate": 1.9525931899772848e-05, + "loss": 0.24054157733917236, + "step": 1013 + }, + { + "epoch": 0.5672727272727273, + "grad_norm": 0.40568696945314003, + "learning_rate": 1.952394972901773e-05, + "loss": 0.14250990748405457, + "step": 1014 + }, + { + "epoch": 0.5678321678321678, + "grad_norm": 0.7293716527286564, + "learning_rate": 1.952196352401801e-05, + "loss": 0.30289068818092346, + "step": 1015 + }, + { + "epoch": 0.5683916083916084, + "grad_norm": 4.246973799326449, + "learning_rate": 1.9519973285615023e-05, + "loss": 0.27133095264434814, + "step": 1016 + }, + { + "epoch": 0.568951048951049, + "grad_norm": 0.44256804482127643, + "learning_rate": 1.9517979014651808e-05, + "loss": 0.1966119110584259, + "step": 1017 + }, + { + "epoch": 0.5695104895104895, + "grad_norm": 0.6099137565113083, + "learning_rate": 1.9515980711973123e-05, + "loss": 0.17500001192092896, + "step": 1018 + }, + { + "epoch": 0.5700699300699301, + "grad_norm": 0.3564897254042853, + "learning_rate": 1.9513978378425427e-05, + "loss": 0.1158447340130806, + "step": 1019 + }, + { + "epoch": 0.5706293706293706, + "grad_norm": 0.40898670868943754, + "learning_rate": 1.9511972014856882e-05, + "loss": 0.15415272116661072, + "step": 1020 + }, + { + "epoch": 0.5711888111888112, + "grad_norm": 0.9798055243217896, + "learning_rate": 1.9509961622117366e-05, + "loss": 0.2781256139278412, + "step": 1021 + }, + { + "epoch": 0.5717482517482517, + "grad_norm": 0.48593700975722587, + "learning_rate": 1.9507947201058464e-05, + "loss": 0.2514978051185608, + "step": 1022 + }, + { + "epoch": 0.5723076923076923, + "grad_norm": 0.3641288776161636, + "learning_rate": 1.9505928752533458e-05, + "loss": 0.09165353327989578, + "step": 1023 + }, + { + "epoch": 0.5728671328671329, + "grad_norm": 0.5376961318366343, + "learning_rate": 1.9503906277397342e-05, + "loss": 0.14220130443572998, + "step": 1024 + }, + { + "epoch": 0.5734265734265734, + "grad_norm": 1.133882342123603, + "learning_rate": 1.950187977650682e-05, + "loss": 0.315986692905426, + "step": 1025 + }, + { + "epoch": 0.573986013986014, + "grad_norm": 0.3197808304545089, + "learning_rate": 1.9499849250720292e-05, + "loss": 0.11382804065942764, + "step": 1026 + }, + { + "epoch": 0.5745454545454546, + "grad_norm": 0.4782667778991907, + "learning_rate": 1.9497814700897866e-05, + "loss": 0.12490015476942062, + "step": 1027 + }, + { + "epoch": 0.5751048951048952, + "grad_norm": 0.2737731129560135, + "learning_rate": 1.9495776127901357e-05, + "loss": 0.0684887021780014, + "step": 1028 + }, + { + "epoch": 0.5756643356643356, + "grad_norm": 0.36365008534101323, + "learning_rate": 1.9493733532594286e-05, + "loss": 0.12830427289009094, + "step": 1029 + }, + { + "epoch": 0.5762237762237762, + "grad_norm": 0.5372221370491576, + "learning_rate": 1.9491686915841876e-05, + "loss": 0.19304972887039185, + "step": 1030 + }, + { + "epoch": 0.5767832167832168, + "grad_norm": 0.6384279525731293, + "learning_rate": 1.9489636278511046e-05, + "loss": 0.34067511558532715, + "step": 1031 + }, + { + "epoch": 0.5773426573426573, + "grad_norm": 0.45394498935639654, + "learning_rate": 1.9487581621470426e-05, + "loss": 0.24409878253936768, + "step": 1032 + }, + { + "epoch": 0.5779020979020979, + "grad_norm": 0.3503362525447514, + "learning_rate": 1.948552294559035e-05, + "loss": 0.11452331393957138, + "step": 1033 + }, + { + "epoch": 0.5784615384615385, + "grad_norm": 0.5293488479208528, + "learning_rate": 1.9483460251742852e-05, + "loss": 0.2811158001422882, + "step": 1034 + }, + { + "epoch": 0.579020979020979, + "grad_norm": 0.2663211892530333, + "learning_rate": 1.9481393540801664e-05, + "loss": 0.15953418612480164, + "step": 1035 + }, + { + "epoch": 0.5795804195804196, + "grad_norm": 0.4150814527417164, + "learning_rate": 1.947932281364222e-05, + "loss": 0.12728016078472137, + "step": 1036 + }, + { + "epoch": 0.5801398601398602, + "grad_norm": 0.6652136343457465, + "learning_rate": 1.947724807114167e-05, + "loss": 0.4229060113430023, + "step": 1037 + }, + { + "epoch": 0.5806993006993006, + "grad_norm": 0.6537982644952748, + "learning_rate": 1.947516931417884e-05, + "loss": 0.37713998556137085, + "step": 1038 + }, + { + "epoch": 0.5812587412587412, + "grad_norm": 0.3735228599673671, + "learning_rate": 1.9473086543634277e-05, + "loss": 0.1380643993616104, + "step": 1039 + }, + { + "epoch": 0.5818181818181818, + "grad_norm": 0.524390344642552, + "learning_rate": 1.9470999760390216e-05, + "loss": 0.2617342174053192, + "step": 1040 + }, + { + "epoch": 0.5823776223776224, + "grad_norm": 0.543674011455634, + "learning_rate": 1.9468908965330603e-05, + "loss": 0.07876032590866089, + "step": 1041 + }, + { + "epoch": 0.5829370629370629, + "grad_norm": 1.5094773427600474, + "learning_rate": 1.9466814159341066e-05, + "loss": 0.3294464647769928, + "step": 1042 + }, + { + "epoch": 0.5834965034965035, + "grad_norm": 0.7485202191558704, + "learning_rate": 1.9464715343308957e-05, + "loss": 0.37307578325271606, + "step": 1043 + }, + { + "epoch": 0.5840559440559441, + "grad_norm": 0.3748100961172418, + "learning_rate": 1.94626125181233e-05, + "loss": 0.11747735738754272, + "step": 1044 + }, + { + "epoch": 0.5846153846153846, + "grad_norm": 0.6335631085530925, + "learning_rate": 1.9460505684674836e-05, + "loss": 0.20038610696792603, + "step": 1045 + }, + { + "epoch": 0.5851748251748252, + "grad_norm": 0.32985957659576787, + "learning_rate": 1.9458394843855998e-05, + "loss": 0.08299502730369568, + "step": 1046 + }, + { + "epoch": 0.5857342657342657, + "grad_norm": 0.5678833052407943, + "learning_rate": 1.9456279996560908e-05, + "loss": 0.4093180298805237, + "step": 1047 + }, + { + "epoch": 0.5862937062937063, + "grad_norm": 0.6525739046688867, + "learning_rate": 1.94541611436854e-05, + "loss": 0.31745290756225586, + "step": 1048 + }, + { + "epoch": 0.5868531468531468, + "grad_norm": 0.47537122258372044, + "learning_rate": 1.9452038286127e-05, + "loss": 0.2111779898405075, + "step": 1049 + }, + { + "epoch": 0.5874125874125874, + "grad_norm": 0.3782099062855441, + "learning_rate": 1.944991142478492e-05, + "loss": 0.18930718302726746, + "step": 1050 + }, + { + "epoch": 0.587972027972028, + "grad_norm": 0.33456693607673427, + "learning_rate": 1.9447780560560085e-05, + "loss": 0.08649227023124695, + "step": 1051 + }, + { + "epoch": 0.5885314685314685, + "grad_norm": 0.3250773606725581, + "learning_rate": 1.9445645694355098e-05, + "loss": 0.06979280710220337, + "step": 1052 + }, + { + "epoch": 0.5890909090909091, + "grad_norm": 0.2995998589876639, + "learning_rate": 1.9443506827074274e-05, + "loss": 0.064674511551857, + "step": 1053 + }, + { + "epoch": 0.5896503496503497, + "grad_norm": 0.3261585494335681, + "learning_rate": 1.94413639596236e-05, + "loss": 0.11316889524459839, + "step": 1054 + }, + { + "epoch": 0.5902097902097903, + "grad_norm": 0.5428823770508131, + "learning_rate": 1.9439217092910795e-05, + "loss": 0.30373477935791016, + "step": 1055 + }, + { + "epoch": 0.5907692307692308, + "grad_norm": 0.5449002298158714, + "learning_rate": 1.943706622784523e-05, + "loss": 0.21477210521697998, + "step": 1056 + }, + { + "epoch": 0.5913286713286713, + "grad_norm": 0.45384775393669335, + "learning_rate": 1.9434911365337995e-05, + "loss": 0.15463876724243164, + "step": 1057 + }, + { + "epoch": 0.5918881118881119, + "grad_norm": 0.47443237016452816, + "learning_rate": 1.943275250630187e-05, + "loss": 0.24767793715000153, + "step": 1058 + }, + { + "epoch": 0.5924475524475524, + "grad_norm": 0.43068061144403197, + "learning_rate": 1.943058965165132e-05, + "loss": 0.18484212458133698, + "step": 1059 + }, + { + "epoch": 0.593006993006993, + "grad_norm": 0.5486009856654529, + "learning_rate": 1.9428422802302508e-05, + "loss": 0.29778236150741577, + "step": 1060 + }, + { + "epoch": 0.5935664335664336, + "grad_norm": 0.5478621092921356, + "learning_rate": 1.9426251959173293e-05, + "loss": 0.24156953394412994, + "step": 1061 + }, + { + "epoch": 0.5941258741258741, + "grad_norm": 0.6019368098466933, + "learning_rate": 1.942407712318322e-05, + "loss": 0.1564061939716339, + "step": 1062 + }, + { + "epoch": 0.5946853146853147, + "grad_norm": 0.4648235351495186, + "learning_rate": 1.942189829525352e-05, + "loss": 0.3132760524749756, + "step": 1063 + }, + { + "epoch": 0.5952447552447553, + "grad_norm": 0.6198615587666126, + "learning_rate": 1.9419715476307132e-05, + "loss": 0.35332822799682617, + "step": 1064 + }, + { + "epoch": 0.5958041958041959, + "grad_norm": 0.6291069180020309, + "learning_rate": 1.9417528667268664e-05, + "loss": 0.4086112976074219, + "step": 1065 + }, + { + "epoch": 0.5963636363636363, + "grad_norm": 0.43207556549375165, + "learning_rate": 1.941533786906443e-05, + "loss": 0.11487323045730591, + "step": 1066 + }, + { + "epoch": 0.5969230769230769, + "grad_norm": 0.5058716235573876, + "learning_rate": 1.9413143082622433e-05, + "loss": 0.2311253547668457, + "step": 1067 + }, + { + "epoch": 0.5974825174825175, + "grad_norm": 0.5716374576454071, + "learning_rate": 1.9410944308872358e-05, + "loss": 0.3118816018104553, + "step": 1068 + }, + { + "epoch": 0.598041958041958, + "grad_norm": 0.43008745317697955, + "learning_rate": 1.9408741548745584e-05, + "loss": 0.15288785099983215, + "step": 1069 + }, + { + "epoch": 0.5986013986013986, + "grad_norm": 0.3243298017982702, + "learning_rate": 1.940653480317517e-05, + "loss": 0.08625251054763794, + "step": 1070 + }, + { + "epoch": 0.5991608391608392, + "grad_norm": 0.5219467928298533, + "learning_rate": 1.9404324073095883e-05, + "loss": 0.31451061367988586, + "step": 1071 + }, + { + "epoch": 0.5997202797202797, + "grad_norm": 1.2014578458466199, + "learning_rate": 1.940210935944415e-05, + "loss": 0.3542405962944031, + "step": 1072 + }, + { + "epoch": 0.6002797202797203, + "grad_norm": 0.5127821292020727, + "learning_rate": 1.9399890663158113e-05, + "loss": 0.18369606137275696, + "step": 1073 + }, + { + "epoch": 0.6008391608391609, + "grad_norm": 0.4595728402230988, + "learning_rate": 1.939766798517758e-05, + "loss": 0.19230841100215912, + "step": 1074 + }, + { + "epoch": 0.6013986013986014, + "grad_norm": 0.21881470252839627, + "learning_rate": 1.939544132644406e-05, + "loss": 0.06281548738479614, + "step": 1075 + }, + { + "epoch": 0.6019580419580419, + "grad_norm": 0.4075118231783373, + "learning_rate": 1.9393210687900735e-05, + "loss": 0.18373005092144012, + "step": 1076 + }, + { + "epoch": 0.6025174825174825, + "grad_norm": 0.5095776475586016, + "learning_rate": 1.9390976070492486e-05, + "loss": 0.18567317724227905, + "step": 1077 + }, + { + "epoch": 0.6030769230769231, + "grad_norm": 0.28257397970255727, + "learning_rate": 1.9388737475165874e-05, + "loss": 0.05393151938915253, + "step": 1078 + }, + { + "epoch": 0.6036363636363636, + "grad_norm": 0.3780427066162853, + "learning_rate": 1.938649490286914e-05, + "loss": 0.11506589502096176, + "step": 1079 + }, + { + "epoch": 0.6041958041958042, + "grad_norm": 0.39845362970112114, + "learning_rate": 1.9384248354552214e-05, + "loss": 0.142998605966568, + "step": 1080 + }, + { + "epoch": 0.6047552447552448, + "grad_norm": 0.4793111819884572, + "learning_rate": 1.9381997831166716e-05, + "loss": 0.21724161505699158, + "step": 1081 + }, + { + "epoch": 0.6053146853146854, + "grad_norm": 0.3007942623953429, + "learning_rate": 1.9379743333665937e-05, + "loss": 0.09619180113077164, + "step": 1082 + }, + { + "epoch": 0.6058741258741259, + "grad_norm": 0.3542238953529059, + "learning_rate": 1.9377484863004863e-05, + "loss": 0.12282408028841019, + "step": 1083 + }, + { + "epoch": 0.6064335664335664, + "grad_norm": 0.5843404588901209, + "learning_rate": 1.9375222420140157e-05, + "loss": 0.32464998960494995, + "step": 1084 + }, + { + "epoch": 0.606993006993007, + "grad_norm": 0.472365709611251, + "learning_rate": 1.9372956006030162e-05, + "loss": 0.21114413440227509, + "step": 1085 + }, + { + "epoch": 0.6075524475524475, + "grad_norm": 0.25996729514109257, + "learning_rate": 1.937068562163491e-05, + "loss": 0.047926660627126694, + "step": 1086 + }, + { + "epoch": 0.6081118881118881, + "grad_norm": 0.48615623591967816, + "learning_rate": 1.9368411267916117e-05, + "loss": 0.3124011158943176, + "step": 1087 + }, + { + "epoch": 0.6086713286713287, + "grad_norm": 0.5749509927500399, + "learning_rate": 1.936613294583717e-05, + "loss": 0.3309420943260193, + "step": 1088 + }, + { + "epoch": 0.6092307692307692, + "grad_norm": 0.6021895428271564, + "learning_rate": 1.936385065636314e-05, + "loss": 0.3941347002983093, + "step": 1089 + }, + { + "epoch": 0.6097902097902098, + "grad_norm": 0.5911937340289877, + "learning_rate": 1.936156440046079e-05, + "loss": 0.25723814964294434, + "step": 1090 + }, + { + "epoch": 0.6103496503496504, + "grad_norm": 0.4300427351987886, + "learning_rate": 1.935927417909854e-05, + "loss": 0.27008870244026184, + "step": 1091 + }, + { + "epoch": 0.610909090909091, + "grad_norm": 0.5899906508471424, + "learning_rate": 1.9356979993246515e-05, + "loss": 0.2866484522819519, + "step": 1092 + }, + { + "epoch": 0.6114685314685314, + "grad_norm": 0.5110557388145952, + "learning_rate": 1.9354681843876504e-05, + "loss": 0.2076454907655716, + "step": 1093 + }, + { + "epoch": 0.612027972027972, + "grad_norm": 0.6116222499727213, + "learning_rate": 1.935237973196198e-05, + "loss": 0.3817736804485321, + "step": 1094 + }, + { + "epoch": 0.6125874125874126, + "grad_norm": 0.6490835457157362, + "learning_rate": 1.935007365847809e-05, + "loss": 0.43747881054878235, + "step": 1095 + }, + { + "epoch": 0.6131468531468531, + "grad_norm": 0.5846704711769108, + "learning_rate": 1.9347763624401664e-05, + "loss": 0.2928038239479065, + "step": 1096 + }, + { + "epoch": 0.6137062937062937, + "grad_norm": 0.4418310190593497, + "learning_rate": 1.934544963071121e-05, + "loss": 0.2194501757621765, + "step": 1097 + }, + { + "epoch": 0.6142657342657343, + "grad_norm": 0.5695508700954969, + "learning_rate": 1.9343131678386908e-05, + "loss": 0.337615430355072, + "step": 1098 + }, + { + "epoch": 0.6148251748251748, + "grad_norm": 0.5877027173258075, + "learning_rate": 1.934080976841062e-05, + "loss": 0.2631508708000183, + "step": 1099 + }, + { + "epoch": 0.6153846153846154, + "grad_norm": 0.5487504183202362, + "learning_rate": 1.9338483901765887e-05, + "loss": 0.22559913992881775, + "step": 1100 + }, + { + "epoch": 0.615944055944056, + "grad_norm": 0.33917764641649967, + "learning_rate": 1.9336154079437915e-05, + "loss": 0.043556440621614456, + "step": 1101 + }, + { + "epoch": 0.6165034965034965, + "grad_norm": 0.4883766738328329, + "learning_rate": 1.9333820302413594e-05, + "loss": 0.26283159852027893, + "step": 1102 + }, + { + "epoch": 0.617062937062937, + "grad_norm": 0.41563887690627627, + "learning_rate": 1.933148257168149e-05, + "loss": 0.19678513705730438, + "step": 1103 + }, + { + "epoch": 0.6176223776223776, + "grad_norm": 0.68734023454053, + "learning_rate": 1.9329140888231834e-05, + "loss": 0.49419569969177246, + "step": 1104 + }, + { + "epoch": 0.6181818181818182, + "grad_norm": 0.49536945715999703, + "learning_rate": 1.9326795253056543e-05, + "loss": 0.22792497277259827, + "step": 1105 + }, + { + "epoch": 0.6187412587412587, + "grad_norm": 0.5380609593483779, + "learning_rate": 1.9324445667149208e-05, + "loss": 0.5005597472190857, + "step": 1106 + }, + { + "epoch": 0.6193006993006993, + "grad_norm": 0.49562764398780523, + "learning_rate": 1.9322092131505084e-05, + "loss": 0.1787535697221756, + "step": 1107 + }, + { + "epoch": 0.6198601398601399, + "grad_norm": 0.45959038903950394, + "learning_rate": 1.9319734647121104e-05, + "loss": 0.26390495896339417, + "step": 1108 + }, + { + "epoch": 0.6204195804195805, + "grad_norm": 0.4492023457070527, + "learning_rate": 1.9317373214995873e-05, + "loss": 0.2035454958677292, + "step": 1109 + }, + { + "epoch": 0.620979020979021, + "grad_norm": 0.5015753782141327, + "learning_rate": 1.9315007836129664e-05, + "loss": 0.25793570280075073, + "step": 1110 + }, + { + "epoch": 0.6215384615384615, + "grad_norm": 0.3755843809856276, + "learning_rate": 1.9312638511524436e-05, + "loss": 0.12532973289489746, + "step": 1111 + }, + { + "epoch": 0.6220979020979021, + "grad_norm": 0.7098455666960787, + "learning_rate": 1.931026524218381e-05, + "loss": 0.35972028970718384, + "step": 1112 + }, + { + "epoch": 0.6226573426573426, + "grad_norm": 0.32886473852213655, + "learning_rate": 1.930788802911307e-05, + "loss": 0.09086612612009048, + "step": 1113 + }, + { + "epoch": 0.6232167832167832, + "grad_norm": 0.5766217310632286, + "learning_rate": 1.930550687331918e-05, + "loss": 0.3678678870201111, + "step": 1114 + }, + { + "epoch": 0.6237762237762238, + "grad_norm": 0.3703812920046654, + "learning_rate": 1.9303121775810777e-05, + "loss": 0.12469181418418884, + "step": 1115 + }, + { + "epoch": 0.6243356643356643, + "grad_norm": 0.5765092163113458, + "learning_rate": 1.930073273759816e-05, + "loss": 0.31603944301605225, + "step": 1116 + }, + { + "epoch": 0.6248951048951049, + "grad_norm": 0.5025584540623939, + "learning_rate": 1.9298339759693304e-05, + "loss": 0.2397618293762207, + "step": 1117 + }, + { + "epoch": 0.6254545454545455, + "grad_norm": 0.6873264764291395, + "learning_rate": 1.9295942843109844e-05, + "loss": 0.3813819885253906, + "step": 1118 + }, + { + "epoch": 0.6260139860139861, + "grad_norm": 0.4875671261707491, + "learning_rate": 1.9293541988863088e-05, + "loss": 0.18809424340724945, + "step": 1119 + }, + { + "epoch": 0.6265734265734266, + "grad_norm": 0.39843608379220036, + "learning_rate": 1.9291137197970018e-05, + "loss": 0.17729797959327698, + "step": 1120 + }, + { + "epoch": 0.6271328671328671, + "grad_norm": 0.44987570883456535, + "learning_rate": 1.9288728471449272e-05, + "loss": 0.18533581495285034, + "step": 1121 + }, + { + "epoch": 0.6276923076923077, + "grad_norm": 0.4023611678221286, + "learning_rate": 1.928631581032117e-05, + "loss": 0.13926270604133606, + "step": 1122 + }, + { + "epoch": 0.6282517482517482, + "grad_norm": 0.38518965671558153, + "learning_rate": 1.9283899215607684e-05, + "loss": 0.12458394467830658, + "step": 1123 + }, + { + "epoch": 0.6288111888111888, + "grad_norm": 0.6046923825353625, + "learning_rate": 1.9281478688332456e-05, + "loss": 0.3207188546657562, + "step": 1124 + }, + { + "epoch": 0.6293706293706294, + "grad_norm": 0.45871967559870747, + "learning_rate": 1.9279054229520796e-05, + "loss": 0.20889319479465485, + "step": 1125 + }, + { + "epoch": 0.62993006993007, + "grad_norm": 0.3574934625913777, + "learning_rate": 1.9276625840199685e-05, + "loss": 0.11756350845098495, + "step": 1126 + }, + { + "epoch": 0.6304895104895105, + "grad_norm": 0.7249889694760484, + "learning_rate": 1.9274193521397758e-05, + "loss": 0.3940296173095703, + "step": 1127 + }, + { + "epoch": 0.6310489510489511, + "grad_norm": 0.9243156965394563, + "learning_rate": 1.927175727414532e-05, + "loss": 0.2546404004096985, + "step": 1128 + }, + { + "epoch": 0.6316083916083917, + "grad_norm": 0.5033061485363374, + "learning_rate": 1.926931709947434e-05, + "loss": 0.3997475802898407, + "step": 1129 + }, + { + "epoch": 0.6321678321678321, + "grad_norm": 0.5213421348308778, + "learning_rate": 1.9266872998418454e-05, + "loss": 0.17789559066295624, + "step": 1130 + }, + { + "epoch": 0.6327272727272727, + "grad_norm": 0.42634403275055444, + "learning_rate": 1.926442497201295e-05, + "loss": 0.16646352410316467, + "step": 1131 + }, + { + "epoch": 0.6332867132867133, + "grad_norm": 0.4809180049979708, + "learning_rate": 1.9261973021294797e-05, + "loss": 0.17646440863609314, + "step": 1132 + }, + { + "epoch": 0.6338461538461538, + "grad_norm": 0.3477682952636734, + "learning_rate": 1.92595171473026e-05, + "loss": 0.11284934729337692, + "step": 1133 + }, + { + "epoch": 0.6344055944055944, + "grad_norm": 0.6499961526813964, + "learning_rate": 1.925705735107666e-05, + "loss": 0.34719952940940857, + "step": 1134 + }, + { + "epoch": 0.634965034965035, + "grad_norm": 0.3958378793396467, + "learning_rate": 1.92545936336589e-05, + "loss": 0.21634995937347412, + "step": 1135 + }, + { + "epoch": 0.6355244755244756, + "grad_norm": 0.38470233863736314, + "learning_rate": 1.9252125996092945e-05, + "loss": 0.17749923467636108, + "step": 1136 + }, + { + "epoch": 0.6360839160839161, + "grad_norm": 0.43928587077397807, + "learning_rate": 1.9249654439424044e-05, + "loss": 0.19224539399147034, + "step": 1137 + }, + { + "epoch": 0.6366433566433567, + "grad_norm": 0.40581842697620807, + "learning_rate": 1.9247178964699133e-05, + "loss": 0.15738870203495026, + "step": 1138 + }, + { + "epoch": 0.6372027972027972, + "grad_norm": 0.5742816210736953, + "learning_rate": 1.924469957296679e-05, + "loss": 0.32128435373306274, + "step": 1139 + }, + { + "epoch": 0.6377622377622377, + "grad_norm": 0.6579005681328847, + "learning_rate": 1.9242216265277265e-05, + "loss": 0.47951018810272217, + "step": 1140 + }, + { + "epoch": 0.6383216783216783, + "grad_norm": 0.6459044007366261, + "learning_rate": 1.923972904268246e-05, + "loss": 0.3300664722919464, + "step": 1141 + }, + { + "epoch": 0.6388811188811189, + "grad_norm": 0.4795681803539708, + "learning_rate": 1.9237237906235933e-05, + "loss": 0.17956653237342834, + "step": 1142 + }, + { + "epoch": 0.6394405594405594, + "grad_norm": 0.3869924538509202, + "learning_rate": 1.92347428569929e-05, + "loss": 0.13673678040504456, + "step": 1143 + }, + { + "epoch": 0.64, + "grad_norm": 0.572182652008674, + "learning_rate": 1.923224389601025e-05, + "loss": 0.22357279062271118, + "step": 1144 + }, + { + "epoch": 0.6405594405594406, + "grad_norm": 0.4102449006822939, + "learning_rate": 1.92297410243465e-05, + "loss": 0.18866273760795593, + "step": 1145 + }, + { + "epoch": 0.6411188811188812, + "grad_norm": 0.6603465233772944, + "learning_rate": 1.9227234243061855e-05, + "loss": 0.3767423629760742, + "step": 1146 + }, + { + "epoch": 0.6416783216783217, + "grad_norm": 0.47218446628891747, + "learning_rate": 1.9224723553218152e-05, + "loss": 0.2464519441127777, + "step": 1147 + }, + { + "epoch": 0.6422377622377622, + "grad_norm": 0.3979259610355061, + "learning_rate": 1.9222208955878896e-05, + "loss": 0.12272395193576813, + "step": 1148 + }, + { + "epoch": 0.6427972027972028, + "grad_norm": 0.28856026186694556, + "learning_rate": 1.9219690452109244e-05, + "loss": 0.09784625470638275, + "step": 1149 + }, + { + "epoch": 0.6433566433566433, + "grad_norm": 0.25826883320998606, + "learning_rate": 1.9217168042976008e-05, + "loss": 0.08572151511907578, + "step": 1150 + }, + { + "epoch": 0.6439160839160839, + "grad_norm": 0.49616652341859646, + "learning_rate": 1.9214641729547647e-05, + "loss": 0.18448996543884277, + "step": 1151 + }, + { + "epoch": 0.6444755244755245, + "grad_norm": 0.42131056138443906, + "learning_rate": 1.921211151289429e-05, + "loss": 0.15158987045288086, + "step": 1152 + }, + { + "epoch": 0.645034965034965, + "grad_norm": 0.324322621180921, + "learning_rate": 1.920957739408771e-05, + "loss": 0.1011272668838501, + "step": 1153 + }, + { + "epoch": 0.6455944055944056, + "grad_norm": 0.5347488707466375, + "learning_rate": 1.920703937420133e-05, + "loss": 0.2921428382396698, + "step": 1154 + }, + { + "epoch": 0.6461538461538462, + "grad_norm": 0.8364561359237235, + "learning_rate": 1.920449745431022e-05, + "loss": 0.48127010464668274, + "step": 1155 + }, + { + "epoch": 0.6467132867132868, + "grad_norm": 0.4505224819113675, + "learning_rate": 1.9201951635491123e-05, + "loss": 0.225082665681839, + "step": 1156 + }, + { + "epoch": 0.6472727272727272, + "grad_norm": 0.5512692398758167, + "learning_rate": 1.919940191882242e-05, + "loss": 0.2524714171886444, + "step": 1157 + }, + { + "epoch": 0.6478321678321678, + "grad_norm": 0.494612276742012, + "learning_rate": 1.9196848305384136e-05, + "loss": 0.24488875269889832, + "step": 1158 + }, + { + "epoch": 0.6483916083916084, + "grad_norm": 0.5481317261566628, + "learning_rate": 1.9194290796257963e-05, + "loss": 0.29227587580680847, + "step": 1159 + }, + { + "epoch": 0.6489510489510489, + "grad_norm": 0.5088271230393152, + "learning_rate": 1.919172939252723e-05, + "loss": 0.2693462371826172, + "step": 1160 + }, + { + "epoch": 0.6495104895104895, + "grad_norm": 0.5037268715417473, + "learning_rate": 1.9189164095276916e-05, + "loss": 0.18121157586574554, + "step": 1161 + }, + { + "epoch": 0.6500699300699301, + "grad_norm": 0.4515603003547127, + "learning_rate": 1.9186594905593665e-05, + "loss": 0.18202300369739532, + "step": 1162 + }, + { + "epoch": 0.6506293706293707, + "grad_norm": 0.5286721739350488, + "learning_rate": 1.9184021824565753e-05, + "loss": 0.29729363322257996, + "step": 1163 + }, + { + "epoch": 0.6511888111888112, + "grad_norm": 0.5849522299891308, + "learning_rate": 1.918144485328311e-05, + "loss": 0.3485714793205261, + "step": 1164 + }, + { + "epoch": 0.6517482517482518, + "grad_norm": 0.5593564311642616, + "learning_rate": 1.917886399283731e-05, + "loss": 0.248091921210289, + "step": 1165 + }, + { + "epoch": 0.6523076923076923, + "grad_norm": 0.528396657157791, + "learning_rate": 1.917627924432159e-05, + "loss": 0.2711426019668579, + "step": 1166 + }, + { + "epoch": 0.6528671328671328, + "grad_norm": 0.426092944121079, + "learning_rate": 1.917369060883081e-05, + "loss": 0.10632821917533875, + "step": 1167 + }, + { + "epoch": 0.6534265734265734, + "grad_norm": 0.3360521774165836, + "learning_rate": 1.9171098087461496e-05, + "loss": 0.11299716681241989, + "step": 1168 + }, + { + "epoch": 0.653986013986014, + "grad_norm": 0.6528684405208269, + "learning_rate": 1.9168501681311806e-05, + "loss": 0.39953672885894775, + "step": 1169 + }, + { + "epoch": 0.6545454545454545, + "grad_norm": 0.5271138359665549, + "learning_rate": 1.9165901391481557e-05, + "loss": 0.26359349489212036, + "step": 1170 + }, + { + "epoch": 0.6551048951048951, + "grad_norm": 0.361710650142558, + "learning_rate": 1.9163297219072195e-05, + "loss": 0.07769016176462173, + "step": 1171 + }, + { + "epoch": 0.6556643356643357, + "grad_norm": 0.7473826960588661, + "learning_rate": 1.9160689165186835e-05, + "loss": 0.18297234177589417, + "step": 1172 + }, + { + "epoch": 0.6562237762237763, + "grad_norm": 0.40334626104256904, + "learning_rate": 1.9158077230930207e-05, + "loss": 0.1556541621685028, + "step": 1173 + }, + { + "epoch": 0.6567832167832168, + "grad_norm": 0.5733742975504827, + "learning_rate": 1.915546141740871e-05, + "loss": 0.2870713174343109, + "step": 1174 + }, + { + "epoch": 0.6573426573426573, + "grad_norm": 0.5177589376724633, + "learning_rate": 1.915284172573036e-05, + "loss": 0.29671698808670044, + "step": 1175 + }, + { + "epoch": 0.6579020979020979, + "grad_norm": 0.5348988171765273, + "learning_rate": 1.915021815700485e-05, + "loss": 0.26493242383003235, + "step": 1176 + }, + { + "epoch": 0.6584615384615384, + "grad_norm": 0.584485300114437, + "learning_rate": 1.914759071234348e-05, + "loss": 0.22414883971214294, + "step": 1177 + }, + { + "epoch": 0.659020979020979, + "grad_norm": 0.39729505343213933, + "learning_rate": 1.9144959392859216e-05, + "loss": 0.1610300987958908, + "step": 1178 + }, + { + "epoch": 0.6595804195804196, + "grad_norm": 0.3880536340107255, + "learning_rate": 1.9142324199666652e-05, + "loss": 0.14239132404327393, + "step": 1179 + }, + { + "epoch": 0.6601398601398601, + "grad_norm": 0.5342701969212307, + "learning_rate": 1.9139685133882036e-05, + "loss": 0.31461817026138306, + "step": 1180 + }, + { + "epoch": 0.6606993006993007, + "grad_norm": 0.43808940140856767, + "learning_rate": 1.913704219662324e-05, + "loss": 0.20869994163513184, + "step": 1181 + }, + { + "epoch": 0.6612587412587413, + "grad_norm": 0.4764339533325639, + "learning_rate": 1.913439538900979e-05, + "loss": 0.2910383641719818, + "step": 1182 + }, + { + "epoch": 0.6618181818181819, + "grad_norm": 0.3532473312145425, + "learning_rate": 1.913174471216284e-05, + "loss": 0.08348032087087631, + "step": 1183 + }, + { + "epoch": 0.6623776223776224, + "grad_norm": 0.5436329309351365, + "learning_rate": 1.9129090167205197e-05, + "loss": 0.30902546644210815, + "step": 1184 + }, + { + "epoch": 0.6629370629370629, + "grad_norm": 0.511802719478397, + "learning_rate": 1.912643175526129e-05, + "loss": 0.14539474248886108, + "step": 1185 + }, + { + "epoch": 0.6634965034965035, + "grad_norm": 0.5514819036236864, + "learning_rate": 1.91237694774572e-05, + "loss": 0.2426593154668808, + "step": 1186 + }, + { + "epoch": 0.664055944055944, + "grad_norm": 0.4901204230776472, + "learning_rate": 1.9121103334920636e-05, + "loss": 0.2552768588066101, + "step": 1187 + }, + { + "epoch": 0.6646153846153846, + "grad_norm": 0.5451523171152032, + "learning_rate": 1.911843332878095e-05, + "loss": 0.2865668535232544, + "step": 1188 + }, + { + "epoch": 0.6651748251748252, + "grad_norm": 0.5475525469224622, + "learning_rate": 1.911575946016913e-05, + "loss": 0.2804175913333893, + "step": 1189 + }, + { + "epoch": 0.6657342657342658, + "grad_norm": 1.127080608286401, + "learning_rate": 1.9113081730217793e-05, + "loss": 0.10971076786518097, + "step": 1190 + }, + { + "epoch": 0.6662937062937063, + "grad_norm": 0.4193587813806797, + "learning_rate": 1.9110400140061203e-05, + "loss": 0.1628573089838028, + "step": 1191 + }, + { + "epoch": 0.6668531468531469, + "grad_norm": 0.3128994400655092, + "learning_rate": 1.9107714690835252e-05, + "loss": 0.05909005552530289, + "step": 1192 + }, + { + "epoch": 0.6674125874125875, + "grad_norm": 0.3316119401337039, + "learning_rate": 1.9105025383677464e-05, + "loss": 0.09491445869207382, + "step": 1193 + }, + { + "epoch": 0.6679720279720279, + "grad_norm": 0.412086261475146, + "learning_rate": 1.9102332219727003e-05, + "loss": 0.11123067140579224, + "step": 1194 + }, + { + "epoch": 0.6685314685314685, + "grad_norm": 0.3760900000219321, + "learning_rate": 1.9099635200124668e-05, + "loss": 0.13746190071105957, + "step": 1195 + }, + { + "epoch": 0.6690909090909091, + "grad_norm": 0.30799540271293874, + "learning_rate": 1.9096934326012885e-05, + "loss": 0.12544505298137665, + "step": 1196 + }, + { + "epoch": 0.6696503496503496, + "grad_norm": 0.42732449059915295, + "learning_rate": 1.9094229598535717e-05, + "loss": 0.18139979243278503, + "step": 1197 + }, + { + "epoch": 0.6702097902097902, + "grad_norm": 0.46994286046468253, + "learning_rate": 1.9091521018838856e-05, + "loss": 0.18799258768558502, + "step": 1198 + }, + { + "epoch": 0.6707692307692308, + "grad_norm": 0.4145370773784259, + "learning_rate": 1.9088808588069633e-05, + "loss": 0.22460326552391052, + "step": 1199 + }, + { + "epoch": 0.6713286713286714, + "grad_norm": 1.5928461684315904, + "learning_rate": 1.9086092307377e-05, + "loss": 0.356277197599411, + "step": 1200 + }, + { + "epoch": 0.6718881118881119, + "grad_norm": 0.711716495164034, + "learning_rate": 1.9083372177911543e-05, + "loss": 0.2535674571990967, + "step": 1201 + }, + { + "epoch": 0.6724475524475525, + "grad_norm": 0.46111872746092186, + "learning_rate": 1.9080648200825487e-05, + "loss": 0.20889690518379211, + "step": 1202 + }, + { + "epoch": 0.673006993006993, + "grad_norm": 0.42875728676019464, + "learning_rate": 1.9077920377272673e-05, + "loss": 0.16600444912910461, + "step": 1203 + }, + { + "epoch": 0.6735664335664335, + "grad_norm": 0.5142087153104058, + "learning_rate": 1.9075188708408586e-05, + "loss": 0.2649494707584381, + "step": 1204 + }, + { + "epoch": 0.6741258741258741, + "grad_norm": 0.5158542717575485, + "learning_rate": 1.9072453195390327e-05, + "loss": 0.09610937535762787, + "step": 1205 + }, + { + "epoch": 0.6746853146853147, + "grad_norm": 0.41971830272515437, + "learning_rate": 1.906971383937663e-05, + "loss": 0.15152233839035034, + "step": 1206 + }, + { + "epoch": 0.6752447552447552, + "grad_norm": 0.38495509680517775, + "learning_rate": 1.9066970641527857e-05, + "loss": 0.27731889486312866, + "step": 1207 + }, + { + "epoch": 0.6758041958041958, + "grad_norm": 0.5563406855603696, + "learning_rate": 1.9064223603006e-05, + "loss": 0.3000246286392212, + "step": 1208 + }, + { + "epoch": 0.6763636363636364, + "grad_norm": 0.3462027431227573, + "learning_rate": 1.9061472724974677e-05, + "loss": 0.145631343126297, + "step": 1209 + }, + { + "epoch": 0.676923076923077, + "grad_norm": 0.4811425445142987, + "learning_rate": 1.9058718008599126e-05, + "loss": 0.257687509059906, + "step": 1210 + }, + { + "epoch": 0.6774825174825175, + "grad_norm": 0.369973056463325, + "learning_rate": 1.9055959455046218e-05, + "loss": 0.08178170025348663, + "step": 1211 + }, + { + "epoch": 0.678041958041958, + "grad_norm": 0.3225510539131323, + "learning_rate": 1.9053197065484447e-05, + "loss": 0.11994513124227524, + "step": 1212 + }, + { + "epoch": 0.6786013986013986, + "grad_norm": 0.05254167105377698, + "learning_rate": 1.905043084108394e-05, + "loss": 0.0022237920202314854, + "step": 1213 + }, + { + "epoch": 0.6791608391608391, + "grad_norm": 0.5675275527981017, + "learning_rate": 1.9047660783016424e-05, + "loss": 0.44033315777778625, + "step": 1214 + }, + { + "epoch": 0.6797202797202797, + "grad_norm": 0.3826190490552237, + "learning_rate": 1.904488689245528e-05, + "loss": 0.09359294921159744, + "step": 1215 + }, + { + "epoch": 0.6802797202797203, + "grad_norm": 0.42364934581382085, + "learning_rate": 1.904210917057549e-05, + "loss": 0.14086344838142395, + "step": 1216 + }, + { + "epoch": 0.6808391608391609, + "grad_norm": 0.2643934873120073, + "learning_rate": 1.9039327618553677e-05, + "loss": 0.05836724489927292, + "step": 1217 + }, + { + "epoch": 0.6813986013986014, + "grad_norm": 0.5072853967656477, + "learning_rate": 1.903654223756807e-05, + "loss": 0.2508912980556488, + "step": 1218 + }, + { + "epoch": 0.681958041958042, + "grad_norm": 0.4899183703256428, + "learning_rate": 1.903375302879853e-05, + "loss": 0.22959785163402557, + "step": 1219 + }, + { + "epoch": 0.6825174825174826, + "grad_norm": 0.5013666364892235, + "learning_rate": 1.903095999342653e-05, + "loss": 0.31373026967048645, + "step": 1220 + }, + { + "epoch": 0.683076923076923, + "grad_norm": 0.4756619903188653, + "learning_rate": 1.9028163132635182e-05, + "loss": 0.25159138441085815, + "step": 1221 + }, + { + "epoch": 0.6836363636363636, + "grad_norm": 0.33198590705659997, + "learning_rate": 1.90253624476092e-05, + "loss": 0.07931622862815857, + "step": 1222 + }, + { + "epoch": 0.6841958041958042, + "grad_norm": 0.4615134785715342, + "learning_rate": 1.9022557939534924e-05, + "loss": 0.18305863440036774, + "step": 1223 + }, + { + "epoch": 0.6847552447552447, + "grad_norm": 0.5059481238920129, + "learning_rate": 1.9019749609600313e-05, + "loss": 0.24432870745658875, + "step": 1224 + }, + { + "epoch": 0.6853146853146853, + "grad_norm": 0.3752575091128949, + "learning_rate": 1.9016937458994952e-05, + "loss": 0.12469576299190521, + "step": 1225 + }, + { + "epoch": 0.6858741258741259, + "grad_norm": 0.36934346256848305, + "learning_rate": 1.9014121488910028e-05, + "loss": 0.2719348967075348, + "step": 1226 + }, + { + "epoch": 0.6864335664335665, + "grad_norm": 0.4719589181427845, + "learning_rate": 1.9011301700538367e-05, + "loss": 0.27376121282577515, + "step": 1227 + }, + { + "epoch": 0.686993006993007, + "grad_norm": 0.625595234479326, + "learning_rate": 1.9008478095074396e-05, + "loss": 0.3813643455505371, + "step": 1228 + }, + { + "epoch": 0.6875524475524476, + "grad_norm": 0.49535816439093255, + "learning_rate": 1.9005650673714163e-05, + "loss": 0.21032418310642242, + "step": 1229 + }, + { + "epoch": 0.6881118881118881, + "grad_norm": 0.5355564020044065, + "learning_rate": 1.9002819437655343e-05, + "loss": 0.1845524162054062, + "step": 1230 + }, + { + "epoch": 0.6886713286713286, + "grad_norm": 0.25783205275457804, + "learning_rate": 1.899998438809721e-05, + "loss": 0.048418037593364716, + "step": 1231 + }, + { + "epoch": 0.6892307692307692, + "grad_norm": 0.4540063037391814, + "learning_rate": 1.8997145526240656e-05, + "loss": 0.13670310378074646, + "step": 1232 + }, + { + "epoch": 0.6897902097902098, + "grad_norm": 0.47824736903942516, + "learning_rate": 1.8994302853288205e-05, + "loss": 0.14143714308738708, + "step": 1233 + }, + { + "epoch": 0.6903496503496503, + "grad_norm": 0.1953521265366616, + "learning_rate": 1.8991456370443973e-05, + "loss": 0.03789675235748291, + "step": 1234 + }, + { + "epoch": 0.6909090909090909, + "grad_norm": 0.42743366564759566, + "learning_rate": 1.898860607891371e-05, + "loss": 0.152780219912529, + "step": 1235 + }, + { + "epoch": 0.6914685314685315, + "grad_norm": 0.3474010204513338, + "learning_rate": 1.898575197990476e-05, + "loss": 0.1169387623667717, + "step": 1236 + }, + { + "epoch": 0.6920279720279721, + "grad_norm": 0.39541233166716894, + "learning_rate": 1.8982894074626094e-05, + "loss": 0.15557576715946198, + "step": 1237 + }, + { + "epoch": 0.6925874125874126, + "grad_norm": 0.36731634119967316, + "learning_rate": 1.8980032364288286e-05, + "loss": 0.14882495999336243, + "step": 1238 + }, + { + "epoch": 0.6931468531468531, + "grad_norm": 0.525537291230844, + "learning_rate": 1.8977166850103535e-05, + "loss": 0.2935725450515747, + "step": 1239 + }, + { + "epoch": 0.6937062937062937, + "grad_norm": 0.49380702474445654, + "learning_rate": 1.8974297533285636e-05, + "loss": 0.25928008556365967, + "step": 1240 + }, + { + "epoch": 0.6942657342657342, + "grad_norm": 0.4989406834620033, + "learning_rate": 1.8971424415050005e-05, + "loss": 0.2947520613670349, + "step": 1241 + }, + { + "epoch": 0.6948251748251748, + "grad_norm": 0.3813848676951959, + "learning_rate": 1.8968547496613657e-05, + "loss": 0.13924340903759003, + "step": 1242 + }, + { + "epoch": 0.6953846153846154, + "grad_norm": 0.438090582280127, + "learning_rate": 1.8965666779195235e-05, + "loss": 0.22265732288360596, + "step": 1243 + }, + { + "epoch": 0.695944055944056, + "grad_norm": 0.38136582482183706, + "learning_rate": 1.896278226401497e-05, + "loss": 0.3004734516143799, + "step": 1244 + }, + { + "epoch": 0.6965034965034965, + "grad_norm": 0.493040783506876, + "learning_rate": 1.895989395229472e-05, + "loss": 0.2936972975730896, + "step": 1245 + }, + { + "epoch": 0.6970629370629371, + "grad_norm": 0.49141436120446186, + "learning_rate": 1.8957001845257947e-05, + "loss": 0.1826832890510559, + "step": 1246 + }, + { + "epoch": 0.6976223776223777, + "grad_norm": 0.4442121656323239, + "learning_rate": 1.8954105944129702e-05, + "loss": 0.15641947090625763, + "step": 1247 + }, + { + "epoch": 0.6981818181818182, + "grad_norm": 0.43052240625335236, + "learning_rate": 1.895120625013667e-05, + "loss": 0.22361494600772858, + "step": 1248 + }, + { + "epoch": 0.6987412587412587, + "grad_norm": 0.32563813899896205, + "learning_rate": 1.894830276450713e-05, + "loss": 0.08146579563617706, + "step": 1249 + }, + { + "epoch": 0.6993006993006993, + "grad_norm": 0.4119214510041244, + "learning_rate": 1.894539548847096e-05, + "loss": 0.16098377108573914, + "step": 1250 + }, + { + "epoch": 0.6998601398601398, + "grad_norm": 0.30738472105934916, + "learning_rate": 1.8942484423259663e-05, + "loss": 0.04757336527109146, + "step": 1251 + }, + { + "epoch": 0.7004195804195804, + "grad_norm": 0.5688401136879419, + "learning_rate": 1.8939569570106326e-05, + "loss": 0.23201589286327362, + "step": 1252 + }, + { + "epoch": 0.700979020979021, + "grad_norm": 0.37687213999601543, + "learning_rate": 1.8936650930245655e-05, + "loss": 0.09793180972337723, + "step": 1253 + }, + { + "epoch": 0.7015384615384616, + "grad_norm": 0.37404122550711294, + "learning_rate": 1.8933728504913953e-05, + "loss": 0.15101146697998047, + "step": 1254 + }, + { + "epoch": 0.7020979020979021, + "grad_norm": 0.6480136528552453, + "learning_rate": 1.893080229534913e-05, + "loss": 0.25147831439971924, + "step": 1255 + }, + { + "epoch": 0.7026573426573427, + "grad_norm": 0.5226137889812511, + "learning_rate": 1.8927872302790696e-05, + "loss": 0.29706037044525146, + "step": 1256 + }, + { + "epoch": 0.7032167832167833, + "grad_norm": 0.4888047949629481, + "learning_rate": 1.8924938528479766e-05, + "loss": 0.20549073815345764, + "step": 1257 + }, + { + "epoch": 0.7037762237762237, + "grad_norm": 0.48999924250823396, + "learning_rate": 1.8922000973659056e-05, + "loss": 0.14198213815689087, + "step": 1258 + }, + { + "epoch": 0.7043356643356643, + "grad_norm": 0.49909125900979895, + "learning_rate": 1.8919059639572882e-05, + "loss": 0.23146186769008636, + "step": 1259 + }, + { + "epoch": 0.7048951048951049, + "grad_norm": 0.5558868477931278, + "learning_rate": 1.891611452746716e-05, + "loss": 0.3991163372993469, + "step": 1260 + }, + { + "epoch": 0.7054545454545454, + "grad_norm": 0.4961964354875628, + "learning_rate": 1.8913165638589416e-05, + "loss": 0.19554615020751953, + "step": 1261 + }, + { + "epoch": 0.706013986013986, + "grad_norm": 0.4060395894095321, + "learning_rate": 1.891021297418876e-05, + "loss": 0.29548728466033936, + "step": 1262 + }, + { + "epoch": 0.7065734265734266, + "grad_norm": 0.5149263175993853, + "learning_rate": 1.8907256535515914e-05, + "loss": 0.2255270779132843, + "step": 1263 + }, + { + "epoch": 0.7071328671328672, + "grad_norm": 0.5954965153402395, + "learning_rate": 1.8904296323823193e-05, + "loss": 0.37839800119400024, + "step": 1264 + }, + { + "epoch": 0.7076923076923077, + "grad_norm": 0.5604076906656097, + "learning_rate": 1.8901332340364512e-05, + "loss": 0.27291709184646606, + "step": 1265 + }, + { + "epoch": 0.7082517482517483, + "grad_norm": 0.5918150786104728, + "learning_rate": 1.8898364586395383e-05, + "loss": 0.3188212215900421, + "step": 1266 + }, + { + "epoch": 0.7088111888111888, + "grad_norm": 0.2834202323052936, + "learning_rate": 1.8895393063172916e-05, + "loss": 0.25226080417633057, + "step": 1267 + }, + { + "epoch": 0.7093706293706293, + "grad_norm": 0.36341050110156886, + "learning_rate": 1.8892417771955815e-05, + "loss": 0.154170960187912, + "step": 1268 + }, + { + "epoch": 0.7099300699300699, + "grad_norm": 0.5037061359510476, + "learning_rate": 1.888943871400438e-05, + "loss": 0.22951939702033997, + "step": 1269 + }, + { + "epoch": 0.7104895104895105, + "grad_norm": 0.5052215858230452, + "learning_rate": 1.8886455890580513e-05, + "loss": 0.1874401867389679, + "step": 1270 + }, + { + "epoch": 0.711048951048951, + "grad_norm": 0.5858963064869629, + "learning_rate": 1.8883469302947705e-05, + "loss": 0.4291166067123413, + "step": 1271 + }, + { + "epoch": 0.7116083916083916, + "grad_norm": 0.16649076522663528, + "learning_rate": 1.8880478952371042e-05, + "loss": 0.01633700728416443, + "step": 1272 + }, + { + "epoch": 0.7121678321678322, + "grad_norm": 0.5207778007370123, + "learning_rate": 1.8877484840117208e-05, + "loss": 0.2168062925338745, + "step": 1273 + }, + { + "epoch": 0.7127272727272728, + "grad_norm": 0.5671017623880911, + "learning_rate": 1.887448696745447e-05, + "loss": 0.31777673959732056, + "step": 1274 + }, + { + "epoch": 0.7132867132867133, + "grad_norm": 0.5043043830593669, + "learning_rate": 1.8871485335652705e-05, + "loss": 0.3538969159126282, + "step": 1275 + }, + { + "epoch": 0.7138461538461538, + "grad_norm": 0.2680147606175334, + "learning_rate": 1.8868479945983367e-05, + "loss": 0.06914018094539642, + "step": 1276 + }, + { + "epoch": 0.7144055944055944, + "grad_norm": 0.21842058692517274, + "learning_rate": 1.886547079971951e-05, + "loss": 0.039529550820589066, + "step": 1277 + }, + { + "epoch": 0.7149650349650349, + "grad_norm": 0.33217790741554737, + "learning_rate": 1.886245789813577e-05, + "loss": 0.10326577723026276, + "step": 1278 + }, + { + "epoch": 0.7155244755244755, + "grad_norm": 0.40893065352144237, + "learning_rate": 1.885944124250839e-05, + "loss": 0.25542083382606506, + "step": 1279 + }, + { + "epoch": 0.7160839160839161, + "grad_norm": 0.6950750017213755, + "learning_rate": 1.8856420834115182e-05, + "loss": 0.35049593448638916, + "step": 1280 + }, + { + "epoch": 0.7166433566433567, + "grad_norm": 0.4618703635611876, + "learning_rate": 1.885339667423557e-05, + "loss": 0.26107078790664673, + "step": 1281 + }, + { + "epoch": 0.7172027972027972, + "grad_norm": 0.4549342134021742, + "learning_rate": 1.8850368764150554e-05, + "loss": 0.1546381711959839, + "step": 1282 + }, + { + "epoch": 0.7177622377622378, + "grad_norm": 0.3364485928510549, + "learning_rate": 1.884733710514272e-05, + "loss": 0.13336928188800812, + "step": 1283 + }, + { + "epoch": 0.7183216783216784, + "grad_norm": 0.5091382751551491, + "learning_rate": 1.884430169849625e-05, + "loss": 0.38645532727241516, + "step": 1284 + }, + { + "epoch": 0.7188811188811188, + "grad_norm": 0.6868927834787496, + "learning_rate": 1.884126254549691e-05, + "loss": 0.5598582029342651, + "step": 1285 + }, + { + "epoch": 0.7194405594405594, + "grad_norm": 0.46457739496894224, + "learning_rate": 1.8838219647432053e-05, + "loss": 0.21529540419578552, + "step": 1286 + }, + { + "epoch": 0.72, + "grad_norm": 0.2863912995869663, + "learning_rate": 1.8835173005590616e-05, + "loss": 0.09676319360733032, + "step": 1287 + }, + { + "epoch": 0.7205594405594405, + "grad_norm": 0.48613662493776555, + "learning_rate": 1.883212262126313e-05, + "loss": 0.3112836480140686, + "step": 1288 + }, + { + "epoch": 0.7211188811188811, + "grad_norm": 0.5320959832186422, + "learning_rate": 1.88290684957417e-05, + "loss": 0.2938607931137085, + "step": 1289 + }, + { + "epoch": 0.7216783216783217, + "grad_norm": 0.4685117313718341, + "learning_rate": 1.8826010630320025e-05, + "loss": 0.230940580368042, + "step": 1290 + }, + { + "epoch": 0.7222377622377623, + "grad_norm": 0.5672234882276421, + "learning_rate": 1.8822949026293383e-05, + "loss": 0.3394428491592407, + "step": 1291 + }, + { + "epoch": 0.7227972027972028, + "grad_norm": 0.3624809055546799, + "learning_rate": 1.881988368495864e-05, + "loss": 0.21149572730064392, + "step": 1292 + }, + { + "epoch": 0.7233566433566434, + "grad_norm": 1.4630785102803525, + "learning_rate": 1.8816814607614234e-05, + "loss": 0.21099887788295746, + "step": 1293 + }, + { + "epoch": 0.7239160839160839, + "grad_norm": 0.3717657561561095, + "learning_rate": 1.8813741795560203e-05, + "loss": 0.16308756172657013, + "step": 1294 + }, + { + "epoch": 0.7244755244755244, + "grad_norm": 0.3042663478051724, + "learning_rate": 1.8810665250098153e-05, + "loss": 0.12714090943336487, + "step": 1295 + }, + { + "epoch": 0.725034965034965, + "grad_norm": 0.5414482044056161, + "learning_rate": 1.880758497253128e-05, + "loss": 0.25265365839004517, + "step": 1296 + }, + { + "epoch": 0.7255944055944056, + "grad_norm": 0.4390656545220634, + "learning_rate": 1.880450096416435e-05, + "loss": 0.13695551455020905, + "step": 1297 + }, + { + "epoch": 0.7261538461538461, + "grad_norm": 0.3015504581194357, + "learning_rate": 1.8801413226303722e-05, + "loss": 0.08582916855812073, + "step": 1298 + }, + { + "epoch": 0.7267132867132867, + "grad_norm": 0.37742620705416025, + "learning_rate": 1.879832176025733e-05, + "loss": 0.130242258310318, + "step": 1299 + }, + { + "epoch": 0.7272727272727273, + "grad_norm": 0.4448069743513974, + "learning_rate": 1.8795226567334683e-05, + "loss": 0.22602578997612, + "step": 1300 + }, + { + "epoch": 0.7278321678321679, + "grad_norm": 0.5507340756120065, + "learning_rate": 1.879212764884687e-05, + "loss": 0.3310128450393677, + "step": 1301 + }, + { + "epoch": 0.7283916083916084, + "grad_norm": 0.5654241564544463, + "learning_rate": 1.878902500610657e-05, + "loss": 0.17944839596748352, + "step": 1302 + }, + { + "epoch": 0.7289510489510489, + "grad_norm": 0.3748298665638449, + "learning_rate": 1.878591864042802e-05, + "loss": 0.12377354502677917, + "step": 1303 + }, + { + "epoch": 0.7295104895104895, + "grad_norm": 0.3755239188800472, + "learning_rate": 1.8782808553127044e-05, + "loss": 0.23134580254554749, + "step": 1304 + }, + { + "epoch": 0.73006993006993, + "grad_norm": 0.530734207986483, + "learning_rate": 1.8779694745521052e-05, + "loss": 0.38295817375183105, + "step": 1305 + }, + { + "epoch": 0.7306293706293706, + "grad_norm": 0.5079821597230926, + "learning_rate": 1.8776577218929007e-05, + "loss": 0.37209686636924744, + "step": 1306 + }, + { + "epoch": 0.7311888111888112, + "grad_norm": 0.3468187284822202, + "learning_rate": 1.877345597467147e-05, + "loss": 0.19278863072395325, + "step": 1307 + }, + { + "epoch": 0.7317482517482518, + "grad_norm": 0.49430601387334405, + "learning_rate": 1.8770331014070563e-05, + "loss": 0.29413846135139465, + "step": 1308 + }, + { + "epoch": 0.7323076923076923, + "grad_norm": 0.5290396815476958, + "learning_rate": 1.876720233844999e-05, + "loss": 0.33733540773391724, + "step": 1309 + }, + { + "epoch": 0.7328671328671329, + "grad_norm": 0.4162281546371152, + "learning_rate": 1.8764069949135018e-05, + "loss": 0.18579035997390747, + "step": 1310 + }, + { + "epoch": 0.7334265734265735, + "grad_norm": 0.4625865612148632, + "learning_rate": 1.87609338474525e-05, + "loss": 0.24365222454071045, + "step": 1311 + }, + { + "epoch": 0.733986013986014, + "grad_norm": 0.29260858475093254, + "learning_rate": 1.875779403473085e-05, + "loss": 0.046830154955387115, + "step": 1312 + }, + { + "epoch": 0.7345454545454545, + "grad_norm": 0.4030582385131079, + "learning_rate": 1.8754650512300067e-05, + "loss": 0.19351214170455933, + "step": 1313 + }, + { + "epoch": 0.7351048951048951, + "grad_norm": 0.2013771135560784, + "learning_rate": 1.8751503281491708e-05, + "loss": 0.03813890367746353, + "step": 1314 + }, + { + "epoch": 0.7356643356643356, + "grad_norm": 0.3436027416041538, + "learning_rate": 1.8748352343638912e-05, + "loss": 0.10832241177558899, + "step": 1315 + }, + { + "epoch": 0.7362237762237762, + "grad_norm": 0.5181843125333964, + "learning_rate": 1.8745197700076376e-05, + "loss": 0.2444397658109665, + "step": 1316 + }, + { + "epoch": 0.7367832167832168, + "grad_norm": 0.4205966576938173, + "learning_rate": 1.874203935214038e-05, + "loss": 0.2845626771450043, + "step": 1317 + }, + { + "epoch": 0.7373426573426574, + "grad_norm": 0.6562588614621905, + "learning_rate": 1.8738877301168766e-05, + "loss": 0.286207914352417, + "step": 1318 + }, + { + "epoch": 0.7379020979020979, + "grad_norm": 0.38431072186268045, + "learning_rate": 1.873571154850094e-05, + "loss": 0.18193387985229492, + "step": 1319 + }, + { + "epoch": 0.7384615384615385, + "grad_norm": 0.23766938670689472, + "learning_rate": 1.8732542095477888e-05, + "loss": 0.09196147322654724, + "step": 1320 + }, + { + "epoch": 0.7390209790209791, + "grad_norm": 0.33511835046084276, + "learning_rate": 1.8729368943442156e-05, + "loss": 0.12661802768707275, + "step": 1321 + }, + { + "epoch": 0.7395804195804195, + "grad_norm": 0.3830630601032016, + "learning_rate": 1.872619209373785e-05, + "loss": 0.14102940261363983, + "step": 1322 + }, + { + "epoch": 0.7401398601398601, + "grad_norm": 0.3912006413137731, + "learning_rate": 1.872301154771066e-05, + "loss": 0.1998492330312729, + "step": 1323 + }, + { + "epoch": 0.7406993006993007, + "grad_norm": 0.9078858852572724, + "learning_rate": 1.871982730670783e-05, + "loss": 0.2623226046562195, + "step": 1324 + }, + { + "epoch": 0.7412587412587412, + "grad_norm": 0.44697423076090625, + "learning_rate": 1.871663937207816e-05, + "loss": 0.2594923973083496, + "step": 1325 + }, + { + "epoch": 0.7418181818181818, + "grad_norm": 0.5751294548919422, + "learning_rate": 1.8713447745172043e-05, + "loss": 0.3071717619895935, + "step": 1326 + }, + { + "epoch": 0.7423776223776224, + "grad_norm": 0.4388516717159596, + "learning_rate": 1.8710252427341408e-05, + "loss": 0.2433147132396698, + "step": 1327 + }, + { + "epoch": 0.742937062937063, + "grad_norm": 0.5301491492363644, + "learning_rate": 1.8707053419939756e-05, + "loss": 0.2626168131828308, + "step": 1328 + }, + { + "epoch": 0.7434965034965035, + "grad_norm": 0.435453580653873, + "learning_rate": 1.8703850724322158e-05, + "loss": 0.22473716735839844, + "step": 1329 + }, + { + "epoch": 0.7440559440559441, + "grad_norm": 0.30013346815142017, + "learning_rate": 1.870064434184524e-05, + "loss": 0.09228631854057312, + "step": 1330 + }, + { + "epoch": 0.7446153846153846, + "grad_norm": 0.40309706907013504, + "learning_rate": 1.8697434273867193e-05, + "loss": 0.12133573740720749, + "step": 1331 + }, + { + "epoch": 0.7451748251748251, + "grad_norm": 0.42969642633057764, + "learning_rate": 1.8694220521747765e-05, + "loss": 0.11321327090263367, + "step": 1332 + }, + { + "epoch": 0.7457342657342657, + "grad_norm": 0.2722734481890191, + "learning_rate": 1.8691003086848273e-05, + "loss": 0.09745318442583084, + "step": 1333 + }, + { + "epoch": 0.7462937062937063, + "grad_norm": 0.6662206176718636, + "learning_rate": 1.868778197053158e-05, + "loss": 0.31245553493499756, + "step": 1334 + }, + { + "epoch": 0.7468531468531469, + "grad_norm": 0.8375900702997423, + "learning_rate": 1.8684557174162123e-05, + "loss": 0.2549764811992645, + "step": 1335 + }, + { + "epoch": 0.7474125874125874, + "grad_norm": 0.6406681644779667, + "learning_rate": 1.868132869910589e-05, + "loss": 0.3100712299346924, + "step": 1336 + }, + { + "epoch": 0.747972027972028, + "grad_norm": 0.48388275042707213, + "learning_rate": 1.8678096546730432e-05, + "loss": 0.12030473351478577, + "step": 1337 + }, + { + "epoch": 0.7485314685314686, + "grad_norm": 0.5524494375353823, + "learning_rate": 1.8674860718404852e-05, + "loss": 0.2433275729417801, + "step": 1338 + }, + { + "epoch": 0.7490909090909091, + "grad_norm": 0.6013119537796081, + "learning_rate": 1.8671621215499812e-05, + "loss": 0.2903353273868561, + "step": 1339 + }, + { + "epoch": 0.7496503496503496, + "grad_norm": 0.5893588684342579, + "learning_rate": 1.8668378039387532e-05, + "loss": 0.35398098826408386, + "step": 1340 + }, + { + "epoch": 0.7502097902097902, + "grad_norm": 0.4119299189699907, + "learning_rate": 1.866513119144179e-05, + "loss": 0.21113407611846924, + "step": 1341 + }, + { + "epoch": 0.7507692307692307, + "grad_norm": 0.42196440216142495, + "learning_rate": 1.8661880673037913e-05, + "loss": 0.1938326060771942, + "step": 1342 + }, + { + "epoch": 0.7513286713286713, + "grad_norm": 0.4047837132859201, + "learning_rate": 1.8658626485552783e-05, + "loss": 0.131469264626503, + "step": 1343 + }, + { + "epoch": 0.7518881118881119, + "grad_norm": 0.29831693633866163, + "learning_rate": 1.8655368630364847e-05, + "loss": 0.09293493628501892, + "step": 1344 + }, + { + "epoch": 0.7524475524475525, + "grad_norm": 0.4317758263202799, + "learning_rate": 1.8652107108854094e-05, + "loss": 0.13842858374118805, + "step": 1345 + }, + { + "epoch": 0.753006993006993, + "grad_norm": 0.4246925787052252, + "learning_rate": 1.864884192240207e-05, + "loss": 0.18156328797340393, + "step": 1346 + }, + { + "epoch": 0.7535664335664336, + "grad_norm": 0.3437202030405131, + "learning_rate": 1.8645573072391877e-05, + "loss": 0.14162813127040863, + "step": 1347 + }, + { + "epoch": 0.7541258741258742, + "grad_norm": 0.6012080434367089, + "learning_rate": 1.8642300560208156e-05, + "loss": 0.21421772241592407, + "step": 1348 + }, + { + "epoch": 0.7546853146853146, + "grad_norm": 0.5595183120884661, + "learning_rate": 1.8639024387237122e-05, + "loss": 0.2547898292541504, + "step": 1349 + }, + { + "epoch": 0.7552447552447552, + "grad_norm": 0.34988418230750673, + "learning_rate": 1.863574455486652e-05, + "loss": 0.15476490557193756, + "step": 1350 + }, + { + "epoch": 0.7558041958041958, + "grad_norm": 0.38510232662654004, + "learning_rate": 1.863246106448565e-05, + "loss": 0.1261695921421051, + "step": 1351 + }, + { + "epoch": 0.7563636363636363, + "grad_norm": 0.49050611674986944, + "learning_rate": 1.8629173917485367e-05, + "loss": 0.2383459210395813, + "step": 1352 + }, + { + "epoch": 0.7569230769230769, + "grad_norm": 0.4482365782038924, + "learning_rate": 1.8625883115258075e-05, + "loss": 0.18813957273960114, + "step": 1353 + }, + { + "epoch": 0.7574825174825175, + "grad_norm": 0.3210977216750005, + "learning_rate": 1.8622588659197717e-05, + "loss": 0.08599026501178741, + "step": 1354 + }, + { + "epoch": 0.7580419580419581, + "grad_norm": 0.3153014724674469, + "learning_rate": 1.8619290550699794e-05, + "loss": 0.13507288694381714, + "step": 1355 + }, + { + "epoch": 0.7586013986013986, + "grad_norm": 0.5759727581607136, + "learning_rate": 1.8615988791161346e-05, + "loss": 0.328048437833786, + "step": 1356 + }, + { + "epoch": 0.7591608391608392, + "grad_norm": 0.4473090511195733, + "learning_rate": 1.861268338198097e-05, + "loss": 0.17863020300865173, + "step": 1357 + }, + { + "epoch": 0.7597202797202797, + "grad_norm": 0.3031396468236042, + "learning_rate": 1.8609374324558794e-05, + "loss": 0.10626602917909622, + "step": 1358 + }, + { + "epoch": 0.7602797202797202, + "grad_norm": 0.39338301917348256, + "learning_rate": 1.8606061620296506e-05, + "loss": 0.13862758874893188, + "step": 1359 + }, + { + "epoch": 0.7608391608391608, + "grad_norm": 0.5024216508369472, + "learning_rate": 1.860274527059733e-05, + "loss": 0.2208954244852066, + "step": 1360 + }, + { + "epoch": 0.7613986013986014, + "grad_norm": 0.5928731132140602, + "learning_rate": 1.8599425276866036e-05, + "loss": 0.24852490425109863, + "step": 1361 + }, + { + "epoch": 0.761958041958042, + "grad_norm": 0.45557585977839504, + "learning_rate": 1.8596101640508943e-05, + "loss": 0.16396808624267578, + "step": 1362 + }, + { + "epoch": 0.7625174825174825, + "grad_norm": 0.42708585006999783, + "learning_rate": 1.8592774362933904e-05, + "loss": 0.1367586851119995, + "step": 1363 + }, + { + "epoch": 0.7630769230769231, + "grad_norm": 0.5129773778301971, + "learning_rate": 1.8589443445550324e-05, + "loss": 0.310060977935791, + "step": 1364 + }, + { + "epoch": 0.7636363636363637, + "grad_norm": 0.3007101367467586, + "learning_rate": 1.8586108889769137e-05, + "loss": 0.05615415796637535, + "step": 1365 + }, + { + "epoch": 0.7641958041958042, + "grad_norm": 0.38084730414141094, + "learning_rate": 1.858277069700283e-05, + "loss": 0.10863764584064484, + "step": 1366 + }, + { + "epoch": 0.7647552447552447, + "grad_norm": 0.33777908956246305, + "learning_rate": 1.857942886866542e-05, + "loss": 0.07619760185480118, + "step": 1367 + }, + { + "epoch": 0.7653146853146853, + "grad_norm": 0.5862493501728807, + "learning_rate": 1.8576083406172478e-05, + "loss": 0.30702123045921326, + "step": 1368 + }, + { + "epoch": 0.7658741258741258, + "grad_norm": 0.34446572907365863, + "learning_rate": 1.8572734310941105e-05, + "loss": 0.15862104296684265, + "step": 1369 + }, + { + "epoch": 0.7664335664335664, + "grad_norm": 0.4298873095841707, + "learning_rate": 1.856938158438994e-05, + "loss": 0.15880060195922852, + "step": 1370 + }, + { + "epoch": 0.766993006993007, + "grad_norm": 0.5173878204007457, + "learning_rate": 1.8566025227939168e-05, + "loss": 0.27035221457481384, + "step": 1371 + }, + { + "epoch": 0.7675524475524476, + "grad_norm": 0.621975471455463, + "learning_rate": 1.85626652430105e-05, + "loss": 0.31578513979911804, + "step": 1372 + }, + { + "epoch": 0.7681118881118881, + "grad_norm": 0.5510339329853917, + "learning_rate": 1.8559301631027192e-05, + "loss": 0.2574416697025299, + "step": 1373 + }, + { + "epoch": 0.7686713286713287, + "grad_norm": 0.48068247451445784, + "learning_rate": 1.8555934393414033e-05, + "loss": 0.16276581585407257, + "step": 1374 + }, + { + "epoch": 0.7692307692307693, + "grad_norm": 0.47141053235812963, + "learning_rate": 1.855256353159735e-05, + "loss": 0.3703039586544037, + "step": 1375 + }, + { + "epoch": 0.7697902097902098, + "grad_norm": 0.37741958080582383, + "learning_rate": 1.854918904700501e-05, + "loss": 0.11179596185684204, + "step": 1376 + }, + { + "epoch": 0.7703496503496503, + "grad_norm": 0.530190788161041, + "learning_rate": 1.8545810941066403e-05, + "loss": 0.22255843877792358, + "step": 1377 + }, + { + "epoch": 0.7709090909090909, + "grad_norm": 0.4009980247475383, + "learning_rate": 1.8542429215212464e-05, + "loss": 0.1687420904636383, + "step": 1378 + }, + { + "epoch": 0.7714685314685314, + "grad_norm": 0.5333102346495306, + "learning_rate": 1.8539043870875648e-05, + "loss": 0.3099929988384247, + "step": 1379 + }, + { + "epoch": 0.772027972027972, + "grad_norm": 0.5057633303685584, + "learning_rate": 1.853565490948996e-05, + "loss": 0.23256784677505493, + "step": 1380 + }, + { + "epoch": 0.7725874125874126, + "grad_norm": 0.3522213561830497, + "learning_rate": 1.853226233249092e-05, + "loss": 0.16178445518016815, + "step": 1381 + }, + { + "epoch": 0.7731468531468532, + "grad_norm": 0.43537488600623797, + "learning_rate": 1.8528866141315597e-05, + "loss": 0.20867958664894104, + "step": 1382 + }, + { + "epoch": 0.7737062937062937, + "grad_norm": 0.6903393146083875, + "learning_rate": 1.8525466337402572e-05, + "loss": 0.390109658241272, + "step": 1383 + }, + { + "epoch": 0.7742657342657343, + "grad_norm": 0.6172104808379274, + "learning_rate": 1.8522062922191976e-05, + "loss": 0.3983703553676605, + "step": 1384 + }, + { + "epoch": 0.7748251748251749, + "grad_norm": 0.6386829111473317, + "learning_rate": 1.8518655897125452e-05, + "loss": 0.3265753388404846, + "step": 1385 + }, + { + "epoch": 0.7753846153846153, + "grad_norm": 0.48208817233640283, + "learning_rate": 1.8515245263646186e-05, + "loss": 0.18475818634033203, + "step": 1386 + }, + { + "epoch": 0.7759440559440559, + "grad_norm": 0.8585309320012744, + "learning_rate": 1.851183102319888e-05, + "loss": 0.5182363390922546, + "step": 1387 + }, + { + "epoch": 0.7765034965034965, + "grad_norm": 0.27568710627043547, + "learning_rate": 1.8508413177229775e-05, + "loss": 0.08971129357814789, + "step": 1388 + }, + { + "epoch": 0.777062937062937, + "grad_norm": 0.5496817934005515, + "learning_rate": 1.8504991727186634e-05, + "loss": 0.3412243127822876, + "step": 1389 + }, + { + "epoch": 0.7776223776223776, + "grad_norm": 0.45811240037407247, + "learning_rate": 1.8501566674518743e-05, + "loss": 0.2338174283504486, + "step": 1390 + }, + { + "epoch": 0.7781818181818182, + "grad_norm": 0.5421095962784012, + "learning_rate": 1.8498138020676924e-05, + "loss": 0.2681596577167511, + "step": 1391 + }, + { + "epoch": 0.7787412587412588, + "grad_norm": 0.5792223498723954, + "learning_rate": 1.8494705767113513e-05, + "loss": 0.29711586236953735, + "step": 1392 + }, + { + "epoch": 0.7793006993006993, + "grad_norm": 0.32367434954961205, + "learning_rate": 1.8491269915282382e-05, + "loss": 0.11838226765394211, + "step": 1393 + }, + { + "epoch": 0.7798601398601399, + "grad_norm": 0.4037869216144031, + "learning_rate": 1.8487830466638916e-05, + "loss": 0.14415180683135986, + "step": 1394 + }, + { + "epoch": 0.7804195804195804, + "grad_norm": 0.39393357361012343, + "learning_rate": 1.8484387422640035e-05, + "loss": 0.0834486111998558, + "step": 1395 + }, + { + "epoch": 0.7809790209790209, + "grad_norm": 0.47178503290559576, + "learning_rate": 1.848094078474417e-05, + "loss": 0.23688632249832153, + "step": 1396 + }, + { + "epoch": 0.7815384615384615, + "grad_norm": 0.5474069632559416, + "learning_rate": 1.8477490554411284e-05, + "loss": 0.21789711713790894, + "step": 1397 + }, + { + "epoch": 0.7820979020979021, + "grad_norm": 0.3500160135924503, + "learning_rate": 1.8474036733102857e-05, + "loss": 0.10302956402301788, + "step": 1398 + }, + { + "epoch": 0.7826573426573427, + "grad_norm": 0.511024252741334, + "learning_rate": 1.8470579322281893e-05, + "loss": 0.23543362319469452, + "step": 1399 + }, + { + "epoch": 0.7832167832167832, + "grad_norm": 0.33227588768811095, + "learning_rate": 1.8467118323412914e-05, + "loss": 0.1557886153459549, + "step": 1400 + }, + { + "epoch": 0.7837762237762238, + "grad_norm": 0.4859737814348562, + "learning_rate": 1.846365373796196e-05, + "loss": 0.21970303356647491, + "step": 1401 + }, + { + "epoch": 0.7843356643356644, + "grad_norm": 0.4358609929463485, + "learning_rate": 1.84601855673966e-05, + "loss": 0.1444120854139328, + "step": 1402 + }, + { + "epoch": 0.784895104895105, + "grad_norm": 0.5612672965209708, + "learning_rate": 1.8456713813185905e-05, + "loss": 0.33197301626205444, + "step": 1403 + }, + { + "epoch": 0.7854545454545454, + "grad_norm": 0.5760570397214995, + "learning_rate": 1.845323847680048e-05, + "loss": 0.17940428853034973, + "step": 1404 + }, + { + "epoch": 0.786013986013986, + "grad_norm": 0.514308692596527, + "learning_rate": 1.8449759559712438e-05, + "loss": 0.28178954124450684, + "step": 1405 + }, + { + "epoch": 0.7865734265734265, + "grad_norm": 0.3077128246756312, + "learning_rate": 1.844627706339541e-05, + "loss": 0.07722143828868866, + "step": 1406 + }, + { + "epoch": 0.7871328671328671, + "grad_norm": 0.6496305456794554, + "learning_rate": 1.844279098932455e-05, + "loss": 0.24841883778572083, + "step": 1407 + }, + { + "epoch": 0.7876923076923077, + "grad_norm": 0.544146269826521, + "learning_rate": 1.8439301338976518e-05, + "loss": 0.3509570360183716, + "step": 1408 + }, + { + "epoch": 0.7882517482517483, + "grad_norm": 0.41706617740195656, + "learning_rate": 1.8435808113829494e-05, + "loss": 0.1569032073020935, + "step": 1409 + }, + { + "epoch": 0.7888111888111888, + "grad_norm": 0.3415037791847888, + "learning_rate": 1.8432311315363173e-05, + "loss": 0.08288843184709549, + "step": 1410 + }, + { + "epoch": 0.7893706293706294, + "grad_norm": 0.5913542016096398, + "learning_rate": 1.842881094505876e-05, + "loss": 0.2277572751045227, + "step": 1411 + }, + { + "epoch": 0.78993006993007, + "grad_norm": 0.6555774206012626, + "learning_rate": 1.8425307004398975e-05, + "loss": 0.24825136363506317, + "step": 1412 + }, + { + "epoch": 0.7904895104895104, + "grad_norm": 0.5265380018866219, + "learning_rate": 1.842179949486805e-05, + "loss": 0.248886838555336, + "step": 1413 + }, + { + "epoch": 0.791048951048951, + "grad_norm": 0.4579257511392682, + "learning_rate": 1.841828841795173e-05, + "loss": 0.25317031145095825, + "step": 1414 + }, + { + "epoch": 0.7916083916083916, + "grad_norm": 0.45349177936600193, + "learning_rate": 1.841477377513727e-05, + "loss": 0.1791454702615738, + "step": 1415 + }, + { + "epoch": 0.7921678321678322, + "grad_norm": 0.29617938379243614, + "learning_rate": 1.8411255567913437e-05, + "loss": 0.038264572620391846, + "step": 1416 + }, + { + "epoch": 0.7927272727272727, + "grad_norm": 0.3047608505693725, + "learning_rate": 1.8407733797770504e-05, + "loss": 0.11317572742700577, + "step": 1417 + }, + { + "epoch": 0.7932867132867133, + "grad_norm": 0.3919304998812546, + "learning_rate": 1.8404208466200254e-05, + "loss": 0.11327424645423889, + "step": 1418 + }, + { + "epoch": 0.7938461538461539, + "grad_norm": 0.4995051354231164, + "learning_rate": 1.840067957469599e-05, + "loss": 0.26368314027786255, + "step": 1419 + }, + { + "epoch": 0.7944055944055944, + "grad_norm": 0.7636363406813251, + "learning_rate": 1.8397147124752495e-05, + "loss": 0.565875768661499, + "step": 1420 + }, + { + "epoch": 0.794965034965035, + "grad_norm": 0.3469240073548153, + "learning_rate": 1.8393611117866098e-05, + "loss": 0.11382953822612762, + "step": 1421 + }, + { + "epoch": 0.7955244755244755, + "grad_norm": 0.6905454038498279, + "learning_rate": 1.83900715555346e-05, + "loss": 0.3906436264514923, + "step": 1422 + }, + { + "epoch": 0.796083916083916, + "grad_norm": 0.5283319170283113, + "learning_rate": 1.8386528439257328e-05, + "loss": 0.318826824426651, + "step": 1423 + }, + { + "epoch": 0.7966433566433566, + "grad_norm": 0.3316900308467073, + "learning_rate": 1.838298177053511e-05, + "loss": 0.09145613759756088, + "step": 1424 + }, + { + "epoch": 0.7972027972027972, + "grad_norm": 0.44356383047759484, + "learning_rate": 1.8379431550870273e-05, + "loss": 0.1778845340013504, + "step": 1425 + }, + { + "epoch": 0.7977622377622378, + "grad_norm": 0.27959458118940406, + "learning_rate": 1.8375877781766655e-05, + "loss": 0.08287844806909561, + "step": 1426 + }, + { + "epoch": 0.7983216783216783, + "grad_norm": 0.5454815223127065, + "learning_rate": 1.8372320464729595e-05, + "loss": 0.2466338872909546, + "step": 1427 + }, + { + "epoch": 0.7988811188811189, + "grad_norm": 0.4429319894742091, + "learning_rate": 1.8368759601265935e-05, + "loss": 0.20674027502536774, + "step": 1428 + }, + { + "epoch": 0.7994405594405595, + "grad_norm": 0.3877647648716051, + "learning_rate": 1.8365195192884016e-05, + "loss": 0.1120634526014328, + "step": 1429 + }, + { + "epoch": 0.8, + "grad_norm": 0.5439156500273681, + "learning_rate": 1.836162724109369e-05, + "loss": 0.20542174577713013, + "step": 1430 + }, + { + "epoch": 0.8005594405594405, + "grad_norm": 0.3391930560420253, + "learning_rate": 1.83580557474063e-05, + "loss": 0.11126194894313812, + "step": 1431 + }, + { + "epoch": 0.8011188811188811, + "grad_norm": 0.4981630218590183, + "learning_rate": 1.8354480713334696e-05, + "loss": 0.13972947001457214, + "step": 1432 + }, + { + "epoch": 0.8016783216783216, + "grad_norm": 0.17976970420995664, + "learning_rate": 1.8350902140393217e-05, + "loss": 0.049743738025426865, + "step": 1433 + }, + { + "epoch": 0.8022377622377622, + "grad_norm": 0.4460579696688039, + "learning_rate": 1.8347320030097714e-05, + "loss": 0.1878945678472519, + "step": 1434 + }, + { + "epoch": 0.8027972027972028, + "grad_norm": 0.36124388809467517, + "learning_rate": 1.8343734383965536e-05, + "loss": 0.14915086328983307, + "step": 1435 + }, + { + "epoch": 0.8033566433566434, + "grad_norm": 0.4882464458179913, + "learning_rate": 1.8340145203515516e-05, + "loss": 0.22911271452903748, + "step": 1436 + }, + { + "epoch": 0.8039160839160839, + "grad_norm": 0.6441536985381489, + "learning_rate": 1.8336552490268e-05, + "loss": 0.32970792055130005, + "step": 1437 + }, + { + "epoch": 0.8044755244755245, + "grad_norm": 0.5791314722112958, + "learning_rate": 1.833295624574482e-05, + "loss": 0.31875988841056824, + "step": 1438 + }, + { + "epoch": 0.8050349650349651, + "grad_norm": 0.5199290275330625, + "learning_rate": 1.8329356471469303e-05, + "loss": 0.27195048332214355, + "step": 1439 + }, + { + "epoch": 0.8055944055944056, + "grad_norm": 0.47708840733377816, + "learning_rate": 1.8325753168966286e-05, + "loss": 0.210830420255661, + "step": 1440 + }, + { + "epoch": 0.8061538461538461, + "grad_norm": 0.3793519692554209, + "learning_rate": 1.8322146339762077e-05, + "loss": 0.19733984768390656, + "step": 1441 + }, + { + "epoch": 0.8067132867132867, + "grad_norm": 1.2665430531987298, + "learning_rate": 1.8318535985384503e-05, + "loss": 0.17760130763053894, + "step": 1442 + }, + { + "epoch": 0.8072727272727273, + "grad_norm": 0.33410227970179474, + "learning_rate": 1.8314922107362866e-05, + "loss": 0.12815436720848083, + "step": 1443 + }, + { + "epoch": 0.8078321678321678, + "grad_norm": 0.7740823776115756, + "learning_rate": 1.8311304707227963e-05, + "loss": 0.41048890352249146, + "step": 1444 + }, + { + "epoch": 0.8083916083916084, + "grad_norm": 0.26296142545178164, + "learning_rate": 1.8307683786512095e-05, + "loss": 0.0654655247926712, + "step": 1445 + }, + { + "epoch": 0.808951048951049, + "grad_norm": 0.33920253513426546, + "learning_rate": 1.8304059346749042e-05, + "loss": 0.13886767625808716, + "step": 1446 + }, + { + "epoch": 0.8095104895104895, + "grad_norm": 0.4109119439705607, + "learning_rate": 1.8300431389474077e-05, + "loss": 0.11193230748176575, + "step": 1447 + }, + { + "epoch": 0.8100699300699301, + "grad_norm": 0.3295416040586028, + "learning_rate": 1.8296799916223963e-05, + "loss": 0.06599109619855881, + "step": 1448 + }, + { + "epoch": 0.8106293706293707, + "grad_norm": 0.41717578772543484, + "learning_rate": 1.8293164928536962e-05, + "loss": 0.19477437436580658, + "step": 1449 + }, + { + "epoch": 0.8111888111888111, + "grad_norm": 0.44781115255962695, + "learning_rate": 1.8289526427952805e-05, + "loss": 0.19513902068138123, + "step": 1450 + }, + { + "epoch": 0.8117482517482517, + "grad_norm": 0.5205554747133834, + "learning_rate": 1.8285884416012727e-05, + "loss": 0.31178590655326843, + "step": 1451 + }, + { + "epoch": 0.8123076923076923, + "grad_norm": 0.6099482916138088, + "learning_rate": 1.8282238894259455e-05, + "loss": 0.4289611577987671, + "step": 1452 + }, + { + "epoch": 0.8128671328671329, + "grad_norm": 0.3858988242311383, + "learning_rate": 1.827858986423718e-05, + "loss": 0.08483538776636124, + "step": 1453 + }, + { + "epoch": 0.8134265734265734, + "grad_norm": 0.39863710187154094, + "learning_rate": 1.82749373274916e-05, + "loss": 0.11833000183105469, + "step": 1454 + }, + { + "epoch": 0.813986013986014, + "grad_norm": 0.4335124075940468, + "learning_rate": 1.8271281285569884e-05, + "loss": 0.18328005075454712, + "step": 1455 + }, + { + "epoch": 0.8145454545454546, + "grad_norm": 0.6918542484252468, + "learning_rate": 1.8267621740020704e-05, + "loss": 0.27575016021728516, + "step": 1456 + }, + { + "epoch": 0.8151048951048951, + "grad_norm": 0.41488162489495173, + "learning_rate": 1.8263958692394196e-05, + "loss": 0.24062323570251465, + "step": 1457 + }, + { + "epoch": 0.8156643356643357, + "grad_norm": 0.5628763681748989, + "learning_rate": 1.826029214424199e-05, + "loss": 0.32751497626304626, + "step": 1458 + }, + { + "epoch": 0.8162237762237762, + "grad_norm": 0.4253206889118255, + "learning_rate": 1.8256622097117197e-05, + "loss": 0.23958967626094818, + "step": 1459 + }, + { + "epoch": 0.8167832167832167, + "grad_norm": 0.3897058450219951, + "learning_rate": 1.8252948552574416e-05, + "loss": 0.16753628849983215, + "step": 1460 + }, + { + "epoch": 0.8173426573426573, + "grad_norm": 0.3578912310841774, + "learning_rate": 1.8249271512169712e-05, + "loss": 0.08340223133563995, + "step": 1461 + }, + { + "epoch": 0.8179020979020979, + "grad_norm": 0.4271239833697933, + "learning_rate": 1.824559097746065e-05, + "loss": 0.16506993770599365, + "step": 1462 + }, + { + "epoch": 0.8184615384615385, + "grad_norm": 0.46605092766974143, + "learning_rate": 1.8241906950006257e-05, + "loss": 0.183750718832016, + "step": 1463 + }, + { + "epoch": 0.819020979020979, + "grad_norm": 0.2190218925983016, + "learning_rate": 1.8238219431367056e-05, + "loss": 0.2130127102136612, + "step": 1464 + }, + { + "epoch": 0.8195804195804196, + "grad_norm": 0.3421354911140324, + "learning_rate": 1.823452842310504e-05, + "loss": 0.1270238757133484, + "step": 1465 + }, + { + "epoch": 0.8201398601398602, + "grad_norm": 0.392803591841699, + "learning_rate": 1.8230833926783673e-05, + "loss": 0.06742765009403229, + "step": 1466 + }, + { + "epoch": 0.8206993006993007, + "grad_norm": 0.4485860741808864, + "learning_rate": 1.8227135943967912e-05, + "loss": 0.2227836698293686, + "step": 1467 + }, + { + "epoch": 0.8212587412587412, + "grad_norm": 0.5079558863837191, + "learning_rate": 1.8223434476224178e-05, + "loss": 0.3019767999649048, + "step": 1468 + }, + { + "epoch": 0.8218181818181818, + "grad_norm": 0.2517465065770272, + "learning_rate": 1.8219729525120383e-05, + "loss": 0.08236641436815262, + "step": 1469 + }, + { + "epoch": 0.8223776223776224, + "grad_norm": 0.41241292824118075, + "learning_rate": 1.82160210922259e-05, + "loss": 0.13357867300510406, + "step": 1470 + }, + { + "epoch": 0.8229370629370629, + "grad_norm": 0.42398940037243404, + "learning_rate": 1.8212309179111578e-05, + "loss": 0.17603634297847748, + "step": 1471 + }, + { + "epoch": 0.8234965034965035, + "grad_norm": 0.5254017894351614, + "learning_rate": 1.820859378734975e-05, + "loss": 0.20842991769313812, + "step": 1472 + }, + { + "epoch": 0.8240559440559441, + "grad_norm": 0.3154224511439607, + "learning_rate": 1.820487491851421e-05, + "loss": 0.04474640265107155, + "step": 1473 + }, + { + "epoch": 0.8246153846153846, + "grad_norm": 0.8128014004930416, + "learning_rate": 1.8201152574180243e-05, + "loss": 0.3176015019416809, + "step": 1474 + }, + { + "epoch": 0.8251748251748252, + "grad_norm": 0.3737082627317614, + "learning_rate": 1.8197426755924587e-05, + "loss": 0.16510044038295746, + "step": 1475 + }, + { + "epoch": 0.8257342657342658, + "grad_norm": 0.25555064011971257, + "learning_rate": 1.8193697465325458e-05, + "loss": 0.03591936081647873, + "step": 1476 + }, + { + "epoch": 0.8262937062937062, + "grad_norm": 0.4330506721467532, + "learning_rate": 1.8189964703962545e-05, + "loss": 0.09115429222583771, + "step": 1477 + }, + { + "epoch": 0.8268531468531468, + "grad_norm": 0.24471406334064924, + "learning_rate": 1.818622847341701e-05, + "loss": 0.055043987929821014, + "step": 1478 + }, + { + "epoch": 0.8274125874125874, + "grad_norm": 0.2872818496611353, + "learning_rate": 1.8182488775271477e-05, + "loss": 0.03424611687660217, + "step": 1479 + }, + { + "epoch": 0.827972027972028, + "grad_norm": 0.7838961889366408, + "learning_rate": 1.8178745611110045e-05, + "loss": 0.1455080360174179, + "step": 1480 + }, + { + "epoch": 0.8285314685314685, + "grad_norm": 0.38058896172549445, + "learning_rate": 1.817499898251828e-05, + "loss": 0.09734778851270676, + "step": 1481 + }, + { + "epoch": 0.8290909090909091, + "grad_norm": 0.8037266952814739, + "learning_rate": 1.8171248891083212e-05, + "loss": 0.3973001539707184, + "step": 1482 + }, + { + "epoch": 0.8296503496503497, + "grad_norm": 0.4448764410165946, + "learning_rate": 1.8167495338393337e-05, + "loss": 0.2914142608642578, + "step": 1483 + }, + { + "epoch": 0.8302097902097902, + "grad_norm": 0.5678575514937263, + "learning_rate": 1.8163738326038625e-05, + "loss": 0.39465975761413574, + "step": 1484 + }, + { + "epoch": 0.8307692307692308, + "grad_norm": 0.47210408337733123, + "learning_rate": 1.8159977855610507e-05, + "loss": 0.30928659439086914, + "step": 1485 + }, + { + "epoch": 0.8313286713286713, + "grad_norm": 0.41397218536811436, + "learning_rate": 1.8156213928701878e-05, + "loss": 0.20367246866226196, + "step": 1486 + }, + { + "epoch": 0.8318881118881118, + "grad_norm": 0.24964794607130558, + "learning_rate": 1.8152446546907095e-05, + "loss": 0.06059672683477402, + "step": 1487 + }, + { + "epoch": 0.8324475524475524, + "grad_norm": 0.35749078567549286, + "learning_rate": 1.8148675711821982e-05, + "loss": 0.15735286474227905, + "step": 1488 + }, + { + "epoch": 0.833006993006993, + "grad_norm": 0.4163539426473246, + "learning_rate": 1.814490142504383e-05, + "loss": 0.16312754154205322, + "step": 1489 + }, + { + "epoch": 0.8335664335664336, + "grad_norm": 0.6187245308326341, + "learning_rate": 1.814112368817138e-05, + "loss": 0.24480485916137695, + "step": 1490 + }, + { + "epoch": 0.8341258741258741, + "grad_norm": 0.46749506213138714, + "learning_rate": 1.8137342502804848e-05, + "loss": 0.23900210857391357, + "step": 1491 + }, + { + "epoch": 0.8346853146853147, + "grad_norm": 0.3070093270419105, + "learning_rate": 1.81335578705459e-05, + "loss": 0.0961148738861084, + "step": 1492 + }, + { + "epoch": 0.8352447552447553, + "grad_norm": 0.4886332425534218, + "learning_rate": 1.8129769792997662e-05, + "loss": 0.22839057445526123, + "step": 1493 + }, + { + "epoch": 0.8358041958041958, + "grad_norm": 0.5731665054590872, + "learning_rate": 1.8125978271764736e-05, + "loss": 0.2662620544433594, + "step": 1494 + }, + { + "epoch": 0.8363636363636363, + "grad_norm": 0.43446074022269476, + "learning_rate": 1.8122183308453157e-05, + "loss": 0.22938987612724304, + "step": 1495 + }, + { + "epoch": 0.8369230769230769, + "grad_norm": 0.5069591980097515, + "learning_rate": 1.811838490467044e-05, + "loss": 0.09429021179676056, + "step": 1496 + }, + { + "epoch": 0.8374825174825175, + "grad_norm": 0.47415506190497186, + "learning_rate": 1.8114583062025547e-05, + "loss": 0.24409779906272888, + "step": 1497 + }, + { + "epoch": 0.838041958041958, + "grad_norm": 0.5112906041816498, + "learning_rate": 1.81107777821289e-05, + "loss": 0.18156293034553528, + "step": 1498 + }, + { + "epoch": 0.8386013986013986, + "grad_norm": 0.5281672400584042, + "learning_rate": 1.8106969066592368e-05, + "loss": 0.25297877192497253, + "step": 1499 + }, + { + "epoch": 0.8391608391608392, + "grad_norm": 0.3888983709219036, + "learning_rate": 1.8103156917029293e-05, + "loss": 0.297002375125885, + "step": 1500 + }, + { + "epoch": 0.8397202797202797, + "grad_norm": 0.4756252536000493, + "learning_rate": 1.8099341335054453e-05, + "loss": 0.1789444386959076, + "step": 1501 + }, + { + "epoch": 0.8402797202797203, + "grad_norm": 0.2312550316193629, + "learning_rate": 1.809552232228409e-05, + "loss": 0.043852709233760834, + "step": 1502 + }, + { + "epoch": 0.8408391608391609, + "grad_norm": 0.5047246705113532, + "learning_rate": 1.8091699880335903e-05, + "loss": 0.11034829914569855, + "step": 1503 + }, + { + "epoch": 0.8413986013986015, + "grad_norm": 0.5162465203046392, + "learning_rate": 1.808787401082903e-05, + "loss": 0.2640632390975952, + "step": 1504 + }, + { + "epoch": 0.8419580419580419, + "grad_norm": 0.5498662940654605, + "learning_rate": 1.808404471538408e-05, + "loss": 0.3162291646003723, + "step": 1505 + }, + { + "epoch": 0.8425174825174825, + "grad_norm": 0.6109716237148386, + "learning_rate": 1.808021199562309e-05, + "loss": 0.4285929501056671, + "step": 1506 + }, + { + "epoch": 0.8430769230769231, + "grad_norm": 0.49648420186743875, + "learning_rate": 1.8076375853169562e-05, + "loss": 0.2947182357311249, + "step": 1507 + }, + { + "epoch": 0.8436363636363636, + "grad_norm": 0.4377206211380998, + "learning_rate": 1.8072536289648455e-05, + "loss": 0.12361947447061539, + "step": 1508 + }, + { + "epoch": 0.8441958041958042, + "grad_norm": 0.5930729822360726, + "learning_rate": 1.8068693306686157e-05, + "loss": 0.2050374448299408, + "step": 1509 + }, + { + "epoch": 0.8447552447552448, + "grad_norm": 0.35025972673977807, + "learning_rate": 1.806484690591052e-05, + "loss": 0.0773276835680008, + "step": 1510 + }, + { + "epoch": 0.8453146853146853, + "grad_norm": 0.5670313694639525, + "learning_rate": 1.8060997088950837e-05, + "loss": 0.13120685517787933, + "step": 1511 + }, + { + "epoch": 0.8458741258741259, + "grad_norm": 0.5400342626191098, + "learning_rate": 1.805714385743785e-05, + "loss": 0.24504128098487854, + "step": 1512 + }, + { + "epoch": 0.8464335664335665, + "grad_norm": 0.5799430984954829, + "learning_rate": 1.805328721300375e-05, + "loss": 0.2342679500579834, + "step": 1513 + }, + { + "epoch": 0.846993006993007, + "grad_norm": 0.45381885838579344, + "learning_rate": 1.8049427157282164e-05, + "loss": 0.15123756229877472, + "step": 1514 + }, + { + "epoch": 0.8475524475524475, + "grad_norm": 0.4719904152179452, + "learning_rate": 1.8045563691908178e-05, + "loss": 0.21666626632213593, + "step": 1515 + }, + { + "epoch": 0.8481118881118881, + "grad_norm": 0.4016475324647, + "learning_rate": 1.8041696818518314e-05, + "loss": 0.16559508442878723, + "step": 1516 + }, + { + "epoch": 0.8486713286713287, + "grad_norm": 0.4621651829525613, + "learning_rate": 1.803782653875054e-05, + "loss": 0.20894457399845123, + "step": 1517 + }, + { + "epoch": 0.8492307692307692, + "grad_norm": 0.4128755369586208, + "learning_rate": 1.8033952854244256e-05, + "loss": 0.14789265394210815, + "step": 1518 + }, + { + "epoch": 0.8497902097902098, + "grad_norm": 0.4318621680166332, + "learning_rate": 1.8030075766640326e-05, + "loss": 0.16713255643844604, + "step": 1519 + }, + { + "epoch": 0.8503496503496504, + "grad_norm": 0.5891858778690118, + "learning_rate": 1.8026195277581035e-05, + "loss": 0.32784438133239746, + "step": 1520 + }, + { + "epoch": 0.850909090909091, + "grad_norm": 0.5319517731137454, + "learning_rate": 1.8022311388710127e-05, + "loss": 0.44883227348327637, + "step": 1521 + }, + { + "epoch": 0.8514685314685315, + "grad_norm": 0.29770337303963595, + "learning_rate": 1.8018424101672762e-05, + "loss": 0.11389957368373871, + "step": 1522 + }, + { + "epoch": 0.852027972027972, + "grad_norm": 0.5079371653697593, + "learning_rate": 1.8014533418115567e-05, + "loss": 0.20411473512649536, + "step": 1523 + }, + { + "epoch": 0.8525874125874126, + "grad_norm": 0.36338656181501744, + "learning_rate": 1.8010639339686585e-05, + "loss": 0.11225219070911407, + "step": 1524 + }, + { + "epoch": 0.8531468531468531, + "grad_norm": 1.4268823738479774, + "learning_rate": 1.8006741868035313e-05, + "loss": 0.36463797092437744, + "step": 1525 + }, + { + "epoch": 0.8537062937062937, + "grad_norm": 0.5115765920698717, + "learning_rate": 1.8002841004812677e-05, + "loss": 0.2551735043525696, + "step": 1526 + }, + { + "epoch": 0.8542657342657343, + "grad_norm": 0.6281628948962924, + "learning_rate": 1.799893675167104e-05, + "loss": 0.4101211428642273, + "step": 1527 + }, + { + "epoch": 0.8548251748251748, + "grad_norm": 0.4258927083772403, + "learning_rate": 1.7995029110264207e-05, + "loss": 0.1648799180984497, + "step": 1528 + }, + { + "epoch": 0.8553846153846154, + "grad_norm": 0.4152828142465705, + "learning_rate": 1.7991118082247405e-05, + "loss": 0.10228888690471649, + "step": 1529 + }, + { + "epoch": 0.855944055944056, + "grad_norm": 0.4062987031402474, + "learning_rate": 1.7987203669277314e-05, + "loss": 0.2177266925573349, + "step": 1530 + }, + { + "epoch": 0.8565034965034966, + "grad_norm": 0.5546002334106177, + "learning_rate": 1.7983285873012027e-05, + "loss": 0.1733018308877945, + "step": 1531 + }, + { + "epoch": 0.857062937062937, + "grad_norm": 0.5546226091291266, + "learning_rate": 1.797936469511109e-05, + "loss": 0.16278310120105743, + "step": 1532 + }, + { + "epoch": 0.8576223776223776, + "grad_norm": 0.30069979268026703, + "learning_rate": 1.7975440137235475e-05, + "loss": 0.14814062416553497, + "step": 1533 + }, + { + "epoch": 0.8581818181818182, + "grad_norm": 0.5586060858001686, + "learning_rate": 1.7971512201047575e-05, + "loss": 0.26308244466781616, + "step": 1534 + }, + { + "epoch": 0.8587412587412587, + "grad_norm": 0.6148685741473294, + "learning_rate": 1.7967580888211226e-05, + "loss": 0.3105056583881378, + "step": 1535 + }, + { + "epoch": 0.8593006993006993, + "grad_norm": 0.5264431987795197, + "learning_rate": 1.7963646200391693e-05, + "loss": 0.2541734576225281, + "step": 1536 + }, + { + "epoch": 0.8598601398601399, + "grad_norm": 0.49601823545546225, + "learning_rate": 1.7959708139255658e-05, + "loss": 0.29362165927886963, + "step": 1537 + }, + { + "epoch": 0.8604195804195804, + "grad_norm": 0.5167707325291908, + "learning_rate": 1.7955766706471257e-05, + "loss": 0.14905357360839844, + "step": 1538 + }, + { + "epoch": 0.860979020979021, + "grad_norm": 0.6921000919189215, + "learning_rate": 1.7951821903708027e-05, + "loss": 0.48252275586128235, + "step": 1539 + }, + { + "epoch": 0.8615384615384616, + "grad_norm": 0.31072911476918497, + "learning_rate": 1.7947873732636954e-05, + "loss": 0.09573337435722351, + "step": 1540 + }, + { + "epoch": 0.862097902097902, + "grad_norm": 0.4423909388913804, + "learning_rate": 1.7943922194930436e-05, + "loss": 0.09844426810741425, + "step": 1541 + }, + { + "epoch": 0.8626573426573426, + "grad_norm": 0.5995897409860936, + "learning_rate": 1.7939967292262305e-05, + "loss": 0.46954768896102905, + "step": 1542 + }, + { + "epoch": 0.8632167832167832, + "grad_norm": 0.3976764993546014, + "learning_rate": 1.7936009026307814e-05, + "loss": 0.21179917454719543, + "step": 1543 + }, + { + "epoch": 0.8637762237762238, + "grad_norm": 0.3398514474143028, + "learning_rate": 1.7932047398743642e-05, + "loss": 0.09582630544900894, + "step": 1544 + }, + { + "epoch": 0.8643356643356643, + "grad_norm": 0.39999128260293587, + "learning_rate": 1.79280824112479e-05, + "loss": 0.11422031372785568, + "step": 1545 + }, + { + "epoch": 0.8648951048951049, + "grad_norm": 0.4204830421688965, + "learning_rate": 1.7924114065500107e-05, + "loss": 0.054923612624406815, + "step": 1546 + }, + { + "epoch": 0.8654545454545455, + "grad_norm": 0.7153393505606297, + "learning_rate": 1.7920142363181212e-05, + "loss": 0.4437777101993561, + "step": 1547 + }, + { + "epoch": 0.866013986013986, + "grad_norm": 0.647075519426563, + "learning_rate": 1.7916167305973592e-05, + "loss": 0.26439574360847473, + "step": 1548 + }, + { + "epoch": 0.8665734265734266, + "grad_norm": 0.4574372470748115, + "learning_rate": 1.7912188895561035e-05, + "loss": 0.19346833229064941, + "step": 1549 + }, + { + "epoch": 0.8671328671328671, + "grad_norm": 0.5358095826473872, + "learning_rate": 1.7908207133628756e-05, + "loss": 0.25684428215026855, + "step": 1550 + }, + { + "epoch": 0.8676923076923077, + "grad_norm": 0.46182190166233394, + "learning_rate": 1.7904222021863385e-05, + "loss": 0.3001222312450409, + "step": 1551 + }, + { + "epoch": 0.8682517482517482, + "grad_norm": 0.7307549806882095, + "learning_rate": 1.7900233561952976e-05, + "loss": 0.27886486053466797, + "step": 1552 + }, + { + "epoch": 0.8688111888111888, + "grad_norm": 0.478994606029666, + "learning_rate": 1.7896241755586994e-05, + "loss": 0.17085976898670197, + "step": 1553 + }, + { + "epoch": 0.8693706293706294, + "grad_norm": 0.3629342150628365, + "learning_rate": 1.7892246604456335e-05, + "loss": 0.10421441495418549, + "step": 1554 + }, + { + "epoch": 0.8699300699300699, + "grad_norm": 0.4574896159448361, + "learning_rate": 1.7888248110253292e-05, + "loss": 0.19437983632087708, + "step": 1555 + }, + { + "epoch": 0.8704895104895105, + "grad_norm": 0.3294379171477935, + "learning_rate": 1.7884246274671592e-05, + "loss": 0.11298567801713943, + "step": 1556 + }, + { + "epoch": 0.8710489510489511, + "grad_norm": 0.4363397345441574, + "learning_rate": 1.7880241099406368e-05, + "loss": 0.1516723930835724, + "step": 1557 + }, + { + "epoch": 0.8716083916083917, + "grad_norm": 0.5659857777029176, + "learning_rate": 1.787623258615417e-05, + "loss": 0.29406455159187317, + "step": 1558 + }, + { + "epoch": 0.8721678321678321, + "grad_norm": 0.3627962680620667, + "learning_rate": 1.7872220736612963e-05, + "loss": 0.10896499454975128, + "step": 1559 + }, + { + "epoch": 0.8727272727272727, + "grad_norm": 0.4834727168508518, + "learning_rate": 1.7868205552482122e-05, + "loss": 0.21315889060497284, + "step": 1560 + }, + { + "epoch": 0.8732867132867133, + "grad_norm": 0.48933081623694075, + "learning_rate": 1.786418703546244e-05, + "loss": 0.27381691336631775, + "step": 1561 + }, + { + "epoch": 0.8738461538461538, + "grad_norm": 0.532303095932101, + "learning_rate": 1.786016518725611e-05, + "loss": 0.2666417956352234, + "step": 1562 + }, + { + "epoch": 0.8744055944055944, + "grad_norm": 0.4216222318443185, + "learning_rate": 1.785614000956676e-05, + "loss": 0.18479011952877045, + "step": 1563 + }, + { + "epoch": 0.874965034965035, + "grad_norm": 0.7386825562889271, + "learning_rate": 1.7852111504099395e-05, + "loss": 0.20558267831802368, + "step": 1564 + }, + { + "epoch": 0.8755244755244755, + "grad_norm": 0.4462965383699469, + "learning_rate": 1.7848079672560462e-05, + "loss": 0.1752689778804779, + "step": 1565 + }, + { + "epoch": 0.8760839160839161, + "grad_norm": 0.35226181373509835, + "learning_rate": 1.7844044516657796e-05, + "loss": 0.12320493161678314, + "step": 1566 + }, + { + "epoch": 0.8766433566433567, + "grad_norm": 0.5452975207012505, + "learning_rate": 1.7840006038100645e-05, + "loss": 0.3106265068054199, + "step": 1567 + }, + { + "epoch": 0.8772027972027973, + "grad_norm": 0.5070810535450645, + "learning_rate": 1.7835964238599666e-05, + "loss": 0.3688770830631256, + "step": 1568 + }, + { + "epoch": 0.8777622377622377, + "grad_norm": 0.32665328109099784, + "learning_rate": 1.783191911986693e-05, + "loss": 0.06713690608739853, + "step": 1569 + }, + { + "epoch": 0.8783216783216783, + "grad_norm": 0.4751896502872442, + "learning_rate": 1.7827870683615894e-05, + "loss": 0.2924523651599884, + "step": 1570 + }, + { + "epoch": 0.8788811188811189, + "grad_norm": 0.48292280917684194, + "learning_rate": 1.7823818931561444e-05, + "loss": 0.21796932816505432, + "step": 1571 + }, + { + "epoch": 0.8794405594405594, + "grad_norm": 0.31526110287225695, + "learning_rate": 1.781976386541985e-05, + "loss": 0.11356610804796219, + "step": 1572 + }, + { + "epoch": 0.88, + "grad_norm": 0.4643420091867196, + "learning_rate": 1.781570548690881e-05, + "loss": 0.24290059506893158, + "step": 1573 + }, + { + "epoch": 0.8805594405594406, + "grad_norm": 0.38110939295011775, + "learning_rate": 1.781164379774739e-05, + "loss": 0.13034270703792572, + "step": 1574 + }, + { + "epoch": 0.8811188811188811, + "grad_norm": 0.3148974445342348, + "learning_rate": 1.78075787996561e-05, + "loss": 0.07894525676965714, + "step": 1575 + }, + { + "epoch": 0.8816783216783217, + "grad_norm": 0.36215832623018, + "learning_rate": 1.780351049435681e-05, + "loss": 0.13240502774715424, + "step": 1576 + }, + { + "epoch": 0.8822377622377623, + "grad_norm": 0.43083366146140056, + "learning_rate": 1.7799438883572825e-05, + "loss": 0.21378064155578613, + "step": 1577 + }, + { + "epoch": 0.8827972027972028, + "grad_norm": 0.37053747914653584, + "learning_rate": 1.7795363969028834e-05, + "loss": 0.1334388554096222, + "step": 1578 + }, + { + "epoch": 0.8833566433566433, + "grad_norm": 0.5013781801856774, + "learning_rate": 1.7791285752450923e-05, + "loss": 0.2582058906555176, + "step": 1579 + }, + { + "epoch": 0.8839160839160839, + "grad_norm": 0.65582662123991, + "learning_rate": 1.7787204235566587e-05, + "loss": 0.4457191526889801, + "step": 1580 + }, + { + "epoch": 0.8844755244755245, + "grad_norm": 0.37369435779925253, + "learning_rate": 1.7783119420104707e-05, + "loss": 0.12105165421962738, + "step": 1581 + }, + { + "epoch": 0.885034965034965, + "grad_norm": 0.5112156427244609, + "learning_rate": 1.7779031307795576e-05, + "loss": 0.2794908881187439, + "step": 1582 + }, + { + "epoch": 0.8855944055944056, + "grad_norm": 0.5790336093648846, + "learning_rate": 1.7774939900370868e-05, + "loss": 0.3630887269973755, + "step": 1583 + }, + { + "epoch": 0.8861538461538462, + "grad_norm": 0.6041666390262403, + "learning_rate": 1.7770845199563665e-05, + "loss": 0.44180774688720703, + "step": 1584 + }, + { + "epoch": 0.8867132867132868, + "grad_norm": 0.32028883864983243, + "learning_rate": 1.776674720710844e-05, + "loss": 0.13922330737113953, + "step": 1585 + }, + { + "epoch": 0.8872727272727273, + "grad_norm": 0.20955813894976666, + "learning_rate": 1.7762645924741052e-05, + "loss": 0.01429430115967989, + "step": 1586 + }, + { + "epoch": 0.8878321678321678, + "grad_norm": 0.7289758799773486, + "learning_rate": 1.7758541354198765e-05, + "loss": 0.4742003083229065, + "step": 1587 + }, + { + "epoch": 0.8883916083916084, + "grad_norm": 0.42912189849127436, + "learning_rate": 1.7754433497220233e-05, + "loss": 0.19136425852775574, + "step": 1588 + }, + { + "epoch": 0.8889510489510489, + "grad_norm": 0.48617375532023893, + "learning_rate": 1.77503223555455e-05, + "loss": 0.1977563202381134, + "step": 1589 + }, + { + "epoch": 0.8895104895104895, + "grad_norm": 0.41342896339346946, + "learning_rate": 1.7746207930916005e-05, + "loss": 0.18498551845550537, + "step": 1590 + }, + { + "epoch": 0.8900699300699301, + "grad_norm": 0.30344443071394306, + "learning_rate": 1.7742090225074567e-05, + "loss": 0.12361852824687958, + "step": 1591 + }, + { + "epoch": 0.8906293706293706, + "grad_norm": 0.3912683659656579, + "learning_rate": 1.7737969239765412e-05, + "loss": 0.2083572894334793, + "step": 1592 + }, + { + "epoch": 0.8911888111888112, + "grad_norm": 0.4885903613238203, + "learning_rate": 1.773384497673414e-05, + "loss": 0.3200020492076874, + "step": 1593 + }, + { + "epoch": 0.8917482517482518, + "grad_norm": 0.4354792586806633, + "learning_rate": 1.7729717437727746e-05, + "loss": 0.12671269476413727, + "step": 1594 + }, + { + "epoch": 0.8923076923076924, + "grad_norm": 0.4508876551445146, + "learning_rate": 1.7725586624494614e-05, + "loss": 0.2997719943523407, + "step": 1595 + }, + { + "epoch": 0.8928671328671328, + "grad_norm": 0.39015262126558725, + "learning_rate": 1.772145253878451e-05, + "loss": 0.2988595962524414, + "step": 1596 + }, + { + "epoch": 0.8934265734265734, + "grad_norm": 0.3435046383219315, + "learning_rate": 1.7717315182348597e-05, + "loss": 0.14170129597187042, + "step": 1597 + }, + { + "epoch": 0.893986013986014, + "grad_norm": 0.39315447690381977, + "learning_rate": 1.7713174556939403e-05, + "loss": 0.14661243557929993, + "step": 1598 + }, + { + "epoch": 0.8945454545454545, + "grad_norm": 0.5780720644868425, + "learning_rate": 1.7709030664310862e-05, + "loss": 0.31579482555389404, + "step": 1599 + }, + { + "epoch": 0.8951048951048951, + "grad_norm": 0.3723679771931456, + "learning_rate": 1.770488350621828e-05, + "loss": 0.17461243271827698, + "step": 1600 + }, + { + "epoch": 0.8956643356643357, + "grad_norm": 0.4626412310417104, + "learning_rate": 1.7700733084418355e-05, + "loss": 0.3058391213417053, + "step": 1601 + }, + { + "epoch": 0.8962237762237762, + "grad_norm": 0.6206657292883363, + "learning_rate": 1.769657940066915e-05, + "loss": 0.2317260205745697, + "step": 1602 + }, + { + "epoch": 0.8967832167832168, + "grad_norm": 0.4521102041324381, + "learning_rate": 1.7692422456730133e-05, + "loss": 0.16295269131660461, + "step": 1603 + }, + { + "epoch": 0.8973426573426574, + "grad_norm": 0.4038732386083825, + "learning_rate": 1.7688262254362135e-05, + "loss": 0.22177726030349731, + "step": 1604 + }, + { + "epoch": 0.8979020979020979, + "grad_norm": 0.5174917689549509, + "learning_rate": 1.768409879532738e-05, + "loss": 0.2447076141834259, + "step": 1605 + }, + { + "epoch": 0.8984615384615384, + "grad_norm": 0.5550225885023431, + "learning_rate": 1.767993208138946e-05, + "loss": 0.2966054677963257, + "step": 1606 + }, + { + "epoch": 0.899020979020979, + "grad_norm": 0.6537235953771467, + "learning_rate": 1.7675762114313348e-05, + "loss": 0.37276527285575867, + "step": 1607 + }, + { + "epoch": 0.8995804195804196, + "grad_norm": 0.5623676601794538, + "learning_rate": 1.767158889586541e-05, + "loss": 0.2815719246864319, + "step": 1608 + }, + { + "epoch": 0.9001398601398601, + "grad_norm": 0.3837789203103388, + "learning_rate": 1.7667412427813362e-05, + "loss": 0.12265042215585709, + "step": 1609 + }, + { + "epoch": 0.9006993006993007, + "grad_norm": 0.3686094169540023, + "learning_rate": 1.7663232711926325e-05, + "loss": 0.08063453435897827, + "step": 1610 + }, + { + "epoch": 0.9012587412587413, + "grad_norm": 0.20024732280701354, + "learning_rate": 1.7659049749974778e-05, + "loss": 0.02443072944879532, + "step": 1611 + }, + { + "epoch": 0.9018181818181819, + "grad_norm": 0.5106437133299264, + "learning_rate": 1.7654863543730576e-05, + "loss": 0.21359990537166595, + "step": 1612 + }, + { + "epoch": 0.9023776223776224, + "grad_norm": 0.46737817076168947, + "learning_rate": 1.7650674094966955e-05, + "loss": 0.2204931527376175, + "step": 1613 + }, + { + "epoch": 0.9029370629370629, + "grad_norm": 0.3254176516699768, + "learning_rate": 1.7646481405458523e-05, + "loss": 0.08348868787288666, + "step": 1614 + }, + { + "epoch": 0.9034965034965035, + "grad_norm": 0.48509596359162976, + "learning_rate": 1.7642285476981252e-05, + "loss": 0.1931324303150177, + "step": 1615 + }, + { + "epoch": 0.904055944055944, + "grad_norm": 0.7437444153831381, + "learning_rate": 1.76380863113125e-05, + "loss": 0.46750667691230774, + "step": 1616 + }, + { + "epoch": 0.9046153846153846, + "grad_norm": 0.3686687662771808, + "learning_rate": 1.763388391023099e-05, + "loss": 0.19260993599891663, + "step": 1617 + }, + { + "epoch": 0.9051748251748252, + "grad_norm": 0.3630893169256096, + "learning_rate": 1.762967827551681e-05, + "loss": 0.18408820033073425, + "step": 1618 + }, + { + "epoch": 0.9057342657342657, + "grad_norm": 0.5163440145396219, + "learning_rate": 1.7625469408951425e-05, + "loss": 0.20373623073101044, + "step": 1619 + }, + { + "epoch": 0.9062937062937063, + "grad_norm": 0.3437447424677736, + "learning_rate": 1.7621257312317666e-05, + "loss": 0.14490383863449097, + "step": 1620 + }, + { + "epoch": 0.9068531468531469, + "grad_norm": 0.2488641291263141, + "learning_rate": 1.761704198739973e-05, + "loss": 0.07406635582447052, + "step": 1621 + }, + { + "epoch": 0.9074125874125875, + "grad_norm": 0.268402822314311, + "learning_rate": 1.7612823435983187e-05, + "loss": 0.07868371903896332, + "step": 1622 + }, + { + "epoch": 0.9079720279720279, + "grad_norm": 0.3879784412796822, + "learning_rate": 1.7608601659854975e-05, + "loss": 0.18978124856948853, + "step": 1623 + }, + { + "epoch": 0.9085314685314685, + "grad_norm": 0.3332883523788926, + "learning_rate": 1.7604376660803387e-05, + "loss": 0.12689727544784546, + "step": 1624 + }, + { + "epoch": 0.9090909090909091, + "grad_norm": 0.36868501239752544, + "learning_rate": 1.760014844061809e-05, + "loss": 0.0913744792342186, + "step": 1625 + }, + { + "epoch": 0.9096503496503496, + "grad_norm": 0.41356174968963316, + "learning_rate": 1.7595917001090113e-05, + "loss": 0.12943418323993683, + "step": 1626 + }, + { + "epoch": 0.9102097902097902, + "grad_norm": 0.20372723580815705, + "learning_rate": 1.759168234401185e-05, + "loss": 0.08174778521060944, + "step": 1627 + }, + { + "epoch": 0.9107692307692308, + "grad_norm": 0.3253195234298958, + "learning_rate": 1.758744447117706e-05, + "loss": 0.10777575522661209, + "step": 1628 + }, + { + "epoch": 0.9113286713286713, + "grad_norm": 0.4258404270714846, + "learning_rate": 1.7583203384380857e-05, + "loss": 0.18899346888065338, + "step": 1629 + }, + { + "epoch": 0.9118881118881119, + "grad_norm": 0.6579190458747962, + "learning_rate": 1.757895908541972e-05, + "loss": 0.474851131439209, + "step": 1630 + }, + { + "epoch": 0.9124475524475525, + "grad_norm": 0.4095964975005497, + "learning_rate": 1.757471157609149e-05, + "loss": 0.1819888949394226, + "step": 1631 + }, + { + "epoch": 0.9130069930069931, + "grad_norm": 0.4477416198905772, + "learning_rate": 1.7570460858195372e-05, + "loss": 0.2196742594242096, + "step": 1632 + }, + { + "epoch": 0.9135664335664335, + "grad_norm": 0.3146451640320151, + "learning_rate": 1.756620693353192e-05, + "loss": 0.14556241035461426, + "step": 1633 + }, + { + "epoch": 0.9141258741258741, + "grad_norm": 0.21561497540910668, + "learning_rate": 1.7561949803903052e-05, + "loss": 0.04626625403761864, + "step": 1634 + }, + { + "epoch": 0.9146853146853147, + "grad_norm": 0.45010404524452796, + "learning_rate": 1.7557689471112044e-05, + "loss": 0.17609430849552155, + "step": 1635 + }, + { + "epoch": 0.9152447552447552, + "grad_norm": 0.4840527934505853, + "learning_rate": 1.7553425936963532e-05, + "loss": 0.19454896450042725, + "step": 1636 + }, + { + "epoch": 0.9158041958041958, + "grad_norm": 0.4518889867132352, + "learning_rate": 1.75491592032635e-05, + "loss": 0.26230350136756897, + "step": 1637 + }, + { + "epoch": 0.9163636363636364, + "grad_norm": 0.5836440478485843, + "learning_rate": 1.7544889271819285e-05, + "loss": 0.34506380558013916, + "step": 1638 + }, + { + "epoch": 0.916923076923077, + "grad_norm": 0.4103164989028099, + "learning_rate": 1.75406161444396e-05, + "loss": 0.18003425002098083, + "step": 1639 + }, + { + "epoch": 0.9174825174825175, + "grad_norm": 0.6798086736499991, + "learning_rate": 1.7536339822934484e-05, + "loss": 0.4571908116340637, + "step": 1640 + }, + { + "epoch": 0.9180419580419581, + "grad_norm": 0.4890732407554329, + "learning_rate": 1.7532060309115345e-05, + "loss": 0.27332603931427, + "step": 1641 + }, + { + "epoch": 0.9186013986013986, + "grad_norm": 0.6194514946551914, + "learning_rate": 1.7527777604794944e-05, + "loss": 0.2583649754524231, + "step": 1642 + }, + { + "epoch": 0.9191608391608391, + "grad_norm": 0.5998085339049049, + "learning_rate": 1.7523491711787384e-05, + "loss": 0.36741504073143005, + "step": 1643 + }, + { + "epoch": 0.9197202797202797, + "grad_norm": 0.34207481941243867, + "learning_rate": 1.7519202631908126e-05, + "loss": 0.13514187932014465, + "step": 1644 + }, + { + "epoch": 0.9202797202797203, + "grad_norm": 0.548005728855821, + "learning_rate": 1.751491036697398e-05, + "loss": 0.18971621990203857, + "step": 1645 + }, + { + "epoch": 0.9208391608391608, + "grad_norm": 0.44020242510229796, + "learning_rate": 1.7510614918803102e-05, + "loss": 0.27957165241241455, + "step": 1646 + }, + { + "epoch": 0.9213986013986014, + "grad_norm": 0.23100971707282483, + "learning_rate": 1.7506316289215005e-05, + "loss": 0.06857086718082428, + "step": 1647 + }, + { + "epoch": 0.921958041958042, + "grad_norm": 0.5844211296962387, + "learning_rate": 1.7502014480030533e-05, + "loss": 0.4726221561431885, + "step": 1648 + }, + { + "epoch": 0.9225174825174826, + "grad_norm": 0.5796908710123113, + "learning_rate": 1.7497709493071893e-05, + "loss": 0.29996445775032043, + "step": 1649 + }, + { + "epoch": 0.9230769230769231, + "grad_norm": 0.6047630602985847, + "learning_rate": 1.749340133016263e-05, + "loss": 0.33131396770477295, + "step": 1650 + }, + { + "epoch": 0.9236363636363636, + "grad_norm": 0.30668472658124013, + "learning_rate": 1.748908999312764e-05, + "loss": 0.10618998110294342, + "step": 1651 + }, + { + "epoch": 0.9241958041958042, + "grad_norm": 0.4661347471076679, + "learning_rate": 1.7484775483793158e-05, + "loss": 0.2302563637495041, + "step": 1652 + }, + { + "epoch": 0.9247552447552447, + "grad_norm": 0.6857938567486925, + "learning_rate": 1.7480457803986762e-05, + "loss": 0.4540604054927826, + "step": 1653 + }, + { + "epoch": 0.9253146853146853, + "grad_norm": 0.5815117064716567, + "learning_rate": 1.747613695553738e-05, + "loss": 0.349995881319046, + "step": 1654 + }, + { + "epoch": 0.9258741258741259, + "grad_norm": 0.5193310550307928, + "learning_rate": 1.747181294027527e-05, + "loss": 0.284639835357666, + "step": 1655 + }, + { + "epoch": 0.9264335664335664, + "grad_norm": 0.4535889230179055, + "learning_rate": 1.746748576003205e-05, + "loss": 0.19863250851631165, + "step": 1656 + }, + { + "epoch": 0.926993006993007, + "grad_norm": 0.3134041218480337, + "learning_rate": 1.746315541664066e-05, + "loss": 0.03898322209715843, + "step": 1657 + }, + { + "epoch": 0.9275524475524476, + "grad_norm": 0.34315196090314715, + "learning_rate": 1.7458821911935387e-05, + "loss": 0.19983133673667908, + "step": 1658 + }, + { + "epoch": 0.9281118881118882, + "grad_norm": 0.32375806690052866, + "learning_rate": 1.7454485247751868e-05, + "loss": 0.06280991435050964, + "step": 1659 + }, + { + "epoch": 0.9286713286713286, + "grad_norm": 0.4176313013935767, + "learning_rate": 1.7450145425927055e-05, + "loss": 0.1316758096218109, + "step": 1660 + }, + { + "epoch": 0.9292307692307692, + "grad_norm": 0.3851915413950685, + "learning_rate": 1.744580244829926e-05, + "loss": 0.14309416711330414, + "step": 1661 + }, + { + "epoch": 0.9297902097902098, + "grad_norm": 0.8331025315962842, + "learning_rate": 1.7441456316708118e-05, + "loss": 0.17261527478694916, + "step": 1662 + }, + { + "epoch": 0.9303496503496503, + "grad_norm": 0.4487320615389623, + "learning_rate": 1.743710703299461e-05, + "loss": 0.20808139443397522, + "step": 1663 + }, + { + "epoch": 0.9309090909090909, + "grad_norm": 0.551432037707246, + "learning_rate": 1.7432754599001037e-05, + "loss": 0.23567354679107666, + "step": 1664 + }, + { + "epoch": 0.9314685314685315, + "grad_norm": 0.5074235647497862, + "learning_rate": 1.7428399016571052e-05, + "loss": 0.40661072731018066, + "step": 1665 + }, + { + "epoch": 0.932027972027972, + "grad_norm": 0.5008408662231422, + "learning_rate": 1.742404028754963e-05, + "loss": 0.2870525121688843, + "step": 1666 + }, + { + "epoch": 0.9325874125874126, + "grad_norm": 0.1697532222754173, + "learning_rate": 1.7419678413783088e-05, + "loss": 0.014616936445236206, + "step": 1667 + }, + { + "epoch": 0.9331468531468532, + "grad_norm": 0.4556815090804263, + "learning_rate": 1.7415313397119066e-05, + "loss": 0.19322210550308228, + "step": 1668 + }, + { + "epoch": 0.9337062937062937, + "grad_norm": 0.34846135871242945, + "learning_rate": 1.7410945239406533e-05, + "loss": 0.16863524913787842, + "step": 1669 + }, + { + "epoch": 0.9342657342657342, + "grad_norm": 0.46586652551969404, + "learning_rate": 1.7406573942495803e-05, + "loss": 0.20633713901042938, + "step": 1670 + }, + { + "epoch": 0.9348251748251748, + "grad_norm": 0.5844791316669373, + "learning_rate": 1.740219950823851e-05, + "loss": 0.402421236038208, + "step": 1671 + }, + { + "epoch": 0.9353846153846154, + "grad_norm": 0.5233668816420789, + "learning_rate": 1.7397821938487615e-05, + "loss": 0.3429017663002014, + "step": 1672 + }, + { + "epoch": 0.9359440559440559, + "grad_norm": 0.3594795067373239, + "learning_rate": 1.7393441235097417e-05, + "loss": 0.08628326654434204, + "step": 1673 + }, + { + "epoch": 0.9365034965034965, + "grad_norm": 0.40386091296514837, + "learning_rate": 1.7389057399923524e-05, + "loss": 0.1812334507703781, + "step": 1674 + }, + { + "epoch": 0.9370629370629371, + "grad_norm": 0.4468192878549662, + "learning_rate": 1.7384670434822894e-05, + "loss": 0.23956313729286194, + "step": 1675 + }, + { + "epoch": 0.9376223776223777, + "grad_norm": 0.38671082302018844, + "learning_rate": 1.738028034165379e-05, + "loss": 0.19388212263584137, + "step": 1676 + }, + { + "epoch": 0.9381818181818182, + "grad_norm": 0.34673674660231113, + "learning_rate": 1.7375887122275818e-05, + "loss": 0.07962428033351898, + "step": 1677 + }, + { + "epoch": 0.9387412587412587, + "grad_norm": 0.45205727192806505, + "learning_rate": 1.7371490778549894e-05, + "loss": 0.20845860242843628, + "step": 1678 + }, + { + "epoch": 0.9393006993006993, + "grad_norm": 0.4800124056140485, + "learning_rate": 1.7367091312338263e-05, + "loss": 0.21996483206748962, + "step": 1679 + }, + { + "epoch": 0.9398601398601398, + "grad_norm": 0.5339703933136452, + "learning_rate": 1.736268872550449e-05, + "loss": 0.35951757431030273, + "step": 1680 + }, + { + "epoch": 0.9404195804195804, + "grad_norm": 0.5361907528559541, + "learning_rate": 1.7358283019913473e-05, + "loss": 0.29795268177986145, + "step": 1681 + }, + { + "epoch": 0.940979020979021, + "grad_norm": 0.31632032163198665, + "learning_rate": 1.7353874197431412e-05, + "loss": 0.11087039858102798, + "step": 1682 + }, + { + "epoch": 0.9415384615384615, + "grad_norm": 0.32977922929900966, + "learning_rate": 1.7349462259925846e-05, + "loss": 0.10425370931625366, + "step": 1683 + }, + { + "epoch": 0.9420979020979021, + "grad_norm": 0.36623137391877814, + "learning_rate": 1.7345047209265618e-05, + "loss": 0.12108875811100006, + "step": 1684 + }, + { + "epoch": 0.9426573426573427, + "grad_norm": 0.480270919642893, + "learning_rate": 1.7340629047320906e-05, + "loss": 0.32281485199928284, + "step": 1685 + }, + { + "epoch": 0.9432167832167833, + "grad_norm": 0.42272282916646314, + "learning_rate": 1.7336207775963184e-05, + "loss": 0.12829357385635376, + "step": 1686 + }, + { + "epoch": 0.9437762237762237, + "grad_norm": 0.4247855063630827, + "learning_rate": 1.7331783397065267e-05, + "loss": 0.19423940777778625, + "step": 1687 + }, + { + "epoch": 0.9443356643356643, + "grad_norm": 0.27327994474477457, + "learning_rate": 1.7327355912501272e-05, + "loss": 0.09307831525802612, + "step": 1688 + }, + { + "epoch": 0.9448951048951049, + "grad_norm": 0.4598001058779315, + "learning_rate": 1.7322925324146632e-05, + "loss": 0.20622305572032928, + "step": 1689 + }, + { + "epoch": 0.9454545454545454, + "grad_norm": 0.5185409402430922, + "learning_rate": 1.73184916338781e-05, + "loss": 0.16810685396194458, + "step": 1690 + }, + { + "epoch": 0.946013986013986, + "grad_norm": 0.2709856920388912, + "learning_rate": 1.7314054843573744e-05, + "loss": 0.1370825469493866, + "step": 1691 + }, + { + "epoch": 0.9465734265734266, + "grad_norm": 0.3652957532475193, + "learning_rate": 1.7309614955112934e-05, + "loss": 0.1516319215297699, + "step": 1692 + }, + { + "epoch": 0.9471328671328672, + "grad_norm": 0.37195720916528713, + "learning_rate": 1.7305171970376363e-05, + "loss": 0.1676393747329712, + "step": 1693 + }, + { + "epoch": 0.9476923076923077, + "grad_norm": 0.6650740577002039, + "learning_rate": 1.7300725891246034e-05, + "loss": 0.24507887661457062, + "step": 1694 + }, + { + "epoch": 0.9482517482517483, + "grad_norm": 0.4104593747255048, + "learning_rate": 1.729627671960526e-05, + "loss": 0.17468124628067017, + "step": 1695 + }, + { + "epoch": 0.9488111888111889, + "grad_norm": 0.6878953294945831, + "learning_rate": 1.7291824457338657e-05, + "loss": 0.32875490188598633, + "step": 1696 + }, + { + "epoch": 0.9493706293706293, + "grad_norm": 0.5494551520268461, + "learning_rate": 1.7287369106332164e-05, + "loss": 0.27696138620376587, + "step": 1697 + }, + { + "epoch": 0.9499300699300699, + "grad_norm": 0.42856013001731785, + "learning_rate": 1.7282910668473016e-05, + "loss": 0.15715940296649933, + "step": 1698 + }, + { + "epoch": 0.9504895104895105, + "grad_norm": 0.5309501541566418, + "learning_rate": 1.7278449145649763e-05, + "loss": 0.27758514881134033, + "step": 1699 + }, + { + "epoch": 0.951048951048951, + "grad_norm": 0.4960829433801541, + "learning_rate": 1.7273984539752254e-05, + "loss": 0.21390266716480255, + "step": 1700 + }, + { + "epoch": 0.9516083916083916, + "grad_norm": 0.33070841674654544, + "learning_rate": 1.7269516852671654e-05, + "loss": 0.14004793763160706, + "step": 1701 + }, + { + "epoch": 0.9521678321678322, + "grad_norm": 0.4511983365900374, + "learning_rate": 1.7265046086300427e-05, + "loss": 0.15599679946899414, + "step": 1702 + }, + { + "epoch": 0.9527272727272728, + "grad_norm": 0.32540936456167463, + "learning_rate": 1.7260572242532345e-05, + "loss": 0.11955723166465759, + "step": 1703 + }, + { + "epoch": 0.9532867132867133, + "grad_norm": 0.5184428997839865, + "learning_rate": 1.7256095323262476e-05, + "loss": 0.28404897451400757, + "step": 1704 + }, + { + "epoch": 0.9538461538461539, + "grad_norm": 0.34676684841719435, + "learning_rate": 1.7251615330387194e-05, + "loss": 0.14715871214866638, + "step": 1705 + }, + { + "epoch": 0.9544055944055944, + "grad_norm": 0.6045909093111844, + "learning_rate": 1.7247132265804184e-05, + "loss": 0.3182292878627777, + "step": 1706 + }, + { + "epoch": 0.9549650349650349, + "grad_norm": 0.3454559873444579, + "learning_rate": 1.724264613141242e-05, + "loss": 0.1009145975112915, + "step": 1707 + }, + { + "epoch": 0.9555244755244755, + "grad_norm": 0.5160159699705267, + "learning_rate": 1.7238156929112185e-05, + "loss": 0.28568771481513977, + "step": 1708 + }, + { + "epoch": 0.9560839160839161, + "grad_norm": 0.3713600111559413, + "learning_rate": 1.7233664660805055e-05, + "loss": 0.1518329530954361, + "step": 1709 + }, + { + "epoch": 0.9566433566433566, + "grad_norm": 0.459414630840221, + "learning_rate": 1.7229169328393904e-05, + "loss": 0.28611087799072266, + "step": 1710 + }, + { + "epoch": 0.9572027972027972, + "grad_norm": 0.49467038569685046, + "learning_rate": 1.7224670933782913e-05, + "loss": 0.155736044049263, + "step": 1711 + }, + { + "epoch": 0.9577622377622378, + "grad_norm": 0.3623198720099286, + "learning_rate": 1.7220169478877554e-05, + "loss": 0.07055732607841492, + "step": 1712 + }, + { + "epoch": 0.9583216783216784, + "grad_norm": 0.4398538825168541, + "learning_rate": 1.7215664965584593e-05, + "loss": 0.12250161170959473, + "step": 1713 + }, + { + "epoch": 0.9588811188811189, + "grad_norm": 0.4514437758809277, + "learning_rate": 1.7211157395812094e-05, + "loss": 0.2894744873046875, + "step": 1714 + }, + { + "epoch": 0.9594405594405594, + "grad_norm": 0.4216403960186692, + "learning_rate": 1.720664677146942e-05, + "loss": 0.2420210987329483, + "step": 1715 + }, + { + "epoch": 0.96, + "grad_norm": 0.5475729430729264, + "learning_rate": 1.7202133094467223e-05, + "loss": 0.30071884393692017, + "step": 1716 + }, + { + "epoch": 0.9605594405594405, + "grad_norm": 0.28823687431042233, + "learning_rate": 1.7197616366717446e-05, + "loss": 0.06418092548847198, + "step": 1717 + }, + { + "epoch": 0.9611188811188811, + "grad_norm": 0.5165040411886876, + "learning_rate": 1.7193096590133325e-05, + "loss": 0.23366746306419373, + "step": 1718 + }, + { + "epoch": 0.9616783216783217, + "grad_norm": 0.41556826594322427, + "learning_rate": 1.7188573766629398e-05, + "loss": 0.13986369967460632, + "step": 1719 + }, + { + "epoch": 0.9622377622377623, + "grad_norm": 0.34509945889310506, + "learning_rate": 1.7184047898121477e-05, + "loss": 0.10744775831699371, + "step": 1720 + }, + { + "epoch": 0.9627972027972028, + "grad_norm": 0.51628305668275, + "learning_rate": 1.7179518986526676e-05, + "loss": 0.18372637033462524, + "step": 1721 + }, + { + "epoch": 0.9633566433566434, + "grad_norm": 0.4499673686694797, + "learning_rate": 1.71749870337634e-05, + "loss": 0.17784248292446136, + "step": 1722 + }, + { + "epoch": 0.963916083916084, + "grad_norm": 0.4869018160645804, + "learning_rate": 1.7170452041751328e-05, + "loss": 0.17018987238407135, + "step": 1723 + }, + { + "epoch": 0.9644755244755244, + "grad_norm": 0.4803097998608346, + "learning_rate": 1.716591401241143e-05, + "loss": 0.1566029191017151, + "step": 1724 + }, + { + "epoch": 0.965034965034965, + "grad_norm": 0.5089402968079474, + "learning_rate": 1.7161372947665982e-05, + "loss": 0.2766360640525818, + "step": 1725 + }, + { + "epoch": 0.9655944055944056, + "grad_norm": 0.28155329565172865, + "learning_rate": 1.715682884943852e-05, + "loss": 0.08040576428174973, + "step": 1726 + }, + { + "epoch": 0.9661538461538461, + "grad_norm": 0.37838396994517237, + "learning_rate": 1.715228171965388e-05, + "loss": 0.19021615386009216, + "step": 1727 + }, + { + "epoch": 0.9667132867132867, + "grad_norm": 0.44329848105270087, + "learning_rate": 1.714773156023818e-05, + "loss": 0.204390287399292, + "step": 1728 + }, + { + "epoch": 0.9672727272727273, + "grad_norm": 1.65856414147918, + "learning_rate": 1.7143178373118815e-05, + "loss": 0.3924446105957031, + "step": 1729 + }, + { + "epoch": 0.9678321678321679, + "grad_norm": 0.610508831618603, + "learning_rate": 1.713862216022447e-05, + "loss": 0.3880074620246887, + "step": 1730 + }, + { + "epoch": 0.9683916083916084, + "grad_norm": 0.47137033835576614, + "learning_rate": 1.7134062923485106e-05, + "loss": 0.22814473509788513, + "step": 1731 + }, + { + "epoch": 0.968951048951049, + "grad_norm": 0.5503895688551688, + "learning_rate": 1.7129500664831972e-05, + "loss": 0.2966257631778717, + "step": 1732 + }, + { + "epoch": 0.9695104895104895, + "grad_norm": 1.0094040777173745, + "learning_rate": 1.712493538619759e-05, + "loss": 0.2558753192424774, + "step": 1733 + }, + { + "epoch": 0.97006993006993, + "grad_norm": 0.5329545328727171, + "learning_rate": 1.7120367089515763e-05, + "loss": 0.3099202513694763, + "step": 1734 + }, + { + "epoch": 0.9706293706293706, + "grad_norm": 0.26833083573364214, + "learning_rate": 1.7115795776721575e-05, + "loss": 0.09802938997745514, + "step": 1735 + }, + { + "epoch": 0.9711888111888112, + "grad_norm": 0.2580635292678601, + "learning_rate": 1.7111221449751385e-05, + "loss": 0.10171899944543839, + "step": 1736 + }, + { + "epoch": 0.9717482517482517, + "grad_norm": 0.41938537039253193, + "learning_rate": 1.710664411054283e-05, + "loss": 0.12753914296627045, + "step": 1737 + }, + { + "epoch": 0.9723076923076923, + "grad_norm": 0.4857135402008204, + "learning_rate": 1.7102063761034823e-05, + "loss": 0.2561277151107788, + "step": 1738 + }, + { + "epoch": 0.9728671328671329, + "grad_norm": 0.6705812424429759, + "learning_rate": 1.709748040316755e-05, + "loss": 0.28080281615257263, + "step": 1739 + }, + { + "epoch": 0.9734265734265735, + "grad_norm": 0.4448489639942743, + "learning_rate": 1.7092894038882475e-05, + "loss": 0.22109544277191162, + "step": 1740 + }, + { + "epoch": 0.973986013986014, + "grad_norm": 0.3764824299793923, + "learning_rate": 1.708830467012233e-05, + "loss": 0.21845363080501556, + "step": 1741 + }, + { + "epoch": 0.9745454545454545, + "grad_norm": 0.30874744303819807, + "learning_rate": 1.708371229883113e-05, + "loss": 0.13033252954483032, + "step": 1742 + }, + { + "epoch": 0.9751048951048951, + "grad_norm": 0.40152379682962663, + "learning_rate": 1.7079116926954146e-05, + "loss": 0.15591920912265778, + "step": 1743 + }, + { + "epoch": 0.9756643356643356, + "grad_norm": 0.5423496540237306, + "learning_rate": 1.707451855643794e-05, + "loss": 0.2553369402885437, + "step": 1744 + }, + { + "epoch": 0.9762237762237762, + "grad_norm": 0.8277287185793444, + "learning_rate": 1.7069917189230322e-05, + "loss": 0.3380037546157837, + "step": 1745 + }, + { + "epoch": 0.9767832167832168, + "grad_norm": 0.1605213647372991, + "learning_rate": 1.706531282728039e-05, + "loss": 0.0386936292052269, + "step": 1746 + }, + { + "epoch": 0.9773426573426574, + "grad_norm": 0.7149660440404692, + "learning_rate": 1.70607054725385e-05, + "loss": 0.428783118724823, + "step": 1747 + }, + { + "epoch": 0.9779020979020979, + "grad_norm": 0.5289477930133607, + "learning_rate": 1.705609512695628e-05, + "loss": 0.2550695538520813, + "step": 1748 + }, + { + "epoch": 0.9784615384615385, + "grad_norm": 0.47282482666127, + "learning_rate": 1.7051481792486627e-05, + "loss": 0.1777503788471222, + "step": 1749 + }, + { + "epoch": 0.9790209790209791, + "grad_norm": 0.500420094603535, + "learning_rate": 1.704686547108369e-05, + "loss": 0.20227137207984924, + "step": 1750 + }, + { + "epoch": 0.9795804195804195, + "grad_norm": 0.35676174250153575, + "learning_rate": 1.7042246164702904e-05, + "loss": 0.10227454453706741, + "step": 1751 + }, + { + "epoch": 0.9801398601398601, + "grad_norm": 0.4826315246965562, + "learning_rate": 1.7037623875300958e-05, + "loss": 0.29930809140205383, + "step": 1752 + }, + { + "epoch": 0.9806993006993007, + "grad_norm": 0.278361218312106, + "learning_rate": 1.70329986048358e-05, + "loss": 0.03869568556547165, + "step": 1753 + }, + { + "epoch": 0.9812587412587412, + "grad_norm": 0.497087247430351, + "learning_rate": 1.702837035526665e-05, + "loss": 0.17714332044124603, + "step": 1754 + }, + { + "epoch": 0.9818181818181818, + "grad_norm": 0.19646495917555637, + "learning_rate": 1.7023739128553988e-05, + "loss": 0.04542902857065201, + "step": 1755 + }, + { + "epoch": 0.9823776223776224, + "grad_norm": 0.4470227012345435, + "learning_rate": 1.7019104926659542e-05, + "loss": 0.16884341835975647, + "step": 1756 + }, + { + "epoch": 0.982937062937063, + "grad_norm": 0.3168980082681729, + "learning_rate": 1.701446775154632e-05, + "loss": 0.11557105928659439, + "step": 1757 + }, + { + "epoch": 0.9834965034965035, + "grad_norm": 0.46855270772215946, + "learning_rate": 1.7009827605178578e-05, + "loss": 0.2557441294193268, + "step": 1758 + }, + { + "epoch": 0.9840559440559441, + "grad_norm": 0.6624809116301464, + "learning_rate": 1.700518448952184e-05, + "loss": 0.18762089312076569, + "step": 1759 + }, + { + "epoch": 0.9846153846153847, + "grad_norm": 0.351617789439919, + "learning_rate": 1.7000538406542863e-05, + "loss": 0.12941545248031616, + "step": 1760 + }, + { + "epoch": 0.9851748251748251, + "grad_norm": 0.3922542053678613, + "learning_rate": 1.69958893582097e-05, + "loss": 0.19491036236286163, + "step": 1761 + }, + { + "epoch": 0.9857342657342657, + "grad_norm": 0.409538311638206, + "learning_rate": 1.6991237346491623e-05, + "loss": 0.18471840023994446, + "step": 1762 + }, + { + "epoch": 0.9862937062937063, + "grad_norm": 0.5529121053076465, + "learning_rate": 1.6986582373359182e-05, + "loss": 0.3501903712749481, + "step": 1763 + }, + { + "epoch": 0.9868531468531468, + "grad_norm": 0.4000186600291675, + "learning_rate": 1.6981924440784175e-05, + "loss": 0.11872689425945282, + "step": 1764 + }, + { + "epoch": 0.9874125874125874, + "grad_norm": 0.524414512540367, + "learning_rate": 1.697726355073965e-05, + "loss": 0.19359099864959717, + "step": 1765 + }, + { + "epoch": 0.987972027972028, + "grad_norm": 0.34442064735545996, + "learning_rate": 1.697259970519991e-05, + "loss": 0.13528797030448914, + "step": 1766 + }, + { + "epoch": 0.9885314685314686, + "grad_norm": 0.43827230037968123, + "learning_rate": 1.6967932906140517e-05, + "loss": 0.1479535698890686, + "step": 1767 + }, + { + "epoch": 0.9890909090909091, + "grad_norm": 0.4130260352032846, + "learning_rate": 1.696326315553827e-05, + "loss": 0.23039215803146362, + "step": 1768 + }, + { + "epoch": 0.9896503496503497, + "grad_norm": 0.28421507900243104, + "learning_rate": 1.695859045537123e-05, + "loss": 0.09387499839067459, + "step": 1769 + }, + { + "epoch": 0.9902097902097902, + "grad_norm": 0.3965703694473084, + "learning_rate": 1.6953914807618707e-05, + "loss": 0.233891099691391, + "step": 1770 + }, + { + "epoch": 0.9907692307692307, + "grad_norm": 0.5972956075810059, + "learning_rate": 1.6949236214261252e-05, + "loss": 0.20743700861930847, + "step": 1771 + }, + { + "epoch": 0.9913286713286713, + "grad_norm": 0.46205614719610016, + "learning_rate": 1.6944554677280668e-05, + "loss": 0.20164258778095245, + "step": 1772 + }, + { + "epoch": 0.9918881118881119, + "grad_norm": 0.3808706432948349, + "learning_rate": 1.6939870198660005e-05, + "loss": 0.15108591318130493, + "step": 1773 + }, + { + "epoch": 0.9924475524475525, + "grad_norm": 0.5939299293176771, + "learning_rate": 1.6935182780383558e-05, + "loss": 0.21661114692687988, + "step": 1774 + }, + { + "epoch": 0.993006993006993, + "grad_norm": 0.7654951754379522, + "learning_rate": 1.6930492424436864e-05, + "loss": 0.48865288496017456, + "step": 1775 + }, + { + "epoch": 0.9935664335664336, + "grad_norm": 0.4165784821725244, + "learning_rate": 1.692579913280672e-05, + "loss": 0.17697106301784515, + "step": 1776 + }, + { + "epoch": 0.9941258741258742, + "grad_norm": 0.4853944152416085, + "learning_rate": 1.692110290748114e-05, + "loss": 0.2752719223499298, + "step": 1777 + }, + { + "epoch": 0.9946853146853147, + "grad_norm": 0.345117729592068, + "learning_rate": 1.691640375044941e-05, + "loss": 0.12802013754844666, + "step": 1778 + }, + { + "epoch": 0.9952447552447552, + "grad_norm": 0.5403692799831695, + "learning_rate": 1.6911701663702033e-05, + "loss": 0.25361230969429016, + "step": 1779 + }, + { + "epoch": 0.9958041958041958, + "grad_norm": 0.5102029396700746, + "learning_rate": 1.6906996649230766e-05, + "loss": 0.2693482041358948, + "step": 1780 + }, + { + "epoch": 0.9963636363636363, + "grad_norm": 0.7084534575849529, + "learning_rate": 1.690228870902861e-05, + "loss": 0.34110140800476074, + "step": 1781 + }, + { + "epoch": 0.9969230769230769, + "grad_norm": 0.6502350055732851, + "learning_rate": 1.6897577845089785e-05, + "loss": 0.45697373151779175, + "step": 1782 + }, + { + "epoch": 0.9974825174825175, + "grad_norm": 0.4511493205815134, + "learning_rate": 1.689286405940978e-05, + "loss": 0.33918535709381104, + "step": 1783 + }, + { + "epoch": 0.998041958041958, + "grad_norm": 0.33569907848732466, + "learning_rate": 1.6888147353985288e-05, + "loss": 0.11391283571720123, + "step": 1784 + }, + { + "epoch": 0.9986013986013986, + "grad_norm": 0.3653182870124323, + "learning_rate": 1.688342773081427e-05, + "loss": 0.0903647318482399, + "step": 1785 + }, + { + "epoch": 0.9991608391608392, + "grad_norm": 0.437678342233518, + "learning_rate": 1.6878705191895895e-05, + "loss": 0.122248575091362, + "step": 1786 + }, + { + "epoch": 0.9997202797202798, + "grad_norm": 0.42463311206974425, + "learning_rate": 1.6873979739230594e-05, + "loss": 0.2232385128736496, + "step": 1787 + }, + { + "epoch": 1.0, + "grad_norm": 0.7338525808204936, + "learning_rate": 1.686925137482001e-05, + "loss": 0.25163209438323975, + "step": 1788 + }, + { + "epoch": 1.0005594405594405, + "grad_norm": 0.31274922450030174, + "learning_rate": 1.6864520100667036e-05, + "loss": 0.08295510709285736, + "step": 1789 + }, + { + "epoch": 1.0011188811188811, + "grad_norm": 0.3686581937957675, + "learning_rate": 1.685978591877578e-05, + "loss": 0.1294628232717514, + "step": 1790 + }, + { + "epoch": 1.0016783216783216, + "grad_norm": 0.41293603468097995, + "learning_rate": 1.6855048831151596e-05, + "loss": 0.11114399135112762, + "step": 1791 + }, + { + "epoch": 1.0022377622377623, + "grad_norm": 0.1892389225529675, + "learning_rate": 1.6850308839801067e-05, + "loss": 0.12066492438316345, + "step": 1792 + }, + { + "epoch": 1.0027972027972027, + "grad_norm": 0.4297824672040896, + "learning_rate": 1.6845565946732003e-05, + "loss": 0.20680563151836395, + "step": 1793 + }, + { + "epoch": 1.0033566433566434, + "grad_norm": 0.5154768754155762, + "learning_rate": 1.684082015395344e-05, + "loss": 0.20539617538452148, + "step": 1794 + }, + { + "epoch": 1.003916083916084, + "grad_norm": 0.3135760508385155, + "learning_rate": 1.6836071463475646e-05, + "loss": 0.09164722263813019, + "step": 1795 + }, + { + "epoch": 1.0044755244755246, + "grad_norm": 0.33335217026376324, + "learning_rate": 1.6831319877310118e-05, + "loss": 0.13413432240486145, + "step": 1796 + }, + { + "epoch": 1.005034965034965, + "grad_norm": 0.41057516417055934, + "learning_rate": 1.6826565397469577e-05, + "loss": 0.21825465559959412, + "step": 1797 + }, + { + "epoch": 1.0055944055944055, + "grad_norm": 0.22219765169149122, + "learning_rate": 1.682180802596797e-05, + "loss": 0.007302472833544016, + "step": 1798 + }, + { + "epoch": 1.0061538461538462, + "grad_norm": 0.44967957376762835, + "learning_rate": 1.6817047764820472e-05, + "loss": 0.18644356727600098, + "step": 1799 + }, + { + "epoch": 1.0067132867132866, + "grad_norm": 0.44195374557893075, + "learning_rate": 1.6812284616043476e-05, + "loss": 0.1517438292503357, + "step": 1800 + }, + { + "epoch": 1.0072727272727273, + "grad_norm": 0.39826467749557426, + "learning_rate": 1.6807518581654605e-05, + "loss": 0.10193042457103729, + "step": 1801 + }, + { + "epoch": 1.0078321678321678, + "grad_norm": 0.672798149583607, + "learning_rate": 1.6802749663672698e-05, + "loss": 0.19434235990047455, + "step": 1802 + }, + { + "epoch": 1.0083916083916085, + "grad_norm": 0.3457327927604612, + "learning_rate": 1.679797786411782e-05, + "loss": 0.06855184584856033, + "step": 1803 + }, + { + "epoch": 1.008951048951049, + "grad_norm": 0.5076851430182343, + "learning_rate": 1.6793203185011252e-05, + "loss": 0.08057352155447006, + "step": 1804 + }, + { + "epoch": 1.0095104895104896, + "grad_norm": 0.38792493081162954, + "learning_rate": 1.6788425628375504e-05, + "loss": 0.06162555515766144, + "step": 1805 + }, + { + "epoch": 1.01006993006993, + "grad_norm": 1.810359218345408, + "learning_rate": 1.678364519623429e-05, + "loss": 0.4912581443786621, + "step": 1806 + }, + { + "epoch": 1.0106293706293705, + "grad_norm": 0.5364762747246555, + "learning_rate": 1.6778861890612558e-05, + "loss": 0.1503697633743286, + "step": 1807 + }, + { + "epoch": 1.0111888111888112, + "grad_norm": 0.672211704513886, + "learning_rate": 1.677407571353646e-05, + "loss": 0.20396408438682556, + "step": 1808 + }, + { + "epoch": 1.0117482517482517, + "grad_norm": 0.5461627804182001, + "learning_rate": 1.6769286667033374e-05, + "loss": 0.15781229734420776, + "step": 1809 + }, + { + "epoch": 1.0123076923076924, + "grad_norm": 0.6442909507377969, + "learning_rate": 1.6764494753131887e-05, + "loss": 0.216548889875412, + "step": 1810 + }, + { + "epoch": 1.0128671328671328, + "grad_norm": 0.2367161028084213, + "learning_rate": 1.6759699973861808e-05, + "loss": 0.05147993564605713, + "step": 1811 + }, + { + "epoch": 1.0134265734265735, + "grad_norm": 0.5272309973692981, + "learning_rate": 1.6754902331254152e-05, + "loss": 0.14214424788951874, + "step": 1812 + }, + { + "epoch": 1.013986013986014, + "grad_norm": 0.5566237772564367, + "learning_rate": 1.6750101827341146e-05, + "loss": 0.17502613365650177, + "step": 1813 + }, + { + "epoch": 1.0145454545454546, + "grad_norm": 0.5608806001688716, + "learning_rate": 1.6745298464156235e-05, + "loss": 0.15899983048439026, + "step": 1814 + }, + { + "epoch": 1.015104895104895, + "grad_norm": 0.4845941123177395, + "learning_rate": 1.6740492243734078e-05, + "loss": 0.15133816003799438, + "step": 1815 + }, + { + "epoch": 1.0156643356643356, + "grad_norm": 0.4400608472548964, + "learning_rate": 1.6735683168110534e-05, + "loss": 0.08814603090286255, + "step": 1816 + }, + { + "epoch": 1.0162237762237762, + "grad_norm": 0.6917079820425519, + "learning_rate": 1.6730871239322678e-05, + "loss": 0.25201737880706787, + "step": 1817 + }, + { + "epoch": 1.0167832167832167, + "grad_norm": 0.5384334287162728, + "learning_rate": 1.6726056459408792e-05, + "loss": 0.18676358461380005, + "step": 1818 + }, + { + "epoch": 1.0173426573426574, + "grad_norm": 0.4255021351191568, + "learning_rate": 1.6721238830408366e-05, + "loss": 0.12932777404785156, + "step": 1819 + }, + { + "epoch": 1.0179020979020978, + "grad_norm": 0.449611941872502, + "learning_rate": 1.67164183543621e-05, + "loss": 0.17120352387428284, + "step": 1820 + }, + { + "epoch": 1.0184615384615385, + "grad_norm": 0.34756198044777076, + "learning_rate": 1.6711595033311892e-05, + "loss": 0.07034595310688019, + "step": 1821 + }, + { + "epoch": 1.019020979020979, + "grad_norm": 0.4065409256247498, + "learning_rate": 1.670676886930085e-05, + "loss": 0.12487390637397766, + "step": 1822 + }, + { + "epoch": 1.0195804195804197, + "grad_norm": 0.44775124665178573, + "learning_rate": 1.670193986437329e-05, + "loss": 0.15174169838428497, + "step": 1823 + }, + { + "epoch": 1.0201398601398601, + "grad_norm": 0.4597424048543423, + "learning_rate": 1.6697108020574723e-05, + "loss": 0.08087201416492462, + "step": 1824 + }, + { + "epoch": 1.0206993006993006, + "grad_norm": 0.2956980180617255, + "learning_rate": 1.6692273339951875e-05, + "loss": 0.10425123572349548, + "step": 1825 + }, + { + "epoch": 1.0212587412587413, + "grad_norm": 0.48840769718157717, + "learning_rate": 1.6687435824552653e-05, + "loss": 0.20766519010066986, + "step": 1826 + }, + { + "epoch": 1.0218181818181817, + "grad_norm": 0.2917470663791056, + "learning_rate": 1.6682595476426188e-05, + "loss": 0.09538062661886215, + "step": 1827 + }, + { + "epoch": 1.0223776223776224, + "grad_norm": 0.656522925020085, + "learning_rate": 1.6677752297622797e-05, + "loss": 0.1688915640115738, + "step": 1828 + }, + { + "epoch": 1.0229370629370629, + "grad_norm": 0.6009483137325978, + "learning_rate": 1.6672906290194003e-05, + "loss": 0.254028856754303, + "step": 1829 + }, + { + "epoch": 1.0234965034965036, + "grad_norm": 0.2920973690415481, + "learning_rate": 1.6668057456192514e-05, + "loss": 0.04055303707718849, + "step": 1830 + }, + { + "epoch": 1.024055944055944, + "grad_norm": 0.5785224818339686, + "learning_rate": 1.666320579767225e-05, + "loss": 0.12503540515899658, + "step": 1831 + }, + { + "epoch": 1.0246153846153847, + "grad_norm": 0.4691115806034068, + "learning_rate": 1.665835131668833e-05, + "loss": 0.1253487914800644, + "step": 1832 + }, + { + "epoch": 1.0251748251748252, + "grad_norm": 0.4989225062230292, + "learning_rate": 1.6653494015297055e-05, + "loss": 0.168176531791687, + "step": 1833 + }, + { + "epoch": 1.0257342657342656, + "grad_norm": 0.5048823401558591, + "learning_rate": 1.664863389555592e-05, + "loss": 0.1692991852760315, + "step": 1834 + }, + { + "epoch": 1.0262937062937063, + "grad_norm": 0.3137039931050128, + "learning_rate": 1.6643770959523628e-05, + "loss": 0.09577256441116333, + "step": 1835 + }, + { + "epoch": 1.0268531468531468, + "grad_norm": 0.23264938890471074, + "learning_rate": 1.6638905209260067e-05, + "loss": 0.04955904185771942, + "step": 1836 + }, + { + "epoch": 1.0274125874125875, + "grad_norm": 0.4336664040861078, + "learning_rate": 1.6634036646826314e-05, + "loss": 0.13321664929389954, + "step": 1837 + }, + { + "epoch": 1.027972027972028, + "grad_norm": 0.49316287556922833, + "learning_rate": 1.6629165274284648e-05, + "loss": 0.14160437881946564, + "step": 1838 + }, + { + "epoch": 1.0285314685314686, + "grad_norm": 0.42774539067604483, + "learning_rate": 1.662429109369852e-05, + "loss": 0.13516326248645782, + "step": 1839 + }, + { + "epoch": 1.029090909090909, + "grad_norm": 0.1620620946894935, + "learning_rate": 1.661941410713259e-05, + "loss": 0.014976169914007187, + "step": 1840 + }, + { + "epoch": 1.0296503496503497, + "grad_norm": 0.4040477294780048, + "learning_rate": 1.6614534316652694e-05, + "loss": 0.10721161961555481, + "step": 1841 + }, + { + "epoch": 1.0302097902097902, + "grad_norm": 0.3461431907779022, + "learning_rate": 1.6609651724325863e-05, + "loss": 0.05441214144229889, + "step": 1842 + }, + { + "epoch": 1.0307692307692307, + "grad_norm": 0.5273727352482188, + "learning_rate": 1.6604766332220306e-05, + "loss": 0.08860521018505096, + "step": 1843 + }, + { + "epoch": 1.0313286713286713, + "grad_norm": 0.606850132655898, + "learning_rate": 1.659987814240543e-05, + "loss": 0.22397495806217194, + "step": 1844 + }, + { + "epoch": 1.0318881118881118, + "grad_norm": 0.48869387340968873, + "learning_rate": 1.659498715695182e-05, + "loss": 0.1872205138206482, + "step": 1845 + }, + { + "epoch": 1.0324475524475525, + "grad_norm": 0.6079784752382916, + "learning_rate": 1.659009337793124e-05, + "loss": 0.0827154889702797, + "step": 1846 + }, + { + "epoch": 1.033006993006993, + "grad_norm": 0.3745902561272969, + "learning_rate": 1.658519680741665e-05, + "loss": 0.09823012351989746, + "step": 1847 + }, + { + "epoch": 1.0335664335664336, + "grad_norm": 1.8857506518378877, + "learning_rate": 1.658029744748218e-05, + "loss": 0.09591659903526306, + "step": 1848 + }, + { + "epoch": 1.034125874125874, + "grad_norm": 0.302123456667023, + "learning_rate": 1.657539530020315e-05, + "loss": 0.04960941523313522, + "step": 1849 + }, + { + "epoch": 1.0346853146853148, + "grad_norm": 0.423886838596637, + "learning_rate": 1.657049036765606e-05, + "loss": 0.10362362861633301, + "step": 1850 + }, + { + "epoch": 1.0352447552447552, + "grad_norm": 0.5007858063260234, + "learning_rate": 1.656558265191858e-05, + "loss": 0.15144553780555725, + "step": 1851 + }, + { + "epoch": 1.0358041958041957, + "grad_norm": 0.38882430935777873, + "learning_rate": 1.6560672155069573e-05, + "loss": 0.07592298090457916, + "step": 1852 + }, + { + "epoch": 1.0363636363636364, + "grad_norm": 0.6294855673224956, + "learning_rate": 1.6555758879189076e-05, + "loss": 0.1961602121591568, + "step": 1853 + }, + { + "epoch": 1.0369230769230768, + "grad_norm": 0.5671338435113489, + "learning_rate": 1.6550842826358295e-05, + "loss": 0.21520818769931793, + "step": 1854 + }, + { + "epoch": 1.0374825174825175, + "grad_norm": 0.48116600686868166, + "learning_rate": 1.654592399865962e-05, + "loss": 0.1701516956090927, + "step": 1855 + }, + { + "epoch": 1.038041958041958, + "grad_norm": 0.27577189594915974, + "learning_rate": 1.6541002398176614e-05, + "loss": 0.034923385828733444, + "step": 1856 + }, + { + "epoch": 1.0386013986013987, + "grad_norm": 0.5035658218808757, + "learning_rate": 1.6536078026994017e-05, + "loss": 0.1235581561923027, + "step": 1857 + }, + { + "epoch": 1.0391608391608391, + "grad_norm": 0.5715387525286217, + "learning_rate": 1.6531150887197744e-05, + "loss": 0.25293564796447754, + "step": 1858 + }, + { + "epoch": 1.0397202797202798, + "grad_norm": 0.3615970020913046, + "learning_rate": 1.652622098087487e-05, + "loss": 0.10524363070726395, + "step": 1859 + }, + { + "epoch": 1.0402797202797203, + "grad_norm": 0.267492897356417, + "learning_rate": 1.6521288310113654e-05, + "loss": 0.028555942699313164, + "step": 1860 + }, + { + "epoch": 1.0408391608391607, + "grad_norm": 0.500648808317561, + "learning_rate": 1.651635287700353e-05, + "loss": 0.13310854136943817, + "step": 1861 + }, + { + "epoch": 1.0413986013986014, + "grad_norm": 1.1607233074559613, + "learning_rate": 1.651141468363509e-05, + "loss": 0.2666672468185425, + "step": 1862 + }, + { + "epoch": 1.0419580419580419, + "grad_norm": 0.5650275356462355, + "learning_rate": 1.65064737321001e-05, + "loss": 0.22971510887145996, + "step": 1863 + }, + { + "epoch": 1.0425174825174826, + "grad_norm": 0.544926136635955, + "learning_rate": 1.6501530024491493e-05, + "loss": 0.19471578299999237, + "step": 1864 + }, + { + "epoch": 1.043076923076923, + "grad_norm": 0.4116756229575657, + "learning_rate": 1.6496583562903383e-05, + "loss": 0.07229534536600113, + "step": 1865 + }, + { + "epoch": 1.0436363636363637, + "grad_norm": 0.4024258853456086, + "learning_rate": 1.6491634349431024e-05, + "loss": 0.1333303153514862, + "step": 1866 + }, + { + "epoch": 1.0441958041958042, + "grad_norm": 0.6131468369993198, + "learning_rate": 1.6486682386170862e-05, + "loss": 0.22127163410186768, + "step": 1867 + }, + { + "epoch": 1.0447552447552448, + "grad_norm": 0.3274654793885402, + "learning_rate": 1.648172767522049e-05, + "loss": 0.05909726768732071, + "step": 1868 + }, + { + "epoch": 1.0453146853146853, + "grad_norm": 0.25562214433238145, + "learning_rate": 1.647677021867867e-05, + "loss": 0.08577749133110046, + "step": 1869 + }, + { + "epoch": 1.045874125874126, + "grad_norm": 0.15437314353845613, + "learning_rate": 1.6471810018645334e-05, + "loss": 0.017373185604810715, + "step": 1870 + }, + { + "epoch": 1.0464335664335664, + "grad_norm": 0.6277533164475998, + "learning_rate": 1.646684707722157e-05, + "loss": 0.22174407541751862, + "step": 1871 + }, + { + "epoch": 1.046993006993007, + "grad_norm": 0.8892560489621759, + "learning_rate": 1.6461881396509625e-05, + "loss": 0.492086797952652, + "step": 1872 + }, + { + "epoch": 1.0475524475524476, + "grad_norm": 0.2696956497923366, + "learning_rate": 1.6456912978612903e-05, + "loss": 0.052501581609249115, + "step": 1873 + }, + { + "epoch": 1.048111888111888, + "grad_norm": 0.37167261285512593, + "learning_rate": 1.6451941825635987e-05, + "loss": 0.11107093095779419, + "step": 1874 + }, + { + "epoch": 1.0486713286713287, + "grad_norm": 0.2463340723558187, + "learning_rate": 1.6446967939684592e-05, + "loss": 0.04314631596207619, + "step": 1875 + }, + { + "epoch": 1.0492307692307692, + "grad_norm": 0.40376294935353324, + "learning_rate": 1.644199132286561e-05, + "loss": 0.06144296005368233, + "step": 1876 + }, + { + "epoch": 1.0497902097902099, + "grad_norm": 0.36658945491813005, + "learning_rate": 1.6437011977287084e-05, + "loss": 0.06786086410284042, + "step": 1877 + }, + { + "epoch": 1.0503496503496503, + "grad_norm": 0.38671504502138454, + "learning_rate": 1.643202990505821e-05, + "loss": 0.14964109659194946, + "step": 1878 + }, + { + "epoch": 1.050909090909091, + "grad_norm": 0.6917202607016051, + "learning_rate": 1.6427045108289344e-05, + "loss": 0.33606380224227905, + "step": 1879 + }, + { + "epoch": 1.0514685314685315, + "grad_norm": 0.44053981673904596, + "learning_rate": 1.642205758909199e-05, + "loss": 0.157263845205307, + "step": 1880 + }, + { + "epoch": 1.052027972027972, + "grad_norm": 0.44240601917347644, + "learning_rate": 1.641706734957881e-05, + "loss": 0.0952346995472908, + "step": 1881 + }, + { + "epoch": 1.0525874125874126, + "grad_norm": 0.23827261478852022, + "learning_rate": 1.641207439186362e-05, + "loss": 0.08294466882944107, + "step": 1882 + }, + { + "epoch": 1.053146853146853, + "grad_norm": 0.628578081862457, + "learning_rate": 1.6407078718061374e-05, + "loss": 0.26249027252197266, + "step": 1883 + }, + { + "epoch": 1.0537062937062938, + "grad_norm": 0.507792377827534, + "learning_rate": 1.6402080330288198e-05, + "loss": 0.21906225383281708, + "step": 1884 + }, + { + "epoch": 1.0542657342657342, + "grad_norm": 0.540359357848682, + "learning_rate": 1.639707923066135e-05, + "loss": 0.23156702518463135, + "step": 1885 + }, + { + "epoch": 1.054825174825175, + "grad_norm": 0.6359314687900489, + "learning_rate": 1.6392075421299246e-05, + "loss": 0.14702865481376648, + "step": 1886 + }, + { + "epoch": 1.0553846153846154, + "grad_norm": 0.2851165388221565, + "learning_rate": 1.6387068904321445e-05, + "loss": 0.058649949729442596, + "step": 1887 + }, + { + "epoch": 1.055944055944056, + "grad_norm": 0.6857303513880347, + "learning_rate": 1.6382059681848656e-05, + "loss": 0.2501545548439026, + "step": 1888 + }, + { + "epoch": 1.0565034965034965, + "grad_norm": 0.48481109962377356, + "learning_rate": 1.637704775600273e-05, + "loss": 0.17257605493068695, + "step": 1889 + }, + { + "epoch": 1.057062937062937, + "grad_norm": 0.544244557083397, + "learning_rate": 1.637203312890667e-05, + "loss": 0.1971706748008728, + "step": 1890 + }, + { + "epoch": 1.0576223776223777, + "grad_norm": 0.5882085575348375, + "learning_rate": 1.6367015802684618e-05, + "loss": 0.321458637714386, + "step": 1891 + }, + { + "epoch": 1.0581818181818181, + "grad_norm": 0.5478938706269109, + "learning_rate": 1.6361995779461856e-05, + "loss": 0.3172048032283783, + "step": 1892 + }, + { + "epoch": 1.0587412587412588, + "grad_norm": 0.6117362007432704, + "learning_rate": 1.6356973061364822e-05, + "loss": 0.32852277159690857, + "step": 1893 + }, + { + "epoch": 1.0593006993006993, + "grad_norm": 0.4450619952937148, + "learning_rate": 1.6351947650521077e-05, + "loss": 0.19330886006355286, + "step": 1894 + }, + { + "epoch": 1.05986013986014, + "grad_norm": 0.5350555922770339, + "learning_rate": 1.634691954905934e-05, + "loss": 0.1689222753047943, + "step": 1895 + }, + { + "epoch": 1.0604195804195804, + "grad_norm": 0.3924848821278538, + "learning_rate": 1.634188875910945e-05, + "loss": 0.10688693821430206, + "step": 1896 + }, + { + "epoch": 1.060979020979021, + "grad_norm": 0.7512213006623004, + "learning_rate": 1.6336855282802413e-05, + "loss": 0.279634028673172, + "step": 1897 + }, + { + "epoch": 1.0615384615384615, + "grad_norm": 0.46607263876144295, + "learning_rate": 1.6331819122270345e-05, + "loss": 0.1641359180212021, + "step": 1898 + }, + { + "epoch": 1.062097902097902, + "grad_norm": 0.48551977239048866, + "learning_rate": 1.632678027964651e-05, + "loss": 0.13418062031269073, + "step": 1899 + }, + { + "epoch": 1.0626573426573427, + "grad_norm": 0.28763818262971386, + "learning_rate": 1.6321738757065313e-05, + "loss": 0.031035996973514557, + "step": 1900 + }, + { + "epoch": 1.0632167832167831, + "grad_norm": 0.3839823249835374, + "learning_rate": 1.631669455666229e-05, + "loss": 0.1120566800236702, + "step": 1901 + }, + { + "epoch": 1.0637762237762238, + "grad_norm": 0.6025093688193385, + "learning_rate": 1.6311647680574107e-05, + "loss": 0.2700100243091583, + "step": 1902 + }, + { + "epoch": 1.0643356643356643, + "grad_norm": 0.3424515327048564, + "learning_rate": 1.6306598130938573e-05, + "loss": 0.039188630878925323, + "step": 1903 + }, + { + "epoch": 1.064895104895105, + "grad_norm": 0.4615879246285341, + "learning_rate": 1.630154590989462e-05, + "loss": 0.07795591652393341, + "step": 1904 + }, + { + "epoch": 1.0654545454545454, + "grad_norm": 0.506359194947615, + "learning_rate": 1.6296491019582315e-05, + "loss": 0.1720692217350006, + "step": 1905 + }, + { + "epoch": 1.0660139860139861, + "grad_norm": 0.6432250049694156, + "learning_rate": 1.6291433462142858e-05, + "loss": 0.10365504026412964, + "step": 1906 + }, + { + "epoch": 1.0665734265734266, + "grad_norm": 0.4997563461557261, + "learning_rate": 1.6286373239718577e-05, + "loss": 0.19018085300922394, + "step": 1907 + }, + { + "epoch": 1.067132867132867, + "grad_norm": 0.7901027042917784, + "learning_rate": 1.6281310354452928e-05, + "loss": 0.3060765564441681, + "step": 1908 + }, + { + "epoch": 1.0676923076923077, + "grad_norm": 0.397558227339986, + "learning_rate": 1.6276244808490497e-05, + "loss": 0.08396832644939423, + "step": 1909 + }, + { + "epoch": 1.0682517482517482, + "grad_norm": 0.4240001687985713, + "learning_rate": 1.6271176603976994e-05, + "loss": 0.11355772614479065, + "step": 1910 + }, + { + "epoch": 1.0688111888111889, + "grad_norm": 0.5320262615708464, + "learning_rate": 1.626610574305926e-05, + "loss": 0.16688455641269684, + "step": 1911 + }, + { + "epoch": 1.0693706293706293, + "grad_norm": 0.6298413692616952, + "learning_rate": 1.6261032227885253e-05, + "loss": 0.21331313252449036, + "step": 1912 + }, + { + "epoch": 1.06993006993007, + "grad_norm": 0.5005240579317168, + "learning_rate": 1.6255956060604067e-05, + "loss": 0.2213878035545349, + "step": 1913 + }, + { + "epoch": 1.0704895104895105, + "grad_norm": 0.4935023615908714, + "learning_rate": 1.6250877243365906e-05, + "loss": 0.07644247263669968, + "step": 1914 + }, + { + "epoch": 1.0710489510489511, + "grad_norm": 0.3797467083768788, + "learning_rate": 1.6245795778322108e-05, + "loss": 0.10563966631889343, + "step": 1915 + }, + { + "epoch": 1.0716083916083916, + "grad_norm": 0.43716037794063634, + "learning_rate": 1.6240711667625127e-05, + "loss": 0.08450563997030258, + "step": 1916 + }, + { + "epoch": 1.072167832167832, + "grad_norm": 0.4153246170558682, + "learning_rate": 1.623562491342854e-05, + "loss": 0.1038089245557785, + "step": 1917 + }, + { + "epoch": 1.0727272727272728, + "grad_norm": 0.3107047405977357, + "learning_rate": 1.623053551788704e-05, + "loss": 0.059487588703632355, + "step": 1918 + }, + { + "epoch": 1.0732867132867132, + "grad_norm": 0.38329850719465514, + "learning_rate": 1.6225443483156443e-05, + "loss": 0.10687503963708878, + "step": 1919 + }, + { + "epoch": 1.073846153846154, + "grad_norm": 0.34187760308818277, + "learning_rate": 1.622034881139368e-05, + "loss": 0.17818018794059753, + "step": 1920 + }, + { + "epoch": 1.0744055944055944, + "grad_norm": 0.37839498984483627, + "learning_rate": 1.6215251504756796e-05, + "loss": 0.114201121032238, + "step": 1921 + }, + { + "epoch": 1.074965034965035, + "grad_norm": 0.3998303791425286, + "learning_rate": 1.6210151565404967e-05, + "loss": 0.176470085978508, + "step": 1922 + }, + { + "epoch": 1.0755244755244755, + "grad_norm": 0.33165380190577537, + "learning_rate": 1.6205048995498466e-05, + "loss": 0.07264681160449982, + "step": 1923 + }, + { + "epoch": 1.0760839160839162, + "grad_norm": 0.4919165593859058, + "learning_rate": 1.6199943797198687e-05, + "loss": 0.17455509305000305, + "step": 1924 + }, + { + "epoch": 1.0766433566433566, + "grad_norm": 0.5804729083824665, + "learning_rate": 1.6194835972668142e-05, + "loss": 0.3306233882904053, + "step": 1925 + }, + { + "epoch": 1.077202797202797, + "grad_norm": 0.3210748827920078, + "learning_rate": 1.6189725524070445e-05, + "loss": 0.06638755649328232, + "step": 1926 + }, + { + "epoch": 1.0777622377622378, + "grad_norm": 0.31574633175163674, + "learning_rate": 1.6184612453570338e-05, + "loss": 0.11081357300281525, + "step": 1927 + }, + { + "epoch": 1.0783216783216782, + "grad_norm": 0.6735561110640411, + "learning_rate": 1.6179496763333657e-05, + "loss": 0.10380128026008606, + "step": 1928 + }, + { + "epoch": 1.078881118881119, + "grad_norm": 0.588923781669744, + "learning_rate": 1.617437845552736e-05, + "loss": 0.2093890905380249, + "step": 1929 + }, + { + "epoch": 1.0794405594405594, + "grad_norm": 0.23641704168953753, + "learning_rate": 1.61692575323195e-05, + "loss": 0.07261718064546585, + "step": 1930 + }, + { + "epoch": 1.08, + "grad_norm": 0.39850381204767, + "learning_rate": 1.616413399587925e-05, + "loss": 0.1493971049785614, + "step": 1931 + }, + { + "epoch": 1.0805594405594405, + "grad_norm": 0.49600567002959195, + "learning_rate": 1.6159007848376885e-05, + "loss": 0.21584853529930115, + "step": 1932 + }, + { + "epoch": 1.0811188811188812, + "grad_norm": 0.49663498791679184, + "learning_rate": 1.615387909198379e-05, + "loss": 0.14152511954307556, + "step": 1933 + }, + { + "epoch": 1.0816783216783217, + "grad_norm": 0.939656960182029, + "learning_rate": 1.614874772887245e-05, + "loss": 0.08901500701904297, + "step": 1934 + }, + { + "epoch": 1.0822377622377621, + "grad_norm": 0.4483598948103245, + "learning_rate": 1.6143613761216456e-05, + "loss": 0.1626087874174118, + "step": 1935 + }, + { + "epoch": 1.0827972027972028, + "grad_norm": 0.340671280006577, + "learning_rate": 1.6138477191190503e-05, + "loss": 0.10775715112686157, + "step": 1936 + }, + { + "epoch": 1.0833566433566433, + "grad_norm": 0.31638841168600074, + "learning_rate": 1.6133338020970392e-05, + "loss": 0.0633353665471077, + "step": 1937 + }, + { + "epoch": 1.083916083916084, + "grad_norm": 0.4379251097556998, + "learning_rate": 1.6128196252733013e-05, + "loss": 0.12046018242835999, + "step": 1938 + }, + { + "epoch": 1.0844755244755244, + "grad_norm": 0.41892471664338243, + "learning_rate": 1.6123051888656367e-05, + "loss": 0.16687169671058655, + "step": 1939 + }, + { + "epoch": 1.085034965034965, + "grad_norm": 0.5294209274582983, + "learning_rate": 1.6117904930919558e-05, + "loss": 0.1763058304786682, + "step": 1940 + }, + { + "epoch": 1.0855944055944056, + "grad_norm": 0.34428258567859926, + "learning_rate": 1.611275538170278e-05, + "loss": 0.14062942564487457, + "step": 1941 + }, + { + "epoch": 1.0861538461538462, + "grad_norm": 0.5407145902000322, + "learning_rate": 1.6107603243187323e-05, + "loss": 0.15122312307357788, + "step": 1942 + }, + { + "epoch": 1.0867132867132867, + "grad_norm": 0.46522043654712064, + "learning_rate": 1.6102448517555584e-05, + "loss": 0.12458480149507523, + "step": 1943 + }, + { + "epoch": 1.0872727272727274, + "grad_norm": 0.6436486293324727, + "learning_rate": 1.6097291206991052e-05, + "loss": 0.26847195625305176, + "step": 1944 + }, + { + "epoch": 1.0878321678321679, + "grad_norm": 0.3312656299936537, + "learning_rate": 1.6092131313678302e-05, + "loss": 0.08677637577056885, + "step": 1945 + }, + { + "epoch": 1.0883916083916083, + "grad_norm": 0.5815988849778235, + "learning_rate": 1.6086968839803016e-05, + "loss": 0.29111695289611816, + "step": 1946 + }, + { + "epoch": 1.088951048951049, + "grad_norm": 0.40067698919529776, + "learning_rate": 1.6081803787551963e-05, + "loss": 0.10848572850227356, + "step": 1947 + }, + { + "epoch": 1.0895104895104895, + "grad_norm": 0.4591333935193691, + "learning_rate": 1.6076636159113e-05, + "loss": 0.14661894738674164, + "step": 1948 + }, + { + "epoch": 1.0900699300699301, + "grad_norm": 0.6955992082023635, + "learning_rate": 1.607146595667509e-05, + "loss": 0.2794772684574127, + "step": 1949 + }, + { + "epoch": 1.0906293706293706, + "grad_norm": 0.4947956240517011, + "learning_rate": 1.6066293182428264e-05, + "loss": 0.19241252541542053, + "step": 1950 + }, + { + "epoch": 1.0911888111888113, + "grad_norm": 0.39280990792466747, + "learning_rate": 1.6061117838563662e-05, + "loss": 0.10696657001972198, + "step": 1951 + }, + { + "epoch": 1.0917482517482517, + "grad_norm": 0.4805893175649746, + "learning_rate": 1.6055939927273506e-05, + "loss": 0.15843608975410461, + "step": 1952 + }, + { + "epoch": 1.0923076923076924, + "grad_norm": 0.35669681914659135, + "learning_rate": 1.6050759450751097e-05, + "loss": 0.0833943709731102, + "step": 1953 + }, + { + "epoch": 1.0928671328671329, + "grad_norm": 0.5535255015298567, + "learning_rate": 1.604557641119084e-05, + "loss": 0.17700690031051636, + "step": 1954 + }, + { + "epoch": 1.0934265734265733, + "grad_norm": 0.49054704581288766, + "learning_rate": 1.6040390810788203e-05, + "loss": 0.18030565977096558, + "step": 1955 + }, + { + "epoch": 1.093986013986014, + "grad_norm": 0.2994539123199912, + "learning_rate": 1.6035202651739768e-05, + "loss": 0.048740699887275696, + "step": 1956 + }, + { + "epoch": 1.0945454545454545, + "grad_norm": 0.4930670831286999, + "learning_rate": 1.6030011936243173e-05, + "loss": 0.2627163529396057, + "step": 1957 + }, + { + "epoch": 1.0951048951048952, + "grad_norm": 0.4683971263754147, + "learning_rate": 1.6024818666497153e-05, + "loss": 0.24802029132843018, + "step": 1958 + }, + { + "epoch": 1.0956643356643356, + "grad_norm": 0.3103882743023382, + "learning_rate": 1.6019622844701525e-05, + "loss": 0.04002990573644638, + "step": 1959 + }, + { + "epoch": 1.0962237762237763, + "grad_norm": 0.5748687917447187, + "learning_rate": 1.6014424473057177e-05, + "loss": 0.19142946600914001, + "step": 1960 + }, + { + "epoch": 1.0967832167832168, + "grad_norm": 0.5003745791915539, + "learning_rate": 1.6009223553766094e-05, + "loss": 0.21755559742450714, + "step": 1961 + }, + { + "epoch": 1.0973426573426575, + "grad_norm": 0.5115272657135853, + "learning_rate": 1.6004020089031324e-05, + "loss": 0.14702238142490387, + "step": 1962 + }, + { + "epoch": 1.097902097902098, + "grad_norm": 0.5038469464702979, + "learning_rate": 1.5998814081057e-05, + "loss": 0.16499340534210205, + "step": 1963 + }, + { + "epoch": 1.0984615384615384, + "grad_norm": 0.5319391429433344, + "learning_rate": 1.5993605532048336e-05, + "loss": 0.23408597707748413, + "step": 1964 + }, + { + "epoch": 1.099020979020979, + "grad_norm": 0.9078722405373567, + "learning_rate": 1.598839444421161e-05, + "loss": 0.257779985666275, + "step": 1965 + }, + { + "epoch": 1.0995804195804195, + "grad_norm": 0.4121703966314885, + "learning_rate": 1.59831808197542e-05, + "loss": 0.09074355661869049, + "step": 1966 + }, + { + "epoch": 1.1001398601398602, + "grad_norm": 0.5823252382296038, + "learning_rate": 1.597796466088452e-05, + "loss": 0.1712971329689026, + "step": 1967 + }, + { + "epoch": 1.1006993006993007, + "grad_norm": 0.47880860211957843, + "learning_rate": 1.59727459698121e-05, + "loss": 0.17786143720149994, + "step": 1968 + }, + { + "epoch": 1.1012587412587413, + "grad_norm": 0.38994572913490655, + "learning_rate": 1.5967524748747515e-05, + "loss": 0.11310650408267975, + "step": 1969 + }, + { + "epoch": 1.1018181818181818, + "grad_norm": 0.5647839506444448, + "learning_rate": 1.5962300999902416e-05, + "loss": 0.26985836029052734, + "step": 1970 + }, + { + "epoch": 1.1023776223776225, + "grad_norm": 0.3556315245187329, + "learning_rate": 1.595707472548953e-05, + "loss": 0.10740834474563599, + "step": 1971 + }, + { + "epoch": 1.102937062937063, + "grad_norm": 0.32858254097884315, + "learning_rate": 1.5951845927722653e-05, + "loss": 0.05938706174492836, + "step": 1972 + }, + { + "epoch": 1.1034965034965034, + "grad_norm": 0.3554511047225426, + "learning_rate": 1.594661460881665e-05, + "loss": 0.07059267163276672, + "step": 1973 + }, + { + "epoch": 1.104055944055944, + "grad_norm": 0.5103703307212026, + "learning_rate": 1.5941380770987445e-05, + "loss": 0.11098214983940125, + "step": 1974 + }, + { + "epoch": 1.1046153846153846, + "grad_norm": 0.5137531456094215, + "learning_rate": 1.5936144416452048e-05, + "loss": 0.21214474737644196, + "step": 1975 + }, + { + "epoch": 1.1051748251748252, + "grad_norm": 0.46787290708654594, + "learning_rate": 1.593090554742851e-05, + "loss": 0.13516046106815338, + "step": 1976 + }, + { + "epoch": 1.1057342657342657, + "grad_norm": 0.33848572899421225, + "learning_rate": 1.5925664166135972e-05, + "loss": 0.07961449772119522, + "step": 1977 + }, + { + "epoch": 1.1062937062937064, + "grad_norm": 0.3944333605572859, + "learning_rate": 1.592042027479462e-05, + "loss": 0.1356368064880371, + "step": 1978 + }, + { + "epoch": 1.1068531468531468, + "grad_norm": 0.6251866414499749, + "learning_rate": 1.591517387562572e-05, + "loss": 0.18539609014987946, + "step": 1979 + }, + { + "epoch": 1.1074125874125875, + "grad_norm": 0.4522758115478833, + "learning_rate": 1.5909924970851584e-05, + "loss": 0.1736263632774353, + "step": 1980 + }, + { + "epoch": 1.107972027972028, + "grad_norm": 0.42098393281615004, + "learning_rate": 1.5904673562695594e-05, + "loss": 0.1802954524755478, + "step": 1981 + }, + { + "epoch": 1.1085314685314684, + "grad_norm": 0.48806151934864656, + "learning_rate": 1.5899419653382192e-05, + "loss": 0.11812008917331696, + "step": 1982 + }, + { + "epoch": 1.1090909090909091, + "grad_norm": 0.5432200790498781, + "learning_rate": 1.5894163245136874e-05, + "loss": 0.144090473651886, + "step": 1983 + }, + { + "epoch": 1.1096503496503496, + "grad_norm": 0.34236062089085806, + "learning_rate": 1.588890434018621e-05, + "loss": 0.07787076383829117, + "step": 1984 + }, + { + "epoch": 1.1102097902097903, + "grad_norm": 0.5092746915774696, + "learning_rate": 1.588364294075781e-05, + "loss": 0.14424125850200653, + "step": 1985 + }, + { + "epoch": 1.1107692307692307, + "grad_norm": 0.4430125283415431, + "learning_rate": 1.5878379049080344e-05, + "loss": 0.170151025056839, + "step": 1986 + }, + { + "epoch": 1.1113286713286714, + "grad_norm": 0.5382660777109902, + "learning_rate": 1.5873112667383542e-05, + "loss": 0.19215074181556702, + "step": 1987 + }, + { + "epoch": 1.1118881118881119, + "grad_norm": 0.32115144592421563, + "learning_rate": 1.5867843797898194e-05, + "loss": 0.08479930460453033, + "step": 1988 + }, + { + "epoch": 1.1124475524475526, + "grad_norm": 0.47288492716821284, + "learning_rate": 1.586257244285613e-05, + "loss": 0.1669064462184906, + "step": 1989 + }, + { + "epoch": 1.113006993006993, + "grad_norm": 0.32878064355070785, + "learning_rate": 1.585729860449025e-05, + "loss": 0.16119353473186493, + "step": 1990 + }, + { + "epoch": 1.1135664335664335, + "grad_norm": 0.3770989338355465, + "learning_rate": 1.5852022285034485e-05, + "loss": 0.11893858015537262, + "step": 1991 + }, + { + "epoch": 1.1141258741258742, + "grad_norm": 0.5211494382908677, + "learning_rate": 1.5846743486723834e-05, + "loss": 0.22438114881515503, + "step": 1992 + }, + { + "epoch": 1.1146853146853146, + "grad_norm": 0.7303740029167695, + "learning_rate": 1.5841462211794337e-05, + "loss": 0.21321550011634827, + "step": 1993 + }, + { + "epoch": 1.1152447552447553, + "grad_norm": 0.34585226749488246, + "learning_rate": 1.5836178462483092e-05, + "loss": 0.11793461441993713, + "step": 1994 + }, + { + "epoch": 1.1158041958041958, + "grad_norm": 0.5432596406524868, + "learning_rate": 1.583089224102824e-05, + "loss": 0.2472701221704483, + "step": 1995 + }, + { + "epoch": 1.1163636363636364, + "grad_norm": 0.5605224081310403, + "learning_rate": 1.582560354966896e-05, + "loss": 0.23547127842903137, + "step": 1996 + }, + { + "epoch": 1.116923076923077, + "grad_norm": 0.49607525426076016, + "learning_rate": 1.5820312390645488e-05, + "loss": 0.21054032444953918, + "step": 1997 + }, + { + "epoch": 1.1174825174825176, + "grad_norm": 0.3762014716935107, + "learning_rate": 1.5815018766199112e-05, + "loss": 0.04946086183190346, + "step": 1998 + }, + { + "epoch": 1.118041958041958, + "grad_norm": 0.4257222488182966, + "learning_rate": 1.5809722678572147e-05, + "loss": 0.11270518600940704, + "step": 1999 + }, + { + "epoch": 1.1186013986013985, + "grad_norm": 0.5307450643946884, + "learning_rate": 1.5804424130007963e-05, + "loss": 0.1681709885597229, + "step": 2000 + }, + { + "epoch": 1.1191608391608392, + "grad_norm": 0.8572373046427969, + "learning_rate": 1.5799123122750967e-05, + "loss": 0.35935884714126587, + "step": 2001 + }, + { + "epoch": 1.1197202797202797, + "grad_norm": 0.5949752869001521, + "learning_rate": 1.5793819659046615e-05, + "loss": 0.24719780683517456, + "step": 2002 + }, + { + "epoch": 1.1202797202797203, + "grad_norm": 0.45906981271625175, + "learning_rate": 1.5788513741141393e-05, + "loss": 0.25767379999160767, + "step": 2003 + }, + { + "epoch": 1.1208391608391608, + "grad_norm": 0.41369720463852594, + "learning_rate": 1.578320537128283e-05, + "loss": 0.15381020307540894, + "step": 2004 + }, + { + "epoch": 1.1213986013986015, + "grad_norm": 0.579080079510505, + "learning_rate": 1.57778945517195e-05, + "loss": 0.2354571372270584, + "step": 2005 + }, + { + "epoch": 1.121958041958042, + "grad_norm": 0.9731263894721104, + "learning_rate": 1.577258128470101e-05, + "loss": 0.4068896770477295, + "step": 2006 + }, + { + "epoch": 1.1225174825174826, + "grad_norm": 0.2713907392476084, + "learning_rate": 1.5767265572478003e-05, + "loss": 0.04392913728952408, + "step": 2007 + }, + { + "epoch": 1.123076923076923, + "grad_norm": 0.16743376311187513, + "learning_rate": 1.5761947417302156e-05, + "loss": 0.021909723058342934, + "step": 2008 + }, + { + "epoch": 1.1236363636363635, + "grad_norm": 0.3559357060487031, + "learning_rate": 1.575662682142619e-05, + "loss": 0.077776238322258, + "step": 2009 + }, + { + "epoch": 1.1241958041958042, + "grad_norm": 0.33130967380874715, + "learning_rate": 1.5751303787103846e-05, + "loss": 0.03840276971459389, + "step": 2010 + }, + { + "epoch": 1.1247552447552447, + "grad_norm": 0.5471260957160835, + "learning_rate": 1.574597831658991e-05, + "loss": 0.19674792885780334, + "step": 2011 + }, + { + "epoch": 1.1253146853146854, + "grad_norm": 0.6142310271970681, + "learning_rate": 1.5740650412140194e-05, + "loss": 0.16901172697544098, + "step": 2012 + }, + { + "epoch": 1.1258741258741258, + "grad_norm": 0.46043616748667876, + "learning_rate": 1.5735320076011537e-05, + "loss": 0.1439274549484253, + "step": 2013 + }, + { + "epoch": 1.1264335664335665, + "grad_norm": 0.3227565966333693, + "learning_rate": 1.5729987310461822e-05, + "loss": 0.07391933351755142, + "step": 2014 + }, + { + "epoch": 1.126993006993007, + "grad_norm": 0.6286362327198821, + "learning_rate": 1.5724652117749947e-05, + "loss": 0.2885581851005554, + "step": 2015 + }, + { + "epoch": 1.1275524475524477, + "grad_norm": 0.7662188271693107, + "learning_rate": 1.5719314500135844e-05, + "loss": 0.08786419779062271, + "step": 2016 + }, + { + "epoch": 1.1281118881118881, + "grad_norm": 0.42079664356106156, + "learning_rate": 1.5713974459880467e-05, + "loss": 0.14879518747329712, + "step": 2017 + }, + { + "epoch": 1.1286713286713286, + "grad_norm": 0.27186831281207, + "learning_rate": 1.570863199924581e-05, + "loss": 0.071931391954422, + "step": 2018 + }, + { + "epoch": 1.1292307692307693, + "grad_norm": 0.5585962533876995, + "learning_rate": 1.570328712049488e-05, + "loss": 0.213780015707016, + "step": 2019 + }, + { + "epoch": 1.1297902097902097, + "grad_norm": 0.6547341053711909, + "learning_rate": 1.5697939825891707e-05, + "loss": 0.31340765953063965, + "step": 2020 + }, + { + "epoch": 1.1303496503496504, + "grad_norm": 0.4775071271792002, + "learning_rate": 1.569259011770135e-05, + "loss": 0.17053598165512085, + "step": 2021 + }, + { + "epoch": 1.1309090909090909, + "grad_norm": 0.47156693236922875, + "learning_rate": 1.5687237998189892e-05, + "loss": 0.10571020841598511, + "step": 2022 + }, + { + "epoch": 1.1314685314685315, + "grad_norm": 0.589626522477397, + "learning_rate": 1.568188346962443e-05, + "loss": 0.22666051983833313, + "step": 2023 + }, + { + "epoch": 1.132027972027972, + "grad_norm": 0.34784786930374123, + "learning_rate": 1.567652653427309e-05, + "loss": 0.08405138552188873, + "step": 2024 + }, + { + "epoch": 1.1325874125874127, + "grad_norm": 0.43063993273908574, + "learning_rate": 1.567116719440501e-05, + "loss": 0.1688159853219986, + "step": 2025 + }, + { + "epoch": 1.1331468531468532, + "grad_norm": 0.32611570703399456, + "learning_rate": 1.5665805452290354e-05, + "loss": 0.14032645523548126, + "step": 2026 + }, + { + "epoch": 1.1337062937062936, + "grad_norm": 0.4568771885979859, + "learning_rate": 1.5660441310200293e-05, + "loss": 0.10113655775785446, + "step": 2027 + }, + { + "epoch": 1.1342657342657343, + "grad_norm": 0.6140187758899458, + "learning_rate": 1.5655074770407027e-05, + "loss": 0.2282724529504776, + "step": 2028 + }, + { + "epoch": 1.1348251748251748, + "grad_norm": 0.39110200460111205, + "learning_rate": 1.5649705835183767e-05, + "loss": 0.07945576310157776, + "step": 2029 + }, + { + "epoch": 1.1353846153846154, + "grad_norm": 0.46775612373152675, + "learning_rate": 1.5644334506804727e-05, + "loss": 0.17717701196670532, + "step": 2030 + }, + { + "epoch": 1.135944055944056, + "grad_norm": 0.4943878811394455, + "learning_rate": 1.5638960787545154e-05, + "loss": 0.202545627951622, + "step": 2031 + }, + { + "epoch": 1.1365034965034966, + "grad_norm": 0.43298312635904174, + "learning_rate": 1.5633584679681295e-05, + "loss": 0.1656865030527115, + "step": 2032 + }, + { + "epoch": 1.137062937062937, + "grad_norm": 0.592122816290298, + "learning_rate": 1.5628206185490414e-05, + "loss": 0.19921652972698212, + "step": 2033 + }, + { + "epoch": 1.1376223776223777, + "grad_norm": 0.4937127623591173, + "learning_rate": 1.5622825307250787e-05, + "loss": 0.22966957092285156, + "step": 2034 + }, + { + "epoch": 1.1381818181818182, + "grad_norm": 0.35541528210455536, + "learning_rate": 1.5617442047241693e-05, + "loss": 0.09429632127285004, + "step": 2035 + }, + { + "epoch": 1.1387412587412586, + "grad_norm": 0.3579010561520189, + "learning_rate": 1.5612056407743425e-05, + "loss": 0.2207282930612564, + "step": 2036 + }, + { + "epoch": 1.1393006993006993, + "grad_norm": 0.3741647019736701, + "learning_rate": 1.5606668391037283e-05, + "loss": 0.15288980305194855, + "step": 2037 + }, + { + "epoch": 1.1398601398601398, + "grad_norm": 0.5423692235604567, + "learning_rate": 1.560127799940558e-05, + "loss": 0.1533864438533783, + "step": 2038 + }, + { + "epoch": 1.1404195804195805, + "grad_norm": 0.3724728166035704, + "learning_rate": 1.5595885235131616e-05, + "loss": 0.14706389605998993, + "step": 2039 + }, + { + "epoch": 1.140979020979021, + "grad_norm": 0.44497846562062376, + "learning_rate": 1.5590490100499722e-05, + "loss": 0.1838333010673523, + "step": 2040 + }, + { + "epoch": 1.1415384615384616, + "grad_norm": 0.8319057516838321, + "learning_rate": 1.5585092597795216e-05, + "loss": 0.23537740111351013, + "step": 2041 + }, + { + "epoch": 1.142097902097902, + "grad_norm": 0.5082877194117513, + "learning_rate": 1.5579692729304418e-05, + "loss": 0.12658530473709106, + "step": 2042 + }, + { + "epoch": 1.1426573426573428, + "grad_norm": 0.5069266554281983, + "learning_rate": 1.557429049731466e-05, + "loss": 0.2904822826385498, + "step": 2043 + }, + { + "epoch": 1.1432167832167832, + "grad_norm": 0.5482191332765755, + "learning_rate": 1.556888590411427e-05, + "loss": 0.2652897834777832, + "step": 2044 + }, + { + "epoch": 1.1437762237762237, + "grad_norm": 0.3338500466503338, + "learning_rate": 1.5563478951992575e-05, + "loss": 0.08475368469953537, + "step": 2045 + }, + { + "epoch": 1.1443356643356644, + "grad_norm": 0.502058771701403, + "learning_rate": 1.555806964323991e-05, + "loss": 0.1557948887348175, + "step": 2046 + }, + { + "epoch": 1.1448951048951048, + "grad_norm": 0.5280737595428816, + "learning_rate": 1.555265798014759e-05, + "loss": 0.1669459044933319, + "step": 2047 + }, + { + "epoch": 1.1454545454545455, + "grad_norm": 0.3269351563738514, + "learning_rate": 1.5547243965007947e-05, + "loss": 0.0632094219326973, + "step": 2048 + }, + { + "epoch": 1.146013986013986, + "grad_norm": 0.43740701825083356, + "learning_rate": 1.5541827600114293e-05, + "loss": 0.10252562165260315, + "step": 2049 + }, + { + "epoch": 1.1465734265734266, + "grad_norm": 0.4032460665498071, + "learning_rate": 1.5536408887760952e-05, + "loss": 0.06701380014419556, + "step": 2050 + }, + { + "epoch": 1.147132867132867, + "grad_norm": 0.33335088774823424, + "learning_rate": 1.553098783024323e-05, + "loss": 0.05121865123510361, + "step": 2051 + }, + { + "epoch": 1.1476923076923078, + "grad_norm": 0.4743662853245603, + "learning_rate": 1.5525564429857427e-05, + "loss": 0.1669970452785492, + "step": 2052 + }, + { + "epoch": 1.1482517482517482, + "grad_norm": 0.34571890254570103, + "learning_rate": 1.552013868890084e-05, + "loss": 0.056503891944885254, + "step": 2053 + }, + { + "epoch": 1.1488111888111887, + "grad_norm": 0.2872040252045415, + "learning_rate": 1.5514710609671754e-05, + "loss": 0.028636084869503975, + "step": 2054 + }, + { + "epoch": 1.1493706293706294, + "grad_norm": 0.3006623124163333, + "learning_rate": 1.5509280194469447e-05, + "loss": 0.0977797657251358, + "step": 2055 + }, + { + "epoch": 1.1499300699300699, + "grad_norm": 0.4907571002749203, + "learning_rate": 1.5503847445594188e-05, + "loss": 0.14876842498779297, + "step": 2056 + }, + { + "epoch": 1.1504895104895105, + "grad_norm": 0.6006146910861848, + "learning_rate": 1.5498412365347228e-05, + "loss": 0.29423877596855164, + "step": 2057 + }, + { + "epoch": 1.151048951048951, + "grad_norm": 0.6704356884252076, + "learning_rate": 1.5492974956030815e-05, + "loss": 0.22696185111999512, + "step": 2058 + }, + { + "epoch": 1.1516083916083917, + "grad_norm": 0.42791394247030845, + "learning_rate": 1.5487535219948174e-05, + "loss": 0.16424477100372314, + "step": 2059 + }, + { + "epoch": 1.1521678321678321, + "grad_norm": 0.43563890009424355, + "learning_rate": 1.548209315940352e-05, + "loss": 0.108802929520607, + "step": 2060 + }, + { + "epoch": 1.1527272727272728, + "grad_norm": 0.42529692499949556, + "learning_rate": 1.547664877670205e-05, + "loss": 0.06705796718597412, + "step": 2061 + }, + { + "epoch": 1.1532867132867133, + "grad_norm": 0.37219850491933687, + "learning_rate": 1.5471202074149956e-05, + "loss": 0.07843948900699615, + "step": 2062 + }, + { + "epoch": 1.1538461538461537, + "grad_norm": 0.5470885156861128, + "learning_rate": 1.54657530540544e-05, + "loss": 0.16249790787696838, + "step": 2063 + }, + { + "epoch": 1.1544055944055944, + "grad_norm": 0.568482400181953, + "learning_rate": 1.546030171872352e-05, + "loss": 0.23675481975078583, + "step": 2064 + }, + { + "epoch": 1.1549650349650349, + "grad_norm": 0.7502529361898437, + "learning_rate": 1.5454848070466458e-05, + "loss": 0.2960844039916992, + "step": 2065 + }, + { + "epoch": 1.1555244755244756, + "grad_norm": 0.5936748929708618, + "learning_rate": 1.544939211159331e-05, + "loss": 0.33044707775115967, + "step": 2066 + }, + { + "epoch": 1.156083916083916, + "grad_norm": 0.5342957265222156, + "learning_rate": 1.544393384441517e-05, + "loss": 0.18616685271263123, + "step": 2067 + }, + { + "epoch": 1.1566433566433567, + "grad_norm": 0.38940484512521417, + "learning_rate": 1.5438473271244105e-05, + "loss": 0.18014107644557953, + "step": 2068 + }, + { + "epoch": 1.1572027972027972, + "grad_norm": 0.3787908459642924, + "learning_rate": 1.5433010394393147e-05, + "loss": 0.08684305101633072, + "step": 2069 + }, + { + "epoch": 1.1577622377622379, + "grad_norm": 0.6232209367166633, + "learning_rate": 1.5427545216176317e-05, + "loss": 0.3271596431732178, + "step": 2070 + }, + { + "epoch": 1.1583216783216783, + "grad_norm": 0.2931239591763908, + "learning_rate": 1.542207773890861e-05, + "loss": 0.06100625544786453, + "step": 2071 + }, + { + "epoch": 1.1588811188811188, + "grad_norm": 0.43959312900298486, + "learning_rate": 1.5416607964905984e-05, + "loss": 0.19233372807502747, + "step": 2072 + }, + { + "epoch": 1.1594405594405595, + "grad_norm": 0.3875437204089428, + "learning_rate": 1.541113589648539e-05, + "loss": 0.11356054246425629, + "step": 2073 + }, + { + "epoch": 1.16, + "grad_norm": 0.7003863249757152, + "learning_rate": 1.5405661535964726e-05, + "loss": 0.2626035511493683, + "step": 2074 + }, + { + "epoch": 1.1605594405594406, + "grad_norm": 0.5679309785504533, + "learning_rate": 1.5400184885662883e-05, + "loss": 0.22842754423618317, + "step": 2075 + }, + { + "epoch": 1.161118881118881, + "grad_norm": 0.4524642483865695, + "learning_rate": 1.53947059478997e-05, + "loss": 0.21133379638195038, + "step": 2076 + }, + { + "epoch": 1.1616783216783217, + "grad_norm": 0.4654445937863856, + "learning_rate": 1.5389224724996012e-05, + "loss": 0.2178257703781128, + "step": 2077 + }, + { + "epoch": 1.1622377622377622, + "grad_norm": 0.38098396310076943, + "learning_rate": 1.5383741219273596e-05, + "loss": 0.14156261086463928, + "step": 2078 + }, + { + "epoch": 1.162797202797203, + "grad_norm": 0.5703963183341267, + "learning_rate": 1.5378255433055215e-05, + "loss": 0.2085438370704651, + "step": 2079 + }, + { + "epoch": 1.1633566433566433, + "grad_norm": 0.6516858900715636, + "learning_rate": 1.5372767368664584e-05, + "loss": 0.15259240567684174, + "step": 2080 + }, + { + "epoch": 1.1639160839160838, + "grad_norm": 0.2513273579993092, + "learning_rate": 1.536727702842639e-05, + "loss": 0.033811211585998535, + "step": 2081 + }, + { + "epoch": 1.1644755244755245, + "grad_norm": 0.28542274794106515, + "learning_rate": 1.5361784414666286e-05, + "loss": 0.06193928048014641, + "step": 2082 + }, + { + "epoch": 1.165034965034965, + "grad_norm": 0.29138464766394867, + "learning_rate": 1.5356289529710885e-05, + "loss": 0.09337253868579865, + "step": 2083 + }, + { + "epoch": 1.1655944055944056, + "grad_norm": 0.35763710444455565, + "learning_rate": 1.535079237588776e-05, + "loss": 0.11022676527500153, + "step": 2084 + }, + { + "epoch": 1.166153846153846, + "grad_norm": 0.44645429020986604, + "learning_rate": 1.5345292955525452e-05, + "loss": 0.13584671914577484, + "step": 2085 + }, + { + "epoch": 1.1667132867132868, + "grad_norm": 0.3028887329019395, + "learning_rate": 1.533979127095345e-05, + "loss": 0.06401629745960236, + "step": 2086 + }, + { + "epoch": 1.1672727272727272, + "grad_norm": 0.3558822526951771, + "learning_rate": 1.5334287324502216e-05, + "loss": 0.16698762774467468, + "step": 2087 + }, + { + "epoch": 1.167832167832168, + "grad_norm": 0.4234144468233863, + "learning_rate": 1.532878111850316e-05, + "loss": 0.16536566615104675, + "step": 2088 + }, + { + "epoch": 1.1683916083916084, + "grad_norm": 0.31214312260289395, + "learning_rate": 1.5323272655288655e-05, + "loss": 0.05860889330506325, + "step": 2089 + }, + { + "epoch": 1.1689510489510488, + "grad_norm": 0.41681068638416896, + "learning_rate": 1.5317761937192023e-05, + "loss": 0.16140541434288025, + "step": 2090 + }, + { + "epoch": 1.1695104895104895, + "grad_norm": 0.5599402069001224, + "learning_rate": 1.5312248966547555e-05, + "loss": 0.23804160952568054, + "step": 2091 + }, + { + "epoch": 1.17006993006993, + "grad_norm": 0.548251335207275, + "learning_rate": 1.530673374569048e-05, + "loss": 0.08927642554044724, + "step": 2092 + }, + { + "epoch": 1.1706293706293707, + "grad_norm": 0.38916840666237573, + "learning_rate": 1.530121627695699e-05, + "loss": 0.08895355463027954, + "step": 2093 + }, + { + "epoch": 1.1711888111888111, + "grad_norm": 0.4102778934509906, + "learning_rate": 1.5295696562684226e-05, + "loss": 0.1526639759540558, + "step": 2094 + }, + { + "epoch": 1.1717482517482518, + "grad_norm": 0.5143109629290507, + "learning_rate": 1.529017460521028e-05, + "loss": 0.17482587695121765, + "step": 2095 + }, + { + "epoch": 1.1723076923076923, + "grad_norm": 0.3681917197127282, + "learning_rate": 1.5284650406874197e-05, + "loss": 0.09317049384117126, + "step": 2096 + }, + { + "epoch": 1.172867132867133, + "grad_norm": 0.455008868073321, + "learning_rate": 1.527912397001597e-05, + "loss": 0.18568839132785797, + "step": 2097 + }, + { + "epoch": 1.1734265734265734, + "grad_norm": 0.6040982882147787, + "learning_rate": 1.527359529697653e-05, + "loss": 0.2325122058391571, + "step": 2098 + }, + { + "epoch": 1.1739860139860139, + "grad_norm": 0.7673006401173544, + "learning_rate": 1.5268064390097777e-05, + "loss": 0.17921952903270721, + "step": 2099 + }, + { + "epoch": 1.1745454545454546, + "grad_norm": 0.6731350264762506, + "learning_rate": 1.5262531251722535e-05, + "loss": 0.1375807523727417, + "step": 2100 + }, + { + "epoch": 1.175104895104895, + "grad_norm": 0.4315171453049918, + "learning_rate": 1.5256995884194593e-05, + "loss": 0.15246731042861938, + "step": 2101 + }, + { + "epoch": 1.1756643356643357, + "grad_norm": 0.4901234668840061, + "learning_rate": 1.5251458289858665e-05, + "loss": 0.15595918893814087, + "step": 2102 + }, + { + "epoch": 1.1762237762237762, + "grad_norm": 0.8054219798928296, + "learning_rate": 1.5245918471060423e-05, + "loss": 0.40096986293792725, + "step": 2103 + }, + { + "epoch": 1.1767832167832168, + "grad_norm": 0.5507990423127248, + "learning_rate": 1.5240376430146473e-05, + "loss": 0.3100760579109192, + "step": 2104 + }, + { + "epoch": 1.1773426573426573, + "grad_norm": 0.5673900667582227, + "learning_rate": 1.5234832169464367e-05, + "loss": 0.16883251070976257, + "step": 2105 + }, + { + "epoch": 1.177902097902098, + "grad_norm": 0.3195774337754071, + "learning_rate": 1.5229285691362594e-05, + "loss": 0.053090669214725494, + "step": 2106 + }, + { + "epoch": 1.1784615384615384, + "grad_norm": 0.3951650134358521, + "learning_rate": 1.5223736998190584e-05, + "loss": 0.09708268940448761, + "step": 2107 + }, + { + "epoch": 1.179020979020979, + "grad_norm": 0.6376365710157335, + "learning_rate": 1.5218186092298703e-05, + "loss": 0.3400949537754059, + "step": 2108 + }, + { + "epoch": 1.1795804195804196, + "grad_norm": 0.567436799834298, + "learning_rate": 1.5212632976038259e-05, + "loss": 0.11515244096517563, + "step": 2109 + }, + { + "epoch": 1.18013986013986, + "grad_norm": 0.6431422591840589, + "learning_rate": 1.5207077651761494e-05, + "loss": 0.20428475737571716, + "step": 2110 + }, + { + "epoch": 1.1806993006993007, + "grad_norm": 0.5719334365101207, + "learning_rate": 1.5201520121821583e-05, + "loss": 0.18052145838737488, + "step": 2111 + }, + { + "epoch": 1.1812587412587412, + "grad_norm": 0.35554852152988314, + "learning_rate": 1.5195960388572636e-05, + "loss": 0.06973645836114883, + "step": 2112 + }, + { + "epoch": 1.1818181818181819, + "grad_norm": 0.42231356012650934, + "learning_rate": 1.5190398454369702e-05, + "loss": 0.17916619777679443, + "step": 2113 + }, + { + "epoch": 1.1823776223776223, + "grad_norm": 0.47058325890077707, + "learning_rate": 1.518483432156875e-05, + "loss": 0.14442399144172668, + "step": 2114 + }, + { + "epoch": 1.182937062937063, + "grad_norm": 0.5949725491554787, + "learning_rate": 1.5179267992526692e-05, + "loss": 0.2838698625564575, + "step": 2115 + }, + { + "epoch": 1.1834965034965035, + "grad_norm": 0.43152070117098645, + "learning_rate": 1.5173699469601369e-05, + "loss": 0.14477328956127167, + "step": 2116 + }, + { + "epoch": 1.184055944055944, + "grad_norm": 0.3696238586060693, + "learning_rate": 1.5168128755151546e-05, + "loss": 0.14884303510189056, + "step": 2117 + }, + { + "epoch": 1.1846153846153846, + "grad_norm": 0.500056707806488, + "learning_rate": 1.5162555851536916e-05, + "loss": 0.19063085317611694, + "step": 2118 + }, + { + "epoch": 1.185174825174825, + "grad_norm": 0.43930162605421164, + "learning_rate": 1.5156980761118107e-05, + "loss": 0.12622380256652832, + "step": 2119 + }, + { + "epoch": 1.1857342657342658, + "grad_norm": 1.0228420945222447, + "learning_rate": 1.5151403486256668e-05, + "loss": 0.14665603637695312, + "step": 2120 + }, + { + "epoch": 1.1862937062937062, + "grad_norm": 0.3180409285710375, + "learning_rate": 1.5145824029315069e-05, + "loss": 0.10852214694023132, + "step": 2121 + }, + { + "epoch": 1.186853146853147, + "grad_norm": 0.5481985251648946, + "learning_rate": 1.5140242392656715e-05, + "loss": 0.12678593397140503, + "step": 2122 + }, + { + "epoch": 1.1874125874125874, + "grad_norm": 0.47593285238786986, + "learning_rate": 1.5134658578645924e-05, + "loss": 0.09018857777118683, + "step": 2123 + }, + { + "epoch": 1.187972027972028, + "grad_norm": 0.37038823931945103, + "learning_rate": 1.5129072589647945e-05, + "loss": 0.11733860522508621, + "step": 2124 + }, + { + "epoch": 1.1885314685314685, + "grad_norm": 0.3891911307569987, + "learning_rate": 1.5123484428028938e-05, + "loss": 0.0967697724699974, + "step": 2125 + }, + { + "epoch": 1.189090909090909, + "grad_norm": 0.6347831633859475, + "learning_rate": 1.5117894096155992e-05, + "loss": 0.18317899107933044, + "step": 2126 + }, + { + "epoch": 1.1896503496503497, + "grad_norm": 0.5118873211248536, + "learning_rate": 1.5112301596397117e-05, + "loss": 0.15682217478752136, + "step": 2127 + }, + { + "epoch": 1.1902097902097901, + "grad_norm": 0.3913669387514239, + "learning_rate": 1.510670693112123e-05, + "loss": 0.0676366314291954, + "step": 2128 + }, + { + "epoch": 1.1907692307692308, + "grad_norm": 0.3999162521364176, + "learning_rate": 1.5101110102698178e-05, + "loss": 0.11411210894584656, + "step": 2129 + }, + { + "epoch": 1.1913286713286713, + "grad_norm": 0.5368501039588917, + "learning_rate": 1.5095511113498713e-05, + "loss": 0.15485668182373047, + "step": 2130 + }, + { + "epoch": 1.191888111888112, + "grad_norm": 0.4901323556544949, + "learning_rate": 1.5089909965894513e-05, + "loss": 0.2711738348007202, + "step": 2131 + }, + { + "epoch": 1.1924475524475524, + "grad_norm": 0.6097937645729039, + "learning_rate": 1.5084306662258162e-05, + "loss": 0.1966949701309204, + "step": 2132 + }, + { + "epoch": 1.193006993006993, + "grad_norm": 0.8209830530865314, + "learning_rate": 1.507870120496316e-05, + "loss": 0.23829127848148346, + "step": 2133 + }, + { + "epoch": 1.1935664335664335, + "grad_norm": 0.2639974523600241, + "learning_rate": 1.5073093596383918e-05, + "loss": 0.06377270817756653, + "step": 2134 + }, + { + "epoch": 1.194125874125874, + "grad_norm": 0.600205783682875, + "learning_rate": 1.5067483838895767e-05, + "loss": 0.29641926288604736, + "step": 2135 + }, + { + "epoch": 1.1946853146853147, + "grad_norm": 0.6985850908549078, + "learning_rate": 1.5061871934874934e-05, + "loss": 0.29890936613082886, + "step": 2136 + }, + { + "epoch": 1.1952447552447552, + "grad_norm": 0.3208994839519978, + "learning_rate": 1.5056257886698566e-05, + "loss": 0.0704234391450882, + "step": 2137 + }, + { + "epoch": 1.1958041958041958, + "grad_norm": 1.029106271507462, + "learning_rate": 1.5050641696744711e-05, + "loss": 0.1706911325454712, + "step": 2138 + }, + { + "epoch": 1.1963636363636363, + "grad_norm": 0.6095530934747178, + "learning_rate": 1.504502336739233e-05, + "loss": 0.2793387174606323, + "step": 2139 + }, + { + "epoch": 1.196923076923077, + "grad_norm": 0.25963512130568717, + "learning_rate": 1.5039402901021292e-05, + "loss": 0.029729411005973816, + "step": 2140 + }, + { + "epoch": 1.1974825174825174, + "grad_norm": 0.5425161682820868, + "learning_rate": 1.5033780300012362e-05, + "loss": 0.23659291863441467, + "step": 2141 + }, + { + "epoch": 1.1980419580419581, + "grad_norm": 0.4225070280721075, + "learning_rate": 1.502815556674721e-05, + "loss": 0.10613439232110977, + "step": 2142 + }, + { + "epoch": 1.1986013986013986, + "grad_norm": 1.4623339247510971, + "learning_rate": 1.5022528703608425e-05, + "loss": 0.34353798627853394, + "step": 2143 + }, + { + "epoch": 1.199160839160839, + "grad_norm": 0.5620928807091438, + "learning_rate": 1.5016899712979478e-05, + "loss": 0.2048952281475067, + "step": 2144 + }, + { + "epoch": 1.1997202797202797, + "grad_norm": 0.6491489522913763, + "learning_rate": 1.5011268597244756e-05, + "loss": 0.17696765065193176, + "step": 2145 + }, + { + "epoch": 1.2002797202797202, + "grad_norm": 0.4120451859779918, + "learning_rate": 1.5005635358789536e-05, + "loss": 0.09329492598772049, + "step": 2146 + }, + { + "epoch": 1.2008391608391609, + "grad_norm": 0.29221510981681, + "learning_rate": 1.5000000000000002e-05, + "loss": 0.060803212225437164, + "step": 2147 + }, + { + "epoch": 1.2013986013986013, + "grad_norm": 0.5981813074972334, + "learning_rate": 1.4994362523263229e-05, + "loss": 0.1763313263654709, + "step": 2148 + }, + { + "epoch": 1.201958041958042, + "grad_norm": 0.7616535874945668, + "learning_rate": 1.4988722930967191e-05, + "loss": 0.3552679419517517, + "step": 2149 + }, + { + "epoch": 1.2025174825174825, + "grad_norm": 0.3534272942026646, + "learning_rate": 1.4983081225500766e-05, + "loss": 0.08860057592391968, + "step": 2150 + }, + { + "epoch": 1.2030769230769232, + "grad_norm": 0.3677115385766433, + "learning_rate": 1.4977437409253717e-05, + "loss": 0.06494182348251343, + "step": 2151 + }, + { + "epoch": 1.2036363636363636, + "grad_norm": 0.7398053198689172, + "learning_rate": 1.4971791484616707e-05, + "loss": 0.2027955800294876, + "step": 2152 + }, + { + "epoch": 1.204195804195804, + "grad_norm": 0.4821051739673429, + "learning_rate": 1.4966143453981286e-05, + "loss": 0.17163538932800293, + "step": 2153 + }, + { + "epoch": 1.2047552447552448, + "grad_norm": 0.3778403968407402, + "learning_rate": 1.4960493319739905e-05, + "loss": 0.12215086817741394, + "step": 2154 + }, + { + "epoch": 1.2053146853146852, + "grad_norm": 0.562766609756667, + "learning_rate": 1.4954841084285899e-05, + "loss": 0.15273746848106384, + "step": 2155 + }, + { + "epoch": 1.205874125874126, + "grad_norm": 0.3033043158416881, + "learning_rate": 1.4949186750013493e-05, + "loss": 0.0647292509675026, + "step": 2156 + }, + { + "epoch": 1.2064335664335664, + "grad_norm": 0.13880203947471295, + "learning_rate": 1.4943530319317806e-05, + "loss": 0.012553836219012737, + "step": 2157 + }, + { + "epoch": 1.206993006993007, + "grad_norm": 0.39072561827718466, + "learning_rate": 1.4937871794594847e-05, + "loss": 0.09680251032114029, + "step": 2158 + }, + { + "epoch": 1.2075524475524475, + "grad_norm": 0.41605778369843266, + "learning_rate": 1.49322111782415e-05, + "loss": 0.12145292013883591, + "step": 2159 + }, + { + "epoch": 1.2081118881118882, + "grad_norm": 0.5798678943078637, + "learning_rate": 1.4926548472655545e-05, + "loss": 0.31201595067977905, + "step": 2160 + }, + { + "epoch": 1.2086713286713286, + "grad_norm": 0.5492861718132338, + "learning_rate": 1.4920883680235643e-05, + "loss": 0.11184802651405334, + "step": 2161 + }, + { + "epoch": 1.209230769230769, + "grad_norm": 0.5228364063094828, + "learning_rate": 1.4915216803381341e-05, + "loss": 0.18981671333312988, + "step": 2162 + }, + { + "epoch": 1.2097902097902098, + "grad_norm": 0.3224726550738173, + "learning_rate": 1.4909547844493074e-05, + "loss": 0.0899486094713211, + "step": 2163 + }, + { + "epoch": 1.2103496503496505, + "grad_norm": 0.4206728207129065, + "learning_rate": 1.4903876805972142e-05, + "loss": 0.06900287419557571, + "step": 2164 + }, + { + "epoch": 1.210909090909091, + "grad_norm": 0.3943173985634231, + "learning_rate": 1.4898203690220742e-05, + "loss": 0.13189850747585297, + "step": 2165 + }, + { + "epoch": 1.2114685314685314, + "grad_norm": 0.31451158473107244, + "learning_rate": 1.4892528499641952e-05, + "loss": 0.10305362939834595, + "step": 2166 + }, + { + "epoch": 1.212027972027972, + "grad_norm": 0.4883191623420662, + "learning_rate": 1.4886851236639715e-05, + "loss": 0.25402671098709106, + "step": 2167 + }, + { + "epoch": 1.2125874125874125, + "grad_norm": 0.38481890504466865, + "learning_rate": 1.4881171903618863e-05, + "loss": 0.07016675174236298, + "step": 2168 + }, + { + "epoch": 1.2131468531468532, + "grad_norm": 0.4338785028882996, + "learning_rate": 1.48754905029851e-05, + "loss": 0.1338023543357849, + "step": 2169 + }, + { + "epoch": 1.2137062937062937, + "grad_norm": 0.5597899671198167, + "learning_rate": 1.4869807037145009e-05, + "loss": 0.2672594487667084, + "step": 2170 + }, + { + "epoch": 1.2142657342657341, + "grad_norm": 0.3843695509112722, + "learning_rate": 1.4864121508506046e-05, + "loss": 0.0858537033200264, + "step": 2171 + }, + { + "epoch": 1.2148251748251748, + "grad_norm": 0.2678222015355069, + "learning_rate": 1.4858433919476543e-05, + "loss": 0.037628188729286194, + "step": 2172 + }, + { + "epoch": 1.2153846153846155, + "grad_norm": 0.25684813197301565, + "learning_rate": 1.4852744272465696e-05, + "loss": 0.05113109573721886, + "step": 2173 + }, + { + "epoch": 1.215944055944056, + "grad_norm": 0.5599814581007939, + "learning_rate": 1.4847052569883585e-05, + "loss": 0.18549975752830505, + "step": 2174 + }, + { + "epoch": 1.2165034965034964, + "grad_norm": 0.24693071341600326, + "learning_rate": 1.4841358814141154e-05, + "loss": 0.04100503772497177, + "step": 2175 + }, + { + "epoch": 1.2170629370629371, + "grad_norm": 0.39989817401928085, + "learning_rate": 1.4835663007650218e-05, + "loss": 0.1358257234096527, + "step": 2176 + }, + { + "epoch": 1.2176223776223776, + "grad_norm": 0.5328542524451414, + "learning_rate": 1.482996515282346e-05, + "loss": 0.16321472823619843, + "step": 2177 + }, + { + "epoch": 1.2181818181818183, + "grad_norm": 0.34567088858643125, + "learning_rate": 1.482426525207443e-05, + "loss": 0.13497884571552277, + "step": 2178 + }, + { + "epoch": 1.2187412587412587, + "grad_norm": 0.7419650486989876, + "learning_rate": 1.4818563307817546e-05, + "loss": 0.21831157803535461, + "step": 2179 + }, + { + "epoch": 1.2193006993006994, + "grad_norm": 0.577484095675839, + "learning_rate": 1.4812859322468099e-05, + "loss": 0.241571843624115, + "step": 2180 + }, + { + "epoch": 1.2198601398601399, + "grad_norm": 0.25180582361600845, + "learning_rate": 1.4807153298442226e-05, + "loss": 0.029868807643651962, + "step": 2181 + }, + { + "epoch": 1.2204195804195805, + "grad_norm": 0.42025612983693245, + "learning_rate": 1.4801445238156942e-05, + "loss": 0.10821191221475601, + "step": 2182 + }, + { + "epoch": 1.220979020979021, + "grad_norm": 0.3869840698752478, + "learning_rate": 1.4795735144030121e-05, + "loss": 0.0779876857995987, + "step": 2183 + }, + { + "epoch": 1.2215384615384615, + "grad_norm": 0.4895092611832519, + "learning_rate": 1.4790023018480502e-05, + "loss": 0.10421731323003769, + "step": 2184 + }, + { + "epoch": 1.2220979020979021, + "grad_norm": 0.48467653424518053, + "learning_rate": 1.478430886392768e-05, + "loss": 0.1073048859834671, + "step": 2185 + }, + { + "epoch": 1.2226573426573426, + "grad_norm": 0.40414508817660105, + "learning_rate": 1.477859268279211e-05, + "loss": 0.15002557635307312, + "step": 2186 + }, + { + "epoch": 1.2232167832167833, + "grad_norm": 1.419764031118243, + "learning_rate": 1.4772874477495104e-05, + "loss": 0.38225501775741577, + "step": 2187 + }, + { + "epoch": 1.2237762237762237, + "grad_norm": 0.47094846488038816, + "learning_rate": 1.4767154250458836e-05, + "loss": 0.12482350319623947, + "step": 2188 + }, + { + "epoch": 1.2243356643356644, + "grad_norm": 0.33510368605229757, + "learning_rate": 1.4761432004106331e-05, + "loss": 0.10538751631975174, + "step": 2189 + }, + { + "epoch": 1.224895104895105, + "grad_norm": 0.3079599406259472, + "learning_rate": 1.4755707740861476e-05, + "loss": 0.06819538027048111, + "step": 2190 + }, + { + "epoch": 1.2254545454545456, + "grad_norm": 0.6232425755597811, + "learning_rate": 1.4749981463149005e-05, + "loss": 0.24481016397476196, + "step": 2191 + }, + { + "epoch": 1.226013986013986, + "grad_norm": 0.5865762120114516, + "learning_rate": 1.4744253173394509e-05, + "loss": 0.22804947197437286, + "step": 2192 + }, + { + "epoch": 1.2265734265734265, + "grad_norm": 0.3774281460337346, + "learning_rate": 1.4738522874024435e-05, + "loss": 0.08752113580703735, + "step": 2193 + }, + { + "epoch": 1.2271328671328672, + "grad_norm": 0.5650527903451826, + "learning_rate": 1.4732790567466074e-05, + "loss": 0.17888322472572327, + "step": 2194 + }, + { + "epoch": 1.2276923076923076, + "grad_norm": 0.2855667774933255, + "learning_rate": 1.4727056256147569e-05, + "loss": 0.0452457033097744, + "step": 2195 + }, + { + "epoch": 1.2282517482517483, + "grad_norm": 0.6256343757382681, + "learning_rate": 1.4721319942497915e-05, + "loss": 0.24215678870677948, + "step": 2196 + }, + { + "epoch": 1.2288111888111888, + "grad_norm": 0.2252184842528938, + "learning_rate": 1.471558162894696e-05, + "loss": 0.03775346279144287, + "step": 2197 + }, + { + "epoch": 1.2293706293706295, + "grad_norm": 0.19882300955537868, + "learning_rate": 1.4709841317925384e-05, + "loss": 0.030302587896585464, + "step": 2198 + }, + { + "epoch": 1.22993006993007, + "grad_norm": 0.3570893777563371, + "learning_rate": 1.470409901186473e-05, + "loss": 0.08427294343709946, + "step": 2199 + }, + { + "epoch": 1.2304895104895106, + "grad_norm": 0.45570588283334795, + "learning_rate": 1.4698354713197371e-05, + "loss": 0.15774285793304443, + "step": 2200 + }, + { + "epoch": 1.231048951048951, + "grad_norm": 0.3380465810305629, + "learning_rate": 1.4692608424356536e-05, + "loss": 0.15683630108833313, + "step": 2201 + }, + { + "epoch": 1.2316083916083915, + "grad_norm": 0.4321632519958417, + "learning_rate": 1.4686860147776292e-05, + "loss": 0.13205881416797638, + "step": 2202 + }, + { + "epoch": 1.2321678321678322, + "grad_norm": 0.9251500944667977, + "learning_rate": 1.4681109885891546e-05, + "loss": 0.16652852296829224, + "step": 2203 + }, + { + "epoch": 1.2327272727272727, + "grad_norm": 0.5388221162456206, + "learning_rate": 1.4675357641138051e-05, + "loss": 0.17532968521118164, + "step": 2204 + }, + { + "epoch": 1.2332867132867134, + "grad_norm": 0.3047627186430227, + "learning_rate": 1.466960341595239e-05, + "loss": 0.0872238501906395, + "step": 2205 + }, + { + "epoch": 1.2338461538461538, + "grad_norm": 0.7647052086342258, + "learning_rate": 1.4663847212772005e-05, + "loss": 0.19453716278076172, + "step": 2206 + }, + { + "epoch": 1.2344055944055945, + "grad_norm": 0.5030466287817682, + "learning_rate": 1.4658089034035152e-05, + "loss": 0.1096837967634201, + "step": 2207 + }, + { + "epoch": 1.234965034965035, + "grad_norm": 0.2704196493986812, + "learning_rate": 1.4652328882180942e-05, + "loss": 0.155201256275177, + "step": 2208 + }, + { + "epoch": 1.2355244755244756, + "grad_norm": 0.29619933036845686, + "learning_rate": 1.4646566759649308e-05, + "loss": 0.06926161050796509, + "step": 2209 + }, + { + "epoch": 1.236083916083916, + "grad_norm": 0.40967494364252827, + "learning_rate": 1.4640802668881027e-05, + "loss": 0.13178397715091705, + "step": 2210 + }, + { + "epoch": 1.2366433566433566, + "grad_norm": 0.21609408115809756, + "learning_rate": 1.4635036612317708e-05, + "loss": 0.026109972968697548, + "step": 2211 + }, + { + "epoch": 1.2372027972027972, + "grad_norm": 0.3934970178281393, + "learning_rate": 1.4629268592401794e-05, + "loss": 0.11033852398395538, + "step": 2212 + }, + { + "epoch": 1.2377622377622377, + "grad_norm": 0.6683918842101253, + "learning_rate": 1.4623498611576549e-05, + "loss": 0.13531993329524994, + "step": 2213 + }, + { + "epoch": 1.2383216783216784, + "grad_norm": 0.4277170713038595, + "learning_rate": 1.4617726672286088e-05, + "loss": 0.11963668465614319, + "step": 2214 + }, + { + "epoch": 1.2388811188811188, + "grad_norm": 0.3210570391291762, + "learning_rate": 1.461195277697533e-05, + "loss": 0.08609425276517868, + "step": 2215 + }, + { + "epoch": 1.2394405594405595, + "grad_norm": 0.4511289602518505, + "learning_rate": 1.460617692809005e-05, + "loss": 0.16657137870788574, + "step": 2216 + }, + { + "epoch": 1.24, + "grad_norm": 0.6330739429696652, + "learning_rate": 1.4600399128076825e-05, + "loss": 0.23990467190742493, + "step": 2217 + }, + { + "epoch": 1.2405594405594407, + "grad_norm": 0.4289560792136445, + "learning_rate": 1.4594619379383075e-05, + "loss": 0.2109047770500183, + "step": 2218 + }, + { + "epoch": 1.2411188811188811, + "grad_norm": 0.352613605643488, + "learning_rate": 1.4588837684457043e-05, + "loss": 0.14184972643852234, + "step": 2219 + }, + { + "epoch": 1.2416783216783216, + "grad_norm": 0.5719003411695138, + "learning_rate": 1.4583054045747789e-05, + "loss": 0.21545779705047607, + "step": 2220 + }, + { + "epoch": 1.2422377622377623, + "grad_norm": 0.3750221666886002, + "learning_rate": 1.4577268465705205e-05, + "loss": 0.10148483514785767, + "step": 2221 + }, + { + "epoch": 1.2427972027972027, + "grad_norm": 0.44695694587863244, + "learning_rate": 1.457148094678e-05, + "loss": 0.11905775219202042, + "step": 2222 + }, + { + "epoch": 1.2433566433566434, + "grad_norm": 0.5940058138199108, + "learning_rate": 1.4565691491423705e-05, + "loss": 0.18257169425487518, + "step": 2223 + }, + { + "epoch": 1.2439160839160839, + "grad_norm": 0.47915529512070787, + "learning_rate": 1.4559900102088676e-05, + "loss": 0.1472766399383545, + "step": 2224 + }, + { + "epoch": 1.2444755244755246, + "grad_norm": 0.5685707329483147, + "learning_rate": 1.4554106781228084e-05, + "loss": 0.12070463597774506, + "step": 2225 + }, + { + "epoch": 1.245034965034965, + "grad_norm": 0.37441796478444317, + "learning_rate": 1.4548311531295913e-05, + "loss": 0.09103168547153473, + "step": 2226 + }, + { + "epoch": 1.2455944055944057, + "grad_norm": 0.3837728357568189, + "learning_rate": 1.4542514354746973e-05, + "loss": 0.11128625273704529, + "step": 2227 + }, + { + "epoch": 1.2461538461538462, + "grad_norm": 0.5191750215615499, + "learning_rate": 1.4536715254036885e-05, + "loss": 0.08916842937469482, + "step": 2228 + }, + { + "epoch": 1.2467132867132866, + "grad_norm": 0.5346177086321672, + "learning_rate": 1.4530914231622095e-05, + "loss": 0.17227885127067566, + "step": 2229 + }, + { + "epoch": 1.2472727272727273, + "grad_norm": 0.7068287181982126, + "learning_rate": 1.4525111289959845e-05, + "loss": 0.13030363619327545, + "step": 2230 + }, + { + "epoch": 1.2478321678321678, + "grad_norm": 0.28085702350789704, + "learning_rate": 1.4519306431508204e-05, + "loss": 0.055643633008003235, + "step": 2231 + }, + { + "epoch": 1.2483916083916085, + "grad_norm": 0.3318122913617973, + "learning_rate": 1.4513499658726049e-05, + "loss": 0.07306533306837082, + "step": 2232 + }, + { + "epoch": 1.248951048951049, + "grad_norm": 0.4945013817831093, + "learning_rate": 1.4507690974073066e-05, + "loss": 0.15479454398155212, + "step": 2233 + }, + { + "epoch": 1.2495104895104896, + "grad_norm": 0.5528420296879332, + "learning_rate": 1.4501880380009756e-05, + "loss": 0.1688430905342102, + "step": 2234 + }, + { + "epoch": 1.25006993006993, + "grad_norm": 0.6148910398104782, + "learning_rate": 1.4496067878997419e-05, + "loss": 0.25149592757225037, + "step": 2235 + }, + { + "epoch": 1.2506293706293707, + "grad_norm": 0.7602044064363678, + "learning_rate": 1.4490253473498177e-05, + "loss": 0.4366171956062317, + "step": 2236 + }, + { + "epoch": 1.2511888111888112, + "grad_norm": 0.2561165915543763, + "learning_rate": 1.4484437165974945e-05, + "loss": 0.033375732600688934, + "step": 2237 + }, + { + "epoch": 1.2517482517482517, + "grad_norm": 0.6293840022704872, + "learning_rate": 1.4478618958891452e-05, + "loss": 0.25928834080696106, + "step": 2238 + }, + { + "epoch": 1.2523076923076923, + "grad_norm": 0.5667250548920968, + "learning_rate": 1.4472798854712229e-05, + "loss": 0.2190554291009903, + "step": 2239 + }, + { + "epoch": 1.2528671328671328, + "grad_norm": 0.4504836023795229, + "learning_rate": 1.4466976855902609e-05, + "loss": 0.16690543293952942, + "step": 2240 + }, + { + "epoch": 1.2534265734265735, + "grad_norm": 0.2250763439015012, + "learning_rate": 1.4461152964928729e-05, + "loss": 0.0784834772348404, + "step": 2241 + }, + { + "epoch": 1.253986013986014, + "grad_norm": 0.43508787330345916, + "learning_rate": 1.4455327184257533e-05, + "loss": 0.16247597336769104, + "step": 2242 + }, + { + "epoch": 1.2545454545454544, + "grad_norm": 0.5330798008359404, + "learning_rate": 1.4449499516356755e-05, + "loss": 0.1705142706632614, + "step": 2243 + }, + { + "epoch": 1.255104895104895, + "grad_norm": 0.21793937900878946, + "learning_rate": 1.4443669963694934e-05, + "loss": 0.021440725773572922, + "step": 2244 + }, + { + "epoch": 1.2556643356643358, + "grad_norm": 0.5104672014576728, + "learning_rate": 1.4437838528741408e-05, + "loss": 0.18354830145835876, + "step": 2245 + }, + { + "epoch": 1.2562237762237762, + "grad_norm": 0.6231297057051832, + "learning_rate": 1.4432005213966312e-05, + "loss": 0.1939404457807541, + "step": 2246 + }, + { + "epoch": 1.2567832167832167, + "grad_norm": 0.29276073380367745, + "learning_rate": 1.4426170021840577e-05, + "loss": 0.07651235908269882, + "step": 2247 + }, + { + "epoch": 1.2573426573426574, + "grad_norm": 0.4059617833753533, + "learning_rate": 1.4420332954835927e-05, + "loss": 0.046428948640823364, + "step": 2248 + }, + { + "epoch": 1.2579020979020978, + "grad_norm": 0.5068925348669214, + "learning_rate": 1.4414494015424883e-05, + "loss": 0.1285722702741623, + "step": 2249 + }, + { + "epoch": 1.2584615384615385, + "grad_norm": 0.27353046632125894, + "learning_rate": 1.4408653206080756e-05, + "loss": 0.06627713143825531, + "step": 2250 + }, + { + "epoch": 1.259020979020979, + "grad_norm": 0.5589701655551564, + "learning_rate": 1.4402810529277656e-05, + "loss": 0.23785154521465302, + "step": 2251 + }, + { + "epoch": 1.2595804195804197, + "grad_norm": 0.465047307720766, + "learning_rate": 1.4396965987490472e-05, + "loss": 0.16897568106651306, + "step": 2252 + }, + { + "epoch": 1.2601398601398601, + "grad_norm": 0.4493914707773903, + "learning_rate": 1.4391119583194899e-05, + "loss": 0.2043132334947586, + "step": 2253 + }, + { + "epoch": 1.2606993006993008, + "grad_norm": 0.2926174009288681, + "learning_rate": 1.4385271318867405e-05, + "loss": 0.0723540410399437, + "step": 2254 + }, + { + "epoch": 1.2612587412587413, + "grad_norm": 0.40488717260924656, + "learning_rate": 1.4379421196985258e-05, + "loss": 0.06294068694114685, + "step": 2255 + }, + { + "epoch": 1.2618181818181817, + "grad_norm": 0.503234017018595, + "learning_rate": 1.4373569220026504e-05, + "loss": 0.17759248614311218, + "step": 2256 + }, + { + "epoch": 1.2623776223776224, + "grad_norm": 0.4988077571969662, + "learning_rate": 1.436771539046998e-05, + "loss": 0.2183111011981964, + "step": 2257 + }, + { + "epoch": 1.2629370629370629, + "grad_norm": 0.24959146993093967, + "learning_rate": 1.4361859710795307e-05, + "loss": 0.06851965188980103, + "step": 2258 + }, + { + "epoch": 1.2634965034965036, + "grad_norm": 0.604266701628363, + "learning_rate": 1.435600218348289e-05, + "loss": 0.1771463304758072, + "step": 2259 + }, + { + "epoch": 1.264055944055944, + "grad_norm": 0.2610140486520115, + "learning_rate": 1.4350142811013915e-05, + "loss": 0.06010114401578903, + "step": 2260 + }, + { + "epoch": 1.2646153846153847, + "grad_norm": 0.41020986598833575, + "learning_rate": 1.4344281595870349e-05, + "loss": 0.13397587835788727, + "step": 2261 + }, + { + "epoch": 1.2651748251748252, + "grad_norm": 0.11000928604532365, + "learning_rate": 1.433841854053494e-05, + "loss": 0.011446855030953884, + "step": 2262 + }, + { + "epoch": 1.2657342657342658, + "grad_norm": 0.26166930455766546, + "learning_rate": 1.4332553647491218e-05, + "loss": 0.042227908968925476, + "step": 2263 + }, + { + "epoch": 1.2662937062937063, + "grad_norm": 0.25722810509775146, + "learning_rate": 1.4326686919223493e-05, + "loss": 0.06095722317695618, + "step": 2264 + }, + { + "epoch": 1.2668531468531468, + "grad_norm": 0.5834599914454022, + "learning_rate": 1.4320818358216848e-05, + "loss": 0.28107377886772156, + "step": 2265 + }, + { + "epoch": 1.2674125874125874, + "grad_norm": 0.3966476182234331, + "learning_rate": 1.4314947966957138e-05, + "loss": 0.1023840680718422, + "step": 2266 + }, + { + "epoch": 1.267972027972028, + "grad_norm": 0.37276247895294784, + "learning_rate": 1.4309075747931002e-05, + "loss": 0.10148217529058456, + "step": 2267 + }, + { + "epoch": 1.2685314685314686, + "grad_norm": 0.3588546188524183, + "learning_rate": 1.4303201703625853e-05, + "loss": 0.11574014276266098, + "step": 2268 + }, + { + "epoch": 1.269090909090909, + "grad_norm": 0.6075291358307413, + "learning_rate": 1.4297325836529872e-05, + "loss": 0.1353980004787445, + "step": 2269 + }, + { + "epoch": 1.2696503496503497, + "grad_norm": 0.24436666151926964, + "learning_rate": 1.4291448149132015e-05, + "loss": 0.0474870502948761, + "step": 2270 + }, + { + "epoch": 1.2702097902097902, + "grad_norm": 0.3406818593962306, + "learning_rate": 1.4285568643922004e-05, + "loss": 0.07048232853412628, + "step": 2271 + }, + { + "epoch": 1.2707692307692309, + "grad_norm": 0.5982876453594432, + "learning_rate": 1.427968732339034e-05, + "loss": 0.12050151824951172, + "step": 2272 + }, + { + "epoch": 1.2713286713286713, + "grad_norm": 0.5309432566176314, + "learning_rate": 1.4273804190028284e-05, + "loss": 0.10930024087429047, + "step": 2273 + }, + { + "epoch": 1.2718881118881118, + "grad_norm": 0.4404751225018033, + "learning_rate": 1.4267919246327877e-05, + "loss": 0.21798774600028992, + "step": 2274 + }, + { + "epoch": 1.2724475524475525, + "grad_norm": 0.47937397258066533, + "learning_rate": 1.4262032494781915e-05, + "loss": 0.21044299006462097, + "step": 2275 + }, + { + "epoch": 1.273006993006993, + "grad_norm": 0.46267144147520006, + "learning_rate": 1.425614393788396e-05, + "loss": 0.12780770659446716, + "step": 2276 + }, + { + "epoch": 1.2735664335664336, + "grad_norm": 0.6216691381992442, + "learning_rate": 1.4250253578128343e-05, + "loss": 0.27876508235931396, + "step": 2277 + }, + { + "epoch": 1.274125874125874, + "grad_norm": 0.38303293671138255, + "learning_rate": 1.4244361418010169e-05, + "loss": 0.05382530391216278, + "step": 2278 + }, + { + "epoch": 1.2746853146853148, + "grad_norm": 0.4892218545188831, + "learning_rate": 1.423846746002528e-05, + "loss": 0.2046329528093338, + "step": 2279 + }, + { + "epoch": 1.2752447552447552, + "grad_norm": 0.9026003781610328, + "learning_rate": 1.4232571706670303e-05, + "loss": 0.3457147777080536, + "step": 2280 + }, + { + "epoch": 1.275804195804196, + "grad_norm": 0.34271324463750696, + "learning_rate": 1.4226674160442617e-05, + "loss": 0.1682617962360382, + "step": 2281 + }, + { + "epoch": 1.2763636363636364, + "grad_norm": 0.4627974092258886, + "learning_rate": 1.4220774823840355e-05, + "loss": 0.2219056487083435, + "step": 2282 + }, + { + "epoch": 1.2769230769230768, + "grad_norm": 0.24758556515331948, + "learning_rate": 1.4214873699362422e-05, + "loss": 0.030912689864635468, + "step": 2283 + }, + { + "epoch": 1.2774825174825175, + "grad_norm": 0.3179845383994855, + "learning_rate": 1.4208970789508463e-05, + "loss": 0.09763117879629135, + "step": 2284 + }, + { + "epoch": 1.278041958041958, + "grad_norm": 0.5111342677735178, + "learning_rate": 1.4203066096778893e-05, + "loss": 0.13511262834072113, + "step": 2285 + }, + { + "epoch": 1.2786013986013987, + "grad_norm": 0.41743382316396443, + "learning_rate": 1.419715962367488e-05, + "loss": 0.12463656812906265, + "step": 2286 + }, + { + "epoch": 1.2791608391608391, + "grad_norm": 0.4289038737962521, + "learning_rate": 1.419125137269834e-05, + "loss": 0.12693163752555847, + "step": 2287 + }, + { + "epoch": 1.2797202797202798, + "grad_norm": 0.36060178112606994, + "learning_rate": 1.4185341346351947e-05, + "loss": 0.10232726484537125, + "step": 2288 + }, + { + "epoch": 1.2802797202797203, + "grad_norm": 0.310914330714065, + "learning_rate": 1.417942954713912e-05, + "loss": 0.09676478058099747, + "step": 2289 + }, + { + "epoch": 1.280839160839161, + "grad_norm": 0.4489932801672495, + "learning_rate": 1.4173515977564045e-05, + "loss": 0.14718085527420044, + "step": 2290 + }, + { + "epoch": 1.2813986013986014, + "grad_norm": 0.5743794508802168, + "learning_rate": 1.4167600640131643e-05, + "loss": 0.28424859046936035, + "step": 2291 + }, + { + "epoch": 1.2819580419580419, + "grad_norm": 0.46627907978214367, + "learning_rate": 1.4161683537347589e-05, + "loss": 0.12832795083522797, + "step": 2292 + }, + { + "epoch": 1.2825174825174825, + "grad_norm": 0.34102631404179823, + "learning_rate": 1.4155764671718302e-05, + "loss": 0.0876542329788208, + "step": 2293 + }, + { + "epoch": 1.283076923076923, + "grad_norm": 0.3893720564290554, + "learning_rate": 1.4149844045750953e-05, + "loss": 0.20542249083518982, + "step": 2294 + }, + { + "epoch": 1.2836363636363637, + "grad_norm": 0.48229221867265865, + "learning_rate": 1.414392166195346e-05, + "loss": 0.13296125829219818, + "step": 2295 + }, + { + "epoch": 1.2841958041958041, + "grad_norm": 0.6929403901849925, + "learning_rate": 1.4137997522834477e-05, + "loss": 0.29220935702323914, + "step": 2296 + }, + { + "epoch": 1.2847552447552448, + "grad_norm": 0.27183656604953715, + "learning_rate": 1.413207163090341e-05, + "loss": 0.07354001700878143, + "step": 2297 + }, + { + "epoch": 1.2853146853146853, + "grad_norm": 0.24262991277829585, + "learning_rate": 1.4126143988670406e-05, + "loss": 0.049216367304325104, + "step": 2298 + }, + { + "epoch": 1.285874125874126, + "grad_norm": 0.6306026669591583, + "learning_rate": 1.4120214598646347e-05, + "loss": 0.15531830489635468, + "step": 2299 + }, + { + "epoch": 1.2864335664335664, + "grad_norm": 0.5119732664779477, + "learning_rate": 1.4114283463342867e-05, + "loss": 0.2570089101791382, + "step": 2300 + }, + { + "epoch": 1.286993006993007, + "grad_norm": 0.19206685657684291, + "learning_rate": 1.4108350585272321e-05, + "loss": 0.017475413158535957, + "step": 2301 + }, + { + "epoch": 1.2875524475524476, + "grad_norm": 0.7352072009425686, + "learning_rate": 1.4102415966947822e-05, + "loss": 0.2496194839477539, + "step": 2302 + }, + { + "epoch": 1.288111888111888, + "grad_norm": 0.5385210681364542, + "learning_rate": 1.4096479610883213e-05, + "loss": 0.10613247752189636, + "step": 2303 + }, + { + "epoch": 1.2886713286713287, + "grad_norm": 0.48675571693089714, + "learning_rate": 1.4090541519593067e-05, + "loss": 0.24276016652584076, + "step": 2304 + }, + { + "epoch": 1.2892307692307692, + "grad_norm": 0.6739256651788433, + "learning_rate": 1.4084601695592699e-05, + "loss": 0.3647720217704773, + "step": 2305 + }, + { + "epoch": 1.2897902097902099, + "grad_norm": 0.31732296318202363, + "learning_rate": 1.4078660141398154e-05, + "loss": 0.06622689217329025, + "step": 2306 + }, + { + "epoch": 1.2903496503496503, + "grad_norm": 0.3739862397183727, + "learning_rate": 1.4072716859526214e-05, + "loss": 0.19807204604148865, + "step": 2307 + }, + { + "epoch": 1.290909090909091, + "grad_norm": 0.23201129001335333, + "learning_rate": 1.406677185249439e-05, + "loss": 0.03155851364135742, + "step": 2308 + }, + { + "epoch": 1.2914685314685315, + "grad_norm": 0.6018760195583196, + "learning_rate": 1.4060825122820925e-05, + "loss": 0.2588830888271332, + "step": 2309 + }, + { + "epoch": 1.292027972027972, + "grad_norm": 0.3713608213668674, + "learning_rate": 1.405487667302479e-05, + "loss": 0.05991354212164879, + "step": 2310 + }, + { + "epoch": 1.2925874125874126, + "grad_norm": 0.3589887529358321, + "learning_rate": 1.4048926505625682e-05, + "loss": 0.10276368260383606, + "step": 2311 + }, + { + "epoch": 1.293146853146853, + "grad_norm": 0.457337663485681, + "learning_rate": 1.404297462314403e-05, + "loss": 0.09990692883729935, + "step": 2312 + }, + { + "epoch": 1.2937062937062938, + "grad_norm": 0.4922921231834744, + "learning_rate": 1.4037021028100996e-05, + "loss": 0.15268345177173615, + "step": 2313 + }, + { + "epoch": 1.2942657342657342, + "grad_norm": 0.3787407097919712, + "learning_rate": 1.4031065723018455e-05, + "loss": 0.1448216736316681, + "step": 2314 + }, + { + "epoch": 1.294825174825175, + "grad_norm": 0.15077737736299748, + "learning_rate": 1.402510871041901e-05, + "loss": 0.022810693830251694, + "step": 2315 + }, + { + "epoch": 1.2953846153846154, + "grad_norm": 0.47166478122340666, + "learning_rate": 1.4019149992825991e-05, + "loss": 0.20819419622421265, + "step": 2316 + }, + { + "epoch": 1.295944055944056, + "grad_norm": 0.47674991541515493, + "learning_rate": 1.4013189572763443e-05, + "loss": 0.1419975608587265, + "step": 2317 + }, + { + "epoch": 1.2965034965034965, + "grad_norm": 0.38805604597219906, + "learning_rate": 1.4007227452756147e-05, + "loss": 0.09433316439390182, + "step": 2318 + }, + { + "epoch": 1.297062937062937, + "grad_norm": 0.47889049278830526, + "learning_rate": 1.4001263635329584e-05, + "loss": 0.1789778769016266, + "step": 2319 + }, + { + "epoch": 1.2976223776223776, + "grad_norm": 0.4179724742610904, + "learning_rate": 1.3995298123009971e-05, + "loss": 0.1831597089767456, + "step": 2320 + }, + { + "epoch": 1.298181818181818, + "grad_norm": 0.48689873816092955, + "learning_rate": 1.3989330918324231e-05, + "loss": 0.17378035187721252, + "step": 2321 + }, + { + "epoch": 1.2987412587412588, + "grad_norm": 0.2808646836530253, + "learning_rate": 1.3983362023800011e-05, + "loss": 0.08113575726747513, + "step": 2322 + }, + { + "epoch": 1.2993006993006992, + "grad_norm": 0.5167316959476286, + "learning_rate": 1.3977391441965676e-05, + "loss": 0.14387616515159607, + "step": 2323 + }, + { + "epoch": 1.29986013986014, + "grad_norm": 0.3822722767192902, + "learning_rate": 1.3971419175350294e-05, + "loss": 0.08410822600126266, + "step": 2324 + }, + { + "epoch": 1.3004195804195804, + "grad_norm": 0.5058388440965752, + "learning_rate": 1.3965445226483655e-05, + "loss": 0.15957938134670258, + "step": 2325 + }, + { + "epoch": 1.300979020979021, + "grad_norm": 0.3665546023214084, + "learning_rate": 1.3959469597896269e-05, + "loss": 0.07351817190647125, + "step": 2326 + }, + { + "epoch": 1.3015384615384615, + "grad_norm": 0.34664808903443384, + "learning_rate": 1.3953492292119343e-05, + "loss": 0.10773013532161713, + "step": 2327 + }, + { + "epoch": 1.302097902097902, + "grad_norm": 0.47309256549178724, + "learning_rate": 1.3947513311684798e-05, + "loss": 0.13009023666381836, + "step": 2328 + }, + { + "epoch": 1.3026573426573427, + "grad_norm": 0.4266867180772355, + "learning_rate": 1.394153265912527e-05, + "loss": 0.07286220788955688, + "step": 2329 + }, + { + "epoch": 1.3032167832167831, + "grad_norm": 0.5744959118586335, + "learning_rate": 1.3935550336974102e-05, + "loss": 0.2962875962257385, + "step": 2330 + }, + { + "epoch": 1.3037762237762238, + "grad_norm": 0.4054875751554663, + "learning_rate": 1.3929566347765341e-05, + "loss": 0.1040998324751854, + "step": 2331 + }, + { + "epoch": 1.3043356643356643, + "grad_norm": 0.520680346107766, + "learning_rate": 1.3923580694033745e-05, + "loss": 0.1252475082874298, + "step": 2332 + }, + { + "epoch": 1.304895104895105, + "grad_norm": 0.4740287283793865, + "learning_rate": 1.3917593378314767e-05, + "loss": 0.15631037950515747, + "step": 2333 + }, + { + "epoch": 1.3054545454545454, + "grad_norm": 0.47513585117436613, + "learning_rate": 1.3911604403144572e-05, + "loss": 0.1472390592098236, + "step": 2334 + }, + { + "epoch": 1.306013986013986, + "grad_norm": 0.380513704309546, + "learning_rate": 1.390561377106003e-05, + "loss": 0.08398403972387314, + "step": 2335 + }, + { + "epoch": 1.3065734265734266, + "grad_norm": 0.4810479613246694, + "learning_rate": 1.389962148459871e-05, + "loss": 0.16348221898078918, + "step": 2336 + }, + { + "epoch": 1.307132867132867, + "grad_norm": 0.48314978162428013, + "learning_rate": 1.3893627546298882e-05, + "loss": 0.08259494602680206, + "step": 2337 + }, + { + "epoch": 1.3076923076923077, + "grad_norm": 0.5120073399509159, + "learning_rate": 1.388763195869951e-05, + "loss": 0.21214108169078827, + "step": 2338 + }, + { + "epoch": 1.3082517482517482, + "grad_norm": 0.5123619848082921, + "learning_rate": 1.3881634724340263e-05, + "loss": 0.18700185418128967, + "step": 2339 + }, + { + "epoch": 1.3088111888111889, + "grad_norm": 0.5182689384766702, + "learning_rate": 1.387563584576151e-05, + "loss": 0.2143566608428955, + "step": 2340 + }, + { + "epoch": 1.3093706293706293, + "grad_norm": 0.29770826875655165, + "learning_rate": 1.3869635325504307e-05, + "loss": 0.0645679384469986, + "step": 2341 + }, + { + "epoch": 1.30993006993007, + "grad_norm": 0.45247555241339243, + "learning_rate": 1.386363316611042e-05, + "loss": 0.12910711765289307, + "step": 2342 + }, + { + "epoch": 1.3104895104895105, + "grad_norm": 0.355835188593108, + "learning_rate": 1.385762937012229e-05, + "loss": 0.03921044245362282, + "step": 2343 + }, + { + "epoch": 1.3110489510489511, + "grad_norm": 0.4481319757130604, + "learning_rate": 1.3851623940083068e-05, + "loss": 0.14040474593639374, + "step": 2344 + }, + { + "epoch": 1.3116083916083916, + "grad_norm": 0.6249847304020423, + "learning_rate": 1.3845616878536588e-05, + "loss": 0.09478513151407242, + "step": 2345 + }, + { + "epoch": 1.312167832167832, + "grad_norm": 0.4786783816333721, + "learning_rate": 1.383960818802738e-05, + "loss": 0.15669310092926025, + "step": 2346 + }, + { + "epoch": 1.3127272727272727, + "grad_norm": 0.2740515178786061, + "learning_rate": 1.383359787110066e-05, + "loss": 0.07649362087249756, + "step": 2347 + }, + { + "epoch": 1.3132867132867134, + "grad_norm": 0.2990568302832124, + "learning_rate": 1.3827585930302336e-05, + "loss": 0.08076094090938568, + "step": 2348 + }, + { + "epoch": 1.3138461538461539, + "grad_norm": 0.3792379473580695, + "learning_rate": 1.3821572368179007e-05, + "loss": 0.0926351398229599, + "step": 2349 + }, + { + "epoch": 1.3144055944055943, + "grad_norm": 0.600180101950028, + "learning_rate": 1.3815557187277944e-05, + "loss": 0.28519099950790405, + "step": 2350 + }, + { + "epoch": 1.314965034965035, + "grad_norm": 0.455306550074267, + "learning_rate": 1.3809540390147121e-05, + "loss": 0.14812952280044556, + "step": 2351 + }, + { + "epoch": 1.3155244755244755, + "grad_norm": 0.6009864741887014, + "learning_rate": 1.3803521979335189e-05, + "loss": 0.1255715787410736, + "step": 2352 + }, + { + "epoch": 1.3160839160839162, + "grad_norm": 0.4036349732755222, + "learning_rate": 1.3797501957391483e-05, + "loss": 0.0979311466217041, + "step": 2353 + }, + { + "epoch": 1.3166433566433566, + "grad_norm": 0.27573591160158145, + "learning_rate": 1.3791480326866022e-05, + "loss": 0.02696126326918602, + "step": 2354 + }, + { + "epoch": 1.317202797202797, + "grad_norm": 0.5351925259331445, + "learning_rate": 1.3785457090309504e-05, + "loss": 0.27828383445739746, + "step": 2355 + }, + { + "epoch": 1.3177622377622378, + "grad_norm": 0.36479631805973944, + "learning_rate": 1.3779432250273306e-05, + "loss": 0.07887479662895203, + "step": 2356 + }, + { + "epoch": 1.3183216783216785, + "grad_norm": 0.478523119592265, + "learning_rate": 1.3773405809309491e-05, + "loss": 0.20843224227428436, + "step": 2357 + }, + { + "epoch": 1.318881118881119, + "grad_norm": 0.5453234431917234, + "learning_rate": 1.3767377769970796e-05, + "loss": 0.22880637645721436, + "step": 2358 + }, + { + "epoch": 1.3194405594405594, + "grad_norm": 0.5303484030504658, + "learning_rate": 1.3761348134810631e-05, + "loss": 0.22276586294174194, + "step": 2359 + }, + { + "epoch": 1.32, + "grad_norm": 0.4223540094402885, + "learning_rate": 1.3755316906383085e-05, + "loss": 0.08788898587226868, + "step": 2360 + }, + { + "epoch": 1.3205594405594405, + "grad_norm": 0.32863694578444913, + "learning_rate": 1.3749284087242927e-05, + "loss": 0.05440403148531914, + "step": 2361 + }, + { + "epoch": 1.3211188811188812, + "grad_norm": 0.2535378971453808, + "learning_rate": 1.3743249679945597e-05, + "loss": 0.0408187098801136, + "step": 2362 + }, + { + "epoch": 1.3216783216783217, + "grad_norm": 0.4238563945002769, + "learning_rate": 1.3737213687047199e-05, + "loss": 0.15900954604148865, + "step": 2363 + }, + { + "epoch": 1.3222377622377621, + "grad_norm": 1.032380072354716, + "learning_rate": 1.373117611110452e-05, + "loss": 0.2538132667541504, + "step": 2364 + }, + { + "epoch": 1.3227972027972028, + "grad_norm": 0.39869574962945314, + "learning_rate": 1.3725136954675016e-05, + "loss": 0.11166108399629593, + "step": 2365 + }, + { + "epoch": 1.3233566433566435, + "grad_norm": 0.5639734188842417, + "learning_rate": 1.3719096220316805e-05, + "loss": 0.20830011367797852, + "step": 2366 + }, + { + "epoch": 1.323916083916084, + "grad_norm": 0.5723715629398017, + "learning_rate": 1.3713053910588683e-05, + "loss": 0.2259216010570526, + "step": 2367 + }, + { + "epoch": 1.3244755244755244, + "grad_norm": 0.5630319310406713, + "learning_rate": 1.3707010028050104e-05, + "loss": 0.23482930660247803, + "step": 2368 + }, + { + "epoch": 1.325034965034965, + "grad_norm": 0.5518017662680549, + "learning_rate": 1.3700964575261198e-05, + "loss": 0.30894967913627625, + "step": 2369 + }, + { + "epoch": 1.3255944055944056, + "grad_norm": 0.16065389366588387, + "learning_rate": 1.3694917554782752e-05, + "loss": 0.019383564591407776, + "step": 2370 + }, + { + "epoch": 1.3261538461538462, + "grad_norm": 0.4109978724977815, + "learning_rate": 1.3688868969176217e-05, + "loss": 0.1656731367111206, + "step": 2371 + }, + { + "epoch": 1.3267132867132867, + "grad_norm": 0.44284621784957034, + "learning_rate": 1.3682818821003716e-05, + "loss": 0.11655168235301971, + "step": 2372 + }, + { + "epoch": 1.3272727272727272, + "grad_norm": 0.6477372489125472, + "learning_rate": 1.3676767112828025e-05, + "loss": 0.15491896867752075, + "step": 2373 + }, + { + "epoch": 1.3278321678321678, + "grad_norm": 0.35169762605630045, + "learning_rate": 1.3670713847212583e-05, + "loss": 0.09042191505432129, + "step": 2374 + }, + { + "epoch": 1.3283916083916085, + "grad_norm": 0.42239627017367487, + "learning_rate": 1.3664659026721492e-05, + "loss": 0.12956613302230835, + "step": 2375 + }, + { + "epoch": 1.328951048951049, + "grad_norm": 0.6219110326087679, + "learning_rate": 1.3658602653919509e-05, + "loss": 0.16507470607757568, + "step": 2376 + }, + { + "epoch": 1.3295104895104894, + "grad_norm": 0.40389092325537723, + "learning_rate": 1.3652544731372046e-05, + "loss": 0.08841853588819504, + "step": 2377 + }, + { + "epoch": 1.3300699300699301, + "grad_norm": 0.49308054854452626, + "learning_rate": 1.3646485261645177e-05, + "loss": 0.1448720246553421, + "step": 2378 + }, + { + "epoch": 1.3306293706293706, + "grad_norm": 0.2751766870266479, + "learning_rate": 1.3640424247305633e-05, + "loss": 0.08173331618309021, + "step": 2379 + }, + { + "epoch": 1.3311888111888113, + "grad_norm": 0.6115671040501733, + "learning_rate": 1.3634361690920797e-05, + "loss": 0.2008461356163025, + "step": 2380 + }, + { + "epoch": 1.3317482517482517, + "grad_norm": 0.6211484358920305, + "learning_rate": 1.3628297595058695e-05, + "loss": 0.31576189398765564, + "step": 2381 + }, + { + "epoch": 1.3323076923076922, + "grad_norm": 0.31823779525584733, + "learning_rate": 1.3622231962288022e-05, + "loss": 0.07697045803070068, + "step": 2382 + }, + { + "epoch": 1.3328671328671329, + "grad_norm": 0.36384801307077735, + "learning_rate": 1.361616479517811e-05, + "loss": 0.09908680617809296, + "step": 2383 + }, + { + "epoch": 1.3334265734265736, + "grad_norm": 0.7791798284051571, + "learning_rate": 1.3610096096298953e-05, + "loss": 0.232676699757576, + "step": 2384 + }, + { + "epoch": 1.333986013986014, + "grad_norm": 0.4333571879373032, + "learning_rate": 1.3604025868221184e-05, + "loss": 0.12593509256839752, + "step": 2385 + }, + { + "epoch": 1.3345454545454545, + "grad_norm": 0.4197712665603814, + "learning_rate": 1.359795411351609e-05, + "loss": 0.20413318276405334, + "step": 2386 + }, + { + "epoch": 1.3351048951048952, + "grad_norm": 0.5176179921361973, + "learning_rate": 1.3591880834755599e-05, + "loss": 0.17946910858154297, + "step": 2387 + }, + { + "epoch": 1.3356643356643356, + "grad_norm": 0.8847038785498248, + "learning_rate": 1.358580603451229e-05, + "loss": 0.27608245611190796, + "step": 2388 + }, + { + "epoch": 1.3362237762237763, + "grad_norm": 0.68207380332707, + "learning_rate": 1.3579729715359386e-05, + "loss": 0.2822544276714325, + "step": 2389 + }, + { + "epoch": 1.3367832167832168, + "grad_norm": 0.7587036421655408, + "learning_rate": 1.3573651879870745e-05, + "loss": 0.2236463725566864, + "step": 2390 + }, + { + "epoch": 1.3373426573426572, + "grad_norm": 0.4050480528457885, + "learning_rate": 1.3567572530620883e-05, + "loss": 0.1276034265756607, + "step": 2391 + }, + { + "epoch": 1.337902097902098, + "grad_norm": 0.44008727985364005, + "learning_rate": 1.356149167018494e-05, + "loss": 0.22586891055107117, + "step": 2392 + }, + { + "epoch": 1.3384615384615386, + "grad_norm": 0.483261556204227, + "learning_rate": 1.3555409301138714e-05, + "loss": 0.11792099475860596, + "step": 2393 + }, + { + "epoch": 1.339020979020979, + "grad_norm": 0.32250044095947117, + "learning_rate": 1.3549325426058622e-05, + "loss": 0.07946892082691193, + "step": 2394 + }, + { + "epoch": 1.3395804195804195, + "grad_norm": 0.3638692281518685, + "learning_rate": 1.3543240047521734e-05, + "loss": 0.08701953291893005, + "step": 2395 + }, + { + "epoch": 1.3401398601398602, + "grad_norm": 0.28532027454553727, + "learning_rate": 1.353715316810575e-05, + "loss": 0.06015447527170181, + "step": 2396 + }, + { + "epoch": 1.3406993006993007, + "grad_norm": 0.36409957790359515, + "learning_rate": 1.353106479038901e-05, + "loss": 0.092784583568573, + "step": 2397 + }, + { + "epoch": 1.3412587412587413, + "grad_norm": 0.33296399910365143, + "learning_rate": 1.3524974916950493e-05, + "loss": 0.12548783421516418, + "step": 2398 + }, + { + "epoch": 1.3418181818181818, + "grad_norm": 0.7062522787577235, + "learning_rate": 1.3518883550369792e-05, + "loss": 0.23377709090709686, + "step": 2399 + }, + { + "epoch": 1.3423776223776223, + "grad_norm": 0.4821394634897971, + "learning_rate": 1.3512790693227151e-05, + "loss": 0.1764272153377533, + "step": 2400 + }, + { + "epoch": 1.342937062937063, + "grad_norm": 0.5354427943883385, + "learning_rate": 1.3506696348103443e-05, + "loss": 0.176019549369812, + "step": 2401 + }, + { + "epoch": 1.3434965034965036, + "grad_norm": 0.3500698572028403, + "learning_rate": 1.3500600517580168e-05, + "loss": 0.04816877096891403, + "step": 2402 + }, + { + "epoch": 1.344055944055944, + "grad_norm": 0.26381495657406334, + "learning_rate": 1.3494503204239451e-05, + "loss": 0.10454526543617249, + "step": 2403 + }, + { + "epoch": 1.3446153846153845, + "grad_norm": 0.5148819278572674, + "learning_rate": 1.3488404410664054e-05, + "loss": 0.20336461067199707, + "step": 2404 + }, + { + "epoch": 1.3451748251748252, + "grad_norm": 0.3082443486119332, + "learning_rate": 1.3482304139437358e-05, + "loss": 0.06843090057373047, + "step": 2405 + }, + { + "epoch": 1.3457342657342657, + "grad_norm": 0.49950834602463356, + "learning_rate": 1.3476202393143377e-05, + "loss": 0.15271905064582825, + "step": 2406 + }, + { + "epoch": 1.3462937062937064, + "grad_norm": 0.3683307710871169, + "learning_rate": 1.3470099174366744e-05, + "loss": 0.09866753220558167, + "step": 2407 + }, + { + "epoch": 1.3468531468531468, + "grad_norm": 0.6110700306833561, + "learning_rate": 1.346399448569272e-05, + "loss": 0.21042823791503906, + "step": 2408 + }, + { + "epoch": 1.3474125874125873, + "grad_norm": 0.44813423761652643, + "learning_rate": 1.3457888329707184e-05, + "loss": 0.11054321378469467, + "step": 2409 + }, + { + "epoch": 1.347972027972028, + "grad_norm": 0.13666744656458696, + "learning_rate": 1.3451780708996642e-05, + "loss": 0.008536288514733315, + "step": 2410 + }, + { + "epoch": 1.3485314685314687, + "grad_norm": 0.38844733951358257, + "learning_rate": 1.3445671626148218e-05, + "loss": 0.09074399620294571, + "step": 2411 + }, + { + "epoch": 1.3490909090909091, + "grad_norm": 0.11320741303455539, + "learning_rate": 1.343956108374965e-05, + "loss": 0.012437272816896439, + "step": 2412 + }, + { + "epoch": 1.3496503496503496, + "grad_norm": 0.20252436324445156, + "learning_rate": 1.3433449084389305e-05, + "loss": 0.013285376131534576, + "step": 2413 + }, + { + "epoch": 1.3502097902097903, + "grad_norm": 0.2684744883511386, + "learning_rate": 1.3427335630656155e-05, + "loss": 0.07112790644168854, + "step": 2414 + }, + { + "epoch": 1.3507692307692307, + "grad_norm": 0.3460967551367594, + "learning_rate": 1.3421220725139805e-05, + "loss": 0.06826774775981903, + "step": 2415 + }, + { + "epoch": 1.3513286713286714, + "grad_norm": 0.44210480580250877, + "learning_rate": 1.3415104370430456e-05, + "loss": 0.1528787612915039, + "step": 2416 + }, + { + "epoch": 1.3518881118881119, + "grad_norm": 0.29916167167178837, + "learning_rate": 1.3408986569118933e-05, + "loss": 0.07178325951099396, + "step": 2417 + }, + { + "epoch": 1.3524475524475523, + "grad_norm": 0.4197982868868124, + "learning_rate": 1.3402867323796672e-05, + "loss": 0.09125544130802155, + "step": 2418 + }, + { + "epoch": 1.353006993006993, + "grad_norm": 0.42175904702492456, + "learning_rate": 1.339674663705572e-05, + "loss": 0.10126252472400665, + "step": 2419 + }, + { + "epoch": 1.3535664335664337, + "grad_norm": 0.2836952678256853, + "learning_rate": 1.3390624511488739e-05, + "loss": 0.04308270290493965, + "step": 2420 + }, + { + "epoch": 1.3541258741258742, + "grad_norm": 0.4588785981577874, + "learning_rate": 1.3384500949688994e-05, + "loss": 0.16726504266262054, + "step": 2421 + }, + { + "epoch": 1.3546853146853146, + "grad_norm": 0.27420420212062085, + "learning_rate": 1.3378375954250363e-05, + "loss": 0.0677989199757576, + "step": 2422 + }, + { + "epoch": 1.3552447552447553, + "grad_norm": 0.38441429334577093, + "learning_rate": 1.3372249527767325e-05, + "loss": 0.12604407966136932, + "step": 2423 + }, + { + "epoch": 1.3558041958041958, + "grad_norm": 0.4574148949045078, + "learning_rate": 1.3366121672834975e-05, + "loss": 0.12370515614748001, + "step": 2424 + }, + { + "epoch": 1.3563636363636364, + "grad_norm": 0.4575244229727621, + "learning_rate": 1.3359992392049002e-05, + "loss": 0.17183727025985718, + "step": 2425 + }, + { + "epoch": 1.356923076923077, + "grad_norm": 0.45399983981987957, + "learning_rate": 1.335386168800571e-05, + "loss": 0.20805403590202332, + "step": 2426 + }, + { + "epoch": 1.3574825174825174, + "grad_norm": 0.5756474198709065, + "learning_rate": 1.3347729563302e-05, + "loss": 0.23937296867370605, + "step": 2427 + }, + { + "epoch": 1.358041958041958, + "grad_norm": 0.45869283795790383, + "learning_rate": 1.334159602053537e-05, + "loss": 0.17234614491462708, + "step": 2428 + }, + { + "epoch": 1.3586013986013987, + "grad_norm": 0.3681707390000976, + "learning_rate": 1.3335461062303932e-05, + "loss": 0.08037231117486954, + "step": 2429 + }, + { + "epoch": 1.3591608391608392, + "grad_norm": 0.5155692817544177, + "learning_rate": 1.3329324691206382e-05, + "loss": 0.2533140182495117, + "step": 2430 + }, + { + "epoch": 1.3597202797202796, + "grad_norm": 0.25651867671863127, + "learning_rate": 1.3323186909842027e-05, + "loss": 0.05418676137924194, + "step": 2431 + }, + { + "epoch": 1.3602797202797203, + "grad_norm": 0.36472850687471814, + "learning_rate": 1.3317047720810766e-05, + "loss": 0.08776415884494781, + "step": 2432 + }, + { + "epoch": 1.3608391608391608, + "grad_norm": 0.5653642105997744, + "learning_rate": 1.331090712671309e-05, + "loss": 0.2018735557794571, + "step": 2433 + }, + { + "epoch": 1.3613986013986015, + "grad_norm": 0.34424642885389634, + "learning_rate": 1.3304765130150096e-05, + "loss": 0.052452098578214645, + "step": 2434 + }, + { + "epoch": 1.361958041958042, + "grad_norm": 0.5124523535138649, + "learning_rate": 1.3298621733723467e-05, + "loss": 0.1496383100748062, + "step": 2435 + }, + { + "epoch": 1.3625174825174824, + "grad_norm": 0.525324871258264, + "learning_rate": 1.3292476940035477e-05, + "loss": 0.17631807923316956, + "step": 2436 + }, + { + "epoch": 1.363076923076923, + "grad_norm": 0.5932128656977157, + "learning_rate": 1.3286330751689e-05, + "loss": 0.3062005341053009, + "step": 2437 + }, + { + "epoch": 1.3636363636363638, + "grad_norm": 0.30172909193356484, + "learning_rate": 1.32801831712875e-05, + "loss": 0.08563196659088135, + "step": 2438 + }, + { + "epoch": 1.3641958041958042, + "grad_norm": 0.6249676905559817, + "learning_rate": 1.3274034201435018e-05, + "loss": 0.27513307332992554, + "step": 2439 + }, + { + "epoch": 1.3647552447552447, + "grad_norm": 0.5564865833687457, + "learning_rate": 1.3267883844736198e-05, + "loss": 0.22013990581035614, + "step": 2440 + }, + { + "epoch": 1.3653146853146854, + "grad_norm": 0.3913335389565812, + "learning_rate": 1.3261732103796264e-05, + "loss": 0.15884727239608765, + "step": 2441 + }, + { + "epoch": 1.3658741258741258, + "grad_norm": 0.5743331212193038, + "learning_rate": 1.3255578981221034e-05, + "loss": 0.24294617772102356, + "step": 2442 + }, + { + "epoch": 1.3664335664335665, + "grad_norm": 0.45219969123099646, + "learning_rate": 1.3249424479616902e-05, + "loss": 0.20455977320671082, + "step": 2443 + }, + { + "epoch": 1.366993006993007, + "grad_norm": 0.4431898035413916, + "learning_rate": 1.3243268601590846e-05, + "loss": 0.0774412453174591, + "step": 2444 + }, + { + "epoch": 1.3675524475524474, + "grad_norm": 0.4785591787403527, + "learning_rate": 1.3237111349750435e-05, + "loss": 0.20527589321136475, + "step": 2445 + }, + { + "epoch": 1.368111888111888, + "grad_norm": 0.4482206525647529, + "learning_rate": 1.3230952726703819e-05, + "loss": 0.11029016971588135, + "step": 2446 + }, + { + "epoch": 1.3686713286713288, + "grad_norm": 0.47484567360703406, + "learning_rate": 1.3224792735059722e-05, + "loss": 0.1555417776107788, + "step": 2447 + }, + { + "epoch": 1.3692307692307693, + "grad_norm": 0.43622515246091437, + "learning_rate": 1.3218631377427448e-05, + "loss": 0.11542768776416779, + "step": 2448 + }, + { + "epoch": 1.3697902097902097, + "grad_norm": 0.4361725436769186, + "learning_rate": 1.3212468656416892e-05, + "loss": 0.08401789516210556, + "step": 2449 + }, + { + "epoch": 1.3703496503496504, + "grad_norm": 0.42839049204767515, + "learning_rate": 1.3206304574638509e-05, + "loss": 0.17544862627983093, + "step": 2450 + }, + { + "epoch": 1.3709090909090909, + "grad_norm": 0.3670394950804874, + "learning_rate": 1.3200139134703346e-05, + "loss": 0.17316821217536926, + "step": 2451 + }, + { + "epoch": 1.3714685314685315, + "grad_norm": 0.509207046779243, + "learning_rate": 1.3193972339223012e-05, + "loss": 0.16223135590553284, + "step": 2452 + }, + { + "epoch": 1.372027972027972, + "grad_norm": 0.5212636306920216, + "learning_rate": 1.31878041908097e-05, + "loss": 0.22562295198440552, + "step": 2453 + }, + { + "epoch": 1.3725874125874125, + "grad_norm": 0.49945780632427483, + "learning_rate": 1.3181634692076173e-05, + "loss": 0.138825923204422, + "step": 2454 + }, + { + "epoch": 1.3731468531468531, + "grad_norm": 0.6065110315456631, + "learning_rate": 1.3175463845635763e-05, + "loss": 0.14975762367248535, + "step": 2455 + }, + { + "epoch": 1.3737062937062938, + "grad_norm": 0.3371201174921658, + "learning_rate": 1.316929165410238e-05, + "loss": 0.09186311811208725, + "step": 2456 + }, + { + "epoch": 1.3742657342657343, + "grad_norm": 0.5689592013463816, + "learning_rate": 1.3163118120090492e-05, + "loss": 0.1791735738515854, + "step": 2457 + }, + { + "epoch": 1.3748251748251747, + "grad_norm": 0.18444801427553145, + "learning_rate": 1.3156943246215148e-05, + "loss": 0.01266934722661972, + "step": 2458 + }, + { + "epoch": 1.3753846153846154, + "grad_norm": 0.26484097762489067, + "learning_rate": 1.3150767035091961e-05, + "loss": 0.03971713408827782, + "step": 2459 + }, + { + "epoch": 1.3759440559440559, + "grad_norm": 0.6162912276790299, + "learning_rate": 1.3144589489337108e-05, + "loss": 0.26717349886894226, + "step": 2460 + }, + { + "epoch": 1.3765034965034966, + "grad_norm": 0.312002562084993, + "learning_rate": 1.3138410611567326e-05, + "loss": 0.028482381254434586, + "step": 2461 + }, + { + "epoch": 1.377062937062937, + "grad_norm": 0.6689296077865052, + "learning_rate": 1.3132230404399926e-05, + "loss": 0.20618808269500732, + "step": 2462 + }, + { + "epoch": 1.3776223776223775, + "grad_norm": 0.4616822449822898, + "learning_rate": 1.312604887045278e-05, + "loss": 0.1066116988658905, + "step": 2463 + }, + { + "epoch": 1.3781818181818182, + "grad_norm": 0.639594405772234, + "learning_rate": 1.3119866012344324e-05, + "loss": 0.2201676070690155, + "step": 2464 + }, + { + "epoch": 1.3787412587412589, + "grad_norm": 0.4199384739513294, + "learning_rate": 1.3113681832693548e-05, + "loss": 0.07959502935409546, + "step": 2465 + }, + { + "epoch": 1.3793006993006993, + "grad_norm": 0.4075191406807767, + "learning_rate": 1.310749633412e-05, + "loss": 0.1027555763721466, + "step": 2466 + }, + { + "epoch": 1.3798601398601398, + "grad_norm": 0.3808020870496113, + "learning_rate": 1.31013095192438e-05, + "loss": 0.11271212995052338, + "step": 2467 + }, + { + "epoch": 1.3804195804195805, + "grad_norm": 0.31179067997863413, + "learning_rate": 1.309512139068561e-05, + "loss": 0.08231376111507416, + "step": 2468 + }, + { + "epoch": 1.380979020979021, + "grad_norm": 0.4715109380699304, + "learning_rate": 1.3088931951066663e-05, + "loss": 0.12280185520648956, + "step": 2469 + }, + { + "epoch": 1.3815384615384616, + "grad_norm": 0.42537495778062745, + "learning_rate": 1.3082741203008737e-05, + "loss": 0.09480578452348709, + "step": 2470 + }, + { + "epoch": 1.382097902097902, + "grad_norm": 0.507604731504221, + "learning_rate": 1.3076549149134167e-05, + "loss": 0.22818467020988464, + "step": 2471 + }, + { + "epoch": 1.3826573426573425, + "grad_norm": 0.32372613648105986, + "learning_rate": 1.3070355792065845e-05, + "loss": 0.1664549559354782, + "step": 2472 + }, + { + "epoch": 1.3832167832167832, + "grad_norm": 0.4407989088619798, + "learning_rate": 1.306416113442721e-05, + "loss": 0.17842406034469604, + "step": 2473 + }, + { + "epoch": 1.383776223776224, + "grad_norm": 0.4284244589969576, + "learning_rate": 1.3057965178842249e-05, + "loss": 0.1323685199022293, + "step": 2474 + }, + { + "epoch": 1.3843356643356644, + "grad_norm": 0.5075666873861242, + "learning_rate": 1.3051767927935508e-05, + "loss": 0.09644083678722382, + "step": 2475 + }, + { + "epoch": 1.3848951048951048, + "grad_norm": 0.6866400422670028, + "learning_rate": 1.3045569384332075e-05, + "loss": 0.3673185110092163, + "step": 2476 + }, + { + "epoch": 1.3854545454545455, + "grad_norm": 0.451667102791491, + "learning_rate": 1.3039369550657594e-05, + "loss": 0.16871073842048645, + "step": 2477 + }, + { + "epoch": 1.386013986013986, + "grad_norm": 0.528498463849333, + "learning_rate": 1.3033168429538243e-05, + "loss": 0.22378090023994446, + "step": 2478 + }, + { + "epoch": 1.3865734265734266, + "grad_norm": 0.3564240348328953, + "learning_rate": 1.3026966023600751e-05, + "loss": 0.11769779771566391, + "step": 2479 + }, + { + "epoch": 1.387132867132867, + "grad_norm": 0.44136408278301414, + "learning_rate": 1.3020762335472393e-05, + "loss": 0.0978880524635315, + "step": 2480 + }, + { + "epoch": 1.3876923076923076, + "grad_norm": 0.5163311468915397, + "learning_rate": 1.3014557367780984e-05, + "loss": 0.229050874710083, + "step": 2481 + }, + { + "epoch": 1.3882517482517482, + "grad_norm": 0.24490277057448775, + "learning_rate": 1.3008351123154889e-05, + "loss": 0.04241913557052612, + "step": 2482 + }, + { + "epoch": 1.388811188811189, + "grad_norm": 0.5242502542298719, + "learning_rate": 1.3002143604223006e-05, + "loss": 0.13000048696994781, + "step": 2483 + }, + { + "epoch": 1.3893706293706294, + "grad_norm": 0.37799318684144706, + "learning_rate": 1.2995934813614768e-05, + "loss": 0.09359068423509598, + "step": 2484 + }, + { + "epoch": 1.3899300699300698, + "grad_norm": 0.5628186290556391, + "learning_rate": 1.2989724753960155e-05, + "loss": 0.3084734380245209, + "step": 2485 + }, + { + "epoch": 1.3904895104895105, + "grad_norm": 0.4865621545689211, + "learning_rate": 1.2983513427889692e-05, + "loss": 0.19326458871364594, + "step": 2486 + }, + { + "epoch": 1.391048951048951, + "grad_norm": 0.8009010711146856, + "learning_rate": 1.2977300838034422e-05, + "loss": 0.26621443033218384, + "step": 2487 + }, + { + "epoch": 1.3916083916083917, + "grad_norm": 0.2852753125759789, + "learning_rate": 1.2971086987025933e-05, + "loss": 0.06607040762901306, + "step": 2488 + }, + { + "epoch": 1.3921678321678321, + "grad_norm": 0.5044080632515159, + "learning_rate": 1.2964871877496347e-05, + "loss": 0.2056734710931778, + "step": 2489 + }, + { + "epoch": 1.3927272727272726, + "grad_norm": 0.5547761969215161, + "learning_rate": 1.2958655512078324e-05, + "loss": 0.21388250589370728, + "step": 2490 + }, + { + "epoch": 1.3932867132867133, + "grad_norm": 0.37581848583864247, + "learning_rate": 1.2952437893405048e-05, + "loss": 0.0713212862610817, + "step": 2491 + }, + { + "epoch": 1.393846153846154, + "grad_norm": 0.2933238331154663, + "learning_rate": 1.2946219024110232e-05, + "loss": 0.06287550926208496, + "step": 2492 + }, + { + "epoch": 1.3944055944055944, + "grad_norm": 0.48177006258308686, + "learning_rate": 1.2939998906828131e-05, + "loss": 0.1461336314678192, + "step": 2493 + }, + { + "epoch": 1.3949650349650349, + "grad_norm": 0.5322627115052949, + "learning_rate": 1.2933777544193514e-05, + "loss": 0.16389258205890656, + "step": 2494 + }, + { + "epoch": 1.3955244755244756, + "grad_norm": 0.3757977572755719, + "learning_rate": 1.2927554938841689e-05, + "loss": 0.07075624167919159, + "step": 2495 + }, + { + "epoch": 1.396083916083916, + "grad_norm": 0.5055200174086182, + "learning_rate": 1.2921331093408487e-05, + "loss": 0.2380932867527008, + "step": 2496 + }, + { + "epoch": 1.3966433566433567, + "grad_norm": 0.278476854972628, + "learning_rate": 1.2915106010530259e-05, + "loss": 0.06349461525678635, + "step": 2497 + }, + { + "epoch": 1.3972027972027972, + "grad_norm": 0.478150638108473, + "learning_rate": 1.2908879692843883e-05, + "loss": 0.19395595788955688, + "step": 2498 + }, + { + "epoch": 1.3977622377622376, + "grad_norm": 0.6611864858378667, + "learning_rate": 1.2902652142986771e-05, + "loss": 0.2335360199213028, + "step": 2499 + }, + { + "epoch": 1.3983216783216783, + "grad_norm": 0.22214124673807756, + "learning_rate": 1.289642336359684e-05, + "loss": 0.018601372838020325, + "step": 2500 + }, + { + "epoch": 1.398881118881119, + "grad_norm": 0.4867816203770311, + "learning_rate": 1.2890193357312534e-05, + "loss": 0.1681356132030487, + "step": 2501 + }, + { + "epoch": 1.3994405594405595, + "grad_norm": 0.26660577689011433, + "learning_rate": 1.2883962126772823e-05, + "loss": 0.0960170328617096, + "step": 2502 + }, + { + "epoch": 1.4, + "grad_norm": 0.3558507087069031, + "learning_rate": 1.2877729674617185e-05, + "loss": 0.06612128019332886, + "step": 2503 + }, + { + "epoch": 1.4005594405594406, + "grad_norm": 0.4876891690886492, + "learning_rate": 1.2871496003485628e-05, + "loss": 0.1970897614955902, + "step": 2504 + }, + { + "epoch": 1.401118881118881, + "grad_norm": 0.5021325081343816, + "learning_rate": 1.2865261116018664e-05, + "loss": 0.29299888014793396, + "step": 2505 + }, + { + "epoch": 1.4016783216783217, + "grad_norm": 0.40226805113128034, + "learning_rate": 1.2859025014857327e-05, + "loss": 0.09906235337257385, + "step": 2506 + }, + { + "epoch": 1.4022377622377622, + "grad_norm": 0.6044651682287532, + "learning_rate": 1.2852787702643162e-05, + "loss": 0.21030399203300476, + "step": 2507 + }, + { + "epoch": 1.4027972027972029, + "grad_norm": 0.46026703001084435, + "learning_rate": 1.284654918201823e-05, + "loss": 0.11501772701740265, + "step": 2508 + }, + { + "epoch": 1.4033566433566433, + "grad_norm": 0.47643083701962524, + "learning_rate": 1.2840309455625103e-05, + "loss": 0.15339097380638123, + "step": 2509 + }, + { + "epoch": 1.403916083916084, + "grad_norm": 0.453494778797613, + "learning_rate": 1.2834068526106865e-05, + "loss": 0.25876539945602417, + "step": 2510 + }, + { + "epoch": 1.4044755244755245, + "grad_norm": 0.48083993952654575, + "learning_rate": 1.2827826396107103e-05, + "loss": 0.16346001625061035, + "step": 2511 + }, + { + "epoch": 1.405034965034965, + "grad_norm": 0.48760512296698655, + "learning_rate": 1.282158306826992e-05, + "loss": 0.1334468573331833, + "step": 2512 + }, + { + "epoch": 1.4055944055944056, + "grad_norm": 0.4918092881128028, + "learning_rate": 1.2815338545239926e-05, + "loss": 0.1394098550081253, + "step": 2513 + }, + { + "epoch": 1.406153846153846, + "grad_norm": 0.34764693096681526, + "learning_rate": 1.2809092829662233e-05, + "loss": 0.061817608773708344, + "step": 2514 + }, + { + "epoch": 1.4067132867132868, + "grad_norm": 0.5508648428859525, + "learning_rate": 1.2802845924182457e-05, + "loss": 0.15196193754673004, + "step": 2515 + }, + { + "epoch": 1.4072727272727272, + "grad_norm": 0.36896890448581243, + "learning_rate": 1.2796597831446728e-05, + "loss": 0.07467684149742126, + "step": 2516 + }, + { + "epoch": 1.407832167832168, + "grad_norm": 0.4682598601092101, + "learning_rate": 1.2790348554101666e-05, + "loss": 0.16751444339752197, + "step": 2517 + }, + { + "epoch": 1.4083916083916084, + "grad_norm": 0.6108607536151017, + "learning_rate": 1.2784098094794406e-05, + "loss": 0.2549717128276825, + "step": 2518 + }, + { + "epoch": 1.408951048951049, + "grad_norm": 0.3936275234064424, + "learning_rate": 1.2777846456172568e-05, + "loss": 0.06868081539869308, + "step": 2519 + }, + { + "epoch": 1.4095104895104895, + "grad_norm": 0.6117413786351957, + "learning_rate": 1.2771593640884283e-05, + "loss": 0.28834298253059387, + "step": 2520 + }, + { + "epoch": 1.41006993006993, + "grad_norm": 0.4682572717246156, + "learning_rate": 1.2765339651578183e-05, + "loss": 0.09444493055343628, + "step": 2521 + }, + { + "epoch": 1.4106293706293707, + "grad_norm": 0.53520672316135, + "learning_rate": 1.2759084490903388e-05, + "loss": 0.20549622178077698, + "step": 2522 + }, + { + "epoch": 1.4111888111888111, + "grad_norm": 0.5484146915498206, + "learning_rate": 1.2752828161509514e-05, + "loss": 0.13213779032230377, + "step": 2523 + }, + { + "epoch": 1.4117482517482518, + "grad_norm": 0.6074415643145639, + "learning_rate": 1.2746570666046681e-05, + "loss": 0.2385183572769165, + "step": 2524 + }, + { + "epoch": 1.4123076923076923, + "grad_norm": 0.5489528561822083, + "learning_rate": 1.2740312007165496e-05, + "loss": 0.12792335450649261, + "step": 2525 + }, + { + "epoch": 1.412867132867133, + "grad_norm": 0.32606437121271203, + "learning_rate": 1.2734052187517065e-05, + "loss": 0.06595709919929504, + "step": 2526 + }, + { + "epoch": 1.4134265734265734, + "grad_norm": 0.5810339141121802, + "learning_rate": 1.2727791209752977e-05, + "loss": 0.39011693000793457, + "step": 2527 + }, + { + "epoch": 1.413986013986014, + "grad_norm": 0.43140659518643454, + "learning_rate": 1.2721529076525318e-05, + "loss": 0.27636608481407166, + "step": 2528 + }, + { + "epoch": 1.4145454545454546, + "grad_norm": 0.4538452733278322, + "learning_rate": 1.271526579048666e-05, + "loss": 0.15842606127262115, + "step": 2529 + }, + { + "epoch": 1.415104895104895, + "grad_norm": 0.3078666686709828, + "learning_rate": 1.2709001354290064e-05, + "loss": 0.0735919326543808, + "step": 2530 + }, + { + "epoch": 1.4156643356643357, + "grad_norm": 0.5852552953287177, + "learning_rate": 1.2702735770589085e-05, + "loss": 0.20895975828170776, + "step": 2531 + }, + { + "epoch": 1.4162237762237762, + "grad_norm": 0.8894252353403835, + "learning_rate": 1.2696469042037752e-05, + "loss": 0.34306687116622925, + "step": 2532 + }, + { + "epoch": 1.4167832167832168, + "grad_norm": 0.31013749485128705, + "learning_rate": 1.2690201171290588e-05, + "loss": 0.1085163950920105, + "step": 2533 + }, + { + "epoch": 1.4173426573426573, + "grad_norm": 0.23145081163468428, + "learning_rate": 1.2683932161002596e-05, + "loss": 0.025592666119337082, + "step": 2534 + }, + { + "epoch": 1.417902097902098, + "grad_norm": 0.29350724366721664, + "learning_rate": 1.2677662013829265e-05, + "loss": 0.09960824251174927, + "step": 2535 + }, + { + "epoch": 1.4184615384615384, + "grad_norm": 0.4141995491826718, + "learning_rate": 1.2671390732426553e-05, + "loss": 0.10015767812728882, + "step": 2536 + }, + { + "epoch": 1.4190209790209791, + "grad_norm": 0.435252123802241, + "learning_rate": 1.2665118319450925e-05, + "loss": 0.11094857007265091, + "step": 2537 + }, + { + "epoch": 1.4195804195804196, + "grad_norm": 0.4228715518674304, + "learning_rate": 1.2658844777559297e-05, + "loss": 0.1043037548661232, + "step": 2538 + }, + { + "epoch": 1.42013986013986, + "grad_norm": 0.5577896040960031, + "learning_rate": 1.2652570109409077e-05, + "loss": 0.26796773076057434, + "step": 2539 + }, + { + "epoch": 1.4206993006993007, + "grad_norm": 0.2740884848772622, + "learning_rate": 1.2646294317658151e-05, + "loss": 0.03737884759902954, + "step": 2540 + }, + { + "epoch": 1.4212587412587412, + "grad_norm": 0.49846874094794946, + "learning_rate": 1.2640017404964877e-05, + "loss": 0.19634473323822021, + "step": 2541 + }, + { + "epoch": 1.4218181818181819, + "grad_norm": 0.7826687130947978, + "learning_rate": 1.263373937398809e-05, + "loss": 0.22753392159938812, + "step": 2542 + }, + { + "epoch": 1.4223776223776223, + "grad_norm": 0.4541928252380977, + "learning_rate": 1.2627460227387095e-05, + "loss": 0.19683019816875458, + "step": 2543 + }, + { + "epoch": 1.422937062937063, + "grad_norm": 0.33423808940376604, + "learning_rate": 1.2621179967821675e-05, + "loss": 0.11294884234666824, + "step": 2544 + }, + { + "epoch": 1.4234965034965035, + "grad_norm": 0.5020077938952384, + "learning_rate": 1.2614898597952084e-05, + "loss": 0.21752138435840607, + "step": 2545 + }, + { + "epoch": 1.4240559440559442, + "grad_norm": 0.40847451529532114, + "learning_rate": 1.2608616120439034e-05, + "loss": 0.1418994665145874, + "step": 2546 + }, + { + "epoch": 1.4246153846153846, + "grad_norm": 0.3556228836352996, + "learning_rate": 1.2602332537943727e-05, + "loss": 0.14981484413146973, + "step": 2547 + }, + { + "epoch": 1.425174825174825, + "grad_norm": 0.4788408690740898, + "learning_rate": 1.2596047853127818e-05, + "loss": 0.16420406103134155, + "step": 2548 + }, + { + "epoch": 1.4257342657342658, + "grad_norm": 0.4648494527617643, + "learning_rate": 1.2589762068653434e-05, + "loss": 0.180812269449234, + "step": 2549 + }, + { + "epoch": 1.4262937062937062, + "grad_norm": 0.41122714407763034, + "learning_rate": 1.258347518718316e-05, + "loss": 0.16212785243988037, + "step": 2550 + }, + { + "epoch": 1.426853146853147, + "grad_norm": 0.3496629741045092, + "learning_rate": 1.2577187211380062e-05, + "loss": 0.11519194394350052, + "step": 2551 + }, + { + "epoch": 1.4274125874125874, + "grad_norm": 0.31045605647536034, + "learning_rate": 1.2570898143907657e-05, + "loss": 0.05389915406703949, + "step": 2552 + }, + { + "epoch": 1.427972027972028, + "grad_norm": 0.4343697140138016, + "learning_rate": 1.2564607987429928e-05, + "loss": 0.20748907327651978, + "step": 2553 + }, + { + "epoch": 1.4285314685314685, + "grad_norm": 0.36142223348223124, + "learning_rate": 1.255831674461132e-05, + "loss": 0.07955549657344818, + "step": 2554 + }, + { + "epoch": 1.4290909090909092, + "grad_norm": 0.475480191210728, + "learning_rate": 1.2552024418116735e-05, + "loss": 0.16949276626110077, + "step": 2555 + }, + { + "epoch": 1.4296503496503496, + "grad_norm": 0.4991975101764169, + "learning_rate": 1.2545731010611535e-05, + "loss": 0.1735232174396515, + "step": 2556 + }, + { + "epoch": 1.43020979020979, + "grad_norm": 0.42294948054166076, + "learning_rate": 1.2539436524761542e-05, + "loss": 0.1384444534778595, + "step": 2557 + }, + { + "epoch": 1.4307692307692308, + "grad_norm": 0.3809482036879326, + "learning_rate": 1.2533140963233037e-05, + "loss": 0.08492761105298996, + "step": 2558 + }, + { + "epoch": 1.4313286713286713, + "grad_norm": 0.5790978807247387, + "learning_rate": 1.2526844328692748e-05, + "loss": 0.1551174521446228, + "step": 2559 + }, + { + "epoch": 1.431888111888112, + "grad_norm": 0.5767742003720998, + "learning_rate": 1.252054662380787e-05, + "loss": 0.2093212604522705, + "step": 2560 + }, + { + "epoch": 1.4324475524475524, + "grad_norm": 0.6202487887471785, + "learning_rate": 1.2514247851246038e-05, + "loss": 0.17169030010700226, + "step": 2561 + }, + { + "epoch": 1.433006993006993, + "grad_norm": 0.19830630472231092, + "learning_rate": 1.2507948013675346e-05, + "loss": 0.015126602724194527, + "step": 2562 + }, + { + "epoch": 1.4335664335664335, + "grad_norm": 0.5658303191506462, + "learning_rate": 1.2501647113764342e-05, + "loss": 0.19893920421600342, + "step": 2563 + }, + { + "epoch": 1.4341258741258742, + "grad_norm": 0.385541421532639, + "learning_rate": 1.2495345154182018e-05, + "loss": 0.29627296328544617, + "step": 2564 + }, + { + "epoch": 1.4346853146853147, + "grad_norm": 0.34546040335160755, + "learning_rate": 1.2489042137597817e-05, + "loss": 0.09909462928771973, + "step": 2565 + }, + { + "epoch": 1.4352447552447551, + "grad_norm": 0.2102809822423523, + "learning_rate": 1.2482738066681636e-05, + "loss": 0.053327735513448715, + "step": 2566 + }, + { + "epoch": 1.4358041958041958, + "grad_norm": 0.3377918872814062, + "learning_rate": 1.2476432944103811e-05, + "loss": 0.0766044333577156, + "step": 2567 + }, + { + "epoch": 1.4363636363636363, + "grad_norm": 0.8605033601117638, + "learning_rate": 1.2470126772535118e-05, + "loss": 0.12152018398046494, + "step": 2568 + }, + { + "epoch": 1.436923076923077, + "grad_norm": 0.4942473483146406, + "learning_rate": 1.2463819554646792e-05, + "loss": 0.18582960963249207, + "step": 2569 + }, + { + "epoch": 1.4374825174825174, + "grad_norm": 0.32474170255803075, + "learning_rate": 1.24575112931105e-05, + "loss": 0.08280140906572342, + "step": 2570 + }, + { + "epoch": 1.4380419580419581, + "grad_norm": 0.5055941663959481, + "learning_rate": 1.2451201990598362e-05, + "loss": 0.17425097525119781, + "step": 2571 + }, + { + "epoch": 1.4386013986013986, + "grad_norm": 0.15353793984887606, + "learning_rate": 1.2444891649782925e-05, + "loss": 0.01326349750161171, + "step": 2572 + }, + { + "epoch": 1.4391608391608393, + "grad_norm": 0.6973030241264379, + "learning_rate": 1.2438580273337185e-05, + "loss": 0.2997506856918335, + "step": 2573 + }, + { + "epoch": 1.4397202797202797, + "grad_norm": 0.33225188501482966, + "learning_rate": 1.243226786393457e-05, + "loss": 0.05882330983877182, + "step": 2574 + }, + { + "epoch": 1.4402797202797202, + "grad_norm": 0.30430660478344274, + "learning_rate": 1.2425954424248959e-05, + "loss": 0.05277079716324806, + "step": 2575 + }, + { + "epoch": 1.4408391608391609, + "grad_norm": 0.5081201463489204, + "learning_rate": 1.2419639956954653e-05, + "loss": 0.15838435292243958, + "step": 2576 + }, + { + "epoch": 1.4413986013986013, + "grad_norm": 0.3799021798411323, + "learning_rate": 1.2413324464726393e-05, + "loss": 0.12325419485569, + "step": 2577 + }, + { + "epoch": 1.441958041958042, + "grad_norm": 0.5521556745577954, + "learning_rate": 1.2407007950239356e-05, + "loss": 0.3183799982070923, + "step": 2578 + }, + { + "epoch": 1.4425174825174825, + "grad_norm": 0.3069883184603065, + "learning_rate": 1.2400690416169153e-05, + "loss": 0.07133820652961731, + "step": 2579 + }, + { + "epoch": 1.4430769230769231, + "grad_norm": 0.35045151224194626, + "learning_rate": 1.2394371865191821e-05, + "loss": 0.17543654143810272, + "step": 2580 + }, + { + "epoch": 1.4436363636363636, + "grad_norm": 0.38845365754259525, + "learning_rate": 1.2388052299983833e-05, + "loss": 0.1259579211473465, + "step": 2581 + }, + { + "epoch": 1.4441958041958043, + "grad_norm": 0.5120885686971979, + "learning_rate": 1.2381731723222092e-05, + "loss": 0.19516217708587646, + "step": 2582 + }, + { + "epoch": 1.4447552447552447, + "grad_norm": 1.1893187922398907, + "learning_rate": 1.2375410137583927e-05, + "loss": 0.3713221549987793, + "step": 2583 + }, + { + "epoch": 1.4453146853146852, + "grad_norm": 0.3862825432632837, + "learning_rate": 1.2369087545747093e-05, + "loss": 0.12221759557723999, + "step": 2584 + }, + { + "epoch": 1.445874125874126, + "grad_norm": 0.6383279488482578, + "learning_rate": 1.2362763950389776e-05, + "loss": 0.24273799359798431, + "step": 2585 + }, + { + "epoch": 1.4464335664335664, + "grad_norm": 0.44816653153497826, + "learning_rate": 1.2356439354190578e-05, + "loss": 0.17039065062999725, + "step": 2586 + }, + { + "epoch": 1.446993006993007, + "grad_norm": 0.3811322292275182, + "learning_rate": 1.235011375982854e-05, + "loss": 0.045549459755420685, + "step": 2587 + }, + { + "epoch": 1.4475524475524475, + "grad_norm": 0.3451624936443498, + "learning_rate": 1.2343787169983112e-05, + "loss": 0.05645807832479477, + "step": 2588 + }, + { + "epoch": 1.4481118881118882, + "grad_norm": 0.2833937999869843, + "learning_rate": 1.2337459587334173e-05, + "loss": 0.03421954810619354, + "step": 2589 + }, + { + "epoch": 1.4486713286713286, + "grad_norm": 0.4245286415459919, + "learning_rate": 1.2331131014562017e-05, + "loss": 0.15909573435783386, + "step": 2590 + }, + { + "epoch": 1.4492307692307693, + "grad_norm": 0.5722517138083721, + "learning_rate": 1.2324801454347362e-05, + "loss": 0.27141815423965454, + "step": 2591 + }, + { + "epoch": 1.4497902097902098, + "grad_norm": 0.4668936365913203, + "learning_rate": 1.2318470909371344e-05, + "loss": 0.13169139623641968, + "step": 2592 + }, + { + "epoch": 1.4503496503496502, + "grad_norm": 0.3929995263451801, + "learning_rate": 1.2312139382315514e-05, + "loss": 0.10707405209541321, + "step": 2593 + }, + { + "epoch": 1.450909090909091, + "grad_norm": 0.5154721894124208, + "learning_rate": 1.230580687586184e-05, + "loss": 0.1734994798898697, + "step": 2594 + }, + { + "epoch": 1.4514685314685314, + "grad_norm": 0.2938263521149777, + "learning_rate": 1.2299473392692707e-05, + "loss": 0.07255610823631287, + "step": 2595 + }, + { + "epoch": 1.452027972027972, + "grad_norm": 0.41641454551835116, + "learning_rate": 1.2293138935490907e-05, + "loss": 0.132124662399292, + "step": 2596 + }, + { + "epoch": 1.4525874125874125, + "grad_norm": 0.5011831949079908, + "learning_rate": 1.2286803506939655e-05, + "loss": 0.1521194577217102, + "step": 2597 + }, + { + "epoch": 1.4531468531468532, + "grad_norm": 0.34696323401412626, + "learning_rate": 1.2280467109722566e-05, + "loss": 0.08124281466007233, + "step": 2598 + }, + { + "epoch": 1.4537062937062937, + "grad_norm": 0.46399336751166637, + "learning_rate": 1.2274129746523678e-05, + "loss": 0.10194060206413269, + "step": 2599 + }, + { + "epoch": 1.4542657342657344, + "grad_norm": 0.36012940059062953, + "learning_rate": 1.2267791420027426e-05, + "loss": 0.11040084064006805, + "step": 2600 + }, + { + "epoch": 1.4548251748251748, + "grad_norm": 0.4637359684465266, + "learning_rate": 1.2261452132918656e-05, + "loss": 0.12690399587154388, + "step": 2601 + }, + { + "epoch": 1.4553846153846153, + "grad_norm": 0.5136509540102716, + "learning_rate": 1.225511188788263e-05, + "loss": 0.17024722695350647, + "step": 2602 + }, + { + "epoch": 1.455944055944056, + "grad_norm": 0.4068310316544381, + "learning_rate": 1.2248770687605002e-05, + "loss": 0.07141393423080444, + "step": 2603 + }, + { + "epoch": 1.4565034965034966, + "grad_norm": 0.2238765679641551, + "learning_rate": 1.2242428534771844e-05, + "loss": 0.025211479514837265, + "step": 2604 + }, + { + "epoch": 1.457062937062937, + "grad_norm": 0.5082974980759168, + "learning_rate": 1.2236085432069623e-05, + "loss": 0.12128719687461853, + "step": 2605 + }, + { + "epoch": 1.4576223776223776, + "grad_norm": 0.6194828312486111, + "learning_rate": 1.2229741382185202e-05, + "loss": 0.2586372494697571, + "step": 2606 + }, + { + "epoch": 1.4581818181818182, + "grad_norm": 0.4032120234149784, + "learning_rate": 1.2223396387805867e-05, + "loss": 0.11020811647176743, + "step": 2607 + }, + { + "epoch": 1.4587412587412587, + "grad_norm": 0.30942108650105227, + "learning_rate": 1.2217050451619284e-05, + "loss": 0.08487781137228012, + "step": 2608 + }, + { + "epoch": 1.4593006993006994, + "grad_norm": 0.2847750745844697, + "learning_rate": 1.2210703576313523e-05, + "loss": 0.057219114154577255, + "step": 2609 + }, + { + "epoch": 1.4598601398601398, + "grad_norm": 0.3937095973238392, + "learning_rate": 1.2204355764577058e-05, + "loss": 0.12177891284227371, + "step": 2610 + }, + { + "epoch": 1.4604195804195803, + "grad_norm": 0.4705151668209875, + "learning_rate": 1.2198007019098751e-05, + "loss": 0.23259848356246948, + "step": 2611 + }, + { + "epoch": 1.460979020979021, + "grad_norm": 0.5884578915263695, + "learning_rate": 1.2191657342567864e-05, + "loss": 0.3108658194541931, + "step": 2612 + }, + { + "epoch": 1.4615384615384617, + "grad_norm": 0.23435596007349385, + "learning_rate": 1.2185306737674055e-05, + "loss": 0.03010696731507778, + "step": 2613 + }, + { + "epoch": 1.4620979020979021, + "grad_norm": 0.38029593018254504, + "learning_rate": 1.217895520710737e-05, + "loss": 0.08955302834510803, + "step": 2614 + }, + { + "epoch": 1.4626573426573426, + "grad_norm": 0.6248268698971846, + "learning_rate": 1.2172602753558259e-05, + "loss": 0.18781223893165588, + "step": 2615 + }, + { + "epoch": 1.4632167832167833, + "grad_norm": 0.4026475240684038, + "learning_rate": 1.2166249379717543e-05, + "loss": 0.10175687819719315, + "step": 2616 + }, + { + "epoch": 1.4637762237762237, + "grad_norm": 0.4211227982548858, + "learning_rate": 1.215989508827645e-05, + "loss": 0.23766371607780457, + "step": 2617 + }, + { + "epoch": 1.4643356643356644, + "grad_norm": 0.4366486078781628, + "learning_rate": 1.2153539881926588e-05, + "loss": 0.1668854057788849, + "step": 2618 + }, + { + "epoch": 1.4648951048951049, + "grad_norm": 0.3839301221002697, + "learning_rate": 1.2147183763359958e-05, + "loss": 0.11260493844747543, + "step": 2619 + }, + { + "epoch": 1.4654545454545453, + "grad_norm": 0.2902345451861635, + "learning_rate": 1.2140826735268947e-05, + "loss": 0.04278929531574249, + "step": 2620 + }, + { + "epoch": 1.466013986013986, + "grad_norm": 0.5461338697686795, + "learning_rate": 1.2134468800346317e-05, + "loss": 0.2554653584957123, + "step": 2621 + }, + { + "epoch": 1.4665734265734267, + "grad_norm": 0.43726805309896644, + "learning_rate": 1.2128109961285233e-05, + "loss": 0.1261654943227768, + "step": 2622 + }, + { + "epoch": 1.4671328671328672, + "grad_norm": 0.6486842044741188, + "learning_rate": 1.2121750220779222e-05, + "loss": 0.19858568906784058, + "step": 2623 + }, + { + "epoch": 1.4676923076923076, + "grad_norm": 0.43558306698335025, + "learning_rate": 1.211538958152221e-05, + "loss": 0.1181764304637909, + "step": 2624 + }, + { + "epoch": 1.4682517482517483, + "grad_norm": 0.5084570987484326, + "learning_rate": 1.210902804620849e-05, + "loss": 0.2142762839794159, + "step": 2625 + }, + { + "epoch": 1.4688111888111888, + "grad_norm": 0.44222867481917255, + "learning_rate": 1.2102665617532748e-05, + "loss": 0.0807376429438591, + "step": 2626 + }, + { + "epoch": 1.4693706293706295, + "grad_norm": 0.3473053724391354, + "learning_rate": 1.2096302298190036e-05, + "loss": 0.05056123435497284, + "step": 2627 + }, + { + "epoch": 1.46993006993007, + "grad_norm": 0.6700280137705067, + "learning_rate": 1.2089938090875795e-05, + "loss": 0.24962349236011505, + "step": 2628 + }, + { + "epoch": 1.4704895104895104, + "grad_norm": 0.4158501492334591, + "learning_rate": 1.2083572998285832e-05, + "loss": 0.0875111073255539, + "step": 2629 + }, + { + "epoch": 1.471048951048951, + "grad_norm": 0.4788499489797384, + "learning_rate": 1.2077207023116333e-05, + "loss": 0.16640730202198029, + "step": 2630 + }, + { + "epoch": 1.4716083916083917, + "grad_norm": 0.42736400948359043, + "learning_rate": 1.2070840168063858e-05, + "loss": 0.17294815182685852, + "step": 2631 + }, + { + "epoch": 1.4721678321678322, + "grad_norm": 0.2698795689106329, + "learning_rate": 1.2064472435825339e-05, + "loss": 0.06078283116221428, + "step": 2632 + }, + { + "epoch": 1.4727272727272727, + "grad_norm": 0.5413746455318379, + "learning_rate": 1.2058103829098087e-05, + "loss": 0.329429566860199, + "step": 2633 + }, + { + "epoch": 1.4732867132867133, + "grad_norm": 0.4061401926392804, + "learning_rate": 1.2051734350579767e-05, + "loss": 0.10931099206209183, + "step": 2634 + }, + { + "epoch": 1.4738461538461538, + "grad_norm": 0.5289791995117499, + "learning_rate": 1.2045364002968425e-05, + "loss": 0.17708632349967957, + "step": 2635 + }, + { + "epoch": 1.4744055944055945, + "grad_norm": 0.5425468874977516, + "learning_rate": 1.2038992788962478e-05, + "loss": 0.2560347318649292, + "step": 2636 + }, + { + "epoch": 1.474965034965035, + "grad_norm": 0.557821274767139, + "learning_rate": 1.2032620711260702e-05, + "loss": 0.3001738488674164, + "step": 2637 + }, + { + "epoch": 1.4755244755244754, + "grad_norm": 0.5392458009268125, + "learning_rate": 1.2026247772562241e-05, + "loss": 0.28792041540145874, + "step": 2638 + }, + { + "epoch": 1.476083916083916, + "grad_norm": 0.5200822030451243, + "learning_rate": 1.2019873975566607e-05, + "loss": 0.1711592823266983, + "step": 2639 + }, + { + "epoch": 1.4766433566433568, + "grad_norm": 0.29004677382270483, + "learning_rate": 1.201349932297367e-05, + "loss": 0.0962599664926529, + "step": 2640 + }, + { + "epoch": 1.4772027972027972, + "grad_norm": 0.3109999537008209, + "learning_rate": 1.2007123817483666e-05, + "loss": 0.059583887457847595, + "step": 2641 + }, + { + "epoch": 1.4777622377622377, + "grad_norm": 0.2777106148813247, + "learning_rate": 1.2000747461797195e-05, + "loss": 0.11674322932958603, + "step": 2642 + }, + { + "epoch": 1.4783216783216784, + "grad_norm": 0.5560719994755344, + "learning_rate": 1.1994370258615211e-05, + "loss": 0.277969628572464, + "step": 2643 + }, + { + "epoch": 1.4788811188811188, + "grad_norm": 0.4065085609617176, + "learning_rate": 1.1987992210639033e-05, + "loss": 0.1078287810087204, + "step": 2644 + }, + { + "epoch": 1.4794405594405595, + "grad_norm": 0.4073175363731146, + "learning_rate": 1.1981613320570328e-05, + "loss": 0.12364526093006134, + "step": 2645 + }, + { + "epoch": 1.48, + "grad_norm": 0.46392156297263903, + "learning_rate": 1.1975233591111138e-05, + "loss": 0.13781434297561646, + "step": 2646 + }, + { + "epoch": 1.4805594405594404, + "grad_norm": 0.38220422362876066, + "learning_rate": 1.1968853024963838e-05, + "loss": 0.11081355810165405, + "step": 2647 + }, + { + "epoch": 1.4811188811188811, + "grad_norm": 0.37670703113654375, + "learning_rate": 1.1962471624831174e-05, + "loss": 0.0981362834572792, + "step": 2648 + }, + { + "epoch": 1.4816783216783218, + "grad_norm": 0.5778010002065278, + "learning_rate": 1.1956089393416237e-05, + "loss": 0.17977339029312134, + "step": 2649 + }, + { + "epoch": 1.4822377622377623, + "grad_norm": 0.16527268411404805, + "learning_rate": 1.1949706333422476e-05, + "loss": 0.032853513956069946, + "step": 2650 + }, + { + "epoch": 1.4827972027972027, + "grad_norm": 0.2011427313611988, + "learning_rate": 1.1943322447553686e-05, + "loss": 0.01158855389803648, + "step": 2651 + }, + { + "epoch": 1.4833566433566434, + "grad_norm": 0.1802975176351583, + "learning_rate": 1.1936937738514015e-05, + "loss": 0.03143032267689705, + "step": 2652 + }, + { + "epoch": 1.4839160839160839, + "grad_norm": 0.47613304307597537, + "learning_rate": 1.1930552209007957e-05, + "loss": 0.23875242471694946, + "step": 2653 + }, + { + "epoch": 1.4844755244755246, + "grad_norm": 0.35468073505435277, + "learning_rate": 1.1924165861740354e-05, + "loss": 0.08013374358415604, + "step": 2654 + }, + { + "epoch": 1.485034965034965, + "grad_norm": 0.33236071945330053, + "learning_rate": 1.1917778699416398e-05, + "loss": 0.10361069440841675, + "step": 2655 + }, + { + "epoch": 1.4855944055944055, + "grad_norm": 0.5160622458772943, + "learning_rate": 1.1911390724741625e-05, + "loss": 0.19031494855880737, + "step": 2656 + }, + { + "epoch": 1.4861538461538462, + "grad_norm": 0.5676890406979732, + "learning_rate": 1.1905001940421906e-05, + "loss": 0.21351051330566406, + "step": 2657 + }, + { + "epoch": 1.4867132867132868, + "grad_norm": 0.5415021445852635, + "learning_rate": 1.189861234916347e-05, + "loss": 0.17911601066589355, + "step": 2658 + }, + { + "epoch": 1.4872727272727273, + "grad_norm": 0.410790808451528, + "learning_rate": 1.1892221953672878e-05, + "loss": 0.0826939269900322, + "step": 2659 + }, + { + "epoch": 1.4878321678321678, + "grad_norm": 0.26888432262329637, + "learning_rate": 1.1885830756657033e-05, + "loss": 0.037227433174848557, + "step": 2660 + }, + { + "epoch": 1.4883916083916084, + "grad_norm": 0.34260633793982836, + "learning_rate": 1.1879438760823181e-05, + "loss": 0.06488566100597382, + "step": 2661 + }, + { + "epoch": 1.488951048951049, + "grad_norm": 0.1577899186871339, + "learning_rate": 1.18730459688789e-05, + "loss": 0.04365688934922218, + "step": 2662 + }, + { + "epoch": 1.4895104895104896, + "grad_norm": 0.47606987506149706, + "learning_rate": 1.1866652383532114e-05, + "loss": 0.1745668649673462, + "step": 2663 + }, + { + "epoch": 1.49006993006993, + "grad_norm": 0.36828473282447366, + "learning_rate": 1.1860258007491077e-05, + "loss": 0.08156780898571014, + "step": 2664 + }, + { + "epoch": 1.4906293706293705, + "grad_norm": 0.5323255863384939, + "learning_rate": 1.1853862843464375e-05, + "loss": 0.2273499071598053, + "step": 2665 + }, + { + "epoch": 1.4911888111888112, + "grad_norm": 0.594509460540588, + "learning_rate": 1.1847466894160937e-05, + "loss": 0.26180756092071533, + "step": 2666 + }, + { + "epoch": 1.4917482517482519, + "grad_norm": 0.42749124993802134, + "learning_rate": 1.1841070162290019e-05, + "loss": 0.1497553139925003, + "step": 2667 + }, + { + "epoch": 1.4923076923076923, + "grad_norm": 0.3750716977902668, + "learning_rate": 1.183467265056121e-05, + "loss": 0.07229838520288467, + "step": 2668 + }, + { + "epoch": 1.4928671328671328, + "grad_norm": 0.6943247416140461, + "learning_rate": 1.1828274361684422e-05, + "loss": 0.36371690034866333, + "step": 2669 + }, + { + "epoch": 1.4934265734265735, + "grad_norm": 0.4512976856043034, + "learning_rate": 1.1821875298369906e-05, + "loss": 0.1397406905889511, + "step": 2670 + }, + { + "epoch": 1.493986013986014, + "grad_norm": 0.5303430518975251, + "learning_rate": 1.1815475463328242e-05, + "loss": 0.17724046111106873, + "step": 2671 + }, + { + "epoch": 1.4945454545454546, + "grad_norm": 0.4275933834219787, + "learning_rate": 1.1809074859270333e-05, + "loss": 0.08694016933441162, + "step": 2672 + }, + { + "epoch": 1.495104895104895, + "grad_norm": 0.6944108514959085, + "learning_rate": 1.1802673488907401e-05, + "loss": 0.3452514111995697, + "step": 2673 + }, + { + "epoch": 1.4956643356643355, + "grad_norm": 0.2849610951442341, + "learning_rate": 1.1796271354951e-05, + "loss": 0.03347887471318245, + "step": 2674 + }, + { + "epoch": 1.4962237762237762, + "grad_norm": 0.40117761438810057, + "learning_rate": 1.1789868460113008e-05, + "loss": 0.07179979979991913, + "step": 2675 + }, + { + "epoch": 1.496783216783217, + "grad_norm": 0.4113183389809946, + "learning_rate": 1.1783464807105625e-05, + "loss": 0.07434782385826111, + "step": 2676 + }, + { + "epoch": 1.4973426573426574, + "grad_norm": 0.37820493797537413, + "learning_rate": 1.177706039864137e-05, + "loss": 0.17787566781044006, + "step": 2677 + }, + { + "epoch": 1.4979020979020978, + "grad_norm": 0.43905949813897255, + "learning_rate": 1.1770655237433084e-05, + "loss": 0.15541744232177734, + "step": 2678 + }, + { + "epoch": 1.4984615384615385, + "grad_norm": 0.32300236858837383, + "learning_rate": 1.1764249326193922e-05, + "loss": 0.06536943465471268, + "step": 2679 + }, + { + "epoch": 1.499020979020979, + "grad_norm": 0.6607102652041649, + "learning_rate": 1.175784266763736e-05, + "loss": 0.15037459135055542, + "step": 2680 + }, + { + "epoch": 1.4995804195804197, + "grad_norm": 0.9694449321880815, + "learning_rate": 1.17514352644772e-05, + "loss": 0.20892234146595, + "step": 2681 + }, + { + "epoch": 1.5001398601398601, + "grad_norm": 0.5192058307899828, + "learning_rate": 1.1745027119427538e-05, + "loss": 0.18815618753433228, + "step": 2682 + }, + { + "epoch": 1.5006993006993006, + "grad_norm": 0.48920343418146006, + "learning_rate": 1.1738618235202805e-05, + "loss": 0.24392755329608917, + "step": 2683 + }, + { + "epoch": 1.5012587412587413, + "grad_norm": 0.4140159732830911, + "learning_rate": 1.1732208614517736e-05, + "loss": 0.14816738665103912, + "step": 2684 + }, + { + "epoch": 1.501818181818182, + "grad_norm": 0.24813287869358588, + "learning_rate": 1.1725798260087374e-05, + "loss": 0.05145620182156563, + "step": 2685 + }, + { + "epoch": 1.5023776223776224, + "grad_norm": 0.4232909003911284, + "learning_rate": 1.1719387174627083e-05, + "loss": 0.14055439829826355, + "step": 2686 + }, + { + "epoch": 1.5029370629370629, + "grad_norm": 0.48829804701548296, + "learning_rate": 1.1712975360852525e-05, + "loss": 0.06928224116563797, + "step": 2687 + }, + { + "epoch": 1.5034965034965035, + "grad_norm": 0.5101767389276161, + "learning_rate": 1.1706562821479683e-05, + "loss": 0.18926560878753662, + "step": 2688 + }, + { + "epoch": 1.504055944055944, + "grad_norm": 0.8479732937560088, + "learning_rate": 1.1700149559224839e-05, + "loss": 0.211931049823761, + "step": 2689 + }, + { + "epoch": 1.5046153846153847, + "grad_norm": 0.40258257558859134, + "learning_rate": 1.1693735576804582e-05, + "loss": 0.08584072440862656, + "step": 2690 + }, + { + "epoch": 1.5051748251748251, + "grad_norm": 0.3972236891489551, + "learning_rate": 1.1687320876935806e-05, + "loss": 0.1607426255941391, + "step": 2691 + }, + { + "epoch": 1.5057342657342656, + "grad_norm": 0.3021274556904061, + "learning_rate": 1.1680905462335711e-05, + "loss": 0.08180231600999832, + "step": 2692 + }, + { + "epoch": 1.5062937062937063, + "grad_norm": 0.615184686964392, + "learning_rate": 1.1674489335721803e-05, + "loss": 0.25060927867889404, + "step": 2693 + }, + { + "epoch": 1.506853146853147, + "grad_norm": 0.2307437318448909, + "learning_rate": 1.1668072499811882e-05, + "loss": 0.08234649151563644, + "step": 2694 + }, + { + "epoch": 1.5074125874125874, + "grad_norm": 0.43315673006205213, + "learning_rate": 1.1661654957324053e-05, + "loss": 0.1301392912864685, + "step": 2695 + }, + { + "epoch": 1.507972027972028, + "grad_norm": 0.5202798050119333, + "learning_rate": 1.1655236710976716e-05, + "loss": 0.30901575088500977, + "step": 2696 + }, + { + "epoch": 1.5085314685314686, + "grad_norm": 0.33833790564109384, + "learning_rate": 1.1648817763488576e-05, + "loss": 0.12982720136642456, + "step": 2697 + }, + { + "epoch": 1.509090909090909, + "grad_norm": 0.5186314255922662, + "learning_rate": 1.164239811757863e-05, + "loss": 0.24956434965133667, + "step": 2698 + }, + { + "epoch": 1.5096503496503497, + "grad_norm": 0.42469341432750096, + "learning_rate": 1.1635977775966174e-05, + "loss": 0.20079776644706726, + "step": 2699 + }, + { + "epoch": 1.5102097902097902, + "grad_norm": 0.3367449575197374, + "learning_rate": 1.1629556741370797e-05, + "loss": 0.06869709491729736, + "step": 2700 + }, + { + "epoch": 1.5107692307692306, + "grad_norm": 0.61414314275935, + "learning_rate": 1.1623135016512377e-05, + "loss": 0.20549604296684265, + "step": 2701 + }, + { + "epoch": 1.5113286713286713, + "grad_norm": 0.48849740844477246, + "learning_rate": 1.1616712604111094e-05, + "loss": 0.18413308262825012, + "step": 2702 + }, + { + "epoch": 1.511888111888112, + "grad_norm": 0.5515656838094445, + "learning_rate": 1.1610289506887409e-05, + "loss": 0.1787792593240738, + "step": 2703 + }, + { + "epoch": 1.5124475524475525, + "grad_norm": 0.5479625961332765, + "learning_rate": 1.1603865727562088e-05, + "loss": 0.17872554063796997, + "step": 2704 + }, + { + "epoch": 1.513006993006993, + "grad_norm": 0.47731054933800876, + "learning_rate": 1.1597441268856163e-05, + "loss": 0.26431435346603394, + "step": 2705 + }, + { + "epoch": 1.5135664335664336, + "grad_norm": 0.42112180060426424, + "learning_rate": 1.1591016133490978e-05, + "loss": 0.10064855217933655, + "step": 2706 + }, + { + "epoch": 1.514125874125874, + "grad_norm": 0.3952446947343075, + "learning_rate": 1.1584590324188146e-05, + "loss": 0.1621956080198288, + "step": 2707 + }, + { + "epoch": 1.5146853146853148, + "grad_norm": 0.41896709334607446, + "learning_rate": 1.1578163843669575e-05, + "loss": 0.16317087411880493, + "step": 2708 + }, + { + "epoch": 1.5152447552447552, + "grad_norm": 0.4790135535077637, + "learning_rate": 1.157173669465745e-05, + "loss": 0.18540197610855103, + "step": 2709 + }, + { + "epoch": 1.5158041958041957, + "grad_norm": 0.4412846949827971, + "learning_rate": 1.1565308879874249e-05, + "loss": 0.08340314030647278, + "step": 2710 + }, + { + "epoch": 1.5163636363636364, + "grad_norm": 0.485048351834888, + "learning_rate": 1.1558880402042725e-05, + "loss": 0.1868424415588379, + "step": 2711 + }, + { + "epoch": 1.516923076923077, + "grad_norm": 0.22481176312181306, + "learning_rate": 1.155245126388591e-05, + "loss": 0.0291266031563282, + "step": 2712 + }, + { + "epoch": 1.5174825174825175, + "grad_norm": 0.39567801529969376, + "learning_rate": 1.1546021468127127e-05, + "loss": 0.1892201006412506, + "step": 2713 + }, + { + "epoch": 1.518041958041958, + "grad_norm": 0.3217565389135706, + "learning_rate": 1.1539591017489962e-05, + "loss": 0.07278981059789658, + "step": 2714 + }, + { + "epoch": 1.5186013986013986, + "grad_norm": 0.5406868326217849, + "learning_rate": 1.1533159914698287e-05, + "loss": 0.13239984214305878, + "step": 2715 + }, + { + "epoch": 1.5191608391608393, + "grad_norm": 0.4177674267944996, + "learning_rate": 1.152672816247625e-05, + "loss": 0.14451396465301514, + "step": 2716 + }, + { + "epoch": 1.5197202797202798, + "grad_norm": 0.5112706164472813, + "learning_rate": 1.1520295763548279e-05, + "loss": 0.12178383767604828, + "step": 2717 + }, + { + "epoch": 1.5202797202797202, + "grad_norm": 0.4495668032171033, + "learning_rate": 1.1513862720639061e-05, + "loss": 0.1048208475112915, + "step": 2718 + }, + { + "epoch": 1.5208391608391607, + "grad_norm": 0.22380467025091064, + "learning_rate": 1.1507429036473573e-05, + "loss": 0.05339743196964264, + "step": 2719 + }, + { + "epoch": 1.5213986013986014, + "grad_norm": 0.4773695217389601, + "learning_rate": 1.1500994713777047e-05, + "loss": 0.14586740732192993, + "step": 2720 + }, + { + "epoch": 1.521958041958042, + "grad_norm": 0.4321702664442346, + "learning_rate": 1.1494559755275003e-05, + "loss": 0.10054674744606018, + "step": 2721 + }, + { + "epoch": 1.5225174825174825, + "grad_norm": 0.35974129970916235, + "learning_rate": 1.1488124163693217e-05, + "loss": 0.05900286138057709, + "step": 2722 + }, + { + "epoch": 1.523076923076923, + "grad_norm": 0.38910250020584747, + "learning_rate": 1.1481687941757743e-05, + "loss": 0.16780465841293335, + "step": 2723 + }, + { + "epoch": 1.5236363636363637, + "grad_norm": 0.45211612280828917, + "learning_rate": 1.1475251092194885e-05, + "loss": 0.15541806817054749, + "step": 2724 + }, + { + "epoch": 1.5241958041958044, + "grad_norm": 0.5240975858695104, + "learning_rate": 1.1468813617731237e-05, + "loss": 0.2158806324005127, + "step": 2725 + }, + { + "epoch": 1.5247552447552448, + "grad_norm": 0.3711594209903294, + "learning_rate": 1.1462375521093638e-05, + "loss": 0.11658114194869995, + "step": 2726 + }, + { + "epoch": 1.5253146853146853, + "grad_norm": 0.26469300885088015, + "learning_rate": 1.14559368050092e-05, + "loss": 0.0538557805120945, + "step": 2727 + }, + { + "epoch": 1.5258741258741257, + "grad_norm": 0.41123073836209784, + "learning_rate": 1.1449497472205296e-05, + "loss": 0.1075274795293808, + "step": 2728 + }, + { + "epoch": 1.5264335664335664, + "grad_norm": 0.46487759864921374, + "learning_rate": 1.1443057525409558e-05, + "loss": 0.10640497505664825, + "step": 2729 + }, + { + "epoch": 1.526993006993007, + "grad_norm": 0.2576281172195759, + "learning_rate": 1.143661696734988e-05, + "loss": 0.08769455552101135, + "step": 2730 + }, + { + "epoch": 1.5275524475524476, + "grad_norm": 0.5889421848675456, + "learning_rate": 1.1430175800754416e-05, + "loss": 0.22475150227546692, + "step": 2731 + }, + { + "epoch": 1.528111888111888, + "grad_norm": 0.46822800804661413, + "learning_rate": 1.1423734028351572e-05, + "loss": 0.14381659030914307, + "step": 2732 + }, + { + "epoch": 1.5286713286713287, + "grad_norm": 0.5141631230433552, + "learning_rate": 1.1417291652870018e-05, + "loss": 0.17542880773544312, + "step": 2733 + }, + { + "epoch": 1.5292307692307694, + "grad_norm": 0.3322610612918832, + "learning_rate": 1.1410848677038677e-05, + "loss": 0.07394293695688248, + "step": 2734 + }, + { + "epoch": 1.5297902097902099, + "grad_norm": 0.41630090457904695, + "learning_rate": 1.1404405103586726e-05, + "loss": 0.15971913933753967, + "step": 2735 + }, + { + "epoch": 1.5303496503496503, + "grad_norm": 0.4291131493833879, + "learning_rate": 1.139796093524359e-05, + "loss": 0.23908200860023499, + "step": 2736 + }, + { + "epoch": 1.5309090909090908, + "grad_norm": 0.38543030318833066, + "learning_rate": 1.1391516174738953e-05, + "loss": 0.1496146023273468, + "step": 2737 + }, + { + "epoch": 1.5314685314685315, + "grad_norm": 0.5925070497996959, + "learning_rate": 1.1385070824802749e-05, + "loss": 0.23367974162101746, + "step": 2738 + }, + { + "epoch": 1.5320279720279721, + "grad_norm": 0.48305110411903435, + "learning_rate": 1.1378624888165162e-05, + "loss": 0.2965750992298126, + "step": 2739 + }, + { + "epoch": 1.5325874125874126, + "grad_norm": 0.44616672638091454, + "learning_rate": 1.1372178367556621e-05, + "loss": 0.10041292011737823, + "step": 2740 + }, + { + "epoch": 1.533146853146853, + "grad_norm": 0.4722135271755495, + "learning_rate": 1.13657312657078e-05, + "loss": 0.24014243483543396, + "step": 2741 + }, + { + "epoch": 1.5337062937062937, + "grad_norm": 0.5527077570558265, + "learning_rate": 1.1359283585349628e-05, + "loss": 0.18213319778442383, + "step": 2742 + }, + { + "epoch": 1.5342657342657344, + "grad_norm": 0.5861159297448009, + "learning_rate": 1.1352835329213272e-05, + "loss": 0.28087472915649414, + "step": 2743 + }, + { + "epoch": 1.5348251748251749, + "grad_norm": 0.48493319154780495, + "learning_rate": 1.1346386500030146e-05, + "loss": 0.15751951932907104, + "step": 2744 + }, + { + "epoch": 1.5353846153846153, + "grad_norm": 0.45816480689568345, + "learning_rate": 1.133993710053191e-05, + "loss": 0.10551661252975464, + "step": 2745 + }, + { + "epoch": 1.5359440559440558, + "grad_norm": 0.3223062055422646, + "learning_rate": 1.1333487133450454e-05, + "loss": 0.07124066352844238, + "step": 2746 + }, + { + "epoch": 1.5365034965034965, + "grad_norm": 0.28714408531558344, + "learning_rate": 1.132703660151792e-05, + "loss": 0.05576277896761894, + "step": 2747 + }, + { + "epoch": 1.5370629370629372, + "grad_norm": 0.27400248299510643, + "learning_rate": 1.1320585507466689e-05, + "loss": 0.1267869472503662, + "step": 2748 + }, + { + "epoch": 1.5376223776223776, + "grad_norm": 0.56595927416101, + "learning_rate": 1.1314133854029367e-05, + "loss": 0.2700503170490265, + "step": 2749 + }, + { + "epoch": 1.538181818181818, + "grad_norm": 0.26245698066605644, + "learning_rate": 1.1307681643938816e-05, + "loss": 0.047837212681770325, + "step": 2750 + }, + { + "epoch": 1.5387412587412588, + "grad_norm": 0.5467478868896998, + "learning_rate": 1.1301228879928118e-05, + "loss": 0.2361563742160797, + "step": 2751 + }, + { + "epoch": 1.5393006993006995, + "grad_norm": 0.40738401680393, + "learning_rate": 1.1294775564730599e-05, + "loss": 0.11354345828294754, + "step": 2752 + }, + { + "epoch": 1.53986013986014, + "grad_norm": 0.4837699207676355, + "learning_rate": 1.1288321701079815e-05, + "loss": 0.10136795043945312, + "step": 2753 + }, + { + "epoch": 1.5404195804195804, + "grad_norm": 0.5371803021151678, + "learning_rate": 1.1281867291709549e-05, + "loss": 0.16819968819618225, + "step": 2754 + }, + { + "epoch": 1.5409790209790208, + "grad_norm": 0.44119261126008924, + "learning_rate": 1.1275412339353826e-05, + "loss": 0.12270970642566681, + "step": 2755 + }, + { + "epoch": 1.5415384615384615, + "grad_norm": 0.4152711918904521, + "learning_rate": 1.1268956846746897e-05, + "loss": 0.2946152687072754, + "step": 2756 + }, + { + "epoch": 1.5420979020979022, + "grad_norm": 0.5016765070619174, + "learning_rate": 1.1262500816623239e-05, + "loss": 0.06950849294662476, + "step": 2757 + }, + { + "epoch": 1.5426573426573427, + "grad_norm": 0.4415394767789302, + "learning_rate": 1.1256044251717555e-05, + "loss": 0.1372474581003189, + "step": 2758 + }, + { + "epoch": 1.5432167832167831, + "grad_norm": 0.3374073525697131, + "learning_rate": 1.124958715476478e-05, + "loss": 0.05484132841229439, + "step": 2759 + }, + { + "epoch": 1.5437762237762238, + "grad_norm": 0.3592342984950759, + "learning_rate": 1.1243129528500073e-05, + "loss": 0.09418675303459167, + "step": 2760 + }, + { + "epoch": 1.5443356643356645, + "grad_norm": 0.4650746470431882, + "learning_rate": 1.123667137565882e-05, + "loss": 0.15810918807983398, + "step": 2761 + }, + { + "epoch": 1.544895104895105, + "grad_norm": 0.4716393916380295, + "learning_rate": 1.1230212698976619e-05, + "loss": 0.12960809469223022, + "step": 2762 + }, + { + "epoch": 1.5454545454545454, + "grad_norm": 0.48816298983104034, + "learning_rate": 1.12237535011893e-05, + "loss": 0.17287057638168335, + "step": 2763 + }, + { + "epoch": 1.5460139860139859, + "grad_norm": 0.4617474422824959, + "learning_rate": 1.1217293785032911e-05, + "loss": 0.11621387302875519, + "step": 2764 + }, + { + "epoch": 1.5465734265734266, + "grad_norm": 0.17404089142424573, + "learning_rate": 1.1210833553243721e-05, + "loss": 0.07183268666267395, + "step": 2765 + }, + { + "epoch": 1.5471328671328672, + "grad_norm": 0.5959121865880257, + "learning_rate": 1.1204372808558218e-05, + "loss": 0.2578980624675751, + "step": 2766 + }, + { + "epoch": 1.5476923076923077, + "grad_norm": 0.7651616342579115, + "learning_rate": 1.1197911553713098e-05, + "loss": 0.18708135187625885, + "step": 2767 + }, + { + "epoch": 1.5482517482517482, + "grad_norm": 0.41211094724360237, + "learning_rate": 1.1191449791445285e-05, + "loss": 0.15511882305145264, + "step": 2768 + }, + { + "epoch": 1.5488111888111888, + "grad_norm": 0.40417129447049893, + "learning_rate": 1.1184987524491915e-05, + "loss": 0.17547817528247833, + "step": 2769 + }, + { + "epoch": 1.5493706293706295, + "grad_norm": 0.4843535986015377, + "learning_rate": 1.1178524755590334e-05, + "loss": 0.1429714411497116, + "step": 2770 + }, + { + "epoch": 1.54993006993007, + "grad_norm": 0.31807438001098637, + "learning_rate": 1.1172061487478103e-05, + "loss": 0.0978471115231514, + "step": 2771 + }, + { + "epoch": 1.5504895104895104, + "grad_norm": 0.4605385453734315, + "learning_rate": 1.1165597722892992e-05, + "loss": 0.1880197525024414, + "step": 2772 + }, + { + "epoch": 1.551048951048951, + "grad_norm": 0.48833686348065997, + "learning_rate": 1.1159133464572989e-05, + "loss": 0.1662192940711975, + "step": 2773 + }, + { + "epoch": 1.5516083916083916, + "grad_norm": 0.5527414526333692, + "learning_rate": 1.115266871525628e-05, + "loss": 0.14673921465873718, + "step": 2774 + }, + { + "epoch": 1.5521678321678323, + "grad_norm": 0.55845026419233, + "learning_rate": 1.1146203477681268e-05, + "loss": 0.33482736349105835, + "step": 2775 + }, + { + "epoch": 1.5527272727272727, + "grad_norm": 0.5209361648111882, + "learning_rate": 1.1139737754586554e-05, + "loss": 0.20802396535873413, + "step": 2776 + }, + { + "epoch": 1.5532867132867132, + "grad_norm": 0.49936592810751357, + "learning_rate": 1.1133271548710958e-05, + "loss": 0.2128729522228241, + "step": 2777 + }, + { + "epoch": 1.5538461538461539, + "grad_norm": 0.8477325413503357, + "learning_rate": 1.112680486279349e-05, + "loss": 0.2423178255558014, + "step": 2778 + }, + { + "epoch": 1.5544055944055946, + "grad_norm": 0.5594639787325337, + "learning_rate": 1.1120337699573378e-05, + "loss": 0.09590500593185425, + "step": 2779 + }, + { + "epoch": 1.554965034965035, + "grad_norm": 0.4858518252052592, + "learning_rate": 1.1113870061790035e-05, + "loss": 0.2182995080947876, + "step": 2780 + }, + { + "epoch": 1.5555244755244755, + "grad_norm": 0.4900758339557702, + "learning_rate": 1.1107401952183088e-05, + "loss": 0.09691740572452545, + "step": 2781 + }, + { + "epoch": 1.556083916083916, + "grad_norm": 0.5230161958351651, + "learning_rate": 1.110093337349236e-05, + "loss": 0.24449270963668823, + "step": 2782 + }, + { + "epoch": 1.5566433566433566, + "grad_norm": 0.433484477870713, + "learning_rate": 1.1094464328457873e-05, + "loss": 0.14741364121437073, + "step": 2783 + }, + { + "epoch": 1.5572027972027973, + "grad_norm": 0.51293148894446, + "learning_rate": 1.1087994819819847e-05, + "loss": 0.17428237199783325, + "step": 2784 + }, + { + "epoch": 1.5577622377622378, + "grad_norm": 0.6144459481468858, + "learning_rate": 1.1081524850318694e-05, + "loss": 0.19588162004947662, + "step": 2785 + }, + { + "epoch": 1.5583216783216782, + "grad_norm": 0.4384355326706255, + "learning_rate": 1.1075054422695027e-05, + "loss": 0.09360744059085846, + "step": 2786 + }, + { + "epoch": 1.558881118881119, + "grad_norm": 0.5458931650099837, + "learning_rate": 1.1068583539689652e-05, + "loss": 0.258054256439209, + "step": 2787 + }, + { + "epoch": 1.5594405594405596, + "grad_norm": 0.2827764274732009, + "learning_rate": 1.1062112204043567e-05, + "loss": 0.05555944889783859, + "step": 2788 + }, + { + "epoch": 1.56, + "grad_norm": 0.44105568521717287, + "learning_rate": 1.1055640418497957e-05, + "loss": 0.10265368223190308, + "step": 2789 + }, + { + "epoch": 1.5605594405594405, + "grad_norm": 0.47492060802525116, + "learning_rate": 1.104916818579421e-05, + "loss": 0.1817026436328888, + "step": 2790 + }, + { + "epoch": 1.561118881118881, + "grad_norm": 1.006106277487736, + "learning_rate": 1.1042695508673886e-05, + "loss": 0.22906853258609772, + "step": 2791 + }, + { + "epoch": 1.5616783216783217, + "grad_norm": 0.43556228605106145, + "learning_rate": 1.1036222389878752e-05, + "loss": 0.10952460765838623, + "step": 2792 + }, + { + "epoch": 1.5622377622377623, + "grad_norm": 0.47325282576576555, + "learning_rate": 1.1029748832150746e-05, + "loss": 0.17642460763454437, + "step": 2793 + }, + { + "epoch": 1.5627972027972028, + "grad_norm": 0.4470815550536847, + "learning_rate": 1.1023274838232e-05, + "loss": 0.13231033086776733, + "step": 2794 + }, + { + "epoch": 1.5633566433566433, + "grad_norm": 0.3690608567575378, + "learning_rate": 1.1016800410864831e-05, + "loss": 0.13153162598609924, + "step": 2795 + }, + { + "epoch": 1.563916083916084, + "grad_norm": 0.1746415973066258, + "learning_rate": 1.1010325552791738e-05, + "loss": 0.014128871262073517, + "step": 2796 + }, + { + "epoch": 1.5644755244755246, + "grad_norm": 0.4257739766716616, + "learning_rate": 1.1003850266755405e-05, + "loss": 0.12916907668113708, + "step": 2797 + }, + { + "epoch": 1.565034965034965, + "grad_norm": 0.40021558307893873, + "learning_rate": 1.099737455549869e-05, + "loss": 0.11345451325178146, + "step": 2798 + }, + { + "epoch": 1.5655944055944055, + "grad_norm": 0.17079439793253023, + "learning_rate": 1.0990898421764638e-05, + "loss": 0.053648822009563446, + "step": 2799 + }, + { + "epoch": 1.566153846153846, + "grad_norm": 0.5032468996304865, + "learning_rate": 1.098442186829647e-05, + "loss": 0.10897266864776611, + "step": 2800 + }, + { + "epoch": 1.5667132867132867, + "grad_norm": 0.9709814709027257, + "learning_rate": 1.0977944897837586e-05, + "loss": 0.07249344885349274, + "step": 2801 + }, + { + "epoch": 1.5672727272727274, + "grad_norm": 0.22787616700781824, + "learning_rate": 1.0971467513131568e-05, + "loss": 0.04046345129609108, + "step": 2802 + }, + { + "epoch": 1.5678321678321678, + "grad_norm": 0.43238909339649567, + "learning_rate": 1.0964989716922158e-05, + "loss": 0.08685645461082458, + "step": 2803 + }, + { + "epoch": 1.5683916083916083, + "grad_norm": 0.5187135504209994, + "learning_rate": 1.0958511511953287e-05, + "loss": 0.14584368467330933, + "step": 2804 + }, + { + "epoch": 1.568951048951049, + "grad_norm": 0.40919396096729327, + "learning_rate": 1.0952032900969056e-05, + "loss": 0.2007332146167755, + "step": 2805 + }, + { + "epoch": 1.5695104895104897, + "grad_norm": 0.583768803330884, + "learning_rate": 1.0945553886713734e-05, + "loss": 0.31702375411987305, + "step": 2806 + }, + { + "epoch": 1.5700699300699301, + "grad_norm": 0.5095161918691351, + "learning_rate": 1.0939074471931766e-05, + "loss": 0.11963526904582977, + "step": 2807 + }, + { + "epoch": 1.5706293706293706, + "grad_norm": 0.36304492217188905, + "learning_rate": 1.0932594659367758e-05, + "loss": 0.0659194216132164, + "step": 2808 + }, + { + "epoch": 1.571188811188811, + "grad_norm": 0.6256324659703801, + "learning_rate": 1.0926114451766491e-05, + "loss": 0.32758593559265137, + "step": 2809 + }, + { + "epoch": 1.5717482517482517, + "grad_norm": 0.8591284331396012, + "learning_rate": 1.0919633851872922e-05, + "loss": 0.31249523162841797, + "step": 2810 + }, + { + "epoch": 1.5723076923076924, + "grad_norm": 0.3592503253528975, + "learning_rate": 1.0913152862432153e-05, + "loss": 0.13023559749126434, + "step": 2811 + }, + { + "epoch": 1.5728671328671329, + "grad_norm": 0.5692675169673967, + "learning_rate": 1.0906671486189469e-05, + "loss": 0.1404387652873993, + "step": 2812 + }, + { + "epoch": 1.5734265734265733, + "grad_norm": 0.46055525002958025, + "learning_rate": 1.090018972589031e-05, + "loss": 0.09957358986139297, + "step": 2813 + }, + { + "epoch": 1.573986013986014, + "grad_norm": 1.3990076524151103, + "learning_rate": 1.0893707584280282e-05, + "loss": 0.2776985168457031, + "step": 2814 + }, + { + "epoch": 1.5745454545454547, + "grad_norm": 0.5993788184156456, + "learning_rate": 1.0887225064105154e-05, + "loss": 0.0713958889245987, + "step": 2815 + }, + { + "epoch": 1.5751048951048952, + "grad_norm": 0.41983044550639737, + "learning_rate": 1.088074216811085e-05, + "loss": 0.188124418258667, + "step": 2816 + }, + { + "epoch": 1.5756643356643356, + "grad_norm": 0.6579626586926081, + "learning_rate": 1.0874258899043456e-05, + "loss": 0.18631935119628906, + "step": 2817 + }, + { + "epoch": 1.576223776223776, + "grad_norm": 0.46130013955007837, + "learning_rate": 1.086777525964922e-05, + "loss": 0.19625741243362427, + "step": 2818 + }, + { + "epoch": 1.5767832167832168, + "grad_norm": 0.36486071916629126, + "learning_rate": 1.0861291252674544e-05, + "loss": 0.11994132399559021, + "step": 2819 + }, + { + "epoch": 1.5773426573426574, + "grad_norm": 0.46007645681972753, + "learning_rate": 1.0854806880865977e-05, + "loss": 0.26735368371009827, + "step": 2820 + }, + { + "epoch": 1.577902097902098, + "grad_norm": 1.0835258908530376, + "learning_rate": 1.0848322146970238e-05, + "loss": 0.37379568815231323, + "step": 2821 + }, + { + "epoch": 1.5784615384615384, + "grad_norm": 0.5239260198080424, + "learning_rate": 1.0841837053734188e-05, + "loss": 0.2209223508834839, + "step": 2822 + }, + { + "epoch": 1.579020979020979, + "grad_norm": 0.29489324713343895, + "learning_rate": 1.0835351603904848e-05, + "loss": 0.05732685327529907, + "step": 2823 + }, + { + "epoch": 1.5795804195804197, + "grad_norm": 0.5282238408168417, + "learning_rate": 1.0828865800229383e-05, + "loss": 0.21869157254695892, + "step": 2824 + }, + { + "epoch": 1.5801398601398602, + "grad_norm": 0.384836033534584, + "learning_rate": 1.082237964545511e-05, + "loss": 0.081295907497406, + "step": 2825 + }, + { + "epoch": 1.5806993006993006, + "grad_norm": 0.42372713587488936, + "learning_rate": 1.0815893142329496e-05, + "loss": 0.13505405187606812, + "step": 2826 + }, + { + "epoch": 1.581258741258741, + "grad_norm": 0.25795482459224156, + "learning_rate": 1.0809406293600158e-05, + "loss": 0.016825826838612556, + "step": 2827 + }, + { + "epoch": 1.5818181818181818, + "grad_norm": 0.40247571847297714, + "learning_rate": 1.0802919102014857e-05, + "loss": 0.08634943515062332, + "step": 2828 + }, + { + "epoch": 1.5823776223776225, + "grad_norm": 0.20117367247494375, + "learning_rate": 1.07964315703215e-05, + "loss": 0.023347143083810806, + "step": 2829 + }, + { + "epoch": 1.582937062937063, + "grad_norm": 0.4174932932089321, + "learning_rate": 1.0789943701268129e-05, + "loss": 0.11973976343870163, + "step": 2830 + }, + { + "epoch": 1.5834965034965034, + "grad_norm": 0.31692841461389737, + "learning_rate": 1.0783455497602942e-05, + "loss": 0.08013211190700531, + "step": 2831 + }, + { + "epoch": 1.584055944055944, + "grad_norm": 0.3940953858813942, + "learning_rate": 1.0776966962074276e-05, + "loss": 0.10090242326259613, + "step": 2832 + }, + { + "epoch": 1.5846153846153848, + "grad_norm": 0.3552407336963213, + "learning_rate": 1.0770478097430601e-05, + "loss": 0.09611985832452774, + "step": 2833 + }, + { + "epoch": 1.5851748251748252, + "grad_norm": 0.35756004304617206, + "learning_rate": 1.0763988906420536e-05, + "loss": 0.04323085770010948, + "step": 2834 + }, + { + "epoch": 1.5857342657342657, + "grad_norm": 0.39096900151491265, + "learning_rate": 1.0757499391792832e-05, + "loss": 0.07391176372766495, + "step": 2835 + }, + { + "epoch": 1.5862937062937061, + "grad_norm": 0.4400945534326678, + "learning_rate": 1.075100955629638e-05, + "loss": 0.2970464527606964, + "step": 2836 + }, + { + "epoch": 1.5868531468531468, + "grad_norm": 0.9473747083266312, + "learning_rate": 1.0744519402680204e-05, + "loss": 0.3290688991546631, + "step": 2837 + }, + { + "epoch": 1.5874125874125875, + "grad_norm": 0.6162444883555247, + "learning_rate": 1.0738028933693465e-05, + "loss": 0.13434967398643494, + "step": 2838 + }, + { + "epoch": 1.587972027972028, + "grad_norm": 0.48859868239670506, + "learning_rate": 1.073153815208546e-05, + "loss": 0.23966121673583984, + "step": 2839 + }, + { + "epoch": 1.5885314685314684, + "grad_norm": 0.40151450692079954, + "learning_rate": 1.072504706060561e-05, + "loss": 0.34113994240760803, + "step": 2840 + }, + { + "epoch": 1.589090909090909, + "grad_norm": 0.34780133908525723, + "learning_rate": 1.071855566200348e-05, + "loss": 0.22892752289772034, + "step": 2841 + }, + { + "epoch": 1.5896503496503498, + "grad_norm": 0.6070404468399897, + "learning_rate": 1.071206395902875e-05, + "loss": 0.21259471774101257, + "step": 2842 + }, + { + "epoch": 1.5902097902097903, + "grad_norm": 1.0051270321468007, + "learning_rate": 1.0705571954431245e-05, + "loss": 0.1700703352689743, + "step": 2843 + }, + { + "epoch": 1.5907692307692307, + "grad_norm": 0.44121591777494157, + "learning_rate": 1.0699079650960902e-05, + "loss": 0.12835678458213806, + "step": 2844 + }, + { + "epoch": 1.5913286713286712, + "grad_norm": 0.4270961176335165, + "learning_rate": 1.0692587051367799e-05, + "loss": 0.136978417634964, + "step": 2845 + }, + { + "epoch": 1.5918881118881119, + "grad_norm": 0.4342035707089759, + "learning_rate": 1.068609415840213e-05, + "loss": 0.14044445753097534, + "step": 2846 + }, + { + "epoch": 1.5924475524475525, + "grad_norm": 0.15368121805568524, + "learning_rate": 1.0679600974814212e-05, + "loss": 0.013831639662384987, + "step": 2847 + }, + { + "epoch": 1.593006993006993, + "grad_norm": 0.3515057315771742, + "learning_rate": 1.0673107503354494e-05, + "loss": 0.1017160415649414, + "step": 2848 + }, + { + "epoch": 1.5935664335664335, + "grad_norm": 0.4697355387278961, + "learning_rate": 1.066661374677354e-05, + "loss": 0.20781946182250977, + "step": 2849 + }, + { + "epoch": 1.5941258741258741, + "grad_norm": 0.09121126603772202, + "learning_rate": 1.0660119707822039e-05, + "loss": 0.006650528870522976, + "step": 2850 + }, + { + "epoch": 1.5946853146853148, + "grad_norm": 0.426921023723342, + "learning_rate": 1.0653625389250795e-05, + "loss": 0.17892834544181824, + "step": 2851 + }, + { + "epoch": 1.5952447552447553, + "grad_norm": 0.24622344554047576, + "learning_rate": 1.0647130793810732e-05, + "loss": 0.030810000374913216, + "step": 2852 + }, + { + "epoch": 1.5958041958041957, + "grad_norm": 0.37067480057630914, + "learning_rate": 1.0640635924252894e-05, + "loss": 0.14324603974819183, + "step": 2853 + }, + { + "epoch": 1.5963636363636362, + "grad_norm": 0.47921734035362223, + "learning_rate": 1.0634140783328441e-05, + "loss": 0.12809765338897705, + "step": 2854 + }, + { + "epoch": 1.596923076923077, + "grad_norm": 0.44852073454930746, + "learning_rate": 1.0627645373788643e-05, + "loss": 0.27224209904670715, + "step": 2855 + }, + { + "epoch": 1.5974825174825176, + "grad_norm": 0.4057868961668423, + "learning_rate": 1.0621149698384885e-05, + "loss": 0.08574418723583221, + "step": 2856 + }, + { + "epoch": 1.598041958041958, + "grad_norm": 0.39181636288752564, + "learning_rate": 1.0614653759868674e-05, + "loss": 0.11137000471353531, + "step": 2857 + }, + { + "epoch": 1.5986013986013985, + "grad_norm": 0.514934319649596, + "learning_rate": 1.0608157560991612e-05, + "loss": 0.14484480023384094, + "step": 2858 + }, + { + "epoch": 1.5991608391608392, + "grad_norm": 0.4967202127822529, + "learning_rate": 1.0601661104505428e-05, + "loss": 0.14593040943145752, + "step": 2859 + }, + { + "epoch": 1.5997202797202799, + "grad_norm": 0.5879543772004645, + "learning_rate": 1.0595164393161948e-05, + "loss": 0.28754502534866333, + "step": 2860 + }, + { + "epoch": 1.6002797202797203, + "grad_norm": 0.9553917648588601, + "learning_rate": 1.0588667429713111e-05, + "loss": 0.47070586681365967, + "step": 2861 + }, + { + "epoch": 1.6008391608391608, + "grad_norm": 0.5030163945675269, + "learning_rate": 1.0582170216910962e-05, + "loss": 0.2060789167881012, + "step": 2862 + }, + { + "epoch": 1.6013986013986012, + "grad_norm": 0.42419967965460564, + "learning_rate": 1.0575672757507653e-05, + "loss": 0.12663966417312622, + "step": 2863 + }, + { + "epoch": 1.601958041958042, + "grad_norm": 0.41276436875376094, + "learning_rate": 1.0569175054255442e-05, + "loss": 0.0831083282828331, + "step": 2864 + }, + { + "epoch": 1.6025174825174826, + "grad_norm": 0.47863215300979656, + "learning_rate": 1.056267710990668e-05, + "loss": 0.16874247789382935, + "step": 2865 + }, + { + "epoch": 1.603076923076923, + "grad_norm": 0.5284703348923648, + "learning_rate": 1.0556178927213834e-05, + "loss": 0.1881970465183258, + "step": 2866 + }, + { + "epoch": 1.6036363636363635, + "grad_norm": 0.4739906034802131, + "learning_rate": 1.0549680508929464e-05, + "loss": 0.14702516794204712, + "step": 2867 + }, + { + "epoch": 1.6041958041958042, + "grad_norm": 0.6835574907427981, + "learning_rate": 1.0543181857806231e-05, + "loss": 0.2145739644765854, + "step": 2868 + }, + { + "epoch": 1.604755244755245, + "grad_norm": 0.5114216891963035, + "learning_rate": 1.0536682976596892e-05, + "loss": 0.14220091700553894, + "step": 2869 + }, + { + "epoch": 1.6053146853146854, + "grad_norm": 0.48959397697827334, + "learning_rate": 1.0530183868054309e-05, + "loss": 0.14121350646018982, + "step": 2870 + }, + { + "epoch": 1.6058741258741258, + "grad_norm": 0.48853613862619916, + "learning_rate": 1.0523684534931434e-05, + "loss": 0.21210911870002747, + "step": 2871 + }, + { + "epoch": 1.6064335664335663, + "grad_norm": 0.2602902687484404, + "learning_rate": 1.0517184979981317e-05, + "loss": 0.04120757058262825, + "step": 2872 + }, + { + "epoch": 1.606993006993007, + "grad_norm": 0.584967238734907, + "learning_rate": 1.0510685205957098e-05, + "loss": 0.21981248259544373, + "step": 2873 + }, + { + "epoch": 1.6075524475524476, + "grad_norm": 0.2932060806718238, + "learning_rate": 1.0504185215612017e-05, + "loss": 0.06539766490459442, + "step": 2874 + }, + { + "epoch": 1.608111888111888, + "grad_norm": 2.197190259009701, + "learning_rate": 1.0497685011699399e-05, + "loss": 0.40811076760292053, + "step": 2875 + }, + { + "epoch": 1.6086713286713286, + "grad_norm": 0.42601332283383725, + "learning_rate": 1.0491184596972658e-05, + "loss": 0.14503905177116394, + "step": 2876 + }, + { + "epoch": 1.6092307692307692, + "grad_norm": 0.6179953878165426, + "learning_rate": 1.0484683974185312e-05, + "loss": 0.23895788192749023, + "step": 2877 + }, + { + "epoch": 1.60979020979021, + "grad_norm": 0.34935247298405137, + "learning_rate": 1.0478183146090943e-05, + "loss": 0.06059478223323822, + "step": 2878 + }, + { + "epoch": 1.6103496503496504, + "grad_norm": 0.3662540295772548, + "learning_rate": 1.0471682115443244e-05, + "loss": 0.08490406721830368, + "step": 2879 + }, + { + "epoch": 1.6109090909090908, + "grad_norm": 0.59874063881367, + "learning_rate": 1.0465180884995976e-05, + "loss": 0.33345717191696167, + "step": 2880 + }, + { + "epoch": 1.6114685314685313, + "grad_norm": 0.37496116321989637, + "learning_rate": 1.0458679457502994e-05, + "loss": 0.1357303261756897, + "step": 2881 + }, + { + "epoch": 1.612027972027972, + "grad_norm": 0.48119716554035635, + "learning_rate": 1.0452177835718234e-05, + "loss": 0.15024752914905548, + "step": 2882 + }, + { + "epoch": 1.6125874125874127, + "grad_norm": 0.5807071247951989, + "learning_rate": 1.0445676022395712e-05, + "loss": 0.12237966805696487, + "step": 2883 + }, + { + "epoch": 1.6131468531468531, + "grad_norm": 0.3862166615995302, + "learning_rate": 1.0439174020289531e-05, + "loss": 0.0868656188249588, + "step": 2884 + }, + { + "epoch": 1.6137062937062936, + "grad_norm": 0.46444736129713654, + "learning_rate": 1.0432671832153873e-05, + "loss": 0.09831293672323227, + "step": 2885 + }, + { + "epoch": 1.6142657342657343, + "grad_norm": 0.32030570213635834, + "learning_rate": 1.0426169460742993e-05, + "loss": 0.08679448068141937, + "step": 2886 + }, + { + "epoch": 1.614825174825175, + "grad_norm": 0.36086994764916697, + "learning_rate": 1.041966690881122e-05, + "loss": 0.09096908569335938, + "step": 2887 + }, + { + "epoch": 1.6153846153846154, + "grad_norm": 0.4852299171913992, + "learning_rate": 1.0413164179112978e-05, + "loss": 0.21412667632102966, + "step": 2888 + }, + { + "epoch": 1.6159440559440559, + "grad_norm": 0.5027132754034461, + "learning_rate": 1.0406661274402747e-05, + "loss": 0.1825488805770874, + "step": 2889 + }, + { + "epoch": 1.6165034965034963, + "grad_norm": 0.21164833856131243, + "learning_rate": 1.0400158197435095e-05, + "loss": 0.02179224044084549, + "step": 2890 + }, + { + "epoch": 1.617062937062937, + "grad_norm": 0.5384834690943247, + "learning_rate": 1.0393654950964651e-05, + "loss": 0.16015592217445374, + "step": 2891 + }, + { + "epoch": 1.6176223776223777, + "grad_norm": 0.345986842108133, + "learning_rate": 1.0387151537746123e-05, + "loss": 0.08906261622905731, + "step": 2892 + }, + { + "epoch": 1.6181818181818182, + "grad_norm": 0.570278392308668, + "learning_rate": 1.0380647960534289e-05, + "loss": 0.2815285921096802, + "step": 2893 + }, + { + "epoch": 1.6187412587412586, + "grad_norm": 0.6712111277879811, + "learning_rate": 1.0374144222083997e-05, + "loss": 0.25456517934799194, + "step": 2894 + }, + { + "epoch": 1.6193006993006993, + "grad_norm": 0.4805620026977263, + "learning_rate": 1.036764032515016e-05, + "loss": 0.12548322975635529, + "step": 2895 + }, + { + "epoch": 1.61986013986014, + "grad_norm": 0.3527157843072519, + "learning_rate": 1.0361136272487761e-05, + "loss": 0.07098907977342606, + "step": 2896 + }, + { + "epoch": 1.6204195804195805, + "grad_norm": 0.47015096350439517, + "learning_rate": 1.0354632066851848e-05, + "loss": 0.14188343286514282, + "step": 2897 + }, + { + "epoch": 1.620979020979021, + "grad_norm": 0.4152579181259369, + "learning_rate": 1.0348127710997534e-05, + "loss": 0.14035232365131378, + "step": 2898 + }, + { + "epoch": 1.6215384615384614, + "grad_norm": 0.5265679889567717, + "learning_rate": 1.0341623207679994e-05, + "loss": 0.24965976178646088, + "step": 2899 + }, + { + "epoch": 1.622097902097902, + "grad_norm": 0.5506190631952943, + "learning_rate": 1.033511855965447e-05, + "loss": 0.22281786799430847, + "step": 2900 + }, + { + "epoch": 1.6226573426573427, + "grad_norm": 0.410629617498307, + "learning_rate": 1.0328613769676258e-05, + "loss": 0.11965949088335037, + "step": 2901 + }, + { + "epoch": 1.6232167832167832, + "grad_norm": 0.22628036035340557, + "learning_rate": 1.0322108840500722e-05, + "loss": 0.04703614115715027, + "step": 2902 + }, + { + "epoch": 1.6237762237762237, + "grad_norm": 0.3856360808364171, + "learning_rate": 1.0315603774883282e-05, + "loss": 0.07611198723316193, + "step": 2903 + }, + { + "epoch": 1.6243356643356643, + "grad_norm": 0.5797053168540631, + "learning_rate": 1.0309098575579413e-05, + "loss": 0.18084417283535004, + "step": 2904 + }, + { + "epoch": 1.624895104895105, + "grad_norm": 0.5558495671410439, + "learning_rate": 1.0302593245344646e-05, + "loss": 0.195207878947258, + "step": 2905 + }, + { + "epoch": 1.6254545454545455, + "grad_norm": 0.47966281883660516, + "learning_rate": 1.0296087786934575e-05, + "loss": 0.14961335062980652, + "step": 2906 + }, + { + "epoch": 1.626013986013986, + "grad_norm": 0.6033491784225403, + "learning_rate": 1.0289582203104843e-05, + "loss": 0.27207764983177185, + "step": 2907 + }, + { + "epoch": 1.6265734265734266, + "grad_norm": 0.4490017165735263, + "learning_rate": 1.0283076496611145e-05, + "loss": 0.13873720169067383, + "step": 2908 + }, + { + "epoch": 1.627132867132867, + "grad_norm": 0.40552389324974175, + "learning_rate": 1.027657067020923e-05, + "loss": 0.3017807602882385, + "step": 2909 + }, + { + "epoch": 1.6276923076923078, + "grad_norm": 0.47550049907725767, + "learning_rate": 1.0270064726654897e-05, + "loss": 0.10161522030830383, + "step": 2910 + }, + { + "epoch": 1.6282517482517482, + "grad_norm": 0.38391960596944674, + "learning_rate": 1.0263558668703995e-05, + "loss": 0.1412440687417984, + "step": 2911 + }, + { + "epoch": 1.6288111888111887, + "grad_norm": 0.473121965886524, + "learning_rate": 1.0257052499112425e-05, + "loss": 0.15437306463718414, + "step": 2912 + }, + { + "epoch": 1.6293706293706294, + "grad_norm": 0.7135904745663675, + "learning_rate": 1.0250546220636128e-05, + "loss": 0.11204564571380615, + "step": 2913 + }, + { + "epoch": 1.62993006993007, + "grad_norm": 0.3753995731502226, + "learning_rate": 1.0244039836031094e-05, + "loss": 0.15478914976119995, + "step": 2914 + }, + { + "epoch": 1.6304895104895105, + "grad_norm": 0.6092927848053656, + "learning_rate": 1.0237533348053361e-05, + "loss": 0.13831430673599243, + "step": 2915 + }, + { + "epoch": 1.631048951048951, + "grad_norm": 0.2707372301504066, + "learning_rate": 1.023102675945901e-05, + "loss": 0.03411205857992172, + "step": 2916 + }, + { + "epoch": 1.6316083916083917, + "grad_norm": 0.30288567738539474, + "learning_rate": 1.022452007300416e-05, + "loss": 0.08191650360822678, + "step": 2917 + }, + { + "epoch": 1.6321678321678321, + "grad_norm": 0.36319401204017315, + "learning_rate": 1.0218013291444979e-05, + "loss": 0.1504591703414917, + "step": 2918 + }, + { + "epoch": 1.6327272727272728, + "grad_norm": 0.4047685167209608, + "learning_rate": 1.0211506417537666e-05, + "loss": 0.11800846457481384, + "step": 2919 + }, + { + "epoch": 1.6332867132867133, + "grad_norm": 0.5741365346693781, + "learning_rate": 1.0204999454038466e-05, + "loss": 0.2851371765136719, + "step": 2920 + }, + { + "epoch": 1.6338461538461537, + "grad_norm": 0.43716404512706925, + "learning_rate": 1.0198492403703662e-05, + "loss": 0.17838457226753235, + "step": 2921 + }, + { + "epoch": 1.6344055944055944, + "grad_norm": 0.49748864135515125, + "learning_rate": 1.0191985269289569e-05, + "loss": 0.027160657569766045, + "step": 2922 + }, + { + "epoch": 1.634965034965035, + "grad_norm": 0.6310205013851521, + "learning_rate": 1.0185478053552543e-05, + "loss": 0.19782215356826782, + "step": 2923 + }, + { + "epoch": 1.6355244755244756, + "grad_norm": 0.35862313548302854, + "learning_rate": 1.017897075924897e-05, + "loss": 0.09065507352352142, + "step": 2924 + }, + { + "epoch": 1.636083916083916, + "grad_norm": 0.4793438051370098, + "learning_rate": 1.0172463389135273e-05, + "loss": 0.15299364924430847, + "step": 2925 + }, + { + "epoch": 1.6366433566433567, + "grad_norm": 0.4019306276943155, + "learning_rate": 1.0165955945967904e-05, + "loss": 0.10404511541128159, + "step": 2926 + }, + { + "epoch": 1.6372027972027972, + "grad_norm": 0.32651537603458425, + "learning_rate": 1.0159448432503345e-05, + "loss": 0.07830087095499039, + "step": 2927 + }, + { + "epoch": 1.6377622377622378, + "grad_norm": 0.4407812576160348, + "learning_rate": 1.0152940851498114e-05, + "loss": 0.19758430123329163, + "step": 2928 + }, + { + "epoch": 1.6383216783216783, + "grad_norm": 0.393414335504958, + "learning_rate": 1.0146433205708755e-05, + "loss": 0.12152351438999176, + "step": 2929 + }, + { + "epoch": 1.6388811188811188, + "grad_norm": 0.4361311998207696, + "learning_rate": 1.0139925497891834e-05, + "loss": 0.12206335365772247, + "step": 2930 + }, + { + "epoch": 1.6394405594405594, + "grad_norm": 0.2550229641804532, + "learning_rate": 1.0133417730803947e-05, + "loss": 0.03285663574934006, + "step": 2931 + }, + { + "epoch": 1.6400000000000001, + "grad_norm": 0.4585508572888316, + "learning_rate": 1.0126909907201717e-05, + "loss": 0.060423556715250015, + "step": 2932 + }, + { + "epoch": 1.6405594405594406, + "grad_norm": 0.5187976561077693, + "learning_rate": 1.012040202984179e-05, + "loss": 0.24861504137516022, + "step": 2933 + }, + { + "epoch": 1.641118881118881, + "grad_norm": 0.4620417418527784, + "learning_rate": 1.0113894101480834e-05, + "loss": 0.18891292810440063, + "step": 2934 + }, + { + "epoch": 1.6416783216783217, + "grad_norm": 0.34021414841714875, + "learning_rate": 1.0107386124875542e-05, + "loss": 0.08980001509189606, + "step": 2935 + }, + { + "epoch": 1.6422377622377622, + "grad_norm": 0.41904349432083154, + "learning_rate": 1.0100878102782617e-05, + "loss": 0.08959965407848358, + "step": 2936 + }, + { + "epoch": 1.6427972027972029, + "grad_norm": 0.38888629351421367, + "learning_rate": 1.009437003795879e-05, + "loss": 0.14102260768413544, + "step": 2937 + }, + { + "epoch": 1.6433566433566433, + "grad_norm": 0.3852752260708846, + "learning_rate": 1.0087861933160811e-05, + "loss": 0.15976952016353607, + "step": 2938 + }, + { + "epoch": 1.6439160839160838, + "grad_norm": 0.38484910440127584, + "learning_rate": 1.008135379114545e-05, + "loss": 0.1374768167734146, + "step": 2939 + }, + { + "epoch": 1.6444755244755245, + "grad_norm": 0.4044884599401813, + "learning_rate": 1.0074845614669477e-05, + "loss": 0.12863121926784515, + "step": 2940 + }, + { + "epoch": 1.6450349650349652, + "grad_norm": 0.43609650979809356, + "learning_rate": 1.0068337406489694e-05, + "loss": 0.08922328054904938, + "step": 2941 + }, + { + "epoch": 1.6455944055944056, + "grad_norm": 0.697227766408502, + "learning_rate": 1.0061829169362905e-05, + "loss": 0.2066095769405365, + "step": 2942 + }, + { + "epoch": 1.646153846153846, + "grad_norm": 0.4125791500933971, + "learning_rate": 1.0055320906045934e-05, + "loss": 0.11847516894340515, + "step": 2943 + }, + { + "epoch": 1.6467132867132868, + "grad_norm": 0.37923457324365933, + "learning_rate": 1.0048812619295607e-05, + "loss": 0.10722637176513672, + "step": 2944 + }, + { + "epoch": 1.6472727272727272, + "grad_norm": 0.41528065853131374, + "learning_rate": 1.004230431186877e-05, + "loss": 0.11080102622509003, + "step": 2945 + }, + { + "epoch": 1.647832167832168, + "grad_norm": 0.5234839592561732, + "learning_rate": 1.0035795986522278e-05, + "loss": 0.27729618549346924, + "step": 2946 + }, + { + "epoch": 1.6483916083916084, + "grad_norm": 0.3613251756580539, + "learning_rate": 1.002928764601298e-05, + "loss": 0.10443106293678284, + "step": 2947 + }, + { + "epoch": 1.6489510489510488, + "grad_norm": 0.6551586031141978, + "learning_rate": 1.0022779293097746e-05, + "loss": 0.23782573640346527, + "step": 2948 + }, + { + "epoch": 1.6495104895104895, + "grad_norm": 0.4889401623815567, + "learning_rate": 1.0016270930533442e-05, + "loss": 0.1809787154197693, + "step": 2949 + }, + { + "epoch": 1.6500699300699302, + "grad_norm": 0.6685558716560108, + "learning_rate": 1.0009762561076943e-05, + "loss": 0.37564533948898315, + "step": 2950 + }, + { + "epoch": 1.6506293706293707, + "grad_norm": 0.5931590601105193, + "learning_rate": 1.0003254187485127e-05, + "loss": 0.19317422807216644, + "step": 2951 + }, + { + "epoch": 1.6511888111888111, + "grad_norm": 0.4596966732163005, + "learning_rate": 9.996745812514876e-06, + "loss": 0.2450426071882248, + "step": 2952 + }, + { + "epoch": 1.6517482517482518, + "grad_norm": 0.39170103814507756, + "learning_rate": 9.990237438923057e-06, + "loss": 0.10384012758731842, + "step": 2953 + }, + { + "epoch": 1.6523076923076923, + "grad_norm": 0.48624406046738683, + "learning_rate": 9.983729069466561e-06, + "loss": 0.1966351568698883, + "step": 2954 + }, + { + "epoch": 1.652867132867133, + "grad_norm": 0.3534519281407321, + "learning_rate": 9.977220706902259e-06, + "loss": 0.0763726457953453, + "step": 2955 + }, + { + "epoch": 1.6534265734265734, + "grad_norm": 0.4521340722222131, + "learning_rate": 9.970712353987022e-06, + "loss": 0.1695699691772461, + "step": 2956 + }, + { + "epoch": 1.6539860139860139, + "grad_norm": 0.44135778096989026, + "learning_rate": 9.964204013477727e-06, + "loss": 0.15836942195892334, + "step": 2957 + }, + { + "epoch": 1.6545454545454545, + "grad_norm": 0.5132442107245544, + "learning_rate": 9.957695688131229e-06, + "loss": 0.22079992294311523, + "step": 2958 + }, + { + "epoch": 1.6551048951048952, + "grad_norm": 0.15482903013604776, + "learning_rate": 9.951187380704396e-06, + "loss": 0.013805646449327469, + "step": 2959 + }, + { + "epoch": 1.6556643356643357, + "grad_norm": 0.5950806203453708, + "learning_rate": 9.944679093954073e-06, + "loss": 0.11412828415632248, + "step": 2960 + }, + { + "epoch": 1.6562237762237761, + "grad_norm": 1.9228194012522422, + "learning_rate": 9.938170830637099e-06, + "loss": 0.4606684744358063, + "step": 2961 + }, + { + "epoch": 1.6567832167832168, + "grad_norm": 0.42121957059928217, + "learning_rate": 9.931662593510312e-06, + "loss": 0.17266225814819336, + "step": 2962 + }, + { + "epoch": 1.6573426573426573, + "grad_norm": 0.4284167113110023, + "learning_rate": 9.925154385330524e-06, + "loss": 0.11028154194355011, + "step": 2963 + }, + { + "epoch": 1.657902097902098, + "grad_norm": 0.5454176971441347, + "learning_rate": 9.918646208854553e-06, + "loss": 0.19698432087898254, + "step": 2964 + }, + { + "epoch": 1.6584615384615384, + "grad_norm": 0.6108667921896949, + "learning_rate": 9.912138066839187e-06, + "loss": 0.29060137271881104, + "step": 2965 + }, + { + "epoch": 1.659020979020979, + "grad_norm": 0.5656210091441917, + "learning_rate": 9.905629962041212e-06, + "loss": 0.23911085724830627, + "step": 2966 + }, + { + "epoch": 1.6595804195804196, + "grad_norm": 0.45315733057143176, + "learning_rate": 9.89912189721739e-06, + "loss": 0.11736980825662613, + "step": 2967 + }, + { + "epoch": 1.6601398601398603, + "grad_norm": 0.572740056691188, + "learning_rate": 9.892613875124461e-06, + "loss": 0.19721703231334686, + "step": 2968 + }, + { + "epoch": 1.6606993006993007, + "grad_norm": 0.39796866199515424, + "learning_rate": 9.886105898519168e-06, + "loss": 0.09309659898281097, + "step": 2969 + }, + { + "epoch": 1.6612587412587412, + "grad_norm": 0.43602240177350404, + "learning_rate": 9.879597970158212e-06, + "loss": 0.14063605666160583, + "step": 2970 + }, + { + "epoch": 1.6618181818181819, + "grad_norm": 0.3266493967559965, + "learning_rate": 9.873090092798285e-06, + "loss": 0.0974130779504776, + "step": 2971 + }, + { + "epoch": 1.6623776223776225, + "grad_norm": 0.49720868719898514, + "learning_rate": 9.866582269196058e-06, + "loss": 0.24445199966430664, + "step": 2972 + }, + { + "epoch": 1.662937062937063, + "grad_norm": 0.5043260390393252, + "learning_rate": 9.86007450210817e-06, + "loss": 0.21387150883674622, + "step": 2973 + }, + { + "epoch": 1.6634965034965035, + "grad_norm": 0.41041952913646756, + "learning_rate": 9.853566794291248e-06, + "loss": 0.04549172520637512, + "step": 2974 + }, + { + "epoch": 1.664055944055944, + "grad_norm": 0.6660951699940785, + "learning_rate": 9.847059148501886e-06, + "loss": 0.15380819141864777, + "step": 2975 + }, + { + "epoch": 1.6646153846153846, + "grad_norm": 0.4510889024416934, + "learning_rate": 9.840551567496657e-06, + "loss": 0.1858704388141632, + "step": 2976 + }, + { + "epoch": 1.6651748251748253, + "grad_norm": 0.5632932025035269, + "learning_rate": 9.834044054032101e-06, + "loss": 0.2124638855457306, + "step": 2977 + }, + { + "epoch": 1.6657342657342658, + "grad_norm": 0.6957515066851443, + "learning_rate": 9.82753661086473e-06, + "loss": 0.2959882318973541, + "step": 2978 + }, + { + "epoch": 1.6662937062937062, + "grad_norm": 0.45973322997655136, + "learning_rate": 9.821029240751035e-06, + "loss": 0.15173383057117462, + "step": 2979 + }, + { + "epoch": 1.666853146853147, + "grad_norm": 0.2656551971988455, + "learning_rate": 9.814521946447459e-06, + "loss": 0.05073995888233185, + "step": 2980 + }, + { + "epoch": 1.6674125874125876, + "grad_norm": 0.2918075165825921, + "learning_rate": 9.808014730710433e-06, + "loss": 0.04940441995859146, + "step": 2981 + }, + { + "epoch": 1.667972027972028, + "grad_norm": 0.48018083497249064, + "learning_rate": 9.80150759629634e-06, + "loss": 0.10461293160915375, + "step": 2982 + }, + { + "epoch": 1.6685314685314685, + "grad_norm": 0.5431110507352617, + "learning_rate": 9.795000545961537e-06, + "loss": 0.3275340795516968, + "step": 2983 + }, + { + "epoch": 1.669090909090909, + "grad_norm": 0.29226962622082203, + "learning_rate": 9.788493582462339e-06, + "loss": 0.06173175200819969, + "step": 2984 + }, + { + "epoch": 1.6696503496503496, + "grad_norm": 0.5355931136146239, + "learning_rate": 9.781986708555023e-06, + "loss": 0.1987234354019165, + "step": 2985 + }, + { + "epoch": 1.6702097902097903, + "grad_norm": 0.36336850942670224, + "learning_rate": 9.775479926995841e-06, + "loss": 0.19026236236095428, + "step": 2986 + }, + { + "epoch": 1.6707692307692308, + "grad_norm": 0.3252324576368939, + "learning_rate": 9.768973240540993e-06, + "loss": 0.06463699042797089, + "step": 2987 + }, + { + "epoch": 1.6713286713286712, + "grad_norm": 0.4037048273101081, + "learning_rate": 9.762466651946642e-06, + "loss": 0.133989155292511, + "step": 2988 + }, + { + "epoch": 1.671888111888112, + "grad_norm": 0.5732010718364146, + "learning_rate": 9.75596016396891e-06, + "loss": 0.3031919002532959, + "step": 2989 + }, + { + "epoch": 1.6724475524475526, + "grad_norm": 0.4206674452553623, + "learning_rate": 9.749453779363875e-06, + "loss": 0.10024263709783554, + "step": 2990 + }, + { + "epoch": 1.673006993006993, + "grad_norm": 0.6511255801239153, + "learning_rate": 9.742947500887578e-06, + "loss": 0.20258700847625732, + "step": 2991 + }, + { + "epoch": 1.6735664335664335, + "grad_norm": 0.39541637246810724, + "learning_rate": 9.736441331296006e-06, + "loss": 0.1786518394947052, + "step": 2992 + }, + { + "epoch": 1.674125874125874, + "grad_norm": 0.44559157070017014, + "learning_rate": 9.729935273345105e-06, + "loss": 0.2345372438430786, + "step": 2993 + }, + { + "epoch": 1.6746853146853147, + "grad_norm": 0.4948562780295993, + "learning_rate": 9.723429329790774e-06, + "loss": 0.22118642926216125, + "step": 2994 + }, + { + "epoch": 1.6752447552447554, + "grad_norm": 0.39973340553041964, + "learning_rate": 9.716923503388856e-06, + "loss": 0.1314345896244049, + "step": 2995 + }, + { + "epoch": 1.6758041958041958, + "grad_norm": 0.4642526545510657, + "learning_rate": 9.710417796895159e-06, + "loss": 0.08646439760923386, + "step": 2996 + }, + { + "epoch": 1.6763636363636363, + "grad_norm": 0.3526107050839664, + "learning_rate": 9.703912213065425e-06, + "loss": 0.05085134878754616, + "step": 2997 + }, + { + "epoch": 1.676923076923077, + "grad_norm": 1.4057971482723521, + "learning_rate": 9.697406754655357e-06, + "loss": 0.1367836743593216, + "step": 2998 + }, + { + "epoch": 1.6774825174825176, + "grad_norm": 0.4350949401271611, + "learning_rate": 9.690901424420588e-06, + "loss": 0.13026833534240723, + "step": 2999 + }, + { + "epoch": 1.678041958041958, + "grad_norm": 0.419240591508611, + "learning_rate": 9.68439622511672e-06, + "loss": 0.11952243000268936, + "step": 3000 + }, + { + "epoch": 1.6786013986013986, + "grad_norm": 0.4239516065330251, + "learning_rate": 9.677891159499282e-06, + "loss": 0.10476125776767731, + "step": 3001 + }, + { + "epoch": 1.679160839160839, + "grad_norm": 0.4559276491656999, + "learning_rate": 9.671386230323744e-06, + "loss": 0.17805519700050354, + "step": 3002 + }, + { + "epoch": 1.6797202797202797, + "grad_norm": 0.40319310931443003, + "learning_rate": 9.664881440345534e-06, + "loss": 0.12021731585264206, + "step": 3003 + }, + { + "epoch": 1.6802797202797204, + "grad_norm": 0.589591264300492, + "learning_rate": 9.658376792320008e-06, + "loss": 0.1792115569114685, + "step": 3004 + }, + { + "epoch": 1.6808391608391609, + "grad_norm": 0.46581164721600765, + "learning_rate": 9.651872289002471e-06, + "loss": 0.20040813088417053, + "step": 3005 + }, + { + "epoch": 1.6813986013986013, + "grad_norm": 0.37498737580920827, + "learning_rate": 9.645367933148158e-06, + "loss": 0.05844014883041382, + "step": 3006 + }, + { + "epoch": 1.681958041958042, + "grad_norm": 0.3426453975057003, + "learning_rate": 9.63886372751224e-06, + "loss": 0.06185554713010788, + "step": 3007 + }, + { + "epoch": 1.6825174825174827, + "grad_norm": 0.33192963497291966, + "learning_rate": 9.632359674849842e-06, + "loss": 0.08320507407188416, + "step": 3008 + }, + { + "epoch": 1.6830769230769231, + "grad_norm": 0.636991155585251, + "learning_rate": 9.625855777916005e-06, + "loss": 0.19655077159404755, + "step": 3009 + }, + { + "epoch": 1.6836363636363636, + "grad_norm": 0.48470589554264937, + "learning_rate": 9.619352039465715e-06, + "loss": 0.1426973193883896, + "step": 3010 + }, + { + "epoch": 1.684195804195804, + "grad_norm": 0.45946678500266813, + "learning_rate": 9.612848462253882e-06, + "loss": 0.14301709830760956, + "step": 3011 + }, + { + "epoch": 1.6847552447552447, + "grad_norm": 0.22887907563717505, + "learning_rate": 9.606345049035352e-06, + "loss": 0.0847204178571701, + "step": 3012 + }, + { + "epoch": 1.6853146853146854, + "grad_norm": 0.354578316212586, + "learning_rate": 9.59984180256491e-06, + "loss": 0.09985585510730743, + "step": 3013 + }, + { + "epoch": 1.6858741258741259, + "grad_norm": 0.3267831865456018, + "learning_rate": 9.593338725597255e-06, + "loss": 0.09818299114704132, + "step": 3014 + }, + { + "epoch": 1.6864335664335663, + "grad_norm": 0.4580448433271002, + "learning_rate": 9.586835820887026e-06, + "loss": 0.15568727254867554, + "step": 3015 + }, + { + "epoch": 1.686993006993007, + "grad_norm": 0.4639313088434833, + "learning_rate": 9.580333091188784e-06, + "loss": 0.13128268718719482, + "step": 3016 + }, + { + "epoch": 1.6875524475524477, + "grad_norm": 0.5235405418925675, + "learning_rate": 9.573830539257012e-06, + "loss": 0.2241504043340683, + "step": 3017 + }, + { + "epoch": 1.6881118881118882, + "grad_norm": 0.49377486564610523, + "learning_rate": 9.56732816784613e-06, + "loss": 0.16132117807865143, + "step": 3018 + }, + { + "epoch": 1.6886713286713286, + "grad_norm": 0.5243527819164711, + "learning_rate": 9.560825979710469e-06, + "loss": 0.15984748303890228, + "step": 3019 + }, + { + "epoch": 1.689230769230769, + "grad_norm": 0.6142077472017384, + "learning_rate": 9.554323977604291e-06, + "loss": 0.25019705295562744, + "step": 3020 + }, + { + "epoch": 1.6897902097902098, + "grad_norm": 0.1152469453848433, + "learning_rate": 9.547822164281766e-06, + "loss": 0.007543501909822226, + "step": 3021 + }, + { + "epoch": 1.6903496503496505, + "grad_norm": 0.4845242475011119, + "learning_rate": 9.541320542497009e-06, + "loss": 0.2000923752784729, + "step": 3022 + }, + { + "epoch": 1.690909090909091, + "grad_norm": 0.596606719820553, + "learning_rate": 9.534819115004027e-06, + "loss": 0.33368125557899475, + "step": 3023 + }, + { + "epoch": 1.6914685314685314, + "grad_norm": 0.4306964980827292, + "learning_rate": 9.528317884556759e-06, + "loss": 0.13940201699733734, + "step": 3024 + }, + { + "epoch": 1.692027972027972, + "grad_norm": 0.40216552637621544, + "learning_rate": 9.521816853909059e-06, + "loss": 0.16801759600639343, + "step": 3025 + }, + { + "epoch": 1.6925874125874127, + "grad_norm": 0.4922263140398671, + "learning_rate": 9.51531602581469e-06, + "loss": 0.110100656747818, + "step": 3026 + }, + { + "epoch": 1.6931468531468532, + "grad_norm": 0.6321621836481088, + "learning_rate": 9.508815403027343e-06, + "loss": 0.2312549203634262, + "step": 3027 + }, + { + "epoch": 1.6937062937062937, + "grad_norm": 0.5789414357305648, + "learning_rate": 9.502314988300606e-06, + "loss": 0.2715122699737549, + "step": 3028 + }, + { + "epoch": 1.6942657342657341, + "grad_norm": 0.48631479137608496, + "learning_rate": 9.495814784387984e-06, + "loss": 0.11637977510690689, + "step": 3029 + }, + { + "epoch": 1.6948251748251748, + "grad_norm": 0.5682251456942948, + "learning_rate": 9.489314794042904e-06, + "loss": 0.24676674604415894, + "step": 3030 + }, + { + "epoch": 1.6953846153846155, + "grad_norm": 1.4071207725563597, + "learning_rate": 9.482815020018686e-06, + "loss": 0.3834327161312103, + "step": 3031 + }, + { + "epoch": 1.695944055944056, + "grad_norm": 0.49890258481349065, + "learning_rate": 9.47631546506857e-06, + "loss": 0.21393917500972748, + "step": 3032 + }, + { + "epoch": 1.6965034965034964, + "grad_norm": 0.326821294323107, + "learning_rate": 9.469816131945698e-06, + "loss": 0.10340915620326996, + "step": 3033 + }, + { + "epoch": 1.697062937062937, + "grad_norm": 0.4163626216550937, + "learning_rate": 9.463317023403112e-06, + "loss": 0.13875150680541992, + "step": 3034 + }, + { + "epoch": 1.6976223776223778, + "grad_norm": 0.4173527970130083, + "learning_rate": 9.456818142193774e-06, + "loss": 0.10419341921806335, + "step": 3035 + }, + { + "epoch": 1.6981818181818182, + "grad_norm": 0.3476178065961141, + "learning_rate": 9.45031949107054e-06, + "loss": 0.12065944820642471, + "step": 3036 + }, + { + "epoch": 1.6987412587412587, + "grad_norm": 0.4808104197467235, + "learning_rate": 9.44382107278617e-06, + "loss": 0.19922295212745667, + "step": 3037 + }, + { + "epoch": 1.6993006993006992, + "grad_norm": 0.38530107011672404, + "learning_rate": 9.43732289009332e-06, + "loss": 0.12009718269109726, + "step": 3038 + }, + { + "epoch": 1.6998601398601398, + "grad_norm": 0.40673063962366685, + "learning_rate": 9.430824945744561e-06, + "loss": 0.15200388431549072, + "step": 3039 + }, + { + "epoch": 1.7004195804195805, + "grad_norm": 0.326091119780937, + "learning_rate": 9.424327242492349e-06, + "loss": 0.08002086728811264, + "step": 3040 + }, + { + "epoch": 1.700979020979021, + "grad_norm": 0.6686106240414941, + "learning_rate": 9.41782978308904e-06, + "loss": 0.332578182220459, + "step": 3041 + }, + { + "epoch": 1.7015384615384614, + "grad_norm": 0.3457627992815248, + "learning_rate": 9.411332570286894e-06, + "loss": 0.13766759634017944, + "step": 3042 + }, + { + "epoch": 1.7020979020979021, + "grad_norm": 0.5766191763410615, + "learning_rate": 9.404835606838053e-06, + "loss": 0.20101264119148254, + "step": 3043 + }, + { + "epoch": 1.7026573426573428, + "grad_norm": 0.5552379339051993, + "learning_rate": 9.398338895494574e-06, + "loss": 0.33135032653808594, + "step": 3044 + }, + { + "epoch": 1.7032167832167833, + "grad_norm": 0.31280126944367814, + "learning_rate": 9.391842439008391e-06, + "loss": 0.026278357952833176, + "step": 3045 + }, + { + "epoch": 1.7037762237762237, + "grad_norm": 0.3930204259793614, + "learning_rate": 9.38534624013133e-06, + "loss": 0.13241004943847656, + "step": 3046 + }, + { + "epoch": 1.7043356643356642, + "grad_norm": 0.22651648477675648, + "learning_rate": 9.378850301615118e-06, + "loss": 0.06253943592309952, + "step": 3047 + }, + { + "epoch": 1.7048951048951049, + "grad_norm": 0.5037480244310341, + "learning_rate": 9.372354626211359e-06, + "loss": 0.16502872109413147, + "step": 3048 + }, + { + "epoch": 1.7054545454545456, + "grad_norm": 0.43582681633890713, + "learning_rate": 9.36585921667156e-06, + "loss": 0.2116778939962387, + "step": 3049 + }, + { + "epoch": 1.706013986013986, + "grad_norm": 0.7570866518977117, + "learning_rate": 9.359364075747111e-06, + "loss": 0.3423939347267151, + "step": 3050 + }, + { + "epoch": 1.7065734265734265, + "grad_norm": 0.27339739809562075, + "learning_rate": 9.35286920618927e-06, + "loss": 0.03352992236614227, + "step": 3051 + }, + { + "epoch": 1.7071328671328672, + "grad_norm": 0.5887844116099638, + "learning_rate": 9.34637461074921e-06, + "loss": 0.21770413219928741, + "step": 3052 + }, + { + "epoch": 1.7076923076923078, + "grad_norm": 0.29438374406575696, + "learning_rate": 9.339880292177964e-06, + "loss": 0.04001019150018692, + "step": 3053 + }, + { + "epoch": 1.7082517482517483, + "grad_norm": 0.40692495892423747, + "learning_rate": 9.333386253226464e-06, + "loss": 0.13102063536643982, + "step": 3054 + }, + { + "epoch": 1.7088111888111888, + "grad_norm": 0.5454506392892898, + "learning_rate": 9.326892496645512e-06, + "loss": 0.14729930460453033, + "step": 3055 + }, + { + "epoch": 1.7093706293706292, + "grad_norm": 0.4223124177010605, + "learning_rate": 9.320399025185791e-06, + "loss": 0.14406803250312805, + "step": 3056 + }, + { + "epoch": 1.70993006993007, + "grad_norm": 0.5078160671321608, + "learning_rate": 9.313905841597875e-06, + "loss": 0.1300947070121765, + "step": 3057 + }, + { + "epoch": 1.7104895104895106, + "grad_norm": 0.5227605416752954, + "learning_rate": 9.307412948632204e-06, + "loss": 0.23029902577400208, + "step": 3058 + }, + { + "epoch": 1.711048951048951, + "grad_norm": 0.49892121669115724, + "learning_rate": 9.3009203490391e-06, + "loss": 0.23121728003025055, + "step": 3059 + }, + { + "epoch": 1.7116083916083915, + "grad_norm": 0.14770498059948153, + "learning_rate": 9.294428045568757e-06, + "loss": 0.0074851601384580135, + "step": 3060 + }, + { + "epoch": 1.7121678321678322, + "grad_norm": 0.4184151147840432, + "learning_rate": 9.287936040971251e-06, + "loss": 0.09484653174877167, + "step": 3061 + }, + { + "epoch": 1.7127272727272729, + "grad_norm": 0.4651299313295798, + "learning_rate": 9.281444337996525e-06, + "loss": 0.13925182819366455, + "step": 3062 + }, + { + "epoch": 1.7132867132867133, + "grad_norm": 0.4875848175753103, + "learning_rate": 9.274952939394391e-06, + "loss": 0.19127994775772095, + "step": 3063 + }, + { + "epoch": 1.7138461538461538, + "grad_norm": 0.5119668489743675, + "learning_rate": 9.268461847914545e-06, + "loss": 0.1682354211807251, + "step": 3064 + }, + { + "epoch": 1.7144055944055943, + "grad_norm": 0.421961737892221, + "learning_rate": 9.261971066306535e-06, + "loss": 0.1610163301229477, + "step": 3065 + }, + { + "epoch": 1.714965034965035, + "grad_norm": 0.6942276306851286, + "learning_rate": 9.255480597319798e-06, + "loss": 0.19976727664470673, + "step": 3066 + }, + { + "epoch": 1.7155244755244756, + "grad_norm": 0.35680684377195887, + "learning_rate": 9.248990443703624e-06, + "loss": 0.15329812467098236, + "step": 3067 + }, + { + "epoch": 1.716083916083916, + "grad_norm": 0.5452816936178774, + "learning_rate": 9.24250060820717e-06, + "loss": 0.2606050670146942, + "step": 3068 + }, + { + "epoch": 1.7166433566433565, + "grad_norm": 0.47892683215865234, + "learning_rate": 9.236011093579468e-06, + "loss": 0.1810053437948227, + "step": 3069 + }, + { + "epoch": 1.7172027972027972, + "grad_norm": 0.4984237354638142, + "learning_rate": 9.229521902569399e-06, + "loss": 0.13593192398548126, + "step": 3070 + }, + { + "epoch": 1.717762237762238, + "grad_norm": 0.4122397608737338, + "learning_rate": 9.223033037925729e-06, + "loss": 0.13370800018310547, + "step": 3071 + }, + { + "epoch": 1.7183216783216784, + "grad_norm": 0.5911908675468669, + "learning_rate": 9.216544502397062e-06, + "loss": 0.3885428309440613, + "step": 3072 + }, + { + "epoch": 1.7188811188811188, + "grad_norm": 0.5477357661821624, + "learning_rate": 9.210056298731876e-06, + "loss": 0.1855706125497818, + "step": 3073 + }, + { + "epoch": 1.7194405594405593, + "grad_norm": 0.4933536426783183, + "learning_rate": 9.203568429678506e-06, + "loss": 0.16089676320552826, + "step": 3074 + }, + { + "epoch": 1.72, + "grad_norm": 0.5941016626555622, + "learning_rate": 9.197080897985143e-06, + "loss": 0.2637489438056946, + "step": 3075 + }, + { + "epoch": 1.7205594405594407, + "grad_norm": 0.21555318834231976, + "learning_rate": 9.190593706399844e-06, + "loss": 0.039192814379930496, + "step": 3076 + }, + { + "epoch": 1.7211188811188811, + "grad_norm": 0.45102587455162074, + "learning_rate": 9.184106857670504e-06, + "loss": 0.23223662376403809, + "step": 3077 + }, + { + "epoch": 1.7216783216783216, + "grad_norm": 0.4389439206665759, + "learning_rate": 9.177620354544895e-06, + "loss": 0.12181822210550308, + "step": 3078 + }, + { + "epoch": 1.7222377622377623, + "grad_norm": 0.3145253883010683, + "learning_rate": 9.171134199770622e-06, + "loss": 0.04619431123137474, + "step": 3079 + }, + { + "epoch": 1.722797202797203, + "grad_norm": 0.44295260361276173, + "learning_rate": 9.164648396095156e-06, + "loss": 0.1830262690782547, + "step": 3080 + }, + { + "epoch": 1.7233566433566434, + "grad_norm": 0.46327304911430345, + "learning_rate": 9.158162946265816e-06, + "loss": 0.18911191821098328, + "step": 3081 + }, + { + "epoch": 1.7239160839160839, + "grad_norm": 0.49793235363474164, + "learning_rate": 9.151677853029764e-06, + "loss": 0.1590443104505539, + "step": 3082 + }, + { + "epoch": 1.7244755244755243, + "grad_norm": 0.38273849335091936, + "learning_rate": 9.145193119134026e-06, + "loss": 0.1471109688282013, + "step": 3083 + }, + { + "epoch": 1.725034965034965, + "grad_norm": 1.0306656503049725, + "learning_rate": 9.138708747325461e-06, + "loss": 0.40186381340026855, + "step": 3084 + }, + { + "epoch": 1.7255944055944057, + "grad_norm": 0.41311087638598004, + "learning_rate": 9.132224740350782e-06, + "loss": 0.11406250298023224, + "step": 3085 + }, + { + "epoch": 1.7261538461538461, + "grad_norm": 0.6116472786014813, + "learning_rate": 9.125741100956546e-06, + "loss": 0.23277802765369415, + "step": 3086 + }, + { + "epoch": 1.7267132867132866, + "grad_norm": 0.5006366467125534, + "learning_rate": 9.119257831889152e-06, + "loss": 0.10958261787891388, + "step": 3087 + }, + { + "epoch": 1.7272727272727273, + "grad_norm": 0.5070223797356671, + "learning_rate": 9.11277493589485e-06, + "loss": 0.14662399888038635, + "step": 3088 + }, + { + "epoch": 1.727832167832168, + "grad_norm": 0.45237688084995903, + "learning_rate": 9.106292415719723e-06, + "loss": 0.2085634171962738, + "step": 3089 + }, + { + "epoch": 1.7283916083916084, + "grad_norm": 0.31388527583382864, + "learning_rate": 9.099810274109694e-06, + "loss": 0.036556847393512726, + "step": 3090 + }, + { + "epoch": 1.728951048951049, + "grad_norm": 0.2635676230753663, + "learning_rate": 9.093328513810536e-06, + "loss": 0.06841983646154404, + "step": 3091 + }, + { + "epoch": 1.7295104895104894, + "grad_norm": 0.2009341267502672, + "learning_rate": 9.086847137567849e-06, + "loss": 0.007382201496511698, + "step": 3092 + }, + { + "epoch": 1.73006993006993, + "grad_norm": 0.5937640482501849, + "learning_rate": 9.080366148127083e-06, + "loss": 0.2771955132484436, + "step": 3093 + }, + { + "epoch": 1.7306293706293707, + "grad_norm": 0.4237621589798544, + "learning_rate": 9.073885548233509e-06, + "loss": 0.1524166762828827, + "step": 3094 + }, + { + "epoch": 1.7311888111888112, + "grad_norm": 0.31174864080122616, + "learning_rate": 9.067405340632246e-06, + "loss": 0.07445754110813141, + "step": 3095 + }, + { + "epoch": 1.7317482517482516, + "grad_norm": 0.30233493088150964, + "learning_rate": 9.060925528068239e-06, + "loss": 0.06812790036201477, + "step": 3096 + }, + { + "epoch": 1.7323076923076923, + "grad_norm": 0.6074746876034635, + "learning_rate": 9.054446113286266e-06, + "loss": 0.31041720509529114, + "step": 3097 + }, + { + "epoch": 1.732867132867133, + "grad_norm": 0.39125467007892467, + "learning_rate": 9.047967099030945e-06, + "loss": 0.13497644662857056, + "step": 3098 + }, + { + "epoch": 1.7334265734265735, + "grad_norm": 0.3943641652799981, + "learning_rate": 9.041488488046713e-06, + "loss": 0.10116368532180786, + "step": 3099 + }, + { + "epoch": 1.733986013986014, + "grad_norm": 0.23524779050808933, + "learning_rate": 9.035010283077845e-06, + "loss": 0.028338227421045303, + "step": 3100 + }, + { + "epoch": 1.7345454545454544, + "grad_norm": 0.2721533675572473, + "learning_rate": 9.028532486868437e-06, + "loss": 0.10103368759155273, + "step": 3101 + }, + { + "epoch": 1.735104895104895, + "grad_norm": 0.40107235252990775, + "learning_rate": 9.022055102162416e-06, + "loss": 0.0964130088686943, + "step": 3102 + }, + { + "epoch": 1.7356643356643358, + "grad_norm": 0.43699628781227884, + "learning_rate": 9.015578131703534e-06, + "loss": 0.14891275763511658, + "step": 3103 + }, + { + "epoch": 1.7362237762237762, + "grad_norm": 0.5246014938774265, + "learning_rate": 9.009101578235364e-06, + "loss": 0.20016345381736755, + "step": 3104 + }, + { + "epoch": 1.7367832167832167, + "grad_norm": 0.45607817578594206, + "learning_rate": 9.002625444501314e-06, + "loss": 0.1195664256811142, + "step": 3105 + }, + { + "epoch": 1.7373426573426574, + "grad_norm": 0.4071603982603155, + "learning_rate": 8.996149733244599e-06, + "loss": 0.0809250921010971, + "step": 3106 + }, + { + "epoch": 1.737902097902098, + "grad_norm": 0.3140482813439461, + "learning_rate": 8.989674447208263e-06, + "loss": 0.05960802361369133, + "step": 3107 + }, + { + "epoch": 1.7384615384615385, + "grad_norm": 0.4359808488575041, + "learning_rate": 8.983199589135172e-06, + "loss": 0.2069295346736908, + "step": 3108 + }, + { + "epoch": 1.739020979020979, + "grad_norm": 0.561258527642534, + "learning_rate": 8.976725161768e-06, + "loss": 0.2972172498703003, + "step": 3109 + }, + { + "epoch": 1.7395804195804194, + "grad_norm": 0.38984747222624117, + "learning_rate": 8.970251167849257e-06, + "loss": 0.11996152251958847, + "step": 3110 + }, + { + "epoch": 1.74013986013986, + "grad_norm": 0.39976292393392976, + "learning_rate": 8.963777610121253e-06, + "loss": 0.08648248761892319, + "step": 3111 + }, + { + "epoch": 1.7406993006993008, + "grad_norm": 0.4486699256979643, + "learning_rate": 8.957304491326115e-06, + "loss": 0.20402149856090546, + "step": 3112 + }, + { + "epoch": 1.7412587412587412, + "grad_norm": 0.4424281044892107, + "learning_rate": 8.950831814205796e-06, + "loss": 0.1489095240831375, + "step": 3113 + }, + { + "epoch": 1.7418181818181817, + "grad_norm": 0.4063348708474474, + "learning_rate": 8.944359581502042e-06, + "loss": 0.13151924312114716, + "step": 3114 + }, + { + "epoch": 1.7423776223776224, + "grad_norm": 0.4441073211589039, + "learning_rate": 8.937887795956436e-06, + "loss": 0.06777593493461609, + "step": 3115 + }, + { + "epoch": 1.742937062937063, + "grad_norm": 0.43736145660804876, + "learning_rate": 8.93141646031035e-06, + "loss": 0.2953047752380371, + "step": 3116 + }, + { + "epoch": 1.7434965034965035, + "grad_norm": 0.4743464847801903, + "learning_rate": 8.924945577304975e-06, + "loss": 0.19538140296936035, + "step": 3117 + }, + { + "epoch": 1.744055944055944, + "grad_norm": 0.43719127361237703, + "learning_rate": 8.91847514968131e-06, + "loss": 0.10583820939064026, + "step": 3118 + }, + { + "epoch": 1.7446153846153845, + "grad_norm": 0.2911651122176028, + "learning_rate": 8.912005180180155e-06, + "loss": 0.04031681641936302, + "step": 3119 + }, + { + "epoch": 1.7451748251748251, + "grad_norm": 0.3775588653465098, + "learning_rate": 8.905535671542129e-06, + "loss": 0.0754489153623581, + "step": 3120 + }, + { + "epoch": 1.7457342657342658, + "grad_norm": 0.5386244198197893, + "learning_rate": 8.899066626507641e-06, + "loss": 0.190330371260643, + "step": 3121 + }, + { + "epoch": 1.7462937062937063, + "grad_norm": 0.5998099777911083, + "learning_rate": 8.892598047816915e-06, + "loss": 0.24131014943122864, + "step": 3122 + }, + { + "epoch": 1.7468531468531467, + "grad_norm": 0.5287215659003242, + "learning_rate": 8.88612993820997e-06, + "loss": 0.15173746645450592, + "step": 3123 + }, + { + "epoch": 1.7474125874125874, + "grad_norm": 0.5773832065286821, + "learning_rate": 8.879662300426623e-06, + "loss": 0.21992839872837067, + "step": 3124 + }, + { + "epoch": 1.747972027972028, + "grad_norm": 0.36792235856151, + "learning_rate": 8.873195137206511e-06, + "loss": 0.09761320054531097, + "step": 3125 + }, + { + "epoch": 1.7485314685314686, + "grad_norm": 0.5357030788598823, + "learning_rate": 8.866728451289043e-06, + "loss": 0.2660660147666931, + "step": 3126 + }, + { + "epoch": 1.749090909090909, + "grad_norm": 0.4099481962578154, + "learning_rate": 8.860262245413448e-06, + "loss": 0.12606793642044067, + "step": 3127 + }, + { + "epoch": 1.7496503496503495, + "grad_norm": 0.39997610602645095, + "learning_rate": 8.853796522318739e-06, + "loss": 0.09767289459705353, + "step": 3128 + }, + { + "epoch": 1.7502097902097902, + "grad_norm": 0.34098518698972113, + "learning_rate": 8.847331284743724e-06, + "loss": 0.05795377492904663, + "step": 3129 + }, + { + "epoch": 1.7507692307692309, + "grad_norm": 0.4436026088535874, + "learning_rate": 8.840866535427016e-06, + "loss": 0.3228636682033539, + "step": 3130 + }, + { + "epoch": 1.7513286713286713, + "grad_norm": 0.5165951250308968, + "learning_rate": 8.834402277107008e-06, + "loss": 0.1834675818681717, + "step": 3131 + }, + { + "epoch": 1.7518881118881118, + "grad_norm": 0.4860377159778398, + "learning_rate": 8.8279385125219e-06, + "loss": 0.16292870044708252, + "step": 3132 + }, + { + "epoch": 1.7524475524475525, + "grad_norm": 0.78010122513191, + "learning_rate": 8.821475244409668e-06, + "loss": 0.3006657660007477, + "step": 3133 + }, + { + "epoch": 1.7530069930069931, + "grad_norm": 0.6165151258837157, + "learning_rate": 8.815012475508088e-06, + "loss": 0.2342771738767624, + "step": 3134 + }, + { + "epoch": 1.7535664335664336, + "grad_norm": 0.33944936859293023, + "learning_rate": 8.808550208554718e-06, + "loss": 0.0401204377412796, + "step": 3135 + }, + { + "epoch": 1.754125874125874, + "grad_norm": 0.5754752921947396, + "learning_rate": 8.802088446286903e-06, + "loss": 0.2918890118598938, + "step": 3136 + }, + { + "epoch": 1.7546853146853145, + "grad_norm": 0.32588069770124517, + "learning_rate": 8.795627191441787e-06, + "loss": 0.0670238733291626, + "step": 3137 + }, + { + "epoch": 1.7552447552447552, + "grad_norm": 0.38641225690477016, + "learning_rate": 8.78916644675628e-06, + "loss": 0.11483792960643768, + "step": 3138 + }, + { + "epoch": 1.7558041958041959, + "grad_norm": 0.38578660991971786, + "learning_rate": 8.782706214967092e-06, + "loss": 0.09549957513809204, + "step": 3139 + }, + { + "epoch": 1.7563636363636363, + "grad_norm": 0.5592934209751709, + "learning_rate": 8.776246498810704e-06, + "loss": 0.14463545382022858, + "step": 3140 + }, + { + "epoch": 1.7569230769230768, + "grad_norm": 0.49182901315905425, + "learning_rate": 8.769787301023383e-06, + "loss": 0.16351813077926636, + "step": 3141 + }, + { + "epoch": 1.7574825174825175, + "grad_norm": 0.5960547695582296, + "learning_rate": 8.763328624341183e-06, + "loss": 0.25015830993652344, + "step": 3142 + }, + { + "epoch": 1.7580419580419582, + "grad_norm": 0.25169220053365426, + "learning_rate": 8.756870471499927e-06, + "loss": 0.02947881817817688, + "step": 3143 + }, + { + "epoch": 1.7586013986013986, + "grad_norm": 0.4390630639544389, + "learning_rate": 8.750412845235221e-06, + "loss": 0.1618751883506775, + "step": 3144 + }, + { + "epoch": 1.759160839160839, + "grad_norm": 0.6141123902399409, + "learning_rate": 8.74395574828245e-06, + "loss": 0.24477940797805786, + "step": 3145 + }, + { + "epoch": 1.7597202797202796, + "grad_norm": 0.5040091331687755, + "learning_rate": 8.737499183376763e-06, + "loss": 0.24421736598014832, + "step": 3146 + }, + { + "epoch": 1.7602797202797202, + "grad_norm": 0.3334677550786011, + "learning_rate": 8.731043153253105e-06, + "loss": 0.10669678449630737, + "step": 3147 + }, + { + "epoch": 1.760839160839161, + "grad_norm": 0.5680894009023356, + "learning_rate": 8.724587660646175e-06, + "loss": 0.19479043781757355, + "step": 3148 + }, + { + "epoch": 1.7613986013986014, + "grad_norm": 0.5088245877512542, + "learning_rate": 8.718132708290454e-06, + "loss": 0.2415938377380371, + "step": 3149 + }, + { + "epoch": 1.7619580419580418, + "grad_norm": 0.5199590442334292, + "learning_rate": 8.711678298920187e-06, + "loss": 0.268382728099823, + "step": 3150 + }, + { + "epoch": 1.7625174825174825, + "grad_norm": 0.4903936355959743, + "learning_rate": 8.705224435269405e-06, + "loss": 0.15624383091926575, + "step": 3151 + }, + { + "epoch": 1.7630769230769232, + "grad_norm": 0.27797873302544956, + "learning_rate": 8.698771120071887e-06, + "loss": 0.06342122703790665, + "step": 3152 + }, + { + "epoch": 1.7636363636363637, + "grad_norm": 0.5074865725660517, + "learning_rate": 8.692318356061186e-06, + "loss": 0.18470287322998047, + "step": 3153 + }, + { + "epoch": 1.7641958041958041, + "grad_norm": 0.48750822656590637, + "learning_rate": 8.685866145970634e-06, + "loss": 0.24003052711486816, + "step": 3154 + }, + { + "epoch": 1.7647552447552446, + "grad_norm": 0.5912458902585866, + "learning_rate": 8.679414492533315e-06, + "loss": 0.27781444787979126, + "step": 3155 + }, + { + "epoch": 1.7653146853146853, + "grad_norm": 0.632297808181952, + "learning_rate": 8.672963398482081e-06, + "loss": 0.3203977346420288, + "step": 3156 + }, + { + "epoch": 1.765874125874126, + "grad_norm": 0.4888321499696355, + "learning_rate": 8.66651286654955e-06, + "loss": 0.14385612308979034, + "step": 3157 + }, + { + "epoch": 1.7664335664335664, + "grad_norm": 0.344435861323144, + "learning_rate": 8.660062899468093e-06, + "loss": 0.06813342869281769, + "step": 3158 + }, + { + "epoch": 1.7669930069930069, + "grad_norm": 0.370937590995787, + "learning_rate": 8.653613499969856e-06, + "loss": 0.09013888239860535, + "step": 3159 + }, + { + "epoch": 1.7675524475524476, + "grad_norm": 0.5734066763737111, + "learning_rate": 8.64716467078673e-06, + "loss": 0.2302234023809433, + "step": 3160 + }, + { + "epoch": 1.7681118881118882, + "grad_norm": 0.4606282300876776, + "learning_rate": 8.640716414650377e-06, + "loss": 0.09541988372802734, + "step": 3161 + }, + { + "epoch": 1.7686713286713287, + "grad_norm": 0.3246984468368925, + "learning_rate": 8.634268734292205e-06, + "loss": 0.12828890979290009, + "step": 3162 + }, + { + "epoch": 1.7692307692307692, + "grad_norm": 0.4139093332593382, + "learning_rate": 8.627821632443382e-06, + "loss": 0.1581895500421524, + "step": 3163 + }, + { + "epoch": 1.7697902097902098, + "grad_norm": 0.46324722702069226, + "learning_rate": 8.62137511183484e-06, + "loss": 0.08003498613834381, + "step": 3164 + }, + { + "epoch": 1.7703496503496503, + "grad_norm": 0.4877810115696498, + "learning_rate": 8.614929175197251e-06, + "loss": 0.1973947286605835, + "step": 3165 + }, + { + "epoch": 1.770909090909091, + "grad_norm": 0.38691878101847715, + "learning_rate": 8.608483825261049e-06, + "loss": 0.10866017639636993, + "step": 3166 + }, + { + "epoch": 1.7714685314685314, + "grad_norm": 0.40399918996148054, + "learning_rate": 8.602039064756416e-06, + "loss": 0.1250346601009369, + "step": 3167 + }, + { + "epoch": 1.772027972027972, + "grad_norm": 0.6070040235278376, + "learning_rate": 8.595594896413277e-06, + "loss": 0.28763458132743835, + "step": 3168 + }, + { + "epoch": 1.7725874125874126, + "grad_norm": 0.2804513327085983, + "learning_rate": 8.589151322961325e-06, + "loss": 0.05523074418306351, + "step": 3169 + }, + { + "epoch": 1.7731468531468533, + "grad_norm": 0.5346616233102769, + "learning_rate": 8.582708347129984e-06, + "loss": 0.2607904076576233, + "step": 3170 + }, + { + "epoch": 1.7737062937062937, + "grad_norm": 0.44131910270514507, + "learning_rate": 8.576265971648432e-06, + "loss": 0.18346716463565826, + "step": 3171 + }, + { + "epoch": 1.7742657342657342, + "grad_norm": 0.3660913777098182, + "learning_rate": 8.569824199245586e-06, + "loss": 0.09570343792438507, + "step": 3172 + }, + { + "epoch": 1.7748251748251749, + "grad_norm": 0.561006296229054, + "learning_rate": 8.563383032650122e-06, + "loss": 0.1438939869403839, + "step": 3173 + }, + { + "epoch": 1.7753846153846153, + "grad_norm": 0.2903615652193841, + "learning_rate": 8.556942474590447e-06, + "loss": 0.06853242218494415, + "step": 3174 + }, + { + "epoch": 1.775944055944056, + "grad_norm": 0.5929401379533877, + "learning_rate": 8.550502527794706e-06, + "loss": 0.20595574378967285, + "step": 3175 + }, + { + "epoch": 1.7765034965034965, + "grad_norm": 0.5596017521217284, + "learning_rate": 8.544063194990803e-06, + "loss": 0.2571282684803009, + "step": 3176 + }, + { + "epoch": 1.777062937062937, + "grad_norm": 0.39066093970799526, + "learning_rate": 8.537624478906364e-06, + "loss": 0.10524804890155792, + "step": 3177 + }, + { + "epoch": 1.7776223776223776, + "grad_norm": 0.428752003198061, + "learning_rate": 8.531186382268768e-06, + "loss": 0.0796380415558815, + "step": 3178 + }, + { + "epoch": 1.7781818181818183, + "grad_norm": 0.5248783986507709, + "learning_rate": 8.52474890780512e-06, + "loss": 0.17871351540088654, + "step": 3179 + }, + { + "epoch": 1.7787412587412588, + "grad_norm": 0.4832891635202594, + "learning_rate": 8.518312058242262e-06, + "loss": 0.19081605970859528, + "step": 3180 + }, + { + "epoch": 1.7793006993006992, + "grad_norm": 0.3636451902625486, + "learning_rate": 8.511875836306785e-06, + "loss": 0.13260558247566223, + "step": 3181 + }, + { + "epoch": 1.77986013986014, + "grad_norm": 0.40551135912271474, + "learning_rate": 8.505440244724999e-06, + "loss": 0.12255693227052689, + "step": 3182 + }, + { + "epoch": 1.7804195804195804, + "grad_norm": 0.49324874081661946, + "learning_rate": 8.499005286222956e-06, + "loss": 0.1830892264842987, + "step": 3183 + }, + { + "epoch": 1.780979020979021, + "grad_norm": 0.5524851779004367, + "learning_rate": 8.492570963526434e-06, + "loss": 0.11568547785282135, + "step": 3184 + }, + { + "epoch": 1.7815384615384615, + "grad_norm": 0.3247755300886649, + "learning_rate": 8.48613727936094e-06, + "loss": 0.18830077350139618, + "step": 3185 + }, + { + "epoch": 1.782097902097902, + "grad_norm": 0.39045209824030175, + "learning_rate": 8.479704236451726e-06, + "loss": 0.11719164997339249, + "step": 3186 + }, + { + "epoch": 1.7826573426573427, + "grad_norm": 0.5156625138917041, + "learning_rate": 8.473271837523752e-06, + "loss": 0.1845027208328247, + "step": 3187 + }, + { + "epoch": 1.7832167832167833, + "grad_norm": 0.4921303575167285, + "learning_rate": 8.466840085301716e-06, + "loss": 0.18442830443382263, + "step": 3188 + }, + { + "epoch": 1.7837762237762238, + "grad_norm": 0.46717709434330773, + "learning_rate": 8.46040898251004e-06, + "loss": 0.13422854244709015, + "step": 3189 + }, + { + "epoch": 1.7843356643356643, + "grad_norm": 0.6989822570384524, + "learning_rate": 8.453978531872874e-06, + "loss": 0.25387439131736755, + "step": 3190 + }, + { + "epoch": 1.784895104895105, + "grad_norm": 0.7953942969794149, + "learning_rate": 8.447548736114091e-06, + "loss": 0.1822403222322464, + "step": 3191 + }, + { + "epoch": 1.7854545454545454, + "grad_norm": 0.46272796028089097, + "learning_rate": 8.441119597957276e-06, + "loss": 0.15814903378486633, + "step": 3192 + }, + { + "epoch": 1.786013986013986, + "grad_norm": 0.5033493709600168, + "learning_rate": 8.434691120125755e-06, + "loss": 0.20812077820301056, + "step": 3193 + }, + { + "epoch": 1.7865734265734265, + "grad_norm": 0.518740092361043, + "learning_rate": 8.42826330534255e-06, + "loss": 0.2235710322856903, + "step": 3194 + }, + { + "epoch": 1.787132867132867, + "grad_norm": 0.3207279427784994, + "learning_rate": 8.42183615633043e-06, + "loss": 0.06347090005874634, + "step": 3195 + }, + { + "epoch": 1.7876923076923077, + "grad_norm": 0.5151523942554032, + "learning_rate": 8.41540967581186e-06, + "loss": 0.21188542246818542, + "step": 3196 + }, + { + "epoch": 1.7882517482517484, + "grad_norm": 0.3867242056014629, + "learning_rate": 8.408983866509027e-06, + "loss": 0.07785777747631073, + "step": 3197 + }, + { + "epoch": 1.7888111888111888, + "grad_norm": 0.3502133578731688, + "learning_rate": 8.40255873114384e-06, + "loss": 0.10463696718215942, + "step": 3198 + }, + { + "epoch": 1.7893706293706293, + "grad_norm": 0.5181820374150858, + "learning_rate": 8.396134272437914e-06, + "loss": 0.12862011790275574, + "step": 3199 + }, + { + "epoch": 1.78993006993007, + "grad_norm": 0.49496337204980523, + "learning_rate": 8.389710493112593e-06, + "loss": 0.21206022799015045, + "step": 3200 + }, + { + "epoch": 1.7904895104895104, + "grad_norm": 0.736389765456481, + "learning_rate": 8.383287395888913e-06, + "loss": 0.10065078735351562, + "step": 3201 + }, + { + "epoch": 1.7910489510489511, + "grad_norm": 0.3973309131230942, + "learning_rate": 8.376864983487627e-06, + "loss": 0.2102937549352646, + "step": 3202 + }, + { + "epoch": 1.7916083916083916, + "grad_norm": 0.43985431013555787, + "learning_rate": 8.370443258629208e-06, + "loss": 0.12319180369377136, + "step": 3203 + }, + { + "epoch": 1.792167832167832, + "grad_norm": 0.2702655327209036, + "learning_rate": 8.364022224033828e-06, + "loss": 0.06279011070728302, + "step": 3204 + }, + { + "epoch": 1.7927272727272727, + "grad_norm": 0.38192121059366924, + "learning_rate": 8.357601882421373e-06, + "loss": 0.12921187281608582, + "step": 3205 + }, + { + "epoch": 1.7932867132867134, + "grad_norm": 0.5927928512201303, + "learning_rate": 8.35118223651143e-06, + "loss": 0.30726486444473267, + "step": 3206 + }, + { + "epoch": 1.7938461538461539, + "grad_norm": 0.47031518948922035, + "learning_rate": 8.344763289023287e-06, + "loss": 0.03763486444950104, + "step": 3207 + }, + { + "epoch": 1.7944055944055943, + "grad_norm": 0.4293459289904279, + "learning_rate": 8.338345042675953e-06, + "loss": 0.14397838711738586, + "step": 3208 + }, + { + "epoch": 1.794965034965035, + "grad_norm": 0.5533170748384957, + "learning_rate": 8.33192750018812e-06, + "loss": 0.1817794144153595, + "step": 3209 + }, + { + "epoch": 1.7955244755244755, + "grad_norm": 0.4349934345574487, + "learning_rate": 8.3255106642782e-06, + "loss": 0.10801716148853302, + "step": 3210 + }, + { + "epoch": 1.7960839160839162, + "grad_norm": 0.41178250326933646, + "learning_rate": 8.319094537664289e-06, + "loss": 0.05909350514411926, + "step": 3211 + }, + { + "epoch": 1.7966433566433566, + "grad_norm": 0.40426440961933313, + "learning_rate": 8.312679123064197e-06, + "loss": 0.13181468844413757, + "step": 3212 + }, + { + "epoch": 1.797202797202797, + "grad_norm": 0.36877130730779467, + "learning_rate": 8.306264423195423e-06, + "loss": 0.12119457870721817, + "step": 3213 + }, + { + "epoch": 1.7977622377622378, + "grad_norm": 0.3493575018210003, + "learning_rate": 8.299850440775163e-06, + "loss": 0.09390321373939514, + "step": 3214 + }, + { + "epoch": 1.7983216783216784, + "grad_norm": 0.42998320587762834, + "learning_rate": 8.29343717852032e-06, + "loss": 0.17034342885017395, + "step": 3215 + }, + { + "epoch": 1.798881118881119, + "grad_norm": 0.4363641452363663, + "learning_rate": 8.287024639147475e-06, + "loss": 0.10654303431510925, + "step": 3216 + }, + { + "epoch": 1.7994405594405594, + "grad_norm": 0.2624709187783446, + "learning_rate": 8.28061282537292e-06, + "loss": 0.06967205554246902, + "step": 3217 + }, + { + "epoch": 1.8, + "grad_norm": 0.22551229550288834, + "learning_rate": 8.27420173991263e-06, + "loss": 0.03199368715286255, + "step": 3218 + }, + { + "epoch": 1.8005594405594405, + "grad_norm": 0.30644933868794844, + "learning_rate": 8.267791385482268e-06, + "loss": 0.05807441845536232, + "step": 3219 + }, + { + "epoch": 1.8011188811188812, + "grad_norm": 0.4321459264595991, + "learning_rate": 8.261381764797198e-06, + "loss": 0.15787997841835022, + "step": 3220 + }, + { + "epoch": 1.8016783216783216, + "grad_norm": 0.3364456308763508, + "learning_rate": 8.254972880572463e-06, + "loss": 0.10627515614032745, + "step": 3221 + }, + { + "epoch": 1.802237762237762, + "grad_norm": 0.38062808607491144, + "learning_rate": 8.248564735522804e-06, + "loss": 0.0636792704463005, + "step": 3222 + }, + { + "epoch": 1.8027972027972028, + "grad_norm": 0.2978693974126301, + "learning_rate": 8.242157332362643e-06, + "loss": 0.05325791984796524, + "step": 3223 + }, + { + "epoch": 1.8033566433566435, + "grad_norm": 0.44746550139286945, + "learning_rate": 8.235750673806082e-06, + "loss": 0.09092092514038086, + "step": 3224 + }, + { + "epoch": 1.803916083916084, + "grad_norm": 0.2985132280522795, + "learning_rate": 8.229344762566921e-06, + "loss": 0.18818290531635284, + "step": 3225 + }, + { + "epoch": 1.8044755244755244, + "grad_norm": 0.5713012185490992, + "learning_rate": 8.222939601358633e-06, + "loss": 0.21693797409534454, + "step": 3226 + }, + { + "epoch": 1.805034965034965, + "grad_norm": 0.44888602703515484, + "learning_rate": 8.216535192894378e-06, + "loss": 0.10395072400569916, + "step": 3227 + }, + { + "epoch": 1.8055944055944058, + "grad_norm": 0.4303436538499434, + "learning_rate": 8.210131539886992e-06, + "loss": 0.12642090022563934, + "step": 3228 + }, + { + "epoch": 1.8061538461538462, + "grad_norm": 0.43798567266779803, + "learning_rate": 8.203728645049002e-06, + "loss": 0.11497074365615845, + "step": 3229 + }, + { + "epoch": 1.8067132867132867, + "grad_norm": 0.6102327865427207, + "learning_rate": 8.197326511092604e-06, + "loss": 0.20979063212871552, + "step": 3230 + }, + { + "epoch": 1.8072727272727271, + "grad_norm": 0.35037036192294624, + "learning_rate": 8.19092514072967e-06, + "loss": 0.044171929359436035, + "step": 3231 + }, + { + "epoch": 1.8078321678321678, + "grad_norm": 0.22937766952183675, + "learning_rate": 8.18452453667176e-06, + "loss": 0.04106922075152397, + "step": 3232 + }, + { + "epoch": 1.8083916083916085, + "grad_norm": 0.5254574450724214, + "learning_rate": 8.178124701630092e-06, + "loss": 0.22721415758132935, + "step": 3233 + }, + { + "epoch": 1.808951048951049, + "grad_norm": 0.263561400974856, + "learning_rate": 8.17172563831558e-06, + "loss": 0.06986412405967712, + "step": 3234 + }, + { + "epoch": 1.8095104895104894, + "grad_norm": 0.5237494113244238, + "learning_rate": 8.165327349438796e-06, + "loss": 0.23333466053009033, + "step": 3235 + }, + { + "epoch": 1.81006993006993, + "grad_norm": 0.3245227964360753, + "learning_rate": 8.158929837709983e-06, + "loss": 0.0997273251414299, + "step": 3236 + }, + { + "epoch": 1.8106293706293708, + "grad_norm": 0.3522334735726338, + "learning_rate": 8.152533105839066e-06, + "loss": 0.3249717950820923, + "step": 3237 + }, + { + "epoch": 1.8111888111888113, + "grad_norm": 0.46820046754004035, + "learning_rate": 8.146137156535625e-06, + "loss": 0.1833459585905075, + "step": 3238 + }, + { + "epoch": 1.8117482517482517, + "grad_norm": 0.4782335385278891, + "learning_rate": 8.139741992508928e-06, + "loss": 0.2868287265300751, + "step": 3239 + }, + { + "epoch": 1.8123076923076922, + "grad_norm": 0.4410149115498927, + "learning_rate": 8.13334761646789e-06, + "loss": 0.1585218608379364, + "step": 3240 + }, + { + "epoch": 1.8128671328671329, + "grad_norm": 0.46237586208732967, + "learning_rate": 8.126954031121103e-06, + "loss": 0.12568265199661255, + "step": 3241 + }, + { + "epoch": 1.8134265734265735, + "grad_norm": 0.3501406980218606, + "learning_rate": 8.120561239176824e-06, + "loss": 0.13871999084949493, + "step": 3242 + }, + { + "epoch": 1.813986013986014, + "grad_norm": 0.16819193352979114, + "learning_rate": 8.114169243342968e-06, + "loss": 0.019778726622462273, + "step": 3243 + }, + { + "epoch": 1.8145454545454545, + "grad_norm": 0.3584488606879959, + "learning_rate": 8.107778046327126e-06, + "loss": 0.07132961601018906, + "step": 3244 + }, + { + "epoch": 1.8151048951048951, + "grad_norm": 0.6940886158841496, + "learning_rate": 8.101387650836531e-06, + "loss": 0.3750250041484833, + "step": 3245 + }, + { + "epoch": 1.8156643356643358, + "grad_norm": 0.5100916667991948, + "learning_rate": 8.094998059578096e-06, + "loss": 0.18472930788993835, + "step": 3246 + }, + { + "epoch": 1.8162237762237763, + "grad_norm": 0.43788693626976416, + "learning_rate": 8.088609275258381e-06, + "loss": 0.1066703274846077, + "step": 3247 + }, + { + "epoch": 1.8167832167832167, + "grad_norm": 0.5456859118511143, + "learning_rate": 8.082221300583603e-06, + "loss": 0.2648044228553772, + "step": 3248 + }, + { + "epoch": 1.8173426573426572, + "grad_norm": 0.3353047492779175, + "learning_rate": 8.075834138259648e-06, + "loss": 0.06476524472236633, + "step": 3249 + }, + { + "epoch": 1.817902097902098, + "grad_norm": 0.536079725566275, + "learning_rate": 8.069447790992044e-06, + "loss": 0.2636910378932953, + "step": 3250 + }, + { + "epoch": 1.8184615384615386, + "grad_norm": 0.42580179010429037, + "learning_rate": 8.063062261485988e-06, + "loss": 0.15321972966194153, + "step": 3251 + }, + { + "epoch": 1.819020979020979, + "grad_norm": 0.36864649479171974, + "learning_rate": 8.056677552446317e-06, + "loss": 0.05630672723054886, + "step": 3252 + }, + { + "epoch": 1.8195804195804195, + "grad_norm": 0.42288788148517836, + "learning_rate": 8.050293666577527e-06, + "loss": 0.09161169826984406, + "step": 3253 + }, + { + "epoch": 1.8201398601398602, + "grad_norm": 0.41612798306658905, + "learning_rate": 8.043910606583767e-06, + "loss": 0.0998765230178833, + "step": 3254 + }, + { + "epoch": 1.8206993006993009, + "grad_norm": 0.6568713027277304, + "learning_rate": 8.037528375168828e-06, + "loss": 0.32714977860450745, + "step": 3255 + }, + { + "epoch": 1.8212587412587413, + "grad_norm": 0.4431579642548325, + "learning_rate": 8.031146975036164e-06, + "loss": 0.13106168806552887, + "step": 3256 + }, + { + "epoch": 1.8218181818181818, + "grad_norm": 0.427112685691691, + "learning_rate": 8.024766408888867e-06, + "loss": 0.15179291367530823, + "step": 3257 + }, + { + "epoch": 1.8223776223776222, + "grad_norm": 0.25217920673519184, + "learning_rate": 8.018386679429673e-06, + "loss": 0.023837648332118988, + "step": 3258 + }, + { + "epoch": 1.822937062937063, + "grad_norm": 0.4134665701873286, + "learning_rate": 8.012007789360972e-06, + "loss": 0.2113943099975586, + "step": 3259 + }, + { + "epoch": 1.8234965034965036, + "grad_norm": 0.7485489594800963, + "learning_rate": 8.00562974138479e-06, + "loss": 0.3009991943836212, + "step": 3260 + }, + { + "epoch": 1.824055944055944, + "grad_norm": 0.3672714182684478, + "learning_rate": 7.999252538202807e-06, + "loss": 0.08428531885147095, + "step": 3261 + }, + { + "epoch": 1.8246153846153845, + "grad_norm": 0.5088186798455904, + "learning_rate": 7.992876182516339e-06, + "loss": 0.11041277647018433, + "step": 3262 + }, + { + "epoch": 1.8251748251748252, + "grad_norm": 0.42668297868506816, + "learning_rate": 7.986500677026334e-06, + "loss": 0.17227564752101898, + "step": 3263 + }, + { + "epoch": 1.825734265734266, + "grad_norm": 0.482484645348215, + "learning_rate": 7.980126024433398e-06, + "loss": 0.19148889183998108, + "step": 3264 + }, + { + "epoch": 1.8262937062937064, + "grad_norm": 0.46725950802667315, + "learning_rate": 7.97375222743776e-06, + "loss": 0.1837417185306549, + "step": 3265 + }, + { + "epoch": 1.8268531468531468, + "grad_norm": 0.5179184252537476, + "learning_rate": 7.9673792887393e-06, + "loss": 0.2764427065849304, + "step": 3266 + }, + { + "epoch": 1.8274125874125873, + "grad_norm": 0.44923373328383853, + "learning_rate": 7.961007211037523e-06, + "loss": 0.1576852649450302, + "step": 3267 + }, + { + "epoch": 1.827972027972028, + "grad_norm": 0.2999779958490929, + "learning_rate": 7.954635997031577e-06, + "loss": 0.06918426603078842, + "step": 3268 + }, + { + "epoch": 1.8285314685314686, + "grad_norm": 0.3902470104587885, + "learning_rate": 7.94826564942024e-06, + "loss": 0.11541706323623657, + "step": 3269 + }, + { + "epoch": 1.829090909090909, + "grad_norm": 0.4705275536787707, + "learning_rate": 7.941896170901916e-06, + "loss": 0.22836562991142273, + "step": 3270 + }, + { + "epoch": 1.8296503496503496, + "grad_norm": 0.4421100375331922, + "learning_rate": 7.935527564174664e-06, + "loss": 0.1463472694158554, + "step": 3271 + }, + { + "epoch": 1.8302097902097902, + "grad_norm": 0.530909236014164, + "learning_rate": 7.929159831936143e-06, + "loss": 0.19594229757785797, + "step": 3272 + }, + { + "epoch": 1.830769230769231, + "grad_norm": 0.3622198517034719, + "learning_rate": 7.92279297688367e-06, + "loss": 0.08977712690830231, + "step": 3273 + }, + { + "epoch": 1.8313286713286714, + "grad_norm": 0.3022253004757961, + "learning_rate": 7.916427001714173e-06, + "loss": 0.10286234319210052, + "step": 3274 + }, + { + "epoch": 1.8318881118881118, + "grad_norm": 0.5231127038275094, + "learning_rate": 7.910061909124209e-06, + "loss": 0.1839471161365509, + "step": 3275 + }, + { + "epoch": 1.8324475524475523, + "grad_norm": 0.4254742382775992, + "learning_rate": 7.903697701809967e-06, + "loss": 0.132829949259758, + "step": 3276 + }, + { + "epoch": 1.833006993006993, + "grad_norm": 0.4204018933890499, + "learning_rate": 7.897334382467254e-06, + "loss": 0.13574588298797607, + "step": 3277 + }, + { + "epoch": 1.8335664335664337, + "grad_norm": 0.36498694334474635, + "learning_rate": 7.890971953791513e-06, + "loss": 0.07292518019676208, + "step": 3278 + }, + { + "epoch": 1.8341258741258741, + "grad_norm": 0.44580898194594953, + "learning_rate": 7.884610418477796e-06, + "loss": 0.11348556727170944, + "step": 3279 + }, + { + "epoch": 1.8346853146853146, + "grad_norm": 0.3995039845184179, + "learning_rate": 7.878249779220781e-06, + "loss": 0.1410672664642334, + "step": 3280 + }, + { + "epoch": 1.8352447552447553, + "grad_norm": 0.20567938601338534, + "learning_rate": 7.871890038714772e-06, + "loss": 0.038923539221286774, + "step": 3281 + }, + { + "epoch": 1.835804195804196, + "grad_norm": 0.45622902619008837, + "learning_rate": 7.865531199653681e-06, + "loss": 0.11697768419981003, + "step": 3282 + }, + { + "epoch": 1.8363636363636364, + "grad_norm": 0.39911538086316706, + "learning_rate": 7.859173264731058e-06, + "loss": 0.08933316171169281, + "step": 3283 + }, + { + "epoch": 1.8369230769230769, + "grad_norm": 0.6380580636902266, + "learning_rate": 7.852816236640042e-06, + "loss": 0.17128747701644897, + "step": 3284 + }, + { + "epoch": 1.8374825174825173, + "grad_norm": 0.5141324447736442, + "learning_rate": 7.846460118073414e-06, + "loss": 0.23782578110694885, + "step": 3285 + }, + { + "epoch": 1.838041958041958, + "grad_norm": 0.3185246075777209, + "learning_rate": 7.840104911723556e-06, + "loss": 0.06162901595234871, + "step": 3286 + }, + { + "epoch": 1.8386013986013987, + "grad_norm": 0.48233579519798925, + "learning_rate": 7.833750620282459e-06, + "loss": 0.14769190549850464, + "step": 3287 + }, + { + "epoch": 1.8391608391608392, + "grad_norm": 0.40269608863029693, + "learning_rate": 7.827397246441747e-06, + "loss": 0.09462253004312515, + "step": 3288 + }, + { + "epoch": 1.8397202797202796, + "grad_norm": 0.8225918385407823, + "learning_rate": 7.82104479289263e-06, + "loss": 0.36728331446647644, + "step": 3289 + }, + { + "epoch": 1.8402797202797203, + "grad_norm": 0.5711909692873306, + "learning_rate": 7.814693262325949e-06, + "loss": 0.13168132305145264, + "step": 3290 + }, + { + "epoch": 1.840839160839161, + "grad_norm": 0.38509219934510297, + "learning_rate": 7.808342657432141e-06, + "loss": 0.11074627935886383, + "step": 3291 + }, + { + "epoch": 1.8413986013986015, + "grad_norm": 0.5526198193138044, + "learning_rate": 7.801992980901252e-06, + "loss": 0.2530818581581116, + "step": 3292 + }, + { + "epoch": 1.841958041958042, + "grad_norm": 0.5156533927185056, + "learning_rate": 7.795644235422946e-06, + "loss": 0.38504558801651, + "step": 3293 + }, + { + "epoch": 1.8425174825174824, + "grad_norm": 0.37639292439234967, + "learning_rate": 7.789296423686479e-06, + "loss": 0.0960349515080452, + "step": 3294 + }, + { + "epoch": 1.843076923076923, + "grad_norm": 0.2889852334095706, + "learning_rate": 7.78294954838072e-06, + "loss": 0.04795685410499573, + "step": 3295 + }, + { + "epoch": 1.8436363636363637, + "grad_norm": 0.41028913266791883, + "learning_rate": 7.776603612194138e-06, + "loss": 0.17522096633911133, + "step": 3296 + }, + { + "epoch": 1.8441958041958042, + "grad_norm": 0.4688820661332411, + "learning_rate": 7.7702586178148e-06, + "loss": 0.10706233978271484, + "step": 3297 + }, + { + "epoch": 1.8447552447552447, + "grad_norm": 0.4402681769274856, + "learning_rate": 7.763914567930384e-06, + "loss": 0.20432394742965698, + "step": 3298 + }, + { + "epoch": 1.8453146853146853, + "grad_norm": 0.298265048833217, + "learning_rate": 7.757571465228158e-06, + "loss": 0.09755592048168182, + "step": 3299 + }, + { + "epoch": 1.845874125874126, + "grad_norm": 0.30550017106759586, + "learning_rate": 7.751229312395e-06, + "loss": 0.06318570673465729, + "step": 3300 + }, + { + "epoch": 1.8464335664335665, + "grad_norm": 0.40667948693971684, + "learning_rate": 7.744888112117372e-06, + "loss": 0.17142446339130402, + "step": 3301 + }, + { + "epoch": 1.846993006993007, + "grad_norm": 0.33096023137403674, + "learning_rate": 7.738547867081347e-06, + "loss": 0.04919543117284775, + "step": 3302 + }, + { + "epoch": 1.8475524475524474, + "grad_norm": 0.3164644256584577, + "learning_rate": 7.73220857997258e-06, + "loss": 0.05155760422348976, + "step": 3303 + }, + { + "epoch": 1.848111888111888, + "grad_norm": 0.4942894098995378, + "learning_rate": 7.725870253476323e-06, + "loss": 0.25951826572418213, + "step": 3304 + }, + { + "epoch": 1.8486713286713288, + "grad_norm": 0.5638069964049093, + "learning_rate": 7.719532890277436e-06, + "loss": 0.21192558109760284, + "step": 3305 + }, + { + "epoch": 1.8492307692307692, + "grad_norm": 0.30287132591323057, + "learning_rate": 7.713196493060347e-06, + "loss": 0.05738449841737747, + "step": 3306 + }, + { + "epoch": 1.8497902097902097, + "grad_norm": 0.20718712633768657, + "learning_rate": 7.706861064509095e-06, + "loss": 0.019613808020949364, + "step": 3307 + }, + { + "epoch": 1.8503496503496504, + "grad_norm": 0.348863984844842, + "learning_rate": 7.700526607307298e-06, + "loss": 0.08770640194416046, + "step": 3308 + }, + { + "epoch": 1.850909090909091, + "grad_norm": 0.401511301494257, + "learning_rate": 7.694193124138161e-06, + "loss": 0.10374884307384491, + "step": 3309 + }, + { + "epoch": 1.8514685314685315, + "grad_norm": 0.5108373161622242, + "learning_rate": 7.68786061768449e-06, + "loss": 0.16725939512252808, + "step": 3310 + }, + { + "epoch": 1.852027972027972, + "grad_norm": 0.4799703301173538, + "learning_rate": 7.681529090628658e-06, + "loss": 0.12029227614402771, + "step": 3311 + }, + { + "epoch": 1.8525874125874124, + "grad_norm": 0.3862506677095787, + "learning_rate": 7.675198545652643e-06, + "loss": 0.14128750562667847, + "step": 3312 + }, + { + "epoch": 1.8531468531468531, + "grad_norm": 0.2170827701236566, + "learning_rate": 7.668868985437988e-06, + "loss": 0.03516771271824837, + "step": 3313 + }, + { + "epoch": 1.8537062937062938, + "grad_norm": 0.37947615122458606, + "learning_rate": 7.66254041266583e-06, + "loss": 0.08889401704072952, + "step": 3314 + }, + { + "epoch": 1.8542657342657343, + "grad_norm": 0.3061808498785176, + "learning_rate": 7.65621283001689e-06, + "loss": 0.07184748351573944, + "step": 3315 + }, + { + "epoch": 1.8548251748251747, + "grad_norm": 0.47827635738884394, + "learning_rate": 7.649886240171462e-06, + "loss": 0.18813163042068481, + "step": 3316 + }, + { + "epoch": 1.8553846153846154, + "grad_norm": 0.536672767583284, + "learning_rate": 7.643560645809423e-06, + "loss": 0.20797763764858246, + "step": 3317 + }, + { + "epoch": 1.855944055944056, + "grad_norm": 0.3684628636146196, + "learning_rate": 7.637236049610229e-06, + "loss": 0.0401025116443634, + "step": 3318 + }, + { + "epoch": 1.8565034965034966, + "grad_norm": 0.2398946479299813, + "learning_rate": 7.630912454252909e-06, + "loss": 0.04069935530424118, + "step": 3319 + }, + { + "epoch": 1.857062937062937, + "grad_norm": 0.48885912382711566, + "learning_rate": 7.6245898624160775e-06, + "loss": 0.23037636280059814, + "step": 3320 + }, + { + "epoch": 1.8576223776223775, + "grad_norm": 0.6937363577084533, + "learning_rate": 7.618268276777909e-06, + "loss": 0.29209572076797485, + "step": 3321 + }, + { + "epoch": 1.8581818181818182, + "grad_norm": 0.4381572806948571, + "learning_rate": 7.611947700016169e-06, + "loss": 0.09035668522119522, + "step": 3322 + }, + { + "epoch": 1.8587412587412588, + "grad_norm": 0.5934394988468752, + "learning_rate": 7.6056281348081815e-06, + "loss": 0.4222186207771301, + "step": 3323 + }, + { + "epoch": 1.8593006993006993, + "grad_norm": 0.24684365144686526, + "learning_rate": 7.599309583830852e-06, + "loss": 0.05275161936879158, + "step": 3324 + }, + { + "epoch": 1.8598601398601398, + "grad_norm": 0.44465947066313655, + "learning_rate": 7.592992049760649e-06, + "loss": 0.1441236287355423, + "step": 3325 + }, + { + "epoch": 1.8604195804195804, + "grad_norm": 0.21883129351293812, + "learning_rate": 7.58667553527361e-06, + "loss": 0.02539297193288803, + "step": 3326 + }, + { + "epoch": 1.8609790209790211, + "grad_norm": 0.34038441998544866, + "learning_rate": 7.580360043045351e-06, + "loss": 0.09021177887916565, + "step": 3327 + }, + { + "epoch": 1.8615384615384616, + "grad_norm": 0.49332604855498424, + "learning_rate": 7.5740455757510434e-06, + "loss": 0.2206195592880249, + "step": 3328 + }, + { + "epoch": 1.862097902097902, + "grad_norm": 0.42452375206883225, + "learning_rate": 7.567732136065432e-06, + "loss": 0.11760076880455017, + "step": 3329 + }, + { + "epoch": 1.8626573426573425, + "grad_norm": 0.38263318819441805, + "learning_rate": 7.561419726662822e-06, + "loss": 0.06637430191040039, + "step": 3330 + }, + { + "epoch": 1.8632167832167832, + "grad_norm": 0.4634512551817729, + "learning_rate": 7.5551083502170775e-06, + "loss": 0.1258738934993744, + "step": 3331 + }, + { + "epoch": 1.8637762237762239, + "grad_norm": 0.2898291703548246, + "learning_rate": 7.548798009401641e-06, + "loss": 0.07161258161067963, + "step": 3332 + }, + { + "epoch": 1.8643356643356643, + "grad_norm": 0.326259879840752, + "learning_rate": 7.5424887068894995e-06, + "loss": 0.08686239272356033, + "step": 3333 + }, + { + "epoch": 1.8648951048951048, + "grad_norm": 0.37079336856522993, + "learning_rate": 7.53618044535321e-06, + "loss": 0.10367372632026672, + "step": 3334 + }, + { + "epoch": 1.8654545454545455, + "grad_norm": 0.515954965489109, + "learning_rate": 7.5298732274648865e-06, + "loss": 0.16587960720062256, + "step": 3335 + }, + { + "epoch": 1.8660139860139862, + "grad_norm": 0.49243930111957324, + "learning_rate": 7.523567055896193e-06, + "loss": 0.20577214658260345, + "step": 3336 + }, + { + "epoch": 1.8665734265734266, + "grad_norm": 0.2079357692441385, + "learning_rate": 7.517261933318365e-06, + "loss": 0.019247375428676605, + "step": 3337 + }, + { + "epoch": 1.867132867132867, + "grad_norm": 0.33148412336913485, + "learning_rate": 7.510957862402182e-06, + "loss": 0.08310063928365707, + "step": 3338 + }, + { + "epoch": 1.8676923076923075, + "grad_norm": 0.3811485400135021, + "learning_rate": 7.504654845817985e-06, + "loss": 0.07640981674194336, + "step": 3339 + }, + { + "epoch": 1.8682517482517482, + "grad_norm": 0.35959408485277833, + "learning_rate": 7.498352886235658e-06, + "loss": 0.12265245616436005, + "step": 3340 + }, + { + "epoch": 1.868811188811189, + "grad_norm": 0.7711929731889879, + "learning_rate": 7.492051986324656e-06, + "loss": 0.2052098512649536, + "step": 3341 + }, + { + "epoch": 1.8693706293706294, + "grad_norm": 0.3782416755128335, + "learning_rate": 7.485752148753966e-06, + "loss": 0.07031114399433136, + "step": 3342 + }, + { + "epoch": 1.8699300699300698, + "grad_norm": 0.4379561876127689, + "learning_rate": 7.479453376192134e-06, + "loss": 0.09227945655584335, + "step": 3343 + }, + { + "epoch": 1.8704895104895105, + "grad_norm": 0.3422110452768986, + "learning_rate": 7.473155671307254e-06, + "loss": 0.20535951852798462, + "step": 3344 + }, + { + "epoch": 1.8710489510489512, + "grad_norm": 0.23144604861995824, + "learning_rate": 7.4668590367669634e-06, + "loss": 0.02173934504389763, + "step": 3345 + }, + { + "epoch": 1.8716083916083917, + "grad_norm": 0.4091947278273256, + "learning_rate": 7.46056347523846e-06, + "loss": 0.1195792481303215, + "step": 3346 + }, + { + "epoch": 1.8721678321678321, + "grad_norm": 0.4713074496848786, + "learning_rate": 7.454268989388471e-06, + "loss": 0.1426640897989273, + "step": 3347 + }, + { + "epoch": 1.8727272727272726, + "grad_norm": 0.42829616560020345, + "learning_rate": 7.447975581883269e-06, + "loss": 0.08933590352535248, + "step": 3348 + }, + { + "epoch": 1.8732867132867133, + "grad_norm": 0.3388373133200613, + "learning_rate": 7.441683255388683e-06, + "loss": 0.03875339776277542, + "step": 3349 + }, + { + "epoch": 1.873846153846154, + "grad_norm": 0.1412349186115566, + "learning_rate": 7.435392012570073e-06, + "loss": 0.0318184420466423, + "step": 3350 + }, + { + "epoch": 1.8744055944055944, + "grad_norm": 0.35128745381710996, + "learning_rate": 7.429101856092346e-06, + "loss": 0.05885808914899826, + "step": 3351 + }, + { + "epoch": 1.8749650349650349, + "grad_norm": 0.586236177563801, + "learning_rate": 7.422812788619942e-06, + "loss": 0.32151180505752563, + "step": 3352 + }, + { + "epoch": 1.8755244755244755, + "grad_norm": 0.5421896994740016, + "learning_rate": 7.416524812816842e-06, + "loss": 0.14148828387260437, + "step": 3353 + }, + { + "epoch": 1.8760839160839162, + "grad_norm": 0.4062343948852509, + "learning_rate": 7.410237931346573e-06, + "loss": 0.08353282511234283, + "step": 3354 + }, + { + "epoch": 1.8766433566433567, + "grad_norm": 0.3544528290649114, + "learning_rate": 7.403952146872186e-06, + "loss": 0.11531280726194382, + "step": 3355 + }, + { + "epoch": 1.8772027972027971, + "grad_norm": 0.5082289104676715, + "learning_rate": 7.397667462056277e-06, + "loss": 0.248196542263031, + "step": 3356 + }, + { + "epoch": 1.8777622377622376, + "grad_norm": 0.5280852575718945, + "learning_rate": 7.391383879560965e-06, + "loss": 0.1805993616580963, + "step": 3357 + }, + { + "epoch": 1.8783216783216783, + "grad_norm": 0.40012098573047417, + "learning_rate": 7.3851014020479205e-06, + "loss": 0.12735983729362488, + "step": 3358 + }, + { + "epoch": 1.878881118881119, + "grad_norm": 0.309334003565052, + "learning_rate": 7.378820032178327e-06, + "loss": 0.04709480702877045, + "step": 3359 + }, + { + "epoch": 1.8794405594405594, + "grad_norm": 0.441829834510686, + "learning_rate": 7.372539772612905e-06, + "loss": 0.1560848355293274, + "step": 3360 + }, + { + "epoch": 1.88, + "grad_norm": 0.5554556173952896, + "learning_rate": 7.366260626011913e-06, + "loss": 0.06077303737401962, + "step": 3361 + }, + { + "epoch": 1.8805594405594406, + "grad_norm": 0.3734705827831097, + "learning_rate": 7.3599825950351226e-06, + "loss": 0.07205686718225479, + "step": 3362 + }, + { + "epoch": 1.8811188811188813, + "grad_norm": 0.47808679486523853, + "learning_rate": 7.3537056823418494e-06, + "loss": 0.16174761950969696, + "step": 3363 + }, + { + "epoch": 1.8816783216783217, + "grad_norm": 0.44367926084224263, + "learning_rate": 7.347429890590926e-06, + "loss": 0.22913508117198944, + "step": 3364 + }, + { + "epoch": 1.8822377622377622, + "grad_norm": 0.47584292646540743, + "learning_rate": 7.341155222440707e-06, + "loss": 0.22197626531124115, + "step": 3365 + }, + { + "epoch": 1.8827972027972026, + "grad_norm": 0.3831930783025013, + "learning_rate": 7.334881680549079e-06, + "loss": 0.09171321988105774, + "step": 3366 + }, + { + "epoch": 1.8833566433566433, + "grad_norm": 0.5978427751995224, + "learning_rate": 7.328609267573446e-06, + "loss": 0.23408979177474976, + "step": 3367 + }, + { + "epoch": 1.883916083916084, + "grad_norm": 0.3990589236985843, + "learning_rate": 7.32233798617074e-06, + "loss": 0.24631021916866302, + "step": 3368 + }, + { + "epoch": 1.8844755244755245, + "grad_norm": 0.4076239937951212, + "learning_rate": 7.316067838997409e-06, + "loss": 0.14442695677280426, + "step": 3369 + }, + { + "epoch": 1.885034965034965, + "grad_norm": 0.34914877316210574, + "learning_rate": 7.309798828709413e-06, + "loss": 0.08294790983200073, + "step": 3370 + }, + { + "epoch": 1.8855944055944056, + "grad_norm": 0.2581241110104319, + "learning_rate": 7.303530957962251e-06, + "loss": 0.06489971280097961, + "step": 3371 + }, + { + "epoch": 1.8861538461538463, + "grad_norm": 0.2242835894146753, + "learning_rate": 7.297264229410918e-06, + "loss": 0.027363475412130356, + "step": 3372 + }, + { + "epoch": 1.8867132867132868, + "grad_norm": 0.3614093672513138, + "learning_rate": 7.290998645709938e-06, + "loss": 0.09044063091278076, + "step": 3373 + }, + { + "epoch": 1.8872727272727272, + "grad_norm": 0.3194825562410812, + "learning_rate": 7.284734209513344e-06, + "loss": 0.09621086716651917, + "step": 3374 + }, + { + "epoch": 1.8878321678321677, + "grad_norm": 0.5489528232691873, + "learning_rate": 7.278470923474685e-06, + "loss": 0.23984023928642273, + "step": 3375 + }, + { + "epoch": 1.8883916083916084, + "grad_norm": 0.3395984139390064, + "learning_rate": 7.2722087902470265e-06, + "loss": 0.05971367657184601, + "step": 3376 + }, + { + "epoch": 1.888951048951049, + "grad_norm": 0.37759419314801435, + "learning_rate": 7.265947812482936e-06, + "loss": 0.04140233248472214, + "step": 3377 + }, + { + "epoch": 1.8895104895104895, + "grad_norm": 0.4216047850104568, + "learning_rate": 7.259687992834505e-06, + "loss": 0.12734784185886383, + "step": 3378 + }, + { + "epoch": 1.89006993006993, + "grad_norm": 0.48657244247450854, + "learning_rate": 7.253429333953319e-06, + "loss": 0.14470745623111725, + "step": 3379 + }, + { + "epoch": 1.8906293706293706, + "grad_norm": 0.5589158707334188, + "learning_rate": 7.2471718384904875e-06, + "loss": 0.1993153691291809, + "step": 3380 + }, + { + "epoch": 1.8911888111888113, + "grad_norm": 0.6140280265999553, + "learning_rate": 7.2409155090966175e-06, + "loss": 0.236606627702713, + "step": 3381 + }, + { + "epoch": 1.8917482517482518, + "grad_norm": 0.48288061901907525, + "learning_rate": 7.234660348421819e-06, + "loss": 0.17082786560058594, + "step": 3382 + }, + { + "epoch": 1.8923076923076922, + "grad_norm": 0.5961960338773331, + "learning_rate": 7.228406359115718e-06, + "loss": 0.14957362413406372, + "step": 3383 + }, + { + "epoch": 1.8928671328671327, + "grad_norm": 0.5688896319122609, + "learning_rate": 7.222153543827432e-06, + "loss": 0.20060384273529053, + "step": 3384 + }, + { + "epoch": 1.8934265734265734, + "grad_norm": 0.5224023369822317, + "learning_rate": 7.215901905205598e-06, + "loss": 0.2028248906135559, + "step": 3385 + }, + { + "epoch": 1.893986013986014, + "grad_norm": 0.408519602655277, + "learning_rate": 7.209651445898336e-06, + "loss": 0.11954338848590851, + "step": 3386 + }, + { + "epoch": 1.8945454545454545, + "grad_norm": 0.5004557064824309, + "learning_rate": 7.203402168553275e-06, + "loss": 0.24602608382701874, + "step": 3387 + }, + { + "epoch": 1.895104895104895, + "grad_norm": 0.361247699801896, + "learning_rate": 7.197154075817545e-06, + "loss": 0.1610192060470581, + "step": 3388 + }, + { + "epoch": 1.8956643356643357, + "grad_norm": 0.27567305260883557, + "learning_rate": 7.190907170337768e-06, + "loss": 0.0561186745762825, + "step": 3389 + }, + { + "epoch": 1.8962237762237764, + "grad_norm": 0.383397589780328, + "learning_rate": 7.184661454760076e-06, + "loss": 0.18576236069202423, + "step": 3390 + }, + { + "epoch": 1.8967832167832168, + "grad_norm": 0.6721800852341708, + "learning_rate": 7.1784169317300825e-06, + "loss": 0.22534814476966858, + "step": 3391 + }, + { + "epoch": 1.8973426573426573, + "grad_norm": 0.3448389199829432, + "learning_rate": 7.1721736038929e-06, + "loss": 0.11819382756948471, + "step": 3392 + }, + { + "epoch": 1.8979020979020977, + "grad_norm": 0.3584149536826586, + "learning_rate": 7.16593147389314e-06, + "loss": 0.13060441613197327, + "step": 3393 + }, + { + "epoch": 1.8984615384615384, + "grad_norm": 0.37007567137850383, + "learning_rate": 7.1596905443748975e-06, + "loss": 0.1372489631175995, + "step": 3394 + }, + { + "epoch": 1.899020979020979, + "grad_norm": 0.45450240815101617, + "learning_rate": 7.153450817981774e-06, + "loss": 0.1455768495798111, + "step": 3395 + }, + { + "epoch": 1.8995804195804196, + "grad_norm": 0.4773749178317382, + "learning_rate": 7.1472122973568406e-06, + "loss": 0.1744433343410492, + "step": 3396 + }, + { + "epoch": 1.90013986013986, + "grad_norm": 0.6195566785515306, + "learning_rate": 7.140974985142677e-06, + "loss": 0.2150195837020874, + "step": 3397 + }, + { + "epoch": 1.9006993006993007, + "grad_norm": 0.582489009244309, + "learning_rate": 7.13473888398134e-06, + "loss": 0.18725216388702393, + "step": 3398 + }, + { + "epoch": 1.9012587412587414, + "grad_norm": 0.4320421598127124, + "learning_rate": 7.128503996514375e-06, + "loss": 0.1420905441045761, + "step": 3399 + }, + { + "epoch": 1.9018181818181819, + "grad_norm": 0.2828450590423948, + "learning_rate": 7.122270325382818e-06, + "loss": 0.06339014321565628, + "step": 3400 + }, + { + "epoch": 1.9023776223776223, + "grad_norm": 0.5073341565048562, + "learning_rate": 7.116037873227178e-06, + "loss": 0.158181831240654, + "step": 3401 + }, + { + "epoch": 1.9029370629370628, + "grad_norm": 0.33134936303099993, + "learning_rate": 7.109806642687468e-06, + "loss": 0.10833267867565155, + "step": 3402 + }, + { + "epoch": 1.9034965034965035, + "grad_norm": 0.41852044465580285, + "learning_rate": 7.103576636403164e-06, + "loss": 0.09106975793838501, + "step": 3403 + }, + { + "epoch": 1.9040559440559441, + "grad_norm": 0.5030246276781317, + "learning_rate": 7.097347857013232e-06, + "loss": 0.1995009183883667, + "step": 3404 + }, + { + "epoch": 1.9046153846153846, + "grad_norm": 0.38373210738460023, + "learning_rate": 7.091120307156118e-06, + "loss": 0.1709326058626175, + "step": 3405 + }, + { + "epoch": 1.905174825174825, + "grad_norm": 0.2153877169144216, + "learning_rate": 7.084893989469743e-06, + "loss": 0.01894509792327881, + "step": 3406 + }, + { + "epoch": 1.9057342657342657, + "grad_norm": 0.452302328098909, + "learning_rate": 7.078668906591517e-06, + "loss": 0.07084296643733978, + "step": 3407 + }, + { + "epoch": 1.9062937062937064, + "grad_norm": 0.3313060257319791, + "learning_rate": 7.072445061158315e-06, + "loss": 0.03980420529842377, + "step": 3408 + }, + { + "epoch": 1.9068531468531469, + "grad_norm": 0.7160739501409119, + "learning_rate": 7.066222455806488e-06, + "loss": 0.22033479809761047, + "step": 3409 + }, + { + "epoch": 1.9074125874125873, + "grad_norm": 0.47937096805551455, + "learning_rate": 7.0600010931718755e-06, + "loss": 0.07911434769630432, + "step": 3410 + }, + { + "epoch": 1.9079720279720278, + "grad_norm": 0.27835308103596823, + "learning_rate": 7.053780975889769e-06, + "loss": 0.059310268610715866, + "step": 3411 + }, + { + "epoch": 1.9085314685314685, + "grad_norm": 0.596130248060901, + "learning_rate": 7.047562106594956e-06, + "loss": 0.10286440700292587, + "step": 3412 + }, + { + "epoch": 1.9090909090909092, + "grad_norm": 0.4220457183677214, + "learning_rate": 7.041344487921679e-06, + "loss": 0.18492405116558075, + "step": 3413 + }, + { + "epoch": 1.9096503496503496, + "grad_norm": 0.5118917792975136, + "learning_rate": 7.035128122503654e-06, + "loss": 0.20573991537094116, + "step": 3414 + }, + { + "epoch": 1.91020979020979, + "grad_norm": 0.3666235792762084, + "learning_rate": 7.0289130129740726e-06, + "loss": 0.08964357525110245, + "step": 3415 + }, + { + "epoch": 1.9107692307692308, + "grad_norm": 0.3730421052072422, + "learning_rate": 7.022699161965581e-06, + "loss": 0.0956602692604065, + "step": 3416 + }, + { + "epoch": 1.9113286713286715, + "grad_norm": 0.5136631650837208, + "learning_rate": 7.0164865721103104e-06, + "loss": 0.11252351850271225, + "step": 3417 + }, + { + "epoch": 1.911888111888112, + "grad_norm": 0.172447562242082, + "learning_rate": 7.010275246039843e-06, + "loss": 0.01893317699432373, + "step": 3418 + }, + { + "epoch": 1.9124475524475524, + "grad_norm": 0.5456018044385736, + "learning_rate": 7.004065186385236e-06, + "loss": 0.18156349658966064, + "step": 3419 + }, + { + "epoch": 1.913006993006993, + "grad_norm": 0.317578093849398, + "learning_rate": 6.997856395776999e-06, + "loss": 0.06971661001443863, + "step": 3420 + }, + { + "epoch": 1.9135664335664335, + "grad_norm": 0.34082035034755237, + "learning_rate": 6.991648876845112e-06, + "loss": 0.05821669101715088, + "step": 3421 + }, + { + "epoch": 1.9141258741258742, + "grad_norm": 0.3846081582898456, + "learning_rate": 6.985442632219017e-06, + "loss": 0.12792979180812836, + "step": 3422 + }, + { + "epoch": 1.9146853146853147, + "grad_norm": 0.4490738534353408, + "learning_rate": 6.979237664527609e-06, + "loss": 0.11260859668254852, + "step": 3423 + }, + { + "epoch": 1.9152447552447551, + "grad_norm": 0.5112939999217723, + "learning_rate": 6.973033976399253e-06, + "loss": 0.22189411520957947, + "step": 3424 + }, + { + "epoch": 1.9158041958041958, + "grad_norm": 0.4411144430435648, + "learning_rate": 6.9668315704617625e-06, + "loss": 0.12042347341775894, + "step": 3425 + }, + { + "epoch": 1.9163636363636365, + "grad_norm": 0.42890318697411994, + "learning_rate": 6.960630449342409e-06, + "loss": 0.08907759934663773, + "step": 3426 + }, + { + "epoch": 1.916923076923077, + "grad_norm": 0.38732606180612694, + "learning_rate": 6.954430615667927e-06, + "loss": 0.16345861554145813, + "step": 3427 + }, + { + "epoch": 1.9174825174825174, + "grad_norm": 0.4946368545373852, + "learning_rate": 6.948232072064493e-06, + "loss": 0.1780029535293579, + "step": 3428 + }, + { + "epoch": 1.918041958041958, + "grad_norm": 0.49128431665656763, + "learning_rate": 6.9420348211577546e-06, + "loss": 0.04524610564112663, + "step": 3429 + }, + { + "epoch": 1.9186013986013986, + "grad_norm": 0.5423301840484069, + "learning_rate": 6.935838865572797e-06, + "loss": 0.22067275643348694, + "step": 3430 + }, + { + "epoch": 1.9191608391608392, + "grad_norm": 0.24293293396813695, + "learning_rate": 6.929644207934158e-06, + "loss": 0.02736462652683258, + "step": 3431 + }, + { + "epoch": 1.9197202797202797, + "grad_norm": 0.3286712941970475, + "learning_rate": 6.923450850865836e-06, + "loss": 0.060130778700113297, + "step": 3432 + }, + { + "epoch": 1.9202797202797202, + "grad_norm": 0.4275367066860061, + "learning_rate": 6.917258796991264e-06, + "loss": 0.12045353651046753, + "step": 3433 + }, + { + "epoch": 1.9208391608391608, + "grad_norm": 0.5214629676825002, + "learning_rate": 6.91106804893334e-06, + "loss": 0.2129320502281189, + "step": 3434 + }, + { + "epoch": 1.9213986013986015, + "grad_norm": 0.5821914474937075, + "learning_rate": 6.904878609314392e-06, + "loss": 0.1366441249847412, + "step": 3435 + }, + { + "epoch": 1.921958041958042, + "grad_norm": 0.4718469740844754, + "learning_rate": 6.898690480756205e-06, + "loss": 0.10970372706651688, + "step": 3436 + }, + { + "epoch": 1.9225174825174824, + "grad_norm": 0.48554145751883593, + "learning_rate": 6.892503665880005e-06, + "loss": 0.18622155487537384, + "step": 3437 + }, + { + "epoch": 1.9230769230769231, + "grad_norm": 0.6368795290196623, + "learning_rate": 6.8863181673064565e-06, + "loss": 0.15056131780147552, + "step": 3438 + }, + { + "epoch": 1.9236363636363636, + "grad_norm": 0.47618412256557674, + "learning_rate": 6.880133987655678e-06, + "loss": 0.18443703651428223, + "step": 3439 + }, + { + "epoch": 1.9241958041958043, + "grad_norm": 0.5948904262600953, + "learning_rate": 6.873951129547219e-06, + "loss": 0.25488942861557007, + "step": 3440 + }, + { + "epoch": 1.9247552447552447, + "grad_norm": 0.4220321825229647, + "learning_rate": 6.867769595600076e-06, + "loss": 0.08821668475866318, + "step": 3441 + }, + { + "epoch": 1.9253146853146852, + "grad_norm": 0.3961564663566034, + "learning_rate": 6.861589388432679e-06, + "loss": 0.08856074512004852, + "step": 3442 + }, + { + "epoch": 1.9258741258741259, + "grad_norm": 0.3307317662447969, + "learning_rate": 6.8554105106628954e-06, + "loss": 0.0803074985742569, + "step": 3443 + }, + { + "epoch": 1.9264335664335666, + "grad_norm": 0.2796797936546954, + "learning_rate": 6.849232964908043e-06, + "loss": 0.03420411795377731, + "step": 3444 + }, + { + "epoch": 1.926993006993007, + "grad_norm": 0.4022698859688085, + "learning_rate": 6.843056753784851e-06, + "loss": 0.16885244846343994, + "step": 3445 + }, + { + "epoch": 1.9275524475524475, + "grad_norm": 0.569953340529058, + "learning_rate": 6.836881879909509e-06, + "loss": 0.2108049839735031, + "step": 3446 + }, + { + "epoch": 1.9281118881118882, + "grad_norm": 0.42057617880127945, + "learning_rate": 6.830708345897625e-06, + "loss": 0.1319536566734314, + "step": 3447 + }, + { + "epoch": 1.9286713286713286, + "grad_norm": 0.4477402208277191, + "learning_rate": 6.824536154364239e-06, + "loss": 0.09764501452445984, + "step": 3448 + }, + { + "epoch": 1.9292307692307693, + "grad_norm": 0.3978422459043464, + "learning_rate": 6.818365307923831e-06, + "loss": 0.1413380354642868, + "step": 3449 + }, + { + "epoch": 1.9297902097902098, + "grad_norm": 0.33284562474847557, + "learning_rate": 6.8121958091903005e-06, + "loss": 0.05311393737792969, + "step": 3450 + }, + { + "epoch": 1.9303496503496502, + "grad_norm": 0.3463095595450734, + "learning_rate": 6.806027660776991e-06, + "loss": 0.19735713303089142, + "step": 3451 + }, + { + "epoch": 1.930909090909091, + "grad_norm": 0.3360851795688373, + "learning_rate": 6.799860865296657e-06, + "loss": 0.11381424218416214, + "step": 3452 + }, + { + "epoch": 1.9314685314685316, + "grad_norm": 0.45637273566225567, + "learning_rate": 6.793695425361493e-06, + "loss": 0.17068126797676086, + "step": 3453 + }, + { + "epoch": 1.932027972027972, + "grad_norm": 0.20298354695509957, + "learning_rate": 6.787531343583113e-06, + "loss": 0.017199130728840828, + "step": 3454 + }, + { + "epoch": 1.9325874125874125, + "grad_norm": 0.3429540265388206, + "learning_rate": 6.781368622572553e-06, + "loss": 0.08450612425804138, + "step": 3455 + }, + { + "epoch": 1.9331468531468532, + "grad_norm": 0.3430425369228152, + "learning_rate": 6.775207264940282e-06, + "loss": 0.04806827753782272, + "step": 3456 + }, + { + "epoch": 1.9337062937062937, + "grad_norm": 0.3862109117464254, + "learning_rate": 6.7690472732961835e-06, + "loss": 0.07865659892559052, + "step": 3457 + }, + { + "epoch": 1.9342657342657343, + "grad_norm": 0.5605965600239812, + "learning_rate": 6.7628886502495674e-06, + "loss": 0.31509312987327576, + "step": 3458 + }, + { + "epoch": 1.9348251748251748, + "grad_norm": 0.5565691627597196, + "learning_rate": 6.75673139840916e-06, + "loss": 0.16801634430885315, + "step": 3459 + }, + { + "epoch": 1.9353846153846153, + "grad_norm": 0.554709430249581, + "learning_rate": 6.750575520383102e-06, + "loss": 0.24640288949012756, + "step": 3460 + }, + { + "epoch": 1.935944055944056, + "grad_norm": 0.5773779072705727, + "learning_rate": 6.74442101877897e-06, + "loss": 0.20307449996471405, + "step": 3461 + }, + { + "epoch": 1.9365034965034966, + "grad_norm": 0.4424386907450472, + "learning_rate": 6.738267896203736e-06, + "loss": 0.15582285821437836, + "step": 3462 + }, + { + "epoch": 1.937062937062937, + "grad_norm": 0.5251610665940847, + "learning_rate": 6.732116155263806e-06, + "loss": 0.2522038519382477, + "step": 3463 + }, + { + "epoch": 1.9376223776223775, + "grad_norm": 0.47571426086568613, + "learning_rate": 6.725965798564987e-06, + "loss": 0.21848735213279724, + "step": 3464 + }, + { + "epoch": 1.9381818181818182, + "grad_norm": 0.3940604360406169, + "learning_rate": 6.719816828712505e-06, + "loss": 0.09598897397518158, + "step": 3465 + }, + { + "epoch": 1.9387412587412587, + "grad_norm": 0.44761139080623724, + "learning_rate": 6.713669248311001e-06, + "loss": 0.17435981333255768, + "step": 3466 + }, + { + "epoch": 1.9393006993006994, + "grad_norm": 0.5605166947206672, + "learning_rate": 6.7075230599645245e-06, + "loss": 0.19949239492416382, + "step": 3467 + }, + { + "epoch": 1.9398601398601398, + "grad_norm": 0.4343410275945147, + "learning_rate": 6.701378266276537e-06, + "loss": 0.12844239175319672, + "step": 3468 + }, + { + "epoch": 1.9404195804195803, + "grad_norm": 0.4861169011893989, + "learning_rate": 6.695234869849908e-06, + "loss": 0.18290919065475464, + "step": 3469 + }, + { + "epoch": 1.940979020979021, + "grad_norm": 0.5243028435545147, + "learning_rate": 6.689092873286913e-06, + "loss": 0.22710451483726501, + "step": 3470 + }, + { + "epoch": 1.9415384615384617, + "grad_norm": 0.6864833193139069, + "learning_rate": 6.6829522791892395e-06, + "loss": 0.2175981104373932, + "step": 3471 + }, + { + "epoch": 1.9420979020979021, + "grad_norm": 0.5770062913715226, + "learning_rate": 6.676813090157974e-06, + "loss": 0.18920888006687164, + "step": 3472 + }, + { + "epoch": 1.9426573426573426, + "grad_norm": 0.5618485636863273, + "learning_rate": 6.67067530879362e-06, + "loss": 0.22502082586288452, + "step": 3473 + }, + { + "epoch": 1.9432167832167833, + "grad_norm": 1.373785013801953, + "learning_rate": 6.664538937696071e-06, + "loss": 0.23475733399391174, + "step": 3474 + }, + { + "epoch": 1.9437762237762237, + "grad_norm": 0.49309314175587815, + "learning_rate": 6.658403979464631e-06, + "loss": 0.10718761384487152, + "step": 3475 + }, + { + "epoch": 1.9443356643356644, + "grad_norm": 0.4178982433004743, + "learning_rate": 6.652270436698005e-06, + "loss": 0.20793825387954712, + "step": 3476 + }, + { + "epoch": 1.9448951048951049, + "grad_norm": 0.4072587079086907, + "learning_rate": 6.646138311994292e-06, + "loss": 0.1298268437385559, + "step": 3477 + }, + { + "epoch": 1.9454545454545453, + "grad_norm": 0.6259130102311717, + "learning_rate": 6.6400076079510004e-06, + "loss": 0.21758905053138733, + "step": 3478 + }, + { + "epoch": 1.946013986013986, + "grad_norm": 0.39740766647546505, + "learning_rate": 6.633878327165028e-06, + "loss": 0.1582922339439392, + "step": 3479 + }, + { + "epoch": 1.9465734265734267, + "grad_norm": 0.8164435779515525, + "learning_rate": 6.627750472232679e-06, + "loss": 0.3181908428668976, + "step": 3480 + }, + { + "epoch": 1.9471328671328672, + "grad_norm": 0.4475078497936409, + "learning_rate": 6.621624045749643e-06, + "loss": 0.19448833167552948, + "step": 3481 + }, + { + "epoch": 1.9476923076923076, + "grad_norm": 0.6871443520874346, + "learning_rate": 6.615499050311007e-06, + "loss": 0.21107375621795654, + "step": 3482 + }, + { + "epoch": 1.9482517482517483, + "grad_norm": 0.2604102280161273, + "learning_rate": 6.609375488511263e-06, + "loss": 0.05676122009754181, + "step": 3483 + }, + { + "epoch": 1.948811188811189, + "grad_norm": 0.5051333849657156, + "learning_rate": 6.603253362944281e-06, + "loss": 0.21985793113708496, + "step": 3484 + }, + { + "epoch": 1.9493706293706294, + "grad_norm": 0.5348030491973162, + "learning_rate": 6.597132676203331e-06, + "loss": 0.36191850900650024, + "step": 3485 + }, + { + "epoch": 1.94993006993007, + "grad_norm": 0.35310549189993473, + "learning_rate": 6.5910134308810725e-06, + "loss": 0.05745324864983559, + "step": 3486 + }, + { + "epoch": 1.9504895104895104, + "grad_norm": 0.45993973774369756, + "learning_rate": 6.584895629569545e-06, + "loss": 0.17055457830429077, + "step": 3487 + }, + { + "epoch": 1.951048951048951, + "grad_norm": 0.45290462804006537, + "learning_rate": 6.578779274860197e-06, + "loss": 0.1341659277677536, + "step": 3488 + }, + { + "epoch": 1.9516083916083917, + "grad_norm": 0.4223534785425047, + "learning_rate": 6.572664369343844e-06, + "loss": 0.10700860619544983, + "step": 3489 + }, + { + "epoch": 1.9521678321678322, + "grad_norm": 0.4867015521264825, + "learning_rate": 6.566550915610699e-06, + "loss": 0.14823457598686218, + "step": 3490 + }, + { + "epoch": 1.9527272727272726, + "grad_norm": 0.48310510376746085, + "learning_rate": 6.56043891625035e-06, + "loss": 0.14534789323806763, + "step": 3491 + }, + { + "epoch": 1.9532867132867133, + "grad_norm": 0.4398997597558771, + "learning_rate": 6.554328373851785e-06, + "loss": 0.15859714150428772, + "step": 3492 + }, + { + "epoch": 1.953846153846154, + "grad_norm": 0.31796932535555594, + "learning_rate": 6.548219291003363e-06, + "loss": 0.050215672701597214, + "step": 3493 + }, + { + "epoch": 1.9544055944055945, + "grad_norm": 0.41357668286175514, + "learning_rate": 6.5421116702928165e-06, + "loss": 0.09473301470279694, + "step": 3494 + }, + { + "epoch": 1.954965034965035, + "grad_norm": 0.3849741937975097, + "learning_rate": 6.536005514307283e-06, + "loss": 0.11327485740184784, + "step": 3495 + }, + { + "epoch": 1.9555244755244754, + "grad_norm": 0.4629525286892661, + "learning_rate": 6.529900825633258e-06, + "loss": 0.1411135047674179, + "step": 3496 + }, + { + "epoch": 1.956083916083916, + "grad_norm": 0.3063673700964594, + "learning_rate": 6.5237976068566266e-06, + "loss": 0.07149860262870789, + "step": 3497 + }, + { + "epoch": 1.9566433566433568, + "grad_norm": 0.3502409133663341, + "learning_rate": 6.517695860562646e-06, + "loss": 0.07713808119297028, + "step": 3498 + }, + { + "epoch": 1.9572027972027972, + "grad_norm": 0.4975181832037392, + "learning_rate": 6.511595589335948e-06, + "loss": 0.17901864647865295, + "step": 3499 + }, + { + "epoch": 1.9577622377622377, + "grad_norm": 0.45018007014481254, + "learning_rate": 6.505496795760553e-06, + "loss": 0.2368810474872589, + "step": 3500 + }, + { + "epoch": 1.9583216783216784, + "grad_norm": 0.48843866621992044, + "learning_rate": 6.499399482419835e-06, + "loss": 0.07429467886686325, + "step": 3501 + }, + { + "epoch": 1.958881118881119, + "grad_norm": 0.3325190573156754, + "learning_rate": 6.493303651896561e-06, + "loss": 0.07490898668766022, + "step": 3502 + }, + { + "epoch": 1.9594405594405595, + "grad_norm": 0.48756018528439626, + "learning_rate": 6.487209306772852e-06, + "loss": 0.15737158060073853, + "step": 3503 + }, + { + "epoch": 1.96, + "grad_norm": 0.6587804748392411, + "learning_rate": 6.4811164496302115e-06, + "loss": 0.2837570607662201, + "step": 3504 + }, + { + "epoch": 1.9605594405594404, + "grad_norm": 0.5468174419590192, + "learning_rate": 6.475025083049513e-06, + "loss": 0.17096087336540222, + "step": 3505 + }, + { + "epoch": 1.961118881118881, + "grad_norm": 0.44520194649563316, + "learning_rate": 6.46893520961099e-06, + "loss": 0.08211810886859894, + "step": 3506 + }, + { + "epoch": 1.9616783216783218, + "grad_norm": 0.2217755387651834, + "learning_rate": 6.462846831894253e-06, + "loss": 0.02249753288924694, + "step": 3507 + }, + { + "epoch": 1.9622377622377623, + "grad_norm": 0.48082406102995834, + "learning_rate": 6.456759952478268e-06, + "loss": 0.16572976112365723, + "step": 3508 + }, + { + "epoch": 1.9627972027972027, + "grad_norm": 0.4834374819206414, + "learning_rate": 6.450674573941381e-06, + "loss": 0.11485010385513306, + "step": 3509 + }, + { + "epoch": 1.9633566433566434, + "grad_norm": 0.1614034052375948, + "learning_rate": 6.44459069886129e-06, + "loss": 0.024327319115400314, + "step": 3510 + }, + { + "epoch": 1.963916083916084, + "grad_norm": 0.4038265878831708, + "learning_rate": 6.438508329815061e-06, + "loss": 0.12833207845687866, + "step": 3511 + }, + { + "epoch": 1.9644755244755245, + "grad_norm": 0.39361362397200195, + "learning_rate": 6.432427469379121e-06, + "loss": 0.09652186930179596, + "step": 3512 + }, + { + "epoch": 1.965034965034965, + "grad_norm": 0.467435032172997, + "learning_rate": 6.426348120129253e-06, + "loss": 0.10218455642461777, + "step": 3513 + }, + { + "epoch": 1.9655944055944055, + "grad_norm": 0.43437770300869705, + "learning_rate": 6.420270284640617e-06, + "loss": 0.12388552725315094, + "step": 3514 + }, + { + "epoch": 1.9661538461538461, + "grad_norm": 0.3266815667334029, + "learning_rate": 6.414193965487712e-06, + "loss": 0.06462287902832031, + "step": 3515 + }, + { + "epoch": 1.9667132867132868, + "grad_norm": 0.7185759174176742, + "learning_rate": 6.408119165244403e-06, + "loss": 0.2515866160392761, + "step": 3516 + }, + { + "epoch": 1.9672727272727273, + "grad_norm": 0.3974821933348702, + "learning_rate": 6.402045886483914e-06, + "loss": 0.14408966898918152, + "step": 3517 + }, + { + "epoch": 1.9678321678321677, + "grad_norm": 0.6240935460404687, + "learning_rate": 6.395974131778816e-06, + "loss": 0.18547114729881287, + "step": 3518 + }, + { + "epoch": 1.9683916083916084, + "grad_norm": 0.3695355346578402, + "learning_rate": 6.389903903701051e-06, + "loss": 0.134500190615654, + "step": 3519 + }, + { + "epoch": 1.968951048951049, + "grad_norm": 0.35094687291021415, + "learning_rate": 6.383835204821895e-06, + "loss": 0.11318524181842804, + "step": 3520 + }, + { + "epoch": 1.9695104895104896, + "grad_norm": 0.3802944599092993, + "learning_rate": 6.377768037711982e-06, + "loss": 0.07681688666343689, + "step": 3521 + }, + { + "epoch": 1.97006993006993, + "grad_norm": 0.5432811700401418, + "learning_rate": 6.371702404941309e-06, + "loss": 0.12052248418331146, + "step": 3522 + }, + { + "epoch": 1.9706293706293705, + "grad_norm": 0.6831476704483054, + "learning_rate": 6.365638309079207e-06, + "loss": 0.30993402004241943, + "step": 3523 + }, + { + "epoch": 1.9711888111888112, + "grad_norm": 0.3097961474449339, + "learning_rate": 6.359575752694369e-06, + "loss": 0.07829742133617401, + "step": 3524 + }, + { + "epoch": 1.9717482517482519, + "grad_norm": 0.39677671382258856, + "learning_rate": 6.3535147383548265e-06, + "loss": 0.09763751924037933, + "step": 3525 + }, + { + "epoch": 1.9723076923076923, + "grad_norm": 0.5244620010415475, + "learning_rate": 6.347455268627958e-06, + "loss": 0.13628189265727997, + "step": 3526 + }, + { + "epoch": 1.9728671328671328, + "grad_norm": 0.46429743019491937, + "learning_rate": 6.341397346080498e-06, + "loss": 0.19780652225017548, + "step": 3527 + }, + { + "epoch": 1.9734265734265735, + "grad_norm": 0.4227208734426038, + "learning_rate": 6.335340973278512e-06, + "loss": 0.0873546376824379, + "step": 3528 + }, + { + "epoch": 1.9739860139860141, + "grad_norm": 0.366361627824157, + "learning_rate": 6.329286152787422e-06, + "loss": 0.12127527594566345, + "step": 3529 + }, + { + "epoch": 1.9745454545454546, + "grad_norm": 0.3577626297965358, + "learning_rate": 6.323232887171978e-06, + "loss": 0.0987105518579483, + "step": 3530 + }, + { + "epoch": 1.975104895104895, + "grad_norm": 0.3974958816813422, + "learning_rate": 6.317181178996286e-06, + "loss": 0.10629743337631226, + "step": 3531 + }, + { + "epoch": 1.9756643356643355, + "grad_norm": 0.28426896282155545, + "learning_rate": 6.311131030823786e-06, + "loss": 0.0655193105340004, + "step": 3532 + }, + { + "epoch": 1.9762237762237762, + "grad_norm": 0.5550641966976677, + "learning_rate": 6.305082445217252e-06, + "loss": 0.14093634486198425, + "step": 3533 + }, + { + "epoch": 1.976783216783217, + "grad_norm": 0.5178769004415351, + "learning_rate": 6.299035424738807e-06, + "loss": 0.10858261585235596, + "step": 3534 + }, + { + "epoch": 1.9773426573426574, + "grad_norm": 0.41891360485690887, + "learning_rate": 6.292989971949896e-06, + "loss": 0.21731534600257874, + "step": 3535 + }, + { + "epoch": 1.9779020979020978, + "grad_norm": 0.44802752192124246, + "learning_rate": 6.286946089411319e-06, + "loss": 0.1250826120376587, + "step": 3536 + }, + { + "epoch": 1.9784615384615385, + "grad_norm": 0.7011177692265023, + "learning_rate": 6.280903779683197e-06, + "loss": 0.20018091797828674, + "step": 3537 + }, + { + "epoch": 1.9790209790209792, + "grad_norm": 0.4519572176087496, + "learning_rate": 6.274863045324987e-06, + "loss": 0.13562744855880737, + "step": 3538 + }, + { + "epoch": 1.9795804195804196, + "grad_norm": 0.5641190003620636, + "learning_rate": 6.2688238888954835e-06, + "loss": 0.21992667019367218, + "step": 3539 + }, + { + "epoch": 1.98013986013986, + "grad_norm": 0.3764968310500614, + "learning_rate": 6.262786312952802e-06, + "loss": 0.09611614048480988, + "step": 3540 + }, + { + "epoch": 1.9806993006993006, + "grad_norm": 0.5184967875800285, + "learning_rate": 6.256750320054408e-06, + "loss": 0.17659834027290344, + "step": 3541 + }, + { + "epoch": 1.9812587412587412, + "grad_norm": 0.4252308841386399, + "learning_rate": 6.2507159127570775e-06, + "loss": 0.077400803565979, + "step": 3542 + }, + { + "epoch": 1.981818181818182, + "grad_norm": 0.4511501460242403, + "learning_rate": 6.244683093616917e-06, + "loss": 0.07500603049993515, + "step": 3543 + }, + { + "epoch": 1.9823776223776224, + "grad_norm": 0.30337405613090945, + "learning_rate": 6.238651865189375e-06, + "loss": 0.07867264002561569, + "step": 3544 + }, + { + "epoch": 1.9829370629370628, + "grad_norm": 1.189958039004775, + "learning_rate": 6.232622230029208e-06, + "loss": 0.2818150520324707, + "step": 3545 + }, + { + "epoch": 1.9834965034965035, + "grad_norm": 0.5692202996789941, + "learning_rate": 6.226594190690512e-06, + "loss": 0.2510114312171936, + "step": 3546 + }, + { + "epoch": 1.9840559440559442, + "grad_norm": 0.3312454538948516, + "learning_rate": 6.220567749726695e-06, + "loss": 0.10725396871566772, + "step": 3547 + }, + { + "epoch": 1.9846153846153847, + "grad_norm": 0.35213912733274894, + "learning_rate": 6.2145429096905e-06, + "loss": 0.056303471326828, + "step": 3548 + }, + { + "epoch": 1.9851748251748251, + "grad_norm": 0.38187277266823944, + "learning_rate": 6.208519673133982e-06, + "loss": 0.10670888423919678, + "step": 3549 + }, + { + "epoch": 1.9857342657342656, + "grad_norm": 0.31152278675764056, + "learning_rate": 6.20249804260852e-06, + "loss": 0.11498218774795532, + "step": 3550 + }, + { + "epoch": 1.9862937062937063, + "grad_norm": 0.4732735309000487, + "learning_rate": 6.196478020664816e-06, + "loss": 0.12872734665870667, + "step": 3551 + }, + { + "epoch": 1.986853146853147, + "grad_norm": 0.36176669078427914, + "learning_rate": 6.1904596098528815e-06, + "loss": 0.14444312453269958, + "step": 3552 + }, + { + "epoch": 1.9874125874125874, + "grad_norm": 0.46837593244168807, + "learning_rate": 6.18444281272206e-06, + "loss": 0.10586313903331757, + "step": 3553 + }, + { + "epoch": 1.9879720279720279, + "grad_norm": 0.5640447305763538, + "learning_rate": 6.178427631821e-06, + "loss": 0.26201170682907104, + "step": 3554 + }, + { + "epoch": 1.9885314685314686, + "grad_norm": 0.3561089418977259, + "learning_rate": 6.172414069697665e-06, + "loss": 0.032081738114356995, + "step": 3555 + }, + { + "epoch": 1.9890909090909092, + "grad_norm": 0.32662189958266785, + "learning_rate": 6.166402128899342e-06, + "loss": 0.0964776948094368, + "step": 3556 + }, + { + "epoch": 1.9896503496503497, + "grad_norm": 0.3064863646994168, + "learning_rate": 6.160391811972621e-06, + "loss": 0.07122602313756943, + "step": 3557 + }, + { + "epoch": 1.9902097902097902, + "grad_norm": 0.46739260049307374, + "learning_rate": 6.154383121463414e-06, + "loss": 0.12019982933998108, + "step": 3558 + }, + { + "epoch": 1.9907692307692306, + "grad_norm": 0.34681824023996916, + "learning_rate": 6.148376059916936e-06, + "loss": 0.1056625172495842, + "step": 3559 + }, + { + "epoch": 1.9913286713286713, + "grad_norm": 0.4466992142575238, + "learning_rate": 6.142370629877713e-06, + "loss": 0.09969577193260193, + "step": 3560 + }, + { + "epoch": 1.991888111888112, + "grad_norm": 0.5111046613821663, + "learning_rate": 6.136366833889585e-06, + "loss": 0.2296912670135498, + "step": 3561 + }, + { + "epoch": 1.9924475524475525, + "grad_norm": 0.6431717871648942, + "learning_rate": 6.130364674495692e-06, + "loss": 0.250143826007843, + "step": 3562 + }, + { + "epoch": 1.993006993006993, + "grad_norm": 0.49424900746684675, + "learning_rate": 6.1243641542384935e-06, + "loss": 0.19993560016155243, + "step": 3563 + }, + { + "epoch": 1.9935664335664336, + "grad_norm": 0.33730918707502133, + "learning_rate": 6.11836527565974e-06, + "loss": 0.07463625073432922, + "step": 3564 + }, + { + "epoch": 1.9941258741258743, + "grad_norm": 0.49830908796272494, + "learning_rate": 6.112368041300494e-06, + "loss": 0.07276284694671631, + "step": 3565 + }, + { + "epoch": 1.9946853146853147, + "grad_norm": 0.3335952795229393, + "learning_rate": 6.1063724537011235e-06, + "loss": 0.1839003711938858, + "step": 3566 + }, + { + "epoch": 1.9952447552447552, + "grad_norm": 0.47031527214264385, + "learning_rate": 6.10037851540129e-06, + "loss": 0.20200061798095703, + "step": 3567 + }, + { + "epoch": 1.9958041958041957, + "grad_norm": 0.7074774625839508, + "learning_rate": 6.094386228939973e-06, + "loss": 0.2476254552602768, + "step": 3568 + }, + { + "epoch": 1.9963636363636363, + "grad_norm": 0.4730661496973922, + "learning_rate": 6.08839559685543e-06, + "loss": 0.14904817938804626, + "step": 3569 + }, + { + "epoch": 1.996923076923077, + "grad_norm": 0.45754446959509204, + "learning_rate": 6.0824066216852376e-06, + "loss": 0.1679742932319641, + "step": 3570 + }, + { + "epoch": 1.9974825174825175, + "grad_norm": 0.5098507358868039, + "learning_rate": 6.076419305966261e-06, + "loss": 0.1967964768409729, + "step": 3571 + }, + { + "epoch": 1.998041958041958, + "grad_norm": 0.3161413689863315, + "learning_rate": 6.07043365223466e-06, + "loss": 0.07619272172451019, + "step": 3572 + }, + { + "epoch": 1.9986013986013986, + "grad_norm": 0.2749688496689265, + "learning_rate": 6.064449663025901e-06, + "loss": 0.07677794247865677, + "step": 3573 + }, + { + "epoch": 1.9991608391608393, + "grad_norm": 0.49452694540929676, + "learning_rate": 6.0584673408747294e-06, + "loss": 0.1892307549715042, + "step": 3574 + }, + { + "epoch": 1.9997202797202798, + "grad_norm": 0.5141170825154593, + "learning_rate": 6.052486688315205e-06, + "loss": 0.26105064153671265, + "step": 3575 + }, + { + "epoch": 2.0, + "grad_norm": 0.4092057065884709, + "learning_rate": 6.0465077078806624e-06, + "loss": 0.06920918077230453, + "step": 3576 + }, + { + "epoch": 2.0005594405594405, + "grad_norm": 0.35978219233657804, + "learning_rate": 6.040530402103733e-06, + "loss": 0.07399743795394897, + "step": 3577 + }, + { + "epoch": 2.001118881118881, + "grad_norm": 0.42726068687381213, + "learning_rate": 6.034554773516345e-06, + "loss": 0.12395565211772919, + "step": 3578 + }, + { + "epoch": 2.001678321678322, + "grad_norm": 0.49725534838504143, + "learning_rate": 6.028580824649708e-06, + "loss": 0.0989704430103302, + "step": 3579 + }, + { + "epoch": 2.0022377622377623, + "grad_norm": 0.4045948537803414, + "learning_rate": 6.022608558034327e-06, + "loss": 0.0751051977276802, + "step": 3580 + }, + { + "epoch": 2.0027972027972027, + "grad_norm": 0.41277344351671674, + "learning_rate": 6.01663797619999e-06, + "loss": 0.04803916811943054, + "step": 3581 + }, + { + "epoch": 2.003356643356643, + "grad_norm": 0.3271614335492754, + "learning_rate": 6.01066908167577e-06, + "loss": 0.08659067004919052, + "step": 3582 + }, + { + "epoch": 2.003916083916084, + "grad_norm": 0.47476889160877705, + "learning_rate": 6.004701876990033e-06, + "loss": 0.14988979697227478, + "step": 3583 + }, + { + "epoch": 2.0044755244755246, + "grad_norm": 0.4904626580470893, + "learning_rate": 5.9987363646704165e-06, + "loss": 0.08567748963832855, + "step": 3584 + }, + { + "epoch": 2.005034965034965, + "grad_norm": 0.5071744802074006, + "learning_rate": 5.992772547243856e-06, + "loss": 0.06175285577774048, + "step": 3585 + }, + { + "epoch": 2.0055944055944055, + "grad_norm": 0.39308048510159155, + "learning_rate": 5.986810427236556e-06, + "loss": 0.06493330746889114, + "step": 3586 + }, + { + "epoch": 2.006153846153846, + "grad_norm": 0.49319573392376725, + "learning_rate": 5.980850007174013e-06, + "loss": 0.15454116463661194, + "step": 3587 + }, + { + "epoch": 2.006713286713287, + "grad_norm": 0.5353366406143473, + "learning_rate": 5.974891289580994e-06, + "loss": 0.0849255844950676, + "step": 3588 + }, + { + "epoch": 2.0072727272727273, + "grad_norm": 0.3605242244360612, + "learning_rate": 5.968934276981546e-06, + "loss": 0.09450843185186386, + "step": 3589 + }, + { + "epoch": 2.007832167832168, + "grad_norm": 0.36870821301451273, + "learning_rate": 5.962978971899005e-06, + "loss": 0.04279526323080063, + "step": 3590 + }, + { + "epoch": 2.0083916083916082, + "grad_norm": 0.5696943378351228, + "learning_rate": 5.95702537685597e-06, + "loss": 0.18739965558052063, + "step": 3591 + }, + { + "epoch": 2.008951048951049, + "grad_norm": 0.5195422799874655, + "learning_rate": 5.951073494374322e-06, + "loss": 0.0595402792096138, + "step": 3592 + }, + { + "epoch": 2.0095104895104896, + "grad_norm": 0.5137655468831066, + "learning_rate": 5.945123326975217e-06, + "loss": 0.15191105008125305, + "step": 3593 + }, + { + "epoch": 2.01006993006993, + "grad_norm": 0.17692481774449162, + "learning_rate": 5.939174877179079e-06, + "loss": 0.008990811184048653, + "step": 3594 + }, + { + "epoch": 2.0106293706293705, + "grad_norm": 0.3318518232043033, + "learning_rate": 5.933228147505614e-06, + "loss": 0.024509279057383537, + "step": 3595 + }, + { + "epoch": 2.011188811188811, + "grad_norm": 0.5865332769328357, + "learning_rate": 5.927283140473786e-06, + "loss": 0.06581870466470718, + "step": 3596 + }, + { + "epoch": 2.011748251748252, + "grad_norm": 0.3337865853000943, + "learning_rate": 5.921339858601848e-06, + "loss": 0.01926508918404579, + "step": 3597 + }, + { + "epoch": 2.0123076923076924, + "grad_norm": 0.33947267458287084, + "learning_rate": 5.915398304407305e-06, + "loss": 0.06254900246858597, + "step": 3598 + }, + { + "epoch": 2.012867132867133, + "grad_norm": 0.3639678334340434, + "learning_rate": 5.909458480406936e-06, + "loss": 0.04075920581817627, + "step": 3599 + }, + { + "epoch": 2.0134265734265733, + "grad_norm": 0.5971145263205969, + "learning_rate": 5.903520389116791e-06, + "loss": 0.14295360445976257, + "step": 3600 + }, + { + "epoch": 2.013986013986014, + "grad_norm": 0.4663310294563994, + "learning_rate": 5.897584033052177e-06, + "loss": 0.09693768620491028, + "step": 3601 + }, + { + "epoch": 2.0145454545454546, + "grad_norm": 0.42361193493744387, + "learning_rate": 5.891649414727681e-06, + "loss": 0.07173395901918411, + "step": 3602 + }, + { + "epoch": 2.015104895104895, + "grad_norm": 0.38530968398688054, + "learning_rate": 5.8857165366571376e-06, + "loss": 0.03856249898672104, + "step": 3603 + }, + { + "epoch": 2.0156643356643356, + "grad_norm": 0.4555599767317801, + "learning_rate": 5.879785401353655e-06, + "loss": 0.1216660887002945, + "step": 3604 + }, + { + "epoch": 2.016223776223776, + "grad_norm": 0.3720973244698482, + "learning_rate": 5.873856011329599e-06, + "loss": 0.05517837777733803, + "step": 3605 + }, + { + "epoch": 2.016783216783217, + "grad_norm": 0.7001920766019045, + "learning_rate": 5.86792836909659e-06, + "loss": 0.1694219708442688, + "step": 3606 + }, + { + "epoch": 2.0173426573426574, + "grad_norm": 0.25070413392477386, + "learning_rate": 5.862002477165525e-06, + "loss": 0.056889232248067856, + "step": 3607 + }, + { + "epoch": 2.017902097902098, + "grad_norm": 0.3757744994037413, + "learning_rate": 5.8560783380465425e-06, + "loss": 0.07298022508621216, + "step": 3608 + }, + { + "epoch": 2.0184615384615383, + "grad_norm": 0.3682329075810601, + "learning_rate": 5.850155954249049e-06, + "loss": 0.05521957203745842, + "step": 3609 + }, + { + "epoch": 2.019020979020979, + "grad_norm": 0.25835913689719603, + "learning_rate": 5.844235328281703e-06, + "loss": 0.01638389565050602, + "step": 3610 + }, + { + "epoch": 2.0195804195804197, + "grad_norm": 0.471959286513821, + "learning_rate": 5.838316462652414e-06, + "loss": 0.06502893567085266, + "step": 3611 + }, + { + "epoch": 2.02013986013986, + "grad_norm": 0.27536697194175397, + "learning_rate": 5.832399359868361e-06, + "loss": 0.04117986932396889, + "step": 3612 + }, + { + "epoch": 2.0206993006993006, + "grad_norm": 0.3994983826905144, + "learning_rate": 5.826484022435954e-06, + "loss": 0.05694076046347618, + "step": 3613 + }, + { + "epoch": 2.021258741258741, + "grad_norm": 0.7388377505985434, + "learning_rate": 5.820570452860881e-06, + "loss": 0.07224252074956894, + "step": 3614 + }, + { + "epoch": 2.021818181818182, + "grad_norm": 0.3812737391875397, + "learning_rate": 5.814658653648062e-06, + "loss": 0.09924677759408951, + "step": 3615 + }, + { + "epoch": 2.0223776223776224, + "grad_norm": 0.17504586101404732, + "learning_rate": 5.808748627301663e-06, + "loss": 0.015852684155106544, + "step": 3616 + }, + { + "epoch": 2.022937062937063, + "grad_norm": 0.5689025785847134, + "learning_rate": 5.802840376325125e-06, + "loss": 0.09133298695087433, + "step": 3617 + }, + { + "epoch": 2.0234965034965033, + "grad_norm": 0.6177896932404707, + "learning_rate": 5.796933903221109e-06, + "loss": 0.18458610773086548, + "step": 3618 + }, + { + "epoch": 2.0240559440559442, + "grad_norm": 0.44864885807466465, + "learning_rate": 5.791029210491539e-06, + "loss": 0.049970563501119614, + "step": 3619 + }, + { + "epoch": 2.0246153846153847, + "grad_norm": 0.4095064388660449, + "learning_rate": 5.7851263006375846e-06, + "loss": 0.057555388659238815, + "step": 3620 + }, + { + "epoch": 2.025174825174825, + "grad_norm": 0.27483395236230124, + "learning_rate": 5.779225176159644e-06, + "loss": 0.015221508219838142, + "step": 3621 + }, + { + "epoch": 2.0257342657342656, + "grad_norm": 0.44035759515809086, + "learning_rate": 5.773325839557389e-06, + "loss": 0.13432274758815765, + "step": 3622 + }, + { + "epoch": 2.026293706293706, + "grad_norm": 0.49856768502239945, + "learning_rate": 5.7674282933297e-06, + "loss": 0.12969036400318146, + "step": 3623 + }, + { + "epoch": 2.026853146853147, + "grad_norm": 0.4728826195090137, + "learning_rate": 5.761532539974723e-06, + "loss": 0.07945172488689423, + "step": 3624 + }, + { + "epoch": 2.0274125874125875, + "grad_norm": 0.4349725363890918, + "learning_rate": 5.755638581989836e-06, + "loss": 0.04767220839858055, + "step": 3625 + }, + { + "epoch": 2.027972027972028, + "grad_norm": 0.5605222708021007, + "learning_rate": 5.749746421871656e-06, + "loss": 0.09730111062526703, + "step": 3626 + }, + { + "epoch": 2.0285314685314684, + "grad_norm": 0.5309360534626533, + "learning_rate": 5.743856062116046e-06, + "loss": 0.07119136303663254, + "step": 3627 + }, + { + "epoch": 2.0290909090909093, + "grad_norm": 0.35813777299678823, + "learning_rate": 5.737967505218092e-06, + "loss": 0.017523549497127533, + "step": 3628 + }, + { + "epoch": 2.0296503496503497, + "grad_norm": 0.4085119298121199, + "learning_rate": 5.732080753672125e-06, + "loss": 0.0720461830496788, + "step": 3629 + }, + { + "epoch": 2.03020979020979, + "grad_norm": 0.4049652031716054, + "learning_rate": 5.726195809971716e-06, + "loss": 0.06236068159341812, + "step": 3630 + }, + { + "epoch": 2.0307692307692307, + "grad_norm": 0.5262831840972194, + "learning_rate": 5.720312676609663e-06, + "loss": 0.08773636817932129, + "step": 3631 + }, + { + "epoch": 2.031328671328671, + "grad_norm": 0.349012979989192, + "learning_rate": 5.714431356078002e-06, + "loss": 0.06945449858903885, + "step": 3632 + }, + { + "epoch": 2.031888111888112, + "grad_norm": 0.30175183999892524, + "learning_rate": 5.708551850867991e-06, + "loss": 0.06037995219230652, + "step": 3633 + }, + { + "epoch": 2.0324475524475525, + "grad_norm": 0.3401986239367232, + "learning_rate": 5.702674163470132e-06, + "loss": 0.02973555400967598, + "step": 3634 + }, + { + "epoch": 2.033006993006993, + "grad_norm": 0.5638604728263719, + "learning_rate": 5.696798296374149e-06, + "loss": 0.16984383761882782, + "step": 3635 + }, + { + "epoch": 2.0335664335664334, + "grad_norm": 0.5010697238528419, + "learning_rate": 5.690924252068998e-06, + "loss": 0.03337867558002472, + "step": 3636 + }, + { + "epoch": 2.0341258741258743, + "grad_norm": 0.24162635232310722, + "learning_rate": 5.685052033042866e-06, + "loss": 0.014196659438312054, + "step": 3637 + }, + { + "epoch": 2.0346853146853148, + "grad_norm": 0.4638428504150314, + "learning_rate": 5.679181641783157e-06, + "loss": 0.1894918531179428, + "step": 3638 + }, + { + "epoch": 2.0352447552447552, + "grad_norm": 0.22788733969469016, + "learning_rate": 5.673313080776508e-06, + "loss": 0.016828207299113274, + "step": 3639 + }, + { + "epoch": 2.0358041958041957, + "grad_norm": 0.23935326507926988, + "learning_rate": 5.667446352508781e-06, + "loss": 0.011041881516575813, + "step": 3640 + }, + { + "epoch": 2.036363636363636, + "grad_norm": 0.38316306257784793, + "learning_rate": 5.661581459465064e-06, + "loss": 0.03983404487371445, + "step": 3641 + }, + { + "epoch": 2.036923076923077, + "grad_norm": 0.5097000963457214, + "learning_rate": 5.655718404129653e-06, + "loss": 0.13667380809783936, + "step": 3642 + }, + { + "epoch": 2.0374825174825175, + "grad_norm": 0.2943381334892028, + "learning_rate": 5.6498571889860895e-06, + "loss": 0.026292983442544937, + "step": 3643 + }, + { + "epoch": 2.038041958041958, + "grad_norm": 1.0648148071899517, + "learning_rate": 5.643997816517113e-06, + "loss": 0.1653956025838852, + "step": 3644 + }, + { + "epoch": 2.0386013986013984, + "grad_norm": 0.2958234046083201, + "learning_rate": 5.638140289204694e-06, + "loss": 0.06516967713832855, + "step": 3645 + }, + { + "epoch": 2.0391608391608393, + "grad_norm": 0.5877306471538616, + "learning_rate": 5.632284609530026e-06, + "loss": 0.15873503684997559, + "step": 3646 + }, + { + "epoch": 2.03972027972028, + "grad_norm": 1.0445558476136994, + "learning_rate": 5.6264307799734975e-06, + "loss": 0.22914071381092072, + "step": 3647 + }, + { + "epoch": 2.0402797202797203, + "grad_norm": 0.2665549225886576, + "learning_rate": 5.620578803014747e-06, + "loss": 0.0227402001619339, + "step": 3648 + }, + { + "epoch": 2.0408391608391607, + "grad_norm": 0.4843898046100904, + "learning_rate": 5.614728681132597e-06, + "loss": 0.07469518482685089, + "step": 3649 + }, + { + "epoch": 2.041398601398601, + "grad_norm": 0.420513358626099, + "learning_rate": 5.608880416805104e-06, + "loss": 0.024593062698841095, + "step": 3650 + }, + { + "epoch": 2.041958041958042, + "grad_norm": 0.27759936879447633, + "learning_rate": 5.6030340125095315e-06, + "loss": 0.012602729722857475, + "step": 3651 + }, + { + "epoch": 2.0425174825174826, + "grad_norm": 0.32277310851925745, + "learning_rate": 5.597189470722346e-06, + "loss": 0.04418928921222687, + "step": 3652 + }, + { + "epoch": 2.043076923076923, + "grad_norm": 0.360862328808045, + "learning_rate": 5.5913467939192475e-06, + "loss": 0.0394764319062233, + "step": 3653 + }, + { + "epoch": 2.0436363636363635, + "grad_norm": 0.4123518350035916, + "learning_rate": 5.585505984575122e-06, + "loss": 0.04966995120048523, + "step": 3654 + }, + { + "epoch": 2.0441958041958044, + "grad_norm": 0.38983659700543016, + "learning_rate": 5.579667045164075e-06, + "loss": 0.056858088821172714, + "step": 3655 + }, + { + "epoch": 2.044755244755245, + "grad_norm": 0.47020113770180444, + "learning_rate": 5.5738299781594286e-06, + "loss": 0.12896110117435455, + "step": 3656 + }, + { + "epoch": 2.0453146853146853, + "grad_norm": 0.3832595305612435, + "learning_rate": 5.5679947860336884e-06, + "loss": 0.07391506433486938, + "step": 3657 + }, + { + "epoch": 2.0458741258741258, + "grad_norm": 0.2887353722135236, + "learning_rate": 5.5621614712585945e-06, + "loss": 0.061476100236177444, + "step": 3658 + }, + { + "epoch": 2.046433566433566, + "grad_norm": 0.4648826566565904, + "learning_rate": 5.556330036305069e-06, + "loss": 0.09372629970312119, + "step": 3659 + }, + { + "epoch": 2.046993006993007, + "grad_norm": 0.6391891169729421, + "learning_rate": 5.550500483643248e-06, + "loss": 0.17711451649665833, + "step": 3660 + }, + { + "epoch": 2.0475524475524476, + "grad_norm": 0.4732211611391292, + "learning_rate": 5.544672815742472e-06, + "loss": 0.10096335411071777, + "step": 3661 + }, + { + "epoch": 2.048111888111888, + "grad_norm": 0.5035580828683603, + "learning_rate": 5.538847035071268e-06, + "loss": 0.11925114691257477, + "step": 3662 + }, + { + "epoch": 2.0486713286713285, + "grad_norm": 0.6030569682663128, + "learning_rate": 5.533023144097394e-06, + "loss": 0.13466320931911469, + "step": 3663 + }, + { + "epoch": 2.0492307692307694, + "grad_norm": 0.5536194034430205, + "learning_rate": 5.527201145287774e-06, + "loss": 0.1448664665222168, + "step": 3664 + }, + { + "epoch": 2.04979020979021, + "grad_norm": 0.48007857818572885, + "learning_rate": 5.5213810411085485e-06, + "loss": 0.05142068862915039, + "step": 3665 + }, + { + "epoch": 2.0503496503496503, + "grad_norm": 0.5445443136220391, + "learning_rate": 5.515562834025059e-06, + "loss": 0.11228204518556595, + "step": 3666 + }, + { + "epoch": 2.050909090909091, + "grad_norm": 0.4818539389764125, + "learning_rate": 5.509746526501827e-06, + "loss": 0.05745711177587509, + "step": 3667 + }, + { + "epoch": 2.0514685314685313, + "grad_norm": 0.3198995442888189, + "learning_rate": 5.503932121002582e-06, + "loss": 0.041490714997053146, + "step": 3668 + }, + { + "epoch": 2.052027972027972, + "grad_norm": 0.4469101592413275, + "learning_rate": 5.498119619990248e-06, + "loss": 0.08994293212890625, + "step": 3669 + }, + { + "epoch": 2.0525874125874126, + "grad_norm": 0.4014275580679061, + "learning_rate": 5.492309025926934e-06, + "loss": 0.11252099275588989, + "step": 3670 + }, + { + "epoch": 2.053146853146853, + "grad_norm": 0.478208282183184, + "learning_rate": 5.486500341273956e-06, + "loss": 0.06122881546616554, + "step": 3671 + }, + { + "epoch": 2.0537062937062935, + "grad_norm": 0.2662500185207163, + "learning_rate": 5.4806935684917995e-06, + "loss": 0.031060781329870224, + "step": 3672 + }, + { + "epoch": 2.0542657342657344, + "grad_norm": 0.4440109835859854, + "learning_rate": 5.4748887100401585e-06, + "loss": 0.07920639216899872, + "step": 3673 + }, + { + "epoch": 2.054825174825175, + "grad_norm": 0.7585302707908824, + "learning_rate": 5.469085768377909e-06, + "loss": 0.14494416117668152, + "step": 3674 + }, + { + "epoch": 2.0553846153846154, + "grad_norm": 0.5757505252746431, + "learning_rate": 5.463284745963114e-06, + "loss": 0.14469082653522491, + "step": 3675 + }, + { + "epoch": 2.055944055944056, + "grad_norm": 0.6454783340955984, + "learning_rate": 5.457485645253031e-06, + "loss": 0.15794336795806885, + "step": 3676 + }, + { + "epoch": 2.0565034965034963, + "grad_norm": 0.44979091261890064, + "learning_rate": 5.451688468704093e-06, + "loss": 0.06799627840518951, + "step": 3677 + }, + { + "epoch": 2.057062937062937, + "grad_norm": 0.3932718471446207, + "learning_rate": 5.44589321877192e-06, + "loss": 0.030850786715745926, + "step": 3678 + }, + { + "epoch": 2.0576223776223777, + "grad_norm": 0.5607185180028378, + "learning_rate": 5.440099897911325e-06, + "loss": 0.13444456458091736, + "step": 3679 + }, + { + "epoch": 2.058181818181818, + "grad_norm": 0.34047391337945915, + "learning_rate": 5.434308508576294e-06, + "loss": 0.027570869773626328, + "step": 3680 + }, + { + "epoch": 2.0587412587412586, + "grad_norm": 0.2793196482785029, + "learning_rate": 5.428519053220001e-06, + "loss": 0.032223835587501526, + "step": 3681 + }, + { + "epoch": 2.0593006993006995, + "grad_norm": 0.4526092185783086, + "learning_rate": 5.422731534294798e-06, + "loss": 0.06694259494543076, + "step": 3682 + }, + { + "epoch": 2.05986013986014, + "grad_norm": 0.3695543829332388, + "learning_rate": 5.416945954252213e-06, + "loss": 0.031319841742515564, + "step": 3683 + }, + { + "epoch": 2.0604195804195804, + "grad_norm": 0.5298096815597152, + "learning_rate": 5.41116231554296e-06, + "loss": 0.0646713376045227, + "step": 3684 + }, + { + "epoch": 2.060979020979021, + "grad_norm": 0.3363106012544741, + "learning_rate": 5.405380620616926e-06, + "loss": 0.019697466865181923, + "step": 3685 + }, + { + "epoch": 2.0615384615384613, + "grad_norm": 0.7205932346679229, + "learning_rate": 5.399600871923176e-06, + "loss": 0.07502421736717224, + "step": 3686 + }, + { + "epoch": 2.0620979020979022, + "grad_norm": 0.598373841757264, + "learning_rate": 5.393823071909956e-06, + "loss": 0.12852756679058075, + "step": 3687 + }, + { + "epoch": 2.0626573426573427, + "grad_norm": 0.6864494564438635, + "learning_rate": 5.388047223024671e-06, + "loss": 0.24184030294418335, + "step": 3688 + }, + { + "epoch": 2.063216783216783, + "grad_norm": 0.8825811438285203, + "learning_rate": 5.382273327713916e-06, + "loss": 0.1504954844713211, + "step": 3689 + }, + { + "epoch": 2.0637762237762236, + "grad_norm": 0.21495900921015693, + "learning_rate": 5.376501388423454e-06, + "loss": 0.012958584353327751, + "step": 3690 + }, + { + "epoch": 2.0643356643356645, + "grad_norm": 0.41194764666844247, + "learning_rate": 5.370731407598209e-06, + "loss": 0.136464461684227, + "step": 3691 + }, + { + "epoch": 2.064895104895105, + "grad_norm": 0.6058791585155276, + "learning_rate": 5.364963387682295e-06, + "loss": 0.047736674547195435, + "step": 3692 + }, + { + "epoch": 2.0654545454545454, + "grad_norm": 0.23607878202573318, + "learning_rate": 5.359197331118975e-06, + "loss": 0.15696394443511963, + "step": 3693 + }, + { + "epoch": 2.066013986013986, + "grad_norm": 0.2966181990404528, + "learning_rate": 5.353433240350694e-06, + "loss": 0.025747347623109818, + "step": 3694 + }, + { + "epoch": 2.0665734265734264, + "grad_norm": 0.3242643003404108, + "learning_rate": 5.3476711178190646e-06, + "loss": 0.06233817711472511, + "step": 3695 + }, + { + "epoch": 2.0671328671328673, + "grad_norm": 0.5130461805871928, + "learning_rate": 5.341910965964847e-06, + "loss": 0.07642102241516113, + "step": 3696 + }, + { + "epoch": 2.0676923076923077, + "grad_norm": 0.5571688174170879, + "learning_rate": 5.336152787227998e-06, + "loss": 0.095681332051754, + "step": 3697 + }, + { + "epoch": 2.068251748251748, + "grad_norm": 0.2600564942004817, + "learning_rate": 5.33039658404761e-06, + "loss": 0.01283261552453041, + "step": 3698 + }, + { + "epoch": 2.0688111888111886, + "grad_norm": 0.40319415687923316, + "learning_rate": 5.324642358861953e-06, + "loss": 0.04589240625500679, + "step": 3699 + }, + { + "epoch": 2.0693706293706295, + "grad_norm": 0.5248797240555537, + "learning_rate": 5.3188901141084595e-06, + "loss": 0.1537456512451172, + "step": 3700 + }, + { + "epoch": 2.06993006993007, + "grad_norm": 0.4090965940361953, + "learning_rate": 5.31313985222371e-06, + "loss": 0.08238714933395386, + "step": 3701 + }, + { + "epoch": 2.0704895104895105, + "grad_norm": 0.4054025947465797, + "learning_rate": 5.307391575643467e-06, + "loss": 0.159087136387825, + "step": 3702 + }, + { + "epoch": 2.071048951048951, + "grad_norm": 0.24142250802755405, + "learning_rate": 5.301645286802631e-06, + "loss": 0.02902410924434662, + "step": 3703 + }, + { + "epoch": 2.0716083916083914, + "grad_norm": 0.4300342453502428, + "learning_rate": 5.295900988135272e-06, + "loss": 0.08081980049610138, + "step": 3704 + }, + { + "epoch": 2.0721678321678323, + "grad_norm": 0.2130032930338116, + "learning_rate": 5.29015868207462e-06, + "loss": 0.009823895990848541, + "step": 3705 + }, + { + "epoch": 2.0727272727272728, + "grad_norm": 0.5974479428406126, + "learning_rate": 5.284418371053039e-06, + "loss": 0.1737496256828308, + "step": 3706 + }, + { + "epoch": 2.073286713286713, + "grad_norm": 0.355739100123085, + "learning_rate": 5.2786800575020855e-06, + "loss": 0.02741951495409012, + "step": 3707 + }, + { + "epoch": 2.0738461538461537, + "grad_norm": 0.3788479267141794, + "learning_rate": 5.2729437438524335e-06, + "loss": 0.03793314844369888, + "step": 3708 + }, + { + "epoch": 2.0744055944055946, + "grad_norm": 0.35463595855975216, + "learning_rate": 5.267209432533929e-06, + "loss": 0.15087255835533142, + "step": 3709 + }, + { + "epoch": 2.074965034965035, + "grad_norm": 0.23743059479925457, + "learning_rate": 5.261477125975571e-06, + "loss": 0.025123829022049904, + "step": 3710 + }, + { + "epoch": 2.0755244755244755, + "grad_norm": 0.3623956298769633, + "learning_rate": 5.25574682660549e-06, + "loss": 0.05077960714697838, + "step": 3711 + }, + { + "epoch": 2.076083916083916, + "grad_norm": 0.41520368154969933, + "learning_rate": 5.250018536851e-06, + "loss": 0.17125150561332703, + "step": 3712 + }, + { + "epoch": 2.0766433566433564, + "grad_norm": 0.47190652002498407, + "learning_rate": 5.244292259138528e-06, + "loss": 0.06280983239412308, + "step": 3713 + }, + { + "epoch": 2.0772027972027973, + "grad_norm": 0.2917179179044393, + "learning_rate": 5.238567995893671e-06, + "loss": 0.030946964398026466, + "step": 3714 + }, + { + "epoch": 2.077762237762238, + "grad_norm": 0.41523135127783195, + "learning_rate": 5.23284574954117e-06, + "loss": 0.06303784251213074, + "step": 3715 + }, + { + "epoch": 2.0783216783216782, + "grad_norm": 0.4982074484604896, + "learning_rate": 5.2271255225049e-06, + "loss": 0.16512702405452728, + "step": 3716 + }, + { + "epoch": 2.0788811188811187, + "grad_norm": 0.3916691250204122, + "learning_rate": 5.221407317207894e-06, + "loss": 0.15320153534412384, + "step": 3717 + }, + { + "epoch": 2.0794405594405596, + "grad_norm": 0.31498819174864545, + "learning_rate": 5.2156911360723204e-06, + "loss": 0.016733992844820023, + "step": 3718 + }, + { + "epoch": 2.08, + "grad_norm": 0.9731971953375126, + "learning_rate": 5.2099769815194975e-06, + "loss": 0.17772963643074036, + "step": 3719 + }, + { + "epoch": 2.0805594405594405, + "grad_norm": 0.3919125559050478, + "learning_rate": 5.2042648559698785e-06, + "loss": 0.059646524488925934, + "step": 3720 + }, + { + "epoch": 2.081118881118881, + "grad_norm": 0.47511146549886973, + "learning_rate": 5.198554761843062e-06, + "loss": 0.10271695256233215, + "step": 3721 + }, + { + "epoch": 2.0816783216783215, + "grad_norm": 0.5032170685530694, + "learning_rate": 5.192846701557779e-06, + "loss": 0.053607027977705, + "step": 3722 + }, + { + "epoch": 2.0822377622377624, + "grad_norm": 0.5530975841106688, + "learning_rate": 5.187140677531905e-06, + "loss": 0.09202791005373001, + "step": 3723 + }, + { + "epoch": 2.082797202797203, + "grad_norm": 0.5560355748477932, + "learning_rate": 5.181436692182453e-06, + "loss": 0.10471323132514954, + "step": 3724 + }, + { + "epoch": 2.0833566433566433, + "grad_norm": 0.5356098577805938, + "learning_rate": 5.175734747925572e-06, + "loss": 0.09282371401786804, + "step": 3725 + }, + { + "epoch": 2.0839160839160837, + "grad_norm": 0.4176588904635729, + "learning_rate": 5.170034847176545e-06, + "loss": 0.035135261714458466, + "step": 3726 + }, + { + "epoch": 2.0844755244755246, + "grad_norm": 0.2373064603626362, + "learning_rate": 5.164336992349786e-06, + "loss": 0.011200472712516785, + "step": 3727 + }, + { + "epoch": 2.085034965034965, + "grad_norm": 0.4205918248316928, + "learning_rate": 5.158641185858848e-06, + "loss": 0.04953043907880783, + "step": 3728 + }, + { + "epoch": 2.0855944055944056, + "grad_norm": 0.31996206955951545, + "learning_rate": 5.152947430116418e-06, + "loss": 0.07613462209701538, + "step": 3729 + }, + { + "epoch": 2.086153846153846, + "grad_norm": 0.585140161242383, + "learning_rate": 5.147255727534305e-06, + "loss": 0.13478463888168335, + "step": 3730 + }, + { + "epoch": 2.0867132867132865, + "grad_norm": 0.4397306316094135, + "learning_rate": 5.141566080523463e-06, + "loss": 0.08629865944385529, + "step": 3731 + }, + { + "epoch": 2.0872727272727274, + "grad_norm": 0.45623252612099474, + "learning_rate": 5.135878491493956e-06, + "loss": 0.06047341600060463, + "step": 3732 + }, + { + "epoch": 2.087832167832168, + "grad_norm": 0.8139877849948387, + "learning_rate": 5.130192962854992e-06, + "loss": 0.1401153951883316, + "step": 3733 + }, + { + "epoch": 2.0883916083916083, + "grad_norm": 0.7231515528317979, + "learning_rate": 5.124509497014901e-06, + "loss": 0.1497367024421692, + "step": 3734 + }, + { + "epoch": 2.0889510489510488, + "grad_norm": 0.49188631981302744, + "learning_rate": 5.118828096381139e-06, + "loss": 0.12784309685230255, + "step": 3735 + }, + { + "epoch": 2.0895104895104897, + "grad_norm": 0.6912538317073462, + "learning_rate": 5.113148763360289e-06, + "loss": 0.21873131394386292, + "step": 3736 + }, + { + "epoch": 2.09006993006993, + "grad_norm": 0.531583772649628, + "learning_rate": 5.107471500358048e-06, + "loss": 0.09356757998466492, + "step": 3737 + }, + { + "epoch": 2.0906293706293706, + "grad_norm": 0.4967114770920321, + "learning_rate": 5.1017963097792585e-06, + "loss": 0.14247947931289673, + "step": 3738 + }, + { + "epoch": 2.091188811188811, + "grad_norm": 0.5950521095397087, + "learning_rate": 5.096123194027864e-06, + "loss": 0.14984065294265747, + "step": 3739 + }, + { + "epoch": 2.091748251748252, + "grad_norm": 1.264844524371985, + "learning_rate": 5.0904521555069306e-06, + "loss": 0.09576083719730377, + "step": 3740 + }, + { + "epoch": 2.0923076923076924, + "grad_norm": 0.49747986423415574, + "learning_rate": 5.084783196618662e-06, + "loss": 0.04242061451077461, + "step": 3741 + }, + { + "epoch": 2.092867132867133, + "grad_norm": 0.789415317260481, + "learning_rate": 5.0791163197643615e-06, + "loss": 0.14414675533771515, + "step": 3742 + }, + { + "epoch": 2.0934265734265733, + "grad_norm": 0.6813495076345872, + "learning_rate": 5.0734515273444585e-06, + "loss": 0.12326627224683762, + "step": 3743 + }, + { + "epoch": 2.093986013986014, + "grad_norm": 0.16370956540438242, + "learning_rate": 5.067788821758506e-06, + "loss": 0.019266968593001366, + "step": 3744 + }, + { + "epoch": 2.0945454545454547, + "grad_norm": 0.4374711390566486, + "learning_rate": 5.0621282054051536e-06, + "loss": 0.048575010150671005, + "step": 3745 + }, + { + "epoch": 2.095104895104895, + "grad_norm": 0.35081009527722423, + "learning_rate": 5.056469680682194e-06, + "loss": 0.04107515886425972, + "step": 3746 + }, + { + "epoch": 2.0956643356643356, + "grad_norm": 0.6562853012383304, + "learning_rate": 5.050813249986508e-06, + "loss": 0.07658736407756805, + "step": 3747 + }, + { + "epoch": 2.096223776223776, + "grad_norm": 0.5908549866414952, + "learning_rate": 5.0451589157141035e-06, + "loss": 0.10319835692644119, + "step": 3748 + }, + { + "epoch": 2.096783216783217, + "grad_norm": 0.6002761284010847, + "learning_rate": 5.0395066802601e-06, + "loss": 0.18548107147216797, + "step": 3749 + }, + { + "epoch": 2.0973426573426575, + "grad_norm": 0.37992393936333196, + "learning_rate": 5.033856546018714e-06, + "loss": 0.05447522923350334, + "step": 3750 + }, + { + "epoch": 2.097902097902098, + "grad_norm": 0.6846206429133925, + "learning_rate": 5.028208515383297e-06, + "loss": 0.1492200642824173, + "step": 3751 + }, + { + "epoch": 2.0984615384615384, + "grad_norm": 0.33261590918083467, + "learning_rate": 5.022562590746285e-06, + "loss": 0.02554817497730255, + "step": 3752 + }, + { + "epoch": 2.099020979020979, + "grad_norm": 0.33174291196653616, + "learning_rate": 5.016918774499235e-06, + "loss": 0.060051798820495605, + "step": 3753 + }, + { + "epoch": 2.0995804195804197, + "grad_norm": 0.6717574905670267, + "learning_rate": 5.011277069032809e-06, + "loss": 0.23002095520496368, + "step": 3754 + }, + { + "epoch": 2.10013986013986, + "grad_norm": 0.6753700535377809, + "learning_rate": 5.005637476736773e-06, + "loss": 0.13818466663360596, + "step": 3755 + }, + { + "epoch": 2.1006993006993007, + "grad_norm": 0.9703323686737847, + "learning_rate": 5.000000000000003e-06, + "loss": 0.12224321067333221, + "step": 3756 + }, + { + "epoch": 2.101258741258741, + "grad_norm": 0.2770874538393943, + "learning_rate": 4.9943646412104665e-06, + "loss": 0.009026276879012585, + "step": 3757 + }, + { + "epoch": 2.101818181818182, + "grad_norm": 0.5016024072570013, + "learning_rate": 4.9887314027552455e-06, + "loss": 0.042649008333683014, + "step": 3758 + }, + { + "epoch": 2.1023776223776225, + "grad_norm": 0.5982271943543598, + "learning_rate": 4.9831002870205216e-06, + "loss": 0.12513811886310577, + "step": 3759 + }, + { + "epoch": 2.102937062937063, + "grad_norm": 0.4462739006792054, + "learning_rate": 4.977471296391575e-06, + "loss": 0.04833667725324631, + "step": 3760 + }, + { + "epoch": 2.1034965034965034, + "grad_norm": 0.2874285865428386, + "learning_rate": 4.971844433252791e-06, + "loss": 0.03420846164226532, + "step": 3761 + }, + { + "epoch": 2.104055944055944, + "grad_norm": 0.7090214984289172, + "learning_rate": 4.966219699987643e-06, + "loss": 0.08666019141674042, + "step": 3762 + }, + { + "epoch": 2.1046153846153848, + "grad_norm": 0.4266949554286465, + "learning_rate": 4.960597098978711e-06, + "loss": 0.07298225909471512, + "step": 3763 + }, + { + "epoch": 2.1051748251748252, + "grad_norm": 0.3912766802897396, + "learning_rate": 4.954976632607669e-06, + "loss": 0.0925387293100357, + "step": 3764 + }, + { + "epoch": 2.1057342657342657, + "grad_norm": 0.6600422670850886, + "learning_rate": 4.949358303255293e-06, + "loss": 0.10714008659124374, + "step": 3765 + }, + { + "epoch": 2.106293706293706, + "grad_norm": 0.20524293525787002, + "learning_rate": 4.943742113301437e-06, + "loss": 0.023769894614815712, + "step": 3766 + }, + { + "epoch": 2.106853146853147, + "grad_norm": 0.29151278200801134, + "learning_rate": 4.9381280651250685e-06, + "loss": 0.01928844302892685, + "step": 3767 + }, + { + "epoch": 2.1074125874125875, + "grad_norm": 0.4562617415369419, + "learning_rate": 4.932516161104235e-06, + "loss": 0.06522198766469955, + "step": 3768 + }, + { + "epoch": 2.107972027972028, + "grad_norm": 0.36721645052664, + "learning_rate": 4.926906403616081e-06, + "loss": 0.05866191163659096, + "step": 3769 + }, + { + "epoch": 2.1085314685314684, + "grad_norm": 0.53461446634668, + "learning_rate": 4.921298795036845e-06, + "loss": 0.16815561056137085, + "step": 3770 + }, + { + "epoch": 2.109090909090909, + "grad_norm": 0.41060116018697684, + "learning_rate": 4.9156933377418425e-06, + "loss": 0.08497156202793121, + "step": 3771 + }, + { + "epoch": 2.10965034965035, + "grad_norm": 0.433693206230795, + "learning_rate": 4.91009003410549e-06, + "loss": 0.13801153004169464, + "step": 3772 + }, + { + "epoch": 2.1102097902097903, + "grad_norm": 0.4911836522930897, + "learning_rate": 4.9044888865012884e-06, + "loss": 0.07784494012594223, + "step": 3773 + }, + { + "epoch": 2.1107692307692307, + "grad_norm": 0.40683713133752514, + "learning_rate": 4.898889897301824e-06, + "loss": 0.03854656592011452, + "step": 3774 + }, + { + "epoch": 2.111328671328671, + "grad_norm": 0.33189599496956135, + "learning_rate": 4.893293068878774e-06, + "loss": 0.117255300283432, + "step": 3775 + }, + { + "epoch": 2.111888111888112, + "grad_norm": 0.45366381004718215, + "learning_rate": 4.887698403602884e-06, + "loss": 0.09304417669773102, + "step": 3776 + }, + { + "epoch": 2.1124475524475526, + "grad_norm": 0.33610822085475284, + "learning_rate": 4.88210590384401e-06, + "loss": 0.03790184110403061, + "step": 3777 + }, + { + "epoch": 2.113006993006993, + "grad_norm": 0.6328726229147658, + "learning_rate": 4.876515571971065e-06, + "loss": 0.15130586922168732, + "step": 3778 + }, + { + "epoch": 2.1135664335664335, + "grad_norm": 0.4212184554478342, + "learning_rate": 4.870927410352059e-06, + "loss": 0.0512462854385376, + "step": 3779 + }, + { + "epoch": 2.114125874125874, + "grad_norm": 0.33209830928085926, + "learning_rate": 4.865341421354081e-06, + "loss": 0.03197665885090828, + "step": 3780 + }, + { + "epoch": 2.114685314685315, + "grad_norm": 0.4436768743667384, + "learning_rate": 4.859757607343286e-06, + "loss": 0.04053480923175812, + "step": 3781 + }, + { + "epoch": 2.1152447552447553, + "grad_norm": 0.08122655799473173, + "learning_rate": 4.854175970684934e-06, + "loss": 0.0027559478767216206, + "step": 3782 + }, + { + "epoch": 2.1158041958041958, + "grad_norm": 0.21418846108405257, + "learning_rate": 4.848596513743336e-06, + "loss": 0.006786721758544445, + "step": 3783 + }, + { + "epoch": 2.1163636363636362, + "grad_norm": 0.5564948842737181, + "learning_rate": 4.843019238881894e-06, + "loss": 0.10572367161512375, + "step": 3784 + }, + { + "epoch": 2.116923076923077, + "grad_norm": 0.4819745625355441, + "learning_rate": 4.837444148463087e-06, + "loss": 0.07152760773897171, + "step": 3785 + }, + { + "epoch": 2.1174825174825176, + "grad_norm": 0.3422066072045526, + "learning_rate": 4.831871244848455e-06, + "loss": 0.05206223577260971, + "step": 3786 + }, + { + "epoch": 2.118041958041958, + "grad_norm": 0.4542248733880757, + "learning_rate": 4.826300530398635e-06, + "loss": 0.09001053869724274, + "step": 3787 + }, + { + "epoch": 2.1186013986013985, + "grad_norm": 0.1867098858850836, + "learning_rate": 4.8207320074733136e-06, + "loss": 0.008701407350599766, + "step": 3788 + }, + { + "epoch": 2.119160839160839, + "grad_norm": 0.35258505096460835, + "learning_rate": 4.815165678431253e-06, + "loss": 0.03854415565729141, + "step": 3789 + }, + { + "epoch": 2.11972027972028, + "grad_norm": 0.19958651321589826, + "learning_rate": 4.809601545630305e-06, + "loss": 0.009942376986145973, + "step": 3790 + }, + { + "epoch": 2.1202797202797203, + "grad_norm": 0.696113538202477, + "learning_rate": 4.804039611427367e-06, + "loss": 0.13618825376033783, + "step": 3791 + }, + { + "epoch": 2.120839160839161, + "grad_norm": 0.31455574767085537, + "learning_rate": 4.798479878178421e-06, + "loss": 0.046507224440574646, + "step": 3792 + }, + { + "epoch": 2.1213986013986013, + "grad_norm": 0.42274780344011437, + "learning_rate": 4.792922348238508e-06, + "loss": 0.1054900586605072, + "step": 3793 + }, + { + "epoch": 2.121958041958042, + "grad_norm": 0.3516097070418653, + "learning_rate": 4.787367023961741e-06, + "loss": 0.02644207701086998, + "step": 3794 + }, + { + "epoch": 2.1225174825174826, + "grad_norm": 0.5382554157735704, + "learning_rate": 4.781813907701301e-06, + "loss": 0.13420142233371735, + "step": 3795 + }, + { + "epoch": 2.123076923076923, + "grad_norm": 0.33949371366087006, + "learning_rate": 4.7762630018094216e-06, + "loss": 0.09387016296386719, + "step": 3796 + }, + { + "epoch": 2.1236363636363635, + "grad_norm": 0.38265930489966254, + "learning_rate": 4.77071430863741e-06, + "loss": 0.04379844665527344, + "step": 3797 + }, + { + "epoch": 2.124195804195804, + "grad_norm": 0.3281385244699593, + "learning_rate": 4.765167830535637e-06, + "loss": 0.05403991416096687, + "step": 3798 + }, + { + "epoch": 2.124755244755245, + "grad_norm": 0.6474416086524375, + "learning_rate": 4.759623569853528e-06, + "loss": 0.1604430228471756, + "step": 3799 + }, + { + "epoch": 2.1253146853146854, + "grad_norm": 0.4519076473523307, + "learning_rate": 4.754081528939581e-06, + "loss": 0.10386520624160767, + "step": 3800 + }, + { + "epoch": 2.125874125874126, + "grad_norm": 0.4231159120754047, + "learning_rate": 4.748541710141339e-06, + "loss": 0.11015531420707703, + "step": 3801 + }, + { + "epoch": 2.1264335664335663, + "grad_norm": 0.3663514603920324, + "learning_rate": 4.7430041158054105e-06, + "loss": 0.12145122140645981, + "step": 3802 + }, + { + "epoch": 2.126993006993007, + "grad_norm": 0.16043461978547224, + "learning_rate": 4.737468748277465e-06, + "loss": 0.00788156595081091, + "step": 3803 + }, + { + "epoch": 2.1275524475524477, + "grad_norm": 0.4146762026881823, + "learning_rate": 4.731935609902225e-06, + "loss": 0.04790748655796051, + "step": 3804 + }, + { + "epoch": 2.128111888111888, + "grad_norm": 0.3930811061312869, + "learning_rate": 4.726404703023474e-06, + "loss": 0.040815550833940506, + "step": 3805 + }, + { + "epoch": 2.1286713286713286, + "grad_norm": 0.46626264160881137, + "learning_rate": 4.720876029984037e-06, + "loss": 0.04240987449884415, + "step": 3806 + }, + { + "epoch": 2.129230769230769, + "grad_norm": 0.23269809351400753, + "learning_rate": 4.715349593125805e-06, + "loss": 0.013739352114498615, + "step": 3807 + }, + { + "epoch": 2.12979020979021, + "grad_norm": 0.4074640712348649, + "learning_rate": 4.709825394789721e-06, + "loss": 0.09310592710971832, + "step": 3808 + }, + { + "epoch": 2.1303496503496504, + "grad_norm": 0.2625608846675623, + "learning_rate": 4.704303437315774e-06, + "loss": 0.04799744114279747, + "step": 3809 + }, + { + "epoch": 2.130909090909091, + "grad_norm": 0.6038500830923805, + "learning_rate": 4.69878372304301e-06, + "loss": 0.12010888010263443, + "step": 3810 + }, + { + "epoch": 2.1314685314685313, + "grad_norm": 0.46867620898177714, + "learning_rate": 4.693266254309522e-06, + "loss": 0.03949858993291855, + "step": 3811 + }, + { + "epoch": 2.1320279720279722, + "grad_norm": 0.3920767103327594, + "learning_rate": 4.687751033452447e-06, + "loss": 0.1470131129026413, + "step": 3812 + }, + { + "epoch": 2.1325874125874127, + "grad_norm": 0.5017840221353354, + "learning_rate": 4.6822380628079755e-06, + "loss": 0.1138114482164383, + "step": 3813 + }, + { + "epoch": 2.133146853146853, + "grad_norm": 1.0616054924319234, + "learning_rate": 4.676727344711351e-06, + "loss": 0.07835336029529572, + "step": 3814 + }, + { + "epoch": 2.1337062937062936, + "grad_norm": 0.23632555700732388, + "learning_rate": 4.67121888149684e-06, + "loss": 0.032216355204582214, + "step": 3815 + }, + { + "epoch": 2.134265734265734, + "grad_norm": 0.5204545998090165, + "learning_rate": 4.665712675497788e-06, + "loss": 0.09749799966812134, + "step": 3816 + }, + { + "epoch": 2.134825174825175, + "grad_norm": 0.34655962613260266, + "learning_rate": 4.660208729046552e-06, + "loss": 0.07754640281200409, + "step": 3817 + }, + { + "epoch": 2.1353846153846154, + "grad_norm": 0.647679232351044, + "learning_rate": 4.654707044474551e-06, + "loss": 0.1486811339855194, + "step": 3818 + }, + { + "epoch": 2.135944055944056, + "grad_norm": 0.4219587193052221, + "learning_rate": 4.649207624112243e-06, + "loss": 0.06307768821716309, + "step": 3819 + }, + { + "epoch": 2.1365034965034964, + "grad_norm": 0.5598186745981235, + "learning_rate": 4.6437104702891146e-06, + "loss": 0.12122844159603119, + "step": 3820 + }, + { + "epoch": 2.1370629370629373, + "grad_norm": 0.13194817461215327, + "learning_rate": 4.638215585333717e-06, + "loss": 0.002600582782179117, + "step": 3821 + }, + { + "epoch": 2.1376223776223777, + "grad_norm": 0.4515996755429112, + "learning_rate": 4.6327229715736125e-06, + "loss": 0.09703333675861359, + "step": 3822 + }, + { + "epoch": 2.138181818181818, + "grad_norm": 0.4944802635887013, + "learning_rate": 4.627232631335419e-06, + "loss": 0.11477223038673401, + "step": 3823 + }, + { + "epoch": 2.1387412587412586, + "grad_norm": 0.5291216998433278, + "learning_rate": 4.621744566944791e-06, + "loss": 0.137488454580307, + "step": 3824 + }, + { + "epoch": 2.139300699300699, + "grad_norm": 0.41362351794453095, + "learning_rate": 4.616258780726404e-06, + "loss": 0.06826088577508926, + "step": 3825 + }, + { + "epoch": 2.13986013986014, + "grad_norm": 0.41473780312733066, + "learning_rate": 4.610775275003992e-06, + "loss": 0.036990582942962646, + "step": 3826 + }, + { + "epoch": 2.1404195804195805, + "grad_norm": 0.7269721293777449, + "learning_rate": 4.605294052100301e-06, + "loss": 0.23528940975666046, + "step": 3827 + }, + { + "epoch": 2.140979020979021, + "grad_norm": 0.3737834767095868, + "learning_rate": 4.599815114337122e-06, + "loss": 0.05900619924068451, + "step": 3828 + }, + { + "epoch": 2.1415384615384614, + "grad_norm": 0.5039629283889282, + "learning_rate": 4.594338464035278e-06, + "loss": 0.08169350028038025, + "step": 3829 + }, + { + "epoch": 2.1420979020979023, + "grad_norm": 0.5614272189355393, + "learning_rate": 4.588864103514612e-06, + "loss": 0.16644254326820374, + "step": 3830 + }, + { + "epoch": 2.1426573426573428, + "grad_norm": 0.5320378581260635, + "learning_rate": 4.583392035094017e-06, + "loss": 0.07631094753742218, + "step": 3831 + }, + { + "epoch": 2.143216783216783, + "grad_norm": 0.6262081200120164, + "learning_rate": 4.577922261091394e-06, + "loss": 0.08085634559392929, + "step": 3832 + }, + { + "epoch": 2.1437762237762237, + "grad_norm": 0.44182703189087263, + "learning_rate": 4.572454783823685e-06, + "loss": 0.08495327830314636, + "step": 3833 + }, + { + "epoch": 2.144335664335664, + "grad_norm": 0.4450053526476975, + "learning_rate": 4.56698960560686e-06, + "loss": 0.05321191996335983, + "step": 3834 + }, + { + "epoch": 2.144895104895105, + "grad_norm": 0.48239743137948027, + "learning_rate": 4.561526728755897e-06, + "loss": 0.12114525586366653, + "step": 3835 + }, + { + "epoch": 2.1454545454545455, + "grad_norm": 0.5728831572983879, + "learning_rate": 4.556066155584832e-06, + "loss": 0.2179345190525055, + "step": 3836 + }, + { + "epoch": 2.146013986013986, + "grad_norm": 0.6499295341955195, + "learning_rate": 4.550607888406692e-06, + "loss": 0.11883588880300522, + "step": 3837 + }, + { + "epoch": 2.1465734265734264, + "grad_norm": 0.6183690540941805, + "learning_rate": 4.545151929533546e-06, + "loss": 0.10780727118253708, + "step": 3838 + }, + { + "epoch": 2.1471328671328673, + "grad_norm": 0.5019199743537712, + "learning_rate": 4.539698281276483e-06, + "loss": 0.07032984495162964, + "step": 3839 + }, + { + "epoch": 2.147692307692308, + "grad_norm": 0.35024170934398385, + "learning_rate": 4.5342469459456065e-06, + "loss": 0.05536240339279175, + "step": 3840 + }, + { + "epoch": 2.1482517482517482, + "grad_norm": 0.26197289492990095, + "learning_rate": 4.5287979258500456e-06, + "loss": 0.03774556517601013, + "step": 3841 + }, + { + "epoch": 2.1488111888111887, + "grad_norm": 0.25235050523940367, + "learning_rate": 4.523351223297949e-06, + "loss": 0.015268439427018166, + "step": 3842 + }, + { + "epoch": 2.149370629370629, + "grad_norm": 0.5448974490876621, + "learning_rate": 4.517906840596482e-06, + "loss": 0.12368149310350418, + "step": 3843 + }, + { + "epoch": 2.14993006993007, + "grad_norm": 0.17772913510186134, + "learning_rate": 4.512464780051831e-06, + "loss": 0.004211365710943937, + "step": 3844 + }, + { + "epoch": 2.1504895104895105, + "grad_norm": 0.5333995740876823, + "learning_rate": 4.507025043969188e-06, + "loss": 0.04077071696519852, + "step": 3845 + }, + { + "epoch": 2.151048951048951, + "grad_norm": 0.614557754341535, + "learning_rate": 4.501587634652773e-06, + "loss": 0.13250434398651123, + "step": 3846 + }, + { + "epoch": 2.1516083916083915, + "grad_norm": 0.5147776666573411, + "learning_rate": 4.496152554405815e-06, + "loss": 0.09368059039115906, + "step": 3847 + }, + { + "epoch": 2.1521678321678324, + "grad_norm": 0.3628310944203244, + "learning_rate": 4.490719805530554e-06, + "loss": 0.05959971621632576, + "step": 3848 + }, + { + "epoch": 2.152727272727273, + "grad_norm": 0.5153859159613096, + "learning_rate": 4.485289390328248e-06, + "loss": 0.07026205211877823, + "step": 3849 + }, + { + "epoch": 2.1532867132867133, + "grad_norm": 0.31394860516477124, + "learning_rate": 4.479861311099165e-06, + "loss": 0.0821579173207283, + "step": 3850 + }, + { + "epoch": 2.1538461538461537, + "grad_norm": 0.4400274521738231, + "learning_rate": 4.474435570142577e-06, + "loss": 0.24506758153438568, + "step": 3851 + }, + { + "epoch": 2.154405594405594, + "grad_norm": 1.6739308156936705, + "learning_rate": 4.469012169756773e-06, + "loss": 0.1311085820198059, + "step": 3852 + }, + { + "epoch": 2.154965034965035, + "grad_norm": 0.5015669429635032, + "learning_rate": 4.463591112239048e-06, + "loss": 0.06196563318371773, + "step": 3853 + }, + { + "epoch": 2.1555244755244756, + "grad_norm": 0.5893217240043253, + "learning_rate": 4.458172399885705e-06, + "loss": 0.13947096467018127, + "step": 3854 + }, + { + "epoch": 2.156083916083916, + "grad_norm": 0.19588068096202657, + "learning_rate": 4.452756034992057e-06, + "loss": 0.014179615303874016, + "step": 3855 + }, + { + "epoch": 2.1566433566433565, + "grad_norm": 0.4651936959008058, + "learning_rate": 4.447342019852412e-06, + "loss": 0.10229015350341797, + "step": 3856 + }, + { + "epoch": 2.1572027972027974, + "grad_norm": 0.48962476479392664, + "learning_rate": 4.441930356760094e-06, + "loss": 0.07068829238414764, + "step": 3857 + }, + { + "epoch": 2.157762237762238, + "grad_norm": 0.4965650387232792, + "learning_rate": 4.436521048007424e-06, + "loss": 0.13860081136226654, + "step": 3858 + }, + { + "epoch": 2.1583216783216783, + "grad_norm": 0.8592392906485635, + "learning_rate": 4.431114095885731e-06, + "loss": 0.08895541727542877, + "step": 3859 + }, + { + "epoch": 2.1588811188811188, + "grad_norm": 0.4620992469891322, + "learning_rate": 4.425709502685344e-06, + "loss": 0.19275455176830292, + "step": 3860 + }, + { + "epoch": 2.1594405594405592, + "grad_norm": 0.40097477757060684, + "learning_rate": 4.420307270695586e-06, + "loss": 0.1312236189842224, + "step": 3861 + }, + { + "epoch": 2.16, + "grad_norm": 0.3635615085838854, + "learning_rate": 4.414907402204789e-06, + "loss": 0.06387532502412796, + "step": 3862 + }, + { + "epoch": 2.1605594405594406, + "grad_norm": 0.3792934226216701, + "learning_rate": 4.409509899500284e-06, + "loss": 0.08187203854322433, + "step": 3863 + }, + { + "epoch": 2.161118881118881, + "grad_norm": 0.4954881384970882, + "learning_rate": 4.404114764868384e-06, + "loss": 0.06419216096401215, + "step": 3864 + }, + { + "epoch": 2.1616783216783215, + "grad_norm": 0.21568140015737655, + "learning_rate": 4.398722000594427e-06, + "loss": 0.06822659820318222, + "step": 3865 + }, + { + "epoch": 2.1622377622377624, + "grad_norm": 0.4608137292700796, + "learning_rate": 4.393331608962718e-06, + "loss": 0.06794237345457077, + "step": 3866 + }, + { + "epoch": 2.162797202797203, + "grad_norm": 0.5811169116721158, + "learning_rate": 4.387943592256577e-06, + "loss": 0.13654986023902893, + "step": 3867 + }, + { + "epoch": 2.1633566433566433, + "grad_norm": 0.36493470409727374, + "learning_rate": 4.382557952758313e-06, + "loss": 0.029762111604213715, + "step": 3868 + }, + { + "epoch": 2.163916083916084, + "grad_norm": 0.433356026418893, + "learning_rate": 4.3771746927492145e-06, + "loss": 0.0476355217397213, + "step": 3869 + }, + { + "epoch": 2.1644755244755243, + "grad_norm": 0.49210536252786563, + "learning_rate": 4.371793814509589e-06, + "loss": 0.07634837925434113, + "step": 3870 + }, + { + "epoch": 2.165034965034965, + "grad_norm": 0.4678666765933248, + "learning_rate": 4.366415320318708e-06, + "loss": 0.036532625555992126, + "step": 3871 + }, + { + "epoch": 2.1655944055944056, + "grad_norm": 0.3589595354321799, + "learning_rate": 4.36103921245485e-06, + "loss": 0.03828028589487076, + "step": 3872 + }, + { + "epoch": 2.166153846153846, + "grad_norm": 0.4158552145751515, + "learning_rate": 4.355665493195279e-06, + "loss": 0.044713299721479416, + "step": 3873 + }, + { + "epoch": 2.1667132867132866, + "grad_norm": 0.3791754507597784, + "learning_rate": 4.350294164816237e-06, + "loss": 0.06631101667881012, + "step": 3874 + }, + { + "epoch": 2.1672727272727275, + "grad_norm": 0.4419945896811687, + "learning_rate": 4.344925229592975e-06, + "loss": 0.1317673623561859, + "step": 3875 + }, + { + "epoch": 2.167832167832168, + "grad_norm": 0.435997145286077, + "learning_rate": 4.339558689799708e-06, + "loss": 0.06341883540153503, + "step": 3876 + }, + { + "epoch": 2.1683916083916084, + "grad_norm": 0.46494250761638206, + "learning_rate": 4.334194547709648e-06, + "loss": 0.10478334128856659, + "step": 3877 + }, + { + "epoch": 2.168951048951049, + "grad_norm": 0.44481634379465523, + "learning_rate": 4.3288328055949935e-06, + "loss": 0.10518811643123627, + "step": 3878 + }, + { + "epoch": 2.1695104895104893, + "grad_norm": 0.5350031067003866, + "learning_rate": 4.323473465726911e-06, + "loss": 0.11992601305246353, + "step": 3879 + }, + { + "epoch": 2.17006993006993, + "grad_norm": 0.7520339293703405, + "learning_rate": 4.3181165303755725e-06, + "loss": 0.1269116997718811, + "step": 3880 + }, + { + "epoch": 2.1706293706293707, + "grad_norm": 0.5457672378885025, + "learning_rate": 4.312762001810111e-06, + "loss": 0.11786731332540512, + "step": 3881 + }, + { + "epoch": 2.171188811188811, + "grad_norm": 0.6160913680254718, + "learning_rate": 4.3074098822986525e-06, + "loss": 0.08864185214042664, + "step": 3882 + }, + { + "epoch": 2.1717482517482516, + "grad_norm": 0.4079189769892911, + "learning_rate": 4.302060174108299e-06, + "loss": 0.056865446269512177, + "step": 3883 + }, + { + "epoch": 2.1723076923076925, + "grad_norm": 0.5927862704604111, + "learning_rate": 4.296712879505122e-06, + "loss": 0.15149927139282227, + "step": 3884 + }, + { + "epoch": 2.172867132867133, + "grad_norm": 0.33682290395193804, + "learning_rate": 4.2913680007541914e-06, + "loss": 0.05931822583079338, + "step": 3885 + }, + { + "epoch": 2.1734265734265734, + "grad_norm": 0.45591297684664156, + "learning_rate": 4.286025540119534e-06, + "loss": 0.026843927800655365, + "step": 3886 + }, + { + "epoch": 2.173986013986014, + "grad_norm": 0.3167897143633637, + "learning_rate": 4.28068549986416e-06, + "loss": 0.06744428724050522, + "step": 3887 + }, + { + "epoch": 2.174545454545455, + "grad_norm": 0.3700282338839621, + "learning_rate": 4.2753478822500545e-06, + "loss": 0.1777801513671875, + "step": 3888 + }, + { + "epoch": 2.1751048951048952, + "grad_norm": 0.5998773016911984, + "learning_rate": 4.270012689538182e-06, + "loss": 0.135552316904068, + "step": 3889 + }, + { + "epoch": 2.1756643356643357, + "grad_norm": 0.5789219510925372, + "learning_rate": 4.264679923988465e-06, + "loss": 0.12418273091316223, + "step": 3890 + }, + { + "epoch": 2.176223776223776, + "grad_norm": 0.33672146398264763, + "learning_rate": 4.25934958785981e-06, + "loss": 0.028251193463802338, + "step": 3891 + }, + { + "epoch": 2.1767832167832166, + "grad_norm": 0.5731696745285129, + "learning_rate": 4.254021683410092e-06, + "loss": 0.15751181542873383, + "step": 3892 + }, + { + "epoch": 2.1773426573426575, + "grad_norm": 0.42179681731896257, + "learning_rate": 4.248696212896154e-06, + "loss": 0.026731228455901146, + "step": 3893 + }, + { + "epoch": 2.177902097902098, + "grad_norm": 0.19726646007500306, + "learning_rate": 4.243373178573814e-06, + "loss": 0.008842997252941132, + "step": 3894 + }, + { + "epoch": 2.1784615384615384, + "grad_norm": 0.4825833101269908, + "learning_rate": 4.2380525826978445e-06, + "loss": 0.07987696677446365, + "step": 3895 + }, + { + "epoch": 2.179020979020979, + "grad_norm": 0.3019418274768594, + "learning_rate": 4.232734427521999e-06, + "loss": 0.13472937047481537, + "step": 3896 + }, + { + "epoch": 2.17958041958042, + "grad_norm": 0.37793501769135235, + "learning_rate": 4.2274187152989905e-06, + "loss": 0.05649025738239288, + "step": 3897 + }, + { + "epoch": 2.1801398601398603, + "grad_norm": 0.5895779828450959, + "learning_rate": 4.2221054482805e-06, + "loss": 0.10967366397380829, + "step": 3898 + }, + { + "epoch": 2.1806993006993007, + "grad_norm": 0.4867417475066692, + "learning_rate": 4.216794628717175e-06, + "loss": 0.13834531605243683, + "step": 3899 + }, + { + "epoch": 2.181258741258741, + "grad_norm": 0.5737034061408215, + "learning_rate": 4.211486258858613e-06, + "loss": 0.1697239875793457, + "step": 3900 + }, + { + "epoch": 2.1818181818181817, + "grad_norm": 0.8074540874970116, + "learning_rate": 4.206180340953389e-06, + "loss": 0.15156517922878265, + "step": 3901 + }, + { + "epoch": 2.1823776223776226, + "grad_norm": 0.48813017762289307, + "learning_rate": 4.200876877249034e-06, + "loss": 0.10627919435501099, + "step": 3902 + }, + { + "epoch": 2.182937062937063, + "grad_norm": 2.346532869636276, + "learning_rate": 4.195575869992038e-06, + "loss": 0.32856282591819763, + "step": 3903 + }, + { + "epoch": 2.1834965034965035, + "grad_norm": 0.7139088415784709, + "learning_rate": 4.190277321427857e-06, + "loss": 0.2708978056907654, + "step": 3904 + }, + { + "epoch": 2.184055944055944, + "grad_norm": 0.372940063418746, + "learning_rate": 4.184981233800888e-06, + "loss": 0.06504981964826584, + "step": 3905 + }, + { + "epoch": 2.184615384615385, + "grad_norm": 0.33537230879631724, + "learning_rate": 4.179687609354512e-06, + "loss": 0.028646428138017654, + "step": 3906 + }, + { + "epoch": 2.1851748251748253, + "grad_norm": 0.49624165433877304, + "learning_rate": 4.174396450331043e-06, + "loss": 0.12064141780138016, + "step": 3907 + }, + { + "epoch": 2.1857342657342658, + "grad_norm": 0.5865613072545661, + "learning_rate": 4.169107758971764e-06, + "loss": 0.09545593708753586, + "step": 3908 + }, + { + "epoch": 2.1862937062937062, + "grad_norm": 0.4223178283563786, + "learning_rate": 4.163821537516911e-06, + "loss": 0.023391593247652054, + "step": 3909 + }, + { + "epoch": 2.1868531468531467, + "grad_norm": 0.5505216941416963, + "learning_rate": 4.158537788205661e-06, + "loss": 0.054874174296855927, + "step": 3910 + }, + { + "epoch": 2.1874125874125876, + "grad_norm": 0.36948819528086874, + "learning_rate": 4.153256513276169e-06, + "loss": 0.06249412149190903, + "step": 3911 + }, + { + "epoch": 2.187972027972028, + "grad_norm": 0.4960574245291073, + "learning_rate": 4.147977714965522e-06, + "loss": 0.0644707977771759, + "step": 3912 + }, + { + "epoch": 2.1885314685314685, + "grad_norm": 0.5713877270462516, + "learning_rate": 4.142701395509754e-06, + "loss": 0.205059215426445, + "step": 3913 + }, + { + "epoch": 2.189090909090909, + "grad_norm": 0.2386147239747397, + "learning_rate": 4.137427557143873e-06, + "loss": 0.020121503621339798, + "step": 3914 + }, + { + "epoch": 2.18965034965035, + "grad_norm": 0.7332311723125273, + "learning_rate": 4.132156202101809e-06, + "loss": 0.1439301073551178, + "step": 3915 + }, + { + "epoch": 2.1902097902097903, + "grad_norm": 0.46680517473527605, + "learning_rate": 4.126887332616459e-06, + "loss": 0.07918552309274673, + "step": 3916 + }, + { + "epoch": 2.190769230769231, + "grad_norm": 0.5046836655491164, + "learning_rate": 4.121620950919662e-06, + "loss": 0.10033179819583893, + "step": 3917 + }, + { + "epoch": 2.1913286713286713, + "grad_norm": 0.6501534927278353, + "learning_rate": 4.116357059242193e-06, + "loss": 0.11881642788648605, + "step": 3918 + }, + { + "epoch": 2.1918881118881117, + "grad_norm": 0.39967224274227664, + "learning_rate": 4.111095659813793e-06, + "loss": 0.039310112595558167, + "step": 3919 + }, + { + "epoch": 2.1924475524475526, + "grad_norm": 0.6282970400257136, + "learning_rate": 4.105836754863126e-06, + "loss": 0.08218531310558319, + "step": 3920 + }, + { + "epoch": 2.193006993006993, + "grad_norm": 0.7349527915327618, + "learning_rate": 4.10058034661781e-06, + "loss": 0.12492300570011139, + "step": 3921 + }, + { + "epoch": 2.1935664335664335, + "grad_norm": 0.4570257233084411, + "learning_rate": 4.095326437304412e-06, + "loss": 0.12985070049762726, + "step": 3922 + }, + { + "epoch": 2.194125874125874, + "grad_norm": 0.2924730104544794, + "learning_rate": 4.090075029148417e-06, + "loss": 0.058650773018598557, + "step": 3923 + }, + { + "epoch": 2.194685314685315, + "grad_norm": 0.3774920026787856, + "learning_rate": 4.084826124374284e-06, + "loss": 0.05051948502659798, + "step": 3924 + }, + { + "epoch": 2.1952447552447554, + "grad_norm": 0.44304383967194805, + "learning_rate": 4.07957972520538e-06, + "loss": 0.05628602206707001, + "step": 3925 + }, + { + "epoch": 2.195804195804196, + "grad_norm": 0.4859411685810675, + "learning_rate": 4.07433583386403e-06, + "loss": 0.09606659412384033, + "step": 3926 + }, + { + "epoch": 2.1963636363636363, + "grad_norm": 0.32101176042279045, + "learning_rate": 4.06909445257149e-06, + "loss": 0.09657777845859528, + "step": 3927 + }, + { + "epoch": 2.1969230769230768, + "grad_norm": 0.444578517060633, + "learning_rate": 4.0638555835479554e-06, + "loss": 0.0757531151175499, + "step": 3928 + }, + { + "epoch": 2.1974825174825177, + "grad_norm": 0.41483380526680597, + "learning_rate": 4.058619229012558e-06, + "loss": 0.08002664148807526, + "step": 3929 + }, + { + "epoch": 2.198041958041958, + "grad_norm": 0.23821140521855205, + "learning_rate": 4.053385391183354e-06, + "loss": 0.047871023416519165, + "step": 3930 + }, + { + "epoch": 2.1986013986013986, + "grad_norm": 0.3318706280808629, + "learning_rate": 4.048154072277348e-06, + "loss": 0.03471294417977333, + "step": 3931 + }, + { + "epoch": 2.199160839160839, + "grad_norm": 0.3319816564733304, + "learning_rate": 4.042925274510471e-06, + "loss": 0.037466805428266525, + "step": 3932 + }, + { + "epoch": 2.19972027972028, + "grad_norm": 1.3221934652761007, + "learning_rate": 4.037699000097585e-06, + "loss": 0.1302366554737091, + "step": 3933 + }, + { + "epoch": 2.2002797202797204, + "grad_norm": 0.2883972288219751, + "learning_rate": 4.032475251252489e-06, + "loss": 0.042289648205041885, + "step": 3934 + }, + { + "epoch": 2.200839160839161, + "grad_norm": 0.37944829854355294, + "learning_rate": 4.0272540301879006e-06, + "loss": 0.03058020770549774, + "step": 3935 + }, + { + "epoch": 2.2013986013986013, + "grad_norm": 0.6750001709148996, + "learning_rate": 4.022035339115479e-06, + "loss": 0.16856679320335388, + "step": 3936 + }, + { + "epoch": 2.201958041958042, + "grad_norm": 0.3270066067960322, + "learning_rate": 4.016819180245805e-06, + "loss": 0.06180114671587944, + "step": 3937 + }, + { + "epoch": 2.2025174825174827, + "grad_norm": 0.478621744882457, + "learning_rate": 4.011605555788392e-06, + "loss": 0.04430089890956879, + "step": 3938 + }, + { + "epoch": 2.203076923076923, + "grad_norm": 0.5627800245554797, + "learning_rate": 4.00639446795167e-06, + "loss": 0.15685822069644928, + "step": 3939 + }, + { + "epoch": 2.2036363636363636, + "grad_norm": 0.6487925237640466, + "learning_rate": 4.001185918943003e-06, + "loss": 0.0799047127366066, + "step": 3940 + }, + { + "epoch": 2.204195804195804, + "grad_norm": 0.4247911411585756, + "learning_rate": 3.99597991096868e-06, + "loss": 0.1177058070898056, + "step": 3941 + }, + { + "epoch": 2.204755244755245, + "grad_norm": 0.28600509169121135, + "learning_rate": 3.990776446233909e-06, + "loss": 0.04188351333141327, + "step": 3942 + }, + { + "epoch": 2.2053146853146854, + "grad_norm": 0.5376005867246233, + "learning_rate": 3.985575526942826e-06, + "loss": 0.0670405924320221, + "step": 3943 + }, + { + "epoch": 2.205874125874126, + "grad_norm": 0.36321145357965035, + "learning_rate": 3.980377155298477e-06, + "loss": 0.07518081367015839, + "step": 3944 + }, + { + "epoch": 2.2064335664335664, + "grad_norm": 0.47450060893711293, + "learning_rate": 3.97518133350285e-06, + "loss": 0.05758193880319595, + "step": 3945 + }, + { + "epoch": 2.206993006993007, + "grad_norm": 0.5094229272492317, + "learning_rate": 3.969988063756829e-06, + "loss": 0.07331477105617523, + "step": 3946 + }, + { + "epoch": 2.2075524475524477, + "grad_norm": 0.49805863243186516, + "learning_rate": 3.964797348260233e-06, + "loss": 0.11568835377693176, + "step": 3947 + }, + { + "epoch": 2.208111888111888, + "grad_norm": 0.4862509030635815, + "learning_rate": 3.959609189211798e-06, + "loss": 0.07070145010948181, + "step": 3948 + }, + { + "epoch": 2.2086713286713286, + "grad_norm": 0.4813765426292812, + "learning_rate": 3.954423588809163e-06, + "loss": 0.07407890260219574, + "step": 3949 + }, + { + "epoch": 2.209230769230769, + "grad_norm": 0.43826593261196445, + "learning_rate": 3.949240549248906e-06, + "loss": 0.07815185189247131, + "step": 3950 + }, + { + "epoch": 2.20979020979021, + "grad_norm": 0.6925381171449159, + "learning_rate": 3.9440600727264985e-06, + "loss": 0.2690117359161377, + "step": 3951 + }, + { + "epoch": 2.2103496503496505, + "grad_norm": 0.3643323169047638, + "learning_rate": 3.938882161436339e-06, + "loss": 0.0678568035364151, + "step": 3952 + }, + { + "epoch": 2.210909090909091, + "grad_norm": 0.32719436073057256, + "learning_rate": 3.933706817571741e-06, + "loss": 0.04408618062734604, + "step": 3953 + }, + { + "epoch": 2.2114685314685314, + "grad_norm": 0.47834801356454393, + "learning_rate": 3.928534043324912e-06, + "loss": 0.17852479219436646, + "step": 3954 + }, + { + "epoch": 2.212027972027972, + "grad_norm": 0.499189315214491, + "learning_rate": 3.923363840887002e-06, + "loss": 0.07297449558973312, + "step": 3955 + }, + { + "epoch": 2.2125874125874128, + "grad_norm": 0.278220922121688, + "learning_rate": 3.91819621244804e-06, + "loss": 0.10242324322462082, + "step": 3956 + }, + { + "epoch": 2.213146853146853, + "grad_norm": 0.9310113423169263, + "learning_rate": 3.913031160196985e-06, + "loss": 0.08546636253595352, + "step": 3957 + }, + { + "epoch": 2.2137062937062937, + "grad_norm": 0.294550734847301, + "learning_rate": 3.907868686321702e-06, + "loss": 0.012173691764473915, + "step": 3958 + }, + { + "epoch": 2.214265734265734, + "grad_norm": 0.531817382038387, + "learning_rate": 3.9027087930089505e-06, + "loss": 0.04898948222398758, + "step": 3959 + }, + { + "epoch": 2.214825174825175, + "grad_norm": 0.5275537968902968, + "learning_rate": 3.8975514824444184e-06, + "loss": 0.1368587762117386, + "step": 3960 + }, + { + "epoch": 2.2153846153846155, + "grad_norm": 0.3927867410479157, + "learning_rate": 3.892396756812679e-06, + "loss": 0.05297800526022911, + "step": 3961 + }, + { + "epoch": 2.215944055944056, + "grad_norm": 0.4645696519699345, + "learning_rate": 3.8872446182972235e-06, + "loss": 0.1151835173368454, + "step": 3962 + }, + { + "epoch": 2.2165034965034964, + "grad_norm": 0.3462028986717498, + "learning_rate": 3.882095069080446e-06, + "loss": 0.0301944799721241, + "step": 3963 + }, + { + "epoch": 2.217062937062937, + "grad_norm": 0.5292409506029238, + "learning_rate": 3.876948111343636e-06, + "loss": 0.12432238459587097, + "step": 3964 + }, + { + "epoch": 2.217622377622378, + "grad_norm": 0.49633858778195306, + "learning_rate": 3.871803747266991e-06, + "loss": 0.0917779952287674, + "step": 3965 + }, + { + "epoch": 2.2181818181818183, + "grad_norm": 0.40992194642896906, + "learning_rate": 3.866661979029612e-06, + "loss": 0.08906366676092148, + "step": 3966 + }, + { + "epoch": 2.2187412587412587, + "grad_norm": 0.46560573594238286, + "learning_rate": 3.861522808809497e-06, + "loss": 0.08712785691022873, + "step": 3967 + }, + { + "epoch": 2.219300699300699, + "grad_norm": 0.5687665315077317, + "learning_rate": 3.856386238783547e-06, + "loss": 0.0677545890212059, + "step": 3968 + }, + { + "epoch": 2.21986013986014, + "grad_norm": 0.4347074032296446, + "learning_rate": 3.851252271127554e-06, + "loss": 0.0493030920624733, + "step": 3969 + }, + { + "epoch": 2.2204195804195805, + "grad_norm": 0.47651579383164216, + "learning_rate": 3.846120908016213e-06, + "loss": 0.10384263098239899, + "step": 3970 + }, + { + "epoch": 2.220979020979021, + "grad_norm": 0.521522838176639, + "learning_rate": 3.840992151623117e-06, + "loss": 0.09387702494859695, + "step": 3971 + }, + { + "epoch": 2.2215384615384615, + "grad_norm": 0.5009208061154036, + "learning_rate": 3.835866004120753e-06, + "loss": 0.09325749427080154, + "step": 3972 + }, + { + "epoch": 2.222097902097902, + "grad_norm": 0.4310635000270723, + "learning_rate": 3.8307424676805076e-06, + "loss": 0.0520363375544548, + "step": 3973 + }, + { + "epoch": 2.222657342657343, + "grad_norm": 0.32363565152083373, + "learning_rate": 3.8256215444726466e-06, + "loss": 0.026735279709100723, + "step": 3974 + }, + { + "epoch": 2.2232167832167833, + "grad_norm": 0.40172553587934084, + "learning_rate": 3.820503236666345e-06, + "loss": 0.04936660826206207, + "step": 3975 + }, + { + "epoch": 2.2237762237762237, + "grad_norm": 0.4215315230974492, + "learning_rate": 3.815387546429663e-06, + "loss": 0.05895145237445831, + "step": 3976 + }, + { + "epoch": 2.224335664335664, + "grad_norm": 0.2903561252121752, + "learning_rate": 3.810274475929554e-06, + "loss": 0.01264667883515358, + "step": 3977 + }, + { + "epoch": 2.224895104895105, + "grad_norm": 0.34273474575258117, + "learning_rate": 3.8051640273318646e-06, + "loss": 0.055085889995098114, + "step": 3978 + }, + { + "epoch": 2.2254545454545456, + "grad_norm": 0.46323302857725185, + "learning_rate": 3.8000562028013177e-06, + "loss": 0.12590879201889038, + "step": 3979 + }, + { + "epoch": 2.226013986013986, + "grad_norm": 0.4575307156129042, + "learning_rate": 3.794951004501539e-06, + "loss": 0.13788959383964539, + "step": 3980 + }, + { + "epoch": 2.2265734265734265, + "grad_norm": 0.37605946665436557, + "learning_rate": 3.789848434595036e-06, + "loss": 0.05030418187379837, + "step": 3981 + }, + { + "epoch": 2.227132867132867, + "grad_norm": 0.5467937960386232, + "learning_rate": 3.7847484952432033e-06, + "loss": 0.06063270568847656, + "step": 3982 + }, + { + "epoch": 2.227692307692308, + "grad_norm": 0.2355029502414884, + "learning_rate": 3.7796511886063227e-06, + "loss": 0.0255948007106781, + "step": 3983 + }, + { + "epoch": 2.2282517482517483, + "grad_norm": 0.5539199602960609, + "learning_rate": 3.7745565168435614e-06, + "loss": 0.09658539295196533, + "step": 3984 + }, + { + "epoch": 2.228811188811189, + "grad_norm": 0.36421733200091205, + "learning_rate": 3.7694644821129635e-06, + "loss": 0.04903118312358856, + "step": 3985 + }, + { + "epoch": 2.2293706293706292, + "grad_norm": 0.3546461979045777, + "learning_rate": 3.764375086571462e-06, + "loss": 0.03705323487520218, + "step": 3986 + }, + { + "epoch": 2.22993006993007, + "grad_norm": 0.47701142971889393, + "learning_rate": 3.7592883323748762e-06, + "loss": 0.08233587443828583, + "step": 3987 + }, + { + "epoch": 2.2304895104895106, + "grad_norm": 0.5452543676600319, + "learning_rate": 3.754204221677892e-06, + "loss": 0.14936336874961853, + "step": 3988 + }, + { + "epoch": 2.231048951048951, + "grad_norm": 0.44084870206617127, + "learning_rate": 3.749122756634097e-06, + "loss": 0.08057862520217896, + "step": 3989 + }, + { + "epoch": 2.2316083916083915, + "grad_norm": 0.6085467495967158, + "learning_rate": 3.744043939395937e-06, + "loss": 0.17346906661987305, + "step": 3990 + }, + { + "epoch": 2.232167832167832, + "grad_norm": 0.5673485284009595, + "learning_rate": 3.738967772114749e-06, + "loss": 0.16614481806755066, + "step": 3991 + }, + { + "epoch": 2.232727272727273, + "grad_norm": 0.37756678555092926, + "learning_rate": 3.7338942569407456e-06, + "loss": 0.05408097058534622, + "step": 3992 + }, + { + "epoch": 2.2332867132867134, + "grad_norm": 0.4438251192725241, + "learning_rate": 3.7288233960230058e-06, + "loss": 0.0801866427063942, + "step": 3993 + }, + { + "epoch": 2.233846153846154, + "grad_norm": 0.2295029218870021, + "learning_rate": 3.723755191509506e-06, + "loss": 0.03332550823688507, + "step": 3994 + }, + { + "epoch": 2.2344055944055943, + "grad_norm": 0.5369337170038493, + "learning_rate": 3.7186896455470745e-06, + "loss": 0.10848326981067657, + "step": 3995 + }, + { + "epoch": 2.234965034965035, + "grad_norm": 0.40343578194062296, + "learning_rate": 3.7136267602814246e-06, + "loss": 0.0342969186604023, + "step": 3996 + }, + { + "epoch": 2.2355244755244756, + "grad_norm": 0.6328476343033618, + "learning_rate": 3.7085665378571465e-06, + "loss": 0.17291346192359924, + "step": 3997 + }, + { + "epoch": 2.236083916083916, + "grad_norm": 0.45141340554287845, + "learning_rate": 3.703508980417686e-06, + "loss": 0.082366943359375, + "step": 3998 + }, + { + "epoch": 2.2366433566433566, + "grad_norm": 0.4703491232235992, + "learning_rate": 3.6984540901053844e-06, + "loss": 0.07363560795783997, + "step": 3999 + }, + { + "epoch": 2.237202797202797, + "grad_norm": 0.5088013079055539, + "learning_rate": 3.69340186906143e-06, + "loss": 0.057901885360479355, + "step": 4000 + }, + { + "epoch": 2.237762237762238, + "grad_norm": 0.3791623755789042, + "learning_rate": 3.688352319425894e-06, + "loss": 0.07826337963342667, + "step": 4001 + }, + { + "epoch": 2.2383216783216784, + "grad_norm": 0.48774061361004595, + "learning_rate": 3.6833054433377145e-06, + "loss": 0.09077227115631104, + "step": 4002 + }, + { + "epoch": 2.238881118881119, + "grad_norm": 0.5434033256511848, + "learning_rate": 3.6782612429346863e-06, + "loss": 0.11784470081329346, + "step": 4003 + }, + { + "epoch": 2.2394405594405593, + "grad_norm": 0.6112429904821101, + "learning_rate": 3.6732197203534937e-06, + "loss": 0.09747405350208282, + "step": 4004 + }, + { + "epoch": 2.24, + "grad_norm": 0.40033507869643103, + "learning_rate": 3.668180877729659e-06, + "loss": 0.10384588688611984, + "step": 4005 + }, + { + "epoch": 2.2405594405594407, + "grad_norm": 0.4107941814076279, + "learning_rate": 3.66314471719759e-06, + "loss": 0.09412307292222977, + "step": 4006 + }, + { + "epoch": 2.241118881118881, + "grad_norm": 1.122582928396912, + "learning_rate": 3.6581112408905517e-06, + "loss": 0.20190943777561188, + "step": 4007 + }, + { + "epoch": 2.2416783216783216, + "grad_norm": 0.48227869870788953, + "learning_rate": 3.6530804509406625e-06, + "loss": 0.056947026401758194, + "step": 4008 + }, + { + "epoch": 2.242237762237762, + "grad_norm": 0.3906032101153484, + "learning_rate": 3.6480523494789257e-06, + "loss": 0.049666061997413635, + "step": 4009 + }, + { + "epoch": 2.242797202797203, + "grad_norm": 0.508391165316692, + "learning_rate": 3.6430269386351813e-06, + "loss": 0.13487812876701355, + "step": 4010 + }, + { + "epoch": 2.2433566433566434, + "grad_norm": 0.3320792647063788, + "learning_rate": 3.638004220538144e-06, + "loss": 0.1209891140460968, + "step": 4011 + }, + { + "epoch": 2.243916083916084, + "grad_norm": 0.21309465014393603, + "learning_rate": 3.6329841973153867e-06, + "loss": 0.037573978304862976, + "step": 4012 + }, + { + "epoch": 2.2444755244755243, + "grad_norm": 0.48563119838397656, + "learning_rate": 3.6279668710933337e-06, + "loss": 0.0838402584195137, + "step": 4013 + }, + { + "epoch": 2.2450349650349652, + "grad_norm": 0.40439660298731084, + "learning_rate": 3.6229522439972733e-06, + "loss": 0.08591905981302261, + "step": 4014 + }, + { + "epoch": 2.2455944055944057, + "grad_norm": 0.44535362764302844, + "learning_rate": 3.6179403181513483e-06, + "loss": 0.0782245621085167, + "step": 4015 + }, + { + "epoch": 2.246153846153846, + "grad_norm": 0.48461815959742605, + "learning_rate": 3.612931095678558e-06, + "loss": 0.17355681955814362, + "step": 4016 + }, + { + "epoch": 2.2467132867132866, + "grad_norm": 0.4745865303506871, + "learning_rate": 3.6079245787007557e-06, + "loss": 0.06991080939769745, + "step": 4017 + }, + { + "epoch": 2.247272727272727, + "grad_norm": 0.5377497765563206, + "learning_rate": 3.6029207693386536e-06, + "loss": 0.05921034514904022, + "step": 4018 + }, + { + "epoch": 2.247832167832168, + "grad_norm": 0.6343287569066993, + "learning_rate": 3.597919669711806e-06, + "loss": 0.11045928299427032, + "step": 4019 + }, + { + "epoch": 2.2483916083916085, + "grad_norm": 0.18551029841890257, + "learning_rate": 3.5929212819386285e-06, + "loss": 0.012293178588151932, + "step": 4020 + }, + { + "epoch": 2.248951048951049, + "grad_norm": 0.3115050590447506, + "learning_rate": 3.5879256081363843e-06, + "loss": 0.03238307312130928, + "step": 4021 + }, + { + "epoch": 2.2495104895104894, + "grad_norm": 0.43024651759596905, + "learning_rate": 3.58293265042119e-06, + "loss": 0.051424071192741394, + "step": 4022 + }, + { + "epoch": 2.2500699300699303, + "grad_norm": 0.5271631478206235, + "learning_rate": 3.577942410908013e-06, + "loss": 0.08836375176906586, + "step": 4023 + }, + { + "epoch": 2.2506293706293707, + "grad_norm": 0.37456513903201427, + "learning_rate": 3.5729548917106584e-06, + "loss": 0.06211835891008377, + "step": 4024 + }, + { + "epoch": 2.251188811188811, + "grad_norm": 0.20303357415475776, + "learning_rate": 3.5679700949417894e-06, + "loss": 0.015144189819693565, + "step": 4025 + }, + { + "epoch": 2.2517482517482517, + "grad_norm": 0.3567222528370749, + "learning_rate": 3.5629880227129154e-06, + "loss": 0.08175687491893768, + "step": 4026 + }, + { + "epoch": 2.252307692307692, + "grad_norm": 0.6093699183434942, + "learning_rate": 3.558008677134389e-06, + "loss": 0.13278828561306, + "step": 4027 + }, + { + "epoch": 2.252867132867133, + "grad_norm": 0.3760729231785968, + "learning_rate": 3.5530320603154113e-06, + "loss": 0.04883341118693352, + "step": 4028 + }, + { + "epoch": 2.2534265734265735, + "grad_norm": 0.5503247543216667, + "learning_rate": 3.5480581743640176e-06, + "loss": 0.10329955071210861, + "step": 4029 + }, + { + "epoch": 2.253986013986014, + "grad_norm": 0.5383509006818986, + "learning_rate": 3.543087021387098e-06, + "loss": 0.09205092489719391, + "step": 4030 + }, + { + "epoch": 2.2545454545454544, + "grad_norm": 0.26388243191330185, + "learning_rate": 3.538118603490379e-06, + "loss": 0.013555154204368591, + "step": 4031 + }, + { + "epoch": 2.2551048951048953, + "grad_norm": 0.5100827764416759, + "learning_rate": 3.5331529227784313e-06, + "loss": 0.09695370495319366, + "step": 4032 + }, + { + "epoch": 2.2556643356643358, + "grad_norm": 0.38163542731840494, + "learning_rate": 3.528189981354667e-06, + "loss": 0.03444574400782585, + "step": 4033 + }, + { + "epoch": 2.2562237762237762, + "grad_norm": 0.45765636041755753, + "learning_rate": 3.523229781321331e-06, + "loss": 0.15827062726020813, + "step": 4034 + }, + { + "epoch": 2.2567832167832167, + "grad_norm": 0.40498115559261233, + "learning_rate": 3.5182723247795124e-06, + "loss": 0.06910944730043411, + "step": 4035 + }, + { + "epoch": 2.257342657342657, + "grad_norm": 0.5105625882418331, + "learning_rate": 3.5133176138291438e-06, + "loss": 0.12077774107456207, + "step": 4036 + }, + { + "epoch": 2.257902097902098, + "grad_norm": 0.41582144089939427, + "learning_rate": 3.5083656505689756e-06, + "loss": 0.06353722512722015, + "step": 4037 + }, + { + "epoch": 2.2584615384615385, + "grad_norm": 0.5581788147287723, + "learning_rate": 3.503416437096622e-06, + "loss": 0.16074544191360474, + "step": 4038 + }, + { + "epoch": 2.259020979020979, + "grad_norm": 0.847192248788569, + "learning_rate": 3.498469975508506e-06, + "loss": 0.1852363497018814, + "step": 4039 + }, + { + "epoch": 2.2595804195804194, + "grad_norm": 0.4335149171091778, + "learning_rate": 3.493526267899903e-06, + "loss": 0.07301154732704163, + "step": 4040 + }, + { + "epoch": 2.2601398601398603, + "grad_norm": 0.45094774784276914, + "learning_rate": 3.4885853163649162e-06, + "loss": 0.06533936411142349, + "step": 4041 + }, + { + "epoch": 2.260699300699301, + "grad_norm": 0.4302137986583994, + "learning_rate": 3.483647122996472e-06, + "loss": 0.0321870818734169, + "step": 4042 + }, + { + "epoch": 2.2612587412587413, + "grad_norm": 0.476429268974784, + "learning_rate": 3.478711689886348e-06, + "loss": 0.08697166293859482, + "step": 4043 + }, + { + "epoch": 2.2618181818181817, + "grad_norm": 0.6227199692834946, + "learning_rate": 3.473779019125134e-06, + "loss": 0.05689668655395508, + "step": 4044 + }, + { + "epoch": 2.262377622377622, + "grad_norm": 0.6140658263158352, + "learning_rate": 3.4688491128022605e-06, + "loss": 0.1464952826499939, + "step": 4045 + }, + { + "epoch": 2.262937062937063, + "grad_norm": 0.2193836753259563, + "learning_rate": 3.463921973005986e-06, + "loss": 0.029885651543736458, + "step": 4046 + }, + { + "epoch": 2.2634965034965036, + "grad_norm": 0.40915772202038836, + "learning_rate": 3.458997601823385e-06, + "loss": 0.03507468104362488, + "step": 4047 + }, + { + "epoch": 2.264055944055944, + "grad_norm": 0.25136680835107317, + "learning_rate": 3.454076001340383e-06, + "loss": 0.011632539331912994, + "step": 4048 + }, + { + "epoch": 2.2646153846153845, + "grad_norm": 0.353706810157251, + "learning_rate": 3.4491571736417074e-06, + "loss": 0.07180996984243393, + "step": 4049 + }, + { + "epoch": 2.2651748251748254, + "grad_norm": 0.5281478482772088, + "learning_rate": 3.444241120810926e-06, + "loss": 0.11113995313644409, + "step": 4050 + }, + { + "epoch": 2.265734265734266, + "grad_norm": 0.17081657197761674, + "learning_rate": 3.43932784493043e-06, + "loss": 0.0048193675465881824, + "step": 4051 + }, + { + "epoch": 2.2662937062937063, + "grad_norm": 0.431591805815967, + "learning_rate": 3.43441734808142e-06, + "loss": 0.0892617478966713, + "step": 4052 + }, + { + "epoch": 2.2668531468531468, + "grad_norm": 1.1452990025411272, + "learning_rate": 3.429509632343946e-06, + "loss": 0.1479639708995819, + "step": 4053 + }, + { + "epoch": 2.267412587412587, + "grad_norm": 0.449840589750116, + "learning_rate": 3.424604699796853e-06, + "loss": 0.09672702103853226, + "step": 4054 + }, + { + "epoch": 2.267972027972028, + "grad_norm": 0.367836774590596, + "learning_rate": 3.419702552517823e-06, + "loss": 0.04358110576868057, + "step": 4055 + }, + { + "epoch": 2.2685314685314686, + "grad_norm": 0.5589733019810754, + "learning_rate": 3.4148031925833525e-06, + "loss": 0.08175045251846313, + "step": 4056 + }, + { + "epoch": 2.269090909090909, + "grad_norm": 0.6464911419907217, + "learning_rate": 3.4099066220687603e-06, + "loss": 0.14600539207458496, + "step": 4057 + }, + { + "epoch": 2.2696503496503495, + "grad_norm": 0.3565724467859798, + "learning_rate": 3.4050128430481844e-06, + "loss": 0.04659982770681381, + "step": 4058 + }, + { + "epoch": 2.2702097902097904, + "grad_norm": 0.7026851967988924, + "learning_rate": 3.4001218575945716e-06, + "loss": 0.15947610139846802, + "step": 4059 + }, + { + "epoch": 2.270769230769231, + "grad_norm": 0.2954683634461207, + "learning_rate": 3.395233667779695e-06, + "loss": 0.0396287702023983, + "step": 4060 + }, + { + "epoch": 2.2713286713286713, + "grad_norm": 0.42994981087816814, + "learning_rate": 3.3903482756741402e-06, + "loss": 0.07789931446313858, + "step": 4061 + }, + { + "epoch": 2.271888111888112, + "grad_norm": 0.43895252474649704, + "learning_rate": 3.38546568334731e-06, + "loss": 0.08802768588066101, + "step": 4062 + }, + { + "epoch": 2.2724475524475523, + "grad_norm": 0.44514956249061005, + "learning_rate": 3.380585892867414e-06, + "loss": 0.06346607953310013, + "step": 4063 + }, + { + "epoch": 2.273006993006993, + "grad_norm": 0.48751436942776405, + "learning_rate": 3.3757089063014836e-06, + "loss": 0.1546359658241272, + "step": 4064 + }, + { + "epoch": 2.2735664335664336, + "grad_norm": 0.462952661135787, + "learning_rate": 3.3708347257153562e-06, + "loss": 0.08639201521873474, + "step": 4065 + }, + { + "epoch": 2.274125874125874, + "grad_norm": 0.6045511528018018, + "learning_rate": 3.3659633531736858e-06, + "loss": 0.05767097324132919, + "step": 4066 + }, + { + "epoch": 2.2746853146853145, + "grad_norm": 0.44370956555379876, + "learning_rate": 3.3610947907399373e-06, + "loss": 0.10328216850757599, + "step": 4067 + }, + { + "epoch": 2.2752447552447554, + "grad_norm": 0.5036973735106417, + "learning_rate": 3.356229040476375e-06, + "loss": 0.06288590282201767, + "step": 4068 + }, + { + "epoch": 2.275804195804196, + "grad_norm": 0.42037428299140284, + "learning_rate": 3.3513661044440825e-06, + "loss": 0.051631905138492584, + "step": 4069 + }, + { + "epoch": 2.2763636363636364, + "grad_norm": 0.4829873598608951, + "learning_rate": 3.346505984702949e-06, + "loss": 0.12575900554656982, + "step": 4070 + }, + { + "epoch": 2.276923076923077, + "grad_norm": 0.5258413338111567, + "learning_rate": 3.34164868331167e-06, + "loss": 0.062405407428741455, + "step": 4071 + }, + { + "epoch": 2.2774825174825173, + "grad_norm": 0.4536695804890835, + "learning_rate": 3.33679420232775e-06, + "loss": 0.1576712429523468, + "step": 4072 + }, + { + "epoch": 2.278041958041958, + "grad_norm": 0.5721614445904575, + "learning_rate": 3.331942543807488e-06, + "loss": 0.15222272276878357, + "step": 4073 + }, + { + "epoch": 2.2786013986013987, + "grad_norm": 0.30290280858534124, + "learning_rate": 3.327093709806002e-06, + "loss": 0.026921415701508522, + "step": 4074 + }, + { + "epoch": 2.279160839160839, + "grad_norm": 0.43343840635360853, + "learning_rate": 3.322247702377204e-06, + "loss": 0.03489012271165848, + "step": 4075 + }, + { + "epoch": 2.2797202797202796, + "grad_norm": 0.4237988340450189, + "learning_rate": 3.3174045235738117e-06, + "loss": 0.03215426206588745, + "step": 4076 + }, + { + "epoch": 2.2802797202797205, + "grad_norm": 0.5970364869157166, + "learning_rate": 3.3125641754473503e-06, + "loss": 0.12308354675769806, + "step": 4077 + }, + { + "epoch": 2.280839160839161, + "grad_norm": 0.5147425591820928, + "learning_rate": 3.3077266600481274e-06, + "loss": 0.07809754461050034, + "step": 4078 + }, + { + "epoch": 2.2813986013986014, + "grad_norm": 0.5636534059960786, + "learning_rate": 3.302891979425278e-06, + "loss": 0.06743837147951126, + "step": 4079 + }, + { + "epoch": 2.281958041958042, + "grad_norm": 0.40399709209953466, + "learning_rate": 3.2980601356267126e-06, + "loss": 0.05035008117556572, + "step": 4080 + }, + { + "epoch": 2.2825174825174823, + "grad_norm": 0.6132218352475859, + "learning_rate": 3.2932311306991504e-06, + "loss": 0.17568081617355347, + "step": 4081 + }, + { + "epoch": 2.2830769230769232, + "grad_norm": 0.5165971985882982, + "learning_rate": 3.2884049666881136e-06, + "loss": 0.15135107934474945, + "step": 4082 + }, + { + "epoch": 2.2836363636363637, + "grad_norm": 0.41535377879082314, + "learning_rate": 3.283581645637901e-06, + "loss": 0.08967706561088562, + "step": 4083 + }, + { + "epoch": 2.284195804195804, + "grad_norm": 0.2923423263543926, + "learning_rate": 3.2787611695916355e-06, + "loss": 0.04252856224775314, + "step": 4084 + }, + { + "epoch": 2.2847552447552446, + "grad_norm": 0.4676146336758347, + "learning_rate": 3.273943540591212e-06, + "loss": 0.05316941440105438, + "step": 4085 + }, + { + "epoch": 2.2853146853146855, + "grad_norm": 0.3615636830495232, + "learning_rate": 3.2691287606773226e-06, + "loss": 0.04093504324555397, + "step": 4086 + }, + { + "epoch": 2.285874125874126, + "grad_norm": 0.6251445414052959, + "learning_rate": 3.2643168318894693e-06, + "loss": 0.18728643655776978, + "step": 4087 + }, + { + "epoch": 2.2864335664335664, + "grad_norm": 0.45201235540859297, + "learning_rate": 3.2595077562659248e-06, + "loss": 0.1192634254693985, + "step": 4088 + }, + { + "epoch": 2.286993006993007, + "grad_norm": 0.3108061231208798, + "learning_rate": 3.254701535843766e-06, + "loss": 0.03759636729955673, + "step": 4089 + }, + { + "epoch": 2.2875524475524474, + "grad_norm": 0.31416363387771745, + "learning_rate": 3.249898172658861e-06, + "loss": 0.03267242759466171, + "step": 4090 + }, + { + "epoch": 2.2881118881118883, + "grad_norm": 0.5864678464127548, + "learning_rate": 3.245097668745851e-06, + "loss": 0.11120981723070145, + "step": 4091 + }, + { + "epoch": 2.2886713286713287, + "grad_norm": 0.48210137182632357, + "learning_rate": 3.240300026138197e-06, + "loss": 0.08536656200885773, + "step": 4092 + }, + { + "epoch": 2.289230769230769, + "grad_norm": 0.46736798508785293, + "learning_rate": 3.2355052468681147e-06, + "loss": 0.13874593377113342, + "step": 4093 + }, + { + "epoch": 2.2897902097902096, + "grad_norm": 0.3987255634853327, + "learning_rate": 3.2307133329666297e-06, + "loss": 0.05582427233457565, + "step": 4094 + }, + { + "epoch": 2.2903496503496505, + "grad_norm": 0.4260332915941483, + "learning_rate": 3.225924286463543e-06, + "loss": 0.08884363621473312, + "step": 4095 + }, + { + "epoch": 2.290909090909091, + "grad_norm": 0.46575679624818783, + "learning_rate": 3.221138109387446e-06, + "loss": 0.07232660055160522, + "step": 4096 + }, + { + "epoch": 2.2914685314685315, + "grad_norm": 0.42173643018719015, + "learning_rate": 3.2163548037657143e-06, + "loss": 0.08202521502971649, + "step": 4097 + }, + { + "epoch": 2.292027972027972, + "grad_norm": 0.4605290307744553, + "learning_rate": 3.211574371624502e-06, + "loss": 0.13254153728485107, + "step": 4098 + }, + { + "epoch": 2.2925874125874124, + "grad_norm": 0.47041096150572986, + "learning_rate": 3.2067968149887496e-06, + "loss": 0.10033851116895676, + "step": 4099 + }, + { + "epoch": 2.2931468531468533, + "grad_norm": 0.4728476488708833, + "learning_rate": 3.202022135882182e-06, + "loss": 0.06426310539245605, + "step": 4100 + }, + { + "epoch": 2.2937062937062938, + "grad_norm": 0.6269701358401174, + "learning_rate": 3.1972503363273034e-06, + "loss": 0.19269779324531555, + "step": 4101 + }, + { + "epoch": 2.294265734265734, + "grad_norm": 0.512630018763121, + "learning_rate": 3.1924814183453977e-06, + "loss": 0.12282806634902954, + "step": 4102 + }, + { + "epoch": 2.2948251748251747, + "grad_norm": 0.5400530933616533, + "learning_rate": 3.1877153839565255e-06, + "loss": 0.08900883048772812, + "step": 4103 + }, + { + "epoch": 2.2953846153846156, + "grad_norm": 0.47819396823832194, + "learning_rate": 3.1829522351795294e-06, + "loss": 0.08213920891284943, + "step": 4104 + }, + { + "epoch": 2.295944055944056, + "grad_norm": 0.3064296239743476, + "learning_rate": 3.17819197403203e-06, + "loss": 0.07684902846813202, + "step": 4105 + }, + { + "epoch": 2.2965034965034965, + "grad_norm": 0.39913492827950353, + "learning_rate": 3.1734346025304242e-06, + "loss": 0.041487667709589005, + "step": 4106 + }, + { + "epoch": 2.297062937062937, + "grad_norm": 0.37282469767580045, + "learning_rate": 3.1686801226898854e-06, + "loss": 0.03580848127603531, + "step": 4107 + }, + { + "epoch": 2.2976223776223774, + "grad_norm": 0.5337312983464462, + "learning_rate": 3.1639285365243575e-06, + "loss": 0.09804193675518036, + "step": 4108 + }, + { + "epoch": 2.2981818181818183, + "grad_norm": 0.5142746629175026, + "learning_rate": 3.159179846046563e-06, + "loss": 0.19630542397499084, + "step": 4109 + }, + { + "epoch": 2.298741258741259, + "grad_norm": 0.4759929149685903, + "learning_rate": 3.1544340532679983e-06, + "loss": 0.08246155083179474, + "step": 4110 + }, + { + "epoch": 2.2993006993006992, + "grad_norm": 0.41041017308238337, + "learning_rate": 3.1496911601989356e-06, + "loss": 0.09034265577793121, + "step": 4111 + }, + { + "epoch": 2.2998601398601397, + "grad_norm": 0.7281734794230877, + "learning_rate": 3.1449511688484024e-06, + "loss": 0.2439144253730774, + "step": 4112 + }, + { + "epoch": 2.3004195804195806, + "grad_norm": 0.3138021871535495, + "learning_rate": 3.1402140812242233e-06, + "loss": 0.053342171013355255, + "step": 4113 + }, + { + "epoch": 2.300979020979021, + "grad_norm": 0.4051621386136702, + "learning_rate": 3.1354798993329673e-06, + "loss": 0.04787937551736832, + "step": 4114 + }, + { + "epoch": 2.3015384615384615, + "grad_norm": 0.62742071918321, + "learning_rate": 3.1307486251799903e-06, + "loss": 0.1571027636528015, + "step": 4115 + }, + { + "epoch": 2.302097902097902, + "grad_norm": 0.273568722370497, + "learning_rate": 3.126020260769409e-06, + "loss": 0.008515046909451485, + "step": 4116 + }, + { + "epoch": 2.3026573426573425, + "grad_norm": 0.6531664691831732, + "learning_rate": 3.1212948081041028e-06, + "loss": 0.08391707390546799, + "step": 4117 + }, + { + "epoch": 2.3032167832167834, + "grad_norm": 0.6505737639873461, + "learning_rate": 3.116572269185735e-06, + "loss": 0.22323326766490936, + "step": 4118 + }, + { + "epoch": 2.303776223776224, + "grad_norm": 0.3662512615884962, + "learning_rate": 3.1118526460147136e-06, + "loss": 0.06100432202219963, + "step": 4119 + }, + { + "epoch": 2.3043356643356643, + "grad_norm": 0.5131061524856819, + "learning_rate": 3.1071359405902246e-06, + "loss": 0.036345768719911575, + "step": 4120 + }, + { + "epoch": 2.3048951048951047, + "grad_norm": 0.3777475561997133, + "learning_rate": 3.102422154910217e-06, + "loss": 0.07189509272575378, + "step": 4121 + }, + { + "epoch": 2.3054545454545456, + "grad_norm": 0.5702115574503347, + "learning_rate": 3.0977112909713923e-06, + "loss": 0.0929940864443779, + "step": 4122 + }, + { + "epoch": 2.306013986013986, + "grad_norm": 0.6433353737199373, + "learning_rate": 3.0930033507692346e-06, + "loss": 0.1341400295495987, + "step": 4123 + }, + { + "epoch": 2.3065734265734266, + "grad_norm": 0.5353758934076167, + "learning_rate": 3.088298336297968e-06, + "loss": 0.2187308520078659, + "step": 4124 + }, + { + "epoch": 2.307132867132867, + "grad_norm": 0.23309275968891738, + "learning_rate": 3.083596249550591e-06, + "loss": 0.027857935056090355, + "step": 4125 + }, + { + "epoch": 2.3076923076923075, + "grad_norm": 0.411809268668689, + "learning_rate": 3.0788970925188622e-06, + "loss": 0.06730443984270096, + "step": 4126 + }, + { + "epoch": 2.3082517482517484, + "grad_norm": 0.37152708483867974, + "learning_rate": 3.0742008671932823e-06, + "loss": 0.061534080654382706, + "step": 4127 + }, + { + "epoch": 2.308811188811189, + "grad_norm": 0.39166063232348686, + "learning_rate": 3.0695075755631376e-06, + "loss": 0.09674179553985596, + "step": 4128 + }, + { + "epoch": 2.3093706293706293, + "grad_norm": 0.24647821961693545, + "learning_rate": 3.064817219616447e-06, + "loss": 0.012089809402823448, + "step": 4129 + }, + { + "epoch": 2.3099300699300698, + "grad_norm": 0.303575513095863, + "learning_rate": 3.0601298013399983e-06, + "loss": 0.023520154878497124, + "step": 4130 + }, + { + "epoch": 2.3104895104895107, + "grad_norm": 0.33008708130483366, + "learning_rate": 3.055445322719337e-06, + "loss": 0.061036668717861176, + "step": 4131 + }, + { + "epoch": 2.311048951048951, + "grad_norm": 0.4810275736234355, + "learning_rate": 3.0507637857387486e-06, + "loss": 0.07786140590906143, + "step": 4132 + }, + { + "epoch": 2.3116083916083916, + "grad_norm": 0.6051715979142173, + "learning_rate": 3.0460851923812953e-06, + "loss": 0.16588738560676575, + "step": 4133 + }, + { + "epoch": 2.312167832167832, + "grad_norm": 0.4873390029605994, + "learning_rate": 3.0414095446287696e-06, + "loss": 0.06233539432287216, + "step": 4134 + }, + { + "epoch": 2.3127272727272725, + "grad_norm": 0.3653340597080994, + "learning_rate": 3.0367368444617307e-06, + "loss": 0.05115117132663727, + "step": 4135 + }, + { + "epoch": 2.3132867132867134, + "grad_norm": 0.32533133216754745, + "learning_rate": 3.0320670938594887e-06, + "loss": 0.021471742540597916, + "step": 4136 + }, + { + "epoch": 2.313846153846154, + "grad_norm": 0.4363020825274332, + "learning_rate": 3.0274002948000925e-06, + "loss": 0.15726208686828613, + "step": 4137 + }, + { + "epoch": 2.3144055944055943, + "grad_norm": 0.3725387477163445, + "learning_rate": 3.022736449260354e-06, + "loss": 0.029623623937368393, + "step": 4138 + }, + { + "epoch": 2.314965034965035, + "grad_norm": 0.7596036689553942, + "learning_rate": 3.0180755592158286e-06, + "loss": 0.13494645059108734, + "step": 4139 + }, + { + "epoch": 2.3155244755244757, + "grad_norm": 0.43292247166330355, + "learning_rate": 3.01341762664082e-06, + "loss": 0.07084748148918152, + "step": 4140 + }, + { + "epoch": 2.316083916083916, + "grad_norm": 0.4192960438724473, + "learning_rate": 3.0087626535083814e-06, + "loss": 0.03671936318278313, + "step": 4141 + }, + { + "epoch": 2.3166433566433566, + "grad_norm": 0.8096814905466239, + "learning_rate": 3.0041106417903056e-06, + "loss": 0.17640714347362518, + "step": 4142 + }, + { + "epoch": 2.317202797202797, + "grad_norm": 0.34425499889470296, + "learning_rate": 2.999461593457137e-06, + "loss": 0.035536833107471466, + "step": 4143 + }, + { + "epoch": 2.3177622377622376, + "grad_norm": 0.2902150891945376, + "learning_rate": 2.994815510478165e-06, + "loss": 0.041797198355197906, + "step": 4144 + }, + { + "epoch": 2.3183216783216785, + "grad_norm": 0.5273119850008087, + "learning_rate": 2.990172394821421e-06, + "loss": 0.10188169777393341, + "step": 4145 + }, + { + "epoch": 2.318881118881119, + "grad_norm": 0.5694206168244444, + "learning_rate": 2.9855322484536833e-06, + "loss": 0.06409934908151627, + "step": 4146 + }, + { + "epoch": 2.3194405594405594, + "grad_norm": 0.3492307219574627, + "learning_rate": 2.9808950733404606e-06, + "loss": 0.06261476874351501, + "step": 4147 + }, + { + "epoch": 2.32, + "grad_norm": 0.5313735820134512, + "learning_rate": 2.9762608714460173e-06, + "loss": 0.10893870145082474, + "step": 4148 + }, + { + "epoch": 2.3205594405594407, + "grad_norm": 0.677716605083071, + "learning_rate": 2.9716296447333506e-06, + "loss": 0.24184440076351166, + "step": 4149 + }, + { + "epoch": 2.321118881118881, + "grad_norm": 0.8119918765186172, + "learning_rate": 2.9670013951641997e-06, + "loss": 0.09185001254081726, + "step": 4150 + }, + { + "epoch": 2.3216783216783217, + "grad_norm": 0.40046048428645004, + "learning_rate": 2.9623761246990423e-06, + "loss": 0.06915062665939331, + "step": 4151 + }, + { + "epoch": 2.322237762237762, + "grad_norm": 0.5324924270700354, + "learning_rate": 2.957753835297098e-06, + "loss": 0.10790784657001495, + "step": 4152 + }, + { + "epoch": 2.3227972027972026, + "grad_norm": 0.33356362679126966, + "learning_rate": 2.9531345289163114e-06, + "loss": 0.025478679686784744, + "step": 4153 + }, + { + "epoch": 2.3233566433566435, + "grad_norm": 0.4085432338324467, + "learning_rate": 2.9485182075133777e-06, + "loss": 0.07190114259719849, + "step": 4154 + }, + { + "epoch": 2.323916083916084, + "grad_norm": 0.30879454225303854, + "learning_rate": 2.9439048730437214e-06, + "loss": 0.06151868775486946, + "step": 4155 + }, + { + "epoch": 2.3244755244755244, + "grad_norm": 0.5340003303628167, + "learning_rate": 2.9392945274615005e-06, + "loss": 0.24332216382026672, + "step": 4156 + }, + { + "epoch": 2.325034965034965, + "grad_norm": 0.3865896995694137, + "learning_rate": 2.9346871727196135e-06, + "loss": 0.030869830399751663, + "step": 4157 + }, + { + "epoch": 2.325594405594406, + "grad_norm": 0.3805556744060159, + "learning_rate": 2.93008281076968e-06, + "loss": 0.05284121632575989, + "step": 4158 + }, + { + "epoch": 2.3261538461538462, + "grad_norm": 0.2030401343179325, + "learning_rate": 2.925481443562064e-06, + "loss": 0.011515405960381031, + "step": 4159 + }, + { + "epoch": 2.3267132867132867, + "grad_norm": 0.3947058387736476, + "learning_rate": 2.920883073045856e-06, + "loss": 0.06661350280046463, + "step": 4160 + }, + { + "epoch": 2.327272727272727, + "grad_norm": 0.45883967073468546, + "learning_rate": 2.916287701168872e-06, + "loss": 0.12680260837078094, + "step": 4161 + }, + { + "epoch": 2.3278321678321676, + "grad_norm": 0.6436523191569041, + "learning_rate": 2.9116953298776717e-06, + "loss": 0.20449991524219513, + "step": 4162 + }, + { + "epoch": 2.3283916083916085, + "grad_norm": 0.35867865020374873, + "learning_rate": 2.907105961117529e-06, + "loss": 0.10279616713523865, + "step": 4163 + }, + { + "epoch": 2.328951048951049, + "grad_norm": 0.4976229658837716, + "learning_rate": 2.9025195968324526e-06, + "loss": 0.055316224694252014, + "step": 4164 + }, + { + "epoch": 2.3295104895104894, + "grad_norm": 0.4155817867323431, + "learning_rate": 2.8979362389651824e-06, + "loss": 0.06609553098678589, + "step": 4165 + }, + { + "epoch": 2.33006993006993, + "grad_norm": 0.43852332868753574, + "learning_rate": 2.8933558894571713e-06, + "loss": 0.06627489626407623, + "step": 4166 + }, + { + "epoch": 2.330629370629371, + "grad_norm": 0.29499603274663455, + "learning_rate": 2.888778550248618e-06, + "loss": 0.015674112364649773, + "step": 4167 + }, + { + "epoch": 2.3311888111888113, + "grad_norm": 0.3564830141881658, + "learning_rate": 2.8842042232784274e-06, + "loss": 0.052654676139354706, + "step": 4168 + }, + { + "epoch": 2.3317482517482517, + "grad_norm": 0.5891644881597351, + "learning_rate": 2.879632910484239e-06, + "loss": 0.15021787583827972, + "step": 4169 + }, + { + "epoch": 2.332307692307692, + "grad_norm": 0.21919869754283414, + "learning_rate": 2.875064613802415e-06, + "loss": 0.05389862507581711, + "step": 4170 + }, + { + "epoch": 2.3328671328671327, + "grad_norm": 0.19059613973852504, + "learning_rate": 2.870499335168029e-06, + "loss": 0.009877381846308708, + "step": 4171 + }, + { + "epoch": 2.3334265734265736, + "grad_norm": 0.6276914217898564, + "learning_rate": 2.8659370765148963e-06, + "loss": 0.09953927993774414, + "step": 4172 + }, + { + "epoch": 2.333986013986014, + "grad_norm": 0.4562099043004734, + "learning_rate": 2.8613778397755332e-06, + "loss": 0.0521346852183342, + "step": 4173 + }, + { + "epoch": 2.3345454545454545, + "grad_norm": 0.443035841521003, + "learning_rate": 2.8568216268811876e-06, + "loss": 0.0974147617816925, + "step": 4174 + }, + { + "epoch": 2.335104895104895, + "grad_norm": 0.41383200408890086, + "learning_rate": 2.8522684397618263e-06, + "loss": 0.03616798669099808, + "step": 4175 + }, + { + "epoch": 2.335664335664336, + "grad_norm": 0.4391515761599591, + "learning_rate": 2.84771828034612e-06, + "loss": 0.04907721281051636, + "step": 4176 + }, + { + "epoch": 2.3362237762237763, + "grad_norm": 0.5577862650432552, + "learning_rate": 2.8431711505614832e-06, + "loss": 0.18164277076721191, + "step": 4177 + }, + { + "epoch": 2.3367832167832168, + "grad_norm": 0.5855559043526604, + "learning_rate": 2.838627052334022e-06, + "loss": 0.11378076672554016, + "step": 4178 + }, + { + "epoch": 2.3373426573426572, + "grad_norm": 0.4720301552282028, + "learning_rate": 2.8340859875885708e-06, + "loss": 0.10286635160446167, + "step": 4179 + }, + { + "epoch": 2.3379020979020977, + "grad_norm": 0.46073118185624384, + "learning_rate": 2.8295479582486794e-06, + "loss": 0.13022872805595398, + "step": 4180 + }, + { + "epoch": 2.3384615384615386, + "grad_norm": 0.4541563472315739, + "learning_rate": 2.825012966236602e-06, + "loss": 0.08087345957756042, + "step": 4181 + }, + { + "epoch": 2.339020979020979, + "grad_norm": 0.45428487352986757, + "learning_rate": 2.820481013473324e-06, + "loss": 0.05846138671040535, + "step": 4182 + }, + { + "epoch": 2.3395804195804195, + "grad_norm": 0.6501690354358819, + "learning_rate": 2.815952101878524e-06, + "loss": 0.11591841280460358, + "step": 4183 + }, + { + "epoch": 2.34013986013986, + "grad_norm": 0.49988429359794484, + "learning_rate": 2.8114262333706045e-06, + "loss": 0.10986776649951935, + "step": 4184 + }, + { + "epoch": 2.340699300699301, + "grad_norm": 0.6631183559013969, + "learning_rate": 2.806903409866678e-06, + "loss": 0.1388150453567505, + "step": 4185 + }, + { + "epoch": 2.3412587412587413, + "grad_norm": 0.39543589117662853, + "learning_rate": 2.8023836332825595e-06, + "loss": 0.06061611324548721, + "step": 4186 + }, + { + "epoch": 2.341818181818182, + "grad_norm": 0.33863931629079863, + "learning_rate": 2.7978669055327812e-06, + "loss": 0.06075330823659897, + "step": 4187 + }, + { + "epoch": 2.3423776223776223, + "grad_norm": 0.5755634367216511, + "learning_rate": 2.793353228530581e-06, + "loss": 0.16876676678657532, + "step": 4188 + }, + { + "epoch": 2.3429370629370627, + "grad_norm": 0.298568069289008, + "learning_rate": 2.7888426041879056e-06, + "loss": 0.09192757308483124, + "step": 4189 + }, + { + "epoch": 2.3434965034965036, + "grad_norm": 0.33163352857752076, + "learning_rate": 2.7843350344154087e-06, + "loss": 0.0636255294084549, + "step": 4190 + }, + { + "epoch": 2.344055944055944, + "grad_norm": 0.42798180782653894, + "learning_rate": 2.7798305211224497e-06, + "loss": 0.05475534498691559, + "step": 4191 + }, + { + "epoch": 2.3446153846153845, + "grad_norm": 0.24828867967650067, + "learning_rate": 2.7753290662170896e-06, + "loss": 0.015556749887764454, + "step": 4192 + }, + { + "epoch": 2.345174825174825, + "grad_norm": 0.5180654607222392, + "learning_rate": 2.770830671606097e-06, + "loss": 0.1287653148174286, + "step": 4193 + }, + { + "epoch": 2.345734265734266, + "grad_norm": 0.6121085957004735, + "learning_rate": 2.7663353391949477e-06, + "loss": 0.13449427485466003, + "step": 4194 + }, + { + "epoch": 2.3462937062937064, + "grad_norm": 0.5736225370768098, + "learning_rate": 2.7618430708878165e-06, + "loss": 0.12945351004600525, + "step": 4195 + }, + { + "epoch": 2.346853146853147, + "grad_norm": 0.4580394353342252, + "learning_rate": 2.757353868587582e-06, + "loss": 0.050352297723293304, + "step": 4196 + }, + { + "epoch": 2.3474125874125873, + "grad_norm": 0.6183583312623002, + "learning_rate": 2.7528677341958175e-06, + "loss": 0.10918182134628296, + "step": 4197 + }, + { + "epoch": 2.3479720279720278, + "grad_norm": 0.38196796254027493, + "learning_rate": 2.7483846696128067e-06, + "loss": 0.0763801857829094, + "step": 4198 + }, + { + "epoch": 2.3485314685314687, + "grad_norm": 0.4373817633942312, + "learning_rate": 2.7439046767375265e-06, + "loss": 0.05111249163746834, + "step": 4199 + }, + { + "epoch": 2.349090909090909, + "grad_norm": 0.2505921663719418, + "learning_rate": 2.7394277574676564e-06, + "loss": 0.018557261675596237, + "step": 4200 + }, + { + "epoch": 2.3496503496503496, + "grad_norm": 0.4856903912577588, + "learning_rate": 2.734953913699574e-06, + "loss": 0.1314995288848877, + "step": 4201 + }, + { + "epoch": 2.35020979020979, + "grad_norm": 0.5530988243223491, + "learning_rate": 2.7304831473283467e-06, + "loss": 0.11927692592144012, + "step": 4202 + }, + { + "epoch": 2.350769230769231, + "grad_norm": 0.349398175129622, + "learning_rate": 2.7260154602477474e-06, + "loss": 0.09752250462770462, + "step": 4203 + }, + { + "epoch": 2.3513286713286714, + "grad_norm": 0.3950308033129619, + "learning_rate": 2.7215508543502402e-06, + "loss": 0.06639213114976883, + "step": 4204 + }, + { + "epoch": 2.351888111888112, + "grad_norm": 0.56125063746809, + "learning_rate": 2.7170893315269842e-06, + "loss": 0.10241188853979111, + "step": 4205 + }, + { + "epoch": 2.3524475524475523, + "grad_norm": 0.30808454874980007, + "learning_rate": 2.712630893667839e-06, + "loss": 0.02258661389350891, + "step": 4206 + }, + { + "epoch": 2.353006993006993, + "grad_norm": 0.413324216440858, + "learning_rate": 2.708175542661342e-06, + "loss": 0.09763824194669724, + "step": 4207 + }, + { + "epoch": 2.3535664335664337, + "grad_norm": 0.6378048713064322, + "learning_rate": 2.7037232803947434e-06, + "loss": 0.0902264192700386, + "step": 4208 + }, + { + "epoch": 2.354125874125874, + "grad_norm": 0.3003584090409068, + "learning_rate": 2.6992741087539697e-06, + "loss": 0.0530712753534317, + "step": 4209 + }, + { + "epoch": 2.3546853146853146, + "grad_norm": 0.4750475273849003, + "learning_rate": 2.6948280296236373e-06, + "loss": 0.06394069641828537, + "step": 4210 + }, + { + "epoch": 2.355244755244755, + "grad_norm": 0.559121498512113, + "learning_rate": 2.6903850448870693e-06, + "loss": 0.1131538450717926, + "step": 4211 + }, + { + "epoch": 2.355804195804196, + "grad_norm": 0.3728169790626828, + "learning_rate": 2.6859451564262595e-06, + "loss": 0.062194716185331345, + "step": 4212 + }, + { + "epoch": 2.3563636363636364, + "grad_norm": 0.46186081799795875, + "learning_rate": 2.6815083661219e-06, + "loss": 0.09972430020570755, + "step": 4213 + }, + { + "epoch": 2.356923076923077, + "grad_norm": 0.6522443172074811, + "learning_rate": 2.6770746758533716e-06, + "loss": 0.1004643365740776, + "step": 4214 + }, + { + "epoch": 2.3574825174825174, + "grad_norm": 0.5647545140604252, + "learning_rate": 2.672644087498729e-06, + "loss": 0.12935084104537964, + "step": 4215 + }, + { + "epoch": 2.358041958041958, + "grad_norm": 1.0217605857985854, + "learning_rate": 2.6682166029347356e-06, + "loss": 0.16304895281791687, + "step": 4216 + }, + { + "epoch": 2.3586013986013987, + "grad_norm": 0.307736736989733, + "learning_rate": 2.6637922240368184e-06, + "loss": 0.060444191098213196, + "step": 4217 + }, + { + "epoch": 2.359160839160839, + "grad_norm": 0.8128947995881212, + "learning_rate": 2.659370952679099e-06, + "loss": 0.14504699409008026, + "step": 4218 + }, + { + "epoch": 2.3597202797202796, + "grad_norm": 0.3684411606052292, + "learning_rate": 2.6549527907343865e-06, + "loss": 0.04341878741979599, + "step": 4219 + }, + { + "epoch": 2.36027972027972, + "grad_norm": 0.43391404521254734, + "learning_rate": 2.650537740074156e-06, + "loss": 0.1116509884595871, + "step": 4220 + }, + { + "epoch": 2.360839160839161, + "grad_norm": 0.7052327255577854, + "learning_rate": 2.646125802568591e-06, + "loss": 0.08082155138254166, + "step": 4221 + }, + { + "epoch": 2.3613986013986015, + "grad_norm": 0.6623060635274876, + "learning_rate": 2.6417169800865306e-06, + "loss": 0.11466452479362488, + "step": 4222 + }, + { + "epoch": 2.361958041958042, + "grad_norm": 0.3542712493637846, + "learning_rate": 2.6373112744955098e-06, + "loss": 0.025832612067461014, + "step": 4223 + }, + { + "epoch": 2.3625174825174824, + "grad_norm": 0.4969604317545119, + "learning_rate": 2.632908687661743e-06, + "loss": 0.1191239133477211, + "step": 4224 + }, + { + "epoch": 2.363076923076923, + "grad_norm": 0.5109913030966607, + "learning_rate": 2.628509221450107e-06, + "loss": 0.06522426009178162, + "step": 4225 + }, + { + "epoch": 2.3636363636363638, + "grad_norm": 0.36205568226762697, + "learning_rate": 2.6241128777241844e-06, + "loss": 0.02990780584514141, + "step": 4226 + }, + { + "epoch": 2.364195804195804, + "grad_norm": 0.5608823152415278, + "learning_rate": 2.6197196583462104e-06, + "loss": 0.06367485970258713, + "step": 4227 + }, + { + "epoch": 2.3647552447552447, + "grad_norm": 0.2839700335351055, + "learning_rate": 2.6153295651771084e-06, + "loss": 0.019065149128437042, + "step": 4228 + }, + { + "epoch": 2.365314685314685, + "grad_norm": 0.33562204498814024, + "learning_rate": 2.610942600076476e-06, + "loss": 0.0381324477493763, + "step": 4229 + }, + { + "epoch": 2.365874125874126, + "grad_norm": 0.4841746950608386, + "learning_rate": 2.606558764902585e-06, + "loss": 0.24262671172618866, + "step": 4230 + }, + { + "epoch": 2.3664335664335665, + "grad_norm": 0.5278484582668506, + "learning_rate": 2.602178061512386e-06, + "loss": 0.133316770195961, + "step": 4231 + }, + { + "epoch": 2.366993006993007, + "grad_norm": 0.4412018453110211, + "learning_rate": 2.597800491761492e-06, + "loss": 0.14182843267917633, + "step": 4232 + }, + { + "epoch": 2.3675524475524474, + "grad_norm": 0.6018920126545529, + "learning_rate": 2.5934260575041968e-06, + "loss": 0.18291951715946198, + "step": 4233 + }, + { + "epoch": 2.368111888111888, + "grad_norm": 0.4138904584306407, + "learning_rate": 2.589054760593468e-06, + "loss": 0.06759610772132874, + "step": 4234 + }, + { + "epoch": 2.368671328671329, + "grad_norm": 0.5658080451313038, + "learning_rate": 2.5846866028809404e-06, + "loss": 0.1317383348941803, + "step": 4235 + }, + { + "epoch": 2.3692307692307693, + "grad_norm": 0.5936334069788327, + "learning_rate": 2.580321586216916e-06, + "loss": 0.12118326127529144, + "step": 4236 + }, + { + "epoch": 2.3697902097902097, + "grad_norm": 0.5590434853622305, + "learning_rate": 2.575959712450371e-06, + "loss": 0.11352255195379257, + "step": 4237 + }, + { + "epoch": 2.37034965034965, + "grad_norm": 0.6153495687365511, + "learning_rate": 2.57160098342895e-06, + "loss": 0.15802036225795746, + "step": 4238 + }, + { + "epoch": 2.370909090909091, + "grad_norm": 0.31440503564237426, + "learning_rate": 2.5672454009989645e-06, + "loss": 0.04245789349079132, + "step": 4239 + }, + { + "epoch": 2.3714685314685315, + "grad_norm": 0.5310291917682838, + "learning_rate": 2.562892967005397e-06, + "loss": 0.14414645731449127, + "step": 4240 + }, + { + "epoch": 2.372027972027972, + "grad_norm": 0.47817502663110284, + "learning_rate": 2.5585436832918853e-06, + "loss": 0.04403708502650261, + "step": 4241 + }, + { + "epoch": 2.3725874125874125, + "grad_norm": 0.5030431272669261, + "learning_rate": 2.5541975517007433e-06, + "loss": 0.11939035356044769, + "step": 4242 + }, + { + "epoch": 2.373146853146853, + "grad_norm": 0.4112692391588895, + "learning_rate": 2.5498545740729474e-06, + "loss": 0.06096184253692627, + "step": 4243 + }, + { + "epoch": 2.373706293706294, + "grad_norm": 0.4732398605939355, + "learning_rate": 2.545514752248136e-06, + "loss": 0.07828359305858612, + "step": 4244 + }, + { + "epoch": 2.3742657342657343, + "grad_norm": 0.35132275310323663, + "learning_rate": 2.541178088064615e-06, + "loss": 0.05178485065698624, + "step": 4245 + }, + { + "epoch": 2.3748251748251747, + "grad_norm": 0.43780290750247336, + "learning_rate": 2.5368445833593423e-06, + "loss": 0.09666630625724792, + "step": 4246 + }, + { + "epoch": 2.375384615384615, + "grad_norm": 0.5641104313001892, + "learning_rate": 2.532514239967955e-06, + "loss": 0.1351543366909027, + "step": 4247 + }, + { + "epoch": 2.375944055944056, + "grad_norm": 0.519941118547285, + "learning_rate": 2.5281870597247315e-06, + "loss": 0.07165377587080002, + "step": 4248 + }, + { + "epoch": 2.3765034965034966, + "grad_norm": 0.36671743109659877, + "learning_rate": 2.5238630444626243e-06, + "loss": 0.05010291934013367, + "step": 4249 + }, + { + "epoch": 2.377062937062937, + "grad_norm": 0.2898776661307772, + "learning_rate": 2.5195421960132416e-06, + "loss": 0.015423586592078209, + "step": 4250 + }, + { + "epoch": 2.3776223776223775, + "grad_norm": 0.39283412902729353, + "learning_rate": 2.515224516206842e-06, + "loss": 0.05329650267958641, + "step": 4251 + }, + { + "epoch": 2.378181818181818, + "grad_norm": 0.4527211648009571, + "learning_rate": 2.5109100068723614e-06, + "loss": 0.11685922741889954, + "step": 4252 + }, + { + "epoch": 2.378741258741259, + "grad_norm": 0.3107902803692862, + "learning_rate": 2.506598669837369e-06, + "loss": 0.01302420161664486, + "step": 4253 + }, + { + "epoch": 2.3793006993006993, + "grad_norm": 0.684051945235495, + "learning_rate": 2.502290506928108e-06, + "loss": 0.12375587224960327, + "step": 4254 + }, + { + "epoch": 2.37986013986014, + "grad_norm": 0.5369358574196714, + "learning_rate": 2.4979855199694704e-06, + "loss": 0.08629684150218964, + "step": 4255 + }, + { + "epoch": 2.3804195804195802, + "grad_norm": 0.35470976293377704, + "learning_rate": 2.4936837107849966e-06, + "loss": 0.13536661863327026, + "step": 4256 + }, + { + "epoch": 2.380979020979021, + "grad_norm": 0.42652289417720424, + "learning_rate": 2.4893850811968977e-06, + "loss": 0.09614861011505127, + "step": 4257 + }, + { + "epoch": 2.3815384615384616, + "grad_norm": 0.48980503856572, + "learning_rate": 2.4850896330260244e-06, + "loss": 0.09313364326953888, + "step": 4258 + }, + { + "epoch": 2.382097902097902, + "grad_norm": 0.2997214579744771, + "learning_rate": 2.480797368091875e-06, + "loss": 0.024739248678088188, + "step": 4259 + }, + { + "epoch": 2.3826573426573425, + "grad_norm": 0.4070006803766942, + "learning_rate": 2.47650828821262e-06, + "loss": 0.058190375566482544, + "step": 4260 + }, + { + "epoch": 2.383216783216783, + "grad_norm": 0.5968629358012509, + "learning_rate": 2.4722223952050606e-06, + "loss": 0.13404443860054016, + "step": 4261 + }, + { + "epoch": 2.383776223776224, + "grad_norm": 0.6514512493680072, + "learning_rate": 2.4679396908846563e-06, + "loss": 0.12624502182006836, + "step": 4262 + }, + { + "epoch": 2.3843356643356644, + "grad_norm": 0.32668258697022806, + "learning_rate": 2.463660177065519e-06, + "loss": 0.05548630282282829, + "step": 4263 + }, + { + "epoch": 2.384895104895105, + "grad_norm": 0.2751610473960363, + "learning_rate": 2.4593838555604035e-06, + "loss": 0.024162402376532555, + "step": 4264 + }, + { + "epoch": 2.3854545454545453, + "grad_norm": 0.39781524802967083, + "learning_rate": 2.455110728180716e-06, + "loss": 0.04397023841738701, + "step": 4265 + }, + { + "epoch": 2.386013986013986, + "grad_norm": 0.08763417974532288, + "learning_rate": 2.4508407967365054e-06, + "loss": 0.007988613098859787, + "step": 4266 + }, + { + "epoch": 2.3865734265734266, + "grad_norm": 0.3845410335908805, + "learning_rate": 2.4465740630364707e-06, + "loss": 0.04263989254832268, + "step": 4267 + }, + { + "epoch": 2.387132867132867, + "grad_norm": 0.5715897409607095, + "learning_rate": 2.4423105288879556e-06, + "loss": 0.13074183464050293, + "step": 4268 + }, + { + "epoch": 2.3876923076923076, + "grad_norm": 0.4678193047692492, + "learning_rate": 2.438050196096948e-06, + "loss": 0.07761231809854507, + "step": 4269 + }, + { + "epoch": 2.388251748251748, + "grad_norm": 0.4475335726082913, + "learning_rate": 2.4337930664680843e-06, + "loss": 0.155867338180542, + "step": 4270 + }, + { + "epoch": 2.388811188811189, + "grad_norm": 0.6421728395434695, + "learning_rate": 2.429539141804631e-06, + "loss": 0.22660008072853088, + "step": 4271 + }, + { + "epoch": 2.3893706293706294, + "grad_norm": 0.6214513096254368, + "learning_rate": 2.4252884239085105e-06, + "loss": 0.2083190381526947, + "step": 4272 + }, + { + "epoch": 2.38993006993007, + "grad_norm": 0.4134965953855934, + "learning_rate": 2.4210409145802825e-06, + "loss": 0.073543980717659, + "step": 4273 + }, + { + "epoch": 2.3904895104895103, + "grad_norm": 0.22550346106679875, + "learning_rate": 2.416796615619146e-06, + "loss": 0.006336988881230354, + "step": 4274 + }, + { + "epoch": 2.391048951048951, + "grad_norm": 0.5562980464337148, + "learning_rate": 2.412555528822944e-06, + "loss": 0.1102716252207756, + "step": 4275 + }, + { + "epoch": 2.3916083916083917, + "grad_norm": 0.4124868018281879, + "learning_rate": 2.4083176559881517e-06, + "loss": 0.16808021068572998, + "step": 4276 + }, + { + "epoch": 2.392167832167832, + "grad_norm": 0.4230582513555001, + "learning_rate": 2.404082998909889e-06, + "loss": 0.13183780014514923, + "step": 4277 + }, + { + "epoch": 2.3927272727272726, + "grad_norm": 0.3727873819963714, + "learning_rate": 2.3998515593819126e-06, + "loss": 0.04239516705274582, + "step": 4278 + }, + { + "epoch": 2.393286713286713, + "grad_norm": 0.3132800339647603, + "learning_rate": 2.395623339196614e-06, + "loss": 0.032993134111166, + "step": 4279 + }, + { + "epoch": 2.393846153846154, + "grad_norm": 0.6698989712609587, + "learning_rate": 2.391398340145028e-06, + "loss": 0.20396141707897186, + "step": 4280 + }, + { + "epoch": 2.3944055944055944, + "grad_norm": 0.2890715602896259, + "learning_rate": 2.387176564016813e-06, + "loss": 0.008912870660424232, + "step": 4281 + }, + { + "epoch": 2.394965034965035, + "grad_norm": 0.6021106399580413, + "learning_rate": 2.3829580126002715e-06, + "loss": 0.06428366899490356, + "step": 4282 + }, + { + "epoch": 2.3955244755244753, + "grad_norm": 0.4798890800081331, + "learning_rate": 2.378742687682336e-06, + "loss": 0.16745463013648987, + "step": 4283 + }, + { + "epoch": 2.3960839160839162, + "grad_norm": 0.4076055644954354, + "learning_rate": 2.3745305910485795e-06, + "loss": 0.053865693509578705, + "step": 4284 + }, + { + "epoch": 2.3966433566433567, + "grad_norm": 0.4360711186977314, + "learning_rate": 2.37032172448319e-06, + "loss": 0.06182485073804855, + "step": 4285 + }, + { + "epoch": 2.397202797202797, + "grad_norm": 0.7655951762436283, + "learning_rate": 2.3661160897690127e-06, + "loss": 0.09955702722072601, + "step": 4286 + }, + { + "epoch": 2.3977622377622376, + "grad_norm": 0.38406201430095055, + "learning_rate": 2.3619136886875005e-06, + "loss": 0.06909747421741486, + "step": 4287 + }, + { + "epoch": 2.398321678321678, + "grad_norm": 0.29134406951836633, + "learning_rate": 2.357714523018748e-06, + "loss": 0.024588461965322495, + "step": 4288 + }, + { + "epoch": 2.398881118881119, + "grad_norm": 0.5575924655362992, + "learning_rate": 2.353518594541483e-06, + "loss": 0.1883913278579712, + "step": 4289 + }, + { + "epoch": 2.3994405594405595, + "grad_norm": 0.7139544739369856, + "learning_rate": 2.3493259050330452e-06, + "loss": 0.2154640108346939, + "step": 4290 + }, + { + "epoch": 2.4, + "grad_norm": 0.3811340549878349, + "learning_rate": 2.3451364562694267e-06, + "loss": 0.039584964513778687, + "step": 4291 + }, + { + "epoch": 2.4005594405594404, + "grad_norm": 0.5999633773203551, + "learning_rate": 2.3409502500252258e-06, + "loss": 0.1151922345161438, + "step": 4292 + }, + { + "epoch": 2.4011188811188813, + "grad_norm": 0.5398394663311386, + "learning_rate": 2.336767288073676e-06, + "loss": 0.0916643887758255, + "step": 4293 + }, + { + "epoch": 2.4016783216783217, + "grad_norm": 0.28651752986154255, + "learning_rate": 2.3325875721866396e-06, + "loss": 0.023895790800452232, + "step": 4294 + }, + { + "epoch": 2.402237762237762, + "grad_norm": 0.33266062450155, + "learning_rate": 2.3284111041345923e-06, + "loss": 0.07810373604297638, + "step": 4295 + }, + { + "epoch": 2.4027972027972027, + "grad_norm": 0.4677620367575912, + "learning_rate": 2.3242378856866534e-06, + "loss": 0.04361053183674812, + "step": 4296 + }, + { + "epoch": 2.403356643356643, + "grad_norm": 0.556346115181841, + "learning_rate": 2.3200679186105446e-06, + "loss": 0.0860578715801239, + "step": 4297 + }, + { + "epoch": 2.403916083916084, + "grad_norm": 0.338190127000119, + "learning_rate": 2.3159012046726228e-06, + "loss": 0.03552931547164917, + "step": 4298 + }, + { + "epoch": 2.4044755244755245, + "grad_norm": 0.4895187587612216, + "learning_rate": 2.3117377456378677e-06, + "loss": 0.1972390115261078, + "step": 4299 + }, + { + "epoch": 2.405034965034965, + "grad_norm": 0.6544756033770618, + "learning_rate": 2.3075775432698678e-06, + "loss": 0.14579619467258453, + "step": 4300 + }, + { + "epoch": 2.4055944055944054, + "grad_norm": 0.5755050857991783, + "learning_rate": 2.3034205993308525e-06, + "loss": 0.12425071001052856, + "step": 4301 + }, + { + "epoch": 2.4061538461538463, + "grad_norm": 0.9849345881866202, + "learning_rate": 2.2992669155816496e-06, + "loss": 0.1435418426990509, + "step": 4302 + }, + { + "epoch": 2.4067132867132868, + "grad_norm": 0.39070909039360324, + "learning_rate": 2.2951164937817206e-06, + "loss": 0.031560156494379044, + "step": 4303 + }, + { + "epoch": 2.4072727272727272, + "grad_norm": 0.43201374576276425, + "learning_rate": 2.2909693356891418e-06, + "loss": 0.09113756567239761, + "step": 4304 + }, + { + "epoch": 2.4078321678321677, + "grad_norm": 0.29830092881970954, + "learning_rate": 2.2868254430605975e-06, + "loss": 0.03356871008872986, + "step": 4305 + }, + { + "epoch": 2.408391608391608, + "grad_norm": 0.6243016366479079, + "learning_rate": 2.2826848176514074e-06, + "loss": 0.14119921624660492, + "step": 4306 + }, + { + "epoch": 2.408951048951049, + "grad_norm": 0.5662355671175362, + "learning_rate": 2.2785474612154902e-06, + "loss": 0.11919906735420227, + "step": 4307 + }, + { + "epoch": 2.4095104895104895, + "grad_norm": 0.37827051445177634, + "learning_rate": 2.2744133755053866e-06, + "loss": 0.0956508070230484, + "step": 4308 + }, + { + "epoch": 2.41006993006993, + "grad_norm": 0.46869086665160015, + "learning_rate": 2.270282562272257e-06, + "loss": 0.07001488655805588, + "step": 4309 + }, + { + "epoch": 2.4106293706293704, + "grad_norm": 0.5127387282191705, + "learning_rate": 2.2661550232658637e-06, + "loss": 0.11895737797021866, + "step": 4310 + }, + { + "epoch": 2.4111888111888113, + "grad_norm": 0.8910865501185814, + "learning_rate": 2.2620307602345916e-06, + "loss": 0.19488152861595154, + "step": 4311 + }, + { + "epoch": 2.411748251748252, + "grad_norm": 0.4154564073010456, + "learning_rate": 2.257909774925434e-06, + "loss": 0.0734558254480362, + "step": 4312 + }, + { + "epoch": 2.4123076923076923, + "grad_norm": 0.3546147012205755, + "learning_rate": 2.2537920690839976e-06, + "loss": 0.061678335070610046, + "step": 4313 + }, + { + "epoch": 2.4128671328671327, + "grad_norm": 0.3873927998560919, + "learning_rate": 2.249677644454503e-06, + "loss": 0.026166999712586403, + "step": 4314 + }, + { + "epoch": 2.413426573426573, + "grad_norm": 0.06610145524882882, + "learning_rate": 2.24556650277977e-06, + "loss": 0.005541364662349224, + "step": 4315 + }, + { + "epoch": 2.413986013986014, + "grad_norm": 0.6431056999313475, + "learning_rate": 2.2414586458012375e-06, + "loss": 0.12966686487197876, + "step": 4316 + }, + { + "epoch": 2.4145454545454546, + "grad_norm": 0.406300505215019, + "learning_rate": 2.237354075258952e-06, + "loss": 0.09022810310125351, + "step": 4317 + }, + { + "epoch": 2.415104895104895, + "grad_norm": 0.4040273626710413, + "learning_rate": 2.2332527928915637e-06, + "loss": 0.038913995027542114, + "step": 4318 + }, + { + "epoch": 2.4156643356643355, + "grad_norm": 0.518359762310585, + "learning_rate": 2.229154800436335e-06, + "loss": 0.07941877841949463, + "step": 4319 + }, + { + "epoch": 2.4162237762237764, + "grad_norm": 0.5914332039550179, + "learning_rate": 2.225060099629134e-06, + "loss": 0.11963687092065811, + "step": 4320 + }, + { + "epoch": 2.416783216783217, + "grad_norm": 0.37186058725384563, + "learning_rate": 2.220968692204426e-06, + "loss": 0.0468156561255455, + "step": 4321 + }, + { + "epoch": 2.4173426573426573, + "grad_norm": 0.47141807401492325, + "learning_rate": 2.216880579895293e-06, + "loss": 0.08945789188146591, + "step": 4322 + }, + { + "epoch": 2.4179020979020978, + "grad_norm": 0.34955083301707396, + "learning_rate": 2.2127957644334154e-06, + "loss": 0.07148704677820206, + "step": 4323 + }, + { + "epoch": 2.418461538461538, + "grad_norm": 0.5708768990644683, + "learning_rate": 2.2087142475490775e-06, + "loss": 0.12272177636623383, + "step": 4324 + }, + { + "epoch": 2.419020979020979, + "grad_norm": 0.5396466162309228, + "learning_rate": 2.20463603097117e-06, + "loss": 0.10421647131443024, + "step": 4325 + }, + { + "epoch": 2.4195804195804196, + "grad_norm": 0.42234614326668823, + "learning_rate": 2.2005611164271766e-06, + "loss": 0.04226556420326233, + "step": 4326 + }, + { + "epoch": 2.42013986013986, + "grad_norm": 0.331529797930077, + "learning_rate": 2.196489505643191e-06, + "loss": 0.010642962530255318, + "step": 4327 + }, + { + "epoch": 2.420699300699301, + "grad_norm": 0.412257896860218, + "learning_rate": 2.192421200343905e-06, + "loss": 0.08660288900136948, + "step": 4328 + }, + { + "epoch": 2.4212587412587414, + "grad_norm": 0.44490079944119376, + "learning_rate": 2.1883562022526083e-06, + "loss": 0.13848651945590973, + "step": 4329 + }, + { + "epoch": 2.421818181818182, + "grad_norm": 0.43944713404358055, + "learning_rate": 2.1842945130911965e-06, + "loss": 0.08055999875068665, + "step": 4330 + }, + { + "epoch": 2.4223776223776223, + "grad_norm": 0.36815972864873525, + "learning_rate": 2.1802361345801493e-06, + "loss": 0.03976232558488846, + "step": 4331 + }, + { + "epoch": 2.422937062937063, + "grad_norm": 0.3922358851924399, + "learning_rate": 2.176181068438559e-06, + "loss": 0.04665861651301384, + "step": 4332 + }, + { + "epoch": 2.4234965034965033, + "grad_norm": 0.36952668818745993, + "learning_rate": 2.1721293163841108e-06, + "loss": 0.021116310730576515, + "step": 4333 + }, + { + "epoch": 2.424055944055944, + "grad_norm": 0.2775703272498425, + "learning_rate": 2.1680808801330733e-06, + "loss": 0.054423797875642776, + "step": 4334 + }, + { + "epoch": 2.4246153846153846, + "grad_norm": 0.29254704292985223, + "learning_rate": 2.164035761400336e-06, + "loss": 0.02370445430278778, + "step": 4335 + }, + { + "epoch": 2.425174825174825, + "grad_norm": 0.44935303325783993, + "learning_rate": 2.159993961899358e-06, + "loss": 0.06737499684095383, + "step": 4336 + }, + { + "epoch": 2.425734265734266, + "grad_norm": 0.5121432414624416, + "learning_rate": 2.155955483342207e-06, + "loss": 0.05144071951508522, + "step": 4337 + }, + { + "epoch": 2.4262937062937064, + "grad_norm": 0.5435570720201567, + "learning_rate": 2.1519203274395417e-06, + "loss": 0.12657935917377472, + "step": 4338 + }, + { + "epoch": 2.426853146853147, + "grad_norm": 0.6924809132183158, + "learning_rate": 2.147888495900604e-06, + "loss": 0.13071051239967346, + "step": 4339 + }, + { + "epoch": 2.4274125874125874, + "grad_norm": 0.4158430622841348, + "learning_rate": 2.1438599904332445e-06, + "loss": 0.07877501845359802, + "step": 4340 + }, + { + "epoch": 2.427972027972028, + "grad_norm": 0.3661548859597376, + "learning_rate": 2.1398348127438905e-06, + "loss": 0.08164171874523163, + "step": 4341 + }, + { + "epoch": 2.4285314685314683, + "grad_norm": 0.32350796235933144, + "learning_rate": 2.1358129645375648e-06, + "loss": 0.01728185825049877, + "step": 4342 + }, + { + "epoch": 2.429090909090909, + "grad_norm": 0.4370602283131204, + "learning_rate": 2.131794447517883e-06, + "loss": 0.06505485624074936, + "step": 4343 + }, + { + "epoch": 2.4296503496503496, + "grad_norm": 0.3953072449143357, + "learning_rate": 2.1277792633870386e-06, + "loss": 0.06129543483257294, + "step": 4344 + }, + { + "epoch": 2.43020979020979, + "grad_norm": 0.6453665454432505, + "learning_rate": 2.123767413845833e-06, + "loss": 0.09214981645345688, + "step": 4345 + }, + { + "epoch": 2.430769230769231, + "grad_norm": 0.5606613053643283, + "learning_rate": 2.1197589005936346e-06, + "loss": 0.13617411255836487, + "step": 4346 + }, + { + "epoch": 2.4313286713286715, + "grad_norm": 0.32285102617263345, + "learning_rate": 2.11575372532841e-06, + "loss": 0.0847213938832283, + "step": 4347 + }, + { + "epoch": 2.431888111888112, + "grad_norm": 0.38824537136755205, + "learning_rate": 2.1117518897467113e-06, + "loss": 0.027891680598258972, + "step": 4348 + }, + { + "epoch": 2.4324475524475524, + "grad_norm": 0.527384877425298, + "learning_rate": 2.107753395543667e-06, + "loss": 0.08984841406345367, + "step": 4349 + }, + { + "epoch": 2.433006993006993, + "grad_norm": 0.6220423218612589, + "learning_rate": 2.1037582444130067e-06, + "loss": 0.09435883164405823, + "step": 4350 + }, + { + "epoch": 2.4335664335664333, + "grad_norm": 0.5253616801330756, + "learning_rate": 2.0997664380470262e-06, + "loss": 0.11791590601205826, + "step": 4351 + }, + { + "epoch": 2.4341258741258742, + "grad_norm": 0.7627284503350098, + "learning_rate": 2.095777978136616e-06, + "loss": 0.1421177089214325, + "step": 4352 + }, + { + "epoch": 2.4346853146853147, + "grad_norm": 0.20581738753832324, + "learning_rate": 2.0917928663712484e-06, + "loss": 0.027294915169477463, + "step": 4353 + }, + { + "epoch": 2.435244755244755, + "grad_norm": 0.4931636666579357, + "learning_rate": 2.087811104438965e-06, + "loss": 0.11855275183916092, + "step": 4354 + }, + { + "epoch": 2.435804195804196, + "grad_norm": 0.5827032862796142, + "learning_rate": 2.083832694026411e-06, + "loss": 0.1294851154088974, + "step": 4355 + }, + { + "epoch": 2.4363636363636365, + "grad_norm": 0.4297615018906224, + "learning_rate": 2.0798576368187893e-06, + "loss": 0.058321014046669006, + "step": 4356 + }, + { + "epoch": 2.436923076923077, + "grad_norm": 0.3387371974219107, + "learning_rate": 2.075885934499896e-06, + "loss": 0.021740924566984177, + "step": 4357 + }, + { + "epoch": 2.4374825174825174, + "grad_norm": 0.4503632573516786, + "learning_rate": 2.0719175887521016e-06, + "loss": 0.08060051500797272, + "step": 4358 + }, + { + "epoch": 2.438041958041958, + "grad_norm": 0.6240243346283557, + "learning_rate": 2.067952601256359e-06, + "loss": 0.08996862918138504, + "step": 4359 + }, + { + "epoch": 2.438601398601399, + "grad_norm": 0.43926704249635384, + "learning_rate": 2.0639909736921892e-06, + "loss": 0.03524692356586456, + "step": 4360 + }, + { + "epoch": 2.4391608391608393, + "grad_norm": 0.48832503048305026, + "learning_rate": 2.060032707737698e-06, + "loss": 0.03271853178739548, + "step": 4361 + }, + { + "epoch": 2.4397202797202797, + "grad_norm": 0.4270769591383042, + "learning_rate": 2.056077805069565e-06, + "loss": 0.06356614083051682, + "step": 4362 + }, + { + "epoch": 2.44027972027972, + "grad_norm": 0.35786073182687217, + "learning_rate": 2.052126267363047e-06, + "loss": 0.032593294978141785, + "step": 4363 + }, + { + "epoch": 2.440839160839161, + "grad_norm": 0.37880238658141546, + "learning_rate": 2.048178096291975e-06, + "loss": 0.07341725379228592, + "step": 4364 + }, + { + "epoch": 2.4413986013986015, + "grad_norm": 0.5110810272381494, + "learning_rate": 2.0442332935287466e-06, + "loss": 0.1449374258518219, + "step": 4365 + }, + { + "epoch": 2.441958041958042, + "grad_norm": 0.3498041928674665, + "learning_rate": 2.040291860744343e-06, + "loss": 0.053326014429330826, + "step": 4366 + }, + { + "epoch": 2.4425174825174825, + "grad_norm": 0.5666833421554016, + "learning_rate": 2.0363537996083114e-06, + "loss": 0.17070254683494568, + "step": 4367 + }, + { + "epoch": 2.443076923076923, + "grad_norm": 0.48786383227587304, + "learning_rate": 2.032419111788776e-06, + "loss": 0.09563805907964706, + "step": 4368 + }, + { + "epoch": 2.443636363636364, + "grad_norm": 0.33663554176328175, + "learning_rate": 2.0284877989524285e-06, + "loss": 0.04321911185979843, + "step": 4369 + }, + { + "epoch": 2.4441958041958043, + "grad_norm": 0.30440972678575967, + "learning_rate": 2.024559862764528e-06, + "loss": 0.03923831507563591, + "step": 4370 + }, + { + "epoch": 2.4447552447552447, + "grad_norm": 0.4343947750352327, + "learning_rate": 2.0206353048889093e-06, + "loss": 0.057186614722013474, + "step": 4371 + }, + { + "epoch": 2.445314685314685, + "grad_norm": 0.30392819393359566, + "learning_rate": 2.016714126987973e-06, + "loss": 0.04584155231714249, + "step": 4372 + }, + { + "epoch": 2.445874125874126, + "grad_norm": 0.23845285161071866, + "learning_rate": 2.0127963307226894e-06, + "loss": 0.02422763779759407, + "step": 4373 + }, + { + "epoch": 2.4464335664335666, + "grad_norm": 0.4529752818402264, + "learning_rate": 2.008881917752599e-06, + "loss": 0.09444719552993774, + "step": 4374 + }, + { + "epoch": 2.446993006993007, + "grad_norm": 0.4635136568118181, + "learning_rate": 2.0049708897357977e-06, + "loss": 0.09193938970565796, + "step": 4375 + }, + { + "epoch": 2.4475524475524475, + "grad_norm": 0.336126969214592, + "learning_rate": 2.0010632483289616e-06, + "loss": 0.04639626666903496, + "step": 4376 + }, + { + "epoch": 2.448111888111888, + "grad_norm": 0.3466302396538429, + "learning_rate": 1.997158995187325e-06, + "loss": 0.03390798345208168, + "step": 4377 + }, + { + "epoch": 2.448671328671329, + "grad_norm": 0.6108162762545862, + "learning_rate": 1.993258131964687e-06, + "loss": 0.1308169662952423, + "step": 4378 + }, + { + "epoch": 2.4492307692307693, + "grad_norm": 0.5081788079666687, + "learning_rate": 1.9893606603134174e-06, + "loss": 0.09462116658687592, + "step": 4379 + }, + { + "epoch": 2.44979020979021, + "grad_norm": 0.4126119457033221, + "learning_rate": 1.985466581884434e-06, + "loss": 0.06645186990499496, + "step": 4380 + }, + { + "epoch": 2.4503496503496502, + "grad_norm": 0.39940370761670246, + "learning_rate": 1.9815758983272404e-06, + "loss": 0.04732794687151909, + "step": 4381 + }, + { + "epoch": 2.450909090909091, + "grad_norm": 0.5373356469815228, + "learning_rate": 1.977688611289881e-06, + "loss": 0.09362974017858505, + "step": 4382 + }, + { + "epoch": 2.4514685314685316, + "grad_norm": 0.4005843824407196, + "learning_rate": 1.973804722418965e-06, + "loss": 0.034056156873703, + "step": 4383 + }, + { + "epoch": 2.452027972027972, + "grad_norm": 0.38440657375489423, + "learning_rate": 1.969924233359678e-06, + "loss": 0.028239693492650986, + "step": 4384 + }, + { + "epoch": 2.4525874125874125, + "grad_norm": 0.2309304218820193, + "learning_rate": 1.9660471457557462e-06, + "loss": 0.007724790833890438, + "step": 4385 + }, + { + "epoch": 2.453146853146853, + "grad_norm": 0.44948734965050036, + "learning_rate": 1.9621734612494646e-06, + "loss": 0.08283065259456635, + "step": 4386 + }, + { + "epoch": 2.453706293706294, + "grad_norm": 0.4623300891126834, + "learning_rate": 1.95830318148169e-06, + "loss": 0.08691463619470596, + "step": 4387 + }, + { + "epoch": 2.4542657342657344, + "grad_norm": 0.3222859037091408, + "learning_rate": 1.9544363080918216e-06, + "loss": 0.03459593653678894, + "step": 4388 + }, + { + "epoch": 2.454825174825175, + "grad_norm": 0.5200569039560013, + "learning_rate": 1.9505728427178384e-06, + "loss": 0.11298046261072159, + "step": 4389 + }, + { + "epoch": 2.4553846153846153, + "grad_norm": 0.5299361858269264, + "learning_rate": 1.9467127869962543e-06, + "loss": 0.12835749983787537, + "step": 4390 + }, + { + "epoch": 2.455944055944056, + "grad_norm": 0.24110456020984405, + "learning_rate": 1.942856142562153e-06, + "loss": 0.013115103356540203, + "step": 4391 + }, + { + "epoch": 2.4565034965034966, + "grad_norm": 0.3621226816991062, + "learning_rate": 1.939002911049168e-06, + "loss": 0.07676021009683609, + "step": 4392 + }, + { + "epoch": 2.457062937062937, + "grad_norm": 0.6363649909437455, + "learning_rate": 1.9351530940894827e-06, + "loss": 0.07798892259597778, + "step": 4393 + }, + { + "epoch": 2.4576223776223776, + "grad_norm": 0.27144975678770517, + "learning_rate": 1.931306693313847e-06, + "loss": 0.056541360914707184, + "step": 4394 + }, + { + "epoch": 2.458181818181818, + "grad_norm": 0.44151449713629115, + "learning_rate": 1.927463710351549e-06, + "loss": 0.08987961709499359, + "step": 4395 + }, + { + "epoch": 2.458741258741259, + "grad_norm": 0.5840018488290561, + "learning_rate": 1.9236241468304385e-06, + "loss": 0.15675915777683258, + "step": 4396 + }, + { + "epoch": 2.4593006993006994, + "grad_norm": 0.5386754083589831, + "learning_rate": 1.9197880043769122e-06, + "loss": 0.04454007372260094, + "step": 4397 + }, + { + "epoch": 2.45986013986014, + "grad_norm": 0.5075682881458985, + "learning_rate": 1.9159552846159233e-06, + "loss": 0.08004562556743622, + "step": 4398 + }, + { + "epoch": 2.4604195804195803, + "grad_norm": 0.39992476807045124, + "learning_rate": 1.912125989170971e-06, + "loss": 0.061470042914152145, + "step": 4399 + }, + { + "epoch": 2.460979020979021, + "grad_norm": 0.6024547933640173, + "learning_rate": 1.9083001196640994e-06, + "loss": 0.12167413532733917, + "step": 4400 + }, + { + "epoch": 2.4615384615384617, + "grad_norm": 0.35760629842233865, + "learning_rate": 1.90447767771591e-06, + "loss": 0.023522404953837395, + "step": 4401 + }, + { + "epoch": 2.462097902097902, + "grad_norm": 0.3188595259261814, + "learning_rate": 1.9006586649455495e-06, + "loss": 0.032641515135765076, + "step": 4402 + }, + { + "epoch": 2.4626573426573426, + "grad_norm": 0.2938349913431669, + "learning_rate": 1.8968430829707097e-06, + "loss": 0.04072485864162445, + "step": 4403 + }, + { + "epoch": 2.463216783216783, + "grad_norm": 0.5321378694202894, + "learning_rate": 1.8930309334076346e-06, + "loss": 0.10253074765205383, + "step": 4404 + }, + { + "epoch": 2.463776223776224, + "grad_norm": 0.48275614481800405, + "learning_rate": 1.889222217871104e-06, + "loss": 0.14560942351818085, + "step": 4405 + }, + { + "epoch": 2.4643356643356644, + "grad_norm": 0.7694279010050857, + "learning_rate": 1.8854169379744548e-06, + "loss": 0.16155387461185455, + "step": 4406 + }, + { + "epoch": 2.464895104895105, + "grad_norm": 0.4755815437229903, + "learning_rate": 1.881615095329561e-06, + "loss": 0.06448402255773544, + "step": 4407 + }, + { + "epoch": 2.4654545454545453, + "grad_norm": 0.2705554872418293, + "learning_rate": 1.8778166915468465e-06, + "loss": 0.07830452173948288, + "step": 4408 + }, + { + "epoch": 2.4660139860139862, + "grad_norm": 0.40035408108792786, + "learning_rate": 1.8740217282352691e-06, + "loss": 0.05782242864370346, + "step": 4409 + }, + { + "epoch": 2.4665734265734267, + "grad_norm": 0.34362919506705597, + "learning_rate": 1.8702302070023392e-06, + "loss": 0.10452114045619965, + "step": 4410 + }, + { + "epoch": 2.467132867132867, + "grad_norm": 0.7794046504943157, + "learning_rate": 1.8664421294541046e-06, + "loss": 0.0969105213880539, + "step": 4411 + }, + { + "epoch": 2.4676923076923076, + "grad_norm": 0.643559937603698, + "learning_rate": 1.862657497195155e-06, + "loss": 0.101371169090271, + "step": 4412 + }, + { + "epoch": 2.468251748251748, + "grad_norm": 0.49273143622853927, + "learning_rate": 1.8588763118286223e-06, + "loss": 0.1356501281261444, + "step": 4413 + }, + { + "epoch": 2.468811188811189, + "grad_norm": 0.528425087313542, + "learning_rate": 1.8550985749561712e-06, + "loss": 0.11275427043437958, + "step": 4414 + }, + { + "epoch": 2.4693706293706295, + "grad_norm": 0.47862918234746044, + "learning_rate": 1.8513242881780192e-06, + "loss": 0.13677522540092468, + "step": 4415 + }, + { + "epoch": 2.46993006993007, + "grad_norm": 0.5642437063761355, + "learning_rate": 1.8475534530929074e-06, + "loss": 0.16282282769680023, + "step": 4416 + }, + { + "epoch": 2.4704895104895104, + "grad_norm": 0.35914642998671, + "learning_rate": 1.843786071298125e-06, + "loss": 0.1123458594083786, + "step": 4417 + }, + { + "epoch": 2.4710489510489513, + "grad_norm": 0.5919524696385559, + "learning_rate": 1.8400221443894971e-06, + "loss": 0.1560671031475067, + "step": 4418 + }, + { + "epoch": 2.4716083916083917, + "grad_norm": 0.5580737880488915, + "learning_rate": 1.836261673961376e-06, + "loss": 0.1323964148759842, + "step": 4419 + }, + { + "epoch": 2.472167832167832, + "grad_norm": 0.22674585596130906, + "learning_rate": 1.8325046616066667e-06, + "loss": 0.016651686280965805, + "step": 4420 + }, + { + "epoch": 2.4727272727272727, + "grad_norm": 0.7437739184674806, + "learning_rate": 1.8287511089167932e-06, + "loss": 0.1184217780828476, + "step": 4421 + }, + { + "epoch": 2.473286713286713, + "grad_norm": 0.6534746572581794, + "learning_rate": 1.8250010174817234e-06, + "loss": 0.17062152922153473, + "step": 4422 + }, + { + "epoch": 2.473846153846154, + "grad_norm": 0.41308990087955016, + "learning_rate": 1.821254388889958e-06, + "loss": 0.08395298570394516, + "step": 4423 + }, + { + "epoch": 2.4744055944055945, + "grad_norm": 0.25277122089669574, + "learning_rate": 1.817511224728523e-06, + "loss": 0.051518164575099945, + "step": 4424 + }, + { + "epoch": 2.474965034965035, + "grad_norm": 1.851501866101314, + "learning_rate": 1.8137715265829937e-06, + "loss": 0.06093098223209381, + "step": 4425 + }, + { + "epoch": 2.4755244755244754, + "grad_norm": 0.40700070927209775, + "learning_rate": 1.8100352960374568e-06, + "loss": 0.050407372415065765, + "step": 4426 + }, + { + "epoch": 2.4760839160839163, + "grad_norm": 0.3894607663929529, + "learning_rate": 1.8063025346745455e-06, + "loss": 0.0873202234506607, + "step": 4427 + }, + { + "epoch": 2.4766433566433568, + "grad_norm": 0.3897363463156646, + "learning_rate": 1.8025732440754184e-06, + "loss": 0.11311569809913635, + "step": 4428 + }, + { + "epoch": 2.4772027972027972, + "grad_norm": 0.40174710821171983, + "learning_rate": 1.7988474258197574e-06, + "loss": 0.15003004670143127, + "step": 4429 + }, + { + "epoch": 2.4777622377622377, + "grad_norm": 1.942134423153005, + "learning_rate": 1.795125081485789e-06, + "loss": 0.1436362862586975, + "step": 4430 + }, + { + "epoch": 2.478321678321678, + "grad_norm": 0.4803510429202762, + "learning_rate": 1.7914062126502551e-06, + "loss": 0.12151581048965454, + "step": 4431 + }, + { + "epoch": 2.478881118881119, + "grad_norm": 0.3122090794320906, + "learning_rate": 1.7876908208884236e-06, + "loss": 0.03245886415243149, + "step": 4432 + }, + { + "epoch": 2.4794405594405595, + "grad_norm": 0.1681364375162296, + "learning_rate": 1.7839789077741043e-06, + "loss": 0.021765537559986115, + "step": 4433 + }, + { + "epoch": 2.48, + "grad_norm": 0.3221750402192679, + "learning_rate": 1.7802704748796185e-06, + "loss": 0.04360999912023544, + "step": 4434 + }, + { + "epoch": 2.4805594405594404, + "grad_norm": 0.4185206153695559, + "learning_rate": 1.7765655237758217e-06, + "loss": 0.06616771966218948, + "step": 4435 + }, + { + "epoch": 2.4811188811188813, + "grad_norm": 0.5750698240724464, + "learning_rate": 1.772864056032091e-06, + "loss": 0.10865949094295502, + "step": 4436 + }, + { + "epoch": 2.481678321678322, + "grad_norm": 0.524388797032406, + "learning_rate": 1.7691660732163297e-06, + "loss": 0.1378260850906372, + "step": 4437 + }, + { + "epoch": 2.4822377622377623, + "grad_norm": 0.45375768340858996, + "learning_rate": 1.765471576894966e-06, + "loss": 0.05870186537504196, + "step": 4438 + }, + { + "epoch": 2.4827972027972027, + "grad_norm": 0.5977252232798682, + "learning_rate": 1.7617805686329458e-06, + "loss": 0.11319805681705475, + "step": 4439 + }, + { + "epoch": 2.483356643356643, + "grad_norm": 0.5219777936807452, + "learning_rate": 1.7580930499937431e-06, + "loss": 0.0752384141087532, + "step": 4440 + }, + { + "epoch": 2.483916083916084, + "grad_norm": 0.2354867850440798, + "learning_rate": 1.7544090225393529e-06, + "loss": 0.0386233776807785, + "step": 4441 + }, + { + "epoch": 2.4844755244755246, + "grad_norm": 0.436013390409342, + "learning_rate": 1.7507284878302887e-06, + "loss": 0.07144824415445328, + "step": 4442 + }, + { + "epoch": 2.485034965034965, + "grad_norm": 0.4830784205048499, + "learning_rate": 1.7470514474255895e-06, + "loss": 0.10955846309661865, + "step": 4443 + }, + { + "epoch": 2.4855944055944055, + "grad_norm": 0.41606645319210983, + "learning_rate": 1.7433779028828046e-06, + "loss": 0.09911411255598068, + "step": 4444 + }, + { + "epoch": 2.4861538461538464, + "grad_norm": 0.4431563014194301, + "learning_rate": 1.7397078557580139e-06, + "loss": 0.07427871227264404, + "step": 4445 + }, + { + "epoch": 2.486713286713287, + "grad_norm": 0.22880785832840544, + "learning_rate": 1.736041307605808e-06, + "loss": 0.011145062744617462, + "step": 4446 + }, + { + "epoch": 2.4872727272727273, + "grad_norm": 0.570748100669533, + "learning_rate": 1.7323782599792994e-06, + "loss": 0.07623221725225449, + "step": 4447 + }, + { + "epoch": 2.4878321678321678, + "grad_norm": 0.5465791632499157, + "learning_rate": 1.7287187144301187e-06, + "loss": 0.08683040738105774, + "step": 4448 + }, + { + "epoch": 2.4883916083916082, + "grad_norm": 0.2758046481555597, + "learning_rate": 1.7250626725084062e-06, + "loss": 0.020749079063534737, + "step": 4449 + }, + { + "epoch": 2.488951048951049, + "grad_norm": 0.28721372850646365, + "learning_rate": 1.7214101357628243e-06, + "loss": 0.03496145084500313, + "step": 4450 + }, + { + "epoch": 2.4895104895104896, + "grad_norm": 0.35272748223220607, + "learning_rate": 1.7177611057405486e-06, + "loss": 0.026832567527890205, + "step": 4451 + }, + { + "epoch": 2.49006993006993, + "grad_norm": 0.4106134124419564, + "learning_rate": 1.7141155839872714e-06, + "loss": 0.036321818828582764, + "step": 4452 + }, + { + "epoch": 2.4906293706293705, + "grad_norm": 0.304971359147559, + "learning_rate": 1.7104735720471954e-06, + "loss": 0.030157815665006638, + "step": 4453 + }, + { + "epoch": 2.4911888111888114, + "grad_norm": 0.6377074795737427, + "learning_rate": 1.7068350714630422e-06, + "loss": 0.15856845676898956, + "step": 4454 + }, + { + "epoch": 2.491748251748252, + "grad_norm": 0.50623780837851, + "learning_rate": 1.7032000837760377e-06, + "loss": 0.0874677449464798, + "step": 4455 + }, + { + "epoch": 2.4923076923076923, + "grad_norm": 0.5066544393207344, + "learning_rate": 1.6995686105259245e-06, + "loss": 0.05432897061109543, + "step": 4456 + }, + { + "epoch": 2.492867132867133, + "grad_norm": 0.5261968832386306, + "learning_rate": 1.6959406532509614e-06, + "loss": 0.15511249005794525, + "step": 4457 + }, + { + "epoch": 2.4934265734265733, + "grad_norm": 0.4615987766287119, + "learning_rate": 1.6923162134879045e-06, + "loss": 0.05809831619262695, + "step": 4458 + }, + { + "epoch": 2.493986013986014, + "grad_norm": 0.3526222804513751, + "learning_rate": 1.688695292772038e-06, + "loss": 0.03205804526805878, + "step": 4459 + }, + { + "epoch": 2.4945454545454546, + "grad_norm": 0.3797283247988356, + "learning_rate": 1.685077892637137e-06, + "loss": 0.02270149625837803, + "step": 4460 + }, + { + "epoch": 2.495104895104895, + "grad_norm": 0.5011763027445314, + "learning_rate": 1.6814640146154993e-06, + "loss": 0.037082523107528687, + "step": 4461 + }, + { + "epoch": 2.4956643356643355, + "grad_norm": 0.5370508788144749, + "learning_rate": 1.677853660237926e-06, + "loss": 0.20298242568969727, + "step": 4462 + }, + { + "epoch": 2.4962237762237764, + "grad_norm": 0.5083657556637048, + "learning_rate": 1.6742468310337178e-06, + "loss": 0.1126210168004036, + "step": 4463 + }, + { + "epoch": 2.496783216783217, + "grad_norm": 0.29206296406322935, + "learning_rate": 1.6706435285306988e-06, + "loss": 0.021989360451698303, + "step": 4464 + }, + { + "epoch": 2.4973426573426574, + "grad_norm": 0.6447573570385207, + "learning_rate": 1.6670437542551843e-06, + "loss": 0.10087059438228607, + "step": 4465 + }, + { + "epoch": 2.497902097902098, + "grad_norm": 0.8499330923316273, + "learning_rate": 1.6634475097320024e-06, + "loss": 0.2375773936510086, + "step": 4466 + }, + { + "epoch": 2.4984615384615383, + "grad_norm": 0.20238221866515294, + "learning_rate": 1.6598547964844869e-06, + "loss": 0.012536962516605854, + "step": 4467 + }, + { + "epoch": 2.499020979020979, + "grad_norm": 0.4586315931763552, + "learning_rate": 1.6562656160344648e-06, + "loss": 0.053984515368938446, + "step": 4468 + }, + { + "epoch": 2.4995804195804197, + "grad_norm": 0.6288914607465032, + "learning_rate": 1.6526799699022865e-06, + "loss": 0.14984266459941864, + "step": 4469 + }, + { + "epoch": 2.50013986013986, + "grad_norm": 0.39069250178311016, + "learning_rate": 1.6490978596067852e-06, + "loss": 0.06923025846481323, + "step": 4470 + }, + { + "epoch": 2.5006993006993006, + "grad_norm": 0.5969424779947693, + "learning_rate": 1.645519286665308e-06, + "loss": 0.15117859840393066, + "step": 4471 + }, + { + "epoch": 2.5012587412587415, + "grad_norm": 0.46437164992319413, + "learning_rate": 1.6419442525937035e-06, + "loss": 0.0656021311879158, + "step": 4472 + }, + { + "epoch": 2.501818181818182, + "grad_norm": 0.6860910814380935, + "learning_rate": 1.6383727589063103e-06, + "loss": 0.17433050274848938, + "step": 4473 + }, + { + "epoch": 2.5023776223776224, + "grad_norm": 0.43191227428700985, + "learning_rate": 1.634804807115985e-06, + "loss": 0.04447856545448303, + "step": 4474 + }, + { + "epoch": 2.502937062937063, + "grad_norm": 0.4709425848041101, + "learning_rate": 1.6312403987340686e-06, + "loss": 0.15348252654075623, + "step": 4475 + }, + { + "epoch": 2.5034965034965033, + "grad_norm": 0.6672262907287791, + "learning_rate": 1.6276795352704077e-06, + "loss": 0.22798123955726624, + "step": 4476 + }, + { + "epoch": 2.504055944055944, + "grad_norm": 0.3991653737523471, + "learning_rate": 1.6241222182333493e-06, + "loss": 0.06473525613546371, + "step": 4477 + }, + { + "epoch": 2.5046153846153847, + "grad_norm": 0.4108787588891321, + "learning_rate": 1.620568449129728e-06, + "loss": 0.07639111578464508, + "step": 4478 + }, + { + "epoch": 2.505174825174825, + "grad_norm": 0.03563478621845586, + "learning_rate": 1.617018229464894e-06, + "loss": 0.0011923732236027718, + "step": 4479 + }, + { + "epoch": 2.5057342657342656, + "grad_norm": 0.4417426784002046, + "learning_rate": 1.6134715607426732e-06, + "loss": 0.07638342678546906, + "step": 4480 + }, + { + "epoch": 2.5062937062937065, + "grad_norm": 0.3219006405469457, + "learning_rate": 1.6099284444654006e-06, + "loss": 0.024365486577153206, + "step": 4481 + }, + { + "epoch": 2.506853146853147, + "grad_norm": 0.4258699061431568, + "learning_rate": 1.606388882133907e-06, + "loss": 0.06167413666844368, + "step": 4482 + }, + { + "epoch": 2.5074125874125874, + "grad_norm": 0.4604376810573585, + "learning_rate": 1.602852875247506e-06, + "loss": 0.08631311357021332, + "step": 4483 + }, + { + "epoch": 2.507972027972028, + "grad_norm": 0.6303978410385384, + "learning_rate": 1.5993204253040162e-06, + "loss": 0.10294310748577118, + "step": 4484 + }, + { + "epoch": 2.5085314685314684, + "grad_norm": 0.32719326536615295, + "learning_rate": 1.5957915337997464e-06, + "loss": 0.07146032899618149, + "step": 4485 + }, + { + "epoch": 2.509090909090909, + "grad_norm": 0.49082100724508343, + "learning_rate": 1.5922662022294987e-06, + "loss": 0.08983171731233597, + "step": 4486 + }, + { + "epoch": 2.5096503496503497, + "grad_norm": 0.5319148581383204, + "learning_rate": 1.5887444320865676e-06, + "loss": 0.10053353756666183, + "step": 4487 + }, + { + "epoch": 2.51020979020979, + "grad_norm": 0.6308635738929826, + "learning_rate": 1.5852262248627326e-06, + "loss": 0.2222030758857727, + "step": 4488 + }, + { + "epoch": 2.5107692307692306, + "grad_norm": 0.2988400345832627, + "learning_rate": 1.581711582048272e-06, + "loss": 0.02491142228245735, + "step": 4489 + }, + { + "epoch": 2.5113286713286715, + "grad_norm": 0.5665688823020886, + "learning_rate": 1.5782005051319517e-06, + "loss": 0.24647054076194763, + "step": 4490 + }, + { + "epoch": 2.511888111888112, + "grad_norm": 0.7147761929852371, + "learning_rate": 1.574692995601027e-06, + "loss": 0.1390518695116043, + "step": 4491 + }, + { + "epoch": 2.5124475524475525, + "grad_norm": 0.4283074929727335, + "learning_rate": 1.5711890549412412e-06, + "loss": 0.04321552813053131, + "step": 4492 + }, + { + "epoch": 2.513006993006993, + "grad_norm": 0.6652149992474509, + "learning_rate": 1.56768868463683e-06, + "loss": 0.08706653863191605, + "step": 4493 + }, + { + "epoch": 2.5135664335664334, + "grad_norm": 0.5107451462272721, + "learning_rate": 1.5641918861705075e-06, + "loss": 0.044477079063653946, + "step": 4494 + }, + { + "epoch": 2.514125874125874, + "grad_norm": 0.5731986447244861, + "learning_rate": 1.5606986610234843e-06, + "loss": 0.09677645564079285, + "step": 4495 + }, + { + "epoch": 2.5146853146853148, + "grad_norm": 0.5509241317259178, + "learning_rate": 1.5572090106754511e-06, + "loss": 0.07995382696390152, + "step": 4496 + }, + { + "epoch": 2.515244755244755, + "grad_norm": 0.5301617085973899, + "learning_rate": 1.5537229366045903e-06, + "loss": 0.11393406242132187, + "step": 4497 + }, + { + "epoch": 2.5158041958041957, + "grad_norm": 0.3805223825782444, + "learning_rate": 1.550240440287566e-06, + "loss": 0.04904080927371979, + "step": 4498 + }, + { + "epoch": 2.5163636363636366, + "grad_norm": 0.3998745656882281, + "learning_rate": 1.5467615231995237e-06, + "loss": 0.03350922837853432, + "step": 4499 + }, + { + "epoch": 2.516923076923077, + "grad_norm": 0.44109055255852037, + "learning_rate": 1.543286186814097e-06, + "loss": 0.04065831005573273, + "step": 4500 + }, + { + "epoch": 2.5174825174825175, + "grad_norm": 0.29750165296873554, + "learning_rate": 1.5398144326034027e-06, + "loss": 0.021185962483286858, + "step": 4501 + }, + { + "epoch": 2.518041958041958, + "grad_norm": 0.3326159288984743, + "learning_rate": 1.5363462620380387e-06, + "loss": 0.022791506722569466, + "step": 4502 + }, + { + "epoch": 2.5186013986013984, + "grad_norm": 0.24140163776085838, + "learning_rate": 1.5328816765870891e-06, + "loss": 0.013553442433476448, + "step": 4503 + }, + { + "epoch": 2.5191608391608393, + "grad_norm": 0.35987218389710757, + "learning_rate": 1.5294206777181087e-06, + "loss": 0.04495007544755936, + "step": 4504 + }, + { + "epoch": 2.51972027972028, + "grad_norm": 0.23093560595038296, + "learning_rate": 1.5259632668971447e-06, + "loss": 0.014980919659137726, + "step": 4505 + }, + { + "epoch": 2.5202797202797202, + "grad_norm": 0.5209933799161336, + "learning_rate": 1.5225094455887202e-06, + "loss": 0.10430073738098145, + "step": 4506 + }, + { + "epoch": 2.5208391608391607, + "grad_norm": 0.6868520956620676, + "learning_rate": 1.5190592152558314e-06, + "loss": 0.2103475034236908, + "step": 4507 + }, + { + "epoch": 2.5213986013986016, + "grad_norm": 0.31054227500070986, + "learning_rate": 1.5156125773599694e-06, + "loss": 0.03322084993124008, + "step": 4508 + }, + { + "epoch": 2.521958041958042, + "grad_norm": 0.34280771208395483, + "learning_rate": 1.5121695333610864e-06, + "loss": 0.04260168969631195, + "step": 4509 + }, + { + "epoch": 2.5225174825174825, + "grad_norm": 0.6741164476231496, + "learning_rate": 1.5087300847176213e-06, + "loss": 0.1739736944437027, + "step": 4510 + }, + { + "epoch": 2.523076923076923, + "grad_norm": 0.5543584112143705, + "learning_rate": 1.5052942328864905e-06, + "loss": 0.16299282014369965, + "step": 4511 + }, + { + "epoch": 2.5236363636363635, + "grad_norm": 0.5193072351226925, + "learning_rate": 1.5018619793230783e-06, + "loss": 0.16750100255012512, + "step": 4512 + }, + { + "epoch": 2.5241958041958044, + "grad_norm": 0.4888147544721058, + "learning_rate": 1.49843332548126e-06, + "loss": 0.09584980458021164, + "step": 4513 + }, + { + "epoch": 2.524755244755245, + "grad_norm": 0.3681327303788436, + "learning_rate": 1.4950082728133698e-06, + "loss": 0.06112157553434372, + "step": 4514 + }, + { + "epoch": 2.5253146853146853, + "grad_norm": 0.24395045672508422, + "learning_rate": 1.4915868227702268e-06, + "loss": 0.0299077108502388, + "step": 4515 + }, + { + "epoch": 2.5258741258741257, + "grad_norm": 0.501725337178453, + "learning_rate": 1.4881689768011232e-06, + "loss": 0.14305603504180908, + "step": 4516 + }, + { + "epoch": 2.5264335664335666, + "grad_norm": 0.21456729716751588, + "learning_rate": 1.484754736353815e-06, + "loss": 0.021605102345347404, + "step": 4517 + }, + { + "epoch": 2.526993006993007, + "grad_norm": 0.2557936194843547, + "learning_rate": 1.481344102874549e-06, + "loss": 0.02200378105044365, + "step": 4518 + }, + { + "epoch": 2.5275524475524476, + "grad_norm": 0.4423484463181601, + "learning_rate": 1.477937077808026e-06, + "loss": 0.1374070793390274, + "step": 4519 + }, + { + "epoch": 2.528111888111888, + "grad_norm": 0.36279594244278796, + "learning_rate": 1.4745336625974282e-06, + "loss": 0.034848652780056, + "step": 4520 + }, + { + "epoch": 2.5286713286713285, + "grad_norm": 0.6119010881178053, + "learning_rate": 1.4711338586844081e-06, + "loss": 0.05923225358128548, + "step": 4521 + }, + { + "epoch": 2.5292307692307694, + "grad_norm": 0.40783914096634577, + "learning_rate": 1.4677376675090805e-06, + "loss": 0.06846512109041214, + "step": 4522 + }, + { + "epoch": 2.52979020979021, + "grad_norm": 0.5772228806210848, + "learning_rate": 1.4643450905100453e-06, + "loss": 0.15426114201545715, + "step": 4523 + }, + { + "epoch": 2.5303496503496503, + "grad_norm": 0.6438033619508343, + "learning_rate": 1.4609561291243547e-06, + "loss": 0.04837312549352646, + "step": 4524 + }, + { + "epoch": 2.5309090909090908, + "grad_norm": 0.39927862251500823, + "learning_rate": 1.4575707847875398e-06, + "loss": 0.0876699835062027, + "step": 4525 + }, + { + "epoch": 2.5314685314685317, + "grad_norm": 0.3558531080966621, + "learning_rate": 1.4541890589335973e-06, + "loss": 0.055250100791454315, + "step": 4526 + }, + { + "epoch": 2.532027972027972, + "grad_norm": 0.42135678287578154, + "learning_rate": 1.4508109529949898e-06, + "loss": 0.17377258837223053, + "step": 4527 + }, + { + "epoch": 2.5325874125874126, + "grad_norm": 0.506942556509911, + "learning_rate": 1.4474364684026499e-06, + "loss": 0.07808820903301239, + "step": 4528 + }, + { + "epoch": 2.533146853146853, + "grad_norm": 0.45514609368293407, + "learning_rate": 1.4440656065859694e-06, + "loss": 0.0693928599357605, + "step": 4529 + }, + { + "epoch": 2.5337062937062935, + "grad_norm": 0.5702477106257547, + "learning_rate": 1.4406983689728117e-06, + "loss": 0.14336681365966797, + "step": 4530 + }, + { + "epoch": 2.5342657342657344, + "grad_norm": 0.26736350794342834, + "learning_rate": 1.4373347569895025e-06, + "loss": 0.08404998481273651, + "step": 4531 + }, + { + "epoch": 2.534825174825175, + "grad_norm": 0.3825212807552491, + "learning_rate": 1.4339747720608355e-06, + "loss": 0.09068721532821655, + "step": 4532 + }, + { + "epoch": 2.5353846153846153, + "grad_norm": 0.35224040550430813, + "learning_rate": 1.43061841561006e-06, + "loss": 0.044285841286182404, + "step": 4533 + }, + { + "epoch": 2.535944055944056, + "grad_norm": 0.15283409947384521, + "learning_rate": 1.4272656890588966e-06, + "loss": 0.004923766478896141, + "step": 4534 + }, + { + "epoch": 2.5365034965034967, + "grad_norm": 0.2708106383649708, + "learning_rate": 1.4239165938275224e-06, + "loss": 0.019866883754730225, + "step": 4535 + }, + { + "epoch": 2.537062937062937, + "grad_norm": 0.523650189033846, + "learning_rate": 1.4205711313345816e-06, + "loss": 0.0585157610476017, + "step": 4536 + }, + { + "epoch": 2.5376223776223776, + "grad_norm": 0.307114444087108, + "learning_rate": 1.4172293029971763e-06, + "loss": 0.018409237265586853, + "step": 4537 + }, + { + "epoch": 2.538181818181818, + "grad_norm": 0.5966018093352803, + "learning_rate": 1.4138911102308672e-06, + "loss": 0.09511364996433258, + "step": 4538 + }, + { + "epoch": 2.5387412587412586, + "grad_norm": 0.4651401404573263, + "learning_rate": 1.4105565544496803e-06, + "loss": 0.16395479440689087, + "step": 4539 + }, + { + "epoch": 2.5393006993006995, + "grad_norm": 0.5143900960729543, + "learning_rate": 1.4072256370660952e-06, + "loss": 0.1138661801815033, + "step": 4540 + }, + { + "epoch": 2.53986013986014, + "grad_norm": 0.3717236909022628, + "learning_rate": 1.403898359491057e-06, + "loss": 0.043411001563072205, + "step": 4541 + }, + { + "epoch": 2.5404195804195804, + "grad_norm": 0.334814219040338, + "learning_rate": 1.4005747231339651e-06, + "loss": 0.06683079898357391, + "step": 4542 + }, + { + "epoch": 2.540979020979021, + "grad_norm": 0.3580459647754164, + "learning_rate": 1.3972547294026728e-06, + "loss": 0.07385344058275223, + "step": 4543 + }, + { + "epoch": 2.5415384615384617, + "grad_norm": 0.7144956495377458, + "learning_rate": 1.3939383797034978e-06, + "loss": 0.16887669265270233, + "step": 4544 + }, + { + "epoch": 2.542097902097902, + "grad_norm": 0.4946714911760177, + "learning_rate": 1.3906256754412084e-06, + "loss": 0.1192508190870285, + "step": 4545 + }, + { + "epoch": 2.5426573426573427, + "grad_norm": 0.4138126419420587, + "learning_rate": 1.3873166180190345e-06, + "loss": 0.13044364750385284, + "step": 4546 + }, + { + "epoch": 2.543216783216783, + "grad_norm": 0.5417572677531166, + "learning_rate": 1.384011208838657e-06, + "loss": 0.11944424360990524, + "step": 4547 + }, + { + "epoch": 2.5437762237762236, + "grad_norm": 0.4308284441379239, + "learning_rate": 1.3807094493002082e-06, + "loss": 0.05917758494615555, + "step": 4548 + }, + { + "epoch": 2.5443356643356645, + "grad_norm": 0.39830799560465363, + "learning_rate": 1.377411340802286e-06, + "loss": 0.09083046764135361, + "step": 4549 + }, + { + "epoch": 2.544895104895105, + "grad_norm": 0.41935761466932997, + "learning_rate": 1.3741168847419284e-06, + "loss": 0.05486498028039932, + "step": 4550 + }, + { + "epoch": 2.5454545454545454, + "grad_norm": 0.4243413830173183, + "learning_rate": 1.3708260825146335e-06, + "loss": 0.029036281630396843, + "step": 4551 + }, + { + "epoch": 2.546013986013986, + "grad_norm": 0.7958948187168551, + "learning_rate": 1.367538935514353e-06, + "loss": 0.11576436460018158, + "step": 4552 + }, + { + "epoch": 2.546573426573427, + "grad_norm": 0.5571840074256849, + "learning_rate": 1.364255445133481e-06, + "loss": 0.21533727645874023, + "step": 4553 + }, + { + "epoch": 2.5471328671328672, + "grad_norm": 0.30517290053623364, + "learning_rate": 1.3609756127628792e-06, + "loss": 0.07367813587188721, + "step": 4554 + }, + { + "epoch": 2.5476923076923077, + "grad_norm": 0.4063461854181082, + "learning_rate": 1.3576994397918453e-06, + "loss": 0.0598619319498539, + "step": 4555 + }, + { + "epoch": 2.548251748251748, + "grad_norm": 0.5779344125829571, + "learning_rate": 1.3544269276081257e-06, + "loss": 0.1191563829779625, + "step": 4556 + }, + { + "epoch": 2.5488111888111886, + "grad_norm": 0.5443206431041567, + "learning_rate": 1.3511580775979328e-06, + "loss": 0.0805964469909668, + "step": 4557 + }, + { + "epoch": 2.5493706293706295, + "grad_norm": 0.5288739067308555, + "learning_rate": 1.3478928911459088e-06, + "loss": 0.06912267208099365, + "step": 4558 + }, + { + "epoch": 2.54993006993007, + "grad_norm": 0.30679385542568594, + "learning_rate": 1.3446313696351542e-06, + "loss": 0.06492966413497925, + "step": 4559 + }, + { + "epoch": 2.5504895104895104, + "grad_norm": 0.4860721246507894, + "learning_rate": 1.3413735144472195e-06, + "loss": 0.05315789580345154, + "step": 4560 + }, + { + "epoch": 2.551048951048951, + "grad_norm": 0.33817513651203707, + "learning_rate": 1.3381193269620896e-06, + "loss": 0.07461623102426529, + "step": 4561 + }, + { + "epoch": 2.551608391608392, + "grad_norm": 0.5776159073008202, + "learning_rate": 1.334868808558213e-06, + "loss": 0.16163872182369232, + "step": 4562 + }, + { + "epoch": 2.5521678321678323, + "grad_norm": 0.2649632808065676, + "learning_rate": 1.331621960612468e-06, + "loss": 0.013561589643359184, + "step": 4563 + }, + { + "epoch": 2.5527272727272727, + "grad_norm": 0.46716021719920214, + "learning_rate": 1.3283787845001884e-06, + "loss": 0.08386319130659103, + "step": 4564 + }, + { + "epoch": 2.553286713286713, + "grad_norm": 0.5923628227617581, + "learning_rate": 1.3251392815951481e-06, + "loss": 0.12511901557445526, + "step": 4565 + }, + { + "epoch": 2.5538461538461537, + "grad_norm": 0.5583269235444146, + "learning_rate": 1.3219034532695663e-06, + "loss": 0.1285950243473053, + "step": 4566 + }, + { + "epoch": 2.5544055944055946, + "grad_norm": 0.16865202131857404, + "learning_rate": 1.3186713008941098e-06, + "loss": 0.008102228865027428, + "step": 4567 + }, + { + "epoch": 2.554965034965035, + "grad_norm": 0.4868514695134091, + "learning_rate": 1.3154428258378782e-06, + "loss": 0.058629631996154785, + "step": 4568 + }, + { + "epoch": 2.5555244755244755, + "grad_norm": 0.38205793492945356, + "learning_rate": 1.3122180294684217e-06, + "loss": 0.042920537292957306, + "step": 4569 + }, + { + "epoch": 2.556083916083916, + "grad_norm": 0.3998709459967448, + "learning_rate": 1.3089969131517311e-06, + "loss": 0.09765076637268066, + "step": 4570 + }, + { + "epoch": 2.556643356643357, + "grad_norm": 0.4623393510818732, + "learning_rate": 1.305779478252236e-06, + "loss": 0.10050079971551895, + "step": 4571 + }, + { + "epoch": 2.5572027972027973, + "grad_norm": 0.5079592832752892, + "learning_rate": 1.3025657261328105e-06, + "loss": 0.17783184349536896, + "step": 4572 + }, + { + "epoch": 2.5577622377622378, + "grad_norm": 0.26452242660240666, + "learning_rate": 1.2993556581547627e-06, + "loss": 0.0397367961704731, + "step": 4573 + }, + { + "epoch": 2.5583216783216782, + "grad_norm": 0.5257140432936166, + "learning_rate": 1.2961492756778437e-06, + "loss": 0.18189845979213715, + "step": 4574 + }, + { + "epoch": 2.5588811188811187, + "grad_norm": 0.3205040448269561, + "learning_rate": 1.2929465800602458e-06, + "loss": 0.11143840849399567, + "step": 4575 + }, + { + "epoch": 2.5594405594405596, + "grad_norm": 0.5808362390264443, + "learning_rate": 1.2897475726585951e-06, + "loss": 0.10607460141181946, + "step": 4576 + }, + { + "epoch": 2.56, + "grad_norm": 0.6686242482211916, + "learning_rate": 1.2865522548279596e-06, + "loss": 0.21376782655715942, + "step": 4577 + }, + { + "epoch": 2.5605594405594405, + "grad_norm": 0.4663053713686167, + "learning_rate": 1.2833606279218392e-06, + "loss": 0.04318004474043846, + "step": 4578 + }, + { + "epoch": 2.561118881118881, + "grad_norm": 0.6210803052878044, + "learning_rate": 1.2801726932921743e-06, + "loss": 0.09065401554107666, + "step": 4579 + }, + { + "epoch": 2.561678321678322, + "grad_norm": 0.4286006979533483, + "learning_rate": 1.2769884522893406e-06, + "loss": 0.06432337313890457, + "step": 4580 + }, + { + "epoch": 2.5622377622377623, + "grad_norm": 0.31518340081378826, + "learning_rate": 1.2738079062621523e-06, + "loss": 0.0363226979970932, + "step": 4581 + }, + { + "epoch": 2.562797202797203, + "grad_norm": 0.34594604284417924, + "learning_rate": 1.2706310565578495e-06, + "loss": 0.08044242858886719, + "step": 4582 + }, + { + "epoch": 2.5633566433566433, + "grad_norm": 0.5263498514678528, + "learning_rate": 1.2674579045221146e-06, + "loss": 0.06646215915679932, + "step": 4583 + }, + { + "epoch": 2.5639160839160837, + "grad_norm": 0.5218679002947159, + "learning_rate": 1.2642884514990617e-06, + "loss": 0.09292811155319214, + "step": 4584 + }, + { + "epoch": 2.5644755244755246, + "grad_norm": 0.42826954820389324, + "learning_rate": 1.2611226988312365e-06, + "loss": 0.03904930502176285, + "step": 4585 + }, + { + "epoch": 2.565034965034965, + "grad_norm": 0.4193783044704173, + "learning_rate": 1.257960647859623e-06, + "loss": 0.0455571785569191, + "step": 4586 + }, + { + "epoch": 2.5655944055944055, + "grad_norm": 0.5070888485823991, + "learning_rate": 1.2548022999236242e-06, + "loss": 0.11863440275192261, + "step": 4587 + }, + { + "epoch": 2.566153846153846, + "grad_norm": 0.47573880650828776, + "learning_rate": 1.2516476563610913e-06, + "loss": 0.09104831516742706, + "step": 4588 + }, + { + "epoch": 2.566713286713287, + "grad_norm": 0.5426160969805378, + "learning_rate": 1.248496718508293e-06, + "loss": 0.10867291688919067, + "step": 4589 + }, + { + "epoch": 2.5672727272727274, + "grad_norm": 0.5225443182021545, + "learning_rate": 1.245349487699935e-06, + "loss": 0.09799507260322571, + "step": 4590 + }, + { + "epoch": 2.567832167832168, + "grad_norm": 0.3195141506744252, + "learning_rate": 1.242205965269152e-06, + "loss": 0.08061123639345169, + "step": 4591 + }, + { + "epoch": 2.5683916083916083, + "grad_norm": 0.4943878490252836, + "learning_rate": 1.2390661525475024e-06, + "loss": 0.0685778334736824, + "step": 4592 + }, + { + "epoch": 2.5689510489510488, + "grad_norm": 0.32895741034227044, + "learning_rate": 1.2359300508649852e-06, + "loss": 0.03397190943360329, + "step": 4593 + }, + { + "epoch": 2.5695104895104897, + "grad_norm": 0.42379331105245027, + "learning_rate": 1.2327976615500136e-06, + "loss": 0.0748581662774086, + "step": 4594 + }, + { + "epoch": 2.57006993006993, + "grad_norm": 0.3579059435876396, + "learning_rate": 1.2296689859294375e-06, + "loss": 0.08978810906410217, + "step": 4595 + }, + { + "epoch": 2.5706293706293706, + "grad_norm": 0.44003419591524223, + "learning_rate": 1.226544025328532e-06, + "loss": 0.1846337616443634, + "step": 4596 + }, + { + "epoch": 2.571188811188811, + "grad_norm": 0.5213592766714341, + "learning_rate": 1.223422781070992e-06, + "loss": 0.17398908734321594, + "step": 4597 + }, + { + "epoch": 2.571748251748252, + "grad_norm": 0.5061556924103352, + "learning_rate": 1.2203052544789518e-06, + "loss": 0.09944969415664673, + "step": 4598 + }, + { + "epoch": 2.5723076923076924, + "grad_norm": 0.4640876185578008, + "learning_rate": 1.2171914468729552e-06, + "loss": 0.059454552829265594, + "step": 4599 + }, + { + "epoch": 2.572867132867133, + "grad_norm": 0.4262652294614994, + "learning_rate": 1.2140813595719825e-06, + "loss": 0.12345708906650543, + "step": 4600 + }, + { + "epoch": 2.5734265734265733, + "grad_norm": 0.5475322029111278, + "learning_rate": 1.2109749938934346e-06, + "loss": 0.06571106612682343, + "step": 4601 + }, + { + "epoch": 2.573986013986014, + "grad_norm": 0.5705999062209591, + "learning_rate": 1.2078723511531288e-06, + "loss": 0.17944902181625366, + "step": 4602 + }, + { + "epoch": 2.5745454545454547, + "grad_norm": 0.4638859614413516, + "learning_rate": 1.2047734326653205e-06, + "loss": 0.06039018556475639, + "step": 4603 + }, + { + "epoch": 2.575104895104895, + "grad_norm": 0.5324028695025897, + "learning_rate": 1.201678239742673e-06, + "loss": 0.15344257652759552, + "step": 4604 + }, + { + "epoch": 2.5756643356643356, + "grad_norm": 0.4938935903799455, + "learning_rate": 1.1985867736962798e-06, + "loss": 0.1656726747751236, + "step": 4605 + }, + { + "epoch": 2.576223776223776, + "grad_norm": 0.3763979079133022, + "learning_rate": 1.1954990358356533e-06, + "loss": 0.05471505597233772, + "step": 4606 + }, + { + "epoch": 2.576783216783217, + "grad_norm": 0.21721774331497704, + "learning_rate": 1.1924150274687252e-06, + "loss": 0.008323697373270988, + "step": 4607 + }, + { + "epoch": 2.5773426573426574, + "grad_norm": 0.30921983885859866, + "learning_rate": 1.1893347499018493e-06, + "loss": 0.055867210030555725, + "step": 4608 + }, + { + "epoch": 2.577902097902098, + "grad_norm": 0.3273550859849165, + "learning_rate": 1.1862582044397985e-06, + "loss": 0.04744353145360947, + "step": 4609 + }, + { + "epoch": 2.5784615384615384, + "grad_norm": 0.3613321889870487, + "learning_rate": 1.183185392385766e-06, + "loss": 0.03727944940328598, + "step": 4610 + }, + { + "epoch": 2.579020979020979, + "grad_norm": 0.4784142919566731, + "learning_rate": 1.1801163150413652e-06, + "loss": 0.06568647921085358, + "step": 4611 + }, + { + "epoch": 2.5795804195804197, + "grad_norm": 0.7853012591479246, + "learning_rate": 1.1770509737066193e-06, + "loss": 0.056148070842027664, + "step": 4612 + }, + { + "epoch": 2.58013986013986, + "grad_norm": 0.5573766784444103, + "learning_rate": 1.1739893696799764e-06, + "loss": 0.08955402672290802, + "step": 4613 + }, + { + "epoch": 2.5806993006993006, + "grad_norm": 0.4806295366957624, + "learning_rate": 1.1709315042583002e-06, + "loss": 0.08713246881961823, + "step": 4614 + }, + { + "epoch": 2.581258741258741, + "grad_norm": 0.49481808313054737, + "learning_rate": 1.1678773787368714e-06, + "loss": 0.1412111520767212, + "step": 4615 + }, + { + "epoch": 2.581818181818182, + "grad_norm": 0.45349789611843666, + "learning_rate": 1.164826994409386e-06, + "loss": 0.16250455379486084, + "step": 4616 + }, + { + "epoch": 2.5823776223776225, + "grad_norm": 0.4503859114489756, + "learning_rate": 1.1617803525679505e-06, + "loss": 0.13743548095226288, + "step": 4617 + }, + { + "epoch": 2.582937062937063, + "grad_norm": 0.8600720618641053, + "learning_rate": 1.1587374545030926e-06, + "loss": 0.08454421162605286, + "step": 4618 + }, + { + "epoch": 2.5834965034965034, + "grad_norm": 0.28841076691014506, + "learning_rate": 1.1556983015037515e-06, + "loss": 0.028263267129659653, + "step": 4619 + }, + { + "epoch": 2.584055944055944, + "grad_norm": 0.4618992554559034, + "learning_rate": 1.1526628948572815e-06, + "loss": 0.0640464574098587, + "step": 4620 + }, + { + "epoch": 2.5846153846153848, + "grad_norm": 0.521173197834153, + "learning_rate": 1.1496312358494477e-06, + "loss": 0.09115159511566162, + "step": 4621 + }, + { + "epoch": 2.585174825174825, + "grad_norm": 0.42442869480596307, + "learning_rate": 1.1466033257644316e-06, + "loss": 0.08466363698244095, + "step": 4622 + }, + { + "epoch": 2.5857342657342657, + "grad_norm": 0.5633783705462175, + "learning_rate": 1.1435791658848194e-06, + "loss": 0.08374187350273132, + "step": 4623 + }, + { + "epoch": 2.586293706293706, + "grad_norm": 0.4205191188860606, + "learning_rate": 1.1405587574916145e-06, + "loss": 0.04049751162528992, + "step": 4624 + }, + { + "epoch": 2.586853146853147, + "grad_norm": 0.26245288807429895, + "learning_rate": 1.1375421018642318e-06, + "loss": 0.07686944305896759, + "step": 4625 + }, + { + "epoch": 2.5874125874125875, + "grad_norm": 0.5685746940965678, + "learning_rate": 1.1345292002804943e-06, + "loss": 0.12538719177246094, + "step": 4626 + }, + { + "epoch": 2.587972027972028, + "grad_norm": 0.49947014220483094, + "learning_rate": 1.131520054016636e-06, + "loss": 0.06467457860708237, + "step": 4627 + }, + { + "epoch": 2.5885314685314684, + "grad_norm": 0.5458676562163716, + "learning_rate": 1.1285146643472978e-06, + "loss": 0.12892964482307434, + "step": 4628 + }, + { + "epoch": 2.589090909090909, + "grad_norm": 0.3267781723729519, + "learning_rate": 1.125513032545531e-06, + "loss": 0.033108122646808624, + "step": 4629 + }, + { + "epoch": 2.58965034965035, + "grad_norm": 0.43592095901239014, + "learning_rate": 1.1225151598827977e-06, + "loss": 0.09453903138637543, + "step": 4630 + }, + { + "epoch": 2.5902097902097903, + "grad_norm": 0.4652988277269842, + "learning_rate": 1.1195210476289597e-06, + "loss": 0.10552264004945755, + "step": 4631 + }, + { + "epoch": 2.5907692307692307, + "grad_norm": 0.45783890394104937, + "learning_rate": 1.1165306970522982e-06, + "loss": 0.12797769904136658, + "step": 4632 + }, + { + "epoch": 2.591328671328671, + "grad_norm": 0.3303735813263733, + "learning_rate": 1.11354410941949e-06, + "loss": 0.0609578937292099, + "step": 4633 + }, + { + "epoch": 2.591888111888112, + "grad_norm": 0.37112305414397523, + "learning_rate": 1.1105612859956227e-06, + "loss": 0.10750634223222733, + "step": 4634 + }, + { + "epoch": 2.5924475524475525, + "grad_norm": 0.525647260420354, + "learning_rate": 1.1075822280441907e-06, + "loss": 0.08289151638746262, + "step": 4635 + }, + { + "epoch": 2.593006993006993, + "grad_norm": 0.5477755860317192, + "learning_rate": 1.1046069368270861e-06, + "loss": 0.08348525315523148, + "step": 4636 + }, + { + "epoch": 2.5935664335664335, + "grad_norm": 0.13451597622436873, + "learning_rate": 1.1016354136046193e-06, + "loss": 0.006321026012301445, + "step": 4637 + }, + { + "epoch": 2.594125874125874, + "grad_norm": 0.3084649206959315, + "learning_rate": 1.0986676596354884e-06, + "loss": 0.03064294531941414, + "step": 4638 + }, + { + "epoch": 2.594685314685315, + "grad_norm": 0.5004473533595007, + "learning_rate": 1.095703676176807e-06, + "loss": 0.14625415205955505, + "step": 4639 + }, + { + "epoch": 2.5952447552447553, + "grad_norm": 0.2963198012403708, + "learning_rate": 1.0927434644840885e-06, + "loss": 0.04113004729151726, + "step": 4640 + }, + { + "epoch": 2.5958041958041957, + "grad_norm": 0.33276015124939606, + "learning_rate": 1.0897870258112396e-06, + "loss": 0.05982401221990585, + "step": 4641 + }, + { + "epoch": 2.596363636363636, + "grad_norm": 0.2511998641823905, + "learning_rate": 1.0868343614105869e-06, + "loss": 0.0202775988727808, + "step": 4642 + }, + { + "epoch": 2.596923076923077, + "grad_norm": 0.7424419642783474, + "learning_rate": 1.0838854725328396e-06, + "loss": 0.10561646521091461, + "step": 4643 + }, + { + "epoch": 2.5974825174825176, + "grad_norm": 0.3921334377917885, + "learning_rate": 1.0809403604271208e-06, + "loss": 0.04724736139178276, + "step": 4644 + }, + { + "epoch": 2.598041958041958, + "grad_norm": 0.4789851877962997, + "learning_rate": 1.0779990263409479e-06, + "loss": 0.10748960822820663, + "step": 4645 + }, + { + "epoch": 2.5986013986013985, + "grad_norm": 0.5650681672753862, + "learning_rate": 1.0750614715202345e-06, + "loss": 0.10508938133716583, + "step": 4646 + }, + { + "epoch": 2.599160839160839, + "grad_norm": 0.26855563849249753, + "learning_rate": 1.0721276972093064e-06, + "loss": 0.04325022175908089, + "step": 4647 + }, + { + "epoch": 2.59972027972028, + "grad_norm": 0.5778236897407188, + "learning_rate": 1.0691977046508718e-06, + "loss": 0.07308591157197952, + "step": 4648 + }, + { + "epoch": 2.6002797202797203, + "grad_norm": 0.3668842675036552, + "learning_rate": 1.0662714950860487e-06, + "loss": 0.036036331206560135, + "step": 4649 + }, + { + "epoch": 2.600839160839161, + "grad_norm": 0.33794532927245613, + "learning_rate": 1.0633490697543481e-06, + "loss": 0.03203625977039337, + "step": 4650 + }, + { + "epoch": 2.6013986013986012, + "grad_norm": 0.27836989029820564, + "learning_rate": 1.060430429893674e-06, + "loss": 0.015751425176858902, + "step": 4651 + }, + { + "epoch": 2.601958041958042, + "grad_norm": 0.9400812863582513, + "learning_rate": 1.05751557674034e-06, + "loss": 0.24254080653190613, + "step": 4652 + }, + { + "epoch": 2.6025174825174826, + "grad_norm": 0.37898925275531303, + "learning_rate": 1.0546045115290405e-06, + "loss": 0.03760292008519173, + "step": 4653 + }, + { + "epoch": 2.603076923076923, + "grad_norm": 0.31855744978593037, + "learning_rate": 1.0516972354928735e-06, + "loss": 0.029210545122623444, + "step": 4654 + }, + { + "epoch": 2.6036363636363635, + "grad_norm": 0.21842456009088282, + "learning_rate": 1.0487937498633327e-06, + "loss": 0.010981056839227676, + "step": 4655 + }, + { + "epoch": 2.604195804195804, + "grad_norm": 0.4015952002871012, + "learning_rate": 1.0458940558703013e-06, + "loss": 0.05680252984166145, + "step": 4656 + }, + { + "epoch": 2.604755244755245, + "grad_norm": 0.6591201520271375, + "learning_rate": 1.0429981547420588e-06, + "loss": 0.058628883212804794, + "step": 4657 + }, + { + "epoch": 2.6053146853146854, + "grad_norm": 0.6597080605516239, + "learning_rate": 1.0401060477052793e-06, + "loss": 0.13952630758285522, + "step": 4658 + }, + { + "epoch": 2.605874125874126, + "grad_norm": 0.6171788196121009, + "learning_rate": 1.03721773598503e-06, + "loss": 0.14513689279556274, + "step": 4659 + }, + { + "epoch": 2.6064335664335663, + "grad_norm": 0.4611243075008831, + "learning_rate": 1.0343332208047675e-06, + "loss": 0.165986105799675, + "step": 4660 + }, + { + "epoch": 2.606993006993007, + "grad_norm": 0.35416729139242337, + "learning_rate": 1.0314525033863455e-06, + "loss": 0.0619468167424202, + "step": 4661 + }, + { + "epoch": 2.6075524475524476, + "grad_norm": 0.25709249372354237, + "learning_rate": 1.0285755849500001e-06, + "loss": 0.02004910632967949, + "step": 4662 + }, + { + "epoch": 2.608111888111888, + "grad_norm": 0.3218273370677025, + "learning_rate": 1.0257024667143667e-06, + "loss": 0.026728782802820206, + "step": 4663 + }, + { + "epoch": 2.6086713286713286, + "grad_norm": 0.44657834891348985, + "learning_rate": 1.0228331498964671e-06, + "loss": 0.08980909734964371, + "step": 4664 + }, + { + "epoch": 2.609230769230769, + "grad_norm": 0.40519497257399045, + "learning_rate": 1.019967635711714e-06, + "loss": 0.0559174083173275, + "step": 4665 + }, + { + "epoch": 2.60979020979021, + "grad_norm": 0.43643889548511794, + "learning_rate": 1.0171059253739112e-06, + "loss": 0.05049540102481842, + "step": 4666 + }, + { + "epoch": 2.6103496503496504, + "grad_norm": 0.5846064941309781, + "learning_rate": 1.0142480200952453e-06, + "loss": 0.18063391745090485, + "step": 4667 + }, + { + "epoch": 2.610909090909091, + "grad_norm": 0.4350409906383758, + "learning_rate": 1.0113939210862955e-06, + "loss": 0.046123601496219635, + "step": 4668 + }, + { + "epoch": 2.6114685314685313, + "grad_norm": 0.642442572742759, + "learning_rate": 1.008543629556028e-06, + "loss": 0.12716144323349, + "step": 4669 + }, + { + "epoch": 2.612027972027972, + "grad_norm": 0.7074749192551028, + "learning_rate": 1.0056971467117982e-06, + "loss": 0.1518527865409851, + "step": 4670 + }, + { + "epoch": 2.6125874125874127, + "grad_norm": 0.36425257752219997, + "learning_rate": 1.0028544737593471e-06, + "loss": 0.10051946341991425, + "step": 4671 + }, + { + "epoch": 2.613146853146853, + "grad_norm": 0.5244584241951495, + "learning_rate": 1.000015611902796e-06, + "loss": 0.08212938159704208, + "step": 4672 + }, + { + "epoch": 2.6137062937062936, + "grad_norm": 0.36038581489839555, + "learning_rate": 9.971805623446596e-07, + "loss": 0.04050418362021446, + "step": 4673 + }, + { + "epoch": 2.614265734265734, + "grad_norm": 0.504207907868682, + "learning_rate": 9.94349326285836e-07, + "loss": 0.07946804910898209, + "step": 4674 + }, + { + "epoch": 2.614825174825175, + "grad_norm": 0.44701335513650176, + "learning_rate": 9.915219049256052e-07, + "loss": 0.10330444574356079, + "step": 4675 + }, + { + "epoch": 2.6153846153846154, + "grad_norm": 0.5060435516033005, + "learning_rate": 9.886982994616357e-07, + "loss": 0.134657084941864, + "step": 4676 + }, + { + "epoch": 2.615944055944056, + "grad_norm": 0.5169713394560126, + "learning_rate": 9.858785110899717e-07, + "loss": 0.01655227690935135, + "step": 4677 + }, + { + "epoch": 2.6165034965034963, + "grad_norm": 0.5269762615147818, + "learning_rate": 9.83062541005053e-07, + "loss": 0.1313818097114563, + "step": 4678 + }, + { + "epoch": 2.6170629370629372, + "grad_norm": 0.4893393812591224, + "learning_rate": 9.802503903996907e-07, + "loss": 0.12005007266998291, + "step": 4679 + }, + { + "epoch": 2.6176223776223777, + "grad_norm": 0.334046453221992, + "learning_rate": 9.774420604650792e-07, + "loss": 0.040859755128622055, + "step": 4680 + }, + { + "epoch": 2.618181818181818, + "grad_norm": 0.46892521240186114, + "learning_rate": 9.746375523908036e-07, + "loss": 0.09826002269983292, + "step": 4681 + }, + { + "epoch": 2.6187412587412586, + "grad_norm": 0.39469627651563044, + "learning_rate": 9.718368673648194e-07, + "loss": 0.04227941110730171, + "step": 4682 + }, + { + "epoch": 2.619300699300699, + "grad_norm": 0.46837227498690376, + "learning_rate": 9.690400065734695e-07, + "loss": 0.18297724425792694, + "step": 4683 + }, + { + "epoch": 2.61986013986014, + "grad_norm": 0.6955021990445678, + "learning_rate": 9.662469712014744e-07, + "loss": 0.23154008388519287, + "step": 4684 + }, + { + "epoch": 2.6204195804195805, + "grad_norm": 0.33217720226049524, + "learning_rate": 9.634577624319318e-07, + "loss": 0.036578088998794556, + "step": 4685 + }, + { + "epoch": 2.620979020979021, + "grad_norm": 0.46368482027869495, + "learning_rate": 9.606723814463258e-07, + "loss": 0.06452926248311996, + "step": 4686 + }, + { + "epoch": 2.6215384615384614, + "grad_norm": 0.3946535884804849, + "learning_rate": 9.578908294245103e-07, + "loss": 0.06394671648740768, + "step": 4687 + }, + { + "epoch": 2.6220979020979023, + "grad_norm": 0.5528772141070742, + "learning_rate": 9.551131075447228e-07, + "loss": 0.12710806727409363, + "step": 4688 + }, + { + "epoch": 2.6226573426573427, + "grad_norm": 0.3474302904683238, + "learning_rate": 9.523392169835799e-07, + "loss": 0.035390134900808334, + "step": 4689 + }, + { + "epoch": 2.623216783216783, + "grad_norm": 4.289777840300856, + "learning_rate": 9.49569158916065e-07, + "loss": 0.19761255383491516, + "step": 4690 + }, + { + "epoch": 2.6237762237762237, + "grad_norm": 0.4423865024804312, + "learning_rate": 9.468029345155527e-07, + "loss": 0.06719936430454254, + "step": 4691 + }, + { + "epoch": 2.624335664335664, + "grad_norm": 0.24670684648924018, + "learning_rate": 9.440405449537838e-07, + "loss": 0.03917856886982918, + "step": 4692 + }, + { + "epoch": 2.624895104895105, + "grad_norm": 0.3900047718361682, + "learning_rate": 9.412819914008764e-07, + "loss": 0.04212659224867821, + "step": 4693 + }, + { + "epoch": 2.6254545454545455, + "grad_norm": 0.3818189361156417, + "learning_rate": 9.385272750253272e-07, + "loss": 0.08476566523313522, + "step": 4694 + }, + { + "epoch": 2.626013986013986, + "grad_norm": 0.7878926828169373, + "learning_rate": 9.357763969940004e-07, + "loss": 0.16443303227424622, + "step": 4695 + }, + { + "epoch": 2.626573426573427, + "grad_norm": 0.435811184912996, + "learning_rate": 9.33029358472145e-07, + "loss": 0.07624956965446472, + "step": 4696 + }, + { + "epoch": 2.6271328671328673, + "grad_norm": 0.40494877680314667, + "learning_rate": 9.302861606233738e-07, + "loss": 0.06449459493160248, + "step": 4697 + }, + { + "epoch": 2.6276923076923078, + "grad_norm": 0.6485706502897199, + "learning_rate": 9.275468046096759e-07, + "loss": 0.12648284435272217, + "step": 4698 + }, + { + "epoch": 2.6282517482517482, + "grad_norm": 0.5337077108171168, + "learning_rate": 9.248112915914153e-07, + "loss": 0.15524229407310486, + "step": 4699 + }, + { + "epoch": 2.6288111888111887, + "grad_norm": 0.547121032317351, + "learning_rate": 9.220796227273254e-07, + "loss": 0.0979311466217041, + "step": 4700 + }, + { + "epoch": 2.629370629370629, + "grad_norm": 0.5609469625863005, + "learning_rate": 9.19351799174516e-07, + "loss": 0.11986128985881805, + "step": 4701 + }, + { + "epoch": 2.62993006993007, + "grad_norm": 0.5659326264140484, + "learning_rate": 9.166278220884584e-07, + "loss": 0.11027545481920242, + "step": 4702 + }, + { + "epoch": 2.6304895104895105, + "grad_norm": 0.3245219720401446, + "learning_rate": 9.13907692623004e-07, + "loss": 0.027227971702814102, + "step": 4703 + }, + { + "epoch": 2.631048951048951, + "grad_norm": 0.6029673009655989, + "learning_rate": 9.1119141193037e-07, + "loss": 0.1036660224199295, + "step": 4704 + }, + { + "epoch": 2.631608391608392, + "grad_norm": 0.31444714257322137, + "learning_rate": 9.084789811611461e-07, + "loss": 0.12865287065505981, + "step": 4705 + }, + { + "epoch": 2.6321678321678323, + "grad_norm": 0.5140777382348938, + "learning_rate": 9.057704014642866e-07, + "loss": 0.1411428302526474, + "step": 4706 + }, + { + "epoch": 2.632727272727273, + "grad_norm": 0.3045475733039163, + "learning_rate": 9.03065673987118e-07, + "loss": 0.014351305551826954, + "step": 4707 + }, + { + "epoch": 2.6332867132867133, + "grad_norm": 0.43946525220281785, + "learning_rate": 9.00364799875334e-07, + "loss": 0.09398384392261505, + "step": 4708 + }, + { + "epoch": 2.6338461538461537, + "grad_norm": 0.40187506234794135, + "learning_rate": 8.976677802729983e-07, + "loss": 0.0376257486641407, + "step": 4709 + }, + { + "epoch": 2.634405594405594, + "grad_norm": 0.22374740011635402, + "learning_rate": 8.949746163225403e-07, + "loss": 0.021047677844762802, + "step": 4710 + }, + { + "epoch": 2.634965034965035, + "grad_norm": 0.3831991239929218, + "learning_rate": 8.922853091647532e-07, + "loss": 0.106009840965271, + "step": 4711 + }, + { + "epoch": 2.6355244755244756, + "grad_norm": 0.43976748471008065, + "learning_rate": 8.895998599388e-07, + "loss": 0.11076105386018753, + "step": 4712 + }, + { + "epoch": 2.636083916083916, + "grad_norm": 0.5449433941489812, + "learning_rate": 8.869182697822087e-07, + "loss": 0.06869295239448547, + "step": 4713 + }, + { + "epoch": 2.636643356643357, + "grad_norm": 0.2832395061051125, + "learning_rate": 8.842405398308729e-07, + "loss": 0.06974361091852188, + "step": 4714 + }, + { + "epoch": 2.6372027972027974, + "grad_norm": 0.03699674882471138, + "learning_rate": 8.815666712190518e-07, + "loss": 0.0003780899860430509, + "step": 4715 + }, + { + "epoch": 2.637762237762238, + "grad_norm": 0.5115732804963549, + "learning_rate": 8.78896665079364e-07, + "loss": 0.1295163333415985, + "step": 4716 + }, + { + "epoch": 2.6383216783216783, + "grad_norm": 0.427514456992195, + "learning_rate": 8.762305225428025e-07, + "loss": 0.04817381128668785, + "step": 4717 + }, + { + "epoch": 2.6388811188811188, + "grad_norm": 0.3545051702788007, + "learning_rate": 8.735682447387117e-07, + "loss": 0.03916630893945694, + "step": 4718 + }, + { + "epoch": 2.639440559440559, + "grad_norm": 0.2429691070309613, + "learning_rate": 8.709098327948051e-07, + "loss": 0.020356670022010803, + "step": 4719 + }, + { + "epoch": 2.64, + "grad_norm": 0.45624953190268497, + "learning_rate": 8.68255287837162e-07, + "loss": 0.07135714590549469, + "step": 4720 + }, + { + "epoch": 2.6405594405594406, + "grad_norm": 0.23868671102245484, + "learning_rate": 8.656046109902116e-07, + "loss": 0.032696399837732315, + "step": 4721 + }, + { + "epoch": 2.641118881118881, + "grad_norm": 0.37076035011889913, + "learning_rate": 8.629578033767616e-07, + "loss": 0.04050498455762863, + "step": 4722 + }, + { + "epoch": 2.641678321678322, + "grad_norm": 0.17429877991877202, + "learning_rate": 8.603148661179672e-07, + "loss": 0.010556847788393497, + "step": 4723 + }, + { + "epoch": 2.6422377622377624, + "grad_norm": 0.4200374430407053, + "learning_rate": 8.576758003333474e-07, + "loss": 0.12463138997554779, + "step": 4724 + }, + { + "epoch": 2.642797202797203, + "grad_norm": 0.2595389418503427, + "learning_rate": 8.55040607140788e-07, + "loss": 0.025165880098938942, + "step": 4725 + }, + { + "epoch": 2.6433566433566433, + "grad_norm": 0.3463949828448519, + "learning_rate": 8.524092876565204e-07, + "loss": 0.069175586104393, + "step": 4726 + }, + { + "epoch": 2.643916083916084, + "grad_norm": 0.5256863430543846, + "learning_rate": 8.497818429951532e-07, + "loss": 0.06919945776462555, + "step": 4727 + }, + { + "epoch": 2.6444755244755243, + "grad_norm": 0.2492137326248089, + "learning_rate": 8.471582742696405e-07, + "loss": 0.021942205727100372, + "step": 4728 + }, + { + "epoch": 2.645034965034965, + "grad_norm": 0.5169927298527381, + "learning_rate": 8.445385825912933e-07, + "loss": 0.10473151504993439, + "step": 4729 + }, + { + "epoch": 2.6455944055944056, + "grad_norm": 0.3707249774328976, + "learning_rate": 8.41922769069794e-07, + "loss": 0.04877055436372757, + "step": 4730 + }, + { + "epoch": 2.646153846153846, + "grad_norm": 0.5399072384671407, + "learning_rate": 8.393108348131674e-07, + "loss": 0.21006245911121368, + "step": 4731 + }, + { + "epoch": 2.646713286713287, + "grad_norm": 0.4700775899820987, + "learning_rate": 8.367027809278039e-07, + "loss": 0.06726476550102234, + "step": 4732 + }, + { + "epoch": 2.6472727272727274, + "grad_norm": 0.4474922430863471, + "learning_rate": 8.340986085184477e-07, + "loss": 0.09445297718048096, + "step": 4733 + }, + { + "epoch": 2.647832167832168, + "grad_norm": 0.4600294061242749, + "learning_rate": 8.314983186881954e-07, + "loss": 0.07439548522233963, + "step": 4734 + }, + { + "epoch": 2.6483916083916084, + "grad_norm": 0.5616974308959732, + "learning_rate": 8.289019125385089e-07, + "loss": 0.06531370431184769, + "step": 4735 + }, + { + "epoch": 2.648951048951049, + "grad_norm": 0.3958776948489961, + "learning_rate": 8.263093911691922e-07, + "loss": 0.04804437607526779, + "step": 4736 + }, + { + "epoch": 2.6495104895104893, + "grad_norm": 0.5471683207447298, + "learning_rate": 8.237207556784121e-07, + "loss": 0.16451652348041534, + "step": 4737 + }, + { + "epoch": 2.65006993006993, + "grad_norm": 0.44768321777197695, + "learning_rate": 8.211360071626873e-07, + "loss": 0.06156668812036514, + "step": 4738 + }, + { + "epoch": 2.6506293706293707, + "grad_norm": 0.5634880714964987, + "learning_rate": 8.185551467168906e-07, + "loss": 0.14925451576709747, + "step": 4739 + }, + { + "epoch": 2.651188811188811, + "grad_norm": 0.3253025577920631, + "learning_rate": 8.159781754342488e-07, + "loss": 0.04738274961709976, + "step": 4740 + }, + { + "epoch": 2.651748251748252, + "grad_norm": 0.5745589850483095, + "learning_rate": 8.134050944063354e-07, + "loss": 0.10660860687494278, + "step": 4741 + }, + { + "epoch": 2.6523076923076925, + "grad_norm": 0.5920815300380432, + "learning_rate": 8.10835904723084e-07, + "loss": 0.06804440915584564, + "step": 4742 + }, + { + "epoch": 2.652867132867133, + "grad_norm": 0.5634126312608378, + "learning_rate": 8.082706074727741e-07, + "loss": 0.16086620092391968, + "step": 4743 + }, + { + "epoch": 2.6534265734265734, + "grad_norm": 0.4135143254609916, + "learning_rate": 8.057092037420399e-07, + "loss": 0.03833664208650589, + "step": 4744 + }, + { + "epoch": 2.653986013986014, + "grad_norm": 0.4884685498327007, + "learning_rate": 8.031516946158657e-07, + "loss": 0.0356685072183609, + "step": 4745 + }, + { + "epoch": 2.6545454545454543, + "grad_norm": 0.33047010866952825, + "learning_rate": 8.005980811775826e-07, + "loss": 0.05908123403787613, + "step": 4746 + }, + { + "epoch": 2.6551048951048952, + "grad_norm": 0.6074591754312414, + "learning_rate": 7.980483645088766e-07, + "loss": 0.11329571902751923, + "step": 4747 + }, + { + "epoch": 2.6556643356643357, + "grad_norm": 0.5210650568692843, + "learning_rate": 7.955025456897791e-07, + "loss": 0.11430435627698898, + "step": 4748 + }, + { + "epoch": 2.656223776223776, + "grad_norm": 0.31152674441470524, + "learning_rate": 7.929606257986744e-07, + "loss": 0.02996731735765934, + "step": 4749 + }, + { + "epoch": 2.656783216783217, + "grad_norm": 0.5540484736364383, + "learning_rate": 7.904226059122921e-07, + "loss": 0.18464013934135437, + "step": 4750 + }, + { + "epoch": 2.6573426573426575, + "grad_norm": 0.4629625391228474, + "learning_rate": 7.87888487105709e-07, + "loss": 0.13110829889774323, + "step": 4751 + }, + { + "epoch": 2.657902097902098, + "grad_norm": 0.3979403324083189, + "learning_rate": 7.853582704523532e-07, + "loss": 0.15432825684547424, + "step": 4752 + }, + { + "epoch": 2.6584615384615384, + "grad_norm": 0.40956599930022236, + "learning_rate": 7.828319570239962e-07, + "loss": 0.13284648954868317, + "step": 4753 + }, + { + "epoch": 2.659020979020979, + "grad_norm": 0.5352375732692416, + "learning_rate": 7.803095478907596e-07, + "loss": 0.13804754614830017, + "step": 4754 + }, + { + "epoch": 2.6595804195804194, + "grad_norm": 0.24260705712596023, + "learning_rate": 7.777910441211045e-07, + "loss": 0.004590172320604324, + "step": 4755 + }, + { + "epoch": 2.6601398601398603, + "grad_norm": 0.6337039834378678, + "learning_rate": 7.752764467818496e-07, + "loss": 0.0848199650645256, + "step": 4756 + }, + { + "epoch": 2.6606993006993007, + "grad_norm": 0.4882221845063981, + "learning_rate": 7.727657569381474e-07, + "loss": 0.08241723477840424, + "step": 4757 + }, + { + "epoch": 2.661258741258741, + "grad_norm": 0.32541607077413814, + "learning_rate": 7.702589756534995e-07, + "loss": 0.03735372796654701, + "step": 4758 + }, + { + "epoch": 2.661818181818182, + "grad_norm": 0.4719447315753768, + "learning_rate": 7.677561039897552e-07, + "loss": 0.09620705991983414, + "step": 4759 + }, + { + "epoch": 2.6623776223776225, + "grad_norm": 0.49514533585203824, + "learning_rate": 7.652571430070987e-07, + "loss": 0.12508580088615417, + "step": 4760 + }, + { + "epoch": 2.662937062937063, + "grad_norm": 0.4932972393262243, + "learning_rate": 7.627620937640712e-07, + "loss": 0.10307709127664566, + "step": 4761 + }, + { + "epoch": 2.6634965034965035, + "grad_norm": 0.5104943819407545, + "learning_rate": 7.602709573175426e-07, + "loss": 0.075468048453331, + "step": 4762 + }, + { + "epoch": 2.664055944055944, + "grad_norm": 0.5981071807728617, + "learning_rate": 7.577837347227346e-07, + "loss": 0.09586608409881592, + "step": 4763 + }, + { + "epoch": 2.6646153846153844, + "grad_norm": 0.47478604006920144, + "learning_rate": 7.55300427033211e-07, + "loss": 0.10071629285812378, + "step": 4764 + }, + { + "epoch": 2.6651748251748253, + "grad_norm": 0.5645558666189217, + "learning_rate": 7.52821035300868e-07, + "loss": 0.1285012662410736, + "step": 4765 + }, + { + "epoch": 2.6657342657342658, + "grad_norm": 0.43833638997842506, + "learning_rate": 7.503455605759568e-07, + "loss": 0.06689862161874771, + "step": 4766 + }, + { + "epoch": 2.666293706293706, + "grad_norm": 0.6589008804618813, + "learning_rate": 7.478740039070587e-07, + "loss": 0.11492881178855896, + "step": 4767 + }, + { + "epoch": 2.666853146853147, + "grad_norm": 0.4619395763858191, + "learning_rate": 7.454063663410993e-07, + "loss": 0.041836850345134735, + "step": 4768 + }, + { + "epoch": 2.6674125874125876, + "grad_norm": 0.5327696621122066, + "learning_rate": 7.42942648923346e-07, + "loss": 0.08689714223146439, + "step": 4769 + }, + { + "epoch": 2.667972027972028, + "grad_norm": 0.17761878583059235, + "learning_rate": 7.404828526973995e-07, + "loss": 0.0075935483910143375, + "step": 4770 + }, + { + "epoch": 2.6685314685314685, + "grad_norm": 0.41702783270400495, + "learning_rate": 7.380269787052074e-07, + "loss": 0.0607183612883091, + "step": 4771 + }, + { + "epoch": 2.669090909090909, + "grad_norm": 0.3685213242200397, + "learning_rate": 7.3557502798705e-07, + "loss": 0.057404614984989166, + "step": 4772 + }, + { + "epoch": 2.6696503496503494, + "grad_norm": 0.39804899145884964, + "learning_rate": 7.331270015815484e-07, + "loss": 0.050055570900440216, + "step": 4773 + }, + { + "epoch": 2.6702097902097903, + "grad_norm": 0.5186615534706113, + "learning_rate": 7.306829005256621e-07, + "loss": 0.10699306428432465, + "step": 4774 + }, + { + "epoch": 2.670769230769231, + "grad_norm": 0.4865733664934921, + "learning_rate": 7.282427258546832e-07, + "loss": 0.11535626649856567, + "step": 4775 + }, + { + "epoch": 2.6713286713286712, + "grad_norm": 0.5077764578927968, + "learning_rate": 7.258064786022456e-07, + "loss": 0.050010714679956436, + "step": 4776 + }, + { + "epoch": 2.671888111888112, + "grad_norm": 0.6749044497405658, + "learning_rate": 7.233741598003175e-07, + "loss": 0.2015446275472641, + "step": 4777 + }, + { + "epoch": 2.6724475524475526, + "grad_norm": 0.499614156645605, + "learning_rate": 7.209457704792044e-07, + "loss": 0.16903989017009735, + "step": 4778 + }, + { + "epoch": 2.673006993006993, + "grad_norm": 0.5040239826972029, + "learning_rate": 7.185213116675482e-07, + "loss": 0.07091642171144485, + "step": 4779 + }, + { + "epoch": 2.6735664335664335, + "grad_norm": 0.4029360608739635, + "learning_rate": 7.161007843923195e-07, + "loss": 0.03246210515499115, + "step": 4780 + }, + { + "epoch": 2.674125874125874, + "grad_norm": 0.31422457400302634, + "learning_rate": 7.136841896788316e-07, + "loss": 0.021352393552660942, + "step": 4781 + }, + { + "epoch": 2.6746853146853145, + "grad_norm": 0.39730140090288874, + "learning_rate": 7.112715285507266e-07, + "loss": 0.07430198788642883, + "step": 4782 + }, + { + "epoch": 2.6752447552447554, + "grad_norm": 0.6584346629914569, + "learning_rate": 7.088628020299826e-07, + "loss": 0.14261582493782043, + "step": 4783 + }, + { + "epoch": 2.675804195804196, + "grad_norm": 0.5355322275771258, + "learning_rate": 7.064580111369135e-07, + "loss": 0.0869755819439888, + "step": 4784 + }, + { + "epoch": 2.6763636363636363, + "grad_norm": 0.48894453749897726, + "learning_rate": 7.040571568901599e-07, + "loss": 0.09919381886720657, + "step": 4785 + }, + { + "epoch": 2.676923076923077, + "grad_norm": 0.41648107356760994, + "learning_rate": 7.016602403066997e-07, + "loss": 0.06281782686710358, + "step": 4786 + }, + { + "epoch": 2.6774825174825176, + "grad_norm": 0.4772607981001247, + "learning_rate": 6.992672624018404e-07, + "loss": 0.050296515226364136, + "step": 4787 + }, + { + "epoch": 2.678041958041958, + "grad_norm": 0.4592568979529306, + "learning_rate": 6.968782241892236e-07, + "loss": 0.08135329186916351, + "step": 4788 + }, + { + "epoch": 2.6786013986013986, + "grad_norm": 0.45578694246887974, + "learning_rate": 6.944931266808208e-07, + "loss": 0.11414910107851028, + "step": 4789 + }, + { + "epoch": 2.679160839160839, + "grad_norm": 0.6564038045385715, + "learning_rate": 6.921119708869329e-07, + "loss": 0.08645497262477875, + "step": 4790 + }, + { + "epoch": 2.6797202797202795, + "grad_norm": 0.6175321063342091, + "learning_rate": 6.897347578161928e-07, + "loss": 0.09158705919981003, + "step": 4791 + }, + { + "epoch": 2.6802797202797204, + "grad_norm": 0.7619437271626575, + "learning_rate": 6.873614884755631e-07, + "loss": 0.17973332107067108, + "step": 4792 + }, + { + "epoch": 2.680839160839161, + "grad_norm": 0.3763750433847247, + "learning_rate": 6.84992163870335e-07, + "loss": 0.059885382652282715, + "step": 4793 + }, + { + "epoch": 2.6813986013986013, + "grad_norm": 0.5914950789572361, + "learning_rate": 6.826267850041313e-07, + "loss": 0.13751119375228882, + "step": 4794 + }, + { + "epoch": 2.681958041958042, + "grad_norm": 0.22696635688845884, + "learning_rate": 6.802653528789005e-07, + "loss": 0.013737626373767853, + "step": 4795 + }, + { + "epoch": 2.6825174825174827, + "grad_norm": 0.4505508317410508, + "learning_rate": 6.779078684949192e-07, + "loss": 0.13206090033054352, + "step": 4796 + }, + { + "epoch": 2.683076923076923, + "grad_norm": 0.46316763830105434, + "learning_rate": 6.75554332850793e-07, + "loss": 0.07776590436697006, + "step": 4797 + }, + { + "epoch": 2.6836363636363636, + "grad_norm": 0.4712889260185119, + "learning_rate": 6.732047469434577e-07, + "loss": 0.038484424352645874, + "step": 4798 + }, + { + "epoch": 2.684195804195804, + "grad_norm": 0.2984676175375853, + "learning_rate": 6.708591117681673e-07, + "loss": 0.044652439653873444, + "step": 4799 + }, + { + "epoch": 2.6847552447552445, + "grad_norm": 0.5556085968290373, + "learning_rate": 6.685174283185159e-07, + "loss": 0.1714843213558197, + "step": 4800 + }, + { + "epoch": 2.6853146853146854, + "grad_norm": 0.4594567561782165, + "learning_rate": 6.661796975864088e-07, + "loss": 0.07104168087244034, + "step": 4801 + }, + { + "epoch": 2.685874125874126, + "grad_norm": 0.46812332545162305, + "learning_rate": 6.638459205620873e-07, + "loss": 0.043704692274332047, + "step": 4802 + }, + { + "epoch": 2.6864335664335663, + "grad_norm": 0.3945443050625129, + "learning_rate": 6.615160982341162e-07, + "loss": 0.07925727218389511, + "step": 4803 + }, + { + "epoch": 2.6869930069930072, + "grad_norm": 0.5990750566696647, + "learning_rate": 6.591902315893794e-07, + "loss": 0.1392529010772705, + "step": 4804 + }, + { + "epoch": 2.6875524475524477, + "grad_norm": 0.6896771922449796, + "learning_rate": 6.568683216130944e-07, + "loss": 0.15204253792762756, + "step": 4805 + }, + { + "epoch": 2.688111888111888, + "grad_norm": 0.538437130315094, + "learning_rate": 6.545503692887934e-07, + "loss": 0.06843969225883484, + "step": 4806 + }, + { + "epoch": 2.6886713286713286, + "grad_norm": 0.5004868076503042, + "learning_rate": 6.522363755983386e-07, + "loss": 0.1008538007736206, + "step": 4807 + }, + { + "epoch": 2.689230769230769, + "grad_norm": 0.5780150137968935, + "learning_rate": 6.499263415219159e-07, + "loss": 0.0961901918053627, + "step": 4808 + }, + { + "epoch": 2.6897902097902096, + "grad_norm": 0.3784987524538954, + "learning_rate": 6.476202680380239e-07, + "loss": 0.053754642605781555, + "step": 4809 + }, + { + "epoch": 2.6903496503496505, + "grad_norm": 0.2839775881193754, + "learning_rate": 6.453181561234989e-07, + "loss": 0.032822247594594955, + "step": 4810 + }, + { + "epoch": 2.690909090909091, + "grad_norm": 0.5491632610256242, + "learning_rate": 6.430200067534875e-07, + "loss": 0.11335393041372299, + "step": 4811 + }, + { + "epoch": 2.6914685314685314, + "grad_norm": 0.3967254707704823, + "learning_rate": 6.407258209014611e-07, + "loss": 0.08931862562894821, + "step": 4812 + }, + { + "epoch": 2.6920279720279723, + "grad_norm": 0.6844491631621199, + "learning_rate": 6.384355995392155e-07, + "loss": 0.15237212181091309, + "step": 4813 + }, + { + "epoch": 2.6925874125874127, + "grad_norm": 0.4154467167576291, + "learning_rate": 6.361493436368593e-07, + "loss": 0.06374464929103851, + "step": 4814 + }, + { + "epoch": 2.693146853146853, + "grad_norm": 0.3251483254317818, + "learning_rate": 6.338670541628322e-07, + "loss": 0.02318044751882553, + "step": 4815 + }, + { + "epoch": 2.6937062937062937, + "grad_norm": 0.4455796548922268, + "learning_rate": 6.315887320838842e-07, + "loss": 0.07137596607208252, + "step": 4816 + }, + { + "epoch": 2.694265734265734, + "grad_norm": 0.5149132436385981, + "learning_rate": 6.29314378365089e-07, + "loss": 0.07192157208919525, + "step": 4817 + }, + { + "epoch": 2.6948251748251746, + "grad_norm": 0.4333220531508339, + "learning_rate": 6.270439939698414e-07, + "loss": 0.08717767149209976, + "step": 4818 + }, + { + "epoch": 2.6953846153846155, + "grad_norm": 0.11353542341861643, + "learning_rate": 6.247775798598466e-07, + "loss": 0.010756965726613998, + "step": 4819 + }, + { + "epoch": 2.695944055944056, + "grad_norm": 1.0597242159678635, + "learning_rate": 6.225151369951399e-07, + "loss": 0.22568590939044952, + "step": 4820 + }, + { + "epoch": 2.6965034965034964, + "grad_norm": 0.8636425954808148, + "learning_rate": 6.202566663340648e-07, + "loss": 0.13338537514209747, + "step": 4821 + }, + { + "epoch": 2.6970629370629373, + "grad_norm": 0.5563255058524349, + "learning_rate": 6.18002168833286e-07, + "loss": 0.21569012105464935, + "step": 4822 + }, + { + "epoch": 2.6976223776223778, + "grad_norm": 0.4972137290109715, + "learning_rate": 6.157516454477874e-07, + "loss": 0.03815922141075134, + "step": 4823 + }, + { + "epoch": 2.6981818181818182, + "grad_norm": 0.5619426627683893, + "learning_rate": 6.13505097130862e-07, + "loss": 0.2077419012784958, + "step": 4824 + }, + { + "epoch": 2.6987412587412587, + "grad_norm": 0.5794222033165627, + "learning_rate": 6.112625248341286e-07, + "loss": 0.18102942407131195, + "step": 4825 + }, + { + "epoch": 2.699300699300699, + "grad_norm": 0.5226775800136841, + "learning_rate": 6.09023929507514e-07, + "loss": 0.17260771989822388, + "step": 4826 + }, + { + "epoch": 2.6998601398601396, + "grad_norm": 0.7368835401566404, + "learning_rate": 6.067893120992652e-07, + "loss": 0.26072627305984497, + "step": 4827 + }, + { + "epoch": 2.7004195804195805, + "grad_norm": 0.4119820101944753, + "learning_rate": 6.045586735559428e-07, + "loss": 0.0822022408246994, + "step": 4828 + }, + { + "epoch": 2.700979020979021, + "grad_norm": 0.4650128874557431, + "learning_rate": 6.023320148224221e-07, + "loss": 0.10042086243629456, + "step": 4829 + }, + { + "epoch": 2.7015384615384614, + "grad_norm": 0.35784846497761535, + "learning_rate": 6.001093368418897e-07, + "loss": 0.1179453581571579, + "step": 4830 + }, + { + "epoch": 2.7020979020979023, + "grad_norm": 0.3975019047249368, + "learning_rate": 5.978906405558504e-07, + "loss": 0.0413038544356823, + "step": 4831 + }, + { + "epoch": 2.702657342657343, + "grad_norm": 0.8608494476664581, + "learning_rate": 5.956759269041201e-07, + "loss": 0.13676247000694275, + "step": 4832 + }, + { + "epoch": 2.7032167832167833, + "grad_norm": 0.3488526678640786, + "learning_rate": 5.934651968248295e-07, + "loss": 0.05617497116327286, + "step": 4833 + }, + { + "epoch": 2.7037762237762237, + "grad_norm": 0.4818130260875163, + "learning_rate": 5.912584512544195e-07, + "loss": 0.13416530191898346, + "step": 4834 + }, + { + "epoch": 2.704335664335664, + "grad_norm": 0.8004503260638752, + "learning_rate": 5.890556911276435e-07, + "loss": 0.1994042694568634, + "step": 4835 + }, + { + "epoch": 2.7048951048951047, + "grad_norm": 0.3854214238800483, + "learning_rate": 5.868569173775673e-07, + "loss": 0.05722580477595329, + "step": 4836 + }, + { + "epoch": 2.7054545454545456, + "grad_norm": 0.43037278076753127, + "learning_rate": 5.846621309355693e-07, + "loss": 0.08654353022575378, + "step": 4837 + }, + { + "epoch": 2.706013986013986, + "grad_norm": 0.9688732020553799, + "learning_rate": 5.82471332731338e-07, + "loss": 0.15309712290763855, + "step": 4838 + }, + { + "epoch": 2.7065734265734265, + "grad_norm": 0.3766526969354465, + "learning_rate": 5.802845236928733e-07, + "loss": 0.06144693121314049, + "step": 4839 + }, + { + "epoch": 2.7071328671328674, + "grad_norm": 0.42049719790313905, + "learning_rate": 5.78101704746482e-07, + "loss": 0.0667698010802269, + "step": 4840 + }, + { + "epoch": 2.707692307692308, + "grad_norm": 0.5712328606530926, + "learning_rate": 5.759228768167835e-07, + "loss": 0.24342583119869232, + "step": 4841 + }, + { + "epoch": 2.7082517482517483, + "grad_norm": 0.3986207530185005, + "learning_rate": 5.737480408267082e-07, + "loss": 0.04426802694797516, + "step": 4842 + }, + { + "epoch": 2.7088111888111888, + "grad_norm": 0.42507582232320085, + "learning_rate": 5.715771976974915e-07, + "loss": 0.13820983469486237, + "step": 4843 + }, + { + "epoch": 2.7093706293706292, + "grad_norm": 0.2960289028152606, + "learning_rate": 5.694103483486835e-07, + "loss": 0.028235316276550293, + "step": 4844 + }, + { + "epoch": 2.7099300699300697, + "grad_norm": 0.3487337957509491, + "learning_rate": 5.672474936981332e-07, + "loss": 0.03292941302061081, + "step": 4845 + }, + { + "epoch": 2.7104895104895106, + "grad_norm": 0.38239161153109746, + "learning_rate": 5.650886346620066e-07, + "loss": 0.04234527796506882, + "step": 4846 + }, + { + "epoch": 2.711048951048951, + "grad_norm": 1.1227131434960178, + "learning_rate": 5.629337721547734e-07, + "loss": 0.27317675948143005, + "step": 4847 + }, + { + "epoch": 2.7116083916083915, + "grad_norm": 0.5478787348110046, + "learning_rate": 5.60782907089208e-07, + "loss": 0.11615654826164246, + "step": 4848 + }, + { + "epoch": 2.7121678321678324, + "grad_norm": 0.4217171289751604, + "learning_rate": 5.586360403763991e-07, + "loss": 0.085695780813694, + "step": 4849 + }, + { + "epoch": 2.712727272727273, + "grad_norm": 0.40400220301301487, + "learning_rate": 5.564931729257317e-07, + "loss": 0.04336924105882645, + "step": 4850 + }, + { + "epoch": 2.7132867132867133, + "grad_norm": 0.36865686394943475, + "learning_rate": 5.543543056449041e-07, + "loss": 0.10591655224561691, + "step": 4851 + }, + { + "epoch": 2.713846153846154, + "grad_norm": 0.39509890929787256, + "learning_rate": 5.522194394399194e-07, + "loss": 0.022871579974889755, + "step": 4852 + }, + { + "epoch": 2.7144055944055943, + "grad_norm": 0.5979420192154814, + "learning_rate": 5.500885752150798e-07, + "loss": 0.11475977301597595, + "step": 4853 + }, + { + "epoch": 2.7149650349650347, + "grad_norm": 0.2528339779355034, + "learning_rate": 5.479617138730031e-07, + "loss": 0.03981172293424606, + "step": 4854 + }, + { + "epoch": 2.7155244755244756, + "grad_norm": 0.4683777920204621, + "learning_rate": 5.458388563146e-07, + "loss": 0.07584191113710403, + "step": 4855 + }, + { + "epoch": 2.716083916083916, + "grad_norm": 0.36681330867634204, + "learning_rate": 5.437200034390932e-07, + "loss": 0.029487628489732742, + "step": 4856 + }, + { + "epoch": 2.7166433566433565, + "grad_norm": 0.5319533150430793, + "learning_rate": 5.416051561440072e-07, + "loss": 0.10549632459878922, + "step": 4857 + }, + { + "epoch": 2.7172027972027974, + "grad_norm": 0.5981260922009679, + "learning_rate": 5.394943153251642e-07, + "loss": 0.12315791845321655, + "step": 4858 + }, + { + "epoch": 2.717762237762238, + "grad_norm": 0.42530466592128696, + "learning_rate": 5.373874818767011e-07, + "loss": 0.0706019476056099, + "step": 4859 + }, + { + "epoch": 2.7183216783216784, + "grad_norm": 0.8973281891723267, + "learning_rate": 5.352846566910453e-07, + "loss": 0.16092607378959656, + "step": 4860 + }, + { + "epoch": 2.718881118881119, + "grad_norm": 0.22954690591170063, + "learning_rate": 5.33185840658933e-07, + "loss": 0.01062640268355608, + "step": 4861 + }, + { + "epoch": 2.7194405594405593, + "grad_norm": 0.5443075772897138, + "learning_rate": 5.310910346694009e-07, + "loss": 0.10521841049194336, + "step": 4862 + }, + { + "epoch": 2.7199999999999998, + "grad_norm": 0.40246180353862177, + "learning_rate": 5.290002396097849e-07, + "loss": 0.056228138506412506, + "step": 4863 + }, + { + "epoch": 2.7205594405594407, + "grad_norm": 0.5540308690450843, + "learning_rate": 5.269134563657264e-07, + "loss": 0.09026574343442917, + "step": 4864 + }, + { + "epoch": 2.721118881118881, + "grad_norm": 0.3066049891318509, + "learning_rate": 5.248306858211627e-07, + "loss": 0.04187865927815437, + "step": 4865 + }, + { + "epoch": 2.7216783216783216, + "grad_norm": 0.3726347381385315, + "learning_rate": 5.227519288583338e-07, + "loss": 0.07991749793291092, + "step": 4866 + }, + { + "epoch": 2.7222377622377625, + "grad_norm": 0.556310370948449, + "learning_rate": 5.206771863577797e-07, + "loss": 0.10445399582386017, + "step": 4867 + }, + { + "epoch": 2.722797202797203, + "grad_norm": 0.5114098147509121, + "learning_rate": 5.186064591983386e-07, + "loss": 0.07234948128461838, + "step": 4868 + }, + { + "epoch": 2.7233566433566434, + "grad_norm": 0.5758765648409454, + "learning_rate": 5.165397482571511e-07, + "loss": 0.14555084705352783, + "step": 4869 + }, + { + "epoch": 2.723916083916084, + "grad_norm": 0.5908776240678022, + "learning_rate": 5.144770544096512e-07, + "loss": 0.11011561751365662, + "step": 4870 + }, + { + "epoch": 2.7244755244755243, + "grad_norm": 0.5072549085281551, + "learning_rate": 5.124183785295755e-07, + "loss": 0.10255992412567139, + "step": 4871 + }, + { + "epoch": 2.725034965034965, + "grad_norm": 0.19096395577568917, + "learning_rate": 5.103637214889568e-07, + "loss": 0.01201770268380642, + "step": 4872 + }, + { + "epoch": 2.7255944055944057, + "grad_norm": 0.8700321732188144, + "learning_rate": 5.083130841581286e-07, + "loss": 0.11315815150737762, + "step": 4873 + }, + { + "epoch": 2.726153846153846, + "grad_norm": 0.42488765028030595, + "learning_rate": 5.06266467405715e-07, + "loss": 0.05612741410732269, + "step": 4874 + }, + { + "epoch": 2.7267132867132866, + "grad_norm": 0.6340268663079893, + "learning_rate": 5.042238720986436e-07, + "loss": 0.2006445676088333, + "step": 4875 + }, + { + "epoch": 2.7272727272727275, + "grad_norm": 0.5360725737229807, + "learning_rate": 5.02185299102137e-07, + "loss": 0.09439338743686676, + "step": 4876 + }, + { + "epoch": 2.727832167832168, + "grad_norm": 0.5196647469596578, + "learning_rate": 5.001507492797119e-07, + "loss": 0.08648508787155151, + "step": 4877 + }, + { + "epoch": 2.7283916083916084, + "grad_norm": 0.5134082014987806, + "learning_rate": 4.981202234931837e-07, + "loss": 0.09073715656995773, + "step": 4878 + }, + { + "epoch": 2.728951048951049, + "grad_norm": 0.27174554367988346, + "learning_rate": 4.960937226026596e-07, + "loss": 0.02977767214179039, + "step": 4879 + }, + { + "epoch": 2.7295104895104894, + "grad_norm": 0.3250346435081105, + "learning_rate": 4.940712474665443e-07, + "loss": 0.06906495243310928, + "step": 4880 + }, + { + "epoch": 2.73006993006993, + "grad_norm": 0.5881946086560043, + "learning_rate": 4.92052798941538e-07, + "loss": 0.14573818445205688, + "step": 4881 + }, + { + "epoch": 2.7306293706293707, + "grad_norm": 0.6509703816444146, + "learning_rate": 4.900383778826345e-07, + "loss": 0.12271745502948761, + "step": 4882 + }, + { + "epoch": 2.731188811188811, + "grad_norm": 0.8818397803799517, + "learning_rate": 4.880279851431214e-07, + "loss": 0.08478693664073944, + "step": 4883 + }, + { + "epoch": 2.7317482517482516, + "grad_norm": 0.40261267028354925, + "learning_rate": 4.860216215745772e-07, + "loss": 0.07550100237131119, + "step": 4884 + }, + { + "epoch": 2.7323076923076925, + "grad_norm": 0.45799363677707355, + "learning_rate": 4.84019288026879e-07, + "loss": 0.12323139607906342, + "step": 4885 + }, + { + "epoch": 2.732867132867133, + "grad_norm": 0.5797879709216026, + "learning_rate": 4.820209853481928e-07, + "loss": 0.14971648156642914, + "step": 4886 + }, + { + "epoch": 2.7334265734265735, + "grad_norm": 0.4501063431132741, + "learning_rate": 4.800267143849791e-07, + "loss": 0.17323225736618042, + "step": 4887 + }, + { + "epoch": 2.733986013986014, + "grad_norm": 0.5440903769682947, + "learning_rate": 4.780364759819911e-07, + "loss": 0.09519673883914948, + "step": 4888 + }, + { + "epoch": 2.7345454545454544, + "grad_norm": 0.34532230935778396, + "learning_rate": 4.7605027098226896e-07, + "loss": 0.044987231492996216, + "step": 4889 + }, + { + "epoch": 2.735104895104895, + "grad_norm": 0.5950370110286977, + "learning_rate": 4.740681002271541e-07, + "loss": 0.139259934425354, + "step": 4890 + }, + { + "epoch": 2.7356643356643358, + "grad_norm": 0.7203951531471128, + "learning_rate": 4.720899645562682e-07, + "loss": 0.1729244440793991, + "step": 4891 + }, + { + "epoch": 2.736223776223776, + "grad_norm": 0.25875740894806315, + "learning_rate": 4.7011586480753105e-07, + "loss": 0.009089294821023941, + "step": 4892 + }, + { + "epoch": 2.7367832167832167, + "grad_norm": 0.4922648404150608, + "learning_rate": 4.681458018171503e-07, + "loss": 0.10830871760845184, + "step": 4893 + }, + { + "epoch": 2.7373426573426576, + "grad_norm": 0.5531009894421701, + "learning_rate": 4.6617977641962075e-07, + "loss": 0.12814119458198547, + "step": 4894 + }, + { + "epoch": 2.737902097902098, + "grad_norm": 0.4083397336276903, + "learning_rate": 4.642177894477351e-07, + "loss": 0.0661759153008461, + "step": 4895 + }, + { + "epoch": 2.7384615384615385, + "grad_norm": 0.5137005360000421, + "learning_rate": 4.6225984173256853e-07, + "loss": 0.20286014676094055, + "step": 4896 + }, + { + "epoch": 2.739020979020979, + "grad_norm": 0.7609374643921859, + "learning_rate": 4.603059341034832e-07, + "loss": 0.06597593426704407, + "step": 4897 + }, + { + "epoch": 2.7395804195804194, + "grad_norm": 0.5694604409823816, + "learning_rate": 4.583560673881404e-07, + "loss": 0.09235873818397522, + "step": 4898 + }, + { + "epoch": 2.74013986013986, + "grad_norm": 0.3421165501165328, + "learning_rate": 4.564102424124772e-07, + "loss": 0.05744379758834839, + "step": 4899 + }, + { + "epoch": 2.740699300699301, + "grad_norm": 0.41322131297235165, + "learning_rate": 4.5446846000072765e-07, + "loss": 0.07656777650117874, + "step": 4900 + }, + { + "epoch": 2.7412587412587412, + "grad_norm": 0.4794835511832768, + "learning_rate": 4.525307209754104e-07, + "loss": 0.10856030136346817, + "step": 4901 + }, + { + "epoch": 2.7418181818181817, + "grad_norm": 0.25959287138522896, + "learning_rate": 4.5059702615732783e-07, + "loss": 0.04348676651716232, + "step": 4902 + }, + { + "epoch": 2.7423776223776226, + "grad_norm": 0.23334537852746856, + "learning_rate": 4.48667376365578e-07, + "loss": 0.0500948429107666, + "step": 4903 + }, + { + "epoch": 2.742937062937063, + "grad_norm": 0.6462335425929755, + "learning_rate": 4.46741772417536e-07, + "loss": 0.1362314522266388, + "step": 4904 + }, + { + "epoch": 2.7434965034965035, + "grad_norm": 0.2849501295286663, + "learning_rate": 4.448202151288683e-07, + "loss": 0.015057351440191269, + "step": 4905 + }, + { + "epoch": 2.744055944055944, + "grad_norm": 0.2930389541172371, + "learning_rate": 4.4290270531352707e-07, + "loss": 0.014581594616174698, + "step": 4906 + }, + { + "epoch": 2.7446153846153845, + "grad_norm": 0.2502326077578538, + "learning_rate": 4.409892437837493e-07, + "loss": 0.04733207821846008, + "step": 4907 + }, + { + "epoch": 2.745174825174825, + "grad_norm": 0.3578377251301733, + "learning_rate": 4.390798313500577e-07, + "loss": 0.06507638096809387, + "step": 4908 + }, + { + "epoch": 2.745734265734266, + "grad_norm": 0.778290408122385, + "learning_rate": 4.371744688212576e-07, + "loss": 0.195505291223526, + "step": 4909 + }, + { + "epoch": 2.7462937062937063, + "grad_norm": 0.39513583583028417, + "learning_rate": 4.3527315700443996e-07, + "loss": 0.061857469379901886, + "step": 4910 + }, + { + "epoch": 2.7468531468531467, + "grad_norm": 0.4904935798491671, + "learning_rate": 4.333758967049828e-07, + "loss": 0.16515888273715973, + "step": 4911 + }, + { + "epoch": 2.7474125874125876, + "grad_norm": 0.5180481659157007, + "learning_rate": 4.314826887265433e-07, + "loss": 0.12153796851634979, + "step": 4912 + }, + { + "epoch": 2.747972027972028, + "grad_norm": 0.4868102762244039, + "learning_rate": 4.2959353387106664e-07, + "loss": 0.13263708353042603, + "step": 4913 + }, + { + "epoch": 2.7485314685314686, + "grad_norm": 0.4642124011611854, + "learning_rate": 4.277084329387771e-07, + "loss": 0.0773947685956955, + "step": 4914 + }, + { + "epoch": 2.749090909090909, + "grad_norm": 0.28412646110783557, + "learning_rate": 4.258273867281837e-07, + "loss": 0.012152200564742088, + "step": 4915 + }, + { + "epoch": 2.7496503496503495, + "grad_norm": 0.37766667409991844, + "learning_rate": 4.2395039603607915e-07, + "loss": 0.06117810308933258, + "step": 4916 + }, + { + "epoch": 2.75020979020979, + "grad_norm": 0.535104413712766, + "learning_rate": 4.2207746165753403e-07, + "loss": 0.16269856691360474, + "step": 4917 + }, + { + "epoch": 2.750769230769231, + "grad_norm": 0.5476969898121803, + "learning_rate": 4.202085843859083e-07, + "loss": 0.12468735873699188, + "step": 4918 + }, + { + "epoch": 2.7513286713286713, + "grad_norm": 0.20775838864046428, + "learning_rate": 4.183437650128341e-07, + "loss": 0.019279032945632935, + "step": 4919 + }, + { + "epoch": 2.7518881118881118, + "grad_norm": 0.4043440622339861, + "learning_rate": 4.1648300432823085e-07, + "loss": 0.030820688232779503, + "step": 4920 + }, + { + "epoch": 2.7524475524475527, + "grad_norm": 0.5399218378399923, + "learning_rate": 4.146263031202968e-07, + "loss": 0.1655111014842987, + "step": 4921 + }, + { + "epoch": 2.753006993006993, + "grad_norm": 0.5023563625642192, + "learning_rate": 4.1277366217551296e-07, + "loss": 0.11915578693151474, + "step": 4922 + }, + { + "epoch": 2.7535664335664336, + "grad_norm": 0.36759375033927216, + "learning_rate": 4.109250822786348e-07, + "loss": 0.07029218971729279, + "step": 4923 + }, + { + "epoch": 2.754125874125874, + "grad_norm": 0.7200391493982657, + "learning_rate": 4.0908056421270606e-07, + "loss": 0.09724392741918564, + "step": 4924 + }, + { + "epoch": 2.7546853146853145, + "grad_norm": 0.5134362359477282, + "learning_rate": 4.072401087590405e-07, + "loss": 0.05131128430366516, + "step": 4925 + }, + { + "epoch": 2.755244755244755, + "grad_norm": 0.18784091286794014, + "learning_rate": 4.0540371669723797e-07, + "loss": 0.00523514486849308, + "step": 4926 + }, + { + "epoch": 2.755804195804196, + "grad_norm": 0.4245385364178885, + "learning_rate": 4.035713888051762e-07, + "loss": 0.06761936843395233, + "step": 4927 + }, + { + "epoch": 2.7563636363636363, + "grad_norm": 0.24067306315016052, + "learning_rate": 4.0174312585900433e-07, + "loss": 0.060123756527900696, + "step": 4928 + }, + { + "epoch": 2.756923076923077, + "grad_norm": 0.39773065299443233, + "learning_rate": 3.999189286331628e-07, + "loss": 0.07848545908927917, + "step": 4929 + }, + { + "epoch": 2.7574825174825177, + "grad_norm": 0.3855485740734452, + "learning_rate": 3.98098797900357e-07, + "loss": 0.016435006633400917, + "step": 4930 + }, + { + "epoch": 2.758041958041958, + "grad_norm": 0.6024034216412593, + "learning_rate": 3.962827344315756e-07, + "loss": 0.12187676876783371, + "step": 4931 + }, + { + "epoch": 2.7586013986013986, + "grad_norm": 0.4284413419350956, + "learning_rate": 3.944707389960878e-07, + "loss": 0.07257356494665146, + "step": 4932 + }, + { + "epoch": 2.759160839160839, + "grad_norm": 0.7438712215344472, + "learning_rate": 3.9266281236142955e-07, + "loss": 0.15309825539588928, + "step": 4933 + }, + { + "epoch": 2.7597202797202796, + "grad_norm": 0.5050064339136041, + "learning_rate": 3.908589552934261e-07, + "loss": 0.061837293207645416, + "step": 4934 + }, + { + "epoch": 2.76027972027972, + "grad_norm": 0.3597293952063821, + "learning_rate": 3.890591685561673e-07, + "loss": 0.015514417551457882, + "step": 4935 + }, + { + "epoch": 2.760839160839161, + "grad_norm": 0.6558928206527522, + "learning_rate": 3.872634529120256e-07, + "loss": 0.17509452998638153, + "step": 4936 + }, + { + "epoch": 2.7613986013986014, + "grad_norm": 0.5555824869127682, + "learning_rate": 3.8547180912164803e-07, + "loss": 0.1754770576953888, + "step": 4937 + }, + { + "epoch": 2.761958041958042, + "grad_norm": 0.3029060418755962, + "learning_rate": 3.836842379439531e-07, + "loss": 0.028663069009780884, + "step": 4938 + }, + { + "epoch": 2.7625174825174827, + "grad_norm": 0.4291378165050777, + "learning_rate": 3.8190074013614184e-07, + "loss": 0.06737642735242844, + "step": 4939 + }, + { + "epoch": 2.763076923076923, + "grad_norm": 0.512039859154867, + "learning_rate": 3.80121316453681e-07, + "loss": 0.07121893018484116, + "step": 4940 + }, + { + "epoch": 2.7636363636363637, + "grad_norm": 0.6663419764390581, + "learning_rate": 3.783459676503165e-07, + "loss": 0.16411644220352173, + "step": 4941 + }, + { + "epoch": 2.764195804195804, + "grad_norm": 0.47699285581778567, + "learning_rate": 3.765746944780702e-07, + "loss": 0.07366091012954712, + "step": 4942 + }, + { + "epoch": 2.7647552447552446, + "grad_norm": 0.4980386771107375, + "learning_rate": 3.7480749768722957e-07, + "loss": 0.08414723724126816, + "step": 4943 + }, + { + "epoch": 2.765314685314685, + "grad_norm": 0.3995268112185005, + "learning_rate": 3.7304437802636486e-07, + "loss": 0.04255523160099983, + "step": 4944 + }, + { + "epoch": 2.765874125874126, + "grad_norm": 0.3973395856436556, + "learning_rate": 3.71285336242313e-07, + "loss": 0.051001179963350296, + "step": 4945 + }, + { + "epoch": 2.7664335664335664, + "grad_norm": 0.2499332833911326, + "learning_rate": 3.695303730801869e-07, + "loss": 0.01567598059773445, + "step": 4946 + }, + { + "epoch": 2.766993006993007, + "grad_norm": 0.5342366773351699, + "learning_rate": 3.677794892833697e-07, + "loss": 0.08082646876573563, + "step": 4947 + }, + { + "epoch": 2.767552447552448, + "grad_norm": 0.5312858852611786, + "learning_rate": 3.66032685593517e-07, + "loss": 0.12716718018054962, + "step": 4948 + }, + { + "epoch": 2.7681118881118882, + "grad_norm": 0.44090861508803575, + "learning_rate": 3.6428996275055585e-07, + "loss": 0.11594318598508835, + "step": 4949 + }, + { + "epoch": 2.7686713286713287, + "grad_norm": 0.4143973579969853, + "learning_rate": 3.6255132149268566e-07, + "loss": 0.0710601657629013, + "step": 4950 + }, + { + "epoch": 2.769230769230769, + "grad_norm": 0.4459346121874162, + "learning_rate": 3.6081676255637743e-07, + "loss": 0.12579065561294556, + "step": 4951 + }, + { + "epoch": 2.7697902097902096, + "grad_norm": 0.32773782031059, + "learning_rate": 3.5908628667637227e-07, + "loss": 0.03872643783688545, + "step": 4952 + }, + { + "epoch": 2.77034965034965, + "grad_norm": 0.3947783154190773, + "learning_rate": 3.5735989458567935e-07, + "loss": 0.04524435102939606, + "step": 4953 + }, + { + "epoch": 2.770909090909091, + "grad_norm": 0.3720351756125673, + "learning_rate": 3.556375870155815e-07, + "loss": 0.0766759067773819, + "step": 4954 + }, + { + "epoch": 2.7714685314685314, + "grad_norm": 0.49524750316201166, + "learning_rate": 3.539193646956307e-07, + "loss": 0.13944050669670105, + "step": 4955 + }, + { + "epoch": 2.772027972027972, + "grad_norm": 0.4534954393007726, + "learning_rate": 3.522052283536459e-07, + "loss": 0.0949394553899765, + "step": 4956 + }, + { + "epoch": 2.772587412587413, + "grad_norm": 0.32766227797349823, + "learning_rate": 3.504951787157207e-07, + "loss": 0.022415973246097565, + "step": 4957 + }, + { + "epoch": 2.7731468531468533, + "grad_norm": 0.5644404358457994, + "learning_rate": 3.4878921650621013e-07, + "loss": 0.16931122541427612, + "step": 4958 + }, + { + "epoch": 2.7737062937062937, + "grad_norm": 0.4058412069661956, + "learning_rate": 3.470873424477428e-07, + "loss": 0.08383970707654953, + "step": 4959 + }, + { + "epoch": 2.774265734265734, + "grad_norm": 0.42311217681404356, + "learning_rate": 3.4538955726121536e-07, + "loss": 0.024366505444049835, + "step": 4960 + }, + { + "epoch": 2.7748251748251747, + "grad_norm": 0.35307568756686214, + "learning_rate": 3.4369586166579263e-07, + "loss": 0.060419823974370956, + "step": 4961 + }, + { + "epoch": 2.775384615384615, + "grad_norm": 0.6605956684495274, + "learning_rate": 3.4200625637890394e-07, + "loss": 0.3101741075515747, + "step": 4962 + }, + { + "epoch": 2.775944055944056, + "grad_norm": 0.4844330830760498, + "learning_rate": 3.403207421162513e-07, + "loss": 0.05042243376374245, + "step": 4963 + }, + { + "epoch": 2.7765034965034965, + "grad_norm": 0.5380542100775341, + "learning_rate": 3.3863931959179697e-07, + "loss": 0.11928140372037888, + "step": 4964 + }, + { + "epoch": 2.777062937062937, + "grad_norm": 0.33493039344875253, + "learning_rate": 3.369619895177756e-07, + "loss": 0.06877477467060089, + "step": 4965 + }, + { + "epoch": 2.777622377622378, + "grad_norm": 0.5995743001537621, + "learning_rate": 3.352887526046877e-07, + "loss": 0.11583748459815979, + "step": 4966 + }, + { + "epoch": 2.7781818181818183, + "grad_norm": 0.3135350199847006, + "learning_rate": 3.3361960956129644e-07, + "loss": 0.030591977760195732, + "step": 4967 + }, + { + "epoch": 2.7787412587412588, + "grad_norm": 0.4922293043480344, + "learning_rate": 3.319545610946362e-07, + "loss": 0.0904340073466301, + "step": 4968 + }, + { + "epoch": 2.7793006993006992, + "grad_norm": 0.6336084973775882, + "learning_rate": 3.302936079100005e-07, + "loss": 0.2623438239097595, + "step": 4969 + }, + { + "epoch": 2.7798601398601397, + "grad_norm": 0.26075602774034057, + "learning_rate": 3.286367507109545e-07, + "loss": 0.04227359592914581, + "step": 4970 + }, + { + "epoch": 2.78041958041958, + "grad_norm": 0.5139184732510992, + "learning_rate": 3.269839901993244e-07, + "loss": 0.08644302934408188, + "step": 4971 + }, + { + "epoch": 2.780979020979021, + "grad_norm": 0.36513075297158504, + "learning_rate": 3.253353270751991e-07, + "loss": 0.03157906234264374, + "step": 4972 + }, + { + "epoch": 2.7815384615384615, + "grad_norm": 0.4682470767146403, + "learning_rate": 3.2369076203694114e-07, + "loss": 0.07002183794975281, + "step": 4973 + }, + { + "epoch": 2.782097902097902, + "grad_norm": 0.3448098625682289, + "learning_rate": 3.220502957811655e-07, + "loss": 0.04069294035434723, + "step": 4974 + }, + { + "epoch": 2.782657342657343, + "grad_norm": 0.34222328829527376, + "learning_rate": 3.204139290027586e-07, + "loss": 0.03844890370965004, + "step": 4975 + }, + { + "epoch": 2.7832167832167833, + "grad_norm": 0.4931608673154309, + "learning_rate": 3.187816623948692e-07, + "loss": 0.1076214462518692, + "step": 4976 + }, + { + "epoch": 2.783776223776224, + "grad_norm": 0.3606352802836697, + "learning_rate": 3.1715349664890425e-07, + "loss": 0.057912588119506836, + "step": 4977 + }, + { + "epoch": 2.7843356643356643, + "grad_norm": 0.4923011272105909, + "learning_rate": 3.1552943245454215e-07, + "loss": 0.14520801603794098, + "step": 4978 + }, + { + "epoch": 2.7848951048951047, + "grad_norm": 0.4021288980968012, + "learning_rate": 3.13909470499717e-07, + "loss": 0.05710513889789581, + "step": 4979 + }, + { + "epoch": 2.785454545454545, + "grad_norm": 0.25908680119295446, + "learning_rate": 3.122936114706276e-07, + "loss": 0.017719509080052376, + "step": 4980 + }, + { + "epoch": 2.786013986013986, + "grad_norm": 0.27734741343009806, + "learning_rate": 3.106818560517366e-07, + "loss": 0.027054881677031517, + "step": 4981 + }, + { + "epoch": 2.7865734265734265, + "grad_norm": 0.3278934600833566, + "learning_rate": 3.0907420492576334e-07, + "loss": 0.02746594324707985, + "step": 4982 + }, + { + "epoch": 2.787132867132867, + "grad_norm": 0.5098919479679965, + "learning_rate": 3.074706587736953e-07, + "loss": 0.1467277556657791, + "step": 4983 + }, + { + "epoch": 2.787692307692308, + "grad_norm": 0.43695784640648994, + "learning_rate": 3.0587121827477473e-07, + "loss": 0.08682317286729813, + "step": 4984 + }, + { + "epoch": 2.7882517482517484, + "grad_norm": 0.3316011328984541, + "learning_rate": 3.0427588410650965e-07, + "loss": 0.050984062254428864, + "step": 4985 + }, + { + "epoch": 2.788811188811189, + "grad_norm": 0.32621095760190205, + "learning_rate": 3.026846569446673e-07, + "loss": 0.06781367212533951, + "step": 4986 + }, + { + "epoch": 2.7893706293706293, + "grad_norm": 0.47748248326823306, + "learning_rate": 3.0109753746327183e-07, + "loss": 0.16698311269283295, + "step": 4987 + }, + { + "epoch": 2.7899300699300698, + "grad_norm": 0.471261085533029, + "learning_rate": 2.995145263346133e-07, + "loss": 0.13171671330928802, + "step": 4988 + }, + { + "epoch": 2.79048951048951, + "grad_norm": 0.4594973071880267, + "learning_rate": 2.979356242292375e-07, + "loss": 0.1539926677942276, + "step": 4989 + }, + { + "epoch": 2.791048951048951, + "grad_norm": 0.25869233017840937, + "learning_rate": 2.9636083181595053e-07, + "loss": 0.027205422520637512, + "step": 4990 + }, + { + "epoch": 2.7916083916083916, + "grad_norm": 0.36339228038700544, + "learning_rate": 2.9479014976181775e-07, + "loss": 0.04700343683362007, + "step": 4991 + }, + { + "epoch": 2.792167832167832, + "grad_norm": 0.3918011213294115, + "learning_rate": 2.9322357873216245e-07, + "loss": 0.05692625418305397, + "step": 4992 + }, + { + "epoch": 2.792727272727273, + "grad_norm": 0.9620374703711249, + "learning_rate": 2.9166111939057053e-07, + "loss": 0.12826746702194214, + "step": 4993 + }, + { + "epoch": 2.7932867132867134, + "grad_norm": 0.5628257230962423, + "learning_rate": 2.901027723988803e-07, + "loss": 0.11037755757570267, + "step": 4994 + }, + { + "epoch": 2.793846153846154, + "grad_norm": 0.48503721906504177, + "learning_rate": 2.8854853841719265e-07, + "loss": 0.06750046461820602, + "step": 4995 + }, + { + "epoch": 2.7944055944055943, + "grad_norm": 0.6879371525889812, + "learning_rate": 2.869984181038654e-07, + "loss": 0.12424492835998535, + "step": 4996 + }, + { + "epoch": 2.794965034965035, + "grad_norm": 1.0323622719268133, + "learning_rate": 2.854524121155111e-07, + "loss": 0.11639942973852158, + "step": 4997 + }, + { + "epoch": 2.7955244755244752, + "grad_norm": 0.5330209762619117, + "learning_rate": 2.839105211070026e-07, + "loss": 0.13956589996814728, + "step": 4998 + }, + { + "epoch": 2.796083916083916, + "grad_norm": 0.3758318085237973, + "learning_rate": 2.8237274573146955e-07, + "loss": 0.05117888003587723, + "step": 4999 + }, + { + "epoch": 2.7966433566433566, + "grad_norm": 0.48712681362470217, + "learning_rate": 2.808390866402955e-07, + "loss": 0.07312138378620148, + "step": 5000 + }, + { + "epoch": 2.797202797202797, + "grad_norm": 0.49352054354040764, + "learning_rate": 2.793095444831229e-07, + "loss": 0.10789519548416138, + "step": 5001 + }, + { + "epoch": 2.797762237762238, + "grad_norm": 0.5002974009388221, + "learning_rate": 2.7778411990785124e-07, + "loss": 0.10860700905323029, + "step": 5002 + }, + { + "epoch": 2.7983216783216784, + "grad_norm": 0.577204203813806, + "learning_rate": 2.762628135606327e-07, + "loss": 0.15171238780021667, + "step": 5003 + }, + { + "epoch": 2.798881118881119, + "grad_norm": 0.24484897642414663, + "learning_rate": 2.747456260858761e-07, + "loss": 0.019601235166192055, + "step": 5004 + }, + { + "epoch": 2.7994405594405594, + "grad_norm": 0.3639286824124896, + "learning_rate": 2.732325581262474e-07, + "loss": 0.06370805203914642, + "step": 5005 + }, + { + "epoch": 2.8, + "grad_norm": 0.5746596353891849, + "learning_rate": 2.7172361032266504e-07, + "loss": 0.14868727326393127, + "step": 5006 + }, + { + "epoch": 2.8005594405594403, + "grad_norm": 0.17376305487723553, + "learning_rate": 2.702187833143055e-07, + "loss": 0.012390978634357452, + "step": 5007 + }, + { + "epoch": 2.801118881118881, + "grad_norm": 0.3601777141781473, + "learning_rate": 2.687180777385956e-07, + "loss": 0.09371516853570938, + "step": 5008 + }, + { + "epoch": 2.8016783216783216, + "grad_norm": 0.595495190115014, + "learning_rate": 2.6722149423121903e-07, + "loss": 0.10727708041667938, + "step": 5009 + }, + { + "epoch": 2.802237762237762, + "grad_norm": 0.477954676402989, + "learning_rate": 2.65729033426112e-07, + "loss": 0.08972786366939545, + "step": 5010 + }, + { + "epoch": 2.802797202797203, + "grad_norm": 0.25695475847028143, + "learning_rate": 2.642406959554655e-07, + "loss": 0.049296434968709946, + "step": 5011 + }, + { + "epoch": 2.8033566433566435, + "grad_norm": 0.648343860283958, + "learning_rate": 2.627564824497253e-07, + "loss": 0.16677223145961761, + "step": 5012 + }, + { + "epoch": 2.803916083916084, + "grad_norm": 0.32922252636848287, + "learning_rate": 2.6127639353758726e-07, + "loss": 0.043198127299547195, + "step": 5013 + }, + { + "epoch": 2.8044755244755244, + "grad_norm": 0.18593675265263768, + "learning_rate": 2.5980042984599994e-07, + "loss": 0.01802062802016735, + "step": 5014 + }, + { + "epoch": 2.805034965034965, + "grad_norm": 0.3435975414327653, + "learning_rate": 2.583285920001677e-07, + "loss": 0.07416543364524841, + "step": 5015 + }, + { + "epoch": 2.8055944055944058, + "grad_norm": 0.6267116584945015, + "learning_rate": 2.568608806235462e-07, + "loss": 0.08955828845500946, + "step": 5016 + }, + { + "epoch": 2.806153846153846, + "grad_norm": 0.3031898002938494, + "learning_rate": 2.553972963378415e-07, + "loss": 0.05864696949720383, + "step": 5017 + }, + { + "epoch": 2.8067132867132867, + "grad_norm": 0.3491119828947996, + "learning_rate": 2.53937839763011e-07, + "loss": 0.12506571412086487, + "step": 5018 + }, + { + "epoch": 2.807272727272727, + "grad_norm": 0.7761357722517344, + "learning_rate": 2.52482511517268e-07, + "loss": 0.09943557530641556, + "step": 5019 + }, + { + "epoch": 2.807832167832168, + "grad_norm": 0.6822418473580225, + "learning_rate": 2.510313122170738e-07, + "loss": 0.19397728145122528, + "step": 5020 + }, + { + "epoch": 2.8083916083916085, + "grad_norm": 0.3374206490016995, + "learning_rate": 2.4958424247713776e-07, + "loss": 0.11500547081232071, + "step": 5021 + }, + { + "epoch": 2.808951048951049, + "grad_norm": 0.2821365962508446, + "learning_rate": 2.481413029104274e-07, + "loss": 0.08655523508787155, + "step": 5022 + }, + { + "epoch": 2.8095104895104894, + "grad_norm": 0.33758019750585594, + "learning_rate": 2.467024941281537e-07, + "loss": 0.025496497750282288, + "step": 5023 + }, + { + "epoch": 2.81006993006993, + "grad_norm": 0.27978230078187755, + "learning_rate": 2.4526781673978263e-07, + "loss": 0.0329664871096611, + "step": 5024 + }, + { + "epoch": 2.810629370629371, + "grad_norm": 0.5819457118518121, + "learning_rate": 2.438372713530279e-07, + "loss": 0.06955914199352264, + "step": 5025 + }, + { + "epoch": 2.8111888111888113, + "grad_norm": 0.3823517869865145, + "learning_rate": 2.424108585738527e-07, + "loss": 0.051360834389925, + "step": 5026 + }, + { + "epoch": 2.8117482517482517, + "grad_norm": 0.2889990044778383, + "learning_rate": 2.409885790064725e-07, + "loss": 0.03358414024114609, + "step": 5027 + }, + { + "epoch": 2.812307692307692, + "grad_norm": 0.4338372129298275, + "learning_rate": 2.3957043325334883e-07, + "loss": 0.06066404655575752, + "step": 5028 + }, + { + "epoch": 2.812867132867133, + "grad_norm": 0.5891857151529228, + "learning_rate": 2.3815642191519107e-07, + "loss": 0.07924894988536835, + "step": 5029 + }, + { + "epoch": 2.8134265734265735, + "grad_norm": 0.4689012119668464, + "learning_rate": 2.3674654559096344e-07, + "loss": 0.10304653644561768, + "step": 5030 + }, + { + "epoch": 2.813986013986014, + "grad_norm": 0.5580362903297925, + "learning_rate": 2.3534080487787158e-07, + "loss": 0.09917331486940384, + "step": 5031 + }, + { + "epoch": 2.8145454545454545, + "grad_norm": 0.481700577510539, + "learning_rate": 2.3393920037137474e-07, + "loss": 0.12496791034936905, + "step": 5032 + }, + { + "epoch": 2.815104895104895, + "grad_norm": 0.6652911670044661, + "learning_rate": 2.3254173266517466e-07, + "loss": 0.13458216190338135, + "step": 5033 + }, + { + "epoch": 2.815664335664336, + "grad_norm": 0.43076114009950484, + "learning_rate": 2.3114840235122783e-07, + "loss": 0.08030325174331665, + "step": 5034 + }, + { + "epoch": 2.8162237762237763, + "grad_norm": 0.4106933760251058, + "learning_rate": 2.2975921001973212e-07, + "loss": 0.08696933090686798, + "step": 5035 + }, + { + "epoch": 2.8167832167832167, + "grad_norm": 0.4154507566779359, + "learning_rate": 2.283741562591324e-07, + "loss": 0.18368548154830933, + "step": 5036 + }, + { + "epoch": 2.817342657342657, + "grad_norm": 0.314104131213745, + "learning_rate": 2.2699324165612824e-07, + "loss": 0.04156312346458435, + "step": 5037 + }, + { + "epoch": 2.817902097902098, + "grad_norm": 0.5882528640705914, + "learning_rate": 2.2561646679565619e-07, + "loss": 0.06862244755029678, + "step": 5038 + }, + { + "epoch": 2.8184615384615386, + "grad_norm": 0.33393791581994003, + "learning_rate": 2.242438322609053e-07, + "loss": 0.045548420399427414, + "step": 5039 + }, + { + "epoch": 2.819020979020979, + "grad_norm": 0.6265466033239815, + "learning_rate": 2.2287533863330824e-07, + "loss": 0.05208326876163483, + "step": 5040 + }, + { + "epoch": 2.8195804195804195, + "grad_norm": 0.3909796454810978, + "learning_rate": 2.2151098649254577e-07, + "loss": 0.04804254695773125, + "step": 5041 + }, + { + "epoch": 2.82013986013986, + "grad_norm": 0.42385764697772543, + "learning_rate": 2.2015077641654225e-07, + "loss": 0.06865841895341873, + "step": 5042 + }, + { + "epoch": 2.820699300699301, + "grad_norm": 0.46240375474549594, + "learning_rate": 2.1879470898146792e-07, + "loss": 0.081064872443676, + "step": 5043 + }, + { + "epoch": 2.8212587412587413, + "grad_norm": 0.3688644746623924, + "learning_rate": 2.1744278476173885e-07, + "loss": 0.04247250407934189, + "step": 5044 + }, + { + "epoch": 2.821818181818182, + "grad_norm": 0.6648619275443681, + "learning_rate": 2.1609500433001583e-07, + "loss": 0.1341962069272995, + "step": 5045 + }, + { + "epoch": 2.8223776223776222, + "grad_norm": 0.4128019414567065, + "learning_rate": 2.1475136825720666e-07, + "loss": 0.056738875806331635, + "step": 5046 + }, + { + "epoch": 2.822937062937063, + "grad_norm": 0.37178378680362356, + "learning_rate": 2.1341187711245713e-07, + "loss": 0.0701540932059288, + "step": 5047 + }, + { + "epoch": 2.8234965034965036, + "grad_norm": 0.6403156324419247, + "learning_rate": 2.1207653146316454e-07, + "loss": 0.18930019438266754, + "step": 5048 + }, + { + "epoch": 2.824055944055944, + "grad_norm": 0.45373028772583235, + "learning_rate": 2.1074533187496638e-07, + "loss": 0.06333623826503754, + "step": 5049 + }, + { + "epoch": 2.8246153846153845, + "grad_norm": 0.36767676242708336, + "learning_rate": 2.0941827891174492e-07, + "loss": 0.05690986290574074, + "step": 5050 + }, + { + "epoch": 2.825174825174825, + "grad_norm": 0.514531442574815, + "learning_rate": 2.080953731356261e-07, + "loss": 0.10526102781295776, + "step": 5051 + }, + { + "epoch": 2.825734265734266, + "grad_norm": 0.12956865464684977, + "learning_rate": 2.0677661510697833e-07, + "loss": 0.0016192694893106818, + "step": 5052 + }, + { + "epoch": 2.8262937062937064, + "grad_norm": 0.6838179977682308, + "learning_rate": 2.0546200538441363e-07, + "loss": 0.17949801683425903, + "step": 5053 + }, + { + "epoch": 2.826853146853147, + "grad_norm": 0.36639706399664235, + "learning_rate": 2.041515445247866e-07, + "loss": 0.03927082195878029, + "step": 5054 + }, + { + "epoch": 2.8274125874125873, + "grad_norm": 0.3948649648789191, + "learning_rate": 2.0284523308319538e-07, + "loss": 0.027906186878681183, + "step": 5055 + }, + { + "epoch": 2.827972027972028, + "grad_norm": 0.6006054285134843, + "learning_rate": 2.015430716129796e-07, + "loss": 0.10151061415672302, + "step": 5056 + }, + { + "epoch": 2.8285314685314686, + "grad_norm": 0.4779716362388702, + "learning_rate": 2.0024506066571915e-07, + "loss": 0.08257726579904556, + "step": 5057 + }, + { + "epoch": 2.829090909090909, + "grad_norm": 0.35130679945298426, + "learning_rate": 1.9895120079124085e-07, + "loss": 0.049539223313331604, + "step": 5058 + }, + { + "epoch": 2.8296503496503496, + "grad_norm": 0.5272403904151606, + "learning_rate": 1.9766149253760858e-07, + "loss": 0.09938731789588928, + "step": 5059 + }, + { + "epoch": 2.83020979020979, + "grad_norm": 0.37737840828432573, + "learning_rate": 1.9637593645112974e-07, + "loss": 0.060706980526447296, + "step": 5060 + }, + { + "epoch": 2.830769230769231, + "grad_norm": 0.48634411048391796, + "learning_rate": 1.9509453307635096e-07, + "loss": 0.0917670726776123, + "step": 5061 + }, + { + "epoch": 2.8313286713286714, + "grad_norm": 0.5253203349071603, + "learning_rate": 1.938172829560614e-07, + "loss": 0.1297025978565216, + "step": 5062 + }, + { + "epoch": 2.831888111888112, + "grad_norm": 0.4623457706561187, + "learning_rate": 1.925441866312927e-07, + "loss": 0.07143554091453552, + "step": 5063 + }, + { + "epoch": 2.8324475524475523, + "grad_norm": 0.3073677591171638, + "learning_rate": 1.9127524464131353e-07, + "loss": 0.019956976175308228, + "step": 5064 + }, + { + "epoch": 2.833006993006993, + "grad_norm": 0.4661428322249231, + "learning_rate": 1.900104575236339e-07, + "loss": 0.06862765550613403, + "step": 5065 + }, + { + "epoch": 2.8335664335664337, + "grad_norm": 0.2400576632752462, + "learning_rate": 1.8874982581400636e-07, + "loss": 0.009724807925522327, + "step": 5066 + }, + { + "epoch": 2.834125874125874, + "grad_norm": 0.5675939431590361, + "learning_rate": 1.874933500464171e-07, + "loss": 0.268549382686615, + "step": 5067 + }, + { + "epoch": 2.8346853146853146, + "grad_norm": 0.4708228984211039, + "learning_rate": 1.862410307531015e-07, + "loss": 0.10982885956764221, + "step": 5068 + }, + { + "epoch": 2.835244755244755, + "grad_norm": 0.2830014565736752, + "learning_rate": 1.8499286846452414e-07, + "loss": 0.08165832608938217, + "step": 5069 + }, + { + "epoch": 2.835804195804196, + "grad_norm": 0.4430977928404783, + "learning_rate": 1.8374886370939538e-07, + "loss": 0.0976344496011734, + "step": 5070 + }, + { + "epoch": 2.8363636363636364, + "grad_norm": 0.5056350442529829, + "learning_rate": 1.825090170146626e-07, + "loss": 0.13575875759124756, + "step": 5071 + }, + { + "epoch": 2.836923076923077, + "grad_norm": 0.41955978797070625, + "learning_rate": 1.812733289055113e-07, + "loss": 0.048558276146650314, + "step": 5072 + }, + { + "epoch": 2.8374825174825173, + "grad_norm": 0.45259569945727673, + "learning_rate": 1.8004179990536607e-07, + "loss": 0.31836220622062683, + "step": 5073 + }, + { + "epoch": 2.8380419580419582, + "grad_norm": 0.39795826745603124, + "learning_rate": 1.7881443053588966e-07, + "loss": 0.06775633245706558, + "step": 5074 + }, + { + "epoch": 2.8386013986013987, + "grad_norm": 0.30253144209574007, + "learning_rate": 1.775912213169828e-07, + "loss": 0.03139274939894676, + "step": 5075 + }, + { + "epoch": 2.839160839160839, + "grad_norm": 0.5658734209484007, + "learning_rate": 1.763721727667844e-07, + "loss": 0.09580185264348984, + "step": 5076 + }, + { + "epoch": 2.8397202797202796, + "grad_norm": 0.3370834785773157, + "learning_rate": 1.751572854016692e-07, + "loss": 0.07367917895317078, + "step": 5077 + }, + { + "epoch": 2.84027972027972, + "grad_norm": 0.46022124423881655, + "learning_rate": 1.7394655973625219e-07, + "loss": 0.06942617893218994, + "step": 5078 + }, + { + "epoch": 2.840839160839161, + "grad_norm": 0.3293263502298244, + "learning_rate": 1.7273999628338311e-07, + "loss": 0.018502244725823402, + "step": 5079 + }, + { + "epoch": 2.8413986013986015, + "grad_norm": 0.39213182375268985, + "learning_rate": 1.7153759555414984e-07, + "loss": 0.035788435488939285, + "step": 5080 + }, + { + "epoch": 2.841958041958042, + "grad_norm": 0.3325613851635554, + "learning_rate": 1.7033935805787715e-07, + "loss": 0.12827466428279877, + "step": 5081 + }, + { + "epoch": 2.8425174825174824, + "grad_norm": 0.4426265928067478, + "learning_rate": 1.6914528430212463e-07, + "loss": 0.09196305274963379, + "step": 5082 + }, + { + "epoch": 2.8430769230769233, + "grad_norm": 0.4209069201228952, + "learning_rate": 1.6795537479268986e-07, + "loss": 0.06387708336114883, + "step": 5083 + }, + { + "epoch": 2.8436363636363637, + "grad_norm": 0.6996468532771095, + "learning_rate": 1.6676963003360636e-07, + "loss": 0.1405351161956787, + "step": 5084 + }, + { + "epoch": 2.844195804195804, + "grad_norm": 0.307219758979532, + "learning_rate": 1.6558805052714234e-07, + "loss": 0.03291900083422661, + "step": 5085 + }, + { + "epoch": 2.8447552447552447, + "grad_norm": 0.4521501746197546, + "learning_rate": 1.644106367738041e-07, + "loss": 0.06944390386343002, + "step": 5086 + }, + { + "epoch": 2.845314685314685, + "grad_norm": 0.3842760165429014, + "learning_rate": 1.6323738927232824e-07, + "loss": 0.06560652703046799, + "step": 5087 + }, + { + "epoch": 2.845874125874126, + "grad_norm": 0.4836884842552966, + "learning_rate": 1.620683085196939e-07, + "loss": 0.09246764332056046, + "step": 5088 + }, + { + "epoch": 2.8464335664335665, + "grad_norm": 0.3387552854406162, + "learning_rate": 1.609033950111083e-07, + "loss": 0.08510936796665192, + "step": 5089 + }, + { + "epoch": 2.846993006993007, + "grad_norm": 0.6836084114191575, + "learning_rate": 1.597426492400189e-07, + "loss": 0.13489529490470886, + "step": 5090 + }, + { + "epoch": 2.8475524475524474, + "grad_norm": 0.4253636045482067, + "learning_rate": 1.5858607169810468e-07, + "loss": 0.0592874176800251, + "step": 5091 + }, + { + "epoch": 2.8481118881118883, + "grad_norm": 0.2164566697046395, + "learning_rate": 1.5743366287527816e-07, + "loss": 0.049627311527729034, + "step": 5092 + }, + { + "epoch": 2.8486713286713288, + "grad_norm": 0.4637991098694843, + "learning_rate": 1.5628542325968887e-07, + "loss": 0.11759379506111145, + "step": 5093 + }, + { + "epoch": 2.8492307692307692, + "grad_norm": 0.3251116488172615, + "learning_rate": 1.5514135333771996e-07, + "loss": 0.049187518656253815, + "step": 5094 + }, + { + "epoch": 2.8497902097902097, + "grad_norm": 0.31198274705165535, + "learning_rate": 1.54001453593986e-07, + "loss": 0.03734589368104935, + "step": 5095 + }, + { + "epoch": 2.85034965034965, + "grad_norm": 0.42395691035354255, + "learning_rate": 1.528657245113352e-07, + "loss": 0.07761257886886597, + "step": 5096 + }, + { + "epoch": 2.850909090909091, + "grad_norm": 0.6310466021513721, + "learning_rate": 1.5173416657085271e-07, + "loss": 0.16581541299819946, + "step": 5097 + }, + { + "epoch": 2.8514685314685315, + "grad_norm": 0.5424352580345334, + "learning_rate": 1.5060678025185404e-07, + "loss": 0.09356231242418289, + "step": 5098 + }, + { + "epoch": 2.852027972027972, + "grad_norm": 0.3719214609014081, + "learning_rate": 1.494835660318883e-07, + "loss": 0.040314048528671265, + "step": 5099 + }, + { + "epoch": 2.8525874125874124, + "grad_norm": 0.5979427034817748, + "learning_rate": 1.4836452438673599e-07, + "loss": 0.0820893794298172, + "step": 5100 + }, + { + "epoch": 2.8531468531468533, + "grad_norm": 0.3410179304589936, + "learning_rate": 1.4724965579041239e-07, + "loss": 0.05582234635949135, + "step": 5101 + }, + { + "epoch": 2.853706293706294, + "grad_norm": 0.28268502233086734, + "learning_rate": 1.461389607151642e-07, + "loss": 0.011546442285180092, + "step": 5102 + }, + { + "epoch": 2.8542657342657343, + "grad_norm": 0.8596136083437055, + "learning_rate": 1.450324396314684e-07, + "loss": 0.15707063674926758, + "step": 5103 + }, + { + "epoch": 2.8548251748251747, + "grad_norm": 0.33367987074221866, + "learning_rate": 1.4393009300803785e-07, + "loss": 0.03148286044597626, + "step": 5104 + }, + { + "epoch": 2.855384615384615, + "grad_norm": 0.3395745483237218, + "learning_rate": 1.4283192131181456e-07, + "loss": 0.06608149409294128, + "step": 5105 + }, + { + "epoch": 2.855944055944056, + "grad_norm": 0.46960636552840423, + "learning_rate": 1.417379250079709e-07, + "loss": 0.05685903877019882, + "step": 5106 + }, + { + "epoch": 2.8565034965034966, + "grad_norm": 0.4401537537424136, + "learning_rate": 1.4064810455991507e-07, + "loss": 0.08568979799747467, + "step": 5107 + }, + { + "epoch": 2.857062937062937, + "grad_norm": 0.19463980063353842, + "learning_rate": 1.3956246042928002e-07, + "loss": 0.010590219870209694, + "step": 5108 + }, + { + "epoch": 2.8576223776223775, + "grad_norm": 0.38129771908492316, + "learning_rate": 1.3848099307593455e-07, + "loss": 0.04023498296737671, + "step": 5109 + }, + { + "epoch": 2.8581818181818184, + "grad_norm": 0.4867190073100135, + "learning_rate": 1.374037029579789e-07, + "loss": 0.0387575663626194, + "step": 5110 + }, + { + "epoch": 2.858741258741259, + "grad_norm": 0.27434391021769194, + "learning_rate": 1.3633059053173804e-07, + "loss": 0.06853128969669342, + "step": 5111 + }, + { + "epoch": 2.8593006993006993, + "grad_norm": 0.3941889310817446, + "learning_rate": 1.352616562517739e-07, + "loss": 0.07619741559028625, + "step": 5112 + }, + { + "epoch": 2.8598601398601398, + "grad_norm": 0.414632406105163, + "learning_rate": 1.3419690057087432e-07, + "loss": 0.08275170624256134, + "step": 5113 + }, + { + "epoch": 2.86041958041958, + "grad_norm": 0.5660698705478232, + "learning_rate": 1.331363239400596e-07, + "loss": 0.20167924463748932, + "step": 5114 + }, + { + "epoch": 2.860979020979021, + "grad_norm": 0.7632408540945325, + "learning_rate": 1.320799268085793e-07, + "loss": 0.18678340315818787, + "step": 5115 + }, + { + "epoch": 2.8615384615384616, + "grad_norm": 0.47237268182042624, + "learning_rate": 1.3102770962390875e-07, + "loss": 0.0465087890625, + "step": 5116 + }, + { + "epoch": 2.862097902097902, + "grad_norm": 0.44668665526330387, + "learning_rate": 1.299796728317604e-07, + "loss": 0.059813205152750015, + "step": 5117 + }, + { + "epoch": 2.8626573426573425, + "grad_norm": 0.763192721430278, + "learning_rate": 1.2893581687606903e-07, + "loss": 0.24804198741912842, + "step": 5118 + }, + { + "epoch": 2.8632167832167834, + "grad_norm": 0.6081076031794271, + "learning_rate": 1.2789614219900103e-07, + "loss": 0.0936436876654625, + "step": 5119 + }, + { + "epoch": 2.863776223776224, + "grad_norm": 0.4960518595877114, + "learning_rate": 1.2686064924095299e-07, + "loss": 0.08598095178604126, + "step": 5120 + }, + { + "epoch": 2.8643356643356643, + "grad_norm": 0.5216456598258115, + "learning_rate": 1.2582933844054733e-07, + "loss": 0.13691863417625427, + "step": 5121 + }, + { + "epoch": 2.864895104895105, + "grad_norm": 0.2695036420822719, + "learning_rate": 1.24802210234638e-07, + "loss": 0.05046981945633888, + "step": 5122 + }, + { + "epoch": 2.8654545454545453, + "grad_norm": 0.30311613663296416, + "learning_rate": 1.2377926505830363e-07, + "loss": 0.05974654480814934, + "step": 5123 + }, + { + "epoch": 2.866013986013986, + "grad_norm": 0.5921983564577964, + "learning_rate": 1.2276050334485424e-07, + "loss": 0.09566514194011688, + "step": 5124 + }, + { + "epoch": 2.8665734265734266, + "grad_norm": 0.39976600685417785, + "learning_rate": 1.2174592552582686e-07, + "loss": 0.07634823769330978, + "step": 5125 + }, + { + "epoch": 2.867132867132867, + "grad_norm": 0.4575814910178799, + "learning_rate": 1.2073553203098552e-07, + "loss": 0.099985271692276, + "step": 5126 + }, + { + "epoch": 2.8676923076923075, + "grad_norm": 0.5513479896863783, + "learning_rate": 1.197293232883212e-07, + "loss": 0.09185055643320084, + "step": 5127 + }, + { + "epoch": 2.8682517482517484, + "grad_norm": 0.5163478464733945, + "learning_rate": 1.1872729972405517e-07, + "loss": 0.071977898478508, + "step": 5128 + }, + { + "epoch": 2.868811188811189, + "grad_norm": 0.4640138704693773, + "learning_rate": 1.1772946176263345e-07, + "loss": 0.08545494079589844, + "step": 5129 + }, + { + "epoch": 2.8693706293706294, + "grad_norm": 0.452131742066645, + "learning_rate": 1.1673580982672905e-07, + "loss": 0.04457031190395355, + "step": 5130 + }, + { + "epoch": 2.86993006993007, + "grad_norm": 0.5145232186117992, + "learning_rate": 1.1574634433724307e-07, + "loss": 0.09940159320831299, + "step": 5131 + }, + { + "epoch": 2.8704895104895103, + "grad_norm": 0.570950024878576, + "learning_rate": 1.1476106571330247e-07, + "loss": 0.12142953276634216, + "step": 5132 + }, + { + "epoch": 2.871048951048951, + "grad_norm": 0.4728111798382547, + "learning_rate": 1.1377997437226008e-07, + "loss": 0.09178759157657623, + "step": 5133 + }, + { + "epoch": 2.8716083916083917, + "grad_norm": 0.30554593338058966, + "learning_rate": 1.1280307072969677e-07, + "loss": 0.045527152717113495, + "step": 5134 + }, + { + "epoch": 2.872167832167832, + "grad_norm": 0.6320214789929364, + "learning_rate": 1.118303551994182e-07, + "loss": 0.09320954978466034, + "step": 5135 + }, + { + "epoch": 2.8727272727272726, + "grad_norm": 0.2533027120497144, + "learning_rate": 1.1086182819345703e-07, + "loss": 0.027723446488380432, + "step": 5136 + }, + { + "epoch": 2.8732867132867135, + "grad_norm": 0.6205580304206739, + "learning_rate": 1.0989749012207063e-07, + "loss": 0.19189725816249847, + "step": 5137 + }, + { + "epoch": 2.873846153846154, + "grad_norm": 0.32777354038534123, + "learning_rate": 1.0893734139374223e-07, + "loss": 0.041648589074611664, + "step": 5138 + }, + { + "epoch": 2.8744055944055944, + "grad_norm": 0.437959838773509, + "learning_rate": 1.0798138241518097e-07, + "loss": 0.05024919658899307, + "step": 5139 + }, + { + "epoch": 2.874965034965035, + "grad_norm": 0.30686626890747043, + "learning_rate": 1.0702961359132069e-07, + "loss": 0.05095798522233963, + "step": 5140 + }, + { + "epoch": 2.8755244755244753, + "grad_norm": 0.3793997627904224, + "learning_rate": 1.0608203532532224e-07, + "loss": 0.03731197863817215, + "step": 5141 + }, + { + "epoch": 2.8760839160839162, + "grad_norm": 0.7712814835315371, + "learning_rate": 1.0513864801856677e-07, + "loss": 0.24399888515472412, + "step": 5142 + }, + { + "epoch": 2.8766433566433567, + "grad_norm": 0.43880865396389873, + "learning_rate": 1.0419945207066573e-07, + "loss": 0.04520170018076897, + "step": 5143 + }, + { + "epoch": 2.877202797202797, + "grad_norm": 0.49960331626338855, + "learning_rate": 1.03264447879452e-07, + "loss": 0.03667888790369034, + "step": 5144 + }, + { + "epoch": 2.8777622377622376, + "grad_norm": 0.37469096777403177, + "learning_rate": 1.0233363584098211e-07, + "loss": 0.05186565965414047, + "step": 5145 + }, + { + "epoch": 2.8783216783216785, + "grad_norm": 0.414147625813024, + "learning_rate": 1.0140701634953954e-07, + "loss": 0.07279538363218307, + "step": 5146 + }, + { + "epoch": 2.878881118881119, + "grad_norm": 0.34584620825685436, + "learning_rate": 1.0048458979763031e-07, + "loss": 0.05483987182378769, + "step": 5147 + }, + { + "epoch": 2.8794405594405594, + "grad_norm": 0.5298945493014151, + "learning_rate": 9.956635657598524e-08, + "loss": 0.08227822184562683, + "step": 5148 + }, + { + "epoch": 2.88, + "grad_norm": 0.41345754580354727, + "learning_rate": 9.86523170735576e-08, + "loss": 0.03304259851574898, + "step": 5149 + }, + { + "epoch": 2.8805594405594404, + "grad_norm": 0.5272871111833105, + "learning_rate": 9.774247167752326e-08, + "loss": 0.10360705852508545, + "step": 5150 + }, + { + "epoch": 2.8811188811188813, + "grad_norm": 0.5329993576556061, + "learning_rate": 9.683682077328616e-08, + "loss": 0.06976587325334549, + "step": 5151 + }, + { + "epoch": 2.8816783216783217, + "grad_norm": 0.7168366752741211, + "learning_rate": 9.59353647444683e-08, + "loss": 0.18481668829917908, + "step": 5152 + }, + { + "epoch": 2.882237762237762, + "grad_norm": 0.5919125267054981, + "learning_rate": 9.503810397291758e-08, + "loss": 0.11155405640602112, + "step": 5153 + }, + { + "epoch": 2.8827972027972026, + "grad_norm": 0.5044101054133162, + "learning_rate": 9.414503883870552e-08, + "loss": 0.16135969758033752, + "step": 5154 + }, + { + "epoch": 2.8833566433566435, + "grad_norm": 0.49943756329844124, + "learning_rate": 9.325616972012175e-08, + "loss": 0.13592611253261566, + "step": 5155 + }, + { + "epoch": 2.883916083916084, + "grad_norm": 0.5151518316575652, + "learning_rate": 9.237149699368508e-08, + "loss": 0.08500874787569046, + "step": 5156 + }, + { + "epoch": 2.8844755244755245, + "grad_norm": 0.6062665186610082, + "learning_rate": 9.149102103413243e-08, + "loss": 0.1580979824066162, + "step": 5157 + }, + { + "epoch": 2.885034965034965, + "grad_norm": 0.6659822369157594, + "learning_rate": 9.061474221442324e-08, + "loss": 0.12007184326648712, + "step": 5158 + }, + { + "epoch": 2.8855944055944054, + "grad_norm": 0.3879087886860332, + "learning_rate": 8.97426609057428e-08, + "loss": 0.018520301207900047, + "step": 5159 + }, + { + "epoch": 2.8861538461538463, + "grad_norm": 0.42236993168179815, + "learning_rate": 8.887477747749118e-08, + "loss": 0.060542263090610504, + "step": 5160 + }, + { + "epoch": 2.8867132867132868, + "grad_norm": 0.46101360448694034, + "learning_rate": 8.801109229729877e-08, + "loss": 0.08037333935499191, + "step": 5161 + }, + { + "epoch": 2.887272727272727, + "grad_norm": 0.45635150806973857, + "learning_rate": 8.715160573101067e-08, + "loss": 0.059672582894563675, + "step": 5162 + }, + { + "epoch": 2.8878321678321677, + "grad_norm": 0.5789040790344269, + "learning_rate": 8.62963181426979e-08, + "loss": 0.0336616113781929, + "step": 5163 + }, + { + "epoch": 2.8883916083916086, + "grad_norm": 0.5356040682114503, + "learning_rate": 8.544522989465064e-08, + "loss": 0.0680554062128067, + "step": 5164 + }, + { + "epoch": 2.888951048951049, + "grad_norm": 0.559150974740085, + "learning_rate": 8.459834134738054e-08, + "loss": 0.12276100367307663, + "step": 5165 + }, + { + "epoch": 2.8895104895104895, + "grad_norm": 0.5795254450515255, + "learning_rate": 8.375565285962173e-08, + "loss": 0.10284769535064697, + "step": 5166 + }, + { + "epoch": 2.89006993006993, + "grad_norm": 0.6120737862015375, + "learning_rate": 8.291716478832757e-08, + "loss": 0.16166862845420837, + "step": 5167 + }, + { + "epoch": 2.8906293706293704, + "grad_norm": 0.3489291250262622, + "learning_rate": 8.208287748867172e-08, + "loss": 0.07191438972949982, + "step": 5168 + }, + { + "epoch": 2.8911888111888113, + "grad_norm": 0.5109187177775074, + "learning_rate": 8.125279131405151e-08, + "loss": 0.2645817697048187, + "step": 5169 + }, + { + "epoch": 2.891748251748252, + "grad_norm": 0.6001480215809134, + "learning_rate": 8.042690661608121e-08, + "loss": 0.07099726796150208, + "step": 5170 + }, + { + "epoch": 2.8923076923076922, + "grad_norm": 0.2876309661902357, + "learning_rate": 7.960522374459767e-08, + "loss": 0.014326678588986397, + "step": 5171 + }, + { + "epoch": 2.8928671328671327, + "grad_norm": 0.296692369382386, + "learning_rate": 7.87877430476569e-08, + "loss": 0.07095473259687424, + "step": 5172 + }, + { + "epoch": 2.8934265734265736, + "grad_norm": 0.3854042926641839, + "learning_rate": 7.797446487153415e-08, + "loss": 0.07701880484819412, + "step": 5173 + }, + { + "epoch": 2.893986013986014, + "grad_norm": 0.3044917963230899, + "learning_rate": 7.716538956072606e-08, + "loss": 0.042147181928157806, + "step": 5174 + }, + { + "epoch": 2.8945454545454545, + "grad_norm": 0.234902190812887, + "learning_rate": 7.636051745794959e-08, + "loss": 0.010980570688843727, + "step": 5175 + }, + { + "epoch": 2.895104895104895, + "grad_norm": 0.39300706785656325, + "learning_rate": 7.55598489041387e-08, + "loss": 0.10655295848846436, + "step": 5176 + }, + { + "epoch": 2.8956643356643355, + "grad_norm": 0.3947027344078852, + "learning_rate": 7.476338423844876e-08, + "loss": 0.033465802669525146, + "step": 5177 + }, + { + "epoch": 2.8962237762237764, + "grad_norm": 0.43819250623991945, + "learning_rate": 7.397112379825322e-08, + "loss": 0.050919309258461, + "step": 5178 + }, + { + "epoch": 2.896783216783217, + "grad_norm": 0.5789011445298275, + "learning_rate": 7.318306791914587e-08, + "loss": 0.1203484758734703, + "step": 5179 + }, + { + "epoch": 2.8973426573426573, + "grad_norm": 0.7525293585723698, + "learning_rate": 7.239921693493857e-08, + "loss": 0.0619923397898674, + "step": 5180 + }, + { + "epoch": 2.8979020979020977, + "grad_norm": 0.5604286104378502, + "learning_rate": 7.161957117766128e-08, + "loss": 0.10916341096162796, + "step": 5181 + }, + { + "epoch": 2.8984615384615386, + "grad_norm": 0.4693532684620626, + "learning_rate": 7.084413097756537e-08, + "loss": 0.1473524123430252, + "step": 5182 + }, + { + "epoch": 2.899020979020979, + "grad_norm": 0.2388544635275206, + "learning_rate": 7.007289666311923e-08, + "loss": 0.011255375109612942, + "step": 5183 + }, + { + "epoch": 2.8995804195804196, + "grad_norm": 0.40197618986749517, + "learning_rate": 6.930586856100818e-08, + "loss": 0.09858214110136032, + "step": 5184 + }, + { + "epoch": 2.90013986013986, + "grad_norm": 0.45329705460812963, + "learning_rate": 6.854304699613789e-08, + "loss": 0.1014171689748764, + "step": 5185 + }, + { + "epoch": 2.9006993006993005, + "grad_norm": 0.35689314954641876, + "learning_rate": 6.778443229163101e-08, + "loss": 0.10433556139469147, + "step": 5186 + }, + { + "epoch": 2.9012587412587414, + "grad_norm": 0.38747150543683034, + "learning_rate": 6.703002476883046e-08, + "loss": 0.03780497610569, + "step": 5187 + }, + { + "epoch": 2.901818181818182, + "grad_norm": 0.44023363634286367, + "learning_rate": 6.627982474729289e-08, + "loss": 0.04920152202248573, + "step": 5188 + }, + { + "epoch": 2.9023776223776223, + "grad_norm": 0.3124126813473744, + "learning_rate": 6.55338325447974e-08, + "loss": 0.0343848317861557, + "step": 5189 + }, + { + "epoch": 2.9029370629370628, + "grad_norm": 0.7598110334879455, + "learning_rate": 6.479204847733678e-08, + "loss": 0.20481495559215546, + "step": 5190 + }, + { + "epoch": 2.9034965034965037, + "grad_norm": 0.5427188786642334, + "learning_rate": 6.405447285912302e-08, + "loss": 0.10618521273136139, + "step": 5191 + }, + { + "epoch": 2.904055944055944, + "grad_norm": 0.6786327920838435, + "learning_rate": 6.332110600258623e-08, + "loss": 0.14831744134426117, + "step": 5192 + }, + { + "epoch": 2.9046153846153846, + "grad_norm": 0.36722854679116707, + "learning_rate": 6.259194821837345e-08, + "loss": 0.13791592419147491, + "step": 5193 + }, + { + "epoch": 2.905174825174825, + "grad_norm": 0.7253962313708597, + "learning_rate": 6.186699981534538e-08, + "loss": 0.21965214610099792, + "step": 5194 + }, + { + "epoch": 2.9057342657342655, + "grad_norm": 0.3388768723510042, + "learning_rate": 6.114626110058641e-08, + "loss": 0.02646435797214508, + "step": 5195 + }, + { + "epoch": 2.9062937062937064, + "grad_norm": 0.5591367023302044, + "learning_rate": 6.04297323793901e-08, + "loss": 0.06938068568706512, + "step": 5196 + }, + { + "epoch": 2.906853146853147, + "grad_norm": 0.2765101586001875, + "learning_rate": 5.971741395527364e-08, + "loss": 0.019158296287059784, + "step": 5197 + }, + { + "epoch": 2.9074125874125873, + "grad_norm": 0.256402343998391, + "learning_rate": 5.900930612996569e-08, + "loss": 0.031087584793567657, + "step": 5198 + }, + { + "epoch": 2.907972027972028, + "grad_norm": 1.4819066926379263, + "learning_rate": 5.830540920341299e-08, + "loss": 0.22102338075637817, + "step": 5199 + }, + { + "epoch": 2.9085314685314687, + "grad_norm": 0.44788436125096837, + "learning_rate": 5.760572347378035e-08, + "loss": 0.12027538567781448, + "step": 5200 + }, + { + "epoch": 2.909090909090909, + "grad_norm": 0.460102047256179, + "learning_rate": 5.691024923744515e-08, + "loss": 0.042640320956707, + "step": 5201 + }, + { + "epoch": 2.9096503496503496, + "grad_norm": 0.32013296207056596, + "learning_rate": 5.621898678900506e-08, + "loss": 0.02074226550757885, + "step": 5202 + }, + { + "epoch": 2.91020979020979, + "grad_norm": 0.6953902685076467, + "learning_rate": 5.5531936421271416e-08, + "loss": 0.13586930930614471, + "step": 5203 + }, + { + "epoch": 2.9107692307692306, + "grad_norm": 0.6448302208320131, + "learning_rate": 5.484909842527031e-08, + "loss": 0.11441251635551453, + "step": 5204 + }, + { + "epoch": 2.9113286713286715, + "grad_norm": 0.06712070138858497, + "learning_rate": 5.417047309024481e-08, + "loss": 0.001350384671241045, + "step": 5205 + }, + { + "epoch": 2.911888111888112, + "grad_norm": 0.501683465886353, + "learning_rate": 5.349606070365498e-08, + "loss": 0.1591944396495819, + "step": 5206 + }, + { + "epoch": 2.9124475524475524, + "grad_norm": 0.395046249117763, + "learning_rate": 5.282586155117231e-08, + "loss": 0.13842764496803284, + "step": 5207 + }, + { + "epoch": 2.9130069930069933, + "grad_norm": 0.42445750324284026, + "learning_rate": 5.2159875916688584e-08, + "loss": 0.06949430704116821, + "step": 5208 + }, + { + "epoch": 2.9135664335664337, + "grad_norm": 0.4545372459852334, + "learning_rate": 5.149810408230704e-08, + "loss": 0.05434008687734604, + "step": 5209 + }, + { + "epoch": 2.914125874125874, + "grad_norm": 0.899628699687208, + "learning_rate": 5.084054632834678e-08, + "loss": 0.2237141728401184, + "step": 5210 + }, + { + "epoch": 2.9146853146853147, + "grad_norm": 0.250048224926423, + "learning_rate": 5.0187202933343894e-08, + "loss": 0.013265509158372879, + "step": 5211 + }, + { + "epoch": 2.915244755244755, + "grad_norm": 0.7183754812848965, + "learning_rate": 4.953807417404588e-08, + "loss": 0.10479851067066193, + "step": 5212 + }, + { + "epoch": 2.9158041958041956, + "grad_norm": 0.4716153129537847, + "learning_rate": 4.8893160325418357e-08, + "loss": 0.07117408514022827, + "step": 5213 + }, + { + "epoch": 2.9163636363636365, + "grad_norm": 0.29638503939382355, + "learning_rate": 4.825246166063835e-08, + "loss": 0.0591374896466732, + "step": 5214 + }, + { + "epoch": 2.916923076923077, + "grad_norm": 0.4626927251327847, + "learning_rate": 4.761597845110211e-08, + "loss": 0.07768068462610245, + "step": 5215 + }, + { + "epoch": 2.9174825174825174, + "grad_norm": 0.16823797800098889, + "learning_rate": 4.698371096641286e-08, + "loss": 0.005802792031317949, + "step": 5216 + }, + { + "epoch": 2.9180419580419583, + "grad_norm": 0.4913667292346074, + "learning_rate": 4.6355659474396355e-08, + "loss": 0.011472095735371113, + "step": 5217 + }, + { + "epoch": 2.9186013986013988, + "grad_norm": 0.3764856313575433, + "learning_rate": 4.573182424108647e-08, + "loss": 0.05923731252551079, + "step": 5218 + }, + { + "epoch": 2.9191608391608392, + "grad_norm": 0.255869098302735, + "learning_rate": 4.511220553073403e-08, + "loss": 0.02489723637700081, + "step": 5219 + }, + { + "epoch": 2.9197202797202797, + "grad_norm": 0.2945950171041178, + "learning_rate": 4.449680360580244e-08, + "loss": 0.05929826572537422, + "step": 5220 + }, + { + "epoch": 2.92027972027972, + "grad_norm": 0.5814294126098861, + "learning_rate": 4.3885618726969836e-08, + "loss": 0.10521118342876434, + "step": 5221 + }, + { + "epoch": 2.9208391608391606, + "grad_norm": 0.36442917237945743, + "learning_rate": 4.327865115312802e-08, + "loss": 0.08411519229412079, + "step": 5222 + }, + { + "epoch": 2.9213986013986015, + "grad_norm": 0.3735906944772255, + "learning_rate": 4.2675901141380204e-08, + "loss": 0.09228894114494324, + "step": 5223 + }, + { + "epoch": 2.921958041958042, + "grad_norm": 0.46767179051699004, + "learning_rate": 4.2077368947047724e-08, + "loss": 0.04566708207130432, + "step": 5224 + }, + { + "epoch": 2.9225174825174824, + "grad_norm": 0.6337444068422258, + "learning_rate": 4.1483054823661104e-08, + "loss": 0.14249449968338013, + "step": 5225 + }, + { + "epoch": 2.9230769230769234, + "grad_norm": 0.5748562186791786, + "learning_rate": 4.089295902296453e-08, + "loss": 0.11930012702941895, + "step": 5226 + }, + { + "epoch": 2.923636363636364, + "grad_norm": 1.06400635443598, + "learning_rate": 4.030708179491805e-08, + "loss": 0.15889689326286316, + "step": 5227 + }, + { + "epoch": 2.9241958041958043, + "grad_norm": 0.7815887728642997, + "learning_rate": 3.972542338769314e-08, + "loss": 0.14439088106155396, + "step": 5228 + }, + { + "epoch": 2.9247552447552447, + "grad_norm": 0.4644692107291607, + "learning_rate": 3.9147984047672724e-08, + "loss": 0.10586054623126984, + "step": 5229 + }, + { + "epoch": 2.925314685314685, + "grad_norm": 0.4069811297111356, + "learning_rate": 3.857476401945337e-08, + "loss": 0.05668766051530838, + "step": 5230 + }, + { + "epoch": 2.9258741258741257, + "grad_norm": 0.40140868433162213, + "learning_rate": 3.8005763545848615e-08, + "loss": 0.057426631450653076, + "step": 5231 + }, + { + "epoch": 2.9264335664335666, + "grad_norm": 0.5850376177380897, + "learning_rate": 3.74409828678779e-08, + "loss": 0.2849140465259552, + "step": 5232 + }, + { + "epoch": 2.926993006993007, + "grad_norm": 0.39766301876417837, + "learning_rate": 3.6880422224776503e-08, + "loss": 0.3134947121143341, + "step": 5233 + }, + { + "epoch": 2.9275524475524475, + "grad_norm": 0.3244723354261282, + "learning_rate": 3.63240818539945e-08, + "loss": 0.031861674040555954, + "step": 5234 + }, + { + "epoch": 2.9281118881118884, + "grad_norm": 0.4633241875263044, + "learning_rate": 3.577196199118893e-08, + "loss": 0.058092888444662094, + "step": 5235 + }, + { + "epoch": 2.928671328671329, + "grad_norm": 0.6355316080544308, + "learning_rate": 3.522406287023272e-08, + "loss": 0.20722651481628418, + "step": 5236 + }, + { + "epoch": 2.9292307692307693, + "grad_norm": 0.49758277924496264, + "learning_rate": 3.4680384723211335e-08, + "loss": 0.11399199068546295, + "step": 5237 + }, + { + "epoch": 2.9297902097902098, + "grad_norm": 0.5004412451177719, + "learning_rate": 3.414092778041944e-08, + "loss": 0.1043810248374939, + "step": 5238 + }, + { + "epoch": 2.9303496503496502, + "grad_norm": 0.4630287422494458, + "learning_rate": 3.360569227036759e-08, + "loss": 0.14572879672050476, + "step": 5239 + }, + { + "epoch": 2.9309090909090907, + "grad_norm": 0.3966348257310289, + "learning_rate": 3.3074678419773296e-08, + "loss": 0.052745770663022995, + "step": 5240 + }, + { + "epoch": 2.9314685314685316, + "grad_norm": 0.3435969368711804, + "learning_rate": 3.254788645356999e-08, + "loss": 0.1724082976579666, + "step": 5241 + }, + { + "epoch": 2.932027972027972, + "grad_norm": 0.5569069406137491, + "learning_rate": 3.202531659490138e-08, + "loss": 0.10703390091657639, + "step": 5242 + }, + { + "epoch": 2.9325874125874125, + "grad_norm": 0.587067604389367, + "learning_rate": 3.150696906512041e-08, + "loss": 0.17900709807872772, + "step": 5243 + }, + { + "epoch": 2.9331468531468534, + "grad_norm": 0.5742460770122021, + "learning_rate": 3.099284408379588e-08, + "loss": 0.08107936382293701, + "step": 5244 + }, + { + "epoch": 2.933706293706294, + "grad_norm": 0.2577019612944207, + "learning_rate": 3.048294186870582e-08, + "loss": 0.02277417853474617, + "step": 5245 + }, + { + "epoch": 2.9342657342657343, + "grad_norm": 0.6200831544507888, + "learning_rate": 2.9977262635837447e-08, + "loss": 0.16745372116565704, + "step": 5246 + }, + { + "epoch": 2.934825174825175, + "grad_norm": 0.29807440145424047, + "learning_rate": 2.9475806599392753e-08, + "loss": 0.025397373363375664, + "step": 5247 + }, + { + "epoch": 2.9353846153846153, + "grad_norm": 0.38001771342332663, + "learning_rate": 2.897857397178294e-08, + "loss": 0.07618074864149094, + "step": 5248 + }, + { + "epoch": 2.9359440559440557, + "grad_norm": 0.5198962622969924, + "learning_rate": 2.848556496362953e-08, + "loss": 0.09367871284484863, + "step": 5249 + }, + { + "epoch": 2.9365034965034966, + "grad_norm": 0.5105154900985371, + "learning_rate": 2.799677978376769e-08, + "loss": 0.10746271163225174, + "step": 5250 + }, + { + "epoch": 2.937062937062937, + "grad_norm": 0.463667428618098, + "learning_rate": 2.7512218639240696e-08, + "loss": 0.09788582473993301, + "step": 5251 + }, + { + "epoch": 2.9376223776223775, + "grad_norm": 0.40652222874290644, + "learning_rate": 2.7031881735303245e-08, + "loss": 0.05372265353798866, + "step": 5252 + }, + { + "epoch": 2.9381818181818184, + "grad_norm": 0.6242734642546979, + "learning_rate": 2.6555769275420362e-08, + "loss": 0.14616882801055908, + "step": 5253 + }, + { + "epoch": 2.938741258741259, + "grad_norm": 0.48711681647482685, + "learning_rate": 2.6083881461270723e-08, + "loss": 0.3291112184524536, + "step": 5254 + }, + { + "epoch": 2.9393006993006994, + "grad_norm": 0.569199732476369, + "learning_rate": 2.5616218492738877e-08, + "loss": 0.2190902978181839, + "step": 5255 + }, + { + "epoch": 2.93986013986014, + "grad_norm": 0.3436579162970652, + "learning_rate": 2.515278056792303e-08, + "loss": 0.0288071446120739, + "step": 5256 + }, + { + "epoch": 2.9404195804195803, + "grad_norm": 0.5658712807077247, + "learning_rate": 2.4693567883129488e-08, + "loss": 0.07421399652957916, + "step": 5257 + }, + { + "epoch": 2.9409790209790208, + "grad_norm": 0.44832637060250485, + "learning_rate": 2.4238580632875984e-08, + "loss": 0.07199421525001526, + "step": 5258 + }, + { + "epoch": 2.9415384615384617, + "grad_norm": 0.5308044709357113, + "learning_rate": 2.378781900989169e-08, + "loss": 0.1259925663471222, + "step": 5259 + }, + { + "epoch": 2.942097902097902, + "grad_norm": 0.3973490413325099, + "learning_rate": 2.3341283205113863e-08, + "loss": 0.03300304710865021, + "step": 5260 + }, + { + "epoch": 2.9426573426573426, + "grad_norm": 0.30048873650261504, + "learning_rate": 2.2898973407688985e-08, + "loss": 0.062602199614048, + "step": 5261 + }, + { + "epoch": 2.9432167832167835, + "grad_norm": 0.45918443845308216, + "learning_rate": 2.2460889804977182e-08, + "loss": 0.072819784283638, + "step": 5262 + }, + { + "epoch": 2.943776223776224, + "grad_norm": 0.435883925554033, + "learning_rate": 2.202703258254446e-08, + "loss": 0.062079332768917084, + "step": 5263 + }, + { + "epoch": 2.9443356643356644, + "grad_norm": 0.262576622289466, + "learning_rate": 2.159740192416826e-08, + "loss": 0.017300274223089218, + "step": 5264 + }, + { + "epoch": 2.944895104895105, + "grad_norm": 0.24671541086933982, + "learning_rate": 2.1171998011836338e-08, + "loss": 0.01414365042001009, + "step": 5265 + }, + { + "epoch": 2.9454545454545453, + "grad_norm": 0.46713448396393925, + "learning_rate": 2.075082102574344e-08, + "loss": 0.1340750902891159, + "step": 5266 + }, + { + "epoch": 2.946013986013986, + "grad_norm": 0.28902716638531867, + "learning_rate": 2.0333871144297966e-08, + "loss": 0.020536599680781364, + "step": 5267 + }, + { + "epoch": 2.9465734265734267, + "grad_norm": 0.5041414640897858, + "learning_rate": 1.992114854411531e-08, + "loss": 0.08401452004909515, + "step": 5268 + }, + { + "epoch": 2.947132867132867, + "grad_norm": 0.3143827736971183, + "learning_rate": 1.9512653400018955e-08, + "loss": 0.0364210270345211, + "step": 5269 + }, + { + "epoch": 2.9476923076923076, + "grad_norm": 0.27316635141260964, + "learning_rate": 1.9108385885043824e-08, + "loss": 0.013481775298714638, + "step": 5270 + }, + { + "epoch": 2.9482517482517485, + "grad_norm": 0.33207200013350363, + "learning_rate": 1.8708346170434043e-08, + "loss": 0.023128246888518333, + "step": 5271 + }, + { + "epoch": 2.948811188811189, + "grad_norm": 0.20025417426846456, + "learning_rate": 1.8312534425640737e-08, + "loss": 0.008567575365304947, + "step": 5272 + }, + { + "epoch": 2.9493706293706294, + "grad_norm": 0.3459382699355651, + "learning_rate": 1.7920950818327566e-08, + "loss": 0.07446427643299103, + "step": 5273 + }, + { + "epoch": 2.94993006993007, + "grad_norm": 0.5972552479977916, + "learning_rate": 1.7533595514362954e-08, + "loss": 0.10595938563346863, + "step": 5274 + }, + { + "epoch": 2.9504895104895104, + "grad_norm": 0.33860115639817917, + "learning_rate": 1.715046867782788e-08, + "loss": 0.013983756303787231, + "step": 5275 + }, + { + "epoch": 2.951048951048951, + "grad_norm": 0.35873285992411813, + "learning_rate": 1.67715704710103e-08, + "loss": 0.07022291421890259, + "step": 5276 + }, + { + "epoch": 2.9516083916083917, + "grad_norm": 0.40820249951923626, + "learning_rate": 1.63969010544085e-08, + "loss": 0.09447485208511353, + "step": 5277 + }, + { + "epoch": 2.952167832167832, + "grad_norm": 0.3126570098592504, + "learning_rate": 1.6026460586727743e-08, + "loss": 0.027280420064926147, + "step": 5278 + }, + { + "epoch": 2.9527272727272726, + "grad_norm": 0.32058470965122965, + "learning_rate": 1.566024922488252e-08, + "loss": 0.030660949647426605, + "step": 5279 + }, + { + "epoch": 2.9532867132867135, + "grad_norm": 0.39423568352962096, + "learning_rate": 1.5298267123996513e-08, + "loss": 0.07480526715517044, + "step": 5280 + }, + { + "epoch": 2.953846153846154, + "grad_norm": 0.4558124519105311, + "learning_rate": 1.4940514437401523e-08, + "loss": 0.05191662162542343, + "step": 5281 + }, + { + "epoch": 2.9544055944055945, + "grad_norm": 0.48075386468438125, + "learning_rate": 1.4586991316636323e-08, + "loss": 0.0955943763256073, + "step": 5282 + }, + { + "epoch": 2.954965034965035, + "grad_norm": 0.3254896365371831, + "learning_rate": 1.4237697911452242e-08, + "loss": 0.07685510814189911, + "step": 5283 + }, + { + "epoch": 2.9555244755244754, + "grad_norm": 0.3582550128447003, + "learning_rate": 1.3892634369804258e-08, + "loss": 0.06673769652843475, + "step": 5284 + }, + { + "epoch": 2.956083916083916, + "grad_norm": 0.3898145693234172, + "learning_rate": 1.3551800837858787e-08, + "loss": 0.023648586124181747, + "step": 5285 + }, + { + "epoch": 2.9566433566433568, + "grad_norm": 0.48682035692266995, + "learning_rate": 1.3215197459988115e-08, + "loss": 0.14275985956192017, + "step": 5286 + }, + { + "epoch": 2.957202797202797, + "grad_norm": 0.31932711422671456, + "learning_rate": 1.288282437877597e-08, + "loss": 0.05834471806883812, + "step": 5287 + }, + { + "epoch": 2.9577622377622377, + "grad_norm": 0.501136088305763, + "learning_rate": 1.2554681735009733e-08, + "loss": 0.08282680064439774, + "step": 5288 + }, + { + "epoch": 2.9583216783216786, + "grad_norm": 0.3728203414455509, + "learning_rate": 1.2230769667687104e-08, + "loss": 0.042859189212322235, + "step": 5289 + }, + { + "epoch": 2.958881118881119, + "grad_norm": 0.9107049187876418, + "learning_rate": 1.1911088314016106e-08, + "loss": 0.19588994979858398, + "step": 5290 + }, + { + "epoch": 2.9594405594405595, + "grad_norm": 0.4708171061755607, + "learning_rate": 1.1595637809409532e-08, + "loss": 0.03237784653902054, + "step": 5291 + }, + { + "epoch": 2.96, + "grad_norm": 0.5077839095484743, + "learning_rate": 1.1284418287488274e-08, + "loss": 0.09009547531604767, + "step": 5292 + }, + { + "epoch": 2.9605594405594404, + "grad_norm": 0.41479234725137204, + "learning_rate": 1.0977429880081325e-08, + "loss": 0.0711212158203125, + "step": 5293 + }, + { + "epoch": 2.961118881118881, + "grad_norm": 0.5445370006848226, + "learning_rate": 1.0674672717225776e-08, + "loss": 0.11750383675098419, + "step": 5294 + }, + { + "epoch": 2.961678321678322, + "grad_norm": 0.3944651535036282, + "learning_rate": 1.0376146927167929e-08, + "loss": 0.021160634234547615, + "step": 5295 + }, + { + "epoch": 2.9622377622377623, + "grad_norm": 0.4258752134754167, + "learning_rate": 1.0081852636358858e-08, + "loss": 0.05732947215437889, + "step": 5296 + }, + { + "epoch": 2.9627972027972027, + "grad_norm": 0.7810650135947191, + "learning_rate": 9.791789969458843e-09, + "loss": 0.2054559290409088, + "step": 5297 + }, + { + "epoch": 2.9633566433566436, + "grad_norm": 0.36782153243918153, + "learning_rate": 9.505959049335155e-09, + "loss": 0.06776751577854156, + "step": 5298 + }, + { + "epoch": 2.963916083916084, + "grad_norm": 0.5083938326040192, + "learning_rate": 9.22435999706206e-09, + "loss": 0.11562247574329376, + "step": 5299 + }, + { + "epoch": 2.9644755244755245, + "grad_norm": 0.31834725172242784, + "learning_rate": 8.946992931923026e-09, + "loss": 0.025352679193019867, + "step": 5300 + }, + { + "epoch": 2.965034965034965, + "grad_norm": 0.4210755299969828, + "learning_rate": 8.673857971408517e-09, + "loss": 0.04324430599808693, + "step": 5301 + }, + { + "epoch": 2.9655944055944055, + "grad_norm": 0.30746060511166756, + "learning_rate": 8.404955231213763e-09, + "loss": 0.02898053079843521, + "step": 5302 + }, + { + "epoch": 2.966153846153846, + "grad_norm": 0.5105640120156102, + "learning_rate": 8.140284825244315e-09, + "loss": 0.10131726413965225, + "step": 5303 + }, + { + "epoch": 2.966713286713287, + "grad_norm": 0.49216533599722534, + "learning_rate": 7.879846865611606e-09, + "loss": 0.19311170279979706, + "step": 5304 + }, + { + "epoch": 2.9672727272727273, + "grad_norm": 0.36095156617296253, + "learning_rate": 7.623641462634057e-09, + "loss": 0.06657704710960388, + "step": 5305 + }, + { + "epoch": 2.9678321678321677, + "grad_norm": 0.5840173266340489, + "learning_rate": 7.371668724837078e-09, + "loss": 0.12123794853687286, + "step": 5306 + }, + { + "epoch": 2.9683916083916086, + "grad_norm": 0.14931284348928345, + "learning_rate": 7.12392875895529e-09, + "loss": 0.005168756935745478, + "step": 5307 + }, + { + "epoch": 2.968951048951049, + "grad_norm": 0.5422160002264499, + "learning_rate": 6.8804216699280836e-09, + "loss": 0.068355992436409, + "step": 5308 + }, + { + "epoch": 2.9695104895104896, + "grad_norm": 0.5187425277185453, + "learning_rate": 6.64114756090184e-09, + "loss": 0.13130581378936768, + "step": 5309 + }, + { + "epoch": 2.97006993006993, + "grad_norm": 0.3829976565407132, + "learning_rate": 6.406106533231038e-09, + "loss": 0.05100453272461891, + "step": 5310 + }, + { + "epoch": 2.9706293706293705, + "grad_norm": 0.4026370656532104, + "learning_rate": 6.175298686476039e-09, + "loss": 0.05833175405859947, + "step": 5311 + }, + { + "epoch": 2.971188811188811, + "grad_norm": 0.5177748873142619, + "learning_rate": 5.948724118405303e-09, + "loss": 0.06594595313072205, + "step": 5312 + }, + { + "epoch": 2.971748251748252, + "grad_norm": 0.6565827478846719, + "learning_rate": 5.726382924992058e-09, + "loss": 0.21014057099819183, + "step": 5313 + }, + { + "epoch": 2.9723076923076923, + "grad_norm": 0.27952011552482764, + "learning_rate": 5.508275200419855e-09, + "loss": 0.0358211025595665, + "step": 5314 + }, + { + "epoch": 2.9728671328671328, + "grad_norm": 0.20732609190074408, + "learning_rate": 5.294401037075902e-09, + "loss": 0.00839526578783989, + "step": 5315 + }, + { + "epoch": 2.9734265734265737, + "grad_norm": 0.27487242922585986, + "learning_rate": 5.0847605255543995e-09, + "loss": 0.039361946284770966, + "step": 5316 + }, + { + "epoch": 2.973986013986014, + "grad_norm": 0.44639472273892516, + "learning_rate": 4.879353754657645e-09, + "loss": 0.08253898471593857, + "step": 5317 + }, + { + "epoch": 2.9745454545454546, + "grad_norm": 0.4053857798529282, + "learning_rate": 4.678180811393818e-09, + "loss": 0.04213719069957733, + "step": 5318 + }, + { + "epoch": 2.975104895104895, + "grad_norm": 0.49154031774173434, + "learning_rate": 4.481241780976975e-09, + "loss": 0.136572003364563, + "step": 5319 + }, + { + "epoch": 2.9756643356643355, + "grad_norm": 0.42832993120112556, + "learning_rate": 4.288536746828164e-09, + "loss": 0.13080723583698273, + "step": 5320 + }, + { + "epoch": 2.976223776223776, + "grad_norm": 1.4963340382517885, + "learning_rate": 4.100065790576535e-09, + "loss": 0.2413359433412552, + "step": 5321 + }, + { + "epoch": 2.976783216783217, + "grad_norm": 0.3914856468383236, + "learning_rate": 3.915828992056003e-09, + "loss": 0.04885898903012276, + "step": 5322 + }, + { + "epoch": 2.9773426573426574, + "grad_norm": 0.35922635754887233, + "learning_rate": 3.735826429305256e-09, + "loss": 0.04643958434462547, + "step": 5323 + }, + { + "epoch": 2.977902097902098, + "grad_norm": 0.9206448482696191, + "learning_rate": 3.5600581785744104e-09, + "loss": 0.128621444106102, + "step": 5324 + }, + { + "epoch": 2.9784615384615387, + "grad_norm": 0.46785341018512616, + "learning_rate": 3.3885243143161327e-09, + "loss": 0.14379192888736725, + "step": 5325 + }, + { + "epoch": 2.979020979020979, + "grad_norm": 0.5493429680758554, + "learning_rate": 3.2212249091900793e-09, + "loss": 0.1060786321759224, + "step": 5326 + }, + { + "epoch": 2.9795804195804196, + "grad_norm": 0.5220154684788509, + "learning_rate": 3.0581600340617857e-09, + "loss": 0.13857942819595337, + "step": 5327 + }, + { + "epoch": 2.98013986013986, + "grad_norm": 0.42861227535802, + "learning_rate": 2.8993297580048874e-09, + "loss": 0.08162636309862137, + "step": 5328 + }, + { + "epoch": 2.9806993006993006, + "grad_norm": 0.344012379371565, + "learning_rate": 2.7447341482977897e-09, + "loss": 0.05172056332230568, + "step": 5329 + }, + { + "epoch": 2.981258741258741, + "grad_norm": 0.36299268798515344, + "learning_rate": 2.594373270424777e-09, + "loss": 0.028990596532821655, + "step": 5330 + }, + { + "epoch": 2.981818181818182, + "grad_norm": 0.4036097308168451, + "learning_rate": 2.4482471880793447e-09, + "loss": 0.046703826636075974, + "step": 5331 + }, + { + "epoch": 2.9823776223776224, + "grad_norm": 0.37772415557721145, + "learning_rate": 2.3063559631575363e-09, + "loss": 0.039994753897190094, + "step": 5332 + }, + { + "epoch": 2.982937062937063, + "grad_norm": 0.378997741866621, + "learning_rate": 2.168699655762385e-09, + "loss": 0.04648361727595329, + "step": 5333 + }, + { + "epoch": 2.9834965034965037, + "grad_norm": 0.5335871549888598, + "learning_rate": 2.0352783242039155e-09, + "loss": 0.10541056096553802, + "step": 5334 + }, + { + "epoch": 2.984055944055944, + "grad_norm": 0.4701246874786772, + "learning_rate": 1.90609202499914e-09, + "loss": 0.10288608074188232, + "step": 5335 + }, + { + "epoch": 2.9846153846153847, + "grad_norm": 0.48778685010938394, + "learning_rate": 1.7811408128687312e-09, + "loss": 0.06302841007709503, + "step": 5336 + }, + { + "epoch": 2.985174825174825, + "grad_norm": 0.36629870258588115, + "learning_rate": 1.660424740741462e-09, + "loss": 0.0783616304397583, + "step": 5337 + }, + { + "epoch": 2.9857342657342656, + "grad_norm": 0.536169338289191, + "learning_rate": 1.5439438597508739e-09, + "loss": 0.11086538434028625, + "step": 5338 + }, + { + "epoch": 2.986293706293706, + "grad_norm": 0.4663742462250939, + "learning_rate": 1.4316982192374984e-09, + "loss": 0.06495922058820724, + "step": 5339 + }, + { + "epoch": 2.986853146853147, + "grad_norm": 0.28595429659942007, + "learning_rate": 1.3236878667477471e-09, + "loss": 0.060855019837617874, + "step": 5340 + }, + { + "epoch": 2.9874125874125874, + "grad_norm": 0.4198180228479673, + "learning_rate": 1.2199128480328004e-09, + "loss": 0.10884314775466919, + "step": 5341 + }, + { + "epoch": 2.987972027972028, + "grad_norm": 0.4410004014570075, + "learning_rate": 1.120373207050829e-09, + "loss": 0.09586424380540848, + "step": 5342 + }, + { + "epoch": 2.988531468531469, + "grad_norm": 0.711909755059331, + "learning_rate": 1.0250689859658824e-09, + "loss": 0.16469401121139526, + "step": 5343 + }, + { + "epoch": 2.9890909090909092, + "grad_norm": 0.4218296061849519, + "learning_rate": 9.340002251478907e-10, + "loss": 0.06607284396886826, + "step": 5344 + }, + { + "epoch": 2.9896503496503497, + "grad_norm": 0.46146177752703593, + "learning_rate": 8.471669631726631e-10, + "loss": 0.06861595809459686, + "step": 5345 + }, + { + "epoch": 2.99020979020979, + "grad_norm": 0.4095522691214845, + "learning_rate": 7.64569236820778e-10, + "loss": 0.05425921082496643, + "step": 5346 + }, + { + "epoch": 2.9907692307692306, + "grad_norm": 0.3161758065970725, + "learning_rate": 6.862070810820243e-10, + "loss": 0.07142813503742218, + "step": 5347 + }, + { + "epoch": 2.991328671328671, + "grad_norm": 0.3547350342539299, + "learning_rate": 6.120805291487397e-10, + "loss": 0.022150443866848946, + "step": 5348 + }, + { + "epoch": 2.991888111888112, + "grad_norm": 0.30854831460255194, + "learning_rate": 5.421896124191417e-10, + "loss": 0.022666743025183678, + "step": 5349 + }, + { + "epoch": 2.9924475524475525, + "grad_norm": 0.14992893012293967, + "learning_rate": 4.765343604995476e-10, + "loss": 0.005687844939529896, + "step": 5350 + }, + { + "epoch": 2.993006993006993, + "grad_norm": 0.48915848380326216, + "learning_rate": 4.1511480119993395e-10, + "loss": 0.1533994823694229, + "step": 5351 + }, + { + "epoch": 2.993566433566434, + "grad_norm": 0.5286437207787819, + "learning_rate": 3.579309605383774e-10, + "loss": 0.1376902461051941, + "step": 5352 + }, + { + "epoch": 2.9941258741258743, + "grad_norm": 0.2907010093243184, + "learning_rate": 3.049828627355034e-10, + "loss": 0.017431624233722687, + "step": 5353 + }, + { + "epoch": 2.9946853146853147, + "grad_norm": 0.4554006127827053, + "learning_rate": 2.562705302211477e-10, + "loss": 0.056222110986709595, + "step": 5354 + }, + { + "epoch": 2.995244755244755, + "grad_norm": 0.5399210949843448, + "learning_rate": 2.11793983627695e-10, + "loss": 0.18948790431022644, + "step": 5355 + }, + { + "epoch": 2.9958041958041957, + "grad_norm": 0.37848901072873925, + "learning_rate": 1.7155324179674026e-10, + "loss": 0.03144899383187294, + "step": 5356 + }, + { + "epoch": 2.996363636363636, + "grad_norm": 0.3713423139153674, + "learning_rate": 1.3554832177242738e-10, + "loss": 0.12154553830623627, + "step": 5357 + }, + { + "epoch": 2.996923076923077, + "grad_norm": 0.5233249992663805, + "learning_rate": 1.0377923880700025e-10, + "loss": 0.08542687445878983, + "step": 5358 + }, + { + "epoch": 2.9974825174825175, + "grad_norm": 0.4747982761111987, + "learning_rate": 7.624600635636192e-11, + "loss": 0.08679330348968506, + "step": 5359 + }, + { + "epoch": 2.998041958041958, + "grad_norm": 0.4871634308332239, + "learning_rate": 5.29486360845155e-11, + "loss": 0.11148659139871597, + "step": 5360 + }, + { + "epoch": 2.998601398601399, + "grad_norm": 0.49771149045931234, + "learning_rate": 3.388713786023345e-11, + "loss": 0.12423374503850937, + "step": 5361 + }, + { + "epoch": 2.9991608391608393, + "grad_norm": 0.6071247750893207, + "learning_rate": 1.906151975594739e-11, + "loss": 0.17181716859340668, + "step": 5362 + }, + { + "epoch": 2.9997202797202798, + "grad_norm": 0.3980499136345401, + "learning_rate": 8.47178805329918e-12, + "loss": 0.04787980765104294, + "step": 5363 + }, + { + "epoch": 3.0, + "grad_norm": 0.6231104160033252, + "learning_rate": 2.117947237589846e-12, + "loss": 0.05316243693232536, + "step": 5364 + }, + { + "epoch": 3.0, + "step": 5364, + "total_flos": 4.717303430077481e+17, + "train_loss": 0.15186350211219263, + "train_runtime": 38487.8681, + "train_samples_per_second": 1.115, + "train_steps_per_second": 0.139 + } + ], + "logging_steps": 1, + "max_steps": 5364, + "num_input_tokens_seen": 0, + "num_train_epochs": 3, + "save_steps": 500, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": true + }, + "attributes": {} + } + }, + "total_flos": 4.717303430077481e+17, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +} diff --git a/training_args.bin b/training_args.bin new file mode 100644 index 0000000..eb1281b --- /dev/null +++ b/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c9ce68ff61f60d33f158a1d69acbeccbbeb7ad84a766e6641a36434769d68d1f +size 7377 diff --git a/training_loss.png b/training_loss.png new file mode 100644 index 0000000..0e2e661 Binary files /dev/null and b/training_loss.png differ