commit fdddb0ddbf8cfa0bc678492db713445f000e4c5f Author: ModelHub XC Date: Fri Jun 26 19:33:13 2026 +0800 初始化项目,由ModelHub XC社区提供模型 Model: mlfoundations-dev/openthoughts3_100k_qwen25_1b_bsz256_lr2e5_epochs5 Source: Original Platform diff --git a/.gitattributes b/.gitattributes new file mode 100644 index 0000000..f429f18 --- /dev/null +++ b/.gitattributes @@ -0,0 +1,53 @@ +*.7z filter=lfs diff=lfs merge=lfs -text +*.arrow filter=lfs diff=lfs merge=lfs -text + + +*.bz2 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 +*.model filter=lfs diff=lfs merge=lfs -text +*.msgpack 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 +*.pt filter=lfs diff=lfs merge=lfs -text +*.pth filter=lfs diff=lfs merge=lfs -text +*.rar filter=lfs diff=lfs merge=lfs -text +saved_model/**/* 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 +*.xz filter=lfs diff=lfs merge=lfs -text +*.zip filter=lfs diff=lfs merge=lfs -text +*.zstandard filter=lfs diff=lfs merge=lfs -text +*.tfevents* filter=lfs diff=lfs merge=lfs -text +*.db* filter=lfs diff=lfs merge=lfs -text +*.ark* filter=lfs diff=lfs merge=lfs -text +**/*ckpt*data* filter=lfs diff=lfs merge=lfs -text +**/*ckpt*.meta filter=lfs diff=lfs merge=lfs -text +**/*ckpt*.index filter=lfs diff=lfs merge=lfs -text + +*.ckpt filter=lfs diff=lfs merge=lfs -text +*.gguf* filter=lfs diff=lfs merge=lfs -text +*.ggml filter=lfs diff=lfs merge=lfs -text +*.llamafile* filter=lfs diff=lfs merge=lfs -text +*.pt2 filter=lfs diff=lfs merge=lfs -text +*.mlmodel filter=lfs diff=lfs merge=lfs -text +*.npy filter=lfs diff=lfs merge=lfs -text +*.npz filter=lfs diff=lfs merge=lfs -text +*.pickle filter=lfs diff=lfs merge=lfs -text +*.pkl filter=lfs diff=lfs merge=lfs -text +*.tar filter=lfs diff=lfs merge=lfs -text +*.wasm filter=lfs diff=lfs merge=lfs -text +*.zst filter=lfs diff=lfs merge=lfs -text +*tfevents* filter=lfs diff=lfs merge=lfs -text + +vocab.json filter=lfs diff=lfs merge=lfs -text +merges.txt filter=lfs diff=lfs merge=lfs -text +model.safetensors filter=lfs diff=lfs merge=lfs -text +training_args.bin filter=lfs diff=lfs merge=lfs -text +tokenizer.json filter=lfs diff=lfs merge=lfs -text \ No newline at end of file diff --git a/README.md b/README.md new file mode 100644 index 0000000..2327ed8 --- /dev/null +++ b/README.md @@ -0,0 +1,60 @@ +--- +library_name: transformers +license: apache-2.0 +base_model: Qwen/Qwen2.5-1.5B-Instruct +tags: +- llama-factory +- full +- generated_from_trainer +model-index: +- name: openthoughts3_100k_qwen25_1b_bsz256_lr2e5_epochs5 + results: [] +--- + + + +# openthoughts3_100k_qwen25_1b_bsz256_lr2e5_epochs5 + +This model is a fine-tuned version of [Qwen/Qwen2.5-1.5B-Instruct](https://huggingface.co/Qwen/Qwen2.5-1.5B-Instruct) on the mlfoundations-dev/openthoughts3_100k 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: 4 +- eval_batch_size: 8 +- seed: 42 +- distributed_type: multi-GPU +- num_devices: 64 +- total_train_batch_size: 256 +- total_eval_batch_size: 512 +- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments +- lr_scheduler_type: cosine +- lr_scheduler_warmup_ratio: 0.1 +- num_epochs: 5.0 + +### Training results + + + +### Framework versions + +- Transformers 4.46.1 +- Pytorch 2.3.0 +- Datasets 3.1.0 +- Tokenizers 0.20.3 diff --git a/added_tokens.json b/added_tokens.json new file mode 100644 index 0000000..482ced4 --- /dev/null +++ b/added_tokens.json @@ -0,0 +1,24 @@ +{ + "": 151658, + "": 151657, + "<|box_end|>": 151649, + "<|box_start|>": 151648, + "<|endoftext|>": 151643, + "<|file_sep|>": 151664, + "<|fim_middle|>": 151660, + "<|fim_pad|>": 151662, + "<|fim_prefix|>": 151659, + "<|fim_suffix|>": 151661, + "<|im_end|>": 151645, + "<|im_start|>": 151644, + "<|image_pad|>": 151655, + "<|object_ref_end|>": 151647, + "<|object_ref_start|>": 151646, + "<|quad_end|>": 151651, + "<|quad_start|>": 151650, + "<|repo_name|>": 151663, + "<|video_pad|>": 151656, + "<|vision_end|>": 151653, + "<|vision_pad|>": 151654, + "<|vision_start|>": 151652 +} diff --git a/all_results.json b/all_results.json new file mode 100644 index 0000000..73c9a0c --- /dev/null +++ b/all_results.json @@ -0,0 +1,8 @@ +{ + "epoch": 5.0, + "total_flos": 7122204608430080.0, + "train_loss": 1.082116018475779, + "train_runtime": 36273.766, + "train_samples_per_second": 13.784, + "train_steps_per_second": 0.054 +} \ No newline at end of file diff --git a/config.json b/config.json new file mode 100644 index 0000000..0e002a8 --- /dev/null +++ b/config.json @@ -0,0 +1,29 @@ +{ + "_name_or_path": "Qwen/Qwen2.5-1.5B-Instruct", + "architectures": [ + "Qwen2ForCausalLM" + ], + "attention_dropout": 0.0, + "bos_token_id": 151643, + "eos_token_id": 151645, + "hidden_act": "silu", + "hidden_size": 1536, + "initializer_range": 0.02, + "intermediate_size": 8960, + "max_position_embeddings": 32768, + "max_window_layers": 21, + "model_type": "qwen2", + "num_attention_heads": 12, + "num_hidden_layers": 28, + "num_key_value_heads": 2, + "rms_norm_eps": 1e-06, + "rope_scaling": null, + "rope_theta": 1000000.0, + "sliding_window": null, + "tie_word_embeddings": true, + "torch_dtype": "bfloat16", + "transformers_version": "4.46.1", + "use_cache": false, + "use_sliding_window": false, + "vocab_size": 151936 +} diff --git a/configs.yaml b/configs.yaml new file mode 100644 index 0000000..1ff119a --- /dev/null +++ b/configs.yaml @@ -0,0 +1,39 @@ +assistant_tag: gpt +bf16: 'True' +content_tag: value +cutoff_len: '16384' +dataloader_num_workers: '4' +dataloader_persistent_workers: 'True' +dataloader_pin_memory: 'True' +dataset: mlfoundations-dev/openthoughts3_100k +dataset_dir: ONLINE +ddp_timeout: '180000000' +deepspeed: /opt/ml/code/zero3.json +do_train: 'True' +enable_liger_kernel: 'True' +finetuning_type: full +formatting: sharegpt +global_batch_size: '256' +gradient_accumulation_steps: '1' +hub_model_id: mlfoundations-dev/openthoughts3_100k_qwen25_1b_bsz256_lr2e5_epochs5 +learning_rate: 2e-05 +logging_steps: '1' +lr_scheduler_type: cosine +messages: conversations +model_name_or_path: Qwen/Qwen2.5-1.5B-Instruct +num_train_epochs: '5.0' +output_dir: /opt/ml/model +overwrite_cache: 'True' +per_device_train_batch_size: '4' +plot_loss: 'True' +preprocessing_num_workers: '16' +push_to_db: 'True' +push_to_hub: 'True' +report_to: wandb +role_tag: from +run_name: openthoughts3_100k_qwen25_1b_bsz256_lr2e5_epochs5 +save_strategy: epoch +stage: sft +template: qwen25 +user_tag: human +warmup_ratio: '0.1' diff --git a/configuration.json b/configuration.json new file mode 100644 index 0000000..bbeeda1 --- /dev/null +++ b/configuration.json @@ -0,0 +1 @@ +{"framework": "pytorch", "task": "text-generation", "allow_remote": true} \ No newline at end of file diff --git a/generation_config.json b/generation_config.json new file mode 100644 index 0000000..16e88f7 --- /dev/null +++ b/generation_config.json @@ -0,0 +1,14 @@ +{ + "bos_token_id": 151643, + "do_sample": true, + "eos_token_id": [ + 151645, + 151643 + ], + "pad_token_id": 151643, + "repetition_penalty": 1.1, + "temperature": 0.7, + "top_k": 20, + "top_p": 0.8, + "transformers_version": "4.46.1" +} diff --git a/merges.txt b/merges.txt new file mode 100644 index 0000000..80c1a19 --- /dev/null +++ b/merges.txt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8831e4f1a044471340f7c0a83d7bd71306a5b867e95fd870f74d0c5308a904d5 +size 1671853 diff --git a/model.safetensors b/model.safetensors new file mode 100644 index 0000000..b38c2cc --- /dev/null +++ b/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8da885df9a99461903e1c0b22dd47d3958acd92eea81a1b46f3fb8dd54f55625 +size 3087467144 diff --git a/special_tokens_map.json b/special_tokens_map.json new file mode 100644 index 0000000..17305b3 --- /dev/null +++ b/special_tokens_map.json @@ -0,0 +1,31 @@ +{ + "additional_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|>" + ], + "eos_token": { + "content": "<|endoftext|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "pad_token": { + "content": "<|endoftext|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + } +} diff --git a/tokenizer.json b/tokenizer.json new file mode 100644 index 0000000..51ebb3b --- /dev/null +++ b/tokenizer.json @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9c5ae00e602b8860cbd784ba82a8aa14e8feecec692e7076590d014d7b7fdafa +size 11421896 diff --git a/tokenizer_config.json b/tokenizer_config.json new file mode 100644 index 0000000..b84f53a --- /dev/null +++ b/tokenizer_config.json @@ -0,0 +1,208 @@ +{ + "add_bos_token": false, + "add_prefix_space": false, + "added_tokens_decoder": { + "151643": { + "content": "<|endoftext|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151644": { + "content": "<|im_start|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151645": { + "content": "<|im_end|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151646": { + "content": "<|object_ref_start|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151647": { + "content": "<|object_ref_end|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151648": { + "content": "<|box_start|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151649": { + "content": "<|box_end|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151650": { + "content": "<|quad_start|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151651": { + "content": "<|quad_end|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151652": { + "content": "<|vision_start|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151653": { + "content": "<|vision_end|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151654": { + "content": "<|vision_pad|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151655": { + "content": "<|image_pad|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151656": { + "content": "<|video_pad|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "151657": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": false + }, + "151658": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": false + }, + "151659": { + "content": "<|fim_prefix|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": false + }, + "151660": { + "content": "<|fim_middle|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": false + }, + "151661": { + "content": "<|fim_suffix|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": false + }, + "151662": { + "content": "<|fim_pad|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": false + }, + "151663": { + "content": "<|repo_name|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": false + }, + "151664": { + "content": "<|file_sep|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": false + } + }, + "additional_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|>" + ], + "bos_token": null, + "chat_template": "{%- if tools %}\n {{- '<|im_start|>system\\n' }}\n {%- if messages[0]['role'] == 'system' %}\n {{- messages[0]['content'] }}\n {%- else %}\n {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}\n {%- endif %}\n {{- \"\\n\\n# 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\" }}\n {%- for tool in tools %}\n {{- \"\\n\" }}\n {{- tool | tojson }}\n {%- endfor %}\n {{- \"\\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\" }}\n{%- else %}\n {%- if messages[0]['role'] == 'system' %}\n {{- '<|im_start|>system\\n' + messages[0]['content'] + '<|im_end|>\\n' }}\n {%- else %}\n {{- '<|im_start|>system\\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\\n' }}\n {%- endif %}\n{%- endif %}\n{%- for message in messages %}\n {%- if (message.role == \"user\") or (message.role == \"system\" and not loop.first) or (message.role == \"assistant\" and not message.tool_calls) %}\n {{- '<|im_start|>' + message.role + '\\n' + message.content + '<|im_end|>' + '\\n' }}\n {%- elif message.role == \"assistant\" %}\n {{- '<|im_start|>' + message.role }}\n {%- if message.content %}\n {{- '\\n' + message.content }}\n {%- endif %}\n {%- for tool_call in message.tool_calls %}\n {%- if tool_call.function is defined %}\n {%- set tool_call = tool_call.function %}\n {%- endif %}\n {{- '\\n\\n{\"name\": \"' }}\n {{- tool_call.name }}\n {{- '\", \"arguments\": ' }}\n {{- tool_call.arguments | tojson }}\n {{- '}\\n' }}\n {%- endfor %}\n {{- '<|im_end|>\\n' }}\n {%- elif message.role == \"tool\" %}\n {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != \"tool\") %}\n {{- '<|im_start|>user' }}\n {%- endif %}\n {{- '\\n\\n' }}\n {{- message.content }}\n {{- '\\n' }}\n {%- if loop.last or (messages[loop.index0 + 1].role != \"tool\") %}\n {{- '<|im_end|>\\n' }}\n {%- endif %}\n {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n {{- '<|im_start|>assistant\\n' }}\n{%- endif %}\n", + "clean_up_tokenization_spaces": false, + "eos_token": "<|endoftext|>", + "errors": "replace", + "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..73c9a0c --- /dev/null +++ b/train_results.json @@ -0,0 +1,8 @@ +{ + "epoch": 5.0, + "total_flos": 7122204608430080.0, + "train_loss": 1.082116018475779, + "train_runtime": 36273.766, + "train_samples_per_second": 13.784, + "train_steps_per_second": 0.054 +} \ No newline at end of file diff --git a/trainer_log.jsonl b/trainer_log.jsonl new file mode 100644 index 0000000..8781c0e --- /dev/null +++ b/trainer_log.jsonl @@ -0,0 +1,1956 @@ +{"current_steps": 1, "total_steps": 1955, "loss": 1.5213, "lr": 1.0204081632653061e-07, "epoch": 0.0025575447570332483, "percentage": 0.05, "elapsed_time": "0:00:25", "remaining_time": "13:57:48"} +{"current_steps": 2, "total_steps": 1955, "loss": 1.4742, "lr": 2.0408163265306121e-07, "epoch": 0.005115089514066497, "percentage": 0.1, "elapsed_time": "0:00:44", "remaining_time": "12:00:52"} +{"current_steps": 3, "total_steps": 1955, "loss": 1.4948, "lr": 3.0612244897959183e-07, "epoch": 0.0076726342710997444, "percentage": 0.15, "elapsed_time": "0:01:02", "remaining_time": "11:19:27"} +{"current_steps": 4, "total_steps": 1955, "loss": 1.4841, "lr": 4.0816326530612243e-07, "epoch": 0.010230179028132993, "percentage": 0.2, "elapsed_time": "0:01:21", "remaining_time": "10:59:10"} +{"current_steps": 5, "total_steps": 1955, "loss": 1.494, "lr": 5.102040816326531e-07, "epoch": 0.01278772378516624, "percentage": 0.26, "elapsed_time": "0:01:39", "remaining_time": "10:46:56"} +{"current_steps": 6, "total_steps": 1955, "loss": 1.5063, "lr": 6.122448979591837e-07, "epoch": 0.015345268542199489, "percentage": 0.31, "elapsed_time": "0:01:58", "remaining_time": "10:42:05"} +{"current_steps": 7, "total_steps": 1955, "loss": 1.4791, "lr": 7.142857142857143e-07, "epoch": 0.017902813299232736, "percentage": 0.36, "elapsed_time": "0:02:17", "remaining_time": "10:36:35"} +{"current_steps": 8, "total_steps": 1955, "loss": 1.4882, "lr": 8.163265306122449e-07, "epoch": 0.020460358056265986, "percentage": 0.41, "elapsed_time": "0:02:36", "remaining_time": "10:34:26"} +{"current_steps": 9, "total_steps": 1955, "loss": 1.5228, "lr": 9.183673469387756e-07, "epoch": 0.023017902813299233, "percentage": 0.46, "elapsed_time": "0:02:54", "remaining_time": "10:28:40"} +{"current_steps": 10, "total_steps": 1955, "loss": 1.5153, "lr": 1.0204081632653063e-06, "epoch": 0.02557544757033248, "percentage": 0.51, "elapsed_time": "0:03:13", "remaining_time": "10:25:59"} +{"current_steps": 11, "total_steps": 1955, "loss": 1.5046, "lr": 1.122448979591837e-06, "epoch": 0.028132992327365727, "percentage": 0.56, "elapsed_time": "0:03:31", "remaining_time": "10:21:51"} +{"current_steps": 12, "total_steps": 1955, "loss": 1.4692, "lr": 1.2244897959183673e-06, "epoch": 0.030690537084398978, "percentage": 0.61, "elapsed_time": "0:03:49", "remaining_time": "10:18:21"} +{"current_steps": 13, "total_steps": 1955, "loss": 1.5132, "lr": 1.3265306122448982e-06, "epoch": 0.03324808184143223, "percentage": 0.66, "elapsed_time": "0:04:07", "remaining_time": "10:16:56"} +{"current_steps": 14, "total_steps": 1955, "loss": 1.4473, "lr": 1.4285714285714286e-06, "epoch": 0.03580562659846547, "percentage": 0.72, "elapsed_time": "0:04:26", "remaining_time": "10:15:38"} +{"current_steps": 15, "total_steps": 1955, "loss": 1.4862, "lr": 1.5306122448979593e-06, "epoch": 0.03836317135549872, "percentage": 0.77, "elapsed_time": "0:04:45", "remaining_time": "10:14:27"} +{"current_steps": 16, "total_steps": 1955, "loss": 1.4457, "lr": 1.6326530612244897e-06, "epoch": 0.04092071611253197, "percentage": 0.82, "elapsed_time": "0:05:04", "remaining_time": "10:14:15"} +{"current_steps": 17, "total_steps": 1955, "loss": 1.4699, "lr": 1.7346938775510206e-06, "epoch": 0.043478260869565216, "percentage": 0.87, "elapsed_time": "0:05:22", "remaining_time": "10:12:07"} +{"current_steps": 18, "total_steps": 1955, "loss": 1.4732, "lr": 1.8367346938775512e-06, "epoch": 0.04603580562659847, "percentage": 0.92, "elapsed_time": "0:05:40", "remaining_time": "10:11:31"} +{"current_steps": 19, "total_steps": 1955, "loss": 1.4474, "lr": 1.938775510204082e-06, "epoch": 0.04859335038363171, "percentage": 0.97, "elapsed_time": "0:05:59", "remaining_time": "10:10:41"} +{"current_steps": 20, "total_steps": 1955, "loss": 1.4366, "lr": 2.0408163265306125e-06, "epoch": 0.05115089514066496, "percentage": 1.02, "elapsed_time": "0:06:18", "remaining_time": "10:11:07"} +{"current_steps": 21, "total_steps": 1955, "loss": 1.4239, "lr": 2.1428571428571427e-06, "epoch": 0.05370843989769821, "percentage": 1.07, "elapsed_time": "0:06:37", "remaining_time": "10:09:22"} +{"current_steps": 22, "total_steps": 1955, "loss": 1.4529, "lr": 2.244897959183674e-06, "epoch": 0.056265984654731455, "percentage": 1.13, "elapsed_time": "0:06:55", "remaining_time": "10:08:41"} +{"current_steps": 23, "total_steps": 1955, "loss": 1.4073, "lr": 2.3469387755102044e-06, "epoch": 0.058823529411764705, "percentage": 1.18, "elapsed_time": "0:07:14", "remaining_time": "10:08:05"} +{"current_steps": 24, "total_steps": 1955, "loss": 1.4322, "lr": 2.4489795918367347e-06, "epoch": 0.061381074168797956, "percentage": 1.23, "elapsed_time": "0:07:32", "remaining_time": "10:07:26"} +{"current_steps": 25, "total_steps": 1955, "loss": 1.4073, "lr": 2.5510204081632657e-06, "epoch": 0.0639386189258312, "percentage": 1.28, "elapsed_time": "0:07:51", "remaining_time": "10:07:00"} +{"current_steps": 26, "total_steps": 1955, "loss": 1.3989, "lr": 2.6530612244897964e-06, "epoch": 0.06649616368286446, "percentage": 1.33, "elapsed_time": "0:08:10", "remaining_time": "10:06:26"} +{"current_steps": 27, "total_steps": 1955, "loss": 1.3804, "lr": 2.7551020408163266e-06, "epoch": 0.06905370843989769, "percentage": 1.38, "elapsed_time": "0:08:29", "remaining_time": "10:06:02"} +{"current_steps": 28, "total_steps": 1955, "loss": 1.4052, "lr": 2.8571428571428573e-06, "epoch": 0.07161125319693094, "percentage": 1.43, "elapsed_time": "0:08:47", "remaining_time": "10:05:28"} +{"current_steps": 29, "total_steps": 1955, "loss": 1.382, "lr": 2.959183673469388e-06, "epoch": 0.0741687979539642, "percentage": 1.48, "elapsed_time": "0:09:06", "remaining_time": "10:04:57"} +{"current_steps": 30, "total_steps": 1955, "loss": 1.3963, "lr": 3.0612244897959185e-06, "epoch": 0.07672634271099744, "percentage": 1.53, "elapsed_time": "0:09:25", "remaining_time": "10:04:34"} +{"current_steps": 31, "total_steps": 1955, "loss": 1.3901, "lr": 3.1632653061224496e-06, "epoch": 0.0792838874680307, "percentage": 1.59, "elapsed_time": "0:09:43", "remaining_time": "10:03:23"} +{"current_steps": 32, "total_steps": 1955, "loss": 1.363, "lr": 3.2653061224489794e-06, "epoch": 0.08184143222506395, "percentage": 1.64, "elapsed_time": "0:10:01", "remaining_time": "10:02:19"} +{"current_steps": 33, "total_steps": 1955, "loss": 1.3952, "lr": 3.3673469387755105e-06, "epoch": 0.08439897698209718, "percentage": 1.69, "elapsed_time": "0:10:20", "remaining_time": "10:02:05"} +{"current_steps": 34, "total_steps": 1955, "loss": 1.4103, "lr": 3.469387755102041e-06, "epoch": 0.08695652173913043, "percentage": 1.74, "elapsed_time": "0:10:38", "remaining_time": "10:01:03"} +{"current_steps": 35, "total_steps": 1955, "loss": 1.3459, "lr": 3.5714285714285718e-06, "epoch": 0.08951406649616368, "percentage": 1.79, "elapsed_time": "0:10:56", "remaining_time": "10:00:37"} +{"current_steps": 36, "total_steps": 1955, "loss": 1.3561, "lr": 3.6734693877551024e-06, "epoch": 0.09207161125319693, "percentage": 1.84, "elapsed_time": "0:11:15", "remaining_time": "10:00:10"} +{"current_steps": 37, "total_steps": 1955, "loss": 1.3694, "lr": 3.7755102040816327e-06, "epoch": 0.09462915601023018, "percentage": 1.89, "elapsed_time": "0:11:34", "remaining_time": "9:59:45"} +{"current_steps": 38, "total_steps": 1955, "loss": 1.3415, "lr": 3.877551020408164e-06, "epoch": 0.09718670076726342, "percentage": 1.94, "elapsed_time": "0:11:53", "remaining_time": "9:59:53"} +{"current_steps": 39, "total_steps": 1955, "loss": 1.3571, "lr": 3.979591836734694e-06, "epoch": 0.09974424552429667, "percentage": 1.99, "elapsed_time": "0:12:12", "remaining_time": "9:59:27"} +{"current_steps": 40, "total_steps": 1955, "loss": 1.2958, "lr": 4.081632653061225e-06, "epoch": 0.10230179028132992, "percentage": 2.05, "elapsed_time": "0:12:30", "remaining_time": "9:59:06"} +{"current_steps": 41, "total_steps": 1955, "loss": 1.3227, "lr": 4.183673469387755e-06, "epoch": 0.10485933503836317, "percentage": 2.1, "elapsed_time": "0:12:49", "remaining_time": "9:58:41"} +{"current_steps": 42, "total_steps": 1955, "loss": 1.3212, "lr": 4.2857142857142855e-06, "epoch": 0.10741687979539642, "percentage": 2.15, "elapsed_time": "0:13:08", "remaining_time": "9:58:16"} +{"current_steps": 43, "total_steps": 1955, "loss": 1.2923, "lr": 4.3877551020408165e-06, "epoch": 0.10997442455242967, "percentage": 2.2, "elapsed_time": "0:13:26", "remaining_time": "9:58:01"} +{"current_steps": 44, "total_steps": 1955, "loss": 1.3017, "lr": 4.489795918367348e-06, "epoch": 0.11253196930946291, "percentage": 2.25, "elapsed_time": "0:13:45", "remaining_time": "9:57:33"} +{"current_steps": 45, "total_steps": 1955, "loss": 1.3013, "lr": 4.591836734693878e-06, "epoch": 0.11508951406649616, "percentage": 2.3, "elapsed_time": "0:14:03", "remaining_time": "9:56:45"} +{"current_steps": 46, "total_steps": 1955, "loss": 1.3225, "lr": 4.693877551020409e-06, "epoch": 0.11764705882352941, "percentage": 2.35, "elapsed_time": "0:14:22", "remaining_time": "9:56:22"} +{"current_steps": 47, "total_steps": 1955, "loss": 1.3221, "lr": 4.795918367346939e-06, "epoch": 0.12020460358056266, "percentage": 2.4, "elapsed_time": "0:14:41", "remaining_time": "9:56:27"} +{"current_steps": 48, "total_steps": 1955, "loss": 1.291, "lr": 4.897959183673469e-06, "epoch": 0.12276214833759591, "percentage": 2.46, "elapsed_time": "0:15:00", "remaining_time": "9:56:04"} +{"current_steps": 49, "total_steps": 1955, "loss": 1.304, "lr": 5e-06, "epoch": 0.12531969309462915, "percentage": 2.51, "elapsed_time": "0:15:18", "remaining_time": "9:55:41"} +{"current_steps": 50, "total_steps": 1955, "loss": 1.2735, "lr": 5.1020408163265315e-06, "epoch": 0.1278772378516624, "percentage": 2.56, "elapsed_time": "0:15:37", "remaining_time": "9:55:18"} +{"current_steps": 51, "total_steps": 1955, "loss": 1.2796, "lr": 5.204081632653062e-06, "epoch": 0.13043478260869565, "percentage": 2.61, "elapsed_time": "0:15:56", "remaining_time": "9:54:56"} +{"current_steps": 52, "total_steps": 1955, "loss": 1.2591, "lr": 5.306122448979593e-06, "epoch": 0.1329923273657289, "percentage": 2.66, "elapsed_time": "0:16:15", "remaining_time": "9:54:48"} +{"current_steps": 53, "total_steps": 1955, "loss": 1.2592, "lr": 5.408163265306123e-06, "epoch": 0.13554987212276215, "percentage": 2.71, "elapsed_time": "0:16:33", "remaining_time": "9:54:03"} +{"current_steps": 54, "total_steps": 1955, "loss": 1.2958, "lr": 5.510204081632653e-06, "epoch": 0.13810741687979539, "percentage": 2.76, "elapsed_time": "0:16:51", "remaining_time": "9:53:36"} +{"current_steps": 55, "total_steps": 1955, "loss": 1.3075, "lr": 5.6122448979591834e-06, "epoch": 0.14066496163682865, "percentage": 2.81, "elapsed_time": "0:17:10", "remaining_time": "9:53:14"} +{"current_steps": 56, "total_steps": 1955, "loss": 1.2947, "lr": 5.7142857142857145e-06, "epoch": 0.1432225063938619, "percentage": 2.86, "elapsed_time": "0:17:29", "remaining_time": "9:52:59"} +{"current_steps": 57, "total_steps": 1955, "loss": 1.2764, "lr": 5.816326530612246e-06, "epoch": 0.14578005115089515, "percentage": 2.92, "elapsed_time": "0:17:47", "remaining_time": "9:52:38"} +{"current_steps": 58, "total_steps": 1955, "loss": 1.2911, "lr": 5.918367346938776e-06, "epoch": 0.1483375959079284, "percentage": 2.97, "elapsed_time": "0:18:06", "remaining_time": "9:52:17"} +{"current_steps": 59, "total_steps": 1955, "loss": 1.2564, "lr": 6.020408163265307e-06, "epoch": 0.15089514066496162, "percentage": 3.02, "elapsed_time": "0:18:25", "remaining_time": "9:51:57"} +{"current_steps": 60, "total_steps": 1955, "loss": 1.2802, "lr": 6.122448979591837e-06, "epoch": 0.1534526854219949, "percentage": 3.07, "elapsed_time": "0:18:44", "remaining_time": "9:51:41"} +{"current_steps": 61, "total_steps": 1955, "loss": 1.2673, "lr": 6.224489795918368e-06, "epoch": 0.15601023017902813, "percentage": 3.12, "elapsed_time": "0:19:02", "remaining_time": "9:51:20"} +{"current_steps": 62, "total_steps": 1955, "loss": 1.2984, "lr": 6.326530612244899e-06, "epoch": 0.1585677749360614, "percentage": 3.17, "elapsed_time": "0:19:21", "remaining_time": "9:50:58"} +{"current_steps": 63, "total_steps": 1955, "loss": 1.2811, "lr": 6.4285714285714295e-06, "epoch": 0.16112531969309463, "percentage": 3.22, "elapsed_time": "0:19:39", "remaining_time": "9:50:24"} +{"current_steps": 64, "total_steps": 1955, "loss": 1.2752, "lr": 6.530612244897959e-06, "epoch": 0.1636828644501279, "percentage": 3.27, "elapsed_time": "0:19:57", "remaining_time": "9:49:41"} +{"current_steps": 65, "total_steps": 1955, "loss": 1.2667, "lr": 6.63265306122449e-06, "epoch": 0.16624040920716113, "percentage": 3.32, "elapsed_time": "0:20:16", "remaining_time": "9:49:24"} +{"current_steps": 66, "total_steps": 1955, "loss": 1.2279, "lr": 6.734693877551021e-06, "epoch": 0.16879795396419436, "percentage": 3.38, "elapsed_time": "0:20:34", "remaining_time": "9:49:01"} +{"current_steps": 67, "total_steps": 1955, "loss": 1.2517, "lr": 6.836734693877551e-06, "epoch": 0.17135549872122763, "percentage": 3.43, "elapsed_time": "0:20:52", "remaining_time": "9:48:23"} +{"current_steps": 68, "total_steps": 1955, "loss": 1.2277, "lr": 6.938775510204082e-06, "epoch": 0.17391304347826086, "percentage": 3.48, "elapsed_time": "0:21:11", "remaining_time": "9:48:06"} +{"current_steps": 69, "total_steps": 1955, "loss": 1.2119, "lr": 7.0408163265306125e-06, "epoch": 0.17647058823529413, "percentage": 3.53, "elapsed_time": "0:21:30", "remaining_time": "9:47:46"} +{"current_steps": 70, "total_steps": 1955, "loss": 1.2234, "lr": 7.1428571428571436e-06, "epoch": 0.17902813299232737, "percentage": 3.58, "elapsed_time": "0:21:48", "remaining_time": "9:47:26"} +{"current_steps": 71, "total_steps": 1955, "loss": 1.2332, "lr": 7.244897959183675e-06, "epoch": 0.1815856777493606, "percentage": 3.63, "elapsed_time": "0:22:07", "remaining_time": "9:47:09"} +{"current_steps": 72, "total_steps": 1955, "loss": 1.2476, "lr": 7.346938775510205e-06, "epoch": 0.18414322250639387, "percentage": 3.68, "elapsed_time": "0:22:26", "remaining_time": "9:46:49"} +{"current_steps": 73, "total_steps": 1955, "loss": 1.2809, "lr": 7.448979591836736e-06, "epoch": 0.1867007672634271, "percentage": 3.73, "elapsed_time": "0:22:44", "remaining_time": "9:46:13"} +{"current_steps": 74, "total_steps": 1955, "loss": 1.228, "lr": 7.551020408163265e-06, "epoch": 0.18925831202046037, "percentage": 3.79, "elapsed_time": "0:23:02", "remaining_time": "9:45:53"} +{"current_steps": 75, "total_steps": 1955, "loss": 1.2256, "lr": 7.653061224489796e-06, "epoch": 0.1918158567774936, "percentage": 3.84, "elapsed_time": "0:23:21", "remaining_time": "9:45:34"} +{"current_steps": 76, "total_steps": 1955, "loss": 1.2277, "lr": 7.755102040816327e-06, "epoch": 0.19437340153452684, "percentage": 3.89, "elapsed_time": "0:23:39", "remaining_time": "9:44:59"} +{"current_steps": 77, "total_steps": 1955, "loss": 1.2383, "lr": 7.857142857142858e-06, "epoch": 0.1969309462915601, "percentage": 3.94, "elapsed_time": "0:23:58", "remaining_time": "9:44:38"} +{"current_steps": 78, "total_steps": 1955, "loss": 1.2176, "lr": 7.959183673469388e-06, "epoch": 0.19948849104859334, "percentage": 3.99, "elapsed_time": "0:24:16", "remaining_time": "9:44:05"} +{"current_steps": 79, "total_steps": 1955, "loss": 1.2119, "lr": 8.06122448979592e-06, "epoch": 0.2020460358056266, "percentage": 4.04, "elapsed_time": "0:24:34", "remaining_time": "9:43:45"} +{"current_steps": 80, "total_steps": 1955, "loss": 1.2285, "lr": 8.16326530612245e-06, "epoch": 0.20460358056265984, "percentage": 4.09, "elapsed_time": "0:24:52", "remaining_time": "9:42:58"} +{"current_steps": 81, "total_steps": 1955, "loss": 1.212, "lr": 8.26530612244898e-06, "epoch": 0.2071611253196931, "percentage": 4.14, "elapsed_time": "0:25:11", "remaining_time": "9:42:42"} +{"current_steps": 82, "total_steps": 1955, "loss": 1.2091, "lr": 8.36734693877551e-06, "epoch": 0.20971867007672634, "percentage": 4.19, "elapsed_time": "0:25:29", "remaining_time": "9:42:10"} +{"current_steps": 83, "total_steps": 1955, "loss": 1.2235, "lr": 8.469387755102042e-06, "epoch": 0.21227621483375958, "percentage": 4.25, "elapsed_time": "0:25:47", "remaining_time": "9:41:51"} +{"current_steps": 84, "total_steps": 1955, "loss": 1.2183, "lr": 8.571428571428571e-06, "epoch": 0.21483375959079284, "percentage": 4.3, "elapsed_time": "0:26:06", "remaining_time": "9:41:31"} +{"current_steps": 85, "total_steps": 1955, "loss": 1.2682, "lr": 8.673469387755103e-06, "epoch": 0.21739130434782608, "percentage": 4.35, "elapsed_time": "0:26:25", "remaining_time": "9:41:13"} +{"current_steps": 86, "total_steps": 1955, "loss": 1.2185, "lr": 8.775510204081633e-06, "epoch": 0.21994884910485935, "percentage": 4.4, "elapsed_time": "0:26:43", "remaining_time": "9:40:41"} +{"current_steps": 87, "total_steps": 1955, "loss": 1.2255, "lr": 8.877551020408163e-06, "epoch": 0.22250639386189258, "percentage": 4.45, "elapsed_time": "0:27:01", "remaining_time": "9:40:22"} +{"current_steps": 88, "total_steps": 1955, "loss": 1.2494, "lr": 8.979591836734695e-06, "epoch": 0.22506393861892582, "percentage": 4.5, "elapsed_time": "0:27:20", "remaining_time": "9:40:03"} +{"current_steps": 89, "total_steps": 1955, "loss": 1.2099, "lr": 9.081632653061225e-06, "epoch": 0.22762148337595908, "percentage": 4.55, "elapsed_time": "0:27:39", "remaining_time": "9:39:44"} +{"current_steps": 90, "total_steps": 1955, "loss": 1.2398, "lr": 9.183673469387756e-06, "epoch": 0.23017902813299232, "percentage": 4.6, "elapsed_time": "0:27:57", "remaining_time": "9:39:25"} +{"current_steps": 91, "total_steps": 1955, "loss": 1.2359, "lr": 9.285714285714288e-06, "epoch": 0.23273657289002558, "percentage": 4.65, "elapsed_time": "0:28:16", "remaining_time": "9:39:06"} +{"current_steps": 92, "total_steps": 1955, "loss": 1.2377, "lr": 9.387755102040818e-06, "epoch": 0.23529411764705882, "percentage": 4.71, "elapsed_time": "0:28:34", "remaining_time": "9:38:46"} +{"current_steps": 93, "total_steps": 1955, "loss": 1.2204, "lr": 9.489795918367348e-06, "epoch": 0.23785166240409208, "percentage": 4.76, "elapsed_time": "0:28:53", "remaining_time": "9:38:28"} +{"current_steps": 94, "total_steps": 1955, "loss": 1.2414, "lr": 9.591836734693878e-06, "epoch": 0.24040920716112532, "percentage": 4.81, "elapsed_time": "0:29:12", "remaining_time": "9:38:09"} +{"current_steps": 95, "total_steps": 1955, "loss": 1.1886, "lr": 9.693877551020408e-06, "epoch": 0.24296675191815856, "percentage": 4.86, "elapsed_time": "0:29:30", "remaining_time": "9:37:50"} +{"current_steps": 96, "total_steps": 1955, "loss": 1.2282, "lr": 9.795918367346939e-06, "epoch": 0.24552429667519182, "percentage": 4.91, "elapsed_time": "0:29:49", "remaining_time": "9:37:36"} +{"current_steps": 97, "total_steps": 1955, "loss": 1.2296, "lr": 9.89795918367347e-06, "epoch": 0.24808184143222506, "percentage": 4.96, "elapsed_time": "0:30:07", "remaining_time": "9:37:06"} +{"current_steps": 98, "total_steps": 1955, "loss": 1.2001, "lr": 1e-05, "epoch": 0.2506393861892583, "percentage": 5.01, "elapsed_time": "0:30:26", "remaining_time": "9:36:47"} +{"current_steps": 99, "total_steps": 1955, "loss": 1.1982, "lr": 1.0102040816326531e-05, "epoch": 0.2531969309462916, "percentage": 5.06, "elapsed_time": "0:30:45", "remaining_time": "9:36:29"} +{"current_steps": 100, "total_steps": 1955, "loss": 1.1914, "lr": 1.0204081632653063e-05, "epoch": 0.2557544757033248, "percentage": 5.12, "elapsed_time": "0:31:03", "remaining_time": "9:36:03"} +{"current_steps": 101, "total_steps": 1955, "loss": 1.2125, "lr": 1.0306122448979591e-05, "epoch": 0.25831202046035806, "percentage": 5.17, "elapsed_time": "0:31:22", "remaining_time": "9:35:51"} +{"current_steps": 102, "total_steps": 1955, "loss": 1.2192, "lr": 1.0408163265306123e-05, "epoch": 0.2608695652173913, "percentage": 5.22, "elapsed_time": "0:31:40", "remaining_time": "9:35:22"} +{"current_steps": 103, "total_steps": 1955, "loss": 1.189, "lr": 1.0510204081632654e-05, "epoch": 0.26342710997442453, "percentage": 5.27, "elapsed_time": "0:31:58", "remaining_time": "9:34:52"} +{"current_steps": 104, "total_steps": 1955, "loss": 1.1791, "lr": 1.0612244897959186e-05, "epoch": 0.2659846547314578, "percentage": 5.32, "elapsed_time": "0:32:16", "remaining_time": "9:34:34"} +{"current_steps": 105, "total_steps": 1955, "loss": 1.1866, "lr": 1.0714285714285714e-05, "epoch": 0.26854219948849106, "percentage": 5.37, "elapsed_time": "0:32:35", "remaining_time": "9:34:15"} +{"current_steps": 106, "total_steps": 1955, "loss": 1.1772, "lr": 1.0816326530612246e-05, "epoch": 0.2710997442455243, "percentage": 5.42, "elapsed_time": "0:32:54", "remaining_time": "9:33:57"} +{"current_steps": 107, "total_steps": 1955, "loss": 1.2043, "lr": 1.0918367346938776e-05, "epoch": 0.27365728900255754, "percentage": 5.47, "elapsed_time": "0:33:12", "remaining_time": "9:33:28"} +{"current_steps": 108, "total_steps": 1955, "loss": 1.1831, "lr": 1.1020408163265306e-05, "epoch": 0.27621483375959077, "percentage": 5.52, "elapsed_time": "0:33:30", "remaining_time": "9:33:11"} +{"current_steps": 109, "total_steps": 1955, "loss": 1.2072, "lr": 1.1122448979591838e-05, "epoch": 0.27877237851662406, "percentage": 5.58, "elapsed_time": "0:33:49", "remaining_time": "9:32:46"} +{"current_steps": 110, "total_steps": 1955, "loss": 1.2077, "lr": 1.1224489795918367e-05, "epoch": 0.2813299232736573, "percentage": 5.63, "elapsed_time": "0:34:07", "remaining_time": "9:32:28"} +{"current_steps": 111, "total_steps": 1955, "loss": 1.2113, "lr": 1.1326530612244899e-05, "epoch": 0.28388746803069054, "percentage": 5.68, "elapsed_time": "0:34:26", "remaining_time": "9:32:08"} +{"current_steps": 112, "total_steps": 1955, "loss": 1.199, "lr": 1.1428571428571429e-05, "epoch": 0.2864450127877238, "percentage": 5.73, "elapsed_time": "0:34:45", "remaining_time": "9:31:51"} +{"current_steps": 113, "total_steps": 1955, "loss": 1.1707, "lr": 1.1530612244897961e-05, "epoch": 0.289002557544757, "percentage": 5.78, "elapsed_time": "0:35:04", "remaining_time": "9:31:45"} +{"current_steps": 114, "total_steps": 1955, "loss": 1.1996, "lr": 1.1632653061224491e-05, "epoch": 0.2915601023017903, "percentage": 5.83, "elapsed_time": "0:35:23", "remaining_time": "9:31:26"} +{"current_steps": 115, "total_steps": 1955, "loss": 1.1818, "lr": 1.1734693877551021e-05, "epoch": 0.29411764705882354, "percentage": 5.88, "elapsed_time": "0:35:41", "remaining_time": "9:31:08"} +{"current_steps": 116, "total_steps": 1955, "loss": 1.1899, "lr": 1.1836734693877552e-05, "epoch": 0.2966751918158568, "percentage": 5.93, "elapsed_time": "0:35:59", "remaining_time": "9:30:40"} +{"current_steps": 117, "total_steps": 1955, "loss": 1.1822, "lr": 1.1938775510204084e-05, "epoch": 0.29923273657289, "percentage": 5.98, "elapsed_time": "0:36:18", "remaining_time": "9:30:19"} +{"current_steps": 118, "total_steps": 1955, "loss": 1.2338, "lr": 1.2040816326530614e-05, "epoch": 0.30179028132992325, "percentage": 6.04, "elapsed_time": "0:36:36", "remaining_time": "9:29:54"} +{"current_steps": 119, "total_steps": 1955, "loss": 1.1949, "lr": 1.2142857142857142e-05, "epoch": 0.30434782608695654, "percentage": 6.09, "elapsed_time": "0:36:55", "remaining_time": "9:29:37"} +{"current_steps": 120, "total_steps": 1955, "loss": 1.2006, "lr": 1.2244897959183674e-05, "epoch": 0.3069053708439898, "percentage": 6.14, "elapsed_time": "0:37:14", "remaining_time": "9:29:21"} +{"current_steps": 121, "total_steps": 1955, "loss": 1.2164, "lr": 1.2346938775510204e-05, "epoch": 0.309462915601023, "percentage": 6.19, "elapsed_time": "0:37:32", "remaining_time": "9:29:03"} +{"current_steps": 122, "total_steps": 1955, "loss": 1.1664, "lr": 1.2448979591836736e-05, "epoch": 0.31202046035805625, "percentage": 6.24, "elapsed_time": "0:37:50", "remaining_time": "9:28:36"} +{"current_steps": 123, "total_steps": 1955, "loss": 1.1875, "lr": 1.2551020408163267e-05, "epoch": 0.3145780051150895, "percentage": 6.29, "elapsed_time": "0:38:08", "remaining_time": "9:28:09"} +{"current_steps": 124, "total_steps": 1955, "loss": 1.1822, "lr": 1.2653061224489798e-05, "epoch": 0.3171355498721228, "percentage": 6.34, "elapsed_time": "0:38:26", "remaining_time": "9:27:41"} +{"current_steps": 125, "total_steps": 1955, "loss": 1.1623, "lr": 1.2755102040816327e-05, "epoch": 0.319693094629156, "percentage": 6.39, "elapsed_time": "0:38:45", "remaining_time": "9:27:24"} +{"current_steps": 126, "total_steps": 1955, "loss": 1.2188, "lr": 1.2857142857142859e-05, "epoch": 0.32225063938618925, "percentage": 6.45, "elapsed_time": "0:39:05", "remaining_time": "9:27:20"} +{"current_steps": 127, "total_steps": 1955, "loss": 1.1605, "lr": 1.2959183673469389e-05, "epoch": 0.3248081841432225, "percentage": 6.5, "elapsed_time": "0:39:23", "remaining_time": "9:27:02"} +{"current_steps": 128, "total_steps": 1955, "loss": 1.1804, "lr": 1.3061224489795918e-05, "epoch": 0.3273657289002558, "percentage": 6.55, "elapsed_time": "0:39:42", "remaining_time": "9:26:46"} +{"current_steps": 129, "total_steps": 1955, "loss": 1.1733, "lr": 1.316326530612245e-05, "epoch": 0.329923273657289, "percentage": 6.6, "elapsed_time": "0:40:01", "remaining_time": "9:26:28"} +{"current_steps": 130, "total_steps": 1955, "loss": 1.2038, "lr": 1.326530612244898e-05, "epoch": 0.33248081841432225, "percentage": 6.65, "elapsed_time": "0:40:19", "remaining_time": "9:26:10"} +{"current_steps": 131, "total_steps": 1955, "loss": 1.1673, "lr": 1.3367346938775512e-05, "epoch": 0.3350383631713555, "percentage": 6.7, "elapsed_time": "0:40:38", "remaining_time": "9:25:52"} +{"current_steps": 132, "total_steps": 1955, "loss": 1.2253, "lr": 1.3469387755102042e-05, "epoch": 0.3375959079283887, "percentage": 6.75, "elapsed_time": "0:40:57", "remaining_time": "9:25:34"} +{"current_steps": 133, "total_steps": 1955, "loss": 1.198, "lr": 1.3571428571428574e-05, "epoch": 0.340153452685422, "percentage": 6.8, "elapsed_time": "0:41:16", "remaining_time": "9:25:21"} +{"current_steps": 134, "total_steps": 1955, "loss": 1.1671, "lr": 1.3673469387755102e-05, "epoch": 0.34271099744245526, "percentage": 6.85, "elapsed_time": "0:41:34", "remaining_time": "9:25:02"} +{"current_steps": 135, "total_steps": 1955, "loss": 1.2261, "lr": 1.3775510204081634e-05, "epoch": 0.3452685421994885, "percentage": 6.91, "elapsed_time": "0:41:53", "remaining_time": "9:24:43"} +{"current_steps": 136, "total_steps": 1955, "loss": 1.1854, "lr": 1.3877551020408165e-05, "epoch": 0.34782608695652173, "percentage": 6.96, "elapsed_time": "0:42:11", "remaining_time": "9:24:16"} +{"current_steps": 137, "total_steps": 1955, "loss": 1.1972, "lr": 1.3979591836734696e-05, "epoch": 0.35038363171355497, "percentage": 7.01, "elapsed_time": "0:42:29", "remaining_time": "9:23:58"} +{"current_steps": 138, "total_steps": 1955, "loss": 1.1841, "lr": 1.4081632653061225e-05, "epoch": 0.35294117647058826, "percentage": 7.06, "elapsed_time": "0:42:49", "remaining_time": "9:23:47"} +{"current_steps": 139, "total_steps": 1955, "loss": 1.1582, "lr": 1.4183673469387755e-05, "epoch": 0.3554987212276215, "percentage": 7.11, "elapsed_time": "0:43:07", "remaining_time": "9:23:29"} +{"current_steps": 140, "total_steps": 1955, "loss": 1.217, "lr": 1.4285714285714287e-05, "epoch": 0.35805626598465473, "percentage": 7.16, "elapsed_time": "0:43:26", "remaining_time": "9:23:12"} +{"current_steps": 141, "total_steps": 1955, "loss": 1.1303, "lr": 1.4387755102040817e-05, "epoch": 0.36061381074168797, "percentage": 7.21, "elapsed_time": "0:43:45", "remaining_time": "9:22:54"} +{"current_steps": 142, "total_steps": 1955, "loss": 1.1808, "lr": 1.448979591836735e-05, "epoch": 0.3631713554987212, "percentage": 7.26, "elapsed_time": "0:44:03", "remaining_time": "9:22:36"} +{"current_steps": 143, "total_steps": 1955, "loss": 1.2083, "lr": 1.4591836734693878e-05, "epoch": 0.3657289002557545, "percentage": 7.31, "elapsed_time": "0:44:23", "remaining_time": "9:22:24"} +{"current_steps": 144, "total_steps": 1955, "loss": 1.1761, "lr": 1.469387755102041e-05, "epoch": 0.36828644501278773, "percentage": 7.37, "elapsed_time": "0:44:41", "remaining_time": "9:22:06"} +{"current_steps": 145, "total_steps": 1955, "loss": 1.1887, "lr": 1.479591836734694e-05, "epoch": 0.37084398976982097, "percentage": 7.42, "elapsed_time": "0:45:00", "remaining_time": "9:21:47"} +{"current_steps": 146, "total_steps": 1955, "loss": 1.1735, "lr": 1.4897959183673472e-05, "epoch": 0.3734015345268542, "percentage": 7.47, "elapsed_time": "0:45:18", "remaining_time": "9:21:29"} +{"current_steps": 147, "total_steps": 1955, "loss": 1.2087, "lr": 1.5000000000000002e-05, "epoch": 0.37595907928388744, "percentage": 7.52, "elapsed_time": "0:45:37", "remaining_time": "9:21:11"} +{"current_steps": 148, "total_steps": 1955, "loss": 1.1688, "lr": 1.510204081632653e-05, "epoch": 0.37851662404092073, "percentage": 7.57, "elapsed_time": "0:45:55", "remaining_time": "9:20:48"} +{"current_steps": 149, "total_steps": 1955, "loss": 1.1553, "lr": 1.5204081632653063e-05, "epoch": 0.38107416879795397, "percentage": 7.62, "elapsed_time": "0:46:14", "remaining_time": "9:20:30"} +{"current_steps": 150, "total_steps": 1955, "loss": 1.1664, "lr": 1.530612244897959e-05, "epoch": 0.3836317135549872, "percentage": 7.67, "elapsed_time": "0:46:33", "remaining_time": "9:20:13"} +{"current_steps": 151, "total_steps": 1955, "loss": 1.17, "lr": 1.5408163265306123e-05, "epoch": 0.38618925831202044, "percentage": 7.72, "elapsed_time": "0:46:51", "remaining_time": "9:19:54"} +{"current_steps": 152, "total_steps": 1955, "loss": 1.175, "lr": 1.5510204081632655e-05, "epoch": 0.3887468030690537, "percentage": 7.77, "elapsed_time": "0:47:10", "remaining_time": "9:19:36"} +{"current_steps": 153, "total_steps": 1955, "loss": 1.1646, "lr": 1.5612244897959187e-05, "epoch": 0.391304347826087, "percentage": 7.83, "elapsed_time": "0:47:29", "remaining_time": "9:19:18"} +{"current_steps": 154, "total_steps": 1955, "loss": 1.1698, "lr": 1.5714285714285715e-05, "epoch": 0.3938618925831202, "percentage": 7.88, "elapsed_time": "0:47:47", "remaining_time": "9:19:00"} +{"current_steps": 155, "total_steps": 1955, "loss": 1.1821, "lr": 1.5816326530612247e-05, "epoch": 0.39641943734015345, "percentage": 7.93, "elapsed_time": "0:48:06", "remaining_time": "9:18:41"} +{"current_steps": 156, "total_steps": 1955, "loss": 1.1864, "lr": 1.5918367346938776e-05, "epoch": 0.3989769820971867, "percentage": 7.98, "elapsed_time": "0:48:24", "remaining_time": "9:18:16"} +{"current_steps": 157, "total_steps": 1955, "loss": 1.1999, "lr": 1.6020408163265308e-05, "epoch": 0.40153452685422, "percentage": 8.03, "elapsed_time": "0:48:43", "remaining_time": "9:17:58"} +{"current_steps": 158, "total_steps": 1955, "loss": 1.1928, "lr": 1.612244897959184e-05, "epoch": 0.4040920716112532, "percentage": 8.08, "elapsed_time": "0:49:01", "remaining_time": "9:17:30"} +{"current_steps": 159, "total_steps": 1955, "loss": 1.1455, "lr": 1.6224489795918368e-05, "epoch": 0.40664961636828645, "percentage": 8.13, "elapsed_time": "0:49:19", "remaining_time": "9:17:12"} +{"current_steps": 160, "total_steps": 1955, "loss": 1.1657, "lr": 1.63265306122449e-05, "epoch": 0.4092071611253197, "percentage": 8.18, "elapsed_time": "0:49:38", "remaining_time": "9:16:54"} +{"current_steps": 161, "total_steps": 1955, "loss": 1.1598, "lr": 1.642857142857143e-05, "epoch": 0.4117647058823529, "percentage": 8.24, "elapsed_time": "0:49:57", "remaining_time": "9:16:36"} +{"current_steps": 162, "total_steps": 1955, "loss": 1.1565, "lr": 1.653061224489796e-05, "epoch": 0.4143222506393862, "percentage": 8.29, "elapsed_time": "0:50:15", "remaining_time": "9:16:16"} +{"current_steps": 163, "total_steps": 1955, "loss": 1.1594, "lr": 1.6632653061224492e-05, "epoch": 0.41687979539641945, "percentage": 8.34, "elapsed_time": "0:50:33", "remaining_time": "9:15:51"} +{"current_steps": 164, "total_steps": 1955, "loss": 1.1699, "lr": 1.673469387755102e-05, "epoch": 0.4194373401534527, "percentage": 8.39, "elapsed_time": "0:50:52", "remaining_time": "9:15:33"} +{"current_steps": 165, "total_steps": 1955, "loss": 1.1459, "lr": 1.6836734693877553e-05, "epoch": 0.4219948849104859, "percentage": 8.44, "elapsed_time": "0:51:10", "remaining_time": "9:15:15"} +{"current_steps": 166, "total_steps": 1955, "loss": 1.1667, "lr": 1.6938775510204085e-05, "epoch": 0.42455242966751916, "percentage": 8.49, "elapsed_time": "0:51:29", "remaining_time": "9:14:56"} +{"current_steps": 167, "total_steps": 1955, "loss": 1.1642, "lr": 1.7040816326530613e-05, "epoch": 0.42710997442455245, "percentage": 8.54, "elapsed_time": "0:51:48", "remaining_time": "9:14:41"} +{"current_steps": 168, "total_steps": 1955, "loss": 1.1741, "lr": 1.7142857142857142e-05, "epoch": 0.4296675191815857, "percentage": 8.59, "elapsed_time": "0:52:07", "remaining_time": "9:14:22"} +{"current_steps": 169, "total_steps": 1955, "loss": 1.1722, "lr": 1.7244897959183674e-05, "epoch": 0.4322250639386189, "percentage": 8.64, "elapsed_time": "0:52:25", "remaining_time": "9:13:59"} +{"current_steps": 170, "total_steps": 1955, "loss": 1.1513, "lr": 1.7346938775510206e-05, "epoch": 0.43478260869565216, "percentage": 8.7, "elapsed_time": "0:52:43", "remaining_time": "9:13:41"} +{"current_steps": 171, "total_steps": 1955, "loss": 1.1641, "lr": 1.7448979591836738e-05, "epoch": 0.4373401534526854, "percentage": 8.75, "elapsed_time": "0:53:02", "remaining_time": "9:13:22"} +{"current_steps": 172, "total_steps": 1955, "loss": 1.1615, "lr": 1.7551020408163266e-05, "epoch": 0.4398976982097187, "percentage": 8.8, "elapsed_time": "0:53:21", "remaining_time": "9:13:04"} +{"current_steps": 173, "total_steps": 1955, "loss": 1.1808, "lr": 1.7653061224489798e-05, "epoch": 0.4424552429667519, "percentage": 8.85, "elapsed_time": "0:53:39", "remaining_time": "9:12:46"} +{"current_steps": 174, "total_steps": 1955, "loss": 1.1897, "lr": 1.7755102040816327e-05, "epoch": 0.44501278772378516, "percentage": 8.9, "elapsed_time": "0:53:57", "remaining_time": "9:12:22"} +{"current_steps": 175, "total_steps": 1955, "loss": 1.161, "lr": 1.785714285714286e-05, "epoch": 0.4475703324808184, "percentage": 8.95, "elapsed_time": "0:54:16", "remaining_time": "9:12:00"} +{"current_steps": 176, "total_steps": 1955, "loss": 1.1912, "lr": 1.795918367346939e-05, "epoch": 0.45012787723785164, "percentage": 9.0, "elapsed_time": "0:54:34", "remaining_time": "9:11:35"} +{"current_steps": 177, "total_steps": 1955, "loss": 1.1648, "lr": 1.806122448979592e-05, "epoch": 0.45268542199488493, "percentage": 9.05, "elapsed_time": "0:54:52", "remaining_time": "9:11:17"} +{"current_steps": 178, "total_steps": 1955, "loss": 1.1639, "lr": 1.816326530612245e-05, "epoch": 0.45524296675191817, "percentage": 9.1, "elapsed_time": "0:55:11", "remaining_time": "9:11:01"} +{"current_steps": 179, "total_steps": 1955, "loss": 1.1479, "lr": 1.826530612244898e-05, "epoch": 0.4578005115089514, "percentage": 9.16, "elapsed_time": "0:55:30", "remaining_time": "9:10:43"} +{"current_steps": 180, "total_steps": 1955, "loss": 1.1568, "lr": 1.836734693877551e-05, "epoch": 0.46035805626598464, "percentage": 9.21, "elapsed_time": "0:55:49", "remaining_time": "9:10:25"} +{"current_steps": 181, "total_steps": 1955, "loss": 1.1822, "lr": 1.8469387755102043e-05, "epoch": 0.4629156010230179, "percentage": 9.26, "elapsed_time": "0:56:07", "remaining_time": "9:10:07"} +{"current_steps": 182, "total_steps": 1955, "loss": 1.1579, "lr": 1.8571428571428575e-05, "epoch": 0.46547314578005117, "percentage": 9.31, "elapsed_time": "0:56:25", "remaining_time": "9:09:42"} +{"current_steps": 183, "total_steps": 1955, "loss": 1.17, "lr": 1.8673469387755104e-05, "epoch": 0.4680306905370844, "percentage": 9.36, "elapsed_time": "0:56:44", "remaining_time": "9:09:27"} +{"current_steps": 184, "total_steps": 1955, "loss": 1.1657, "lr": 1.8775510204081636e-05, "epoch": 0.47058823529411764, "percentage": 9.41, "elapsed_time": "0:57:02", "remaining_time": "9:09:02"} +{"current_steps": 185, "total_steps": 1955, "loss": 1.1727, "lr": 1.8877551020408164e-05, "epoch": 0.4731457800511509, "percentage": 9.46, "elapsed_time": "0:57:21", "remaining_time": "9:08:44"} +{"current_steps": 186, "total_steps": 1955, "loss": 1.1697, "lr": 1.8979591836734696e-05, "epoch": 0.47570332480818417, "percentage": 9.51, "elapsed_time": "0:57:39", "remaining_time": "9:08:25"} +{"current_steps": 187, "total_steps": 1955, "loss": 1.1504, "lr": 1.9081632653061225e-05, "epoch": 0.4782608695652174, "percentage": 9.57, "elapsed_time": "0:57:58", "remaining_time": "9:08:09"} +{"current_steps": 188, "total_steps": 1955, "loss": 1.1771, "lr": 1.9183673469387756e-05, "epoch": 0.48081841432225064, "percentage": 9.62, "elapsed_time": "0:58:17", "remaining_time": "9:07:50"} +{"current_steps": 189, "total_steps": 1955, "loss": 1.1576, "lr": 1.928571428571429e-05, "epoch": 0.4833759590792839, "percentage": 9.67, "elapsed_time": "0:58:36", "remaining_time": "9:07:33"} +{"current_steps": 190, "total_steps": 1955, "loss": 1.1597, "lr": 1.9387755102040817e-05, "epoch": 0.4859335038363171, "percentage": 9.72, "elapsed_time": "0:58:54", "remaining_time": "9:07:15"} +{"current_steps": 191, "total_steps": 1955, "loss": 1.1084, "lr": 1.948979591836735e-05, "epoch": 0.4884910485933504, "percentage": 9.77, "elapsed_time": "0:59:13", "remaining_time": "9:06:56"} +{"current_steps": 192, "total_steps": 1955, "loss": 1.1505, "lr": 1.9591836734693877e-05, "epoch": 0.49104859335038364, "percentage": 9.82, "elapsed_time": "0:59:31", "remaining_time": "9:06:38"} +{"current_steps": 193, "total_steps": 1955, "loss": 1.1214, "lr": 1.969387755102041e-05, "epoch": 0.4936061381074169, "percentage": 9.87, "elapsed_time": "0:59:51", "remaining_time": "9:06:27"} +{"current_steps": 194, "total_steps": 1955, "loss": 1.1197, "lr": 1.979591836734694e-05, "epoch": 0.4961636828644501, "percentage": 9.92, "elapsed_time": "1:00:10", "remaining_time": "9:06:11"} +{"current_steps": 195, "total_steps": 1955, "loss": 1.195, "lr": 1.9897959183673473e-05, "epoch": 0.49872122762148335, "percentage": 9.97, "elapsed_time": "1:00:28", "remaining_time": "9:05:47"} +{"current_steps": 196, "total_steps": 1955, "loss": 1.1522, "lr": 2e-05, "epoch": 0.5012787723785166, "percentage": 10.03, "elapsed_time": "1:00:46", "remaining_time": "9:05:29"} +{"current_steps": 197, "total_steps": 1955, "loss": 1.1393, "lr": 1.9999984050834843e-05, "epoch": 0.5038363171355499, "percentage": 10.08, "elapsed_time": "1:01:06", "remaining_time": "9:05:15"} +{"current_steps": 198, "total_steps": 1955, "loss": 1.172, "lr": 1.999993620339024e-05, "epoch": 0.5063938618925832, "percentage": 10.13, "elapsed_time": "1:01:24", "remaining_time": "9:04:52"} +{"current_steps": 199, "total_steps": 1955, "loss": 1.1374, "lr": 1.9999856457818812e-05, "epoch": 0.5089514066496164, "percentage": 10.18, "elapsed_time": "1:01:42", "remaining_time": "9:04:34"} +{"current_steps": 200, "total_steps": 1955, "loss": 1.1511, "lr": 1.9999744814374943e-05, "epoch": 0.5115089514066496, "percentage": 10.23, "elapsed_time": "1:02:01", "remaining_time": "9:04:15"} +{"current_steps": 201, "total_steps": 1955, "loss": 1.1342, "lr": 1.999960127341475e-05, "epoch": 0.5140664961636828, "percentage": 10.28, "elapsed_time": "1:02:19", "remaining_time": "9:03:53"} +{"current_steps": 202, "total_steps": 1955, "loss": 1.1282, "lr": 1.9999425835396113e-05, "epoch": 0.5166240409207161, "percentage": 10.33, "elapsed_time": "1:02:38", "remaining_time": "9:03:35"} +{"current_steps": 203, "total_steps": 1955, "loss": 1.1589, "lr": 1.9999218500878645e-05, "epoch": 0.5191815856777494, "percentage": 10.38, "elapsed_time": "1:02:56", "remaining_time": "9:03:17"} +{"current_steps": 204, "total_steps": 1955, "loss": 1.1291, "lr": 1.9998979270523704e-05, "epoch": 0.5217391304347826, "percentage": 10.43, "elapsed_time": "1:03:15", "remaining_time": "9:02:58"} +{"current_steps": 205, "total_steps": 1955, "loss": 1.1529, "lr": 1.99987081450944e-05, "epoch": 0.5242966751918159, "percentage": 10.49, "elapsed_time": "1:03:34", "remaining_time": "9:02:40"} +{"current_steps": 206, "total_steps": 1955, "loss": 1.1418, "lr": 1.999840512545558e-05, "epoch": 0.5268542199488491, "percentage": 10.54, "elapsed_time": "1:03:52", "remaining_time": "9:02:22"} +{"current_steps": 207, "total_steps": 1955, "loss": 1.1596, "lr": 1.9998070212573823e-05, "epoch": 0.5294117647058824, "percentage": 10.59, "elapsed_time": "1:04:10", "remaining_time": "9:01:58"} +{"current_steps": 208, "total_steps": 1955, "loss": 1.1457, "lr": 1.9997703407517442e-05, "epoch": 0.5319693094629157, "percentage": 10.64, "elapsed_time": "1:04:29", "remaining_time": "9:01:39"} +{"current_steps": 209, "total_steps": 1955, "loss": 1.135, "lr": 1.999730471145649e-05, "epoch": 0.5345268542199488, "percentage": 10.69, "elapsed_time": "1:04:47", "remaining_time": "9:01:13"} +{"current_steps": 210, "total_steps": 1955, "loss": 1.1286, "lr": 1.999687412566274e-05, "epoch": 0.5370843989769821, "percentage": 10.74, "elapsed_time": "1:05:05", "remaining_time": "9:00:50"} +{"current_steps": 211, "total_steps": 1955, "loss": 1.1488, "lr": 1.9996411651509683e-05, "epoch": 0.5396419437340153, "percentage": 10.79, "elapsed_time": "1:05:23", "remaining_time": "9:00:26"} +{"current_steps": 212, "total_steps": 1955, "loss": 1.1172, "lr": 1.999591729047254e-05, "epoch": 0.5421994884910486, "percentage": 10.84, "elapsed_time": "1:05:41", "remaining_time": "9:00:08"} +{"current_steps": 213, "total_steps": 1955, "loss": 1.162, "lr": 1.9995391044128242e-05, "epoch": 0.5447570332480819, "percentage": 10.9, "elapsed_time": "1:06:00", "remaining_time": "8:59:50"} +{"current_steps": 214, "total_steps": 1955, "loss": 1.1111, "lr": 1.9994832914155417e-05, "epoch": 0.5473145780051151, "percentage": 10.95, "elapsed_time": "1:06:19", "remaining_time": "8:59:31"} +{"current_steps": 215, "total_steps": 1955, "loss": 1.118, "lr": 1.9994242902334418e-05, "epoch": 0.5498721227621484, "percentage": 11.0, "elapsed_time": "1:06:37", "remaining_time": "8:59:08"} +{"current_steps": 216, "total_steps": 1955, "loss": 1.1515, "lr": 1.9993621010547277e-05, "epoch": 0.5524296675191815, "percentage": 11.05, "elapsed_time": "1:06:55", "remaining_time": "8:58:50"} +{"current_steps": 217, "total_steps": 1955, "loss": 1.125, "lr": 1.9992967240777727e-05, "epoch": 0.5549872122762148, "percentage": 11.1, "elapsed_time": "1:07:13", "remaining_time": "8:58:27"} +{"current_steps": 218, "total_steps": 1955, "loss": 1.0982, "lr": 1.9992281595111186e-05, "epoch": 0.5575447570332481, "percentage": 11.15, "elapsed_time": "1:07:32", "remaining_time": "8:58:06"} +{"current_steps": 219, "total_steps": 1955, "loss": 1.121, "lr": 1.9991564075734746e-05, "epoch": 0.5601023017902813, "percentage": 11.2, "elapsed_time": "1:07:50", "remaining_time": "8:57:48"} +{"current_steps": 220, "total_steps": 1955, "loss": 1.1759, "lr": 1.9990814684937177e-05, "epoch": 0.5626598465473146, "percentage": 11.25, "elapsed_time": "1:08:09", "remaining_time": "8:57:30"} +{"current_steps": 221, "total_steps": 1955, "loss": 1.1183, "lr": 1.9990033425108906e-05, "epoch": 0.5652173913043478, "percentage": 11.3, "elapsed_time": "1:08:28", "remaining_time": "8:57:12"} +{"current_steps": 222, "total_steps": 1955, "loss": 1.1193, "lr": 1.998922029874203e-05, "epoch": 0.5677749360613811, "percentage": 11.36, "elapsed_time": "1:08:46", "remaining_time": "8:56:50"} +{"current_steps": 223, "total_steps": 1955, "loss": 1.1249, "lr": 1.9988375308430275e-05, "epoch": 0.5703324808184144, "percentage": 11.41, "elapsed_time": "1:09:04", "remaining_time": "8:56:26"} +{"current_steps": 224, "total_steps": 1955, "loss": 1.1553, "lr": 1.9987498456869025e-05, "epoch": 0.5728900255754475, "percentage": 11.46, "elapsed_time": "1:09:22", "remaining_time": "8:56:09"} +{"current_steps": 225, "total_steps": 1955, "loss": 1.1531, "lr": 1.9986589746855295e-05, "epoch": 0.5754475703324808, "percentage": 11.51, "elapsed_time": "1:09:40", "remaining_time": "8:55:46"} +{"current_steps": 226, "total_steps": 1955, "loss": 1.1252, "lr": 1.998564918128771e-05, "epoch": 0.578005115089514, "percentage": 11.56, "elapsed_time": "1:09:58", "remaining_time": "8:55:23"} +{"current_steps": 227, "total_steps": 1955, "loss": 1.1352, "lr": 1.998467676316652e-05, "epoch": 0.5805626598465473, "percentage": 11.61, "elapsed_time": "1:10:17", "remaining_time": "8:55:04"} +{"current_steps": 228, "total_steps": 1955, "loss": 1.1211, "lr": 1.9983672495593577e-05, "epoch": 0.5831202046035806, "percentage": 11.66, "elapsed_time": "1:10:36", "remaining_time": "8:54:46"} +{"current_steps": 229, "total_steps": 1955, "loss": 1.14, "lr": 1.9982636381772328e-05, "epoch": 0.5856777493606138, "percentage": 11.71, "elapsed_time": "1:10:54", "remaining_time": "8:54:28"} +{"current_steps": 230, "total_steps": 1955, "loss": 1.1445, "lr": 1.99815684250078e-05, "epoch": 0.5882352941176471, "percentage": 11.76, "elapsed_time": "1:11:13", "remaining_time": "8:54:10"} +{"current_steps": 231, "total_steps": 1955, "loss": 1.1456, "lr": 1.9980468628706606e-05, "epoch": 0.5907928388746803, "percentage": 11.82, "elapsed_time": "1:11:32", "remaining_time": "8:53:52"} +{"current_steps": 232, "total_steps": 1955, "loss": 1.1299, "lr": 1.9979336996376896e-05, "epoch": 0.5933503836317136, "percentage": 11.87, "elapsed_time": "1:11:51", "remaining_time": "8:53:38"} +{"current_steps": 233, "total_steps": 1955, "loss": 1.1344, "lr": 1.99781735316284e-05, "epoch": 0.5959079283887468, "percentage": 11.92, "elapsed_time": "1:12:09", "remaining_time": "8:53:17"} +{"current_steps": 234, "total_steps": 1955, "loss": 1.0922, "lr": 1.997697823817237e-05, "epoch": 0.59846547314578, "percentage": 11.97, "elapsed_time": "1:12:28", "remaining_time": "8:52:59"} +{"current_steps": 235, "total_steps": 1955, "loss": 1.1096, "lr": 1.99757511198216e-05, "epoch": 0.6010230179028133, "percentage": 12.02, "elapsed_time": "1:12:46", "remaining_time": "8:52:41"} +{"current_steps": 236, "total_steps": 1955, "loss": 1.1395, "lr": 1.997449218049039e-05, "epoch": 0.6035805626598465, "percentage": 12.07, "elapsed_time": "1:13:05", "remaining_time": "8:52:20"} +{"current_steps": 237, "total_steps": 1955, "loss": 1.1245, "lr": 1.9973201424194544e-05, "epoch": 0.6061381074168798, "percentage": 12.12, "elapsed_time": "1:13:24", "remaining_time": "8:52:07"} +{"current_steps": 238, "total_steps": 1955, "loss": 1.1064, "lr": 1.997187885505136e-05, "epoch": 0.6086956521739131, "percentage": 12.17, "elapsed_time": "1:13:43", "remaining_time": "8:51:48"} +{"current_steps": 239, "total_steps": 1955, "loss": 1.1517, "lr": 1.997052447727961e-05, "epoch": 0.6112531969309463, "percentage": 12.23, "elapsed_time": "1:14:01", "remaining_time": "8:51:30"} +{"current_steps": 240, "total_steps": 1955, "loss": 1.1178, "lr": 1.996913829519954e-05, "epoch": 0.6138107416879796, "percentage": 12.28, "elapsed_time": "1:14:20", "remaining_time": "8:51:12"} +{"current_steps": 241, "total_steps": 1955, "loss": 1.1105, "lr": 1.996772031323283e-05, "epoch": 0.6163682864450127, "percentage": 12.33, "elapsed_time": "1:14:38", "remaining_time": "8:50:54"} +{"current_steps": 242, "total_steps": 1955, "loss": 1.1373, "lr": 1.9966270535902618e-05, "epoch": 0.618925831202046, "percentage": 12.38, "elapsed_time": "1:14:57", "remaining_time": "8:50:36"} +{"current_steps": 243, "total_steps": 1955, "loss": 1.1335, "lr": 1.9964788967833438e-05, "epoch": 0.6214833759590793, "percentage": 12.43, "elapsed_time": "1:15:15", "remaining_time": "8:50:14"} +{"current_steps": 244, "total_steps": 1955, "loss": 1.1791, "lr": 1.9963275613751255e-05, "epoch": 0.6240409207161125, "percentage": 12.48, "elapsed_time": "1:15:34", "remaining_time": "8:49:56"} +{"current_steps": 245, "total_steps": 1955, "loss": 1.138, "lr": 1.996173047848341e-05, "epoch": 0.6265984654731458, "percentage": 12.53, "elapsed_time": "1:15:52", "remaining_time": "8:49:36"} +{"current_steps": 246, "total_steps": 1955, "loss": 1.1343, "lr": 1.996015356695863e-05, "epoch": 0.629156010230179, "percentage": 12.58, "elapsed_time": "1:16:11", "remaining_time": "8:49:17"} +{"current_steps": 247, "total_steps": 1955, "loss": 1.127, "lr": 1.9958544884207e-05, "epoch": 0.6317135549872123, "percentage": 12.63, "elapsed_time": "1:16:29", "remaining_time": "8:48:59"} +{"current_steps": 248, "total_steps": 1955, "loss": 1.1735, "lr": 1.9956904435359945e-05, "epoch": 0.6342710997442456, "percentage": 12.69, "elapsed_time": "1:16:48", "remaining_time": "8:48:40"} +{"current_steps": 249, "total_steps": 1955, "loss": 1.1535, "lr": 1.9955232225650223e-05, "epoch": 0.6368286445012787, "percentage": 12.74, "elapsed_time": "1:17:07", "remaining_time": "8:48:22"} +{"current_steps": 250, "total_steps": 1955, "loss": 1.1335, "lr": 1.9953528260411912e-05, "epoch": 0.639386189258312, "percentage": 12.79, "elapsed_time": "1:17:25", "remaining_time": "8:48:04"} +{"current_steps": 251, "total_steps": 1955, "loss": 1.1401, "lr": 1.9951792545080368e-05, "epoch": 0.6419437340153452, "percentage": 12.84, "elapsed_time": "1:17:44", "remaining_time": "8:47:45"} +{"current_steps": 252, "total_steps": 1955, "loss": 1.1454, "lr": 1.995002508519223e-05, "epoch": 0.6445012787723785, "percentage": 12.89, "elapsed_time": "1:18:03", "remaining_time": "8:47:27"} +{"current_steps": 253, "total_steps": 1955, "loss": 1.124, "lr": 1.9948225886385415e-05, "epoch": 0.6470588235294118, "percentage": 12.94, "elapsed_time": "1:18:21", "remaining_time": "8:47:10"} +{"current_steps": 254, "total_steps": 1955, "loss": 1.1745, "lr": 1.9946394954399055e-05, "epoch": 0.649616368286445, "percentage": 12.99, "elapsed_time": "1:18:40", "remaining_time": "8:46:52"} +{"current_steps": 255, "total_steps": 1955, "loss": 1.1316, "lr": 1.994453229507352e-05, "epoch": 0.6521739130434783, "percentage": 13.04, "elapsed_time": "1:18:59", "remaining_time": "8:46:35"} +{"current_steps": 256, "total_steps": 1955, "loss": 1.1208, "lr": 1.9942637914350378e-05, "epoch": 0.6547314578005116, "percentage": 13.09, "elapsed_time": "1:19:17", "remaining_time": "8:46:17"} +{"current_steps": 257, "total_steps": 1955, "loss": 1.156, "lr": 1.9940711818272395e-05, "epoch": 0.6572890025575447, "percentage": 13.15, "elapsed_time": "1:19:36", "remaining_time": "8:45:58"} +{"current_steps": 258, "total_steps": 1955, "loss": 1.1182, "lr": 1.993875401298349e-05, "epoch": 0.659846547314578, "percentage": 13.2, "elapsed_time": "1:19:55", "remaining_time": "8:45:40"} +{"current_steps": 259, "total_steps": 1955, "loss": 1.1447, "lr": 1.993676450472874e-05, "epoch": 0.6624040920716112, "percentage": 13.25, "elapsed_time": "1:20:13", "remaining_time": "8:45:22"} +{"current_steps": 260, "total_steps": 1955, "loss": 1.1273, "lr": 1.993474329985434e-05, "epoch": 0.6649616368286445, "percentage": 13.3, "elapsed_time": "1:20:32", "remaining_time": "8:45:04"} +{"current_steps": 261, "total_steps": 1955, "loss": 1.0992, "lr": 1.99326904048076e-05, "epoch": 0.6675191815856778, "percentage": 13.35, "elapsed_time": "1:20:50", "remaining_time": "8:44:42"} +{"current_steps": 262, "total_steps": 1955, "loss": 1.1151, "lr": 1.9930605826136904e-05, "epoch": 0.670076726342711, "percentage": 13.4, "elapsed_time": "1:21:09", "remaining_time": "8:44:24"} +{"current_steps": 263, "total_steps": 1955, "loss": 1.1594, "lr": 1.992848957049172e-05, "epoch": 0.6726342710997443, "percentage": 13.45, "elapsed_time": "1:21:27", "remaining_time": "8:44:06"} +{"current_steps": 264, "total_steps": 1955, "loss": 1.1374, "lr": 1.9926341644622544e-05, "epoch": 0.6751918158567775, "percentage": 13.5, "elapsed_time": "1:21:45", "remaining_time": "8:43:44"} +{"current_steps": 265, "total_steps": 1955, "loss": 1.1654, "lr": 1.9924162055380905e-05, "epoch": 0.6777493606138107, "percentage": 13.55, "elapsed_time": "1:22:04", "remaining_time": "8:43:25"} +{"current_steps": 266, "total_steps": 1955, "loss": 1.1324, "lr": 1.9921950809719326e-05, "epoch": 0.680306905370844, "percentage": 13.61, "elapsed_time": "1:22:23", "remaining_time": "8:43:08"} +{"current_steps": 267, "total_steps": 1955, "loss": 1.0858, "lr": 1.991970791469131e-05, "epoch": 0.6828644501278772, "percentage": 13.66, "elapsed_time": "1:22:41", "remaining_time": "8:42:50"} +{"current_steps": 268, "total_steps": 1955, "loss": 1.13, "lr": 1.991743337745132e-05, "epoch": 0.6854219948849105, "percentage": 13.71, "elapsed_time": "1:23:01", "remaining_time": "8:42:36"} +{"current_steps": 269, "total_steps": 1955, "loss": 1.1065, "lr": 1.991512720525475e-05, "epoch": 0.6879795396419437, "percentage": 13.76, "elapsed_time": "1:23:19", "remaining_time": "8:42:17"} +{"current_steps": 270, "total_steps": 1955, "loss": 1.1469, "lr": 1.9912789405457907e-05, "epoch": 0.690537084398977, "percentage": 13.81, "elapsed_time": "1:23:38", "remaining_time": "8:41:59"} +{"current_steps": 271, "total_steps": 1955, "loss": 1.1456, "lr": 1.9910419985517976e-05, "epoch": 0.6930946291560103, "percentage": 13.86, "elapsed_time": "1:23:56", "remaining_time": "8:41:36"} +{"current_steps": 272, "total_steps": 1955, "loss": 1.1063, "lr": 1.9908018952993017e-05, "epoch": 0.6956521739130435, "percentage": 13.91, "elapsed_time": "1:24:14", "remaining_time": "8:41:14"} +{"current_steps": 273, "total_steps": 1955, "loss": 1.1221, "lr": 1.9905586315541917e-05, "epoch": 0.6982097186700768, "percentage": 13.96, "elapsed_time": "1:24:32", "remaining_time": "8:40:51"} +{"current_steps": 274, "total_steps": 1955, "loss": 1.1442, "lr": 1.9903122080924385e-05, "epoch": 0.7007672634271099, "percentage": 14.02, "elapsed_time": "1:24:50", "remaining_time": "8:40:32"} +{"current_steps": 275, "total_steps": 1955, "loss": 1.1211, "lr": 1.9900626257000924e-05, "epoch": 0.7033248081841432, "percentage": 14.07, "elapsed_time": "1:25:09", "remaining_time": "8:40:14"} +{"current_steps": 276, "total_steps": 1955, "loss": 1.1387, "lr": 1.9898098851732787e-05, "epoch": 0.7058823529411765, "percentage": 14.12, "elapsed_time": "1:25:28", "remaining_time": "8:39:56"} +{"current_steps": 277, "total_steps": 1955, "loss": 1.1531, "lr": 1.989553987318198e-05, "epoch": 0.7084398976982097, "percentage": 14.17, "elapsed_time": "1:25:46", "remaining_time": "8:39:38"} +{"current_steps": 278, "total_steps": 1955, "loss": 1.119, "lr": 1.9892949329511213e-05, "epoch": 0.710997442455243, "percentage": 14.22, "elapsed_time": "1:26:05", "remaining_time": "8:39:20"} +{"current_steps": 279, "total_steps": 1955, "loss": 1.1492, "lr": 1.989032722898389e-05, "epoch": 0.7135549872122762, "percentage": 14.27, "elapsed_time": "1:26:24", "remaining_time": "8:39:05"} +{"current_steps": 280, "total_steps": 1955, "loss": 1.1162, "lr": 1.988767357996408e-05, "epoch": 0.7161125319693095, "percentage": 14.32, "elapsed_time": "1:26:43", "remaining_time": "8:38:47"} +{"current_steps": 281, "total_steps": 1955, "loss": 1.1092, "lr": 1.988498839091647e-05, "epoch": 0.7186700767263428, "percentage": 14.37, "elapsed_time": "1:27:02", "remaining_time": "8:38:30"} +{"current_steps": 282, "total_steps": 1955, "loss": 1.1357, "lr": 1.9882271670406373e-05, "epoch": 0.7212276214833759, "percentage": 14.42, "elapsed_time": "1:27:20", "remaining_time": "8:38:12"} +{"current_steps": 283, "total_steps": 1955, "loss": 1.0653, "lr": 1.9879523427099668e-05, "epoch": 0.7237851662404092, "percentage": 14.48, "elapsed_time": "1:27:39", "remaining_time": "8:37:54"} +{"current_steps": 284, "total_steps": 1955, "loss": 1.131, "lr": 1.9876743669762795e-05, "epoch": 0.7263427109974424, "percentage": 14.53, "elapsed_time": "1:27:58", "remaining_time": "8:37:36"} +{"current_steps": 285, "total_steps": 1955, "loss": 1.1131, "lr": 1.9873932407262715e-05, "epoch": 0.7289002557544757, "percentage": 14.58, "elapsed_time": "1:28:16", "remaining_time": "8:37:14"} +{"current_steps": 286, "total_steps": 1955, "loss": 1.1392, "lr": 1.9871089648566887e-05, "epoch": 0.731457800511509, "percentage": 14.63, "elapsed_time": "1:28:34", "remaining_time": "8:36:56"} +{"current_steps": 287, "total_steps": 1955, "loss": 1.1369, "lr": 1.9868215402743236e-05, "epoch": 0.7340153452685422, "percentage": 14.68, "elapsed_time": "1:28:53", "remaining_time": "8:36:38"} +{"current_steps": 288, "total_steps": 1955, "loss": 1.0986, "lr": 1.9865309678960125e-05, "epoch": 0.7365728900255755, "percentage": 14.73, "elapsed_time": "1:29:12", "remaining_time": "8:36:22"} +{"current_steps": 289, "total_steps": 1955, "loss": 1.1309, "lr": 1.986237248648633e-05, "epoch": 0.7391304347826086, "percentage": 14.78, "elapsed_time": "1:29:31", "remaining_time": "8:36:04"} +{"current_steps": 290, "total_steps": 1955, "loss": 1.1421, "lr": 1.9859403834691006e-05, "epoch": 0.7416879795396419, "percentage": 14.83, "elapsed_time": "1:29:49", "remaining_time": "8:35:45"} +{"current_steps": 291, "total_steps": 1955, "loss": 1.1239, "lr": 1.985640373304365e-05, "epoch": 0.7442455242966752, "percentage": 14.88, "elapsed_time": "1:30:08", "remaining_time": "8:35:27"} +{"current_steps": 292, "total_steps": 1955, "loss": 1.0883, "lr": 1.985337219111409e-05, "epoch": 0.7468030690537084, "percentage": 14.94, "elapsed_time": "1:30:26", "remaining_time": "8:35:05"} +{"current_steps": 293, "total_steps": 1955, "loss": 1.1165, "lr": 1.9850309218572437e-05, "epoch": 0.7493606138107417, "percentage": 14.99, "elapsed_time": "1:30:45", "remaining_time": "8:34:47"} +{"current_steps": 294, "total_steps": 1955, "loss": 1.1125, "lr": 1.984721482518907e-05, "epoch": 0.7519181585677749, "percentage": 15.04, "elapsed_time": "1:31:03", "remaining_time": "8:34:28"} +{"current_steps": 295, "total_steps": 1955, "loss": 1.1041, "lr": 1.9844089020834573e-05, "epoch": 0.7544757033248082, "percentage": 15.09, "elapsed_time": "1:31:22", "remaining_time": "8:34:10"} +{"current_steps": 296, "total_steps": 1955, "loss": 1.1122, "lr": 1.9840931815479747e-05, "epoch": 0.7570332480818415, "percentage": 15.14, "elapsed_time": "1:31:41", "remaining_time": "8:33:52"} +{"current_steps": 297, "total_steps": 1955, "loss": 1.1508, "lr": 1.9837743219195554e-05, "epoch": 0.7595907928388747, "percentage": 15.19, "elapsed_time": "1:31:59", "remaining_time": "8:33:33"} +{"current_steps": 298, "total_steps": 1955, "loss": 1.1296, "lr": 1.983452324215308e-05, "epoch": 0.7621483375959079, "percentage": 15.24, "elapsed_time": "1:32:18", "remaining_time": "8:33:16"} +{"current_steps": 299, "total_steps": 1955, "loss": 1.1396, "lr": 1.983127189462351e-05, "epoch": 0.7647058823529411, "percentage": 15.29, "elapsed_time": "1:32:37", "remaining_time": "8:32:57"} +{"current_steps": 300, "total_steps": 1955, "loss": 1.1025, "lr": 1.9827989186978105e-05, "epoch": 0.7672634271099744, "percentage": 15.35, "elapsed_time": "1:32:55", "remaining_time": "8:32:40"} +{"current_steps": 301, "total_steps": 1955, "loss": 1.1349, "lr": 1.9824675129688152e-05, "epoch": 0.7698209718670077, "percentage": 15.4, "elapsed_time": "1:33:14", "remaining_time": "8:32:22"} +{"current_steps": 302, "total_steps": 1955, "loss": 1.1333, "lr": 1.9821329733324943e-05, "epoch": 0.7723785166240409, "percentage": 15.45, "elapsed_time": "1:33:33", "remaining_time": "8:32:04"} +{"current_steps": 303, "total_steps": 1955, "loss": 1.1024, "lr": 1.9817953008559733e-05, "epoch": 0.7749360613810742, "percentage": 15.5, "elapsed_time": "1:33:51", "remaining_time": "8:31:46"} +{"current_steps": 304, "total_steps": 1955, "loss": 1.1046, "lr": 1.9814544966163708e-05, "epoch": 0.7774936061381074, "percentage": 15.55, "elapsed_time": "1:34:10", "remaining_time": "8:31:28"} +{"current_steps": 305, "total_steps": 1955, "loss": 1.1209, "lr": 1.9811105617007958e-05, "epoch": 0.7800511508951407, "percentage": 15.6, "elapsed_time": "1:34:29", "remaining_time": "8:31:10"} +{"current_steps": 306, "total_steps": 1955, "loss": 1.1121, "lr": 1.980763497206343e-05, "epoch": 0.782608695652174, "percentage": 15.65, "elapsed_time": "1:34:48", "remaining_time": "8:30:52"} +{"current_steps": 307, "total_steps": 1955, "loss": 1.0798, "lr": 1.98041330424009e-05, "epoch": 0.7851662404092071, "percentage": 15.7, "elapsed_time": "1:35:06", "remaining_time": "8:30:30"} +{"current_steps": 308, "total_steps": 1955, "loss": 1.1121, "lr": 1.9800599839190942e-05, "epoch": 0.7877237851662404, "percentage": 15.75, "elapsed_time": "1:35:24", "remaining_time": "8:30:08"} +{"current_steps": 309, "total_steps": 1955, "loss": 1.1301, "lr": 1.979703537370388e-05, "epoch": 0.7902813299232737, "percentage": 15.81, "elapsed_time": "1:35:42", "remaining_time": "8:29:51"} +{"current_steps": 310, "total_steps": 1955, "loss": 1.0972, "lr": 1.979343965730977e-05, "epoch": 0.7928388746803069, "percentage": 15.86, "elapsed_time": "1:36:01", "remaining_time": "8:29:33"} +{"current_steps": 311, "total_steps": 1955, "loss": 1.1292, "lr": 1.9789812701478346e-05, "epoch": 0.7953964194373402, "percentage": 15.91, "elapsed_time": "1:36:20", "remaining_time": "8:29:14"} +{"current_steps": 312, "total_steps": 1955, "loss": 1.1467, "lr": 1.9786154517778987e-05, "epoch": 0.7979539641943734, "percentage": 15.96, "elapsed_time": "1:36:38", "remaining_time": "8:28:56"} +{"current_steps": 313, "total_steps": 1955, "loss": 1.109, "lr": 1.9782465117880694e-05, "epoch": 0.8005115089514067, "percentage": 16.01, "elapsed_time": "1:36:56", "remaining_time": "8:28:35"} +{"current_steps": 314, "total_steps": 1955, "loss": 1.1127, "lr": 1.977874451355203e-05, "epoch": 0.80306905370844, "percentage": 16.06, "elapsed_time": "1:37:14", "remaining_time": "8:28:09"} +{"current_steps": 315, "total_steps": 1955, "loss": 1.1059, "lr": 1.9774992716661106e-05, "epoch": 0.8056265984654731, "percentage": 16.11, "elapsed_time": "1:37:32", "remaining_time": "8:27:51"} +{"current_steps": 316, "total_steps": 1955, "loss": 1.102, "lr": 1.9771209739175522e-05, "epoch": 0.8081841432225064, "percentage": 16.16, "elapsed_time": "1:37:50", "remaining_time": "8:27:30"} +{"current_steps": 317, "total_steps": 1955, "loss": 1.131, "lr": 1.9767395593162356e-05, "epoch": 0.8107416879795396, "percentage": 16.21, "elapsed_time": "1:38:09", "remaining_time": "8:27:12"} +{"current_steps": 318, "total_steps": 1955, "loss": 1.1236, "lr": 1.9763550290788086e-05, "epoch": 0.8132992327365729, "percentage": 16.27, "elapsed_time": "1:38:28", "remaining_time": "8:26:54"} +{"current_steps": 319, "total_steps": 1955, "loss": 1.0826, "lr": 1.9759673844318592e-05, "epoch": 0.8158567774936062, "percentage": 16.32, "elapsed_time": "1:38:46", "remaining_time": "8:26:35"} +{"current_steps": 320, "total_steps": 1955, "loss": 1.1101, "lr": 1.9755766266119085e-05, "epoch": 0.8184143222506394, "percentage": 16.37, "elapsed_time": "1:39:05", "remaining_time": "8:26:17"} +{"current_steps": 321, "total_steps": 1955, "loss": 1.0896, "lr": 1.975182756865409e-05, "epoch": 0.8209718670076727, "percentage": 16.42, "elapsed_time": "1:39:24", "remaining_time": "8:25:59"} +{"current_steps": 322, "total_steps": 1955, "loss": 1.119, "lr": 1.9747857764487395e-05, "epoch": 0.8235294117647058, "percentage": 16.47, "elapsed_time": "1:39:42", "remaining_time": "8:25:41"} +{"current_steps": 323, "total_steps": 1955, "loss": 1.139, "lr": 1.974385686628201e-05, "epoch": 0.8260869565217391, "percentage": 16.52, "elapsed_time": "1:40:00", "remaining_time": "8:25:19"} +{"current_steps": 324, "total_steps": 1955, "loss": 1.1387, "lr": 1.9739824886800134e-05, "epoch": 0.8286445012787724, "percentage": 16.57, "elapsed_time": "1:40:19", "remaining_time": "8:25:01"} +{"current_steps": 325, "total_steps": 1955, "loss": 1.0992, "lr": 1.9735761838903106e-05, "epoch": 0.8312020460358056, "percentage": 16.62, "elapsed_time": "1:40:38", "remaining_time": "8:24:43"} +{"current_steps": 326, "total_steps": 1955, "loss": 1.1099, "lr": 1.9731667735551376e-05, "epoch": 0.8337595907928389, "percentage": 16.68, "elapsed_time": "1:40:56", "remaining_time": "8:24:25"} +{"current_steps": 327, "total_steps": 1955, "loss": 1.1504, "lr": 1.9727542589804446e-05, "epoch": 0.8363171355498721, "percentage": 16.73, "elapsed_time": "1:41:15", "remaining_time": "8:24:07"} +{"current_steps": 328, "total_steps": 1955, "loss": 1.1214, "lr": 1.9723386414820844e-05, "epoch": 0.8388746803069054, "percentage": 16.78, "elapsed_time": "1:41:34", "remaining_time": "8:23:48"} +{"current_steps": 329, "total_steps": 1955, "loss": 1.1606, "lr": 1.971919922385807e-05, "epoch": 0.8414322250639387, "percentage": 16.83, "elapsed_time": "1:41:52", "remaining_time": "8:23:30"} +{"current_steps": 330, "total_steps": 1955, "loss": 1.1306, "lr": 1.9714981030272566e-05, "epoch": 0.8439897698209718, "percentage": 16.88, "elapsed_time": "1:42:11", "remaining_time": "8:23:13"} +{"current_steps": 331, "total_steps": 1955, "loss": 1.1045, "lr": 1.9710731847519665e-05, "epoch": 0.8465473145780051, "percentage": 16.93, "elapsed_time": "1:42:30", "remaining_time": "8:22:55"} +{"current_steps": 332, "total_steps": 1955, "loss": 1.0964, "lr": 1.9706451689153556e-05, "epoch": 0.8491048593350383, "percentage": 16.98, "elapsed_time": "1:42:49", "remaining_time": "8:22:37"} +{"current_steps": 333, "total_steps": 1955, "loss": 1.1434, "lr": 1.9702140568827224e-05, "epoch": 0.8516624040920716, "percentage": 17.03, "elapsed_time": "1:43:07", "remaining_time": "8:22:16"} +{"current_steps": 334, "total_steps": 1955, "loss": 1.0752, "lr": 1.9697798500292422e-05, "epoch": 0.8542199488491049, "percentage": 17.08, "elapsed_time": "1:43:25", "remaining_time": "8:21:58"} +{"current_steps": 335, "total_steps": 1955, "loss": 1.102, "lr": 1.969342549739963e-05, "epoch": 0.8567774936061381, "percentage": 17.14, "elapsed_time": "1:43:44", "remaining_time": "8:21:40"} +{"current_steps": 336, "total_steps": 1955, "loss": 1.1016, "lr": 1.9689021574097988e-05, "epoch": 0.8593350383631714, "percentage": 17.19, "elapsed_time": "1:44:02", "remaining_time": "8:21:19"} +{"current_steps": 337, "total_steps": 1955, "loss": 1.125, "lr": 1.9684586744435285e-05, "epoch": 0.8618925831202046, "percentage": 17.24, "elapsed_time": "1:44:21", "remaining_time": "8:21:01"} +{"current_steps": 338, "total_steps": 1955, "loss": 1.1295, "lr": 1.9680121022557882e-05, "epoch": 0.8644501278772379, "percentage": 17.29, "elapsed_time": "1:44:39", "remaining_time": "8:20:42"} +{"current_steps": 339, "total_steps": 1955, "loss": 1.0954, "lr": 1.9675624422710685e-05, "epoch": 0.8670076726342711, "percentage": 17.34, "elapsed_time": "1:44:58", "remaining_time": "8:20:23"} +{"current_steps": 340, "total_steps": 1955, "loss": 1.1348, "lr": 1.9671096959237102e-05, "epoch": 0.8695652173913043, "percentage": 17.39, "elapsed_time": "1:45:16", "remaining_time": "8:20:05"} +{"current_steps": 341, "total_steps": 1955, "loss": 1.091, "lr": 1.9666538646578978e-05, "epoch": 0.8721227621483376, "percentage": 17.44, "elapsed_time": "1:45:35", "remaining_time": "8:19:47"} +{"current_steps": 342, "total_steps": 1955, "loss": 1.1175, "lr": 1.9661949499276577e-05, "epoch": 0.8746803069053708, "percentage": 17.49, "elapsed_time": "1:45:54", "remaining_time": "8:19:28"} +{"current_steps": 343, "total_steps": 1955, "loss": 1.1149, "lr": 1.965732953196851e-05, "epoch": 0.8772378516624041, "percentage": 17.54, "elapsed_time": "1:46:12", "remaining_time": "8:19:10"} +{"current_steps": 344, "total_steps": 1955, "loss": 1.1224, "lr": 1.96526787593917e-05, "epoch": 0.8797953964194374, "percentage": 17.6, "elapsed_time": "1:46:30", "remaining_time": "8:18:48"} +{"current_steps": 345, "total_steps": 1955, "loss": 1.1039, "lr": 1.964799719638134e-05, "epoch": 0.8823529411764706, "percentage": 17.65, "elapsed_time": "1:46:49", "remaining_time": "8:18:30"} +{"current_steps": 346, "total_steps": 1955, "loss": 1.1686, "lr": 1.9643284857870824e-05, "epoch": 0.8849104859335039, "percentage": 17.7, "elapsed_time": "1:47:08", "remaining_time": "8:18:12"} +{"current_steps": 347, "total_steps": 1955, "loss": 1.0509, "lr": 1.9638541758891733e-05, "epoch": 0.887468030690537, "percentage": 17.75, "elapsed_time": "1:47:26", "remaining_time": "8:17:54"} +{"current_steps": 348, "total_steps": 1955, "loss": 1.1162, "lr": 1.963376791457376e-05, "epoch": 0.8900255754475703, "percentage": 17.8, "elapsed_time": "1:47:45", "remaining_time": "8:17:36"} +{"current_steps": 349, "total_steps": 1955, "loss": 1.1, "lr": 1.962896334014467e-05, "epoch": 0.8925831202046036, "percentage": 17.85, "elapsed_time": "1:48:03", "remaining_time": "8:17:15"} +{"current_steps": 350, "total_steps": 1955, "loss": 1.1228, "lr": 1.9624128050930252e-05, "epoch": 0.8951406649616368, "percentage": 17.9, "elapsed_time": "1:48:22", "remaining_time": "8:16:57"} +{"current_steps": 351, "total_steps": 1955, "loss": 1.1403, "lr": 1.9619262062354273e-05, "epoch": 0.8976982097186701, "percentage": 17.95, "elapsed_time": "1:48:40", "remaining_time": "8:16:35"} +{"current_steps": 352, "total_steps": 1955, "loss": 1.0943, "lr": 1.9614365389938425e-05, "epoch": 0.9002557544757033, "percentage": 18.01, "elapsed_time": "1:48:58", "remaining_time": "8:16:17"} +{"current_steps": 353, "total_steps": 1955, "loss": 1.1167, "lr": 1.9609438049302274e-05, "epoch": 0.9028132992327366, "percentage": 18.06, "elapsed_time": "1:49:17", "remaining_time": "8:15:59"} +{"current_steps": 354, "total_steps": 1955, "loss": 1.1109, "lr": 1.9604480056163215e-05, "epoch": 0.9053708439897699, "percentage": 18.11, "elapsed_time": "1:49:36", "remaining_time": "8:15:41"} +{"current_steps": 355, "total_steps": 1955, "loss": 1.0884, "lr": 1.9599491426336415e-05, "epoch": 0.907928388746803, "percentage": 18.16, "elapsed_time": "1:49:55", "remaining_time": "8:15:24"} +{"current_steps": 356, "total_steps": 1955, "loss": 1.1013, "lr": 1.9594472175734776e-05, "epoch": 0.9104859335038363, "percentage": 18.21, "elapsed_time": "1:50:13", "remaining_time": "8:15:07"} +{"current_steps": 357, "total_steps": 1955, "loss": 1.112, "lr": 1.9589422320368863e-05, "epoch": 0.9130434782608695, "percentage": 18.26, "elapsed_time": "1:50:32", "remaining_time": "8:14:50"} +{"current_steps": 358, "total_steps": 1955, "loss": 1.1338, "lr": 1.9584341876346874e-05, "epoch": 0.9156010230179028, "percentage": 18.31, "elapsed_time": "1:50:51", "remaining_time": "8:14:32"} +{"current_steps": 359, "total_steps": 1955, "loss": 1.115, "lr": 1.957923085987458e-05, "epoch": 0.9181585677749361, "percentage": 18.36, "elapsed_time": "1:51:10", "remaining_time": "8:14:14"} +{"current_steps": 360, "total_steps": 1955, "loss": 1.0819, "lr": 1.957408928725527e-05, "epoch": 0.9207161125319693, "percentage": 18.41, "elapsed_time": "1:51:28", "remaining_time": "8:13:55"} +{"current_steps": 361, "total_steps": 1955, "loss": 1.1301, "lr": 1.9568917174889696e-05, "epoch": 0.9232736572890026, "percentage": 18.47, "elapsed_time": "1:51:47", "remaining_time": "8:13:37"} +{"current_steps": 362, "total_steps": 1955, "loss": 1.0582, "lr": 1.9563714539276037e-05, "epoch": 0.9258312020460358, "percentage": 18.52, "elapsed_time": "1:52:06", "remaining_time": "8:13:19"} +{"current_steps": 363, "total_steps": 1955, "loss": 1.1201, "lr": 1.9558481397009832e-05, "epoch": 0.928388746803069, "percentage": 18.57, "elapsed_time": "1:52:24", "remaining_time": "8:13:01"} +{"current_steps": 364, "total_steps": 1955, "loss": 1.1031, "lr": 1.9553217764783928e-05, "epoch": 0.9309462915601023, "percentage": 18.62, "elapsed_time": "1:52:43", "remaining_time": "8:12:42"} +{"current_steps": 365, "total_steps": 1955, "loss": 1.1201, "lr": 1.954792365938844e-05, "epoch": 0.9335038363171355, "percentage": 18.67, "elapsed_time": "1:53:02", "remaining_time": "8:12:24"} +{"current_steps": 366, "total_steps": 1955, "loss": 1.1431, "lr": 1.9542599097710677e-05, "epoch": 0.9360613810741688, "percentage": 18.72, "elapsed_time": "1:53:20", "remaining_time": "8:12:06"} +{"current_steps": 367, "total_steps": 1955, "loss": 1.1213, "lr": 1.9537244096735096e-05, "epoch": 0.9386189258312021, "percentage": 18.77, "elapsed_time": "1:53:38", "remaining_time": "8:11:45"} +{"current_steps": 368, "total_steps": 1955, "loss": 1.1147, "lr": 1.9531858673543266e-05, "epoch": 0.9411764705882353, "percentage": 18.82, "elapsed_time": "1:53:57", "remaining_time": "8:11:27"} +{"current_steps": 369, "total_steps": 1955, "loss": 1.1096, "lr": 1.952644284531378e-05, "epoch": 0.9437340153452686, "percentage": 18.87, "elapsed_time": "1:54:15", "remaining_time": "8:11:06"} +{"current_steps": 370, "total_steps": 1955, "loss": 1.1056, "lr": 1.9520996629322228e-05, "epoch": 0.9462915601023018, "percentage": 18.93, "elapsed_time": "1:54:33", "remaining_time": "8:10:45"} +{"current_steps": 371, "total_steps": 1955, "loss": 1.1329, "lr": 1.9515520042941133e-05, "epoch": 0.948849104859335, "percentage": 18.98, "elapsed_time": "1:54:52", "remaining_time": "8:10:26"} +{"current_steps": 372, "total_steps": 1955, "loss": 1.1139, "lr": 1.9510013103639885e-05, "epoch": 0.9514066496163683, "percentage": 19.03, "elapsed_time": "1:55:10", "remaining_time": "8:10:08"} +{"current_steps": 373, "total_steps": 1955, "loss": 1.1172, "lr": 1.9504475828984706e-05, "epoch": 0.9539641943734015, "percentage": 19.08, "elapsed_time": "1:55:29", "remaining_time": "8:09:50"} +{"current_steps": 374, "total_steps": 1955, "loss": 1.117, "lr": 1.9498908236638572e-05, "epoch": 0.9565217391304348, "percentage": 19.13, "elapsed_time": "1:55:48", "remaining_time": "8:09:31"} +{"current_steps": 375, "total_steps": 1955, "loss": 1.1392, "lr": 1.949331034436118e-05, "epoch": 0.959079283887468, "percentage": 19.18, "elapsed_time": "1:56:06", "remaining_time": "8:09:10"} +{"current_steps": 376, "total_steps": 1955, "loss": 1.0888, "lr": 1.9487682170008864e-05, "epoch": 0.9616368286445013, "percentage": 19.23, "elapsed_time": "1:56:24", "remaining_time": "8:08:52"} +{"current_steps": 377, "total_steps": 1955, "loss": 1.1222, "lr": 1.948202373153457e-05, "epoch": 0.9641943734015346, "percentage": 19.28, "elapsed_time": "1:56:43", "remaining_time": "8:08:34"} +{"current_steps": 378, "total_steps": 1955, "loss": 1.1274, "lr": 1.9476335046987765e-05, "epoch": 0.9667519181585678, "percentage": 19.34, "elapsed_time": "1:57:02", "remaining_time": "8:08:15"} +{"current_steps": 379, "total_steps": 1955, "loss": 1.0955, "lr": 1.9470616134514405e-05, "epoch": 0.969309462915601, "percentage": 19.39, "elapsed_time": "1:57:20", "remaining_time": "8:07:54"} +{"current_steps": 380, "total_steps": 1955, "loss": 1.0795, "lr": 1.9464867012356866e-05, "epoch": 0.9718670076726342, "percentage": 19.44, "elapsed_time": "1:57:38", "remaining_time": "8:07:35"} +{"current_steps": 381, "total_steps": 1955, "loss": 1.1092, "lr": 1.9459087698853884e-05, "epoch": 0.9744245524296675, "percentage": 19.49, "elapsed_time": "1:57:57", "remaining_time": "8:07:17"} +{"current_steps": 382, "total_steps": 1955, "loss": 1.113, "lr": 1.945327821244051e-05, "epoch": 0.9769820971867008, "percentage": 19.54, "elapsed_time": "1:58:15", "remaining_time": "8:06:59"} +{"current_steps": 383, "total_steps": 1955, "loss": 1.1169, "lr": 1.944743857164803e-05, "epoch": 0.979539641943734, "percentage": 19.59, "elapsed_time": "1:58:34", "remaining_time": "8:06:41"} +{"current_steps": 384, "total_steps": 1955, "loss": 1.0704, "lr": 1.9441568795103932e-05, "epoch": 0.9820971867007673, "percentage": 19.64, "elapsed_time": "1:58:53", "remaining_time": "8:06:23"} +{"current_steps": 385, "total_steps": 1955, "loss": 1.1517, "lr": 1.9435668901531813e-05, "epoch": 0.9846547314578005, "percentage": 19.69, "elapsed_time": "1:59:12", "remaining_time": "8:06:05"} +{"current_steps": 386, "total_steps": 1955, "loss": 1.1052, "lr": 1.9429738909751355e-05, "epoch": 0.9872122762148338, "percentage": 19.74, "elapsed_time": "1:59:30", "remaining_time": "8:05:47"} +{"current_steps": 387, "total_steps": 1955, "loss": 1.1192, "lr": 1.9423778838678236e-05, "epoch": 0.989769820971867, "percentage": 19.8, "elapsed_time": "1:59:49", "remaining_time": "8:05:29"} +{"current_steps": 388, "total_steps": 1955, "loss": 1.133, "lr": 1.9417788707324095e-05, "epoch": 0.9923273657289002, "percentage": 19.85, "elapsed_time": "2:00:08", "remaining_time": "8:05:11"} +{"current_steps": 389, "total_steps": 1955, "loss": 1.1161, "lr": 1.9411768534796445e-05, "epoch": 0.9948849104859335, "percentage": 19.9, "elapsed_time": "2:00:26", "remaining_time": "8:04:52"} +{"current_steps": 390, "total_steps": 1955, "loss": 1.0955, "lr": 1.9405718340298633e-05, "epoch": 0.9974424552429667, "percentage": 19.95, "elapsed_time": "2:00:45", "remaining_time": "8:04:35"} +{"current_steps": 391, "total_steps": 1955, "loss": 1.1251, "lr": 1.9399638143129768e-05, "epoch": 1.0, "percentage": 20.0, "elapsed_time": "2:01:04", "remaining_time": "8:04:16"} +{"current_steps": 392, "total_steps": 1955, "loss": 1.1002, "lr": 1.9393527962684665e-05, "epoch": 1.0025575447570332, "percentage": 20.05, "elapsed_time": "2:01:34", "remaining_time": "8:04:44"} +{"current_steps": 393, "total_steps": 1955, "loss": 1.1167, "lr": 1.938738781845378e-05, "epoch": 1.0051150895140666, "percentage": 20.1, "elapsed_time": "2:01:52", "remaining_time": "8:04:25"} +{"current_steps": 394, "total_steps": 1955, "loss": 1.0882, "lr": 1.938121773002315e-05, "epoch": 1.0076726342710998, "percentage": 20.15, "elapsed_time": "2:02:11", "remaining_time": "8:04:06"} +{"current_steps": 395, "total_steps": 1955, "loss": 1.0969, "lr": 1.9375017717074322e-05, "epoch": 1.010230179028133, "percentage": 20.2, "elapsed_time": "2:02:30", "remaining_time": "8:03:47"} +{"current_steps": 396, "total_steps": 1955, "loss": 1.1139, "lr": 1.9368787799384303e-05, "epoch": 1.0127877237851663, "percentage": 20.26, "elapsed_time": "2:02:48", "remaining_time": "8:03:27"} +{"current_steps": 397, "total_steps": 1955, "loss": 1.1153, "lr": 1.936252799682549e-05, "epoch": 1.0153452685421995, "percentage": 20.31, "elapsed_time": "2:03:06", "remaining_time": "8:03:09"} +{"current_steps": 398, "total_steps": 1955, "loss": 1.1053, "lr": 1.9356238329365614e-05, "epoch": 1.0179028132992327, "percentage": 20.36, "elapsed_time": "2:03:25", "remaining_time": "8:02:50"} +{"current_steps": 399, "total_steps": 1955, "loss": 1.0765, "lr": 1.9349918817067658e-05, "epoch": 1.020460358056266, "percentage": 20.41, "elapsed_time": "2:03:43", "remaining_time": "8:02:31"} +{"current_steps": 400, "total_steps": 1955, "loss": 1.1314, "lr": 1.934356948008981e-05, "epoch": 1.0230179028132993, "percentage": 20.46, "elapsed_time": "2:04:02", "remaining_time": "8:02:12"} +{"current_steps": 401, "total_steps": 1955, "loss": 1.11, "lr": 1.93371903386854e-05, "epoch": 1.0255754475703325, "percentage": 20.51, "elapsed_time": "2:04:21", "remaining_time": "8:01:55"} +{"current_steps": 402, "total_steps": 1955, "loss": 1.1286, "lr": 1.933078141320282e-05, "epoch": 1.0281329923273657, "percentage": 20.56, "elapsed_time": "2:04:40", "remaining_time": "8:01:36"} +{"current_steps": 403, "total_steps": 1955, "loss": 1.0938, "lr": 1.9324342724085473e-05, "epoch": 1.030690537084399, "percentage": 20.61, "elapsed_time": "2:04:58", "remaining_time": "8:01:18"} +{"current_steps": 404, "total_steps": 1955, "loss": 1.1004, "lr": 1.9317874291871703e-05, "epoch": 1.0332480818414322, "percentage": 20.66, "elapsed_time": "2:05:17", "remaining_time": "8:00:59"} +{"current_steps": 405, "total_steps": 1955, "loss": 1.0987, "lr": 1.931137613719473e-05, "epoch": 1.0358056265984654, "percentage": 20.72, "elapsed_time": "2:05:35", "remaining_time": "8:00:41"} +{"current_steps": 406, "total_steps": 1955, "loss": 1.1313, "lr": 1.9304848280782582e-05, "epoch": 1.0383631713554988, "percentage": 20.77, "elapsed_time": "2:05:54", "remaining_time": "8:00:22"} +{"current_steps": 407, "total_steps": 1955, "loss": 1.0649, "lr": 1.9298290743458027e-05, "epoch": 1.040920716112532, "percentage": 20.82, "elapsed_time": "2:06:13", "remaining_time": "8:00:05"} +{"current_steps": 408, "total_steps": 1955, "loss": 1.0967, "lr": 1.929170354613852e-05, "epoch": 1.0434782608695652, "percentage": 20.87, "elapsed_time": "2:06:32", "remaining_time": "7:59:46"} +{"current_steps": 409, "total_steps": 1955, "loss": 1.0775, "lr": 1.9285086709836115e-05, "epoch": 1.0460358056265984, "percentage": 20.92, "elapsed_time": "2:06:50", "remaining_time": "7:59:27"} +{"current_steps": 410, "total_steps": 1955, "loss": 1.0716, "lr": 1.927844025565742e-05, "epoch": 1.0485933503836318, "percentage": 20.97, "elapsed_time": "2:07:09", "remaining_time": "7:59:08"} +{"current_steps": 411, "total_steps": 1955, "loss": 1.0984, "lr": 1.9271764204803512e-05, "epoch": 1.051150895140665, "percentage": 21.02, "elapsed_time": "2:07:27", "remaining_time": "7:58:49"} +{"current_steps": 412, "total_steps": 1955, "loss": 1.079, "lr": 1.9265058578569878e-05, "epoch": 1.0537084398976981, "percentage": 21.07, "elapsed_time": "2:07:45", "remaining_time": "7:58:28"} +{"current_steps": 413, "total_steps": 1955, "loss": 1.0881, "lr": 1.9258323398346345e-05, "epoch": 1.0562659846547315, "percentage": 21.13, "elapsed_time": "2:08:04", "remaining_time": "7:58:10"} +{"current_steps": 414, "total_steps": 1955, "loss": 1.1338, "lr": 1.9251558685617015e-05, "epoch": 1.0588235294117647, "percentage": 21.18, "elapsed_time": "2:08:22", "remaining_time": "7:57:51"} +{"current_steps": 415, "total_steps": 1955, "loss": 1.1055, "lr": 1.9244764461960193e-05, "epoch": 1.061381074168798, "percentage": 21.23, "elapsed_time": "2:08:41", "remaining_time": "7:57:33"} +{"current_steps": 416, "total_steps": 1955, "loss": 1.0866, "lr": 1.923794074904832e-05, "epoch": 1.0639386189258313, "percentage": 21.28, "elapsed_time": "2:09:00", "remaining_time": "7:57:15"} +{"current_steps": 417, "total_steps": 1955, "loss": 1.0705, "lr": 1.9231087568647892e-05, "epoch": 1.0664961636828645, "percentage": 21.33, "elapsed_time": "2:09:18", "remaining_time": "7:56:56"} +{"current_steps": 418, "total_steps": 1955, "loss": 1.0802, "lr": 1.9224204942619418e-05, "epoch": 1.0690537084398977, "percentage": 21.38, "elapsed_time": "2:09:37", "remaining_time": "7:56:38"} +{"current_steps": 419, "total_steps": 1955, "loss": 1.1369, "lr": 1.9217292892917326e-05, "epoch": 1.0716112531969308, "percentage": 21.43, "elapsed_time": "2:09:55", "remaining_time": "7:56:17"} +{"current_steps": 420, "total_steps": 1955, "loss": 1.0851, "lr": 1.9210351441589898e-05, "epoch": 1.0741687979539642, "percentage": 21.48, "elapsed_time": "2:10:13", "remaining_time": "7:55:57"} +{"current_steps": 421, "total_steps": 1955, "loss": 1.0983, "lr": 1.92033806107792e-05, "epoch": 1.0767263427109974, "percentage": 21.53, "elapsed_time": "2:10:32", "remaining_time": "7:55:38"} +{"current_steps": 422, "total_steps": 1955, "loss": 1.0899, "lr": 1.919638042272103e-05, "epoch": 1.0792838874680306, "percentage": 21.59, "elapsed_time": "2:10:51", "remaining_time": "7:55:21"} +{"current_steps": 423, "total_steps": 1955, "loss": 1.0748, "lr": 1.9189350899744807e-05, "epoch": 1.081841432225064, "percentage": 21.64, "elapsed_time": "2:11:09", "remaining_time": "7:55:02"} +{"current_steps": 424, "total_steps": 1955, "loss": 1.1209, "lr": 1.9182292064273544e-05, "epoch": 1.0843989769820972, "percentage": 21.69, "elapsed_time": "2:11:28", "remaining_time": "7:54:44"} +{"current_steps": 425, "total_steps": 1955, "loss": 1.0853, "lr": 1.9175203938823747e-05, "epoch": 1.0869565217391304, "percentage": 21.74, "elapsed_time": "2:11:47", "remaining_time": "7:54:25"} +{"current_steps": 426, "total_steps": 1955, "loss": 1.0746, "lr": 1.9168086546005348e-05, "epoch": 1.0895140664961638, "percentage": 21.79, "elapsed_time": "2:12:05", "remaining_time": "7:54:06"} +{"current_steps": 427, "total_steps": 1955, "loss": 1.0991, "lr": 1.9160939908521642e-05, "epoch": 1.092071611253197, "percentage": 21.84, "elapsed_time": "2:12:24", "remaining_time": "7:53:48"} +{"current_steps": 428, "total_steps": 1955, "loss": 1.0799, "lr": 1.9153764049169213e-05, "epoch": 1.0946291560102301, "percentage": 21.89, "elapsed_time": "2:12:42", "remaining_time": "7:53:27"} +{"current_steps": 429, "total_steps": 1955, "loss": 1.1098, "lr": 1.9146558990837853e-05, "epoch": 1.0971867007672633, "percentage": 21.94, "elapsed_time": "2:13:00", "remaining_time": "7:53:06"} +{"current_steps": 430, "total_steps": 1955, "loss": 1.0854, "lr": 1.9139324756510496e-05, "epoch": 1.0997442455242967, "percentage": 21.99, "elapsed_time": "2:13:18", "remaining_time": "7:52:46"} +{"current_steps": 431, "total_steps": 1955, "loss": 1.1157, "lr": 1.913206136926314e-05, "epoch": 1.10230179028133, "percentage": 22.05, "elapsed_time": "2:13:37", "remaining_time": "7:52:29"} +{"current_steps": 432, "total_steps": 1955, "loss": 1.134, "lr": 1.9124768852264774e-05, "epoch": 1.104859335038363, "percentage": 22.1, "elapsed_time": "2:13:55", "remaining_time": "7:52:09"} +{"current_steps": 433, "total_steps": 1955, "loss": 1.1111, "lr": 1.9117447228777317e-05, "epoch": 1.1074168797953965, "percentage": 22.15, "elapsed_time": "2:14:14", "remaining_time": "7:51:50"} +{"current_steps": 434, "total_steps": 1955, "loss": 1.0996, "lr": 1.9110096522155525e-05, "epoch": 1.1099744245524297, "percentage": 22.2, "elapsed_time": "2:14:32", "remaining_time": "7:51:31"} +{"current_steps": 435, "total_steps": 1955, "loss": 1.0916, "lr": 1.9102716755846916e-05, "epoch": 1.1125319693094629, "percentage": 22.25, "elapsed_time": "2:14:51", "remaining_time": "7:51:13"} +{"current_steps": 436, "total_steps": 1955, "loss": 1.0894, "lr": 1.909530795339172e-05, "epoch": 1.1150895140664963, "percentage": 22.3, "elapsed_time": "2:15:10", "remaining_time": "7:50:55"} +{"current_steps": 437, "total_steps": 1955, "loss": 1.0702, "lr": 1.9087870138422777e-05, "epoch": 1.1176470588235294, "percentage": 22.35, "elapsed_time": "2:15:28", "remaining_time": "7:50:36"} +{"current_steps": 438, "total_steps": 1955, "loss": 1.0831, "lr": 1.9080403334665475e-05, "epoch": 1.1202046035805626, "percentage": 22.4, "elapsed_time": "2:15:47", "remaining_time": "7:50:19"} +{"current_steps": 439, "total_steps": 1955, "loss": 1.1038, "lr": 1.9072907565937675e-05, "epoch": 1.1227621483375958, "percentage": 22.46, "elapsed_time": "2:16:05", "remaining_time": "7:49:58"} +{"current_steps": 440, "total_steps": 1955, "loss": 1.0813, "lr": 1.9065382856149623e-05, "epoch": 1.1253196930946292, "percentage": 22.51, "elapsed_time": "2:16:23", "remaining_time": "7:49:38"} +{"current_steps": 441, "total_steps": 1955, "loss": 1.1008, "lr": 1.905782922930389e-05, "epoch": 1.1278772378516624, "percentage": 22.56, "elapsed_time": "2:16:42", "remaining_time": "7:49:19"} +{"current_steps": 442, "total_steps": 1955, "loss": 1.0811, "lr": 1.9050246709495283e-05, "epoch": 1.1304347826086956, "percentage": 22.61, "elapsed_time": "2:17:00", "remaining_time": "7:49:00"} +{"current_steps": 443, "total_steps": 1955, "loss": 1.1057, "lr": 1.904263532091077e-05, "epoch": 1.132992327365729, "percentage": 22.66, "elapsed_time": "2:17:19", "remaining_time": "7:48:42"} +{"current_steps": 444, "total_steps": 1955, "loss": 1.0716, "lr": 1.903499508782942e-05, "epoch": 1.1355498721227621, "percentage": 22.71, "elapsed_time": "2:17:37", "remaining_time": "7:48:22"} +{"current_steps": 445, "total_steps": 1955, "loss": 1.0921, "lr": 1.902732603462229e-05, "epoch": 1.1381074168797953, "percentage": 22.76, "elapsed_time": "2:17:56", "remaining_time": "7:48:03"} +{"current_steps": 446, "total_steps": 1955, "loss": 1.0648, "lr": 1.9019628185752384e-05, "epoch": 1.1406649616368287, "percentage": 22.81, "elapsed_time": "2:18:15", "remaining_time": "7:47:45"} +{"current_steps": 447, "total_steps": 1955, "loss": 1.0533, "lr": 1.9011901565774555e-05, "epoch": 1.143222506393862, "percentage": 22.86, "elapsed_time": "2:18:33", "remaining_time": "7:47:26"} +{"current_steps": 448, "total_steps": 1955, "loss": 1.0858, "lr": 1.900414619933543e-05, "epoch": 1.145780051150895, "percentage": 22.92, "elapsed_time": "2:18:52", "remaining_time": "7:47:08"} +{"current_steps": 449, "total_steps": 1955, "loss": 1.0772, "lr": 1.8996362111173338e-05, "epoch": 1.1483375959079285, "percentage": 22.97, "elapsed_time": "2:19:10", "remaining_time": "7:46:48"} +{"current_steps": 450, "total_steps": 1955, "loss": 1.1276, "lr": 1.898854932611821e-05, "epoch": 1.1508951406649617, "percentage": 23.02, "elapsed_time": "2:19:29", "remaining_time": "7:46:30"} +{"current_steps": 451, "total_steps": 1955, "loss": 1.1, "lr": 1.8980707869091534e-05, "epoch": 1.1534526854219949, "percentage": 23.07, "elapsed_time": "2:19:47", "remaining_time": "7:46:11"} +{"current_steps": 452, "total_steps": 1955, "loss": 1.1032, "lr": 1.8972837765106244e-05, "epoch": 1.156010230179028, "percentage": 23.12, "elapsed_time": "2:20:06", "remaining_time": "7:45:53"} +{"current_steps": 453, "total_steps": 1955, "loss": 1.1185, "lr": 1.896493903926666e-05, "epoch": 1.1585677749360614, "percentage": 23.17, "elapsed_time": "2:20:24", "remaining_time": "7:45:34"} +{"current_steps": 454, "total_steps": 1955, "loss": 1.0895, "lr": 1.8957011716768403e-05, "epoch": 1.1611253196930946, "percentage": 23.22, "elapsed_time": "2:20:43", "remaining_time": "7:45:16"} +{"current_steps": 455, "total_steps": 1955, "loss": 1.0851, "lr": 1.89490558228983e-05, "epoch": 1.1636828644501278, "percentage": 23.27, "elapsed_time": "2:21:02", "remaining_time": "7:44:58"} +{"current_steps": 456, "total_steps": 1955, "loss": 1.0773, "lr": 1.8941071383034328e-05, "epoch": 1.1662404092071612, "percentage": 23.32, "elapsed_time": "2:21:21", "remaining_time": "7:44:39"} +{"current_steps": 457, "total_steps": 1955, "loss": 1.0699, "lr": 1.8933058422645516e-05, "epoch": 1.1687979539641944, "percentage": 23.38, "elapsed_time": "2:21:39", "remaining_time": "7:44:21"} +{"current_steps": 458, "total_steps": 1955, "loss": 1.0742, "lr": 1.8925016967291873e-05, "epoch": 1.1713554987212276, "percentage": 23.43, "elapsed_time": "2:21:58", "remaining_time": "7:44:03"} +{"current_steps": 459, "total_steps": 1955, "loss": 1.0827, "lr": 1.8916947042624294e-05, "epoch": 1.1739130434782608, "percentage": 23.48, "elapsed_time": "2:22:17", "remaining_time": "7:43:44"} +{"current_steps": 460, "total_steps": 1955, "loss": 1.1101, "lr": 1.8908848674384494e-05, "epoch": 1.1764705882352942, "percentage": 23.53, "elapsed_time": "2:22:35", "remaining_time": "7:43:25"} +{"current_steps": 461, "total_steps": 1955, "loss": 1.1335, "lr": 1.8900721888404917e-05, "epoch": 1.1790281329923273, "percentage": 23.58, "elapsed_time": "2:22:54", "remaining_time": "7:43:07"} +{"current_steps": 462, "total_steps": 1955, "loss": 1.0758, "lr": 1.8892566710608648e-05, "epoch": 1.1815856777493605, "percentage": 23.63, "elapsed_time": "2:23:12", "remaining_time": "7:42:48"} +{"current_steps": 463, "total_steps": 1955, "loss": 1.0658, "lr": 1.888438316700935e-05, "epoch": 1.184143222506394, "percentage": 23.68, "elapsed_time": "2:23:31", "remaining_time": "7:42:29"} +{"current_steps": 464, "total_steps": 1955, "loss": 1.0937, "lr": 1.8876171283711153e-05, "epoch": 1.186700767263427, "percentage": 23.73, "elapsed_time": "2:23:49", "remaining_time": "7:42:09"} +{"current_steps": 465, "total_steps": 1955, "loss": 1.1014, "lr": 1.88679310869086e-05, "epoch": 1.1892583120204603, "percentage": 23.79, "elapsed_time": "2:24:07", "remaining_time": "7:41:49"} +{"current_steps": 466, "total_steps": 1955, "loss": 1.0993, "lr": 1.885966260288654e-05, "epoch": 1.1918158567774937, "percentage": 23.84, "elapsed_time": "2:24:26", "remaining_time": "7:41:30"} +{"current_steps": 467, "total_steps": 1955, "loss": 1.0761, "lr": 1.8851365858020054e-05, "epoch": 1.1943734015345269, "percentage": 23.89, "elapsed_time": "2:24:44", "remaining_time": "7:41:11"} +{"current_steps": 468, "total_steps": 1955, "loss": 1.1276, "lr": 1.884304087877438e-05, "epoch": 1.19693094629156, "percentage": 23.94, "elapsed_time": "2:25:03", "remaining_time": "7:40:52"} +{"current_steps": 469, "total_steps": 1955, "loss": 1.0915, "lr": 1.8834687691704807e-05, "epoch": 1.1994884910485935, "percentage": 23.99, "elapsed_time": "2:25:21", "remaining_time": "7:40:33"} +{"current_steps": 470, "total_steps": 1955, "loss": 1.1133, "lr": 1.8826306323456603e-05, "epoch": 1.2020460358056266, "percentage": 24.04, "elapsed_time": "2:25:39", "remaining_time": "7:40:13"} +{"current_steps": 471, "total_steps": 1955, "loss": 1.085, "lr": 1.881789680076494e-05, "epoch": 1.2046035805626598, "percentage": 24.09, "elapsed_time": "2:25:57", "remaining_time": "7:39:53"} +{"current_steps": 472, "total_steps": 1955, "loss": 1.0432, "lr": 1.8809459150454788e-05, "epoch": 1.207161125319693, "percentage": 24.14, "elapsed_time": "2:26:15", "remaining_time": "7:39:32"} +{"current_steps": 473, "total_steps": 1955, "loss": 1.0998, "lr": 1.8800993399440845e-05, "epoch": 1.2097186700767264, "percentage": 24.19, "elapsed_time": "2:26:34", "remaining_time": "7:39:14"} +{"current_steps": 474, "total_steps": 1955, "loss": 1.0828, "lr": 1.879249957472744e-05, "epoch": 1.2122762148337596, "percentage": 24.25, "elapsed_time": "2:26:53", "remaining_time": "7:38:56"} +{"current_steps": 475, "total_steps": 1955, "loss": 1.0979, "lr": 1.878397770340846e-05, "epoch": 1.2148337595907928, "percentage": 24.3, "elapsed_time": "2:27:11", "remaining_time": "7:38:37"} +{"current_steps": 476, "total_steps": 1955, "loss": 1.0667, "lr": 1.877542781266725e-05, "epoch": 1.2173913043478262, "percentage": 24.35, "elapsed_time": "2:27:29", "remaining_time": "7:38:15"} +{"current_steps": 477, "total_steps": 1955, "loss": 1.1219, "lr": 1.876684992977653e-05, "epoch": 1.2199488491048593, "percentage": 24.4, "elapsed_time": "2:27:47", "remaining_time": "7:37:56"} +{"current_steps": 478, "total_steps": 1955, "loss": 1.1044, "lr": 1.875824408209832e-05, "epoch": 1.2225063938618925, "percentage": 24.45, "elapsed_time": "2:28:06", "remaining_time": "7:37:38"} +{"current_steps": 479, "total_steps": 1955, "loss": 1.113, "lr": 1.874961029708383e-05, "epoch": 1.2250639386189257, "percentage": 24.5, "elapsed_time": "2:28:25", "remaining_time": "7:37:21"} +{"current_steps": 480, "total_steps": 1955, "loss": 1.064, "lr": 1.8740948602273402e-05, "epoch": 1.227621483375959, "percentage": 24.55, "elapsed_time": "2:28:43", "remaining_time": "7:37:01"} +{"current_steps": 481, "total_steps": 1955, "loss": 1.0728, "lr": 1.873225902529639e-05, "epoch": 1.2301790281329923, "percentage": 24.6, "elapsed_time": "2:29:02", "remaining_time": "7:36:43"} +{"current_steps": 482, "total_steps": 1955, "loss": 1.0938, "lr": 1.872354159387109e-05, "epoch": 1.2327365728900257, "percentage": 24.65, "elapsed_time": "2:29:20", "remaining_time": "7:36:24"} +{"current_steps": 483, "total_steps": 1955, "loss": 1.0849, "lr": 1.8714796335804665e-05, "epoch": 1.2352941176470589, "percentage": 24.71, "elapsed_time": "2:29:39", "remaining_time": "7:36:06"} +{"current_steps": 484, "total_steps": 1955, "loss": 1.1004, "lr": 1.8706023278993015e-05, "epoch": 1.237851662404092, "percentage": 24.76, "elapsed_time": "2:29:57", "remaining_time": "7:35:46"} +{"current_steps": 485, "total_steps": 1955, "loss": 1.1043, "lr": 1.8697222451420735e-05, "epoch": 1.2404092071611252, "percentage": 24.81, "elapsed_time": "2:30:15", "remaining_time": "7:35:26"} +{"current_steps": 486, "total_steps": 1955, "loss": 1.0987, "lr": 1.8688393881160992e-05, "epoch": 1.2429667519181586, "percentage": 24.86, "elapsed_time": "2:30:34", "remaining_time": "7:35:07"} +{"current_steps": 487, "total_steps": 1955, "loss": 1.0696, "lr": 1.867953759637545e-05, "epoch": 1.2455242966751918, "percentage": 24.91, "elapsed_time": "2:30:52", "remaining_time": "7:34:48"} +{"current_steps": 488, "total_steps": 1955, "loss": 1.0895, "lr": 1.8670653625314184e-05, "epoch": 1.248081841432225, "percentage": 24.96, "elapsed_time": "2:31:11", "remaining_time": "7:34:30"} +{"current_steps": 489, "total_steps": 1955, "loss": 1.0975, "lr": 1.8661741996315572e-05, "epoch": 1.2506393861892584, "percentage": 25.01, "elapsed_time": "2:31:29", "remaining_time": "7:34:10"} +{"current_steps": 490, "total_steps": 1955, "loss": 1.119, "lr": 1.8652802737806227e-05, "epoch": 1.2531969309462916, "percentage": 25.06, "elapsed_time": "2:31:48", "remaining_time": "7:33:52"} +{"current_steps": 491, "total_steps": 1955, "loss": 1.0881, "lr": 1.864383587830089e-05, "epoch": 1.2557544757033248, "percentage": 25.12, "elapsed_time": "2:32:06", "remaining_time": "7:33:31"} +{"current_steps": 492, "total_steps": 1955, "loss": 1.1145, "lr": 1.8634841446402344e-05, "epoch": 1.258312020460358, "percentage": 25.17, "elapsed_time": "2:32:24", "remaining_time": "7:33:11"} +{"current_steps": 493, "total_steps": 1955, "loss": 1.0836, "lr": 1.862581947080132e-05, "epoch": 1.2608695652173914, "percentage": 25.22, "elapsed_time": "2:32:43", "remaining_time": "7:32:53"} +{"current_steps": 494, "total_steps": 1955, "loss": 1.0817, "lr": 1.8616769980276427e-05, "epoch": 1.2634271099744245, "percentage": 25.27, "elapsed_time": "2:33:01", "remaining_time": "7:32:33"} +{"current_steps": 495, "total_steps": 1955, "loss": 1.1064, "lr": 1.860769300369402e-05, "epoch": 1.265984654731458, "percentage": 25.32, "elapsed_time": "2:33:19", "remaining_time": "7:32:14"} +{"current_steps": 496, "total_steps": 1955, "loss": 1.0948, "lr": 1.859858857000814e-05, "epoch": 1.2685421994884911, "percentage": 25.37, "elapsed_time": "2:33:38", "remaining_time": "7:31:55"} +{"current_steps": 497, "total_steps": 1955, "loss": 1.0867, "lr": 1.858945670826041e-05, "epoch": 1.2710997442455243, "percentage": 25.42, "elapsed_time": "2:33:56", "remaining_time": "7:31:36"} +{"current_steps": 498, "total_steps": 1955, "loss": 1.0731, "lr": 1.858029744757995e-05, "epoch": 1.2736572890025575, "percentage": 25.47, "elapsed_time": "2:34:14", "remaining_time": "7:31:16"} +{"current_steps": 499, "total_steps": 1955, "loss": 1.0876, "lr": 1.857111081718326e-05, "epoch": 1.2762148337595907, "percentage": 25.52, "elapsed_time": "2:34:33", "remaining_time": "7:30:58"} +{"current_steps": 500, "total_steps": 1955, "loss": 1.0621, "lr": 1.8561896846374167e-05, "epoch": 1.278772378516624, "percentage": 25.58, "elapsed_time": "2:34:52", "remaining_time": "7:30:39"} +{"current_steps": 501, "total_steps": 1955, "loss": 1.0828, "lr": 1.8552655564543695e-05, "epoch": 1.2813299232736572, "percentage": 25.63, "elapsed_time": "2:35:10", "remaining_time": "7:30:20"} +{"current_steps": 502, "total_steps": 1955, "loss": 1.0869, "lr": 1.8543387001169994e-05, "epoch": 1.2838874680306906, "percentage": 25.68, "elapsed_time": "2:35:29", "remaining_time": "7:30:02"} +{"current_steps": 503, "total_steps": 1955, "loss": 1.1069, "lr": 1.853409118581823e-05, "epoch": 1.2864450127877238, "percentage": 25.73, "elapsed_time": "2:35:47", "remaining_time": "7:29:44"} +{"current_steps": 504, "total_steps": 1955, "loss": 1.0826, "lr": 1.8524768148140504e-05, "epoch": 1.289002557544757, "percentage": 25.78, "elapsed_time": "2:36:06", "remaining_time": "7:29:24"} +{"current_steps": 505, "total_steps": 1955, "loss": 1.0964, "lr": 1.851541791787575e-05, "epoch": 1.2915601023017902, "percentage": 25.83, "elapsed_time": "2:36:24", "remaining_time": "7:29:05"} +{"current_steps": 506, "total_steps": 1955, "loss": 1.0721, "lr": 1.850604052484964e-05, "epoch": 1.2941176470588236, "percentage": 25.88, "elapsed_time": "2:36:42", "remaining_time": "7:28:45"} +{"current_steps": 507, "total_steps": 1955, "loss": 1.1063, "lr": 1.8496635998974487e-05, "epoch": 1.2966751918158568, "percentage": 25.93, "elapsed_time": "2:37:00", "remaining_time": "7:28:25"} +{"current_steps": 508, "total_steps": 1955, "loss": 1.0664, "lr": 1.848720437024917e-05, "epoch": 1.29923273657289, "percentage": 25.98, "elapsed_time": "2:37:19", "remaining_time": "7:28:07"} +{"current_steps": 509, "total_steps": 1955, "loss": 1.1, "lr": 1.8477745668758998e-05, "epoch": 1.3017902813299234, "percentage": 26.04, "elapsed_time": "2:37:38", "remaining_time": "7:27:49"} +{"current_steps": 510, "total_steps": 1955, "loss": 1.068, "lr": 1.8468259924675655e-05, "epoch": 1.3043478260869565, "percentage": 26.09, "elapsed_time": "2:37:56", "remaining_time": "7:27:30"} +{"current_steps": 511, "total_steps": 1955, "loss": 1.0768, "lr": 1.8458747168257086e-05, "epoch": 1.3069053708439897, "percentage": 26.14, "elapsed_time": "2:38:15", "remaining_time": "7:27:12"} +{"current_steps": 512, "total_steps": 1955, "loss": 1.1078, "lr": 1.8449207429847383e-05, "epoch": 1.309462915601023, "percentage": 26.19, "elapsed_time": "2:38:34", "remaining_time": "7:26:54"} +{"current_steps": 513, "total_steps": 1955, "loss": 1.0634, "lr": 1.8439640739876732e-05, "epoch": 1.3120204603580563, "percentage": 26.24, "elapsed_time": "2:38:52", "remaining_time": "7:26:34"} +{"current_steps": 514, "total_steps": 1955, "loss": 1.0846, "lr": 1.8430047128861267e-05, "epoch": 1.3145780051150895, "percentage": 26.29, "elapsed_time": "2:39:10", "remaining_time": "7:26:15"} +{"current_steps": 515, "total_steps": 1955, "loss": 1.076, "lr": 1.8420426627403008e-05, "epoch": 1.317135549872123, "percentage": 26.34, "elapsed_time": "2:39:29", "remaining_time": "7:25:57"} +{"current_steps": 516, "total_steps": 1955, "loss": 1.0715, "lr": 1.8410779266189752e-05, "epoch": 1.319693094629156, "percentage": 26.39, "elapsed_time": "2:39:47", "remaining_time": "7:25:36"} +{"current_steps": 517, "total_steps": 1955, "loss": 1.0676, "lr": 1.840110507599497e-05, "epoch": 1.3222506393861893, "percentage": 26.45, "elapsed_time": "2:40:06", "remaining_time": "7:25:18"} +{"current_steps": 518, "total_steps": 1955, "loss": 1.0847, "lr": 1.8391404087677704e-05, "epoch": 1.3248081841432224, "percentage": 26.5, "elapsed_time": "2:40:24", "remaining_time": "7:24:59"} +{"current_steps": 519, "total_steps": 1955, "loss": 1.0749, "lr": 1.8381676332182498e-05, "epoch": 1.3273657289002558, "percentage": 26.55, "elapsed_time": "2:40:43", "remaining_time": "7:24:41"} +{"current_steps": 520, "total_steps": 1955, "loss": 1.1205, "lr": 1.8371921840539264e-05, "epoch": 1.329923273657289, "percentage": 26.6, "elapsed_time": "2:41:01", "remaining_time": "7:24:21"} +{"current_steps": 521, "total_steps": 1955, "loss": 1.1202, "lr": 1.83621406438632e-05, "epoch": 1.3324808184143222, "percentage": 26.65, "elapsed_time": "2:41:18", "remaining_time": "7:24:00"} +{"current_steps": 522, "total_steps": 1955, "loss": 1.0973, "lr": 1.8352332773354696e-05, "epoch": 1.3350383631713556, "percentage": 26.7, "elapsed_time": "2:41:36", "remaining_time": "7:23:40"} +{"current_steps": 523, "total_steps": 1955, "loss": 1.0864, "lr": 1.8342498260299212e-05, "epoch": 1.3375959079283888, "percentage": 26.75, "elapsed_time": "2:41:55", "remaining_time": "7:23:21"} +{"current_steps": 524, "total_steps": 1955, "loss": 1.1128, "lr": 1.833263713606721e-05, "epoch": 1.340153452685422, "percentage": 26.8, "elapsed_time": "2:42:14", "remaining_time": "7:23:02"} +{"current_steps": 525, "total_steps": 1955, "loss": 1.083, "lr": 1.8322749432114028e-05, "epoch": 1.3427109974424551, "percentage": 26.85, "elapsed_time": "2:42:32", "remaining_time": "7:22:44"} +{"current_steps": 526, "total_steps": 1955, "loss": 1.1089, "lr": 1.8312835179979788e-05, "epoch": 1.3452685421994885, "percentage": 26.91, "elapsed_time": "2:42:51", "remaining_time": "7:22:27"} +{"current_steps": 527, "total_steps": 1955, "loss": 1.0576, "lr": 1.8302894411289303e-05, "epoch": 1.3478260869565217, "percentage": 26.96, "elapsed_time": "2:43:10", "remaining_time": "7:22:08"} +{"current_steps": 528, "total_steps": 1955, "loss": 1.1227, "lr": 1.8292927157751964e-05, "epoch": 1.350383631713555, "percentage": 27.01, "elapsed_time": "2:43:29", "remaining_time": "7:21:50"} +{"current_steps": 529, "total_steps": 1955, "loss": 1.0839, "lr": 1.8282933451161645e-05, "epoch": 1.3529411764705883, "percentage": 27.06, "elapsed_time": "2:43:47", "remaining_time": "7:21:32"} +{"current_steps": 530, "total_steps": 1955, "loss": 1.0556, "lr": 1.8272913323396598e-05, "epoch": 1.3554987212276215, "percentage": 27.11, "elapsed_time": "2:44:05", "remaining_time": "7:21:12"} +{"current_steps": 531, "total_steps": 1955, "loss": 1.0941, "lr": 1.8262866806419365e-05, "epoch": 1.3580562659846547, "percentage": 27.16, "elapsed_time": "2:44:24", "remaining_time": "7:20:53"} +{"current_steps": 532, "total_steps": 1955, "loss": 1.0782, "lr": 1.8252793932276652e-05, "epoch": 1.3606138107416879, "percentage": 27.21, "elapsed_time": "2:44:43", "remaining_time": "7:20:35"} +{"current_steps": 533, "total_steps": 1955, "loss": 1.1175, "lr": 1.8242694733099248e-05, "epoch": 1.3631713554987213, "percentage": 27.26, "elapsed_time": "2:45:01", "remaining_time": "7:20:17"} +{"current_steps": 534, "total_steps": 1955, "loss": 1.0834, "lr": 1.823256924110191e-05, "epoch": 1.3657289002557544, "percentage": 27.31, "elapsed_time": "2:45:20", "remaining_time": "7:19:58"} +{"current_steps": 535, "total_steps": 1955, "loss": 1.0943, "lr": 1.822241748858327e-05, "epoch": 1.3682864450127878, "percentage": 27.37, "elapsed_time": "2:45:38", "remaining_time": "7:19:39"} +{"current_steps": 536, "total_steps": 1955, "loss": 1.1287, "lr": 1.821223950792572e-05, "epoch": 1.370843989769821, "percentage": 27.42, "elapsed_time": "2:45:57", "remaining_time": "7:19:21"} +{"current_steps": 537, "total_steps": 1955, "loss": 1.0849, "lr": 1.8202035331595322e-05, "epoch": 1.3734015345268542, "percentage": 27.47, "elapsed_time": "2:46:16", "remaining_time": "7:19:03"} +{"current_steps": 538, "total_steps": 1955, "loss": 1.0976, "lr": 1.8191804992141694e-05, "epoch": 1.3759590792838874, "percentage": 27.52, "elapsed_time": "2:46:34", "remaining_time": "7:18:44"} +{"current_steps": 539, "total_steps": 1955, "loss": 1.082, "lr": 1.818154852219791e-05, "epoch": 1.3785166240409208, "percentage": 27.57, "elapsed_time": "2:46:53", "remaining_time": "7:18:26"} +{"current_steps": 540, "total_steps": 1955, "loss": 1.0908, "lr": 1.8171265954480396e-05, "epoch": 1.381074168797954, "percentage": 27.62, "elapsed_time": "2:47:11", "remaining_time": "7:18:06"} +{"current_steps": 541, "total_steps": 1955, "loss": 1.1171, "lr": 1.8160957321788826e-05, "epoch": 1.3836317135549872, "percentage": 27.67, "elapsed_time": "2:47:30", "remaining_time": "7:17:48"} +{"current_steps": 542, "total_steps": 1955, "loss": 1.1102, "lr": 1.8150622657006017e-05, "epoch": 1.3861892583120206, "percentage": 27.72, "elapsed_time": "2:47:49", "remaining_time": "7:17:30"} +{"current_steps": 543, "total_steps": 1955, "loss": 1.0649, "lr": 1.814026199309783e-05, "epoch": 1.3887468030690537, "percentage": 27.77, "elapsed_time": "2:48:07", "remaining_time": "7:17:11"} +{"current_steps": 544, "total_steps": 1955, "loss": 1.0855, "lr": 1.8129875363113045e-05, "epoch": 1.391304347826087, "percentage": 27.83, "elapsed_time": "2:48:26", "remaining_time": "7:16:52"} +{"current_steps": 545, "total_steps": 1955, "loss": 1.074, "lr": 1.811946280018328e-05, "epoch": 1.39386189258312, "percentage": 27.88, "elapsed_time": "2:48:44", "remaining_time": "7:16:33"} +{"current_steps": 546, "total_steps": 1955, "loss": 1.053, "lr": 1.8109024337522875e-05, "epoch": 1.3964194373401535, "percentage": 27.93, "elapsed_time": "2:49:02", "remaining_time": "7:16:13"} +{"current_steps": 547, "total_steps": 1955, "loss": 1.0931, "lr": 1.8098560008428778e-05, "epoch": 1.3989769820971867, "percentage": 27.98, "elapsed_time": "2:49:21", "remaining_time": "7:15:55"} +{"current_steps": 548, "total_steps": 1955, "loss": 1.0836, "lr": 1.8088069846280456e-05, "epoch": 1.40153452685422, "percentage": 28.03, "elapsed_time": "2:49:39", "remaining_time": "7:15:36"} +{"current_steps": 549, "total_steps": 1955, "loss": 1.1068, "lr": 1.8077553884539772e-05, "epoch": 1.4040920716112533, "percentage": 28.08, "elapsed_time": "2:49:58", "remaining_time": "7:15:18"} +{"current_steps": 550, "total_steps": 1955, "loss": 1.0469, "lr": 1.8067012156750892e-05, "epoch": 1.4066496163682864, "percentage": 28.13, "elapsed_time": "2:50:16", "remaining_time": "7:14:58"} +{"current_steps": 551, "total_steps": 1955, "loss": 1.0579, "lr": 1.8056444696540162e-05, "epoch": 1.4092071611253196, "percentage": 28.18, "elapsed_time": "2:50:35", "remaining_time": "7:14:40"} +{"current_steps": 552, "total_steps": 1955, "loss": 1.1032, "lr": 1.804585153761602e-05, "epoch": 1.4117647058823528, "percentage": 28.24, "elapsed_time": "2:50:53", "remaining_time": "7:14:21"} +{"current_steps": 553, "total_steps": 1955, "loss": 1.087, "lr": 1.803523271376887e-05, "epoch": 1.4143222506393862, "percentage": 28.29, "elapsed_time": "2:51:12", "remaining_time": "7:14:03"} +{"current_steps": 554, "total_steps": 1955, "loss": 1.0976, "lr": 1.802458825887099e-05, "epoch": 1.4168797953964194, "percentage": 28.34, "elapsed_time": "2:51:31", "remaining_time": "7:13:45"} +{"current_steps": 555, "total_steps": 1955, "loss": 1.1109, "lr": 1.8013918206876416e-05, "epoch": 1.4194373401534528, "percentage": 28.39, "elapsed_time": "2:51:49", "remaining_time": "7:13:26"} +{"current_steps": 556, "total_steps": 1955, "loss": 1.0955, "lr": 1.8003222591820827e-05, "epoch": 1.421994884910486, "percentage": 28.44, "elapsed_time": "2:52:08", "remaining_time": "7:13:08"} +{"current_steps": 557, "total_steps": 1955, "loss": 1.0815, "lr": 1.7992501447821454e-05, "epoch": 1.4245524296675192, "percentage": 28.49, "elapsed_time": "2:52:26", "remaining_time": "7:12:48"} +{"current_steps": 558, "total_steps": 1955, "loss": 1.0743, "lr": 1.7981754809076952e-05, "epoch": 1.4271099744245523, "percentage": 28.54, "elapsed_time": "2:52:44", "remaining_time": "7:12:28"} +{"current_steps": 559, "total_steps": 1955, "loss": 1.0665, "lr": 1.7970982709867307e-05, "epoch": 1.4296675191815857, "percentage": 28.59, "elapsed_time": "2:53:03", "remaining_time": "7:12:09"} +{"current_steps": 560, "total_steps": 1955, "loss": 1.0576, "lr": 1.7960185184553717e-05, "epoch": 1.432225063938619, "percentage": 28.64, "elapsed_time": "2:53:21", "remaining_time": "7:11:50"} +{"current_steps": 561, "total_steps": 1955, "loss": 1.07, "lr": 1.7949362267578485e-05, "epoch": 1.434782608695652, "percentage": 28.7, "elapsed_time": "2:53:39", "remaining_time": "7:11:32"} +{"current_steps": 562, "total_steps": 1955, "loss": 1.0837, "lr": 1.793851399346491e-05, "epoch": 1.4373401534526855, "percentage": 28.75, "elapsed_time": "2:53:58", "remaining_time": "7:11:13"} +{"current_steps": 563, "total_steps": 1955, "loss": 1.1108, "lr": 1.7927640396817172e-05, "epoch": 1.4398976982097187, "percentage": 28.8, "elapsed_time": "2:54:17", "remaining_time": "7:10:55"} +{"current_steps": 564, "total_steps": 1955, "loss": 1.109, "lr": 1.7916741512320227e-05, "epoch": 1.4424552429667519, "percentage": 28.85, "elapsed_time": "2:54:35", "remaining_time": "7:10:35"} +{"current_steps": 565, "total_steps": 1955, "loss": 1.1167, "lr": 1.7905817374739706e-05, "epoch": 1.445012787723785, "percentage": 28.9, "elapsed_time": "2:54:53", "remaining_time": "7:10:16"} +{"current_steps": 566, "total_steps": 1955, "loss": 1.0784, "lr": 1.789486801892177e-05, "epoch": 1.4475703324808185, "percentage": 28.95, "elapsed_time": "2:55:11", "remaining_time": "7:09:56"} +{"current_steps": 567, "total_steps": 1955, "loss": 1.0596, "lr": 1.7883893479793052e-05, "epoch": 1.4501278772378516, "percentage": 29.0, "elapsed_time": "2:55:30", "remaining_time": "7:09:39"} +{"current_steps": 568, "total_steps": 1955, "loss": 1.0907, "lr": 1.7872893792360488e-05, "epoch": 1.452685421994885, "percentage": 29.05, "elapsed_time": "2:55:49", "remaining_time": "7:09:20"} +{"current_steps": 569, "total_steps": 1955, "loss": 1.0833, "lr": 1.7861868991711245e-05, "epoch": 1.4552429667519182, "percentage": 29.1, "elapsed_time": "2:56:08", "remaining_time": "7:09:02"} +{"current_steps": 570, "total_steps": 1955, "loss": 1.0702, "lr": 1.78508191130126e-05, "epoch": 1.4578005115089514, "percentage": 29.16, "elapsed_time": "2:56:26", "remaining_time": "7:08:43"} +{"current_steps": 571, "total_steps": 1955, "loss": 1.092, "lr": 1.783974419151182e-05, "epoch": 1.4603580562659846, "percentage": 29.21, "elapsed_time": "2:56:45", "remaining_time": "7:08:25"} +{"current_steps": 572, "total_steps": 1955, "loss": 1.0969, "lr": 1.782864426253606e-05, "epoch": 1.4629156010230178, "percentage": 29.26, "elapsed_time": "2:57:04", "remaining_time": "7:08:08"} +{"current_steps": 573, "total_steps": 1955, "loss": 1.0826, "lr": 1.7817519361492227e-05, "epoch": 1.4654731457800512, "percentage": 29.31, "elapsed_time": "2:57:23", "remaining_time": "7:07:49"} +{"current_steps": 574, "total_steps": 1955, "loss": 1.0674, "lr": 1.7806369523866914e-05, "epoch": 1.4680306905370843, "percentage": 29.36, "elapsed_time": "2:57:41", "remaining_time": "7:07:31"} +{"current_steps": 575, "total_steps": 1955, "loss": 1.1002, "lr": 1.779519478522623e-05, "epoch": 1.4705882352941178, "percentage": 29.41, "elapsed_time": "2:58:00", "remaining_time": "7:07:12"} +{"current_steps": 576, "total_steps": 1955, "loss": 1.1042, "lr": 1.7783995181215727e-05, "epoch": 1.473145780051151, "percentage": 29.46, "elapsed_time": "2:58:19", "remaining_time": "7:06:54"} +{"current_steps": 577, "total_steps": 1955, "loss": 1.0958, "lr": 1.7772770747560276e-05, "epoch": 1.4757033248081841, "percentage": 29.51, "elapsed_time": "2:58:37", "remaining_time": "7:06:36"} +{"current_steps": 578, "total_steps": 1955, "loss": 1.0857, "lr": 1.7761521520063945e-05, "epoch": 1.4782608695652173, "percentage": 29.57, "elapsed_time": "2:58:56", "remaining_time": "7:06:18"} +{"current_steps": 579, "total_steps": 1955, "loss": 1.0638, "lr": 1.775024753460989e-05, "epoch": 1.4808184143222507, "percentage": 29.62, "elapsed_time": "2:59:15", "remaining_time": "7:05:59"} +{"current_steps": 580, "total_steps": 1955, "loss": 1.081, "lr": 1.773894882716024e-05, "epoch": 1.4833759590792839, "percentage": 29.67, "elapsed_time": "2:59:33", "remaining_time": "7:05:41"} +{"current_steps": 581, "total_steps": 1955, "loss": 1.0979, "lr": 1.7727625433755992e-05, "epoch": 1.485933503836317, "percentage": 29.72, "elapsed_time": "2:59:52", "remaining_time": "7:05:22"} +{"current_steps": 582, "total_steps": 1955, "loss": 1.0872, "lr": 1.7716277390516875e-05, "epoch": 1.4884910485933505, "percentage": 29.77, "elapsed_time": "3:00:11", "remaining_time": "7:05:04"} +{"current_steps": 583, "total_steps": 1955, "loss": 1.1129, "lr": 1.7704904733641255e-05, "epoch": 1.4910485933503836, "percentage": 29.82, "elapsed_time": "3:00:29", "remaining_time": "7:04:45"} +{"current_steps": 584, "total_steps": 1955, "loss": 1.1183, "lr": 1.7693507499406003e-05, "epoch": 1.4936061381074168, "percentage": 29.87, "elapsed_time": "3:00:48", "remaining_time": "7:04:27"} +{"current_steps": 585, "total_steps": 1955, "loss": 1.0564, "lr": 1.76820857241664e-05, "epoch": 1.49616368286445, "percentage": 29.92, "elapsed_time": "3:01:07", "remaining_time": "7:04:09"} +{"current_steps": 586, "total_steps": 1955, "loss": 1.0723, "lr": 1.7670639444355998e-05, "epoch": 1.4987212276214834, "percentage": 29.97, "elapsed_time": "3:01:25", "remaining_time": "7:03:50"} +{"current_steps": 587, "total_steps": 1955, "loss": 1.0807, "lr": 1.7659168696486523e-05, "epoch": 1.5012787723785166, "percentage": 30.03, "elapsed_time": "3:01:43", "remaining_time": "7:03:31"} +{"current_steps": 588, "total_steps": 1955, "loss": 1.1072, "lr": 1.764767351714774e-05, "epoch": 1.50383631713555, "percentage": 30.08, "elapsed_time": "3:02:02", "remaining_time": "7:03:12"} +{"current_steps": 589, "total_steps": 1955, "loss": 1.048, "lr": 1.763615394300735e-05, "epoch": 1.5063938618925832, "percentage": 30.13, "elapsed_time": "3:02:20", "remaining_time": "7:02:52"} +{"current_steps": 590, "total_steps": 1955, "loss": 1.0773, "lr": 1.762461001081088e-05, "epoch": 1.5089514066496164, "percentage": 30.18, "elapsed_time": "3:02:39", "remaining_time": "7:02:34"} +{"current_steps": 591, "total_steps": 1955, "loss": 1.0509, "lr": 1.7613041757381538e-05, "epoch": 1.5115089514066495, "percentage": 30.23, "elapsed_time": "3:02:57", "remaining_time": "7:02:15"} +{"current_steps": 592, "total_steps": 1955, "loss": 1.1046, "lr": 1.7601449219620127e-05, "epoch": 1.5140664961636827, "percentage": 30.28, "elapsed_time": "3:03:16", "remaining_time": "7:01:57"} +{"current_steps": 593, "total_steps": 1955, "loss": 1.0329, "lr": 1.7589832434504903e-05, "epoch": 1.5166240409207161, "percentage": 30.33, "elapsed_time": "3:03:34", "remaining_time": "7:01:37"} +{"current_steps": 594, "total_steps": 1955, "loss": 1.1, "lr": 1.757819143909147e-05, "epoch": 1.5191815856777495, "percentage": 30.38, "elapsed_time": "3:03:53", "remaining_time": "7:01:19"} +{"current_steps": 595, "total_steps": 1955, "loss": 1.0706, "lr": 1.7566526270512663e-05, "epoch": 1.5217391304347827, "percentage": 30.43, "elapsed_time": "3:04:11", "remaining_time": "7:01:01"} +{"current_steps": 596, "total_steps": 1955, "loss": 1.0811, "lr": 1.755483696597842e-05, "epoch": 1.5242966751918159, "percentage": 30.49, "elapsed_time": "3:04:30", "remaining_time": "7:00:42"} +{"current_steps": 597, "total_steps": 1955, "loss": 1.0892, "lr": 1.754312356277567e-05, "epoch": 1.526854219948849, "percentage": 30.54, "elapsed_time": "3:04:48", "remaining_time": "7:00:23"} +{"current_steps": 598, "total_steps": 1955, "loss": 1.0919, "lr": 1.753138609826822e-05, "epoch": 1.5294117647058822, "percentage": 30.59, "elapsed_time": "3:05:07", "remaining_time": "7:00:04"} +{"current_steps": 599, "total_steps": 1955, "loss": 1.0679, "lr": 1.7519624609896614e-05, "epoch": 1.5319693094629157, "percentage": 30.64, "elapsed_time": "3:05:25", "remaining_time": "6:59:46"} +{"current_steps": 600, "total_steps": 1955, "loss": 1.1173, "lr": 1.750783913517804e-05, "epoch": 1.5345268542199488, "percentage": 30.69, "elapsed_time": "3:05:43", "remaining_time": "6:59:26"} +{"current_steps": 601, "total_steps": 1955, "loss": 1.065, "lr": 1.749602971170619e-05, "epoch": 1.5370843989769822, "percentage": 30.74, "elapsed_time": "3:06:02", "remaining_time": "6:59:07"} +{"current_steps": 602, "total_steps": 1955, "loss": 1.0628, "lr": 1.748419637715116e-05, "epoch": 1.5396419437340154, "percentage": 30.79, "elapsed_time": "3:06:20", "remaining_time": "6:58:47"} +{"current_steps": 603, "total_steps": 1955, "loss": 1.0757, "lr": 1.7472339169259306e-05, "epoch": 1.5421994884910486, "percentage": 30.84, "elapsed_time": "3:06:39", "remaining_time": "6:58:29"} +{"current_steps": 604, "total_steps": 1955, "loss": 1.0865, "lr": 1.7460458125853143e-05, "epoch": 1.5447570332480818, "percentage": 30.9, "elapsed_time": "3:06:57", "remaining_time": "6:58:10"} +{"current_steps": 605, "total_steps": 1955, "loss": 1.0731, "lr": 1.744855328483122e-05, "epoch": 1.547314578005115, "percentage": 30.95, "elapsed_time": "3:07:15", "remaining_time": "6:57:51"} +{"current_steps": 606, "total_steps": 1955, "loss": 1.0495, "lr": 1.7436624684167984e-05, "epoch": 1.5498721227621484, "percentage": 31.0, "elapsed_time": "3:07:34", "remaining_time": "6:57:33"} +{"current_steps": 607, "total_steps": 1955, "loss": 1.0736, "lr": 1.7424672361913685e-05, "epoch": 1.5524296675191815, "percentage": 31.05, "elapsed_time": "3:07:52", "remaining_time": "6:57:13"} +{"current_steps": 608, "total_steps": 1955, "loss": 1.0983, "lr": 1.7412696356194233e-05, "epoch": 1.554987212276215, "percentage": 31.1, "elapsed_time": "3:08:10", "remaining_time": "6:56:54"} +{"current_steps": 609, "total_steps": 1955, "loss": 1.0942, "lr": 1.740069670521109e-05, "epoch": 1.5575447570332481, "percentage": 31.15, "elapsed_time": "3:08:29", "remaining_time": "6:56:35"} +{"current_steps": 610, "total_steps": 1955, "loss": 1.0643, "lr": 1.738867344724114e-05, "epoch": 1.5601023017902813, "percentage": 31.2, "elapsed_time": "3:08:48", "remaining_time": "6:56:17"} +{"current_steps": 611, "total_steps": 1955, "loss": 1.0981, "lr": 1.7376626620636558e-05, "epoch": 1.5626598465473145, "percentage": 31.25, "elapsed_time": "3:09:06", "remaining_time": "6:55:58"} +{"current_steps": 612, "total_steps": 1955, "loss": 1.0884, "lr": 1.7364556263824718e-05, "epoch": 1.5652173913043477, "percentage": 31.3, "elapsed_time": "3:09:24", "remaining_time": "6:55:39"} +{"current_steps": 613, "total_steps": 1955, "loss": 1.0784, "lr": 1.7352462415308043e-05, "epoch": 1.567774936061381, "percentage": 31.36, "elapsed_time": "3:09:43", "remaining_time": "6:55:20"} +{"current_steps": 614, "total_steps": 1955, "loss": 1.0824, "lr": 1.734034511366389e-05, "epoch": 1.5703324808184145, "percentage": 31.41, "elapsed_time": "3:10:01", "remaining_time": "6:55:02"} +{"current_steps": 615, "total_steps": 1955, "loss": 1.1013, "lr": 1.7328204397544426e-05, "epoch": 1.5728900255754477, "percentage": 31.46, "elapsed_time": "3:10:20", "remaining_time": "6:54:43"} +{"current_steps": 616, "total_steps": 1955, "loss": 1.0553, "lr": 1.731604030567651e-05, "epoch": 1.5754475703324808, "percentage": 31.51, "elapsed_time": "3:10:39", "remaining_time": "6:54:25"} +{"current_steps": 617, "total_steps": 1955, "loss": 1.0892, "lr": 1.730385287686156e-05, "epoch": 1.578005115089514, "percentage": 31.56, "elapsed_time": "3:10:57", "remaining_time": "6:54:06"} +{"current_steps": 618, "total_steps": 1955, "loss": 1.0825, "lr": 1.7291642149975447e-05, "epoch": 1.5805626598465472, "percentage": 31.61, "elapsed_time": "3:11:16", "remaining_time": "6:53:48"} +{"current_steps": 619, "total_steps": 1955, "loss": 1.0682, "lr": 1.7279408163968344e-05, "epoch": 1.5831202046035806, "percentage": 31.66, "elapsed_time": "3:11:35", "remaining_time": "6:53:29"} +{"current_steps": 620, "total_steps": 1955, "loss": 1.0562, "lr": 1.7267150957864626e-05, "epoch": 1.5856777493606138, "percentage": 31.71, "elapsed_time": "3:11:53", "remaining_time": "6:53:11"} +{"current_steps": 621, "total_steps": 1955, "loss": 1.0859, "lr": 1.7254870570762734e-05, "epoch": 1.5882352941176472, "percentage": 31.76, "elapsed_time": "3:12:12", "remaining_time": "6:52:53"} +{"current_steps": 622, "total_steps": 1955, "loss": 1.0983, "lr": 1.724256704183505e-05, "epoch": 1.5907928388746804, "percentage": 31.82, "elapsed_time": "3:12:30", "remaining_time": "6:52:34"} +{"current_steps": 623, "total_steps": 1955, "loss": 1.0722, "lr": 1.7230240410327784e-05, "epoch": 1.5933503836317136, "percentage": 31.87, "elapsed_time": "3:12:49", "remaining_time": "6:52:16"} +{"current_steps": 624, "total_steps": 1955, "loss": 1.1117, "lr": 1.7217890715560822e-05, "epoch": 1.5959079283887467, "percentage": 31.92, "elapsed_time": "3:13:08", "remaining_time": "6:51:57"} +{"current_steps": 625, "total_steps": 1955, "loss": 1.1324, "lr": 1.720551799692764e-05, "epoch": 1.59846547314578, "percentage": 31.97, "elapsed_time": "3:13:26", "remaining_time": "6:51:38"} +{"current_steps": 626, "total_steps": 1955, "loss": 1.0592, "lr": 1.7193122293895137e-05, "epoch": 1.6010230179028133, "percentage": 32.02, "elapsed_time": "3:13:45", "remaining_time": "6:51:21"} +{"current_steps": 627, "total_steps": 1955, "loss": 1.0697, "lr": 1.7180703646003536e-05, "epoch": 1.6035805626598465, "percentage": 32.07, "elapsed_time": "3:14:04", "remaining_time": "6:51:02"} +{"current_steps": 628, "total_steps": 1955, "loss": 1.1086, "lr": 1.7168262092866252e-05, "epoch": 1.60613810741688, "percentage": 32.12, "elapsed_time": "3:14:22", "remaining_time": "6:50:44"} +{"current_steps": 629, "total_steps": 1955, "loss": 1.0619, "lr": 1.715579767416976e-05, "epoch": 1.608695652173913, "percentage": 32.17, "elapsed_time": "3:14:41", "remaining_time": "6:50:26"} +{"current_steps": 630, "total_steps": 1955, "loss": 1.0676, "lr": 1.714331042967348e-05, "epoch": 1.6112531969309463, "percentage": 32.23, "elapsed_time": "3:15:00", "remaining_time": "6:50:07"} +{"current_steps": 631, "total_steps": 1955, "loss": 1.0759, "lr": 1.7130800399209632e-05, "epoch": 1.6138107416879794, "percentage": 32.28, "elapsed_time": "3:15:18", "remaining_time": "6:49:49"} +{"current_steps": 632, "total_steps": 1955, "loss": 1.0696, "lr": 1.7118267622683123e-05, "epoch": 1.6163682864450126, "percentage": 32.33, "elapsed_time": "3:15:37", "remaining_time": "6:49:30"} +{"current_steps": 633, "total_steps": 1955, "loss": 1.052, "lr": 1.7105712140071427e-05, "epoch": 1.618925831202046, "percentage": 32.38, "elapsed_time": "3:15:55", "remaining_time": "6:49:11"} +{"current_steps": 634, "total_steps": 1955, "loss": 1.1003, "lr": 1.7093133991424425e-05, "epoch": 1.6214833759590794, "percentage": 32.43, "elapsed_time": "3:16:14", "remaining_time": "6:48:52"} +{"current_steps": 635, "total_steps": 1955, "loss": 1.053, "lr": 1.7080533216864318e-05, "epoch": 1.6240409207161126, "percentage": 32.48, "elapsed_time": "3:16:32", "remaining_time": "6:48:32"} +{"current_steps": 636, "total_steps": 1955, "loss": 1.0994, "lr": 1.7067909856585474e-05, "epoch": 1.6265984654731458, "percentage": 32.53, "elapsed_time": "3:16:51", "remaining_time": "6:48:15"} +{"current_steps": 637, "total_steps": 1955, "loss": 1.0932, "lr": 1.70552639508543e-05, "epoch": 1.629156010230179, "percentage": 32.58, "elapsed_time": "3:17:09", "remaining_time": "6:47:55"} +{"current_steps": 638, "total_steps": 1955, "loss": 1.0777, "lr": 1.7042595540009123e-05, "epoch": 1.6317135549872122, "percentage": 32.63, "elapsed_time": "3:17:27", "remaining_time": "6:47:36"} +{"current_steps": 639, "total_steps": 1955, "loss": 1.1112, "lr": 1.7029904664460066e-05, "epoch": 1.6342710997442456, "percentage": 32.69, "elapsed_time": "3:17:45", "remaining_time": "6:47:17"} +{"current_steps": 640, "total_steps": 1955, "loss": 1.0452, "lr": 1.7017191364688896e-05, "epoch": 1.6368286445012787, "percentage": 32.74, "elapsed_time": "3:18:04", "remaining_time": "6:46:58"} +{"current_steps": 641, "total_steps": 1955, "loss": 1.0658, "lr": 1.7004455681248918e-05, "epoch": 1.6393861892583121, "percentage": 32.79, "elapsed_time": "3:18:23", "remaining_time": "6:46:40"} +{"current_steps": 642, "total_steps": 1955, "loss": 1.0679, "lr": 1.699169765476484e-05, "epoch": 1.6419437340153453, "percentage": 32.84, "elapsed_time": "3:18:41", "remaining_time": "6:46:21"} +{"current_steps": 643, "total_steps": 1955, "loss": 1.0833, "lr": 1.6978917325932628e-05, "epoch": 1.6445012787723785, "percentage": 32.89, "elapsed_time": "3:19:00", "remaining_time": "6:46:03"} +{"current_steps": 644, "total_steps": 1955, "loss": 1.1294, "lr": 1.6966114735519408e-05, "epoch": 1.6470588235294117, "percentage": 32.94, "elapsed_time": "3:19:18", "remaining_time": "6:45:43"} +{"current_steps": 645, "total_steps": 1955, "loss": 1.0381, "lr": 1.6953289924363297e-05, "epoch": 1.6496163682864449, "percentage": 32.99, "elapsed_time": "3:19:36", "remaining_time": "6:45:25"} +{"current_steps": 646, "total_steps": 1955, "loss": 1.0953, "lr": 1.6940442933373305e-05, "epoch": 1.6521739130434783, "percentage": 33.04, "elapsed_time": "3:19:55", "remaining_time": "6:45:06"} +{"current_steps": 647, "total_steps": 1955, "loss": 1.0531, "lr": 1.6927573803529186e-05, "epoch": 1.6547314578005117, "percentage": 33.09, "elapsed_time": "3:20:14", "remaining_time": "6:44:48"} +{"current_steps": 648, "total_steps": 1955, "loss": 1.0712, "lr": 1.6914682575881312e-05, "epoch": 1.6572890025575449, "percentage": 33.15, "elapsed_time": "3:20:33", "remaining_time": "6:44:30"} +{"current_steps": 649, "total_steps": 1955, "loss": 1.0762, "lr": 1.690176929155056e-05, "epoch": 1.659846547314578, "percentage": 33.2, "elapsed_time": "3:20:51", "remaining_time": "6:44:12"} +{"current_steps": 650, "total_steps": 1955, "loss": 1.0598, "lr": 1.6888833991728137e-05, "epoch": 1.6624040920716112, "percentage": 33.25, "elapsed_time": "3:21:10", "remaining_time": "6:43:53"} +{"current_steps": 651, "total_steps": 1955, "loss": 1.0865, "lr": 1.68758767176755e-05, "epoch": 1.6649616368286444, "percentage": 33.3, "elapsed_time": "3:21:29", "remaining_time": "6:43:35"} +{"current_steps": 652, "total_steps": 1955, "loss": 1.1109, "lr": 1.6862897510724176e-05, "epoch": 1.6675191815856778, "percentage": 33.35, "elapsed_time": "3:21:47", "remaining_time": "6:43:16"} +{"current_steps": 653, "total_steps": 1955, "loss": 1.086, "lr": 1.6849896412275684e-05, "epoch": 1.670076726342711, "percentage": 33.4, "elapsed_time": "3:22:05", "remaining_time": "6:42:56"} +{"current_steps": 654, "total_steps": 1955, "loss": 1.0907, "lr": 1.6836873463801348e-05, "epoch": 1.6726342710997444, "percentage": 33.45, "elapsed_time": "3:22:23", "remaining_time": "6:42:37"} +{"current_steps": 655, "total_steps": 1955, "loss": 1.0798, "lr": 1.68238287068422e-05, "epoch": 1.6751918158567776, "percentage": 33.5, "elapsed_time": "3:22:41", "remaining_time": "6:42:17"} +{"current_steps": 656, "total_steps": 1955, "loss": 1.0662, "lr": 1.6810762183008845e-05, "epoch": 1.6777493606138107, "percentage": 33.55, "elapsed_time": "3:22:59", "remaining_time": "6:41:58"} +{"current_steps": 657, "total_steps": 1955, "loss": 1.0685, "lr": 1.67976739339813e-05, "epoch": 1.680306905370844, "percentage": 33.61, "elapsed_time": "3:23:18", "remaining_time": "6:41:39"} +{"current_steps": 658, "total_steps": 1955, "loss": 1.0904, "lr": 1.67845640015089e-05, "epoch": 1.682864450127877, "percentage": 33.66, "elapsed_time": "3:23:37", "remaining_time": "6:41:21"} +{"current_steps": 659, "total_steps": 1955, "loss": 1.09, "lr": 1.6771432427410136e-05, "epoch": 1.6854219948849105, "percentage": 33.71, "elapsed_time": "3:23:55", "remaining_time": "6:41:02"} +{"current_steps": 660, "total_steps": 1955, "loss": 1.107, "lr": 1.6758279253572548e-05, "epoch": 1.6879795396419437, "percentage": 33.76, "elapsed_time": "3:24:14", "remaining_time": "6:40:44"} +{"current_steps": 661, "total_steps": 1955, "loss": 1.0859, "lr": 1.6745104521952552e-05, "epoch": 1.690537084398977, "percentage": 33.81, "elapsed_time": "3:24:32", "remaining_time": "6:40:26"} +{"current_steps": 662, "total_steps": 1955, "loss": 1.078, "lr": 1.673190827457535e-05, "epoch": 1.6930946291560103, "percentage": 33.86, "elapsed_time": "3:24:51", "remaining_time": "6:40:07"} +{"current_steps": 663, "total_steps": 1955, "loss": 1.0998, "lr": 1.6718690553534766e-05, "epoch": 1.6956521739130435, "percentage": 33.91, "elapsed_time": "3:25:10", "remaining_time": "6:39:49"} +{"current_steps": 664, "total_steps": 1955, "loss": 1.0856, "lr": 1.670545140099312e-05, "epoch": 1.6982097186700766, "percentage": 33.96, "elapsed_time": "3:25:28", "remaining_time": "6:39:30"} +{"current_steps": 665, "total_steps": 1955, "loss": 1.0898, "lr": 1.6692190859181102e-05, "epoch": 1.7007672634271098, "percentage": 34.02, "elapsed_time": "3:25:47", "remaining_time": "6:39:11"} +{"current_steps": 666, "total_steps": 1955, "loss": 1.0609, "lr": 1.6678908970397623e-05, "epoch": 1.7033248081841432, "percentage": 34.07, "elapsed_time": "3:26:04", "remaining_time": "6:38:51"} +{"current_steps": 667, "total_steps": 1955, "loss": 1.0816, "lr": 1.6665605777009698e-05, "epoch": 1.7058823529411766, "percentage": 34.12, "elapsed_time": "3:26:23", "remaining_time": "6:38:32"} +{"current_steps": 668, "total_steps": 1955, "loss": 1.0705, "lr": 1.665228132145228e-05, "epoch": 1.7084398976982098, "percentage": 34.17, "elapsed_time": "3:26:42", "remaining_time": "6:38:14"} +{"current_steps": 669, "total_steps": 1955, "loss": 1.0969, "lr": 1.6638935646228168e-05, "epoch": 1.710997442455243, "percentage": 34.22, "elapsed_time": "3:27:00", "remaining_time": "6:37:56"} +{"current_steps": 670, "total_steps": 1955, "loss": 1.0758, "lr": 1.6625568793907835e-05, "epoch": 1.7135549872122762, "percentage": 34.27, "elapsed_time": "3:27:18", "remaining_time": "6:37:36"} +{"current_steps": 671, "total_steps": 1955, "loss": 1.0928, "lr": 1.6612180807129305e-05, "epoch": 1.7161125319693094, "percentage": 34.32, "elapsed_time": "3:27:37", "remaining_time": "6:37:18"} +{"current_steps": 672, "total_steps": 1955, "loss": 1.0943, "lr": 1.6598771728598024e-05, "epoch": 1.7186700767263428, "percentage": 34.37, "elapsed_time": "3:27:56", "remaining_time": "6:37:00"} +{"current_steps": 673, "total_steps": 1955, "loss": 1.0714, "lr": 1.6585341601086713e-05, "epoch": 1.721227621483376, "percentage": 34.42, "elapsed_time": "3:28:15", "remaining_time": "6:36:42"} +{"current_steps": 674, "total_steps": 1955, "loss": 1.1156, "lr": 1.6571890467435233e-05, "epoch": 1.7237851662404093, "percentage": 34.48, "elapsed_time": "3:28:33", "remaining_time": "6:36:23"} +{"current_steps": 675, "total_steps": 1955, "loss": 1.0854, "lr": 1.655841837055046e-05, "epoch": 1.7263427109974425, "percentage": 34.53, "elapsed_time": "3:28:51", "remaining_time": "6:36:04"} +{"current_steps": 676, "total_steps": 1955, "loss": 1.0516, "lr": 1.6544925353406127e-05, "epoch": 1.7289002557544757, "percentage": 34.58, "elapsed_time": "3:29:09", "remaining_time": "6:35:44"} +{"current_steps": 677, "total_steps": 1955, "loss": 1.0684, "lr": 1.653141145904271e-05, "epoch": 1.7314578005115089, "percentage": 34.63, "elapsed_time": "3:29:28", "remaining_time": "6:35:26"} +{"current_steps": 678, "total_steps": 1955, "loss": 1.07, "lr": 1.651787673056728e-05, "epoch": 1.734015345268542, "percentage": 34.68, "elapsed_time": "3:29:47", "remaining_time": "6:35:07"} +{"current_steps": 679, "total_steps": 1955, "loss": 1.0711, "lr": 1.6504321211153355e-05, "epoch": 1.7365728900255755, "percentage": 34.73, "elapsed_time": "3:30:05", "remaining_time": "6:34:49"} +{"current_steps": 680, "total_steps": 1955, "loss": 1.0726, "lr": 1.6490744944040776e-05, "epoch": 1.7391304347826086, "percentage": 34.78, "elapsed_time": "3:30:24", "remaining_time": "6:34:31"} +{"current_steps": 681, "total_steps": 1955, "loss": 1.061, "lr": 1.6477147972535577e-05, "epoch": 1.741687979539642, "percentage": 34.83, "elapsed_time": "3:30:43", "remaining_time": "6:34:13"} +{"current_steps": 682, "total_steps": 1955, "loss": 1.1065, "lr": 1.6463530340009817e-05, "epoch": 1.7442455242966752, "percentage": 34.88, "elapsed_time": "3:31:01", "remaining_time": "6:33:53"} +{"current_steps": 683, "total_steps": 1955, "loss": 1.0993, "lr": 1.644989208990148e-05, "epoch": 1.7468030690537084, "percentage": 34.94, "elapsed_time": "3:31:20", "remaining_time": "6:33:35"} +{"current_steps": 684, "total_steps": 1955, "loss": 1.084, "lr": 1.6436233265714297e-05, "epoch": 1.7493606138107416, "percentage": 34.99, "elapsed_time": "3:31:38", "remaining_time": "6:33:16"} +{"current_steps": 685, "total_steps": 1955, "loss": 1.0761, "lr": 1.6422553911017642e-05, "epoch": 1.7519181585677748, "percentage": 35.04, "elapsed_time": "3:31:56", "remaining_time": "6:32:57"} +{"current_steps": 686, "total_steps": 1955, "loss": 1.1069, "lr": 1.6408854069446374e-05, "epoch": 1.7544757033248082, "percentage": 35.09, "elapsed_time": "3:32:15", "remaining_time": "6:32:38"} +{"current_steps": 687, "total_steps": 1955, "loss": 1.0813, "lr": 1.6395133784700696e-05, "epoch": 1.7570332480818416, "percentage": 35.14, "elapsed_time": "3:32:34", "remaining_time": "6:32:20"} +{"current_steps": 688, "total_steps": 1955, "loss": 1.0692, "lr": 1.638139310054603e-05, "epoch": 1.7595907928388748, "percentage": 35.19, "elapsed_time": "3:32:52", "remaining_time": "6:32:01"} +{"current_steps": 689, "total_steps": 1955, "loss": 1.0506, "lr": 1.6367632060812856e-05, "epoch": 1.762148337595908, "percentage": 35.24, "elapsed_time": "3:33:10", "remaining_time": "6:31:42"} +{"current_steps": 690, "total_steps": 1955, "loss": 1.0764, "lr": 1.6353850709396605e-05, "epoch": 1.7647058823529411, "percentage": 35.29, "elapsed_time": "3:33:29", "remaining_time": "6:31:24"} +{"current_steps": 691, "total_steps": 1955, "loss": 1.0887, "lr": 1.6340049090257476e-05, "epoch": 1.7672634271099743, "percentage": 35.35, "elapsed_time": "3:33:48", "remaining_time": "6:31:05"} +{"current_steps": 692, "total_steps": 1955, "loss": 1.0812, "lr": 1.6326227247420337e-05, "epoch": 1.7698209718670077, "percentage": 35.4, "elapsed_time": "3:34:06", "remaining_time": "6:30:47"} +{"current_steps": 693, "total_steps": 1955, "loss": 1.0971, "lr": 1.6312385224974555e-05, "epoch": 1.772378516624041, "percentage": 35.45, "elapsed_time": "3:34:25", "remaining_time": "6:30:29"} +{"current_steps": 694, "total_steps": 1955, "loss": 1.0485, "lr": 1.6298523067073872e-05, "epoch": 1.7749360613810743, "percentage": 35.5, "elapsed_time": "3:34:44", "remaining_time": "6:30:10"} +{"current_steps": 695, "total_steps": 1955, "loss": 1.0629, "lr": 1.6284640817936254e-05, "epoch": 1.7774936061381075, "percentage": 35.55, "elapsed_time": "3:35:02", "remaining_time": "6:29:52"} +{"current_steps": 696, "total_steps": 1955, "loss": 1.097, "lr": 1.6270738521843762e-05, "epoch": 1.7800511508951407, "percentage": 35.6, "elapsed_time": "3:35:21", "remaining_time": "6:29:34"} +{"current_steps": 697, "total_steps": 1955, "loss": 1.074, "lr": 1.62568162231424e-05, "epoch": 1.7826086956521738, "percentage": 35.65, "elapsed_time": "3:35:40", "remaining_time": "6:29:15"} +{"current_steps": 698, "total_steps": 1955, "loss": 1.09, "lr": 1.6242873966241975e-05, "epoch": 1.785166240409207, "percentage": 35.7, "elapsed_time": "3:35:58", "remaining_time": "6:28:57"} +{"current_steps": 699, "total_steps": 1955, "loss": 1.0786, "lr": 1.6228911795615953e-05, "epoch": 1.7877237851662404, "percentage": 35.75, "elapsed_time": "3:36:16", "remaining_time": "6:28:37"} +{"current_steps": 700, "total_steps": 1955, "loss": 1.0959, "lr": 1.6214929755801335e-05, "epoch": 1.7902813299232738, "percentage": 35.81, "elapsed_time": "3:36:35", "remaining_time": "6:28:19"} +{"current_steps": 701, "total_steps": 1955, "loss": 1.0617, "lr": 1.6200927891398486e-05, "epoch": 1.792838874680307, "percentage": 35.86, "elapsed_time": "3:36:54", "remaining_time": "6:28:00"} +{"current_steps": 702, "total_steps": 1955, "loss": 1.082, "lr": 1.6186906247071026e-05, "epoch": 1.7953964194373402, "percentage": 35.91, "elapsed_time": "3:37:12", "remaining_time": "6:27:42"} +{"current_steps": 703, "total_steps": 1955, "loss": 1.0842, "lr": 1.6172864867545653e-05, "epoch": 1.7979539641943734, "percentage": 35.96, "elapsed_time": "3:37:30", "remaining_time": "6:27:22"} +{"current_steps": 704, "total_steps": 1955, "loss": 1.0643, "lr": 1.615880379761202e-05, "epoch": 1.8005115089514065, "percentage": 36.01, "elapsed_time": "3:37:49", "remaining_time": "6:27:03"} +{"current_steps": 705, "total_steps": 1955, "loss": 1.0953, "lr": 1.6144723082122596e-05, "epoch": 1.80306905370844, "percentage": 36.06, "elapsed_time": "3:38:08", "remaining_time": "6:26:45"} +{"current_steps": 706, "total_steps": 1955, "loss": 1.0457, "lr": 1.613062276599251e-05, "epoch": 1.8056265984654731, "percentage": 36.11, "elapsed_time": "3:38:26", "remaining_time": "6:26:27"} +{"current_steps": 707, "total_steps": 1955, "loss": 1.0503, "lr": 1.611650289419942e-05, "epoch": 1.8081841432225065, "percentage": 36.16, "elapsed_time": "3:38:45", "remaining_time": "6:26:08"} +{"current_steps": 708, "total_steps": 1955, "loss": 1.0763, "lr": 1.610236351178336e-05, "epoch": 1.8107416879795397, "percentage": 36.21, "elapsed_time": "3:39:03", "remaining_time": "6:25:49"} +{"current_steps": 709, "total_steps": 1955, "loss": 1.0253, "lr": 1.6088204663846596e-05, "epoch": 1.813299232736573, "percentage": 36.27, "elapsed_time": "3:39:22", "remaining_time": "6:25:31"} +{"current_steps": 710, "total_steps": 1955, "loss": 1.0709, "lr": 1.6074026395553487e-05, "epoch": 1.815856777493606, "percentage": 36.32, "elapsed_time": "3:39:40", "remaining_time": "6:25:12"} +{"current_steps": 711, "total_steps": 1955, "loss": 1.1258, "lr": 1.6059828752130347e-05, "epoch": 1.8184143222506393, "percentage": 36.37, "elapsed_time": "3:39:58", "remaining_time": "6:24:53"} +{"current_steps": 712, "total_steps": 1955, "loss": 1.0728, "lr": 1.6045611778865282e-05, "epoch": 1.8209718670076727, "percentage": 36.42, "elapsed_time": "3:40:17", "remaining_time": "6:24:34"} +{"current_steps": 713, "total_steps": 1955, "loss": 1.0817, "lr": 1.6031375521108068e-05, "epoch": 1.8235294117647058, "percentage": 36.47, "elapsed_time": "3:40:36", "remaining_time": "6:24:16"} +{"current_steps": 714, "total_steps": 1955, "loss": 1.0707, "lr": 1.601712002426999e-05, "epoch": 1.8260869565217392, "percentage": 36.52, "elapsed_time": "3:40:54", "remaining_time": "6:23:58"} +{"current_steps": 715, "total_steps": 1955, "loss": 1.0537, "lr": 1.6002845333823696e-05, "epoch": 1.8286445012787724, "percentage": 36.57, "elapsed_time": "3:41:13", "remaining_time": "6:23:39"} +{"current_steps": 716, "total_steps": 1955, "loss": 1.0707, "lr": 1.5988551495303072e-05, "epoch": 1.8312020460358056, "percentage": 36.62, "elapsed_time": "3:41:32", "remaining_time": "6:23:21"} +{"current_steps": 717, "total_steps": 1955, "loss": 1.041, "lr": 1.5974238554303078e-05, "epoch": 1.8337595907928388, "percentage": 36.68, "elapsed_time": "3:41:50", "remaining_time": "6:23:02"} +{"current_steps": 718, "total_steps": 1955, "loss": 1.0912, "lr": 1.5959906556479595e-05, "epoch": 1.836317135549872, "percentage": 36.73, "elapsed_time": "3:42:08", "remaining_time": "6:22:43"} +{"current_steps": 719, "total_steps": 1955, "loss": 1.0905, "lr": 1.5945555547549316e-05, "epoch": 1.8388746803069054, "percentage": 36.78, "elapsed_time": "3:42:27", "remaining_time": "6:22:25"} +{"current_steps": 720, "total_steps": 1955, "loss": 1.0977, "lr": 1.5931185573289557e-05, "epoch": 1.8414322250639388, "percentage": 36.83, "elapsed_time": "3:42:46", "remaining_time": "6:22:06"} +{"current_steps": 721, "total_steps": 1955, "loss": 1.0983, "lr": 1.5916796679538137e-05, "epoch": 1.843989769820972, "percentage": 36.88, "elapsed_time": "3:43:04", "remaining_time": "6:21:48"} +{"current_steps": 722, "total_steps": 1955, "loss": 1.0538, "lr": 1.5902388912193223e-05, "epoch": 1.8465473145780051, "percentage": 36.93, "elapsed_time": "3:43:23", "remaining_time": "6:21:29"} +{"current_steps": 723, "total_steps": 1955, "loss": 1.0815, "lr": 1.588796231721319e-05, "epoch": 1.8491048593350383, "percentage": 36.98, "elapsed_time": "3:43:42", "remaining_time": "6:21:11"} +{"current_steps": 724, "total_steps": 1955, "loss": 1.0713, "lr": 1.5873516940616468e-05, "epoch": 1.8516624040920715, "percentage": 37.03, "elapsed_time": "3:44:00", "remaining_time": "6:20:52"} +{"current_steps": 725, "total_steps": 1955, "loss": 1.1034, "lr": 1.5859052828481395e-05, "epoch": 1.854219948849105, "percentage": 37.08, "elapsed_time": "3:44:18", "remaining_time": "6:20:33"} +{"current_steps": 726, "total_steps": 1955, "loss": 1.0254, "lr": 1.5844570026946068e-05, "epoch": 1.856777493606138, "percentage": 37.14, "elapsed_time": "3:44:37", "remaining_time": "6:20:15"} +{"current_steps": 727, "total_steps": 1955, "loss": 1.0513, "lr": 1.5830068582208217e-05, "epoch": 1.8593350383631715, "percentage": 37.19, "elapsed_time": "3:44:56", "remaining_time": "6:19:56"} +{"current_steps": 728, "total_steps": 1955, "loss": 1.0559, "lr": 1.581554854052502e-05, "epoch": 1.8618925831202047, "percentage": 37.24, "elapsed_time": "3:45:14", "remaining_time": "6:19:38"} +{"current_steps": 729, "total_steps": 1955, "loss": 1.1154, "lr": 1.5801009948212992e-05, "epoch": 1.8644501278772379, "percentage": 37.29, "elapsed_time": "3:45:33", "remaining_time": "6:19:20"} +{"current_steps": 730, "total_steps": 1955, "loss": 1.0687, "lr": 1.578645285164781e-05, "epoch": 1.867007672634271, "percentage": 37.34, "elapsed_time": "3:45:51", "remaining_time": "6:19:00"} +{"current_steps": 731, "total_steps": 1955, "loss": 1.0398, "lr": 1.5771877297264184e-05, "epoch": 1.8695652173913042, "percentage": 37.39, "elapsed_time": "3:46:10", "remaining_time": "6:18:42"} +{"current_steps": 732, "total_steps": 1955, "loss": 1.0769, "lr": 1.5757283331555698e-05, "epoch": 1.8721227621483376, "percentage": 37.44, "elapsed_time": "3:46:28", "remaining_time": "6:18:23"} +{"current_steps": 733, "total_steps": 1955, "loss": 1.0889, "lr": 1.5742671001074666e-05, "epoch": 1.8746803069053708, "percentage": 37.49, "elapsed_time": "3:46:46", "remaining_time": "6:18:04"} +{"current_steps": 734, "total_steps": 1955, "loss": 1.0472, "lr": 1.5728040352431983e-05, "epoch": 1.8772378516624042, "percentage": 37.54, "elapsed_time": "3:47:05", "remaining_time": "6:17:46"} +{"current_steps": 735, "total_steps": 1955, "loss": 1.0766, "lr": 1.5713391432296976e-05, "epoch": 1.8797953964194374, "percentage": 37.6, "elapsed_time": "3:47:24", "remaining_time": "6:17:27"} +{"current_steps": 736, "total_steps": 1955, "loss": 1.1086, "lr": 1.5698724287397254e-05, "epoch": 1.8823529411764706, "percentage": 37.65, "elapsed_time": "3:47:43", "remaining_time": "6:17:09"} +{"current_steps": 737, "total_steps": 1955, "loss": 1.0814, "lr": 1.5684038964518557e-05, "epoch": 1.8849104859335037, "percentage": 37.7, "elapsed_time": "3:48:01", "remaining_time": "6:16:51"} +{"current_steps": 738, "total_steps": 1955, "loss": 1.0886, "lr": 1.566933551050462e-05, "epoch": 1.887468030690537, "percentage": 37.75, "elapsed_time": "3:48:20", "remaining_time": "6:16:32"} +{"current_steps": 739, "total_steps": 1955, "loss": 1.0723, "lr": 1.5654613972256997e-05, "epoch": 1.8900255754475703, "percentage": 37.8, "elapsed_time": "3:48:39", "remaining_time": "6:16:14"} +{"current_steps": 740, "total_steps": 1955, "loss": 1.0432, "lr": 1.5639874396734943e-05, "epoch": 1.8925831202046037, "percentage": 37.85, "elapsed_time": "3:48:57", "remaining_time": "6:15:55"} +{"current_steps": 741, "total_steps": 1955, "loss": 1.0531, "lr": 1.5625116830955246e-05, "epoch": 1.895140664961637, "percentage": 37.9, "elapsed_time": "3:49:16", "remaining_time": "6:15:37"} +{"current_steps": 742, "total_steps": 1955, "loss": 1.0597, "lr": 1.5610341321992068e-05, "epoch": 1.89769820971867, "percentage": 37.95, "elapsed_time": "3:49:34", "remaining_time": "6:15:18"} +{"current_steps": 743, "total_steps": 1955, "loss": 1.0988, "lr": 1.559554791697682e-05, "epoch": 1.9002557544757033, "percentage": 38.01, "elapsed_time": "3:49:53", "remaining_time": "6:15:00"} +{"current_steps": 744, "total_steps": 1955, "loss": 1.0533, "lr": 1.5580736663097998e-05, "epoch": 1.9028132992327365, "percentage": 38.06, "elapsed_time": "3:50:12", "remaining_time": "6:14:42"} +{"current_steps": 745, "total_steps": 1955, "loss": 1.0502, "lr": 1.5565907607601025e-05, "epoch": 1.9053708439897699, "percentage": 38.11, "elapsed_time": "3:50:30", "remaining_time": "6:14:22"} +{"current_steps": 746, "total_steps": 1955, "loss": 1.0742, "lr": 1.5551060797788107e-05, "epoch": 1.907928388746803, "percentage": 38.16, "elapsed_time": "3:50:48", "remaining_time": "6:14:03"} +{"current_steps": 747, "total_steps": 1955, "loss": 1.0656, "lr": 1.5536196281018096e-05, "epoch": 1.9104859335038364, "percentage": 38.21, "elapsed_time": "3:51:06", "remaining_time": "6:13:43"} +{"current_steps": 748, "total_steps": 1955, "loss": 1.0813, "lr": 1.5521314104706318e-05, "epoch": 1.9130434782608696, "percentage": 38.26, "elapsed_time": "3:51:24", "remaining_time": "6:13:24"} +{"current_steps": 749, "total_steps": 1955, "loss": 1.1034, "lr": 1.5506414316324426e-05, "epoch": 1.9156010230179028, "percentage": 38.31, "elapsed_time": "3:51:43", "remaining_time": "6:13:06"} +{"current_steps": 750, "total_steps": 1955, "loss": 1.0563, "lr": 1.549149696340026e-05, "epoch": 1.918158567774936, "percentage": 38.36, "elapsed_time": "3:52:01", "remaining_time": "6:12:47"} +{"current_steps": 751, "total_steps": 1955, "loss": 1.1094, "lr": 1.547656209351769e-05, "epoch": 1.9207161125319692, "percentage": 38.41, "elapsed_time": "3:52:19", "remaining_time": "6:12:28"} +{"current_steps": 752, "total_steps": 1955, "loss": 1.0598, "lr": 1.5461609754316446e-05, "epoch": 1.9232736572890026, "percentage": 38.47, "elapsed_time": "3:52:38", "remaining_time": "6:12:09"} +{"current_steps": 753, "total_steps": 1955, "loss": 1.0752, "lr": 1.5446639993492003e-05, "epoch": 1.9258312020460358, "percentage": 38.52, "elapsed_time": "3:52:56", "remaining_time": "6:11:51"} +{"current_steps": 754, "total_steps": 1955, "loss": 1.0701, "lr": 1.5431652858795394e-05, "epoch": 1.9283887468030692, "percentage": 38.57, "elapsed_time": "3:53:15", "remaining_time": "6:11:32"} +{"current_steps": 755, "total_steps": 1955, "loss": 1.0393, "lr": 1.5416648398033076e-05, "epoch": 1.9309462915601023, "percentage": 38.62, "elapsed_time": "3:53:34", "remaining_time": "6:11:14"} +{"current_steps": 756, "total_steps": 1955, "loss": 1.0779, "lr": 1.5401626659066774e-05, "epoch": 1.9335038363171355, "percentage": 38.67, "elapsed_time": "3:53:52", "remaining_time": "6:10:55"} +{"current_steps": 757, "total_steps": 1955, "loss": 1.0741, "lr": 1.5386587689813333e-05, "epoch": 1.9360613810741687, "percentage": 38.72, "elapsed_time": "3:54:11", "remaining_time": "6:10:36"} +{"current_steps": 758, "total_steps": 1955, "loss": 1.0554, "lr": 1.5371531538244547e-05, "epoch": 1.938618925831202, "percentage": 38.77, "elapsed_time": "3:54:29", "remaining_time": "6:10:17"} +{"current_steps": 759, "total_steps": 1955, "loss": 1.102, "lr": 1.5356458252387025e-05, "epoch": 1.9411764705882353, "percentage": 38.82, "elapsed_time": "3:54:47", "remaining_time": "6:09:58"} +{"current_steps": 760, "total_steps": 1955, "loss": 1.1062, "lr": 1.534136788032204e-05, "epoch": 1.9437340153452687, "percentage": 38.87, "elapsed_time": "3:55:06", "remaining_time": "6:09:40"} +{"current_steps": 761, "total_steps": 1955, "loss": 1.1062, "lr": 1.5326260470185353e-05, "epoch": 1.9462915601023019, "percentage": 38.93, "elapsed_time": "3:55:25", "remaining_time": "6:09:22"} +{"current_steps": 762, "total_steps": 1955, "loss": 1.0834, "lr": 1.5311136070167076e-05, "epoch": 1.948849104859335, "percentage": 38.98, "elapsed_time": "3:55:43", "remaining_time": "6:09:03"} +{"current_steps": 763, "total_steps": 1955, "loss": 1.0716, "lr": 1.529599472851153e-05, "epoch": 1.9514066496163682, "percentage": 39.03, "elapsed_time": "3:56:02", "remaining_time": "6:08:45"} +{"current_steps": 764, "total_steps": 1955, "loss": 1.0872, "lr": 1.5280836493517058e-05, "epoch": 1.9539641943734014, "percentage": 39.08, "elapsed_time": "3:56:21", "remaining_time": "6:08:27"} +{"current_steps": 765, "total_steps": 1955, "loss": 1.0184, "lr": 1.5265661413535907e-05, "epoch": 1.9565217391304348, "percentage": 39.13, "elapsed_time": "3:56:39", "remaining_time": "6:08:08"} +{"current_steps": 766, "total_steps": 1955, "loss": 1.0848, "lr": 1.5250469536974043e-05, "epoch": 1.959079283887468, "percentage": 39.18, "elapsed_time": "3:56:58", "remaining_time": "6:07:50"} +{"current_steps": 767, "total_steps": 1955, "loss": 1.0531, "lr": 1.5235260912291012e-05, "epoch": 1.9616368286445014, "percentage": 39.23, "elapsed_time": "3:57:17", "remaining_time": "6:07:31"} +{"current_steps": 768, "total_steps": 1955, "loss": 1.0798, "lr": 1.5220035587999795e-05, "epoch": 1.9641943734015346, "percentage": 39.28, "elapsed_time": "3:57:35", "remaining_time": "6:07:12"} +{"current_steps": 769, "total_steps": 1955, "loss": 1.0884, "lr": 1.5204793612666628e-05, "epoch": 1.9667519181585678, "percentage": 39.34, "elapsed_time": "3:57:53", "remaining_time": "6:06:53"} +{"current_steps": 770, "total_steps": 1955, "loss": 1.07, "lr": 1.5189535034910874e-05, "epoch": 1.969309462915601, "percentage": 39.39, "elapsed_time": "3:58:12", "remaining_time": "6:06:35"} +{"current_steps": 771, "total_steps": 1955, "loss": 1.0618, "lr": 1.5174259903404844e-05, "epoch": 1.9718670076726341, "percentage": 39.44, "elapsed_time": "3:58:30", "remaining_time": "6:06:16"} +{"current_steps": 772, "total_steps": 1955, "loss": 1.0774, "lr": 1.5158968266873657e-05, "epoch": 1.9744245524296675, "percentage": 39.49, "elapsed_time": "3:58:49", "remaining_time": "6:05:58"} +{"current_steps": 773, "total_steps": 1955, "loss": 1.0865, "lr": 1.514366017409508e-05, "epoch": 1.976982097186701, "percentage": 39.54, "elapsed_time": "3:59:08", "remaining_time": "6:05:40"} +{"current_steps": 774, "total_steps": 1955, "loss": 1.0534, "lr": 1.5128335673899377e-05, "epoch": 1.979539641943734, "percentage": 39.59, "elapsed_time": "3:59:26", "remaining_time": "6:05:21"} +{"current_steps": 775, "total_steps": 1955, "loss": 1.0643, "lr": 1.5112994815169143e-05, "epoch": 1.9820971867007673, "percentage": 39.64, "elapsed_time": "3:59:45", "remaining_time": "6:05:02"} +{"current_steps": 776, "total_steps": 1955, "loss": 1.0827, "lr": 1.509763764683915e-05, "epoch": 1.9846547314578005, "percentage": 39.69, "elapsed_time": "4:00:03", "remaining_time": "6:04:43"} +{"current_steps": 777, "total_steps": 1955, "loss": 1.0622, "lr": 1.5082264217896208e-05, "epoch": 1.9872122762148337, "percentage": 39.74, "elapsed_time": "4:00:22", "remaining_time": "6:04:25"} +{"current_steps": 778, "total_steps": 1955, "loss": 1.1021, "lr": 1.5066874577378987e-05, "epoch": 1.989769820971867, "percentage": 39.8, "elapsed_time": "4:00:40", "remaining_time": "6:04:05"} +{"current_steps": 779, "total_steps": 1955, "loss": 1.0833, "lr": 1.5051468774377868e-05, "epoch": 1.9923273657289002, "percentage": 39.85, "elapsed_time": "4:00:58", "remaining_time": "6:03:47"} +{"current_steps": 780, "total_steps": 1955, "loss": 1.0656, "lr": 1.5036046858034796e-05, "epoch": 1.9948849104859336, "percentage": 39.9, "elapsed_time": "4:01:16", "remaining_time": "6:03:28"} +{"current_steps": 781, "total_steps": 1955, "loss": 1.0718, "lr": 1.5020608877543103e-05, "epoch": 1.9974424552429668, "percentage": 39.95, "elapsed_time": "4:01:35", "remaining_time": "6:03:10"} +{"current_steps": 782, "total_steps": 1955, "loss": 1.0498, "lr": 1.5005154882147373e-05, "epoch": 2.0, "percentage": 40.0, "elapsed_time": "4:01:54", "remaining_time": "6:02:51"} +{"current_steps": 783, "total_steps": 1955, "loss": 1.0568, "lr": 1.4989684921143268e-05, "epoch": 2.002557544757033, "percentage": 40.05, "elapsed_time": "4:02:23", "remaining_time": "6:02:48"} +{"current_steps": 784, "total_steps": 1955, "loss": 1.044, "lr": 1.4974199043877383e-05, "epoch": 2.0051150895140664, "percentage": 40.1, "elapsed_time": "4:02:41", "remaining_time": "6:02:30"} +{"current_steps": 785, "total_steps": 1955, "loss": 1.0467, "lr": 1.495869729974708e-05, "epoch": 2.0076726342710995, "percentage": 40.15, "elapsed_time": "4:03:00", "remaining_time": "6:02:11"} +{"current_steps": 786, "total_steps": 1955, "loss": 1.1337, "lr": 1.4943179738200334e-05, "epoch": 2.010230179028133, "percentage": 40.2, "elapsed_time": "4:03:18", "remaining_time": "6:01:52"} +{"current_steps": 787, "total_steps": 1955, "loss": 1.0676, "lr": 1.4927646408735576e-05, "epoch": 2.0127877237851663, "percentage": 40.26, "elapsed_time": "4:03:37", "remaining_time": "6:01:34"} +{"current_steps": 788, "total_steps": 1955, "loss": 1.0775, "lr": 1.4912097360901534e-05, "epoch": 2.0153452685421995, "percentage": 40.31, "elapsed_time": "4:03:56", "remaining_time": "6:01:15"} +{"current_steps": 789, "total_steps": 1955, "loss": 1.0604, "lr": 1.489653264429707e-05, "epoch": 2.0179028132992327, "percentage": 40.36, "elapsed_time": "4:04:14", "remaining_time": "6:00:57"} +{"current_steps": 790, "total_steps": 1955, "loss": 1.0497, "lr": 1.4880952308571039e-05, "epoch": 2.020460358056266, "percentage": 40.41, "elapsed_time": "4:04:33", "remaining_time": "6:00:39"} +{"current_steps": 791, "total_steps": 1955, "loss": 1.0481, "lr": 1.4865356403422104e-05, "epoch": 2.023017902813299, "percentage": 40.46, "elapsed_time": "4:04:51", "remaining_time": "6:00:19"} +{"current_steps": 792, "total_steps": 1955, "loss": 1.0529, "lr": 1.4849744978598604e-05, "epoch": 2.0255754475703327, "percentage": 40.51, "elapsed_time": "4:05:10", "remaining_time": "6:00:01"} +{"current_steps": 793, "total_steps": 1955, "loss": 1.0449, "lr": 1.4834118083898373e-05, "epoch": 2.028132992327366, "percentage": 40.56, "elapsed_time": "4:05:29", "remaining_time": "5:59:43"} +{"current_steps": 794, "total_steps": 1955, "loss": 1.0448, "lr": 1.4818475769168594e-05, "epoch": 2.030690537084399, "percentage": 40.61, "elapsed_time": "4:05:47", "remaining_time": "5:59:24"} +{"current_steps": 795, "total_steps": 1955, "loss": 1.0491, "lr": 1.4802818084305646e-05, "epoch": 2.0332480818414322, "percentage": 40.66, "elapsed_time": "4:06:06", "remaining_time": "5:59:06"} +{"current_steps": 796, "total_steps": 1955, "loss": 1.0631, "lr": 1.4787145079254926e-05, "epoch": 2.0358056265984654, "percentage": 40.72, "elapsed_time": "4:06:24", "remaining_time": "5:58:46"} +{"current_steps": 797, "total_steps": 1955, "loss": 1.0844, "lr": 1.4771456804010703e-05, "epoch": 2.0383631713554986, "percentage": 40.77, "elapsed_time": "4:06:43", "remaining_time": "5:58:28"} +{"current_steps": 798, "total_steps": 1955, "loss": 1.0561, "lr": 1.4755753308615953e-05, "epoch": 2.040920716112532, "percentage": 40.82, "elapsed_time": "4:07:02", "remaining_time": "5:58:10"} +{"current_steps": 799, "total_steps": 1955, "loss": 1.0572, "lr": 1.474003464316221e-05, "epoch": 2.0434782608695654, "percentage": 40.87, "elapsed_time": "4:07:20", "remaining_time": "5:57:51"} +{"current_steps": 800, "total_steps": 1955, "loss": 1.1033, "lr": 1.4724300857789387e-05, "epoch": 2.0460358056265986, "percentage": 40.92, "elapsed_time": "4:07:39", "remaining_time": "5:57:33"} +{"current_steps": 801, "total_steps": 1955, "loss": 1.0545, "lr": 1.4708552002685634e-05, "epoch": 2.0485933503836318, "percentage": 40.97, "elapsed_time": "4:07:58", "remaining_time": "5:57:14"} +{"current_steps": 802, "total_steps": 1955, "loss": 1.06, "lr": 1.4692788128087174e-05, "epoch": 2.051150895140665, "percentage": 41.02, "elapsed_time": "4:08:16", "remaining_time": "5:56:56"} +{"current_steps": 803, "total_steps": 1955, "loss": 1.0685, "lr": 1.4677009284278128e-05, "epoch": 2.053708439897698, "percentage": 41.07, "elapsed_time": "4:08:34", "remaining_time": "5:56:36"} +{"current_steps": 804, "total_steps": 1955, "loss": 1.0969, "lr": 1.4661215521590375e-05, "epoch": 2.0562659846547313, "percentage": 41.13, "elapsed_time": "4:08:53", "remaining_time": "5:56:18"} +{"current_steps": 805, "total_steps": 1955, "loss": 1.0437, "lr": 1.4645406890403384e-05, "epoch": 2.0588235294117645, "percentage": 41.18, "elapsed_time": "4:09:11", "remaining_time": "5:55:59"} +{"current_steps": 806, "total_steps": 1955, "loss": 1.0571, "lr": 1.4629583441144044e-05, "epoch": 2.061381074168798, "percentage": 41.23, "elapsed_time": "4:09:30", "remaining_time": "5:55:41"} +{"current_steps": 807, "total_steps": 1955, "loss": 1.0992, "lr": 1.4613745224286523e-05, "epoch": 2.0639386189258313, "percentage": 41.28, "elapsed_time": "4:09:48", "remaining_time": "5:55:22"} +{"current_steps": 808, "total_steps": 1955, "loss": 1.071, "lr": 1.459789229035208e-05, "epoch": 2.0664961636828645, "percentage": 41.33, "elapsed_time": "4:10:07", "remaining_time": "5:55:03"} +{"current_steps": 809, "total_steps": 1955, "loss": 1.0472, "lr": 1.4582024689908933e-05, "epoch": 2.0690537084398977, "percentage": 41.38, "elapsed_time": "4:10:25", "remaining_time": "5:54:44"} +{"current_steps": 810, "total_steps": 1955, "loss": 1.0509, "lr": 1.456614247357208e-05, "epoch": 2.071611253196931, "percentage": 41.43, "elapsed_time": "4:10:43", "remaining_time": "5:54:25"} +{"current_steps": 811, "total_steps": 1955, "loss": 1.0754, "lr": 1.455024569200313e-05, "epoch": 2.074168797953964, "percentage": 41.48, "elapsed_time": "4:11:02", "remaining_time": "5:54:07"} +{"current_steps": 812, "total_steps": 1955, "loss": 1.0487, "lr": 1.4534334395910173e-05, "epoch": 2.0767263427109977, "percentage": 41.53, "elapsed_time": "4:11:21", "remaining_time": "5:53:48"} +{"current_steps": 813, "total_steps": 1955, "loss": 1.049, "lr": 1.4518408636047579e-05, "epoch": 2.079283887468031, "percentage": 41.59, "elapsed_time": "4:11:39", "remaining_time": "5:53:29"} +{"current_steps": 814, "total_steps": 1955, "loss": 1.111, "lr": 1.4502468463215866e-05, "epoch": 2.081841432225064, "percentage": 41.64, "elapsed_time": "4:11:58", "remaining_time": "5:53:11"} +{"current_steps": 815, "total_steps": 1955, "loss": 1.0377, "lr": 1.4486513928261525e-05, "epoch": 2.084398976982097, "percentage": 41.69, "elapsed_time": "4:12:16", "remaining_time": "5:52:52"} +{"current_steps": 816, "total_steps": 1955, "loss": 1.0498, "lr": 1.4470545082076855e-05, "epoch": 2.0869565217391304, "percentage": 41.74, "elapsed_time": "4:12:35", "remaining_time": "5:52:34"} +{"current_steps": 817, "total_steps": 1955, "loss": 1.0612, "lr": 1.445456197559981e-05, "epoch": 2.0895140664961636, "percentage": 41.79, "elapsed_time": "4:12:53", "remaining_time": "5:52:15"} +{"current_steps": 818, "total_steps": 1955, "loss": 1.0406, "lr": 1.4438564659813833e-05, "epoch": 2.0920716112531967, "percentage": 41.84, "elapsed_time": "4:13:12", "remaining_time": "5:51:57"} +{"current_steps": 819, "total_steps": 1955, "loss": 1.0436, "lr": 1.4422553185747693e-05, "epoch": 2.0946291560102304, "percentage": 41.89, "elapsed_time": "4:13:31", "remaining_time": "5:51:38"} +{"current_steps": 820, "total_steps": 1955, "loss": 1.0457, "lr": 1.4406527604475309e-05, "epoch": 2.0971867007672635, "percentage": 41.94, "elapsed_time": "4:13:49", "remaining_time": "5:51:20"} +{"current_steps": 821, "total_steps": 1955, "loss": 1.0576, "lr": 1.4390487967115619e-05, "epoch": 2.0997442455242967, "percentage": 41.99, "elapsed_time": "4:14:08", "remaining_time": "5:51:02"} +{"current_steps": 822, "total_steps": 1955, "loss": 1.0421, "lr": 1.4374434324832386e-05, "epoch": 2.10230179028133, "percentage": 42.05, "elapsed_time": "4:14:27", "remaining_time": "5:50:43"} +{"current_steps": 823, "total_steps": 1955, "loss": 1.0737, "lr": 1.4358366728834043e-05, "epoch": 2.104859335038363, "percentage": 42.1, "elapsed_time": "4:14:46", "remaining_time": "5:50:25"} +{"current_steps": 824, "total_steps": 1955, "loss": 1.067, "lr": 1.434228523037355e-05, "epoch": 2.1074168797953963, "percentage": 42.15, "elapsed_time": "4:15:04", "remaining_time": "5:50:06"} +{"current_steps": 825, "total_steps": 1955, "loss": 1.067, "lr": 1.4326189880748187e-05, "epoch": 2.10997442455243, "percentage": 42.2, "elapsed_time": "4:15:23", "remaining_time": "5:49:48"} +{"current_steps": 826, "total_steps": 1955, "loss": 1.0641, "lr": 1.4310080731299443e-05, "epoch": 2.112531969309463, "percentage": 42.25, "elapsed_time": "4:15:41", "remaining_time": "5:49:29"} +{"current_steps": 827, "total_steps": 1955, "loss": 1.0547, "lr": 1.4293957833412813e-05, "epoch": 2.1150895140664963, "percentage": 42.3, "elapsed_time": "4:16:00", "remaining_time": "5:49:11"} +{"current_steps": 828, "total_steps": 1955, "loss": 1.066, "lr": 1.4277821238517644e-05, "epoch": 2.1176470588235294, "percentage": 42.35, "elapsed_time": "4:16:19", "remaining_time": "5:48:52"} +{"current_steps": 829, "total_steps": 1955, "loss": 1.0691, "lr": 1.4261670998086988e-05, "epoch": 2.1202046035805626, "percentage": 42.4, "elapsed_time": "4:16:37", "remaining_time": "5:48:34"} +{"current_steps": 830, "total_steps": 1955, "loss": 1.0967, "lr": 1.4245507163637406e-05, "epoch": 2.122762148337596, "percentage": 42.46, "elapsed_time": "4:16:56", "remaining_time": "5:48:15"} +{"current_steps": 831, "total_steps": 1955, "loss": 1.0879, "lr": 1.4229329786728839e-05, "epoch": 2.125319693094629, "percentage": 42.51, "elapsed_time": "4:17:15", "remaining_time": "5:47:57"} +{"current_steps": 832, "total_steps": 1955, "loss": 1.0847, "lr": 1.4213138918964415e-05, "epoch": 2.1278772378516626, "percentage": 42.56, "elapsed_time": "4:17:33", "remaining_time": "5:47:38"} +{"current_steps": 833, "total_steps": 1955, "loss": 1.0836, "lr": 1.4196934611990296e-05, "epoch": 2.130434782608696, "percentage": 42.61, "elapsed_time": "4:17:52", "remaining_time": "5:47:20"} +{"current_steps": 834, "total_steps": 1955, "loss": 1.0431, "lr": 1.4180716917495521e-05, "epoch": 2.132992327365729, "percentage": 42.66, "elapsed_time": "4:18:11", "remaining_time": "5:47:02"} +{"current_steps": 835, "total_steps": 1955, "loss": 1.0748, "lr": 1.4164485887211825e-05, "epoch": 2.135549872122762, "percentage": 42.71, "elapsed_time": "4:18:30", "remaining_time": "5:46:43"} +{"current_steps": 836, "total_steps": 1955, "loss": 1.0915, "lr": 1.4148241572913482e-05, "epoch": 2.1381074168797953, "percentage": 42.76, "elapsed_time": "4:18:48", "remaining_time": "5:46:25"} +{"current_steps": 837, "total_steps": 1955, "loss": 1.0289, "lr": 1.4131984026417147e-05, "epoch": 2.1406649616368285, "percentage": 42.81, "elapsed_time": "4:19:07", "remaining_time": "5:46:06"} +{"current_steps": 838, "total_steps": 1955, "loss": 1.0898, "lr": 1.4115713299581677e-05, "epoch": 2.1432225063938617, "percentage": 42.86, "elapsed_time": "4:19:25", "remaining_time": "5:45:48"} +{"current_steps": 839, "total_steps": 1955, "loss": 1.067, "lr": 1.4099429444307973e-05, "epoch": 2.1457800511508953, "percentage": 42.92, "elapsed_time": "4:19:44", "remaining_time": "5:45:29"} +{"current_steps": 840, "total_steps": 1955, "loss": 1.0905, "lr": 1.4083132512538815e-05, "epoch": 2.1483375959079285, "percentage": 42.97, "elapsed_time": "4:20:03", "remaining_time": "5:45:11"} +{"current_steps": 841, "total_steps": 1955, "loss": 1.0666, "lr": 1.4066822556258694e-05, "epoch": 2.1508951406649617, "percentage": 43.02, "elapsed_time": "4:20:21", "remaining_time": "5:44:52"} +{"current_steps": 842, "total_steps": 1955, "loss": 1.0798, "lr": 1.4050499627493648e-05, "epoch": 2.153452685421995, "percentage": 43.07, "elapsed_time": "4:20:39", "remaining_time": "5:44:33"} +{"current_steps": 843, "total_steps": 1955, "loss": 1.0529, "lr": 1.4034163778311096e-05, "epoch": 2.156010230179028, "percentage": 43.12, "elapsed_time": "4:20:58", "remaining_time": "5:44:15"} +{"current_steps": 844, "total_steps": 1955, "loss": 1.1025, "lr": 1.4017815060819667e-05, "epoch": 2.1585677749360612, "percentage": 43.17, "elapsed_time": "4:21:16", "remaining_time": "5:43:55"} +{"current_steps": 845, "total_steps": 1955, "loss": 1.0612, "lr": 1.4001453527169035e-05, "epoch": 2.1611253196930944, "percentage": 43.22, "elapsed_time": "4:21:35", "remaining_time": "5:43:37"} +{"current_steps": 846, "total_steps": 1955, "loss": 1.0569, "lr": 1.3985079229549771e-05, "epoch": 2.163682864450128, "percentage": 43.27, "elapsed_time": "4:21:53", "remaining_time": "5:43:18"} +{"current_steps": 847, "total_steps": 1955, "loss": 1.0174, "lr": 1.3968692220193145e-05, "epoch": 2.166240409207161, "percentage": 43.32, "elapsed_time": "4:22:11", "remaining_time": "5:42:59"} +{"current_steps": 848, "total_steps": 1955, "loss": 1.11, "lr": 1.3952292551370978e-05, "epoch": 2.1687979539641944, "percentage": 43.38, "elapsed_time": "4:22:30", "remaining_time": "5:42:40"} +{"current_steps": 849, "total_steps": 1955, "loss": 1.0695, "lr": 1.3935880275395483e-05, "epoch": 2.1713554987212276, "percentage": 43.43, "elapsed_time": "4:22:48", "remaining_time": "5:42:21"} +{"current_steps": 850, "total_steps": 1955, "loss": 1.07, "lr": 1.3919455444619074e-05, "epoch": 2.1739130434782608, "percentage": 43.48, "elapsed_time": "4:23:06", "remaining_time": "5:42:02"} +{"current_steps": 851, "total_steps": 1955, "loss": 1.0748, "lr": 1.390301811143422e-05, "epoch": 2.176470588235294, "percentage": 43.53, "elapsed_time": "4:23:25", "remaining_time": "5:41:44"} +{"current_steps": 852, "total_steps": 1955, "loss": 1.0886, "lr": 1.388656832827327e-05, "epoch": 2.1790281329923276, "percentage": 43.58, "elapsed_time": "4:23:44", "remaining_time": "5:41:26"} +{"current_steps": 853, "total_steps": 1955, "loss": 1.026, "lr": 1.3870106147608282e-05, "epoch": 2.1815856777493607, "percentage": 43.63, "elapsed_time": "4:24:02", "remaining_time": "5:41:07"} +{"current_steps": 854, "total_steps": 1955, "loss": 1.0095, "lr": 1.385363162195087e-05, "epoch": 2.184143222506394, "percentage": 43.68, "elapsed_time": "4:24:21", "remaining_time": "5:40:48"} +{"current_steps": 855, "total_steps": 1955, "loss": 1.0625, "lr": 1.3837144803852016e-05, "epoch": 2.186700767263427, "percentage": 43.73, "elapsed_time": "4:24:39", "remaining_time": "5:40:30"} +{"current_steps": 856, "total_steps": 1955, "loss": 1.0819, "lr": 1.3820645745901917e-05, "epoch": 2.1892583120204603, "percentage": 43.79, "elapsed_time": "4:24:57", "remaining_time": "5:40:11"} +{"current_steps": 857, "total_steps": 1955, "loss": 1.0987, "lr": 1.3804134500729817e-05, "epoch": 2.1918158567774935, "percentage": 43.84, "elapsed_time": "4:25:16", "remaining_time": "5:39:52"} +{"current_steps": 858, "total_steps": 1955, "loss": 1.091, "lr": 1.3787611121003825e-05, "epoch": 2.1943734015345266, "percentage": 43.89, "elapsed_time": "4:25:35", "remaining_time": "5:39:34"} +{"current_steps": 859, "total_steps": 1955, "loss": 1.0872, "lr": 1.377107565943077e-05, "epoch": 2.1969309462915603, "percentage": 43.94, "elapsed_time": "4:25:53", "remaining_time": "5:39:15"} +{"current_steps": 860, "total_steps": 1955, "loss": 1.0834, "lr": 1.3754528168756007e-05, "epoch": 2.1994884910485935, "percentage": 43.99, "elapsed_time": "4:26:12", "remaining_time": "5:38:57"} +{"current_steps": 861, "total_steps": 1955, "loss": 1.0766, "lr": 1.3737968701763275e-05, "epoch": 2.2020460358056266, "percentage": 44.04, "elapsed_time": "4:26:30", "remaining_time": "5:38:37"} +{"current_steps": 862, "total_steps": 1955, "loss": 1.0488, "lr": 1.3721397311274505e-05, "epoch": 2.20460358056266, "percentage": 44.09, "elapsed_time": "4:26:49", "remaining_time": "5:38:19"} +{"current_steps": 863, "total_steps": 1955, "loss": 1.0714, "lr": 1.3704814050149664e-05, "epoch": 2.207161125319693, "percentage": 44.14, "elapsed_time": "4:27:08", "remaining_time": "5:38:01"} +{"current_steps": 864, "total_steps": 1955, "loss": 1.0745, "lr": 1.3688218971286588e-05, "epoch": 2.209718670076726, "percentage": 44.19, "elapsed_time": "4:27:27", "remaining_time": "5:37:43"} +{"current_steps": 865, "total_steps": 1955, "loss": 1.0298, "lr": 1.367161212762081e-05, "epoch": 2.21227621483376, "percentage": 44.25, "elapsed_time": "4:27:46", "remaining_time": "5:37:25"} +{"current_steps": 866, "total_steps": 1955, "loss": 1.0609, "lr": 1.3654993572125385e-05, "epoch": 2.214833759590793, "percentage": 44.3, "elapsed_time": "4:28:04", "remaining_time": "5:37:05"} +{"current_steps": 867, "total_steps": 1955, "loss": 1.1041, "lr": 1.3638363357810734e-05, "epoch": 2.217391304347826, "percentage": 44.35, "elapsed_time": "4:28:22", "remaining_time": "5:36:47"} +{"current_steps": 868, "total_steps": 1955, "loss": 1.0437, "lr": 1.362172153772446e-05, "epoch": 2.2199488491048593, "percentage": 44.4, "elapsed_time": "4:28:41", "remaining_time": "5:36:29"} +{"current_steps": 869, "total_steps": 1955, "loss": 1.0818, "lr": 1.3605068164951195e-05, "epoch": 2.2225063938618925, "percentage": 44.45, "elapsed_time": "4:29:00", "remaining_time": "5:36:10"} +{"current_steps": 870, "total_steps": 1955, "loss": 1.0667, "lr": 1.3588403292612409e-05, "epoch": 2.2250639386189257, "percentage": 44.5, "elapsed_time": "4:29:18", "remaining_time": "5:35:52"} +{"current_steps": 871, "total_steps": 1955, "loss": 1.0468, "lr": 1.3571726973866275e-05, "epoch": 2.227621483375959, "percentage": 44.55, "elapsed_time": "4:29:37", "remaining_time": "5:35:33"} +{"current_steps": 872, "total_steps": 1955, "loss": 1.0681, "lr": 1.3555039261907452e-05, "epoch": 2.2301790281329925, "percentage": 44.6, "elapsed_time": "4:29:56", "remaining_time": "5:35:15"} +{"current_steps": 873, "total_steps": 1955, "loss": 1.0846, "lr": 1.3538340209966965e-05, "epoch": 2.2327365728900257, "percentage": 44.65, "elapsed_time": "4:30:14", "remaining_time": "5:34:55"} +{"current_steps": 874, "total_steps": 1955, "loss": 1.0731, "lr": 1.3521629871311997e-05, "epoch": 2.235294117647059, "percentage": 44.71, "elapsed_time": "4:30:32", "remaining_time": "5:34:37"} +{"current_steps": 875, "total_steps": 1955, "loss": 1.0393, "lr": 1.3504908299245738e-05, "epoch": 2.237851662404092, "percentage": 44.76, "elapsed_time": "4:30:50", "remaining_time": "5:34:18"} +{"current_steps": 876, "total_steps": 1955, "loss": 1.0592, "lr": 1.3488175547107214e-05, "epoch": 2.2404092071611252, "percentage": 44.81, "elapsed_time": "4:31:09", "remaining_time": "5:33:59"} +{"current_steps": 877, "total_steps": 1955, "loss": 1.0488, "lr": 1.3471431668271105e-05, "epoch": 2.2429667519181584, "percentage": 44.86, "elapsed_time": "4:31:28", "remaining_time": "5:33:41"} +{"current_steps": 878, "total_steps": 1955, "loss": 1.0557, "lr": 1.3454676716147595e-05, "epoch": 2.2455242966751916, "percentage": 44.91, "elapsed_time": "4:31:46", "remaining_time": "5:33:22"} +{"current_steps": 879, "total_steps": 1955, "loss": 1.0557, "lr": 1.343791074418218e-05, "epoch": 2.2480818414322252, "percentage": 44.96, "elapsed_time": "4:32:05", "remaining_time": "5:33:04"} +{"current_steps": 880, "total_steps": 1955, "loss": 1.0305, "lr": 1.342113380585551e-05, "epoch": 2.2506393861892584, "percentage": 45.01, "elapsed_time": "4:32:24", "remaining_time": "5:32:46"} +{"current_steps": 881, "total_steps": 1955, "loss": 1.0502, "lr": 1.3404345954683223e-05, "epoch": 2.2531969309462916, "percentage": 45.06, "elapsed_time": "4:32:43", "remaining_time": "5:32:27"} +{"current_steps": 882, "total_steps": 1955, "loss": 1.017, "lr": 1.3387547244215755e-05, "epoch": 2.2557544757033248, "percentage": 45.12, "elapsed_time": "4:33:01", "remaining_time": "5:32:09"} +{"current_steps": 883, "total_steps": 1955, "loss": 1.0577, "lr": 1.337073772803819e-05, "epoch": 2.258312020460358, "percentage": 45.17, "elapsed_time": "4:33:20", "remaining_time": "5:31:50"} +{"current_steps": 884, "total_steps": 1955, "loss": 1.0523, "lr": 1.3353917459770078e-05, "epoch": 2.260869565217391, "percentage": 45.22, "elapsed_time": "4:33:38", "remaining_time": "5:31:31"} +{"current_steps": 885, "total_steps": 1955, "loss": 1.0727, "lr": 1.3337086493065266e-05, "epoch": 2.2634271099744243, "percentage": 45.27, "elapsed_time": "4:33:56", "remaining_time": "5:31:12"} +{"current_steps": 886, "total_steps": 1955, "loss": 1.0801, "lr": 1.3320244881611728e-05, "epoch": 2.265984654731458, "percentage": 45.32, "elapsed_time": "4:34:15", "remaining_time": "5:30:54"} +{"current_steps": 887, "total_steps": 1955, "loss": 1.0342, "lr": 1.3303392679131393e-05, "epoch": 2.268542199488491, "percentage": 45.37, "elapsed_time": "4:34:33", "remaining_time": "5:30:35"} +{"current_steps": 888, "total_steps": 1955, "loss": 1.0403, "lr": 1.3286529939379969e-05, "epoch": 2.2710997442455243, "percentage": 45.42, "elapsed_time": "4:34:52", "remaining_time": "5:30:17"} +{"current_steps": 889, "total_steps": 1955, "loss": 1.0657, "lr": 1.3269656716146786e-05, "epoch": 2.2736572890025575, "percentage": 45.47, "elapsed_time": "4:35:11", "remaining_time": "5:29:58"} +{"current_steps": 890, "total_steps": 1955, "loss": 1.0996, "lr": 1.3252773063254609e-05, "epoch": 2.2762148337595907, "percentage": 45.52, "elapsed_time": "4:35:29", "remaining_time": "5:29:39"} +{"current_steps": 891, "total_steps": 1955, "loss": 1.0756, "lr": 1.3235879034559467e-05, "epoch": 2.2787723785166243, "percentage": 45.58, "elapsed_time": "4:35:48", "remaining_time": "5:29:21"} +{"current_steps": 892, "total_steps": 1955, "loss": 1.0592, "lr": 1.321897468395049e-05, "epoch": 2.2813299232736575, "percentage": 45.63, "elapsed_time": "4:36:07", "remaining_time": "5:29:02"} +{"current_steps": 893, "total_steps": 1955, "loss": 1.0656, "lr": 1.320206006534974e-05, "epoch": 2.2838874680306906, "percentage": 45.68, "elapsed_time": "4:36:24", "remaining_time": "5:28:43"} +{"current_steps": 894, "total_steps": 1955, "loss": 1.0862, "lr": 1.3185135232712024e-05, "epoch": 2.286445012787724, "percentage": 45.73, "elapsed_time": "4:36:43", "remaining_time": "5:28:25"} +{"current_steps": 895, "total_steps": 1955, "loss": 1.0757, "lr": 1.316820024002473e-05, "epoch": 2.289002557544757, "percentage": 45.78, "elapsed_time": "4:37:02", "remaining_time": "5:28:06"} +{"current_steps": 896, "total_steps": 1955, "loss": 1.0618, "lr": 1.3151255141307658e-05, "epoch": 2.29156010230179, "percentage": 45.83, "elapsed_time": "4:37:20", "remaining_time": "5:27:48"} +{"current_steps": 897, "total_steps": 1955, "loss": 1.0607, "lr": 1.3134299990612843e-05, "epoch": 2.2941176470588234, "percentage": 45.88, "elapsed_time": "4:37:38", "remaining_time": "5:27:28"} +{"current_steps": 898, "total_steps": 1955, "loss": 1.036, "lr": 1.3117334842024386e-05, "epoch": 2.296675191815857, "percentage": 45.93, "elapsed_time": "4:37:57", "remaining_time": "5:27:10"} +{"current_steps": 899, "total_steps": 1955, "loss": 1.0722, "lr": 1.310035974965828e-05, "epoch": 2.29923273657289, "percentage": 45.98, "elapsed_time": "4:38:16", "remaining_time": "5:26:51"} +{"current_steps": 900, "total_steps": 1955, "loss": 1.0559, "lr": 1.308337476766223e-05, "epoch": 2.3017902813299234, "percentage": 46.04, "elapsed_time": "4:38:34", "remaining_time": "5:26:32"} +{"current_steps": 901, "total_steps": 1955, "loss": 1.0473, "lr": 1.3066379950215499e-05, "epoch": 2.3043478260869565, "percentage": 46.09, "elapsed_time": "4:38:52", "remaining_time": "5:26:13"} +{"current_steps": 902, "total_steps": 1955, "loss": 1.0525, "lr": 1.3049375351528713e-05, "epoch": 2.3069053708439897, "percentage": 46.14, "elapsed_time": "4:39:11", "remaining_time": "5:25:55"} +{"current_steps": 903, "total_steps": 1955, "loss": 1.0763, "lr": 1.3032361025843704e-05, "epoch": 2.309462915601023, "percentage": 46.19, "elapsed_time": "4:39:30", "remaining_time": "5:25:37"} +{"current_steps": 904, "total_steps": 1955, "loss": 0.9991, "lr": 1.301533702743333e-05, "epoch": 2.312020460358056, "percentage": 46.24, "elapsed_time": "4:39:48", "remaining_time": "5:25:18"} +{"current_steps": 905, "total_steps": 1955, "loss": 1.0369, "lr": 1.29983034106013e-05, "epoch": 2.3145780051150897, "percentage": 46.29, "elapsed_time": "4:40:07", "remaining_time": "5:25:00"} +{"current_steps": 906, "total_steps": 1955, "loss": 1.0461, "lr": 1.2981260229682018e-05, "epoch": 2.317135549872123, "percentage": 46.34, "elapsed_time": "4:40:25", "remaining_time": "5:24:41"} +{"current_steps": 907, "total_steps": 1955, "loss": 1.0505, "lr": 1.2964207539040373e-05, "epoch": 2.319693094629156, "percentage": 46.39, "elapsed_time": "4:40:44", "remaining_time": "5:24:22"} +{"current_steps": 908, "total_steps": 1955, "loss": 1.0912, "lr": 1.2947145393071607e-05, "epoch": 2.3222506393861893, "percentage": 46.45, "elapsed_time": "4:41:02", "remaining_time": "5:24:03"} +{"current_steps": 909, "total_steps": 1955, "loss": 1.0261, "lr": 1.2930073846201116e-05, "epoch": 2.3248081841432224, "percentage": 46.5, "elapsed_time": "4:41:20", "remaining_time": "5:23:45"} +{"current_steps": 910, "total_steps": 1955, "loss": 1.1043, "lr": 1.2912992952884283e-05, "epoch": 2.3273657289002556, "percentage": 46.55, "elapsed_time": "4:41:39", "remaining_time": "5:23:26"} +{"current_steps": 911, "total_steps": 1955, "loss": 1.0806, "lr": 1.289590276760631e-05, "epoch": 2.329923273657289, "percentage": 46.6, "elapsed_time": "4:41:58", "remaining_time": "5:23:07"} +{"current_steps": 912, "total_steps": 1955, "loss": 1.0392, "lr": 1.2878803344882029e-05, "epoch": 2.3324808184143224, "percentage": 46.65, "elapsed_time": "4:42:16", "remaining_time": "5:22:49"} +{"current_steps": 913, "total_steps": 1955, "loss": 1.0713, "lr": 1.2861694739255746e-05, "epoch": 2.3350383631713556, "percentage": 46.7, "elapsed_time": "4:42:35", "remaining_time": "5:22:30"} +{"current_steps": 914, "total_steps": 1955, "loss": 1.0704, "lr": 1.2844577005301055e-05, "epoch": 2.337595907928389, "percentage": 46.75, "elapsed_time": "4:42:53", "remaining_time": "5:22:12"} +{"current_steps": 915, "total_steps": 1955, "loss": 1.0685, "lr": 1.2827450197620673e-05, "epoch": 2.340153452685422, "percentage": 46.8, "elapsed_time": "4:43:12", "remaining_time": "5:21:53"} +{"current_steps": 916, "total_steps": 1955, "loss": 1.0895, "lr": 1.2810314370846252e-05, "epoch": 2.342710997442455, "percentage": 46.85, "elapsed_time": "4:43:30", "remaining_time": "5:21:34"} +{"current_steps": 917, "total_steps": 1955, "loss": 1.0299, "lr": 1.2793169579638222e-05, "epoch": 2.3452685421994883, "percentage": 46.91, "elapsed_time": "4:43:49", "remaining_time": "5:21:16"} +{"current_steps": 918, "total_steps": 1955, "loss": 1.0582, "lr": 1.2776015878685603e-05, "epoch": 2.3478260869565215, "percentage": 46.96, "elapsed_time": "4:44:07", "remaining_time": "5:20:57"} +{"current_steps": 919, "total_steps": 1955, "loss": 1.0451, "lr": 1.2758853322705835e-05, "epoch": 2.350383631713555, "percentage": 47.01, "elapsed_time": "4:44:26", "remaining_time": "5:20:38"} +{"current_steps": 920, "total_steps": 1955, "loss": 1.045, "lr": 1.2741681966444608e-05, "epoch": 2.3529411764705883, "percentage": 47.06, "elapsed_time": "4:44:44", "remaining_time": "5:20:20"} +{"current_steps": 921, "total_steps": 1955, "loss": 1.0487, "lr": 1.2724501864675682e-05, "epoch": 2.3554987212276215, "percentage": 47.11, "elapsed_time": "4:45:03", "remaining_time": "5:20:02"} +{"current_steps": 922, "total_steps": 1955, "loss": 1.0481, "lr": 1.270731307220071e-05, "epoch": 2.3580562659846547, "percentage": 47.16, "elapsed_time": "4:45:21", "remaining_time": "5:19:42"} +{"current_steps": 923, "total_steps": 1955, "loss": 1.0515, "lr": 1.269011564384908e-05, "epoch": 2.360613810741688, "percentage": 47.21, "elapsed_time": "4:45:40", "remaining_time": "5:19:24"} +{"current_steps": 924, "total_steps": 1955, "loss": 1.0593, "lr": 1.2672909634477701e-05, "epoch": 2.363171355498721, "percentage": 47.26, "elapsed_time": "4:45:58", "remaining_time": "5:19:05"} +{"current_steps": 925, "total_steps": 1955, "loss": 1.0468, "lr": 1.2655695098970881e-05, "epoch": 2.3657289002557547, "percentage": 47.31, "elapsed_time": "4:46:16", "remaining_time": "5:18:46"} +{"current_steps": 926, "total_steps": 1955, "loss": 1.0224, "lr": 1.2638472092240112e-05, "epoch": 2.368286445012788, "percentage": 47.37, "elapsed_time": "4:46:34", "remaining_time": "5:18:27"} +{"current_steps": 927, "total_steps": 1955, "loss": 1.0799, "lr": 1.2621240669223906e-05, "epoch": 2.370843989769821, "percentage": 47.42, "elapsed_time": "4:46:53", "remaining_time": "5:18:08"} +{"current_steps": 928, "total_steps": 1955, "loss": 1.0855, "lr": 1.2604000884887636e-05, "epoch": 2.373401534526854, "percentage": 47.47, "elapsed_time": "4:47:12", "remaining_time": "5:17:50"} +{"current_steps": 929, "total_steps": 1955, "loss": 1.0541, "lr": 1.2586752794223321e-05, "epoch": 2.3759590792838874, "percentage": 47.52, "elapsed_time": "4:47:30", "remaining_time": "5:17:31"} +{"current_steps": 930, "total_steps": 1955, "loss": 1.0786, "lr": 1.25694964522495e-05, "epoch": 2.3785166240409206, "percentage": 47.57, "elapsed_time": "4:47:49", "remaining_time": "5:17:13"} +{"current_steps": 931, "total_steps": 1955, "loss": 1.0452, "lr": 1.2552231914011017e-05, "epoch": 2.381074168797954, "percentage": 47.62, "elapsed_time": "4:48:08", "remaining_time": "5:16:55"} +{"current_steps": 932, "total_steps": 1955, "loss": 1.071, "lr": 1.2534959234578872e-05, "epoch": 2.3836317135549874, "percentage": 47.67, "elapsed_time": "4:48:26", "remaining_time": "5:16:36"} +{"current_steps": 933, "total_steps": 1955, "loss": 1.053, "lr": 1.2517678469050023e-05, "epoch": 2.3861892583120206, "percentage": 47.72, "elapsed_time": "4:48:45", "remaining_time": "5:16:18"} +{"current_steps": 934, "total_steps": 1955, "loss": 1.0411, "lr": 1.2500389672547233e-05, "epoch": 2.3887468030690537, "percentage": 47.77, "elapsed_time": "4:49:04", "remaining_time": "5:15:59"} +{"current_steps": 935, "total_steps": 1955, "loss": 1.0397, "lr": 1.2483092900218872e-05, "epoch": 2.391304347826087, "percentage": 47.83, "elapsed_time": "4:49:22", "remaining_time": "5:15:41"} +{"current_steps": 936, "total_steps": 1955, "loss": 1.0579, "lr": 1.2465788207238754e-05, "epoch": 2.39386189258312, "percentage": 47.88, "elapsed_time": "4:49:41", "remaining_time": "5:15:22"} +{"current_steps": 937, "total_steps": 1955, "loss": 1.06, "lr": 1.2448475648805965e-05, "epoch": 2.3964194373401533, "percentage": 47.93, "elapsed_time": "4:50:00", "remaining_time": "5:15:04"} +{"current_steps": 938, "total_steps": 1955, "loss": 1.0693, "lr": 1.243115528014467e-05, "epoch": 2.398976982097187, "percentage": 47.98, "elapsed_time": "4:50:18", "remaining_time": "5:14:45"} +{"current_steps": 939, "total_steps": 1955, "loss": 1.0517, "lr": 1.2413827156503959e-05, "epoch": 2.40153452685422, "percentage": 48.03, "elapsed_time": "4:50:36", "remaining_time": "5:14:26"} +{"current_steps": 940, "total_steps": 1955, "loss": 1.0618, "lr": 1.2396491333157653e-05, "epoch": 2.4040920716112533, "percentage": 48.08, "elapsed_time": "4:50:55", "remaining_time": "5:14:08"} +{"current_steps": 941, "total_steps": 1955, "loss": 1.0503, "lr": 1.2379147865404126e-05, "epoch": 2.4066496163682864, "percentage": 48.13, "elapsed_time": "4:51:14", "remaining_time": "5:13:50"} +{"current_steps": 942, "total_steps": 1955, "loss": 1.0608, "lr": 1.2361796808566154e-05, "epoch": 2.4092071611253196, "percentage": 48.18, "elapsed_time": "4:51:32", "remaining_time": "5:13:31"} +{"current_steps": 943, "total_steps": 1955, "loss": 1.0632, "lr": 1.2344438217990706e-05, "epoch": 2.411764705882353, "percentage": 48.24, "elapsed_time": "4:51:51", "remaining_time": "5:13:12"} +{"current_steps": 944, "total_steps": 1955, "loss": 1.0639, "lr": 1.2327072149048787e-05, "epoch": 2.414322250639386, "percentage": 48.29, "elapsed_time": "4:52:09", "remaining_time": "5:12:53"} +{"current_steps": 945, "total_steps": 1955, "loss": 1.0162, "lr": 1.2309698657135263e-05, "epoch": 2.4168797953964196, "percentage": 48.34, "elapsed_time": "4:52:28", "remaining_time": "5:12:35"} +{"current_steps": 946, "total_steps": 1955, "loss": 1.0363, "lr": 1.2292317797668664e-05, "epoch": 2.419437340153453, "percentage": 48.39, "elapsed_time": "4:52:47", "remaining_time": "5:12:17"} +{"current_steps": 947, "total_steps": 1955, "loss": 1.0881, "lr": 1.2274929626091035e-05, "epoch": 2.421994884910486, "percentage": 48.44, "elapsed_time": "4:53:05", "remaining_time": "5:11:58"} +{"current_steps": 948, "total_steps": 1955, "loss": 1.0446, "lr": 1.2257534197867744e-05, "epoch": 2.424552429667519, "percentage": 48.49, "elapsed_time": "4:53:24", "remaining_time": "5:11:39"} +{"current_steps": 949, "total_steps": 1955, "loss": 1.0616, "lr": 1.2240131568487292e-05, "epoch": 2.4271099744245523, "percentage": 48.54, "elapsed_time": "4:53:42", "remaining_time": "5:11:21"} +{"current_steps": 950, "total_steps": 1955, "loss": 1.0419, "lr": 1.222272179346117e-05, "epoch": 2.4296675191815855, "percentage": 48.59, "elapsed_time": "4:54:01", "remaining_time": "5:11:02"} +{"current_steps": 951, "total_steps": 1955, "loss": 1.0595, "lr": 1.2205304928323649e-05, "epoch": 2.4322250639386187, "percentage": 48.64, "elapsed_time": "4:54:20", "remaining_time": "5:10:44"} +{"current_steps": 952, "total_steps": 1955, "loss": 1.022, "lr": 1.2187881028631621e-05, "epoch": 2.4347826086956523, "percentage": 48.7, "elapsed_time": "4:54:38", "remaining_time": "5:10:26"} +{"current_steps": 953, "total_steps": 1955, "loss": 1.0756, "lr": 1.2170450149964422e-05, "epoch": 2.4373401534526855, "percentage": 48.75, "elapsed_time": "4:54:57", "remaining_time": "5:10:07"} +{"current_steps": 954, "total_steps": 1955, "loss": 1.0771, "lr": 1.2153012347923635e-05, "epoch": 2.4398976982097187, "percentage": 48.8, "elapsed_time": "4:55:16", "remaining_time": "5:09:49"} +{"current_steps": 955, "total_steps": 1955, "loss": 1.0523, "lr": 1.2135567678132942e-05, "epoch": 2.442455242966752, "percentage": 48.85, "elapsed_time": "4:55:34", "remaining_time": "5:09:30"} +{"current_steps": 956, "total_steps": 1955, "loss": 1.059, "lr": 1.211811619623793e-05, "epoch": 2.445012787723785, "percentage": 48.9, "elapsed_time": "4:55:52", "remaining_time": "5:09:11"} +{"current_steps": 957, "total_steps": 1955, "loss": 1.0307, "lr": 1.2100657957905908e-05, "epoch": 2.4475703324808182, "percentage": 48.95, "elapsed_time": "4:56:11", "remaining_time": "5:08:53"} +{"current_steps": 958, "total_steps": 1955, "loss": 1.025, "lr": 1.2083193018825745e-05, "epoch": 2.4501278772378514, "percentage": 49.0, "elapsed_time": "4:56:30", "remaining_time": "5:08:34"} +{"current_steps": 959, "total_steps": 1955, "loss": 1.0474, "lr": 1.2065721434707677e-05, "epoch": 2.452685421994885, "percentage": 49.05, "elapsed_time": "4:56:49", "remaining_time": "5:08:16"} +{"current_steps": 960, "total_steps": 1955, "loss": 1.0226, "lr": 1.2048243261283144e-05, "epoch": 2.455242966751918, "percentage": 49.1, "elapsed_time": "4:57:07", "remaining_time": "5:07:57"} +{"current_steps": 961, "total_steps": 1955, "loss": 1.0924, "lr": 1.2030758554304593e-05, "epoch": 2.4578005115089514, "percentage": 49.16, "elapsed_time": "4:57:25", "remaining_time": "5:07:38"} +{"current_steps": 962, "total_steps": 1955, "loss": 1.0667, "lr": 1.2013267369545329e-05, "epoch": 2.4603580562659846, "percentage": 49.21, "elapsed_time": "4:57:44", "remaining_time": "5:07:20"} +{"current_steps": 963, "total_steps": 1955, "loss": 1.0754, "lr": 1.1995769762799307e-05, "epoch": 2.4629156010230178, "percentage": 49.26, "elapsed_time": "4:58:02", "remaining_time": "5:07:01"} +{"current_steps": 964, "total_steps": 1955, "loss": 1.0808, "lr": 1.1978265789880974e-05, "epoch": 2.4654731457800514, "percentage": 49.31, "elapsed_time": "4:58:21", "remaining_time": "5:06:42"} +{"current_steps": 965, "total_steps": 1955, "loss": 1.0307, "lr": 1.1960755506625077e-05, "epoch": 2.4680306905370846, "percentage": 49.36, "elapsed_time": "4:58:40", "remaining_time": "5:06:24"} +{"current_steps": 966, "total_steps": 1955, "loss": 1.046, "lr": 1.1943238968886492e-05, "epoch": 2.4705882352941178, "percentage": 49.41, "elapsed_time": "4:58:58", "remaining_time": "5:06:05"} +{"current_steps": 967, "total_steps": 1955, "loss": 1.0571, "lr": 1.1925716232540062e-05, "epoch": 2.473145780051151, "percentage": 49.46, "elapsed_time": "4:59:17", "remaining_time": "5:05:47"} +{"current_steps": 968, "total_steps": 1955, "loss": 1.0621, "lr": 1.1908187353480381e-05, "epoch": 2.475703324808184, "percentage": 49.51, "elapsed_time": "4:59:35", "remaining_time": "5:05:28"} +{"current_steps": 969, "total_steps": 1955, "loss": 1.0603, "lr": 1.1890652387621645e-05, "epoch": 2.4782608695652173, "percentage": 49.57, "elapsed_time": "4:59:54", "remaining_time": "5:05:09"} +{"current_steps": 970, "total_steps": 1955, "loss": 1.0301, "lr": 1.1873111390897475e-05, "epoch": 2.4808184143222505, "percentage": 49.62, "elapsed_time": "5:00:12", "remaining_time": "5:04:51"} +{"current_steps": 971, "total_steps": 1955, "loss": 1.0531, "lr": 1.1855564419260714e-05, "epoch": 2.483375959079284, "percentage": 49.67, "elapsed_time": "5:00:31", "remaining_time": "5:04:32"} +{"current_steps": 972, "total_steps": 1955, "loss": 1.074, "lr": 1.1838011528683279e-05, "epoch": 2.4859335038363173, "percentage": 49.72, "elapsed_time": "5:00:49", "remaining_time": "5:04:13"} +{"current_steps": 973, "total_steps": 1955, "loss": 1.0536, "lr": 1.1820452775155957e-05, "epoch": 2.4884910485933505, "percentage": 49.77, "elapsed_time": "5:01:08", "remaining_time": "5:03:55"} +{"current_steps": 974, "total_steps": 1955, "loss": 1.0526, "lr": 1.1802888214688235e-05, "epoch": 2.4910485933503836, "percentage": 49.82, "elapsed_time": "5:01:26", "remaining_time": "5:03:36"} +{"current_steps": 975, "total_steps": 1955, "loss": 1.0319, "lr": 1.1785317903308138e-05, "epoch": 2.493606138107417, "percentage": 49.87, "elapsed_time": "5:01:44", "remaining_time": "5:03:17"} +{"current_steps": 976, "total_steps": 1955, "loss": 1.0613, "lr": 1.1767741897062018e-05, "epoch": 2.49616368286445, "percentage": 49.92, "elapsed_time": "5:02:02", "remaining_time": "5:02:58"} +{"current_steps": 977, "total_steps": 1955, "loss": 1.0498, "lr": 1.1750160252014401e-05, "epoch": 2.498721227621483, "percentage": 49.97, "elapsed_time": "5:02:21", "remaining_time": "5:02:39"} +{"current_steps": 978, "total_steps": 1955, "loss": 1.0659, "lr": 1.1732573024247803e-05, "epoch": 2.501278772378517, "percentage": 50.03, "elapsed_time": "5:02:40", "remaining_time": "5:02:21"} +{"current_steps": 979, "total_steps": 1955, "loss": 1.0702, "lr": 1.171498026986254e-05, "epoch": 2.50383631713555, "percentage": 50.08, "elapsed_time": "5:02:58", "remaining_time": "5:02:03"} +{"current_steps": 980, "total_steps": 1955, "loss": 1.0288, "lr": 1.1697382044976563e-05, "epoch": 2.506393861892583, "percentage": 50.13, "elapsed_time": "5:03:16", "remaining_time": "5:01:43"} +{"current_steps": 981, "total_steps": 1955, "loss": 1.0376, "lr": 1.1679778405725274e-05, "epoch": 2.5089514066496164, "percentage": 50.18, "elapsed_time": "5:03:35", "remaining_time": "5:01:25"} +{"current_steps": 982, "total_steps": 1955, "loss": 1.0187, "lr": 1.166216940826134e-05, "epoch": 2.5115089514066495, "percentage": 50.23, "elapsed_time": "5:03:54", "remaining_time": "5:01:06"} +{"current_steps": 983, "total_steps": 1955, "loss": 1.0308, "lr": 1.1644555108754518e-05, "epoch": 2.5140664961636827, "percentage": 50.28, "elapsed_time": "5:04:12", "remaining_time": "5:00:48"} +{"current_steps": 984, "total_steps": 1955, "loss": 1.0618, "lr": 1.1626935563391491e-05, "epoch": 2.516624040920716, "percentage": 50.33, "elapsed_time": "5:04:31", "remaining_time": "5:00:29"} +{"current_steps": 985, "total_steps": 1955, "loss": 1.0594, "lr": 1.1609310828375662e-05, "epoch": 2.5191815856777495, "percentage": 50.38, "elapsed_time": "5:04:50", "remaining_time": "5:00:11"} +{"current_steps": 986, "total_steps": 1955, "loss": 1.0322, "lr": 1.1591680959926994e-05, "epoch": 2.5217391304347827, "percentage": 50.43, "elapsed_time": "5:05:08", "remaining_time": "4:59:53"} +{"current_steps": 987, "total_steps": 1955, "loss": 1.044, "lr": 1.1574046014281823e-05, "epoch": 2.524296675191816, "percentage": 50.49, "elapsed_time": "5:05:26", "remaining_time": "4:59:34"} +{"current_steps": 988, "total_steps": 1955, "loss": 1.0695, "lr": 1.1556406047692681e-05, "epoch": 2.526854219948849, "percentage": 50.54, "elapsed_time": "5:05:45", "remaining_time": "4:59:15"} +{"current_steps": 989, "total_steps": 1955, "loss": 1.0685, "lr": 1.153876111642812e-05, "epoch": 2.5294117647058822, "percentage": 50.59, "elapsed_time": "5:06:03", "remaining_time": "4:58:56"} +{"current_steps": 990, "total_steps": 1955, "loss": 1.0778, "lr": 1.1521111276772519e-05, "epoch": 2.531969309462916, "percentage": 50.64, "elapsed_time": "5:06:21", "remaining_time": "4:58:37"} +{"current_steps": 991, "total_steps": 1955, "loss": 1.0573, "lr": 1.1503456585025918e-05, "epoch": 2.5345268542199486, "percentage": 50.69, "elapsed_time": "5:06:40", "remaining_time": "4:58:19"} +{"current_steps": 992, "total_steps": 1955, "loss": 1.0753, "lr": 1.1485797097503848e-05, "epoch": 2.5370843989769822, "percentage": 50.74, "elapsed_time": "5:06:59", "remaining_time": "4:58:00"} +{"current_steps": 993, "total_steps": 1955, "loss": 1.0565, "lr": 1.1468132870537112e-05, "epoch": 2.5396419437340154, "percentage": 50.79, "elapsed_time": "5:07:17", "remaining_time": "4:57:41"} +{"current_steps": 994, "total_steps": 1955, "loss": 1.0388, "lr": 1.1450463960471652e-05, "epoch": 2.5421994884910486, "percentage": 50.84, "elapsed_time": "5:07:35", "remaining_time": "4:57:22"} +{"current_steps": 995, "total_steps": 1955, "loss": 1.0315, "lr": 1.143279042366834e-05, "epoch": 2.544757033248082, "percentage": 50.9, "elapsed_time": "5:07:53", "remaining_time": "4:57:03"} +{"current_steps": 996, "total_steps": 1955, "loss": 1.0494, "lr": 1.1415112316502803e-05, "epoch": 2.547314578005115, "percentage": 50.95, "elapsed_time": "5:08:11", "remaining_time": "4:56:44"} +{"current_steps": 997, "total_steps": 1955, "loss": 1.0528, "lr": 1.139742969536526e-05, "epoch": 2.5498721227621486, "percentage": 51.0, "elapsed_time": "5:08:29", "remaining_time": "4:56:25"} +{"current_steps": 998, "total_steps": 1955, "loss": 1.0512, "lr": 1.137974261666031e-05, "epoch": 2.5524296675191813, "percentage": 51.05, "elapsed_time": "5:08:48", "remaining_time": "4:56:07"} +{"current_steps": 999, "total_steps": 1955, "loss": 1.0457, "lr": 1.1362051136806789e-05, "epoch": 2.554987212276215, "percentage": 51.1, "elapsed_time": "5:09:07", "remaining_time": "4:55:48"} +{"current_steps": 1000, "total_steps": 1955, "loss": 1.0489, "lr": 1.134435531223756e-05, "epoch": 2.557544757033248, "percentage": 51.15, "elapsed_time": "5:09:25", "remaining_time": "4:55:29"} +{"current_steps": 1001, "total_steps": 1955, "loss": 1.0505, "lr": 1.1326655199399345e-05, "epoch": 2.5601023017902813, "percentage": 51.2, "elapsed_time": "5:09:43", "remaining_time": "4:55:11"} +{"current_steps": 1002, "total_steps": 1955, "loss": 1.0412, "lr": 1.1308950854752558e-05, "epoch": 2.5626598465473145, "percentage": 51.25, "elapsed_time": "5:10:01", "remaining_time": "4:54:52"} +{"current_steps": 1003, "total_steps": 1955, "loss": 1.069, "lr": 1.1291242334771095e-05, "epoch": 2.5652173913043477, "percentage": 51.3, "elapsed_time": "5:10:20", "remaining_time": "4:54:33"} +{"current_steps": 1004, "total_steps": 1955, "loss": 1.0759, "lr": 1.1273529695942182e-05, "epoch": 2.5677749360613813, "percentage": 51.36, "elapsed_time": "5:10:39", "remaining_time": "4:54:15"} +{"current_steps": 1005, "total_steps": 1955, "loss": 1.0684, "lr": 1.1255812994766175e-05, "epoch": 2.5703324808184145, "percentage": 51.41, "elapsed_time": "5:10:58", "remaining_time": "4:53:57"} +{"current_steps": 1006, "total_steps": 1955, "loss": 1.0048, "lr": 1.1238092287756398e-05, "epoch": 2.5728900255754477, "percentage": 51.46, "elapsed_time": "5:11:17", "remaining_time": "4:53:38"} +{"current_steps": 1007, "total_steps": 1955, "loss": 1.0829, "lr": 1.1220367631438942e-05, "epoch": 2.575447570332481, "percentage": 51.51, "elapsed_time": "5:11:35", "remaining_time": "4:53:19"} +{"current_steps": 1008, "total_steps": 1955, "loss": 1.0709, "lr": 1.1202639082352507e-05, "epoch": 2.578005115089514, "percentage": 51.56, "elapsed_time": "5:11:53", "remaining_time": "4:53:01"} +{"current_steps": 1009, "total_steps": 1955, "loss": 1.0693, "lr": 1.1184906697048202e-05, "epoch": 2.580562659846547, "percentage": 51.61, "elapsed_time": "5:12:12", "remaining_time": "4:52:42"} +{"current_steps": 1010, "total_steps": 1955, "loss": 1.0562, "lr": 1.1167170532089369e-05, "epoch": 2.5831202046035804, "percentage": 51.66, "elapsed_time": "5:12:30", "remaining_time": "4:52:23"} +{"current_steps": 1011, "total_steps": 1955, "loss": 1.0568, "lr": 1.1149430644051424e-05, "epoch": 2.585677749360614, "percentage": 51.71, "elapsed_time": "5:12:48", "remaining_time": "4:52:04"} +{"current_steps": 1012, "total_steps": 1955, "loss": 1.0614, "lr": 1.1131687089521642e-05, "epoch": 2.588235294117647, "percentage": 51.76, "elapsed_time": "5:13:06", "remaining_time": "4:51:45"} +{"current_steps": 1013, "total_steps": 1955, "loss": 1.0446, "lr": 1.1113939925098997e-05, "epoch": 2.5907928388746804, "percentage": 51.82, "elapsed_time": "5:13:24", "remaining_time": "4:51:26"} +{"current_steps": 1014, "total_steps": 1955, "loss": 1.062, "lr": 1.1096189207393988e-05, "epoch": 2.5933503836317136, "percentage": 51.87, "elapsed_time": "5:13:43", "remaining_time": "4:51:08"} +{"current_steps": 1015, "total_steps": 1955, "loss": 1.0439, "lr": 1.107843499302843e-05, "epoch": 2.5959079283887467, "percentage": 51.92, "elapsed_time": "5:14:01", "remaining_time": "4:50:49"} +{"current_steps": 1016, "total_steps": 1955, "loss": 1.044, "lr": 1.106067733863531e-05, "epoch": 2.59846547314578, "percentage": 51.97, "elapsed_time": "5:14:20", "remaining_time": "4:50:31"} +{"current_steps": 1017, "total_steps": 1955, "loss": 1.0535, "lr": 1.1042916300858584e-05, "epoch": 2.601023017902813, "percentage": 52.02, "elapsed_time": "5:14:38", "remaining_time": "4:50:12"} +{"current_steps": 1018, "total_steps": 1955, "loss": 1.0573, "lr": 1.1025151936352987e-05, "epoch": 2.6035805626598467, "percentage": 52.07, "elapsed_time": "5:14:56", "remaining_time": "4:49:53"} +{"current_steps": 1019, "total_steps": 1955, "loss": 1.0287, "lr": 1.1007384301783884e-05, "epoch": 2.60613810741688, "percentage": 52.12, "elapsed_time": "5:15:14", "remaining_time": "4:49:34"} +{"current_steps": 1020, "total_steps": 1955, "loss": 1.0376, "lr": 1.0989613453827058e-05, "epoch": 2.608695652173913, "percentage": 52.17, "elapsed_time": "5:15:33", "remaining_time": "4:49:15"} +{"current_steps": 1021, "total_steps": 1955, "loss": 1.05, "lr": 1.0971839449168543e-05, "epoch": 2.6112531969309463, "percentage": 52.23, "elapsed_time": "5:15:52", "remaining_time": "4:48:57"} +{"current_steps": 1022, "total_steps": 1955, "loss": 1.0431, "lr": 1.095406234450446e-05, "epoch": 2.6138107416879794, "percentage": 52.28, "elapsed_time": "5:16:10", "remaining_time": "4:48:38"} +{"current_steps": 1023, "total_steps": 1955, "loss": 1.0668, "lr": 1.0936282196540788e-05, "epoch": 2.6163682864450126, "percentage": 52.33, "elapsed_time": "5:16:29", "remaining_time": "4:48:20"} +{"current_steps": 1024, "total_steps": 1955, "loss": 1.0454, "lr": 1.0918499061993241e-05, "epoch": 2.618925831202046, "percentage": 52.38, "elapsed_time": "5:16:48", "remaining_time": "4:48:01"} +{"current_steps": 1025, "total_steps": 1955, "loss": 1.0405, "lr": 1.0900712997587048e-05, "epoch": 2.6214833759590794, "percentage": 52.43, "elapsed_time": "5:17:06", "remaining_time": "4:47:42"} +{"current_steps": 1026, "total_steps": 1955, "loss": 1.0487, "lr": 1.0882924060056782e-05, "epoch": 2.6240409207161126, "percentage": 52.48, "elapsed_time": "5:17:24", "remaining_time": "4:47:24"} +{"current_steps": 1027, "total_steps": 1955, "loss": 1.0561, "lr": 1.0865132306146182e-05, "epoch": 2.626598465473146, "percentage": 52.53, "elapsed_time": "5:17:43", "remaining_time": "4:47:05"} +{"current_steps": 1028, "total_steps": 1955, "loss": 1.0662, "lr": 1.0847337792607978e-05, "epoch": 2.629156010230179, "percentage": 52.58, "elapsed_time": "5:18:01", "remaining_time": "4:46:46"} +{"current_steps": 1029, "total_steps": 1955, "loss": 1.069, "lr": 1.0829540576203695e-05, "epoch": 2.631713554987212, "percentage": 52.63, "elapsed_time": "5:18:20", "remaining_time": "4:46:28"} +{"current_steps": 1030, "total_steps": 1955, "loss": 1.0545, "lr": 1.0811740713703476e-05, "epoch": 2.634271099744246, "percentage": 52.69, "elapsed_time": "5:18:38", "remaining_time": "4:46:09"} +{"current_steps": 1031, "total_steps": 1955, "loss": 1.0598, "lr": 1.0793938261885917e-05, "epoch": 2.6368286445012785, "percentage": 52.74, "elapsed_time": "5:18:57", "remaining_time": "4:45:51"} +{"current_steps": 1032, "total_steps": 1955, "loss": 1.0253, "lr": 1.0776133277537865e-05, "epoch": 2.639386189258312, "percentage": 52.79, "elapsed_time": "5:19:15", "remaining_time": "4:45:32"} +{"current_steps": 1033, "total_steps": 1955, "loss": 1.0847, "lr": 1.075832581745425e-05, "epoch": 2.6419437340153453, "percentage": 52.84, "elapsed_time": "5:19:34", "remaining_time": "4:45:14"} +{"current_steps": 1034, "total_steps": 1955, "loss": 1.0618, "lr": 1.074051593843789e-05, "epoch": 2.6445012787723785, "percentage": 52.89, "elapsed_time": "5:19:53", "remaining_time": "4:44:55"} +{"current_steps": 1035, "total_steps": 1955, "loss": 1.0662, "lr": 1.0722703697299328e-05, "epoch": 2.6470588235294117, "percentage": 52.94, "elapsed_time": "5:20:11", "remaining_time": "4:44:37"} +{"current_steps": 1036, "total_steps": 1955, "loss": 1.0387, "lr": 1.070488915085664e-05, "epoch": 2.649616368286445, "percentage": 52.99, "elapsed_time": "5:20:30", "remaining_time": "4:44:18"} +{"current_steps": 1037, "total_steps": 1955, "loss": 1.0458, "lr": 1.0687072355935257e-05, "epoch": 2.6521739130434785, "percentage": 53.04, "elapsed_time": "5:20:49", "remaining_time": "4:44:00"} +{"current_steps": 1038, "total_steps": 1955, "loss": 1.0501, "lr": 1.0669253369367776e-05, "epoch": 2.6547314578005117, "percentage": 53.09, "elapsed_time": "5:21:07", "remaining_time": "4:43:41"} +{"current_steps": 1039, "total_steps": 1955, "loss": 1.0585, "lr": 1.0651432247993793e-05, "epoch": 2.657289002557545, "percentage": 53.15, "elapsed_time": "5:21:26", "remaining_time": "4:43:23"} +{"current_steps": 1040, "total_steps": 1955, "loss": 1.0691, "lr": 1.0633609048659706e-05, "epoch": 2.659846547314578, "percentage": 53.2, "elapsed_time": "5:21:45", "remaining_time": "4:43:04"} +{"current_steps": 1041, "total_steps": 1955, "loss": 1.0482, "lr": 1.0615783828218546e-05, "epoch": 2.662404092071611, "percentage": 53.25, "elapsed_time": "5:22:03", "remaining_time": "4:42:46"} +{"current_steps": 1042, "total_steps": 1955, "loss": 1.0795, "lr": 1.0597956643529789e-05, "epoch": 2.6649616368286444, "percentage": 53.3, "elapsed_time": "5:22:22", "remaining_time": "4:42:27"} +{"current_steps": 1043, "total_steps": 1955, "loss": 1.0123, "lr": 1.0580127551459178e-05, "epoch": 2.6675191815856776, "percentage": 53.35, "elapsed_time": "5:22:41", "remaining_time": "4:42:09"} +{"current_steps": 1044, "total_steps": 1955, "loss": 1.0545, "lr": 1.0562296608878546e-05, "epoch": 2.670076726342711, "percentage": 53.4, "elapsed_time": "5:22:59", "remaining_time": "4:41:50"} +{"current_steps": 1045, "total_steps": 1955, "loss": 1.0412, "lr": 1.0544463872665612e-05, "epoch": 2.6726342710997444, "percentage": 53.45, "elapsed_time": "5:23:17", "remaining_time": "4:41:31"} +{"current_steps": 1046, "total_steps": 1955, "loss": 1.0252, "lr": 1.0526629399703834e-05, "epoch": 2.6751918158567776, "percentage": 53.5, "elapsed_time": "5:23:35", "remaining_time": "4:41:12"} +{"current_steps": 1047, "total_steps": 1955, "loss": 1.0589, "lr": 1.0508793246882202e-05, "epoch": 2.6777493606138107, "percentage": 53.55, "elapsed_time": "5:23:53", "remaining_time": "4:40:53"} +{"current_steps": 1048, "total_steps": 1955, "loss": 1.0584, "lr": 1.0490955471095062e-05, "epoch": 2.680306905370844, "percentage": 53.61, "elapsed_time": "5:24:12", "remaining_time": "4:40:35"} +{"current_steps": 1049, "total_steps": 1955, "loss": 1.0329, "lr": 1.0473116129241944e-05, "epoch": 2.682864450127877, "percentage": 53.66, "elapsed_time": "5:24:30", "remaining_time": "4:40:16"} +{"current_steps": 1050, "total_steps": 1955, "loss": 1.06, "lr": 1.045527527822737e-05, "epoch": 2.6854219948849103, "percentage": 53.71, "elapsed_time": "5:24:49", "remaining_time": "4:39:57"} +{"current_steps": 1051, "total_steps": 1955, "loss": 1.0564, "lr": 1.0437432974960676e-05, "epoch": 2.687979539641944, "percentage": 53.76, "elapsed_time": "5:25:07", "remaining_time": "4:39:39"} +{"current_steps": 1052, "total_steps": 1955, "loss": 1.0475, "lr": 1.0419589276355826e-05, "epoch": 2.690537084398977, "percentage": 53.81, "elapsed_time": "5:25:26", "remaining_time": "4:39:20"} +{"current_steps": 1053, "total_steps": 1955, "loss": 1.0614, "lr": 1.0401744239331244e-05, "epoch": 2.6930946291560103, "percentage": 53.86, "elapsed_time": "5:25:44", "remaining_time": "4:39:02"} +{"current_steps": 1054, "total_steps": 1955, "loss": 1.0241, "lr": 1.0383897920809618e-05, "epoch": 2.6956521739130435, "percentage": 53.91, "elapsed_time": "5:26:03", "remaining_time": "4:38:43"} +{"current_steps": 1055, "total_steps": 1955, "loss": 1.0386, "lr": 1.0366050377717723e-05, "epoch": 2.6982097186700766, "percentage": 53.96, "elapsed_time": "5:26:22", "remaining_time": "4:38:25"} +{"current_steps": 1056, "total_steps": 1955, "loss": 1.0587, "lr": 1.0348201666986241e-05, "epoch": 2.70076726342711, "percentage": 54.02, "elapsed_time": "5:26:40", "remaining_time": "4:38:06"} +{"current_steps": 1057, "total_steps": 1955, "loss": 1.0586, "lr": 1.0330351845549577e-05, "epoch": 2.703324808184143, "percentage": 54.07, "elapsed_time": "5:26:59", "remaining_time": "4:37:48"} +{"current_steps": 1058, "total_steps": 1955, "loss": 1.0476, "lr": 1.031250097034569e-05, "epoch": 2.7058823529411766, "percentage": 54.12, "elapsed_time": "5:27:17", "remaining_time": "4:37:29"} +{"current_steps": 1059, "total_steps": 1955, "loss": 1.0758, "lr": 1.029464909831588e-05, "epoch": 2.70843989769821, "percentage": 54.17, "elapsed_time": "5:27:36", "remaining_time": "4:37:10"} +{"current_steps": 1060, "total_steps": 1955, "loss": 1.0346, "lr": 1.0276796286404644e-05, "epoch": 2.710997442455243, "percentage": 54.22, "elapsed_time": "5:27:54", "remaining_time": "4:36:52"} +{"current_steps": 1061, "total_steps": 1955, "loss": 1.0741, "lr": 1.0258942591559475e-05, "epoch": 2.713554987212276, "percentage": 54.27, "elapsed_time": "5:28:13", "remaining_time": "4:36:33"} +{"current_steps": 1062, "total_steps": 1955, "loss": 1.062, "lr": 1.024108807073067e-05, "epoch": 2.7161125319693094, "percentage": 54.32, "elapsed_time": "5:28:32", "remaining_time": "4:36:15"} +{"current_steps": 1063, "total_steps": 1955, "loss": 1.0958, "lr": 1.0223232780871173e-05, "epoch": 2.718670076726343, "percentage": 54.37, "elapsed_time": "5:28:50", "remaining_time": "4:35:56"} +{"current_steps": 1064, "total_steps": 1955, "loss": 1.0572, "lr": 1.020537677893638e-05, "epoch": 2.7212276214833757, "percentage": 54.42, "elapsed_time": "5:29:08", "remaining_time": "4:35:37"} +{"current_steps": 1065, "total_steps": 1955, "loss": 1.0266, "lr": 1.0187520121883952e-05, "epoch": 2.7237851662404093, "percentage": 54.48, "elapsed_time": "5:29:27", "remaining_time": "4:35:19"} +{"current_steps": 1066, "total_steps": 1955, "loss": 1.0456, "lr": 1.0169662866673647e-05, "epoch": 2.7263427109974425, "percentage": 54.53, "elapsed_time": "5:29:45", "remaining_time": "4:35:00"} +{"current_steps": 1067, "total_steps": 1955, "loss": 1.0525, "lr": 1.0151805070267123e-05, "epoch": 2.7289002557544757, "percentage": 54.58, "elapsed_time": "5:30:04", "remaining_time": "4:34:41"} +{"current_steps": 1068, "total_steps": 1955, "loss": 1.0714, "lr": 1.0133946789627773e-05, "epoch": 2.731457800511509, "percentage": 54.63, "elapsed_time": "5:30:22", "remaining_time": "4:34:23"} +{"current_steps": 1069, "total_steps": 1955, "loss": 1.0476, "lr": 1.0116088081720528e-05, "epoch": 2.734015345268542, "percentage": 54.68, "elapsed_time": "5:30:41", "remaining_time": "4:34:04"} +{"current_steps": 1070, "total_steps": 1955, "loss": 1.0935, "lr": 1.0098229003511684e-05, "epoch": 2.7365728900255757, "percentage": 54.73, "elapsed_time": "5:30:59", "remaining_time": "4:33:45"} +{"current_steps": 1071, "total_steps": 1955, "loss": 1.042, "lr": 1.0080369611968724e-05, "epoch": 2.7391304347826084, "percentage": 54.78, "elapsed_time": "5:31:18", "remaining_time": "4:33:27"} +{"current_steps": 1072, "total_steps": 1955, "loss": 1.0744, "lr": 1.0062509964060119e-05, "epoch": 2.741687979539642, "percentage": 54.83, "elapsed_time": "5:31:36", "remaining_time": "4:33:08"} +{"current_steps": 1073, "total_steps": 1955, "loss": 1.056, "lr": 1.0044650116755166e-05, "epoch": 2.7442455242966752, "percentage": 54.88, "elapsed_time": "5:31:55", "remaining_time": "4:32:50"} +{"current_steps": 1074, "total_steps": 1955, "loss": 1.0406, "lr": 1.0026790127023793e-05, "epoch": 2.7468030690537084, "percentage": 54.94, "elapsed_time": "5:32:13", "remaining_time": "4:32:31"} +{"current_steps": 1075, "total_steps": 1955, "loss": 1.0957, "lr": 1.0008930051836389e-05, "epoch": 2.7493606138107416, "percentage": 54.99, "elapsed_time": "5:32:32", "remaining_time": "4:32:13"} +{"current_steps": 1076, "total_steps": 1955, "loss": 1.0509, "lr": 9.991069948163616e-06, "epoch": 2.7519181585677748, "percentage": 55.04, "elapsed_time": "5:32:50", "remaining_time": "4:31:54"} +{"current_steps": 1077, "total_steps": 1955, "loss": 1.0727, "lr": 9.97320987297621e-06, "epoch": 2.7544757033248084, "percentage": 55.09, "elapsed_time": "5:33:09", "remaining_time": "4:31:35"} +{"current_steps": 1078, "total_steps": 1955, "loss": 1.0343, "lr": 9.955349883244837e-06, "epoch": 2.7570332480818416, "percentage": 55.14, "elapsed_time": "5:33:26", "remaining_time": "4:31:16"} +{"current_steps": 1079, "total_steps": 1955, "loss": 1.0119, "lr": 9.937490035939884e-06, "epoch": 2.7595907928388748, "percentage": 55.19, "elapsed_time": "5:33:45", "remaining_time": "4:30:57"} +{"current_steps": 1080, "total_steps": 1955, "loss": 1.0598, "lr": 9.919630388031278e-06, "epoch": 2.762148337595908, "percentage": 55.24, "elapsed_time": "5:34:03", "remaining_time": "4:30:38"} +{"current_steps": 1081, "total_steps": 1955, "loss": 1.0224, "lr": 9.901770996488316e-06, "epoch": 2.764705882352941, "percentage": 55.29, "elapsed_time": "5:34:21", "remaining_time": "4:30:20"} +{"current_steps": 1082, "total_steps": 1955, "loss": 1.0609, "lr": 9.883911918279476e-06, "epoch": 2.7672634271099743, "percentage": 55.35, "elapsed_time": "5:34:40", "remaining_time": "4:30:01"} +{"current_steps": 1083, "total_steps": 1955, "loss": 1.0578, "lr": 9.86605321037223e-06, "epoch": 2.7698209718670075, "percentage": 55.4, "elapsed_time": "5:34:58", "remaining_time": "4:29:43"} +{"current_steps": 1084, "total_steps": 1955, "loss": 1.0606, "lr": 9.84819492973288e-06, "epoch": 2.772378516624041, "percentage": 55.45, "elapsed_time": "5:35:17", "remaining_time": "4:29:24"} +{"current_steps": 1085, "total_steps": 1955, "loss": 1.0351, "lr": 9.830337133326355e-06, "epoch": 2.7749360613810743, "percentage": 55.5, "elapsed_time": "5:35:36", "remaining_time": "4:29:06"} +{"current_steps": 1086, "total_steps": 1955, "loss": 1.0897, "lr": 9.81247987811605e-06, "epoch": 2.7774936061381075, "percentage": 55.55, "elapsed_time": "5:35:54", "remaining_time": "4:28:47"} +{"current_steps": 1087, "total_steps": 1955, "loss": 1.0146, "lr": 9.794623221063625e-06, "epoch": 2.7800511508951407, "percentage": 55.6, "elapsed_time": "5:36:13", "remaining_time": "4:28:29"} +{"current_steps": 1088, "total_steps": 1955, "loss": 1.042, "lr": 9.776767219128829e-06, "epoch": 2.782608695652174, "percentage": 55.65, "elapsed_time": "5:36:31", "remaining_time": "4:28:10"} +{"current_steps": 1089, "total_steps": 1955, "loss": 1.0633, "lr": 9.758911929269334e-06, "epoch": 2.785166240409207, "percentage": 55.7, "elapsed_time": "5:36:50", "remaining_time": "4:27:51"} +{"current_steps": 1090, "total_steps": 1955, "loss": 1.0285, "lr": 9.741057408440527e-06, "epoch": 2.78772378516624, "percentage": 55.75, "elapsed_time": "5:37:08", "remaining_time": "4:27:32"} +{"current_steps": 1091, "total_steps": 1955, "loss": 1.004, "lr": 9.723203713595356e-06, "epoch": 2.790281329923274, "percentage": 55.81, "elapsed_time": "5:37:26", "remaining_time": "4:27:13"} +{"current_steps": 1092, "total_steps": 1955, "loss": 1.0651, "lr": 9.70535090168412e-06, "epoch": 2.792838874680307, "percentage": 55.86, "elapsed_time": "5:37:44", "remaining_time": "4:26:54"} +{"current_steps": 1093, "total_steps": 1955, "loss": 1.068, "lr": 9.687499029654314e-06, "epoch": 2.79539641943734, "percentage": 55.91, "elapsed_time": "5:38:02", "remaining_time": "4:26:36"} +{"current_steps": 1094, "total_steps": 1955, "loss": 1.0499, "lr": 9.669648154450425e-06, "epoch": 2.7979539641943734, "percentage": 55.96, "elapsed_time": "5:38:21", "remaining_time": "4:26:17"} +{"current_steps": 1095, "total_steps": 1955, "loss": 1.0455, "lr": 9.651798333013762e-06, "epoch": 2.8005115089514065, "percentage": 56.01, "elapsed_time": "5:38:39", "remaining_time": "4:25:59"} +{"current_steps": 1096, "total_steps": 1955, "loss": 1.0956, "lr": 9.63394962228228e-06, "epoch": 2.80306905370844, "percentage": 56.06, "elapsed_time": "5:38:57", "remaining_time": "4:25:39"} +{"current_steps": 1097, "total_steps": 1955, "loss": 1.062, "lr": 9.616102079190382e-06, "epoch": 2.805626598465473, "percentage": 56.11, "elapsed_time": "5:39:16", "remaining_time": "4:25:21"} +{"current_steps": 1098, "total_steps": 1955, "loss": 1.0769, "lr": 9.59825576066876e-06, "epoch": 2.8081841432225065, "percentage": 56.16, "elapsed_time": "5:39:34", "remaining_time": "4:25:02"} +{"current_steps": 1099, "total_steps": 1955, "loss": 1.0665, "lr": 9.580410723644177e-06, "epoch": 2.8107416879795397, "percentage": 56.21, "elapsed_time": "5:39:53", "remaining_time": "4:24:44"} +{"current_steps": 1100, "total_steps": 1955, "loss": 1.0403, "lr": 9.562567025039326e-06, "epoch": 2.813299232736573, "percentage": 56.27, "elapsed_time": "5:40:11", "remaining_time": "4:24:25"} +{"current_steps": 1101, "total_steps": 1955, "loss": 1.0679, "lr": 9.544724721772632e-06, "epoch": 2.815856777493606, "percentage": 56.32, "elapsed_time": "5:40:29", "remaining_time": "4:24:06"} +{"current_steps": 1102, "total_steps": 1955, "loss": 1.0499, "lr": 9.526883870758056e-06, "epoch": 2.8184143222506393, "percentage": 56.37, "elapsed_time": "5:40:49", "remaining_time": "4:23:48"} +{"current_steps": 1103, "total_steps": 1955, "loss": 1.0652, "lr": 9.50904452890494e-06, "epoch": 2.820971867007673, "percentage": 56.42, "elapsed_time": "5:41:07", "remaining_time": "4:23:30"} +{"current_steps": 1104, "total_steps": 1955, "loss": 1.0413, "lr": 9.491206753117803e-06, "epoch": 2.8235294117647056, "percentage": 56.47, "elapsed_time": "5:41:25", "remaining_time": "4:23:11"} +{"current_steps": 1105, "total_steps": 1955, "loss": 1.0316, "lr": 9.47337060029617e-06, "epoch": 2.8260869565217392, "percentage": 56.52, "elapsed_time": "5:41:43", "remaining_time": "4:22:52"} +{"current_steps": 1106, "total_steps": 1955, "loss": 1.0618, "lr": 9.45553612733439e-06, "epoch": 2.8286445012787724, "percentage": 56.57, "elapsed_time": "5:42:02", "remaining_time": "4:22:33"} +{"current_steps": 1107, "total_steps": 1955, "loss": 1.0577, "lr": 9.437703391121457e-06, "epoch": 2.8312020460358056, "percentage": 56.62, "elapsed_time": "5:42:21", "remaining_time": "4:22:15"} +{"current_steps": 1108, "total_steps": 1955, "loss": 1.069, "lr": 9.419872448540821e-06, "epoch": 2.833759590792839, "percentage": 56.68, "elapsed_time": "5:42:39", "remaining_time": "4:21:56"} +{"current_steps": 1109, "total_steps": 1955, "loss": 1.073, "lr": 9.402043356470214e-06, "epoch": 2.836317135549872, "percentage": 56.73, "elapsed_time": "5:42:58", "remaining_time": "4:21:38"} +{"current_steps": 1110, "total_steps": 1955, "loss": 1.0535, "lr": 9.384216171781458e-06, "epoch": 2.8388746803069056, "percentage": 56.78, "elapsed_time": "5:43:16", "remaining_time": "4:21:19"} +{"current_steps": 1111, "total_steps": 1955, "loss": 1.053, "lr": 9.366390951340298e-06, "epoch": 2.8414322250639388, "percentage": 56.83, "elapsed_time": "5:43:34", "remaining_time": "4:21:00"} +{"current_steps": 1112, "total_steps": 1955, "loss": 1.0618, "lr": 9.348567752006208e-06, "epoch": 2.843989769820972, "percentage": 56.88, "elapsed_time": "5:43:53", "remaining_time": "4:20:42"} +{"current_steps": 1113, "total_steps": 1955, "loss": 1.0511, "lr": 9.330746630632224e-06, "epoch": 2.846547314578005, "percentage": 56.93, "elapsed_time": "5:44:12", "remaining_time": "4:20:23"} +{"current_steps": 1114, "total_steps": 1955, "loss": 1.0518, "lr": 9.312927644064741e-06, "epoch": 2.8491048593350383, "percentage": 56.98, "elapsed_time": "5:44:30", "remaining_time": "4:20:05"} +{"current_steps": 1115, "total_steps": 1955, "loss": 1.0568, "lr": 9.295110849143361e-06, "epoch": 2.8516624040920715, "percentage": 57.03, "elapsed_time": "5:44:48", "remaining_time": "4:19:46"} +{"current_steps": 1116, "total_steps": 1955, "loss": 1.0484, "lr": 9.277296302700675e-06, "epoch": 2.8542199488491047, "percentage": 57.08, "elapsed_time": "5:45:07", "remaining_time": "4:19:27"} +{"current_steps": 1117, "total_steps": 1955, "loss": 1.0334, "lr": 9.259484061562114e-06, "epoch": 2.8567774936061383, "percentage": 57.14, "elapsed_time": "5:45:25", "remaining_time": "4:19:09"} +{"current_steps": 1118, "total_steps": 1955, "loss": 1.0649, "lr": 9.241674182545755e-06, "epoch": 2.8593350383631715, "percentage": 57.19, "elapsed_time": "5:45:44", "remaining_time": "4:18:50"} +{"current_steps": 1119, "total_steps": 1955, "loss": 1.0379, "lr": 9.223866722462134e-06, "epoch": 2.8618925831202047, "percentage": 57.24, "elapsed_time": "5:46:02", "remaining_time": "4:18:31"} +{"current_steps": 1120, "total_steps": 1955, "loss": 1.0941, "lr": 9.206061738114087e-06, "epoch": 2.864450127877238, "percentage": 57.29, "elapsed_time": "5:46:20", "remaining_time": "4:18:12"} +{"current_steps": 1121, "total_steps": 1955, "loss": 1.062, "lr": 9.188259286296527e-06, "epoch": 2.867007672634271, "percentage": 57.34, "elapsed_time": "5:46:38", "remaining_time": "4:17:53"} +{"current_steps": 1122, "total_steps": 1955, "loss": 1.0418, "lr": 9.170459423796309e-06, "epoch": 2.869565217391304, "percentage": 57.39, "elapsed_time": "5:46:57", "remaining_time": "4:17:35"} +{"current_steps": 1123, "total_steps": 1955, "loss": 1.0117, "lr": 9.152662207392025e-06, "epoch": 2.8721227621483374, "percentage": 57.44, "elapsed_time": "5:47:16", "remaining_time": "4:17:16"} +{"current_steps": 1124, "total_steps": 1955, "loss": 1.0953, "lr": 9.134867693853818e-06, "epoch": 2.874680306905371, "percentage": 57.49, "elapsed_time": "5:47:34", "remaining_time": "4:16:58"} +{"current_steps": 1125, "total_steps": 1955, "loss": 1.0714, "lr": 9.11707593994322e-06, "epoch": 2.877237851662404, "percentage": 57.54, "elapsed_time": "5:47:52", "remaining_time": "4:16:39"} +{"current_steps": 1126, "total_steps": 1955, "loss": 1.0621, "lr": 9.099287002412957e-06, "epoch": 2.8797953964194374, "percentage": 57.6, "elapsed_time": "5:48:11", "remaining_time": "4:16:20"} +{"current_steps": 1127, "total_steps": 1955, "loss": 1.0292, "lr": 9.08150093800676e-06, "epoch": 2.8823529411764706, "percentage": 57.65, "elapsed_time": "5:48:29", "remaining_time": "4:16:02"} +{"current_steps": 1128, "total_steps": 1955, "loss": 1.0336, "lr": 9.063717803459215e-06, "epoch": 2.8849104859335037, "percentage": 57.7, "elapsed_time": "5:48:48", "remaining_time": "4:15:43"} +{"current_steps": 1129, "total_steps": 1955, "loss": 1.0408, "lr": 9.045937655495544e-06, "epoch": 2.887468030690537, "percentage": 57.75, "elapsed_time": "5:49:06", "remaining_time": "4:15:24"} +{"current_steps": 1130, "total_steps": 1955, "loss": 1.0667, "lr": 9.028160550831458e-06, "epoch": 2.89002557544757, "percentage": 57.8, "elapsed_time": "5:49:25", "remaining_time": "4:15:06"} +{"current_steps": 1131, "total_steps": 1955, "loss": 1.0792, "lr": 9.01038654617295e-06, "epoch": 2.8925831202046037, "percentage": 57.85, "elapsed_time": "5:49:43", "remaining_time": "4:14:47"} +{"current_steps": 1132, "total_steps": 1955, "loss": 1.015, "lr": 8.992615698216121e-06, "epoch": 2.895140664961637, "percentage": 57.9, "elapsed_time": "5:50:01", "remaining_time": "4:14:28"} +{"current_steps": 1133, "total_steps": 1955, "loss": 1.0342, "lr": 8.974848063647017e-06, "epoch": 2.89769820971867, "percentage": 57.95, "elapsed_time": "5:50:19", "remaining_time": "4:14:09"} +{"current_steps": 1134, "total_steps": 1955, "loss": 1.0361, "lr": 8.95708369914142e-06, "epoch": 2.9002557544757033, "percentage": 58.01, "elapsed_time": "5:50:38", "remaining_time": "4:13:51"} +{"current_steps": 1135, "total_steps": 1955, "loss": 1.0608, "lr": 8.93932266136469e-06, "epoch": 2.9028132992327365, "percentage": 58.06, "elapsed_time": "5:50:56", "remaining_time": "4:13:32"} +{"current_steps": 1136, "total_steps": 1955, "loss": 1.0324, "lr": 8.92156500697157e-06, "epoch": 2.90537084398977, "percentage": 58.11, "elapsed_time": "5:51:15", "remaining_time": "4:13:14"} +{"current_steps": 1137, "total_steps": 1955, "loss": 1.0974, "lr": 8.903810792606019e-06, "epoch": 2.907928388746803, "percentage": 58.16, "elapsed_time": "5:51:33", "remaining_time": "4:12:55"} +{"current_steps": 1138, "total_steps": 1955, "loss": 1.0339, "lr": 8.886060074901006e-06, "epoch": 2.9104859335038364, "percentage": 58.21, "elapsed_time": "5:51:52", "remaining_time": "4:12:37"} +{"current_steps": 1139, "total_steps": 1955, "loss": 1.0478, "lr": 8.868312910478362e-06, "epoch": 2.9130434782608696, "percentage": 58.26, "elapsed_time": "5:52:11", "remaining_time": "4:12:18"} +{"current_steps": 1140, "total_steps": 1955, "loss": 1.0411, "lr": 8.850569355948579e-06, "epoch": 2.915601023017903, "percentage": 58.31, "elapsed_time": "5:52:30", "remaining_time": "4:12:00"} +{"current_steps": 1141, "total_steps": 1955, "loss": 1.0576, "lr": 8.832829467910631e-06, "epoch": 2.918158567774936, "percentage": 58.36, "elapsed_time": "5:52:48", "remaining_time": "4:11:42"} +{"current_steps": 1142, "total_steps": 1955, "loss": 1.0676, "lr": 8.815093302951805e-06, "epoch": 2.920716112531969, "percentage": 58.41, "elapsed_time": "5:53:07", "remaining_time": "4:11:23"} +{"current_steps": 1143, "total_steps": 1955, "loss": 1.0635, "lr": 8.797360917647498e-06, "epoch": 2.923273657289003, "percentage": 58.47, "elapsed_time": "5:53:26", "remaining_time": "4:11:05"} +{"current_steps": 1144, "total_steps": 1955, "loss": 1.0418, "lr": 8.77963236856106e-06, "epoch": 2.9258312020460355, "percentage": 58.52, "elapsed_time": "5:53:44", "remaining_time": "4:10:46"} +{"current_steps": 1145, "total_steps": 1955, "loss": 1.0491, "lr": 8.761907712243606e-06, "epoch": 2.928388746803069, "percentage": 58.57, "elapsed_time": "5:54:02", "remaining_time": "4:10:27"} +{"current_steps": 1146, "total_steps": 1955, "loss": 1.0361, "lr": 8.744187005233826e-06, "epoch": 2.9309462915601023, "percentage": 58.62, "elapsed_time": "5:54:21", "remaining_time": "4:10:09"} +{"current_steps": 1147, "total_steps": 1955, "loss": 1.0541, "lr": 8.726470304057821e-06, "epoch": 2.9335038363171355, "percentage": 58.67, "elapsed_time": "5:54:39", "remaining_time": "4:09:50"} +{"current_steps": 1148, "total_steps": 1955, "loss": 1.0523, "lr": 8.708757665228909e-06, "epoch": 2.9360613810741687, "percentage": 58.72, "elapsed_time": "5:54:58", "remaining_time": "4:09:31"} +{"current_steps": 1149, "total_steps": 1955, "loss": 1.0681, "lr": 8.691049145247446e-06, "epoch": 2.938618925831202, "percentage": 58.77, "elapsed_time": "5:55:16", "remaining_time": "4:09:12"} +{"current_steps": 1150, "total_steps": 1955, "loss": 1.0573, "lr": 8.673344800600657e-06, "epoch": 2.9411764705882355, "percentage": 58.82, "elapsed_time": "5:55:35", "remaining_time": "4:08:54"} +{"current_steps": 1151, "total_steps": 1955, "loss": 1.0722, "lr": 8.655644687762443e-06, "epoch": 2.9437340153452687, "percentage": 58.87, "elapsed_time": "5:55:53", "remaining_time": "4:08:36"} +{"current_steps": 1152, "total_steps": 1955, "loss": 1.0431, "lr": 8.637948863193215e-06, "epoch": 2.946291560102302, "percentage": 58.93, "elapsed_time": "5:56:11", "remaining_time": "4:08:17"} +{"current_steps": 1153, "total_steps": 1955, "loss": 1.0658, "lr": 8.620257383339694e-06, "epoch": 2.948849104859335, "percentage": 58.98, "elapsed_time": "5:56:30", "remaining_time": "4:07:58"} +{"current_steps": 1154, "total_steps": 1955, "loss": 1.0701, "lr": 8.602570304634745e-06, "epoch": 2.9514066496163682, "percentage": 59.03, "elapsed_time": "5:56:48", "remaining_time": "4:07:39"} +{"current_steps": 1155, "total_steps": 1955, "loss": 1.086, "lr": 8.5848876834972e-06, "epoch": 2.9539641943734014, "percentage": 59.08, "elapsed_time": "5:57:06", "remaining_time": "4:07:20"} +{"current_steps": 1156, "total_steps": 1955, "loss": 1.059, "lr": 8.567209576331662e-06, "epoch": 2.9565217391304346, "percentage": 59.13, "elapsed_time": "5:57:25", "remaining_time": "4:07:02"} +{"current_steps": 1157, "total_steps": 1955, "loss": 1.0505, "lr": 8.549536039528351e-06, "epoch": 2.959079283887468, "percentage": 59.18, "elapsed_time": "5:57:43", "remaining_time": "4:06:43"} +{"current_steps": 1158, "total_steps": 1955, "loss": 1.0448, "lr": 8.53186712946289e-06, "epoch": 2.9616368286445014, "percentage": 59.23, "elapsed_time": "5:58:02", "remaining_time": "4:06:25"} +{"current_steps": 1159, "total_steps": 1955, "loss": 1.045, "lr": 8.514202902496157e-06, "epoch": 2.9641943734015346, "percentage": 59.28, "elapsed_time": "5:58:21", "remaining_time": "4:06:07"} +{"current_steps": 1160, "total_steps": 1955, "loss": 1.0672, "lr": 8.496543414974084e-06, "epoch": 2.9667519181585678, "percentage": 59.34, "elapsed_time": "5:58:40", "remaining_time": "4:05:48"} +{"current_steps": 1161, "total_steps": 1955, "loss": 1.0364, "lr": 8.478888723227486e-06, "epoch": 2.969309462915601, "percentage": 59.39, "elapsed_time": "5:58:58", "remaining_time": "4:05:30"} +{"current_steps": 1162, "total_steps": 1955, "loss": 1.0413, "lr": 8.461238883571885e-06, "epoch": 2.971867007672634, "percentage": 59.44, "elapsed_time": "5:59:17", "remaining_time": "4:05:11"} +{"current_steps": 1163, "total_steps": 1955, "loss": 1.0833, "lr": 8.443593952307319e-06, "epoch": 2.9744245524296673, "percentage": 59.49, "elapsed_time": "5:59:35", "remaining_time": "4:04:53"} +{"current_steps": 1164, "total_steps": 1955, "loss": 1.036, "lr": 8.425953985718182e-06, "epoch": 2.976982097186701, "percentage": 59.54, "elapsed_time": "5:59:54", "remaining_time": "4:04:34"} +{"current_steps": 1165, "total_steps": 1955, "loss": 1.0447, "lr": 8.408319040073011e-06, "epoch": 2.979539641943734, "percentage": 59.59, "elapsed_time": "6:00:13", "remaining_time": "4:04:16"} +{"current_steps": 1166, "total_steps": 1955, "loss": 1.0771, "lr": 8.390689171624341e-06, "epoch": 2.9820971867007673, "percentage": 59.64, "elapsed_time": "6:00:31", "remaining_time": "4:03:57"} +{"current_steps": 1167, "total_steps": 1955, "loss": 1.053, "lr": 8.373064436608512e-06, "epoch": 2.9846547314578005, "percentage": 59.69, "elapsed_time": "6:00:49", "remaining_time": "4:03:38"} +{"current_steps": 1168, "total_steps": 1955, "loss": 1.0684, "lr": 8.355444891245482e-06, "epoch": 2.9872122762148337, "percentage": 59.74, "elapsed_time": "6:01:08", "remaining_time": "4:03:20"} +{"current_steps": 1169, "total_steps": 1955, "loss": 1.0437, "lr": 8.337830591738664e-06, "epoch": 2.9897698209718673, "percentage": 59.8, "elapsed_time": "6:01:26", "remaining_time": "4:03:01"} +{"current_steps": 1170, "total_steps": 1955, "loss": 1.0422, "lr": 8.32022159427473e-06, "epoch": 2.9923273657289, "percentage": 59.85, "elapsed_time": "6:01:44", "remaining_time": "4:02:42"} +{"current_steps": 1171, "total_steps": 1955, "loss": 1.0383, "lr": 8.302617955023438e-06, "epoch": 2.9948849104859336, "percentage": 59.9, "elapsed_time": "6:02:03", "remaining_time": "4:02:24"} +{"current_steps": 1172, "total_steps": 1955, "loss": 1.0814, "lr": 8.285019730137462e-06, "epoch": 2.997442455242967, "percentage": 59.95, "elapsed_time": "6:02:22", "remaining_time": "4:02:05"} +{"current_steps": 1173, "total_steps": 1955, "loss": 1.0749, "lr": 8.2674269757522e-06, "epoch": 3.0, "percentage": 60.0, "elapsed_time": "6:02:41", "remaining_time": "4:01:47"} +{"current_steps": 1174, "total_steps": 1955, "loss": 1.0689, "lr": 8.2498397479856e-06, "epoch": 3.002557544757033, "percentage": 60.05, "elapsed_time": "6:03:09", "remaining_time": "4:01:35"} +{"current_steps": 1175, "total_steps": 1955, "loss": 1.0685, "lr": 8.232258102937987e-06, "epoch": 3.0051150895140664, "percentage": 60.1, "elapsed_time": "6:03:28", "remaining_time": "4:01:17"} +{"current_steps": 1176, "total_steps": 1955, "loss": 1.0474, "lr": 8.214682096691866e-06, "epoch": 3.0076726342710995, "percentage": 60.15, "elapsed_time": "6:03:47", "remaining_time": "4:00:58"} +{"current_steps": 1177, "total_steps": 1955, "loss": 1.0858, "lr": 8.197111785311767e-06, "epoch": 3.010230179028133, "percentage": 60.2, "elapsed_time": "6:04:05", "remaining_time": "4:00:40"} +{"current_steps": 1178, "total_steps": 1955, "loss": 1.0743, "lr": 8.179547224844046e-06, "epoch": 3.0127877237851663, "percentage": 60.26, "elapsed_time": "6:04:24", "remaining_time": "4:00:21"} +{"current_steps": 1179, "total_steps": 1955, "loss": 1.0228, "lr": 8.161988471316723e-06, "epoch": 3.0153452685421995, "percentage": 60.31, "elapsed_time": "6:04:43", "remaining_time": "4:00:03"} +{"current_steps": 1180, "total_steps": 1955, "loss": 1.0336, "lr": 8.144435580739285e-06, "epoch": 3.0179028132992327, "percentage": 60.36, "elapsed_time": "6:05:01", "remaining_time": "3:59:44"} +{"current_steps": 1181, "total_steps": 1955, "loss": 1.0455, "lr": 8.12688860910253e-06, "epoch": 3.020460358056266, "percentage": 60.41, "elapsed_time": "6:05:20", "remaining_time": "3:59:26"} +{"current_steps": 1182, "total_steps": 1955, "loss": 1.0395, "lr": 8.109347612378357e-06, "epoch": 3.023017902813299, "percentage": 60.46, "elapsed_time": "6:05:38", "remaining_time": "3:59:07"} +{"current_steps": 1183, "total_steps": 1955, "loss": 1.0623, "lr": 8.091812646519624e-06, "epoch": 3.0255754475703327, "percentage": 60.51, "elapsed_time": "6:05:57", "remaining_time": "3:58:49"} +{"current_steps": 1184, "total_steps": 1955, "loss": 1.0264, "lr": 8.074283767459941e-06, "epoch": 3.028132992327366, "percentage": 60.56, "elapsed_time": "6:06:16", "remaining_time": "3:58:30"} +{"current_steps": 1185, "total_steps": 1955, "loss": 1.0419, "lr": 8.056761031113506e-06, "epoch": 3.030690537084399, "percentage": 60.61, "elapsed_time": "6:06:35", "remaining_time": "3:58:12"} +{"current_steps": 1186, "total_steps": 1955, "loss": 1.0239, "lr": 8.03924449337493e-06, "epoch": 3.0332480818414322, "percentage": 60.66, "elapsed_time": "6:06:53", "remaining_time": "3:57:53"} +{"current_steps": 1187, "total_steps": 1955, "loss": 1.0482, "lr": 8.021734210119031e-06, "epoch": 3.0358056265984654, "percentage": 60.72, "elapsed_time": "6:07:11", "remaining_time": "3:57:34"} +{"current_steps": 1188, "total_steps": 1955, "loss": 1.0117, "lr": 8.004230237200695e-06, "epoch": 3.0383631713554986, "percentage": 60.77, "elapsed_time": "6:07:30", "remaining_time": "3:57:16"} +{"current_steps": 1189, "total_steps": 1955, "loss": 1.0469, "lr": 7.986732630454673e-06, "epoch": 3.040920716112532, "percentage": 60.82, "elapsed_time": "6:07:48", "remaining_time": "3:56:57"} +{"current_steps": 1190, "total_steps": 1955, "loss": 1.0341, "lr": 7.969241445695407e-06, "epoch": 3.0434782608695654, "percentage": 60.87, "elapsed_time": "6:08:07", "remaining_time": "3:56:39"} +{"current_steps": 1191, "total_steps": 1955, "loss": 1.0147, "lr": 7.951756738716861e-06, "epoch": 3.0460358056265986, "percentage": 60.92, "elapsed_time": "6:08:26", "remaining_time": "3:56:20"} +{"current_steps": 1192, "total_steps": 1955, "loss": 1.0535, "lr": 7.934278565292328e-06, "epoch": 3.0485933503836318, "percentage": 60.97, "elapsed_time": "6:08:44", "remaining_time": "3:56:02"} +{"current_steps": 1193, "total_steps": 1955, "loss": 1.0605, "lr": 7.916806981174258e-06, "epoch": 3.051150895140665, "percentage": 61.02, "elapsed_time": "6:09:02", "remaining_time": "3:55:43"} +{"current_steps": 1194, "total_steps": 1955, "loss": 1.0316, "lr": 7.899342042094094e-06, "epoch": 3.053708439897698, "percentage": 61.07, "elapsed_time": "6:09:21", "remaining_time": "3:55:24"} +{"current_steps": 1195, "total_steps": 1955, "loss": 1.0367, "lr": 7.881883803762071e-06, "epoch": 3.0562659846547313, "percentage": 61.13, "elapsed_time": "6:09:39", "remaining_time": "3:55:05"} +{"current_steps": 1196, "total_steps": 1955, "loss": 1.0504, "lr": 7.864432321867056e-06, "epoch": 3.0588235294117645, "percentage": 61.18, "elapsed_time": "6:09:58", "remaining_time": "3:54:47"} +{"current_steps": 1197, "total_steps": 1955, "loss": 1.0511, "lr": 7.846987652076372e-06, "epoch": 3.061381074168798, "percentage": 61.23, "elapsed_time": "6:10:17", "remaining_time": "3:54:28"} +{"current_steps": 1198, "total_steps": 1955, "loss": 1.0484, "lr": 7.829549850035585e-06, "epoch": 3.0639386189258313, "percentage": 61.28, "elapsed_time": "6:10:35", "remaining_time": "3:54:10"} +{"current_steps": 1199, "total_steps": 1955, "loss": 1.0661, "lr": 7.812118971368384e-06, "epoch": 3.0664961636828645, "percentage": 61.33, "elapsed_time": "6:10:54", "remaining_time": "3:53:51"} +{"current_steps": 1200, "total_steps": 1955, "loss": 1.0245, "lr": 7.794695071676355e-06, "epoch": 3.0690537084398977, "percentage": 61.38, "elapsed_time": "6:11:12", "remaining_time": "3:53:33"} +{"current_steps": 1201, "total_steps": 1955, "loss": 1.0368, "lr": 7.777278206538832e-06, "epoch": 3.071611253196931, "percentage": 61.43, "elapsed_time": "6:11:31", "remaining_time": "3:53:14"} +{"current_steps": 1202, "total_steps": 1955, "loss": 1.0603, "lr": 7.75986843151271e-06, "epoch": 3.074168797953964, "percentage": 61.48, "elapsed_time": "6:11:49", "remaining_time": "3:52:55"} +{"current_steps": 1203, "total_steps": 1955, "loss": 1.0666, "lr": 7.742465802132261e-06, "epoch": 3.0767263427109977, "percentage": 61.53, "elapsed_time": "6:12:08", "remaining_time": "3:52:37"} +{"current_steps": 1204, "total_steps": 1955, "loss": 1.0653, "lr": 7.725070373908967e-06, "epoch": 3.079283887468031, "percentage": 61.59, "elapsed_time": "6:12:26", "remaining_time": "3:52:18"} +{"current_steps": 1205, "total_steps": 1955, "loss": 1.0659, "lr": 7.707682202331338e-06, "epoch": 3.081841432225064, "percentage": 61.64, "elapsed_time": "6:12:45", "remaining_time": "3:52:00"} +{"current_steps": 1206, "total_steps": 1955, "loss": 1.0361, "lr": 7.690301342864738e-06, "epoch": 3.084398976982097, "percentage": 61.69, "elapsed_time": "6:13:04", "remaining_time": "3:51:42"} +{"current_steps": 1207, "total_steps": 1955, "loss": 1.0579, "lr": 7.672927850951213e-06, "epoch": 3.0869565217391304, "percentage": 61.74, "elapsed_time": "6:13:22", "remaining_time": "3:51:23"} +{"current_steps": 1208, "total_steps": 1955, "loss": 1.0867, "lr": 7.6555617820093e-06, "epoch": 3.0895140664961636, "percentage": 61.79, "elapsed_time": "6:13:42", "remaining_time": "3:51:05"} +{"current_steps": 1209, "total_steps": 1955, "loss": 1.0312, "lr": 7.638203191433848e-06, "epoch": 3.0920716112531967, "percentage": 61.84, "elapsed_time": "6:14:00", "remaining_time": "3:50:46"} +{"current_steps": 1210, "total_steps": 1955, "loss": 1.0553, "lr": 7.620852134595875e-06, "epoch": 3.0946291560102304, "percentage": 61.89, "elapsed_time": "6:14:19", "remaining_time": "3:50:28"} +{"current_steps": 1211, "total_steps": 1955, "loss": 1.0566, "lr": 7.60350866684235e-06, "epoch": 3.0971867007672635, "percentage": 61.94, "elapsed_time": "6:14:37", "remaining_time": "3:50:09"} +{"current_steps": 1212, "total_steps": 1955, "loss": 1.0194, "lr": 7.586172843496042e-06, "epoch": 3.0997442455242967, "percentage": 61.99, "elapsed_time": "6:14:55", "remaining_time": "3:49:50"} +{"current_steps": 1213, "total_steps": 1955, "loss": 1.0887, "lr": 7.568844719855328e-06, "epoch": 3.10230179028133, "percentage": 62.05, "elapsed_time": "6:15:13", "remaining_time": "3:49:31"} +{"current_steps": 1214, "total_steps": 1955, "loss": 1.0367, "lr": 7.5515243511940395e-06, "epoch": 3.104859335038363, "percentage": 62.1, "elapsed_time": "6:15:32", "remaining_time": "3:49:13"} +{"current_steps": 1215, "total_steps": 1955, "loss": 1.0784, "lr": 7.534211792761249e-06, "epoch": 3.1074168797953963, "percentage": 62.15, "elapsed_time": "6:15:50", "remaining_time": "3:48:54"} +{"current_steps": 1216, "total_steps": 1955, "loss": 1.0298, "lr": 7.516907099781131e-06, "epoch": 3.10997442455243, "percentage": 62.2, "elapsed_time": "6:16:09", "remaining_time": "3:48:36"} +{"current_steps": 1217, "total_steps": 1955, "loss": 1.0678, "lr": 7.49961032745277e-06, "epoch": 3.112531969309463, "percentage": 62.25, "elapsed_time": "6:16:28", "remaining_time": "3:48:17"} +{"current_steps": 1218, "total_steps": 1955, "loss": 1.0397, "lr": 7.4823215309499766e-06, "epoch": 3.1150895140664963, "percentage": 62.3, "elapsed_time": "6:16:46", "remaining_time": "3:47:59"} +{"current_steps": 1219, "total_steps": 1955, "loss": 1.058, "lr": 7.465040765421132e-06, "epoch": 3.1176470588235294, "percentage": 62.35, "elapsed_time": "6:17:04", "remaining_time": "3:47:40"} +{"current_steps": 1220, "total_steps": 1955, "loss": 1.0461, "lr": 7.447768085988987e-06, "epoch": 3.1202046035805626, "percentage": 62.4, "elapsed_time": "6:17:23", "remaining_time": "3:47:21"} +{"current_steps": 1221, "total_steps": 1955, "loss": 1.0425, "lr": 7.430503547750505e-06, "epoch": 3.122762148337596, "percentage": 62.46, "elapsed_time": "6:17:41", "remaining_time": "3:47:02"} +{"current_steps": 1222, "total_steps": 1955, "loss": 0.9899, "lr": 7.413247205776684e-06, "epoch": 3.125319693094629, "percentage": 62.51, "elapsed_time": "6:17:59", "remaining_time": "3:46:44"} +{"current_steps": 1223, "total_steps": 1955, "loss": 1.0532, "lr": 7.3959991151123685e-06, "epoch": 3.1278772378516626, "percentage": 62.56, "elapsed_time": "6:18:18", "remaining_time": "3:46:25"} +{"current_steps": 1224, "total_steps": 1955, "loss": 1.0212, "lr": 7.378759330776093e-06, "epoch": 3.130434782608696, "percentage": 62.61, "elapsed_time": "6:18:37", "remaining_time": "3:46:07"} +{"current_steps": 1225, "total_steps": 1955, "loss": 1.0279, "lr": 7.361527907759893e-06, "epoch": 3.132992327365729, "percentage": 62.66, "elapsed_time": "6:18:55", "remaining_time": "3:45:48"} +{"current_steps": 1226, "total_steps": 1955, "loss": 1.0474, "lr": 7.344304901029121e-06, "epoch": 3.135549872122762, "percentage": 62.71, "elapsed_time": "6:19:14", "remaining_time": "3:45:30"} +{"current_steps": 1227, "total_steps": 1955, "loss": 1.0457, "lr": 7.327090365522303e-06, "epoch": 3.1381074168797953, "percentage": 62.76, "elapsed_time": "6:19:32", "remaining_time": "3:45:11"} +{"current_steps": 1228, "total_steps": 1955, "loss": 1.0644, "lr": 7.309884356150925e-06, "epoch": 3.1406649616368285, "percentage": 62.81, "elapsed_time": "6:19:51", "remaining_time": "3:44:52"} +{"current_steps": 1229, "total_steps": 1955, "loss": 1.0704, "lr": 7.2926869277992884e-06, "epoch": 3.1432225063938617, "percentage": 62.86, "elapsed_time": "6:20:09", "remaining_time": "3:44:34"} +{"current_steps": 1230, "total_steps": 1955, "loss": 1.0518, "lr": 7.275498135324322e-06, "epoch": 3.1457800511508953, "percentage": 62.92, "elapsed_time": "6:20:28", "remaining_time": "3:44:15"} +{"current_steps": 1231, "total_steps": 1955, "loss": 1.0622, "lr": 7.258318033555394e-06, "epoch": 3.1483375959079285, "percentage": 62.97, "elapsed_time": "6:20:47", "remaining_time": "3:43:57"} +{"current_steps": 1232, "total_steps": 1955, "loss": 1.0389, "lr": 7.2411466772941684e-06, "epoch": 3.1508951406649617, "percentage": 63.02, "elapsed_time": "6:21:05", "remaining_time": "3:43:38"} +{"current_steps": 1233, "total_steps": 1955, "loss": 1.0802, "lr": 7.2239841213144e-06, "epoch": 3.153452685421995, "percentage": 63.07, "elapsed_time": "6:21:24", "remaining_time": "3:43:20"} +{"current_steps": 1234, "total_steps": 1955, "loss": 1.0616, "lr": 7.20683042036178e-06, "epoch": 3.156010230179028, "percentage": 63.12, "elapsed_time": "6:21:42", "remaining_time": "3:43:01"} +{"current_steps": 1235, "total_steps": 1955, "loss": 1.0518, "lr": 7.189685629153749e-06, "epoch": 3.1585677749360612, "percentage": 63.17, "elapsed_time": "6:22:00", "remaining_time": "3:42:42"} +{"current_steps": 1236, "total_steps": 1955, "loss": 1.055, "lr": 7.1725498023793305e-06, "epoch": 3.1611253196930944, "percentage": 63.22, "elapsed_time": "6:22:19", "remaining_time": "3:42:23"} +{"current_steps": 1237, "total_steps": 1955, "loss": 1.044, "lr": 7.155422994698948e-06, "epoch": 3.163682864450128, "percentage": 63.27, "elapsed_time": "6:22:37", "remaining_time": "3:42:05"} +{"current_steps": 1238, "total_steps": 1955, "loss": 1.0489, "lr": 7.138305260744256e-06, "epoch": 3.166240409207161, "percentage": 63.32, "elapsed_time": "6:22:56", "remaining_time": "3:41:47"} +{"current_steps": 1239, "total_steps": 1955, "loss": 1.0133, "lr": 7.121196655117975e-06, "epoch": 3.1687979539641944, "percentage": 63.38, "elapsed_time": "6:23:15", "remaining_time": "3:41:28"} +{"current_steps": 1240, "total_steps": 1955, "loss": 1.039, "lr": 7.104097232393692e-06, "epoch": 3.1713554987212276, "percentage": 63.43, "elapsed_time": "6:23:33", "remaining_time": "3:41:10"} +{"current_steps": 1241, "total_steps": 1955, "loss": 1.0275, "lr": 7.087007047115719e-06, "epoch": 3.1739130434782608, "percentage": 63.48, "elapsed_time": "6:23:52", "remaining_time": "3:40:51"} +{"current_steps": 1242, "total_steps": 1955, "loss": 1.0343, "lr": 7.069926153798889e-06, "epoch": 3.176470588235294, "percentage": 63.53, "elapsed_time": "6:24:11", "remaining_time": "3:40:32"} +{"current_steps": 1243, "total_steps": 1955, "loss": 1.0442, "lr": 7.052854606928396e-06, "epoch": 3.1790281329923276, "percentage": 63.58, "elapsed_time": "6:24:29", "remaining_time": "3:40:14"} +{"current_steps": 1244, "total_steps": 1955, "loss": 1.0272, "lr": 7.0357924609596305e-06, "epoch": 3.1815856777493607, "percentage": 63.63, "elapsed_time": "6:24:48", "remaining_time": "3:39:56"} +{"current_steps": 1245, "total_steps": 1955, "loss": 1.0039, "lr": 7.018739770317985e-06, "epoch": 3.184143222506394, "percentage": 63.68, "elapsed_time": "6:25:07", "remaining_time": "3:39:37"} +{"current_steps": 1246, "total_steps": 1955, "loss": 1.0569, "lr": 7.001696589398699e-06, "epoch": 3.186700767263427, "percentage": 63.73, "elapsed_time": "6:25:25", "remaining_time": "3:39:18"} +{"current_steps": 1247, "total_steps": 1955, "loss": 1.0636, "lr": 6.984662972566674e-06, "epoch": 3.1892583120204603, "percentage": 63.79, "elapsed_time": "6:25:43", "remaining_time": "3:39:00"} +{"current_steps": 1248, "total_steps": 1955, "loss": 1.0439, "lr": 6.967638974156299e-06, "epoch": 3.1918158567774935, "percentage": 63.84, "elapsed_time": "6:26:02", "remaining_time": "3:38:41"} +{"current_steps": 1249, "total_steps": 1955, "loss": 1.033, "lr": 6.950624648471289e-06, "epoch": 3.1943734015345266, "percentage": 63.89, "elapsed_time": "6:26:21", "remaining_time": "3:38:23"} +{"current_steps": 1250, "total_steps": 1955, "loss": 1.0437, "lr": 6.933620049784502e-06, "epoch": 3.1969309462915603, "percentage": 63.94, "elapsed_time": "6:26:39", "remaining_time": "3:38:04"} +{"current_steps": 1251, "total_steps": 1955, "loss": 1.0201, "lr": 6.9166252323377705e-06, "epoch": 3.1994884910485935, "percentage": 63.99, "elapsed_time": "6:26:58", "remaining_time": "3:37:46"} +{"current_steps": 1252, "total_steps": 1955, "loss": 1.0607, "lr": 6.899640250341726e-06, "epoch": 3.2020460358056266, "percentage": 64.04, "elapsed_time": "6:27:16", "remaining_time": "3:37:27"} +{"current_steps": 1253, "total_steps": 1955, "loss": 1.0151, "lr": 6.882665157975617e-06, "epoch": 3.20460358056266, "percentage": 64.09, "elapsed_time": "6:27:35", "remaining_time": "3:37:09"} +{"current_steps": 1254, "total_steps": 1955, "loss": 1.0398, "lr": 6.865700009387161e-06, "epoch": 3.207161125319693, "percentage": 64.14, "elapsed_time": "6:27:54", "remaining_time": "3:36:50"} +{"current_steps": 1255, "total_steps": 1955, "loss": 1.0411, "lr": 6.8487448586923445e-06, "epoch": 3.209718670076726, "percentage": 64.19, "elapsed_time": "6:28:12", "remaining_time": "3:36:31"} +{"current_steps": 1256, "total_steps": 1955, "loss": 1.0312, "lr": 6.831799759975273e-06, "epoch": 3.21227621483376, "percentage": 64.25, "elapsed_time": "6:28:30", "remaining_time": "3:36:13"} +{"current_steps": 1257, "total_steps": 1955, "loss": 1.0364, "lr": 6.814864767287977e-06, "epoch": 3.214833759590793, "percentage": 64.3, "elapsed_time": "6:28:48", "remaining_time": "3:35:54"} +{"current_steps": 1258, "total_steps": 1955, "loss": 1.0324, "lr": 6.797939934650263e-06, "epoch": 3.217391304347826, "percentage": 64.35, "elapsed_time": "6:29:06", "remaining_time": "3:35:35"} +{"current_steps": 1259, "total_steps": 1955, "loss": 1.0794, "lr": 6.781025316049512e-06, "epoch": 3.2199488491048593, "percentage": 64.4, "elapsed_time": "6:29:25", "remaining_time": "3:35:16"} +{"current_steps": 1260, "total_steps": 1955, "loss": 1.0463, "lr": 6.7641209654405375e-06, "epoch": 3.2225063938618925, "percentage": 64.45, "elapsed_time": "6:29:44", "remaining_time": "3:34:58"} +{"current_steps": 1261, "total_steps": 1955, "loss": 1.0292, "lr": 6.747226936745395e-06, "epoch": 3.2250639386189257, "percentage": 64.5, "elapsed_time": "6:30:02", "remaining_time": "3:34:39"} +{"current_steps": 1262, "total_steps": 1955, "loss": 1.0104, "lr": 6.730343283853213e-06, "epoch": 3.227621483375959, "percentage": 64.55, "elapsed_time": "6:30:21", "remaining_time": "3:34:21"} +{"current_steps": 1263, "total_steps": 1955, "loss": 1.0333, "lr": 6.713470060620033e-06, "epoch": 3.2301790281329925, "percentage": 64.6, "elapsed_time": "6:30:40", "remaining_time": "3:34:02"} +{"current_steps": 1264, "total_steps": 1955, "loss": 1.0392, "lr": 6.6966073208686135e-06, "epoch": 3.2327365728900257, "percentage": 64.65, "elapsed_time": "6:30:58", "remaining_time": "3:33:44"} +{"current_steps": 1265, "total_steps": 1955, "loss": 1.0333, "lr": 6.679755118388275e-06, "epoch": 3.235294117647059, "percentage": 64.71, "elapsed_time": "6:31:16", "remaining_time": "3:33:25"} +{"current_steps": 1266, "total_steps": 1955, "loss": 1.0404, "lr": 6.662913506934737e-06, "epoch": 3.237851662404092, "percentage": 64.76, "elapsed_time": "6:31:35", "remaining_time": "3:33:07"} +{"current_steps": 1267, "total_steps": 1955, "loss": 1.0341, "lr": 6.646082540229923e-06, "epoch": 3.2404092071611252, "percentage": 64.81, "elapsed_time": "6:31:54", "remaining_time": "3:32:48"} +{"current_steps": 1268, "total_steps": 1955, "loss": 0.9955, "lr": 6.629262271961811e-06, "epoch": 3.2429667519181584, "percentage": 64.86, "elapsed_time": "6:32:12", "remaining_time": "3:32:30"} +{"current_steps": 1269, "total_steps": 1955, "loss": 1.0588, "lr": 6.61245275578425e-06, "epoch": 3.2455242966751916, "percentage": 64.91, "elapsed_time": "6:32:31", "remaining_time": "3:32:11"} +{"current_steps": 1270, "total_steps": 1955, "loss": 1.0342, "lr": 6.595654045316781e-06, "epoch": 3.2480818414322252, "percentage": 64.96, "elapsed_time": "6:32:50", "remaining_time": "3:31:53"} +{"current_steps": 1271, "total_steps": 1955, "loss": 1.0512, "lr": 6.5788661941444934e-06, "epoch": 3.2506393861892584, "percentage": 65.01, "elapsed_time": "6:33:08", "remaining_time": "3:31:34"} +{"current_steps": 1272, "total_steps": 1955, "loss": 1.0344, "lr": 6.562089255817823e-06, "epoch": 3.2531969309462916, "percentage": 65.06, "elapsed_time": "6:33:27", "remaining_time": "3:31:16"} +{"current_steps": 1273, "total_steps": 1955, "loss": 1.0343, "lr": 6.545323283852407e-06, "epoch": 3.2557544757033248, "percentage": 65.12, "elapsed_time": "6:33:46", "remaining_time": "3:30:57"} +{"current_steps": 1274, "total_steps": 1955, "loss": 1.0476, "lr": 6.528568331728895e-06, "epoch": 3.258312020460358, "percentage": 65.17, "elapsed_time": "6:34:04", "remaining_time": "3:30:38"} +{"current_steps": 1275, "total_steps": 1955, "loss": 1.0539, "lr": 6.511824452892791e-06, "epoch": 3.260869565217391, "percentage": 65.22, "elapsed_time": "6:34:22", "remaining_time": "3:30:20"} +{"current_steps": 1276, "total_steps": 1955, "loss": 1.009, "lr": 6.495091700754266e-06, "epoch": 3.2634271099744243, "percentage": 65.27, "elapsed_time": "6:34:41", "remaining_time": "3:30:01"} +{"current_steps": 1277, "total_steps": 1955, "loss": 1.0654, "lr": 6.478370128688005e-06, "epoch": 3.265984654731458, "percentage": 65.32, "elapsed_time": "6:35:00", "remaining_time": "3:29:43"} +{"current_steps": 1278, "total_steps": 1955, "loss": 1.0333, "lr": 6.461659790033038e-06, "epoch": 3.268542199488491, "percentage": 65.37, "elapsed_time": "6:35:18", "remaining_time": "3:29:24"} +{"current_steps": 1279, "total_steps": 1955, "loss": 1.0848, "lr": 6.444960738092548e-06, "epoch": 3.2710997442455243, "percentage": 65.42, "elapsed_time": "6:35:37", "remaining_time": "3:29:06"} +{"current_steps": 1280, "total_steps": 1955, "loss": 1.0533, "lr": 6.428273026133731e-06, "epoch": 3.2736572890025575, "percentage": 65.47, "elapsed_time": "6:35:56", "remaining_time": "3:28:47"} +{"current_steps": 1281, "total_steps": 1955, "loss": 1.0621, "lr": 6.411596707387594e-06, "epoch": 3.2762148337595907, "percentage": 65.52, "elapsed_time": "6:36:14", "remaining_time": "3:28:29"} +{"current_steps": 1282, "total_steps": 1955, "loss": 1.0477, "lr": 6.39493183504881e-06, "epoch": 3.2787723785166243, "percentage": 65.58, "elapsed_time": "6:36:33", "remaining_time": "3:28:10"} +{"current_steps": 1283, "total_steps": 1955, "loss": 1.0468, "lr": 6.378278462275542e-06, "epoch": 3.2813299232736575, "percentage": 65.63, "elapsed_time": "6:36:52", "remaining_time": "3:27:52"} +{"current_steps": 1284, "total_steps": 1955, "loss": 1.0426, "lr": 6.361636642189269e-06, "epoch": 3.2838874680306906, "percentage": 65.68, "elapsed_time": "6:37:10", "remaining_time": "3:27:33"} +{"current_steps": 1285, "total_steps": 1955, "loss": 1.0453, "lr": 6.345006427874616e-06, "epoch": 3.286445012787724, "percentage": 65.73, "elapsed_time": "6:37:28", "remaining_time": "3:27:14"} +{"current_steps": 1286, "total_steps": 1955, "loss": 1.0424, "lr": 6.328387872379193e-06, "epoch": 3.289002557544757, "percentage": 65.78, "elapsed_time": "6:37:47", "remaining_time": "3:26:56"} +{"current_steps": 1287, "total_steps": 1955, "loss": 1.0337, "lr": 6.311781028713414e-06, "epoch": 3.29156010230179, "percentage": 65.83, "elapsed_time": "6:38:06", "remaining_time": "3:26:37"} +{"current_steps": 1288, "total_steps": 1955, "loss": 1.0185, "lr": 6.29518594985034e-06, "epoch": 3.2941176470588234, "percentage": 65.88, "elapsed_time": "6:38:24", "remaining_time": "3:26:19"} +{"current_steps": 1289, "total_steps": 1955, "loss": 1.019, "lr": 6.278602688725498e-06, "epoch": 3.296675191815857, "percentage": 65.93, "elapsed_time": "6:38:43", "remaining_time": "3:26:00"} +{"current_steps": 1290, "total_steps": 1955, "loss": 1.0163, "lr": 6.262031298236728e-06, "epoch": 3.29923273657289, "percentage": 65.98, "elapsed_time": "6:39:02", "remaining_time": "3:25:42"} +{"current_steps": 1291, "total_steps": 1955, "loss": 1.0294, "lr": 6.245471831243997e-06, "epoch": 3.3017902813299234, "percentage": 66.04, "elapsed_time": "6:39:20", "remaining_time": "3:25:23"} +{"current_steps": 1292, "total_steps": 1955, "loss": 1.0576, "lr": 6.228924340569234e-06, "epoch": 3.3043478260869565, "percentage": 66.09, "elapsed_time": "6:39:39", "remaining_time": "3:25:05"} +{"current_steps": 1293, "total_steps": 1955, "loss": 0.9984, "lr": 6.212388878996177e-06, "epoch": 3.3069053708439897, "percentage": 66.14, "elapsed_time": "6:39:58", "remaining_time": "3:24:46"} +{"current_steps": 1294, "total_steps": 1955, "loss": 1.0298, "lr": 6.195865499270186e-06, "epoch": 3.309462915601023, "percentage": 66.19, "elapsed_time": "6:40:16", "remaining_time": "3:24:28"} +{"current_steps": 1295, "total_steps": 1955, "loss": 1.0431, "lr": 6.179354254098085e-06, "epoch": 3.312020460358056, "percentage": 66.24, "elapsed_time": "6:40:35", "remaining_time": "3:24:09"} +{"current_steps": 1296, "total_steps": 1955, "loss": 1.0544, "lr": 6.1628551961479856e-06, "epoch": 3.3145780051150897, "percentage": 66.29, "elapsed_time": "6:40:53", "remaining_time": "3:23:50"} +{"current_steps": 1297, "total_steps": 1955, "loss": 1.0098, "lr": 6.146368378049134e-06, "epoch": 3.317135549872123, "percentage": 66.34, "elapsed_time": "6:41:11", "remaining_time": "3:23:31"} +{"current_steps": 1298, "total_steps": 1955, "loss": 1.0294, "lr": 6.129893852391722e-06, "epoch": 3.319693094629156, "percentage": 66.39, "elapsed_time": "6:41:29", "remaining_time": "3:23:13"} +{"current_steps": 1299, "total_steps": 1955, "loss": 1.0549, "lr": 6.113431671726735e-06, "epoch": 3.3222506393861893, "percentage": 66.45, "elapsed_time": "6:41:48", "remaining_time": "3:22:55"} +{"current_steps": 1300, "total_steps": 1955, "loss": 1.0772, "lr": 6.096981888565784e-06, "epoch": 3.3248081841432224, "percentage": 66.5, "elapsed_time": "6:42:07", "remaining_time": "3:22:36"} +{"current_steps": 1301, "total_steps": 1955, "loss": 1.0184, "lr": 6.080544555380927e-06, "epoch": 3.3273657289002556, "percentage": 66.55, "elapsed_time": "6:42:25", "remaining_time": "3:22:17"} +{"current_steps": 1302, "total_steps": 1955, "loss": 1.0503, "lr": 6.06411972460452e-06, "epoch": 3.329923273657289, "percentage": 66.6, "elapsed_time": "6:42:43", "remaining_time": "3:21:59"} +{"current_steps": 1303, "total_steps": 1955, "loss": 1.0294, "lr": 6.047707448629023e-06, "epoch": 3.3324808184143224, "percentage": 66.65, "elapsed_time": "6:43:01", "remaining_time": "3:21:40"} +{"current_steps": 1304, "total_steps": 1955, "loss": 1.045, "lr": 6.031307779806858e-06, "epoch": 3.3350383631713556, "percentage": 66.7, "elapsed_time": "6:43:19", "remaining_time": "3:21:21"} +{"current_steps": 1305, "total_steps": 1955, "loss": 1.0194, "lr": 6.014920770450232e-06, "epoch": 3.337595907928389, "percentage": 66.75, "elapsed_time": "6:43:38", "remaining_time": "3:21:02"} +{"current_steps": 1306, "total_steps": 1955, "loss": 1.0541, "lr": 5.998546472830965e-06, "epoch": 3.340153452685422, "percentage": 66.8, "elapsed_time": "6:43:57", "remaining_time": "3:20:44"} +{"current_steps": 1307, "total_steps": 1955, "loss": 1.007, "lr": 5.982184939180337e-06, "epoch": 3.342710997442455, "percentage": 66.85, "elapsed_time": "6:44:15", "remaining_time": "3:20:25"} +{"current_steps": 1308, "total_steps": 1955, "loss": 1.0498, "lr": 5.96583622168891e-06, "epoch": 3.3452685421994883, "percentage": 66.91, "elapsed_time": "6:44:34", "remaining_time": "3:20:07"} +{"current_steps": 1309, "total_steps": 1955, "loss": 1.0339, "lr": 5.949500372506354e-06, "epoch": 3.3478260869565215, "percentage": 66.96, "elapsed_time": "6:44:53", "remaining_time": "3:19:48"} +{"current_steps": 1310, "total_steps": 1955, "loss": 1.0739, "lr": 5.9331774437413095e-06, "epoch": 3.350383631713555, "percentage": 67.01, "elapsed_time": "6:45:11", "remaining_time": "3:19:30"} +{"current_steps": 1311, "total_steps": 1955, "loss": 1.0416, "lr": 5.916867487461186e-06, "epoch": 3.3529411764705883, "percentage": 67.06, "elapsed_time": "6:45:29", "remaining_time": "3:19:11"} +{"current_steps": 1312, "total_steps": 1955, "loss": 1.0338, "lr": 5.900570555692029e-06, "epoch": 3.3554987212276215, "percentage": 67.11, "elapsed_time": "6:45:48", "remaining_time": "3:18:52"} +{"current_steps": 1313, "total_steps": 1955, "loss": 1.0369, "lr": 5.884286700418328e-06, "epoch": 3.3580562659846547, "percentage": 67.16, "elapsed_time": "6:46:06", "remaining_time": "3:18:34"} +{"current_steps": 1314, "total_steps": 1955, "loss": 1.0474, "lr": 5.868015973582856e-06, "epoch": 3.360613810741688, "percentage": 67.21, "elapsed_time": "6:46:25", "remaining_time": "3:18:15"} +{"current_steps": 1315, "total_steps": 1955, "loss": 1.0168, "lr": 5.85175842708652e-06, "epoch": 3.363171355498721, "percentage": 67.26, "elapsed_time": "6:46:43", "remaining_time": "3:17:57"} +{"current_steps": 1316, "total_steps": 1955, "loss": 1.0404, "lr": 5.835514112788178e-06, "epoch": 3.3657289002557547, "percentage": 67.31, "elapsed_time": "6:47:02", "remaining_time": "3:17:38"} +{"current_steps": 1317, "total_steps": 1955, "loss": 1.0501, "lr": 5.819283082504482e-06, "epoch": 3.368286445012788, "percentage": 67.37, "elapsed_time": "6:47:21", "remaining_time": "3:17:20"} +{"current_steps": 1318, "total_steps": 1955, "loss": 1.047, "lr": 5.803065388009707e-06, "epoch": 3.370843989769821, "percentage": 67.42, "elapsed_time": "6:47:39", "remaining_time": "3:17:01"} +{"current_steps": 1319, "total_steps": 1955, "loss": 1.0459, "lr": 5.78686108103559e-06, "epoch": 3.373401534526854, "percentage": 67.47, "elapsed_time": "6:47:57", "remaining_time": "3:16:42"} +{"current_steps": 1320, "total_steps": 1955, "loss": 1.0203, "lr": 5.770670213271166e-06, "epoch": 3.3759590792838874, "percentage": 67.52, "elapsed_time": "6:48:16", "remaining_time": "3:16:24"} +{"current_steps": 1321, "total_steps": 1955, "loss": 1.0211, "lr": 5.754492836362597e-06, "epoch": 3.3785166240409206, "percentage": 67.57, "elapsed_time": "6:48:35", "remaining_time": "3:16:05"} +{"current_steps": 1322, "total_steps": 1955, "loss": 1.0312, "lr": 5.7383290019130145e-06, "epoch": 3.381074168797954, "percentage": 67.62, "elapsed_time": "6:48:53", "remaining_time": "3:15:47"} +{"current_steps": 1323, "total_steps": 1955, "loss": 1.0233, "lr": 5.722178761482357e-06, "epoch": 3.3836317135549874, "percentage": 67.67, "elapsed_time": "6:49:12", "remaining_time": "3:15:28"} +{"current_steps": 1324, "total_steps": 1955, "loss": 1.038, "lr": 5.706042166587194e-06, "epoch": 3.3861892583120206, "percentage": 67.72, "elapsed_time": "6:49:31", "remaining_time": "3:15:10"} +{"current_steps": 1325, "total_steps": 1955, "loss": 1.0401, "lr": 5.689919268700559e-06, "epoch": 3.3887468030690537, "percentage": 67.77, "elapsed_time": "6:49:49", "remaining_time": "3:14:51"} +{"current_steps": 1326, "total_steps": 1955, "loss": 1.042, "lr": 5.673810119251815e-06, "epoch": 3.391304347826087, "percentage": 67.83, "elapsed_time": "6:50:08", "remaining_time": "3:14:33"} +{"current_steps": 1327, "total_steps": 1955, "loss": 1.028, "lr": 5.657714769626456e-06, "epoch": 3.39386189258312, "percentage": 67.88, "elapsed_time": "6:50:26", "remaining_time": "3:14:14"} +{"current_steps": 1328, "total_steps": 1955, "loss": 1.0164, "lr": 5.641633271165955e-06, "epoch": 3.3964194373401533, "percentage": 67.93, "elapsed_time": "6:50:44", "remaining_time": "3:13:55"} +{"current_steps": 1329, "total_steps": 1955, "loss": 1.0785, "lr": 5.625565675167614e-06, "epoch": 3.398976982097187, "percentage": 67.98, "elapsed_time": "6:51:01", "remaining_time": "3:13:36"} +{"current_steps": 1330, "total_steps": 1955, "loss": 1.0622, "lr": 5.609512032884385e-06, "epoch": 3.40153452685422, "percentage": 68.03, "elapsed_time": "6:51:20", "remaining_time": "3:13:18"} +{"current_steps": 1331, "total_steps": 1955, "loss": 1.0335, "lr": 5.593472395524692e-06, "epoch": 3.4040920716112533, "percentage": 68.08, "elapsed_time": "6:51:39", "remaining_time": "3:12:59"} +{"current_steps": 1332, "total_steps": 1955, "loss": 1.0527, "lr": 5.577446814252311e-06, "epoch": 3.4066496163682864, "percentage": 68.13, "elapsed_time": "6:51:58", "remaining_time": "3:12:41"} +{"current_steps": 1333, "total_steps": 1955, "loss": 1.0281, "lr": 5.561435340186167e-06, "epoch": 3.4092071611253196, "percentage": 68.18, "elapsed_time": "6:52:16", "remaining_time": "3:12:22"} +{"current_steps": 1334, "total_steps": 1955, "loss": 1.0581, "lr": 5.545438024400193e-06, "epoch": 3.411764705882353, "percentage": 68.24, "elapsed_time": "6:52:35", "remaining_time": "3:12:04"} +{"current_steps": 1335, "total_steps": 1955, "loss": 1.0419, "lr": 5.529454917923149e-06, "epoch": 3.414322250639386, "percentage": 68.29, "elapsed_time": "6:52:54", "remaining_time": "3:11:45"} +{"current_steps": 1336, "total_steps": 1955, "loss": 1.0404, "lr": 5.513486071738481e-06, "epoch": 3.4168797953964196, "percentage": 68.34, "elapsed_time": "6:53:12", "remaining_time": "3:11:27"} +{"current_steps": 1337, "total_steps": 1955, "loss": 1.0375, "lr": 5.497531536784138e-06, "epoch": 3.419437340153453, "percentage": 68.39, "elapsed_time": "6:53:31", "remaining_time": "3:11:08"} +{"current_steps": 1338, "total_steps": 1955, "loss": 1.0262, "lr": 5.481591363952421e-06, "epoch": 3.421994884910486, "percentage": 68.44, "elapsed_time": "6:53:49", "remaining_time": "3:10:49"} +{"current_steps": 1339, "total_steps": 1955, "loss": 1.0317, "lr": 5.465665604089829e-06, "epoch": 3.424552429667519, "percentage": 68.49, "elapsed_time": "6:54:07", "remaining_time": "3:10:31"} +{"current_steps": 1340, "total_steps": 1955, "loss": 1.0867, "lr": 5.449754307996871e-06, "epoch": 3.4271099744245523, "percentage": 68.54, "elapsed_time": "6:54:26", "remaining_time": "3:10:12"} +{"current_steps": 1341, "total_steps": 1955, "loss": 1.0755, "lr": 5.433857526427924e-06, "epoch": 3.4296675191815855, "percentage": 68.59, "elapsed_time": "6:54:44", "remaining_time": "3:09:54"} +{"current_steps": 1342, "total_steps": 1955, "loss": 1.038, "lr": 5.417975310091068e-06, "epoch": 3.4322250639386187, "percentage": 68.64, "elapsed_time": "6:55:03", "remaining_time": "3:09:35"} +{"current_steps": 1343, "total_steps": 1955, "loss": 1.0512, "lr": 5.402107709647921e-06, "epoch": 3.4347826086956523, "percentage": 68.7, "elapsed_time": "6:55:22", "remaining_time": "3:09:16"} +{"current_steps": 1344, "total_steps": 1955, "loss": 1.0333, "lr": 5.386254775713482e-06, "epoch": 3.4373401534526855, "percentage": 68.75, "elapsed_time": "6:55:40", "remaining_time": "3:08:58"} +{"current_steps": 1345, "total_steps": 1955, "loss": 1.0215, "lr": 5.370416558855955e-06, "epoch": 3.4398976982097187, "percentage": 68.8, "elapsed_time": "6:55:59", "remaining_time": "3:08:39"} +{"current_steps": 1346, "total_steps": 1955, "loss": 1.0628, "lr": 5.354593109596621e-06, "epoch": 3.442455242966752, "percentage": 68.85, "elapsed_time": "6:56:18", "remaining_time": "3:08:21"} +{"current_steps": 1347, "total_steps": 1955, "loss": 1.0398, "lr": 5.338784478409629e-06, "epoch": 3.445012787723785, "percentage": 68.9, "elapsed_time": "6:56:36", "remaining_time": "3:08:02"} +{"current_steps": 1348, "total_steps": 1955, "loss": 1.0603, "lr": 5.3229907157218745e-06, "epoch": 3.4475703324808182, "percentage": 68.95, "elapsed_time": "6:56:55", "remaining_time": "3:07:44"} +{"current_steps": 1349, "total_steps": 1955, "loss": 1.0651, "lr": 5.307211871912828e-06, "epoch": 3.4501278772378514, "percentage": 69.0, "elapsed_time": "6:57:13", "remaining_time": "3:07:25"} +{"current_steps": 1350, "total_steps": 1955, "loss": 1.0524, "lr": 5.291447997314367e-06, "epoch": 3.452685421994885, "percentage": 69.05, "elapsed_time": "6:57:31", "remaining_time": "3:07:06"} +{"current_steps": 1351, "total_steps": 1955, "loss": 1.0483, "lr": 5.275699142210615e-06, "epoch": 3.455242966751918, "percentage": 69.1, "elapsed_time": "6:57:50", "remaining_time": "3:06:48"} +{"current_steps": 1352, "total_steps": 1955, "loss": 1.056, "lr": 5.259965356837795e-06, "epoch": 3.4578005115089514, "percentage": 69.16, "elapsed_time": "6:58:09", "remaining_time": "3:06:29"} +{"current_steps": 1353, "total_steps": 1955, "loss": 1.0201, "lr": 5.2442466913840504e-06, "epoch": 3.4603580562659846, "percentage": 69.21, "elapsed_time": "6:58:27", "remaining_time": "3:06:11"} +{"current_steps": 1354, "total_steps": 1955, "loss": 1.042, "lr": 5.228543195989303e-06, "epoch": 3.4629156010230178, "percentage": 69.26, "elapsed_time": "6:58:45", "remaining_time": "3:05:52"} +{"current_steps": 1355, "total_steps": 1955, "loss": 1.0335, "lr": 5.212854920745075e-06, "epoch": 3.4654731457800514, "percentage": 69.31, "elapsed_time": "6:59:04", "remaining_time": "3:05:34"} +{"current_steps": 1356, "total_steps": 1955, "loss": 1.0402, "lr": 5.197181915694355e-06, "epoch": 3.4680306905370846, "percentage": 69.36, "elapsed_time": "6:59:23", "remaining_time": "3:05:15"} +{"current_steps": 1357, "total_steps": 1955, "loss": 1.0441, "lr": 5.18152423083141e-06, "epoch": 3.4705882352941178, "percentage": 69.41, "elapsed_time": "6:59:42", "remaining_time": "3:04:57"} +{"current_steps": 1358, "total_steps": 1955, "loss": 1.0468, "lr": 5.1658819161016296e-06, "epoch": 3.473145780051151, "percentage": 69.46, "elapsed_time": "7:00:00", "remaining_time": "3:04:38"} +{"current_steps": 1359, "total_steps": 1955, "loss": 1.0664, "lr": 5.1502550214013994e-06, "epoch": 3.475703324808184, "percentage": 69.51, "elapsed_time": "7:00:19", "remaining_time": "3:04:20"} +{"current_steps": 1360, "total_steps": 1955, "loss": 1.0536, "lr": 5.134643596577898e-06, "epoch": 3.4782608695652173, "percentage": 69.57, "elapsed_time": "7:00:38", "remaining_time": "3:04:01"} +{"current_steps": 1361, "total_steps": 1955, "loss": 1.0357, "lr": 5.1190476914289645e-06, "epoch": 3.4808184143222505, "percentage": 69.62, "elapsed_time": "7:00:56", "remaining_time": "3:03:43"} +{"current_steps": 1362, "total_steps": 1955, "loss": 1.0428, "lr": 5.103467355702928e-06, "epoch": 3.483375959079284, "percentage": 69.67, "elapsed_time": "7:01:15", "remaining_time": "3:03:24"} +{"current_steps": 1363, "total_steps": 1955, "loss": 1.0586, "lr": 5.087902639098472e-06, "epoch": 3.4859335038363173, "percentage": 69.72, "elapsed_time": "7:01:34", "remaining_time": "3:03:06"} +{"current_steps": 1364, "total_steps": 1955, "loss": 1.0271, "lr": 5.072353591264429e-06, "epoch": 3.4884910485933505, "percentage": 69.77, "elapsed_time": "7:01:52", "remaining_time": "3:02:47"} +{"current_steps": 1365, "total_steps": 1955, "loss": 1.0668, "lr": 5.056820261799668e-06, "epoch": 3.4910485933503836, "percentage": 69.82, "elapsed_time": "7:02:10", "remaining_time": "3:02:28"} +{"current_steps": 1366, "total_steps": 1955, "loss": 1.0853, "lr": 5.041302700252921e-06, "epoch": 3.493606138107417, "percentage": 69.87, "elapsed_time": "7:02:29", "remaining_time": "3:02:10"} +{"current_steps": 1367, "total_steps": 1955, "loss": 1.0664, "lr": 5.02580095612262e-06, "epoch": 3.49616368286445, "percentage": 69.92, "elapsed_time": "7:02:48", "remaining_time": "3:01:51"} +{"current_steps": 1368, "total_steps": 1955, "loss": 1.0043, "lr": 5.010315078856733e-06, "epoch": 3.498721227621483, "percentage": 69.97, "elapsed_time": "7:03:06", "remaining_time": "3:01:33"} +{"current_steps": 1369, "total_steps": 1955, "loss": 1.0535, "lr": 4.99484511785263e-06, "epoch": 3.501278772378517, "percentage": 70.03, "elapsed_time": "7:03:24", "remaining_time": "3:01:14"} +{"current_steps": 1370, "total_steps": 1955, "loss": 1.0565, "lr": 4.979391122456899e-06, "epoch": 3.50383631713555, "percentage": 70.08, "elapsed_time": "7:03:42", "remaining_time": "3:00:55"} +{"current_steps": 1371, "total_steps": 1955, "loss": 1.0531, "lr": 4.963953141965208e-06, "epoch": 3.506393861892583, "percentage": 70.13, "elapsed_time": "7:04:01", "remaining_time": "3:00:37"} +{"current_steps": 1372, "total_steps": 1955, "loss": 1.0707, "lr": 4.94853122562213e-06, "epoch": 3.5089514066496164, "percentage": 70.18, "elapsed_time": "7:04:20", "remaining_time": "3:00:18"} +{"current_steps": 1373, "total_steps": 1955, "loss": 1.0288, "lr": 4.9331254226210136e-06, "epoch": 3.5115089514066495, "percentage": 70.23, "elapsed_time": "7:04:39", "remaining_time": "3:00:00"} +{"current_steps": 1374, "total_steps": 1955, "loss": 1.0516, "lr": 4.917735782103796e-06, "epoch": 3.5140664961636827, "percentage": 70.28, "elapsed_time": "7:04:57", "remaining_time": "2:59:41"} +{"current_steps": 1375, "total_steps": 1955, "loss": 1.0232, "lr": 4.902362353160851e-06, "epoch": 3.516624040920716, "percentage": 70.33, "elapsed_time": "7:05:15", "remaining_time": "2:59:23"} +{"current_steps": 1376, "total_steps": 1955, "loss": 1.0296, "lr": 4.887005184830861e-06, "epoch": 3.5191815856777495, "percentage": 70.38, "elapsed_time": "7:05:34", "remaining_time": "2:59:04"} +{"current_steps": 1377, "total_steps": 1955, "loss": 1.0571, "lr": 4.871664326100625e-06, "epoch": 3.5217391304347827, "percentage": 70.43, "elapsed_time": "7:05:53", "remaining_time": "2:58:46"} +{"current_steps": 1378, "total_steps": 1955, "loss": 1.0221, "lr": 4.856339825904921e-06, "epoch": 3.524296675191816, "percentage": 70.49, "elapsed_time": "7:06:11", "remaining_time": "2:58:27"} +{"current_steps": 1379, "total_steps": 1955, "loss": 1.0872, "lr": 4.841031733126346e-06, "epoch": 3.526854219948849, "percentage": 70.54, "elapsed_time": "7:06:30", "remaining_time": "2:58:08"} +{"current_steps": 1380, "total_steps": 1955, "loss": 1.0272, "lr": 4.825740096595159e-06, "epoch": 3.5294117647058822, "percentage": 70.59, "elapsed_time": "7:06:49", "remaining_time": "2:57:50"} +{"current_steps": 1381, "total_steps": 1955, "loss": 1.0833, "lr": 4.81046496508913e-06, "epoch": 3.531969309462916, "percentage": 70.64, "elapsed_time": "7:07:07", "remaining_time": "2:57:31"} +{"current_steps": 1382, "total_steps": 1955, "loss": 1.0531, "lr": 4.795206387333371e-06, "epoch": 3.5345268542199486, "percentage": 70.69, "elapsed_time": "7:07:25", "remaining_time": "2:57:13"} +{"current_steps": 1383, "total_steps": 1955, "loss": 1.0378, "lr": 4.779964412000207e-06, "epoch": 3.5370843989769822, "percentage": 70.74, "elapsed_time": "7:07:44", "remaining_time": "2:56:54"} +{"current_steps": 1384, "total_steps": 1955, "loss": 1.0668, "lr": 4.764739087708988e-06, "epoch": 3.5396419437340154, "percentage": 70.79, "elapsed_time": "7:08:02", "remaining_time": "2:56:36"} +{"current_steps": 1385, "total_steps": 1955, "loss": 1.0448, "lr": 4.749530463025961e-06, "epoch": 3.5421994884910486, "percentage": 70.84, "elapsed_time": "7:08:21", "remaining_time": "2:56:17"} +{"current_steps": 1386, "total_steps": 1955, "loss": 1.0647, "lr": 4.734338586464096e-06, "epoch": 3.544757033248082, "percentage": 70.9, "elapsed_time": "7:08:40", "remaining_time": "2:55:59"} +{"current_steps": 1387, "total_steps": 1955, "loss": 1.0165, "lr": 4.719163506482942e-06, "epoch": 3.547314578005115, "percentage": 70.95, "elapsed_time": "7:08:58", "remaining_time": "2:55:40"} +{"current_steps": 1388, "total_steps": 1955, "loss": 1.0381, "lr": 4.704005271488472e-06, "epoch": 3.5498721227621486, "percentage": 71.0, "elapsed_time": "7:09:17", "remaining_time": "2:55:22"} +{"current_steps": 1389, "total_steps": 1955, "loss": 1.0659, "lr": 4.688863929832922e-06, "epoch": 3.5524296675191813, "percentage": 71.05, "elapsed_time": "7:09:35", "remaining_time": "2:55:03"} +{"current_steps": 1390, "total_steps": 1955, "loss": 1.0741, "lr": 4.673739529814653e-06, "epoch": 3.554987212276215, "percentage": 71.1, "elapsed_time": "7:09:54", "remaining_time": "2:54:44"} +{"current_steps": 1391, "total_steps": 1955, "loss": 1.0079, "lr": 4.658632119677965e-06, "epoch": 3.557544757033248, "percentage": 71.15, "elapsed_time": "7:10:13", "remaining_time": "2:54:26"} +{"current_steps": 1392, "total_steps": 1955, "loss": 1.0333, "lr": 4.6435417476129745e-06, "epoch": 3.5601023017902813, "percentage": 71.2, "elapsed_time": "7:10:30", "remaining_time": "2:54:07"} +{"current_steps": 1393, "total_steps": 1955, "loss": 1.0531, "lr": 4.628468461755456e-06, "epoch": 3.5626598465473145, "percentage": 71.25, "elapsed_time": "7:10:49", "remaining_time": "2:53:48"} +{"current_steps": 1394, "total_steps": 1955, "loss": 1.0541, "lr": 4.613412310186669e-06, "epoch": 3.5652173913043477, "percentage": 71.3, "elapsed_time": "7:11:07", "remaining_time": "2:53:30"} +{"current_steps": 1395, "total_steps": 1955, "loss": 1.0608, "lr": 4.598373340933226e-06, "epoch": 3.5677749360613813, "percentage": 71.36, "elapsed_time": "7:11:25", "remaining_time": "2:53:11"} +{"current_steps": 1396, "total_steps": 1955, "loss": 1.0385, "lr": 4.583351601966928e-06, "epoch": 3.5703324808184145, "percentage": 71.41, "elapsed_time": "7:11:43", "remaining_time": "2:52:52"} +{"current_steps": 1397, "total_steps": 1955, "loss": 1.019, "lr": 4.568347141204611e-06, "epoch": 3.5728900255754477, "percentage": 71.46, "elapsed_time": "7:12:02", "remaining_time": "2:52:34"} +{"current_steps": 1398, "total_steps": 1955, "loss": 1.0592, "lr": 4.553360006508003e-06, "epoch": 3.575447570332481, "percentage": 71.51, "elapsed_time": "7:12:20", "remaining_time": "2:52:15"} +{"current_steps": 1399, "total_steps": 1955, "loss": 1.0616, "lr": 4.538390245683555e-06, "epoch": 3.578005115089514, "percentage": 71.56, "elapsed_time": "7:12:39", "remaining_time": "2:51:57"} +{"current_steps": 1400, "total_steps": 1955, "loss": 1.0209, "lr": 4.523437906482313e-06, "epoch": 3.580562659846547, "percentage": 71.61, "elapsed_time": "7:12:58", "remaining_time": "2:51:38"} +{"current_steps": 1401, "total_steps": 1955, "loss": 1.0308, "lr": 4.508503036599743e-06, "epoch": 3.5831202046035804, "percentage": 71.66, "elapsed_time": "7:13:16", "remaining_time": "2:51:20"} +{"current_steps": 1402, "total_steps": 1955, "loss": 1.0512, "lr": 4.493585683675575e-06, "epoch": 3.585677749360614, "percentage": 71.71, "elapsed_time": "7:13:35", "remaining_time": "2:51:01"} +{"current_steps": 1403, "total_steps": 1955, "loss": 1.0459, "lr": 4.478685895293685e-06, "epoch": 3.588235294117647, "percentage": 71.76, "elapsed_time": "7:13:54", "remaining_time": "2:50:43"} +{"current_steps": 1404, "total_steps": 1955, "loss": 1.0521, "lr": 4.463803718981905e-06, "epoch": 3.5907928388746804, "percentage": 71.82, "elapsed_time": "7:14:12", "remaining_time": "2:50:24"} +{"current_steps": 1405, "total_steps": 1955, "loss": 1.0426, "lr": 4.448939202211896e-06, "epoch": 3.5933503836317136, "percentage": 71.87, "elapsed_time": "7:14:31", "remaining_time": "2:50:05"} +{"current_steps": 1406, "total_steps": 1955, "loss": 1.0446, "lr": 4.434092392398978e-06, "epoch": 3.5959079283887467, "percentage": 71.92, "elapsed_time": "7:14:49", "remaining_time": "2:49:47"} +{"current_steps": 1407, "total_steps": 1955, "loss": 1.0536, "lr": 4.419263336902006e-06, "epoch": 3.59846547314578, "percentage": 71.97, "elapsed_time": "7:15:08", "remaining_time": "2:49:28"} +{"current_steps": 1408, "total_steps": 1955, "loss": 1.056, "lr": 4.404452083023183e-06, "epoch": 3.601023017902813, "percentage": 72.02, "elapsed_time": "7:15:27", "remaining_time": "2:49:10"} +{"current_steps": 1409, "total_steps": 1955, "loss": 1.0326, "lr": 4.389658678007933e-06, "epoch": 3.6035805626598467, "percentage": 72.07, "elapsed_time": "7:15:45", "remaining_time": "2:48:51"} +{"current_steps": 1410, "total_steps": 1955, "loss": 1.0065, "lr": 4.374883169044757e-06, "epoch": 3.60613810741688, "percentage": 72.12, "elapsed_time": "7:16:03", "remaining_time": "2:48:32"} +{"current_steps": 1411, "total_steps": 1955, "loss": 1.0487, "lr": 4.3601256032650576e-06, "epoch": 3.608695652173913, "percentage": 72.17, "elapsed_time": "7:16:22", "remaining_time": "2:48:14"} +{"current_steps": 1412, "total_steps": 1955, "loss": 1.067, "lr": 4.345386027743005e-06, "epoch": 3.6112531969309463, "percentage": 72.23, "elapsed_time": "7:16:40", "remaining_time": "2:47:55"} +{"current_steps": 1413, "total_steps": 1955, "loss": 1.0233, "lr": 4.330664489495386e-06, "epoch": 3.6138107416879794, "percentage": 72.28, "elapsed_time": "7:16:58", "remaining_time": "2:47:36"} +{"current_steps": 1414, "total_steps": 1955, "loss": 1.0429, "lr": 4.315961035481445e-06, "epoch": 3.6163682864450126, "percentage": 72.33, "elapsed_time": "7:17:16", "remaining_time": "2:47:18"} +{"current_steps": 1415, "total_steps": 1955, "loss": 1.0221, "lr": 4.30127571260275e-06, "epoch": 3.618925831202046, "percentage": 72.38, "elapsed_time": "7:17:35", "remaining_time": "2:46:59"} +{"current_steps": 1416, "total_steps": 1955, "loss": 1.0454, "lr": 4.286608567703025e-06, "epoch": 3.6214833759590794, "percentage": 72.43, "elapsed_time": "7:17:53", "remaining_time": "2:46:41"} +{"current_steps": 1417, "total_steps": 1955, "loss": 1.0208, "lr": 4.271959647568017e-06, "epoch": 3.6240409207161126, "percentage": 72.48, "elapsed_time": "7:18:11", "remaining_time": "2:46:22"} +{"current_steps": 1418, "total_steps": 1955, "loss": 1.0575, "lr": 4.257328998925339e-06, "epoch": 3.626598465473146, "percentage": 72.53, "elapsed_time": "7:18:30", "remaining_time": "2:46:03"} +{"current_steps": 1419, "total_steps": 1955, "loss": 1.0486, "lr": 4.242716668444304e-06, "epoch": 3.629156010230179, "percentage": 72.58, "elapsed_time": "7:18:48", "remaining_time": "2:45:45"} +{"current_steps": 1420, "total_steps": 1955, "loss": 1.0511, "lr": 4.228122702735818e-06, "epoch": 3.631713554987212, "percentage": 72.63, "elapsed_time": "7:19:07", "remaining_time": "2:45:26"} +{"current_steps": 1421, "total_steps": 1955, "loss": 1.0664, "lr": 4.213547148352193e-06, "epoch": 3.634271099744246, "percentage": 72.69, "elapsed_time": "7:19:25", "remaining_time": "2:45:08"} +{"current_steps": 1422, "total_steps": 1955, "loss": 1.0185, "lr": 4.198990051787012e-06, "epoch": 3.6368286445012785, "percentage": 72.74, "elapsed_time": "7:19:44", "remaining_time": "2:44:49"} +{"current_steps": 1423, "total_steps": 1955, "loss": 1.0481, "lr": 4.184451459474983e-06, "epoch": 3.639386189258312, "percentage": 72.79, "elapsed_time": "7:20:03", "remaining_time": "2:44:31"} +{"current_steps": 1424, "total_steps": 1955, "loss": 1.0392, "lr": 4.169931417791788e-06, "epoch": 3.6419437340153453, "percentage": 72.84, "elapsed_time": "7:20:22", "remaining_time": "2:44:12"} +{"current_steps": 1425, "total_steps": 1955, "loss": 1.0682, "lr": 4.155429973053935e-06, "epoch": 3.6445012787723785, "percentage": 72.89, "elapsed_time": "7:20:40", "remaining_time": "2:43:54"} +{"current_steps": 1426, "total_steps": 1955, "loss": 1.0256, "lr": 4.140947171518609e-06, "epoch": 3.6470588235294117, "percentage": 72.94, "elapsed_time": "7:20:59", "remaining_time": "2:43:35"} +{"current_steps": 1427, "total_steps": 1955, "loss": 1.0385, "lr": 4.126483059383534e-06, "epoch": 3.649616368286445, "percentage": 72.99, "elapsed_time": "7:21:18", "remaining_time": "2:43:17"} +{"current_steps": 1428, "total_steps": 1955, "loss": 1.0318, "lr": 4.112037682786811e-06, "epoch": 3.6521739130434785, "percentage": 73.04, "elapsed_time": "7:21:36", "remaining_time": "2:42:58"} +{"current_steps": 1429, "total_steps": 1955, "loss": 1.0216, "lr": 4.097611087806779e-06, "epoch": 3.6547314578005117, "percentage": 73.09, "elapsed_time": "7:21:54", "remaining_time": "2:42:39"} +{"current_steps": 1430, "total_steps": 1955, "loss": 1.0534, "lr": 4.083203320461867e-06, "epoch": 3.657289002557545, "percentage": 73.15, "elapsed_time": "7:22:13", "remaining_time": "2:42:21"} +{"current_steps": 1431, "total_steps": 1955, "loss": 1.0466, "lr": 4.068814426710447e-06, "epoch": 3.659846547314578, "percentage": 73.2, "elapsed_time": "7:22:32", "remaining_time": "2:42:02"} +{"current_steps": 1432, "total_steps": 1955, "loss": 1.0511, "lr": 4.054444452450688e-06, "epoch": 3.662404092071611, "percentage": 73.25, "elapsed_time": "7:22:50", "remaining_time": "2:41:44"} +{"current_steps": 1433, "total_steps": 1955, "loss": 1.0249, "lr": 4.040093443520404e-06, "epoch": 3.6649616368286444, "percentage": 73.3, "elapsed_time": "7:23:09", "remaining_time": "2:41:25"} +{"current_steps": 1434, "total_steps": 1955, "loss": 1.0316, "lr": 4.02576144569693e-06, "epoch": 3.6675191815856776, "percentage": 73.35, "elapsed_time": "7:23:28", "remaining_time": "2:41:07"} +{"current_steps": 1435, "total_steps": 1955, "loss": 1.0269, "lr": 4.011448504696933e-06, "epoch": 3.670076726342711, "percentage": 73.4, "elapsed_time": "7:23:46", "remaining_time": "2:40:48"} +{"current_steps": 1436, "total_steps": 1955, "loss": 1.0582, "lr": 3.997154666176306e-06, "epoch": 3.6726342710997444, "percentage": 73.45, "elapsed_time": "7:24:05", "remaining_time": "2:40:30"} +{"current_steps": 1437, "total_steps": 1955, "loss": 1.0323, "lr": 3.982879975730015e-06, "epoch": 3.6751918158567776, "percentage": 73.5, "elapsed_time": "7:24:24", "remaining_time": "2:40:11"} +{"current_steps": 1438, "total_steps": 1955, "loss": 1.0925, "lr": 3.968624478891935e-06, "epoch": 3.6777493606138107, "percentage": 73.55, "elapsed_time": "7:24:42", "remaining_time": "2:39:53"} +{"current_steps": 1439, "total_steps": 1955, "loss": 1.0265, "lr": 3.954388221134721e-06, "epoch": 3.680306905370844, "percentage": 73.61, "elapsed_time": "7:25:01", "remaining_time": "2:39:34"} +{"current_steps": 1440, "total_steps": 1955, "loss": 1.0263, "lr": 3.9401712478696584e-06, "epoch": 3.682864450127877, "percentage": 73.66, "elapsed_time": "7:25:20", "remaining_time": "2:39:16"} +{"current_steps": 1441, "total_steps": 1955, "loss": 1.0271, "lr": 3.9259736044465165e-06, "epoch": 3.6854219948849103, "percentage": 73.71, "elapsed_time": "7:25:38", "remaining_time": "2:38:57"} +{"current_steps": 1442, "total_steps": 1955, "loss": 1.0233, "lr": 3.91179533615341e-06, "epoch": 3.687979539641944, "percentage": 73.76, "elapsed_time": "7:25:56", "remaining_time": "2:38:38"} +{"current_steps": 1443, "total_steps": 1955, "loss": 1.0072, "lr": 3.897636488216641e-06, "epoch": 3.690537084398977, "percentage": 73.81, "elapsed_time": "7:26:15", "remaining_time": "2:38:20"} +{"current_steps": 1444, "total_steps": 1955, "loss": 1.0349, "lr": 3.88349710580058e-06, "epoch": 3.6930946291560103, "percentage": 73.86, "elapsed_time": "7:26:33", "remaining_time": "2:38:01"} +{"current_steps": 1445, "total_steps": 1955, "loss": 1.057, "lr": 3.8693772340074945e-06, "epoch": 3.6956521739130435, "percentage": 73.91, "elapsed_time": "7:26:52", "remaining_time": "2:37:43"} +{"current_steps": 1446, "total_steps": 1955, "loss": 1.0182, "lr": 3.855276917877408e-06, "epoch": 3.6982097186700766, "percentage": 73.96, "elapsed_time": "7:27:11", "remaining_time": "2:37:24"} +{"current_steps": 1447, "total_steps": 1955, "loss": 1.0478, "lr": 3.8411962023879844e-06, "epoch": 3.70076726342711, "percentage": 74.02, "elapsed_time": "7:27:29", "remaining_time": "2:37:06"} +{"current_steps": 1448, "total_steps": 1955, "loss": 1.0695, "lr": 3.827135132454351e-06, "epoch": 3.703324808184143, "percentage": 74.07, "elapsed_time": "7:27:48", "remaining_time": "2:36:47"} +{"current_steps": 1449, "total_steps": 1955, "loss": 1.0203, "lr": 3.8130937529289734e-06, "epoch": 3.7058823529411766, "percentage": 74.12, "elapsed_time": "7:28:07", "remaining_time": "2:36:29"} +{"current_steps": 1450, "total_steps": 1955, "loss": 1.0344, "lr": 3.7990721086015114e-06, "epoch": 3.70843989769821, "percentage": 74.17, "elapsed_time": "7:28:25", "remaining_time": "2:36:10"} +{"current_steps": 1451, "total_steps": 1955, "loss": 1.0135, "lr": 3.78507024419867e-06, "epoch": 3.710997442455243, "percentage": 74.22, "elapsed_time": "7:28:43", "remaining_time": "2:35:51"} +{"current_steps": 1452, "total_steps": 1955, "loss": 1.0443, "lr": 3.7710882043840513e-06, "epoch": 3.713554987212276, "percentage": 74.27, "elapsed_time": "7:29:01", "remaining_time": "2:35:33"} +{"current_steps": 1453, "total_steps": 1955, "loss": 1.0276, "lr": 3.7571260337580283e-06, "epoch": 3.7161125319693094, "percentage": 74.32, "elapsed_time": "7:29:20", "remaining_time": "2:35:14"} +{"current_steps": 1454, "total_steps": 1955, "loss": 1.0716, "lr": 3.743183776857602e-06, "epoch": 3.718670076726343, "percentage": 74.37, "elapsed_time": "7:29:38", "remaining_time": "2:34:55"} +{"current_steps": 1455, "total_steps": 1955, "loss": 1.026, "lr": 3.7292614781562384e-06, "epoch": 3.7212276214833757, "percentage": 74.42, "elapsed_time": "7:29:57", "remaining_time": "2:34:37"} +{"current_steps": 1456, "total_steps": 1955, "loss": 1.0232, "lr": 3.715359182063748e-06, "epoch": 3.7237851662404093, "percentage": 74.48, "elapsed_time": "7:30:15", "remaining_time": "2:34:18"} +{"current_steps": 1457, "total_steps": 1955, "loss": 1.068, "lr": 3.7014769329261325e-06, "epoch": 3.7263427109974425, "percentage": 74.53, "elapsed_time": "7:30:34", "remaining_time": "2:34:00"} +{"current_steps": 1458, "total_steps": 1955, "loss": 1.0256, "lr": 3.68761477502545e-06, "epoch": 3.7289002557544757, "percentage": 74.58, "elapsed_time": "7:30:53", "remaining_time": "2:33:41"} +{"current_steps": 1459, "total_steps": 1955, "loss": 1.0607, "lr": 3.673772752579665e-06, "epoch": 3.731457800511509, "percentage": 74.63, "elapsed_time": "7:31:11", "remaining_time": "2:33:23"} +{"current_steps": 1460, "total_steps": 1955, "loss": 1.0436, "lr": 3.6599509097425246e-06, "epoch": 3.734015345268542, "percentage": 74.68, "elapsed_time": "7:31:30", "remaining_time": "2:33:04"} +{"current_steps": 1461, "total_steps": 1955, "loss": 1.0202, "lr": 3.646149290603398e-06, "epoch": 3.7365728900255757, "percentage": 74.73, "elapsed_time": "7:31:49", "remaining_time": "2:32:46"} +{"current_steps": 1462, "total_steps": 1955, "loss": 1.0263, "lr": 3.6323679391871447e-06, "epoch": 3.7391304347826084, "percentage": 74.78, "elapsed_time": "7:32:07", "remaining_time": "2:32:27"} +{"current_steps": 1463, "total_steps": 1955, "loss": 1.0356, "lr": 3.6186068994539746e-06, "epoch": 3.741687979539642, "percentage": 74.83, "elapsed_time": "7:32:26", "remaining_time": "2:32:09"} +{"current_steps": 1464, "total_steps": 1955, "loss": 1.0554, "lr": 3.6048662152993062e-06, "epoch": 3.7442455242966752, "percentage": 74.88, "elapsed_time": "7:32:44", "remaining_time": "2:31:50"} +{"current_steps": 1465, "total_steps": 1955, "loss": 1.0428, "lr": 3.5911459305536302e-06, "epoch": 3.7468030690537084, "percentage": 74.94, "elapsed_time": "7:33:03", "remaining_time": "2:31:32"} +{"current_steps": 1466, "total_steps": 1955, "loss": 1.0384, "lr": 3.577446088982357e-06, "epoch": 3.7493606138107416, "percentage": 74.99, "elapsed_time": "7:33:21", "remaining_time": "2:31:13"} +{"current_steps": 1467, "total_steps": 1955, "loss": 1.0403, "lr": 3.563766734285704e-06, "epoch": 3.7519181585677748, "percentage": 75.04, "elapsed_time": "7:33:39", "remaining_time": "2:30:54"} +{"current_steps": 1468, "total_steps": 1955, "loss": 1.036, "lr": 3.550107910098526e-06, "epoch": 3.7544757033248084, "percentage": 75.09, "elapsed_time": "7:33:58", "remaining_time": "2:30:36"} +{"current_steps": 1469, "total_steps": 1955, "loss": 1.0601, "lr": 3.5364696599901837e-06, "epoch": 3.7570332480818416, "percentage": 75.14, "elapsed_time": "7:34:17", "remaining_time": "2:30:17"} +{"current_steps": 1470, "total_steps": 1955, "loss": 1.016, "lr": 3.5228520274644264e-06, "epoch": 3.7595907928388748, "percentage": 75.19, "elapsed_time": "7:34:35", "remaining_time": "2:29:59"} +{"current_steps": 1471, "total_steps": 1955, "loss": 1.0493, "lr": 3.5092550559592243e-06, "epoch": 3.762148337595908, "percentage": 75.24, "elapsed_time": "7:34:53", "remaining_time": "2:29:40"} +{"current_steps": 1472, "total_steps": 1955, "loss": 1.0433, "lr": 3.495678788846648e-06, "epoch": 3.764705882352941, "percentage": 75.29, "elapsed_time": "7:35:12", "remaining_time": "2:29:21"} +{"current_steps": 1473, "total_steps": 1955, "loss": 1.0257, "lr": 3.482123269432722e-06, "epoch": 3.7672634271099743, "percentage": 75.35, "elapsed_time": "7:35:31", "remaining_time": "2:29:03"} +{"current_steps": 1474, "total_steps": 1955, "loss": 1.0329, "lr": 3.4685885409572896e-06, "epoch": 3.7698209718670075, "percentage": 75.4, "elapsed_time": "7:35:49", "remaining_time": "2:28:44"} +{"current_steps": 1475, "total_steps": 1955, "loss": 1.0491, "lr": 3.4550746465938754e-06, "epoch": 3.772378516624041, "percentage": 75.45, "elapsed_time": "7:36:07", "remaining_time": "2:28:26"} +{"current_steps": 1476, "total_steps": 1955, "loss": 1.041, "lr": 3.441581629449542e-06, "epoch": 3.7749360613810743, "percentage": 75.5, "elapsed_time": "7:36:26", "remaining_time": "2:28:07"} +{"current_steps": 1477, "total_steps": 1955, "loss": 1.009, "lr": 3.4281095325647682e-06, "epoch": 3.7774936061381075, "percentage": 75.55, "elapsed_time": "7:36:44", "remaining_time": "2:27:48"} +{"current_steps": 1478, "total_steps": 1955, "loss": 1.0347, "lr": 3.4146583989132897e-06, "epoch": 3.7800511508951407, "percentage": 75.6, "elapsed_time": "7:37:03", "remaining_time": "2:27:30"} +{"current_steps": 1479, "total_steps": 1955, "loss": 1.0205, "lr": 3.4012282714019784e-06, "epoch": 3.782608695652174, "percentage": 75.65, "elapsed_time": "7:37:21", "remaining_time": "2:27:11"} +{"current_steps": 1480, "total_steps": 1955, "loss": 1.0289, "lr": 3.387819192870697e-06, "epoch": 3.785166240409207, "percentage": 75.7, "elapsed_time": "7:37:39", "remaining_time": "2:26:53"} +{"current_steps": 1481, "total_steps": 1955, "loss": 1.0305, "lr": 3.374431206092168e-06, "epoch": 3.78772378516624, "percentage": 75.75, "elapsed_time": "7:37:57", "remaining_time": "2:26:34"} +{"current_steps": 1482, "total_steps": 1955, "loss": 1.0485, "lr": 3.3610643537718347e-06, "epoch": 3.790281329923274, "percentage": 75.81, "elapsed_time": "7:38:15", "remaining_time": "2:26:15"} +{"current_steps": 1483, "total_steps": 1955, "loss": 1.1023, "lr": 3.347718678547719e-06, "epoch": 3.792838874680307, "percentage": 75.86, "elapsed_time": "7:38:34", "remaining_time": "2:25:56"} +{"current_steps": 1484, "total_steps": 1955, "loss": 1.0457, "lr": 3.334394222990307e-06, "epoch": 3.79539641943734, "percentage": 75.91, "elapsed_time": "7:38:52", "remaining_time": "2:25:38"} +{"current_steps": 1485, "total_steps": 1955, "loss": 1.0816, "lr": 3.321091029602378e-06, "epoch": 3.7979539641943734, "percentage": 75.96, "elapsed_time": "7:39:11", "remaining_time": "2:25:19"} +{"current_steps": 1486, "total_steps": 1955, "loss": 1.0702, "lr": 3.3078091408188984e-06, "epoch": 3.8005115089514065, "percentage": 76.01, "elapsed_time": "7:39:29", "remaining_time": "2:25:01"} +{"current_steps": 1487, "total_steps": 1955, "loss": 1.0553, "lr": 3.29454859900688e-06, "epoch": 3.80306905370844, "percentage": 76.06, "elapsed_time": "7:39:48", "remaining_time": "2:24:42"} +{"current_steps": 1488, "total_steps": 1955, "loss": 1.0366, "lr": 3.2813094464652362e-06, "epoch": 3.805626598465473, "percentage": 76.11, "elapsed_time": "7:40:07", "remaining_time": "2:24:24"} +{"current_steps": 1489, "total_steps": 1955, "loss": 1.0434, "lr": 3.2680917254246515e-06, "epoch": 3.8081841432225065, "percentage": 76.16, "elapsed_time": "7:40:25", "remaining_time": "2:24:05"} +{"current_steps": 1490, "total_steps": 1955, "loss": 1.0465, "lr": 3.2548954780474485e-06, "epoch": 3.8107416879795397, "percentage": 76.21, "elapsed_time": "7:40:43", "remaining_time": "2:23:47"} +{"current_steps": 1491, "total_steps": 1955, "loss": 1.0458, "lr": 3.241720746427456e-06, "epoch": 3.813299232736573, "percentage": 76.27, "elapsed_time": "7:41:02", "remaining_time": "2:23:28"} +{"current_steps": 1492, "total_steps": 1955, "loss": 1.0407, "lr": 3.228567572589865e-06, "epoch": 3.815856777493606, "percentage": 76.32, "elapsed_time": "7:41:20", "remaining_time": "2:23:09"} +{"current_steps": 1493, "total_steps": 1955, "loss": 1.0487, "lr": 3.215435998491102e-06, "epoch": 3.8184143222506393, "percentage": 76.37, "elapsed_time": "7:41:38", "remaining_time": "2:22:51"} +{"current_steps": 1494, "total_steps": 1955, "loss": 1.0253, "lr": 3.2023260660187018e-06, "epoch": 3.820971867007673, "percentage": 76.42, "elapsed_time": "7:41:56", "remaining_time": "2:22:32"} +{"current_steps": 1495, "total_steps": 1955, "loss": 1.0735, "lr": 3.189237816991161e-06, "epoch": 3.8235294117647056, "percentage": 76.47, "elapsed_time": "7:42:15", "remaining_time": "2:22:14"} +{"current_steps": 1496, "total_steps": 1955, "loss": 1.0434, "lr": 3.176171293157798e-06, "epoch": 3.8260869565217392, "percentage": 76.52, "elapsed_time": "7:42:34", "remaining_time": "2:21:55"} +{"current_steps": 1497, "total_steps": 1955, "loss": 1.0568, "lr": 3.163126536198653e-06, "epoch": 3.8286445012787724, "percentage": 76.57, "elapsed_time": "7:42:52", "remaining_time": "2:21:36"} +{"current_steps": 1498, "total_steps": 1955, "loss": 1.0244, "lr": 3.150103587724318e-06, "epoch": 3.8312020460358056, "percentage": 76.62, "elapsed_time": "7:43:11", "remaining_time": "2:21:18"} +{"current_steps": 1499, "total_steps": 1955, "loss": 1.0333, "lr": 3.137102489275824e-06, "epoch": 3.833759590792839, "percentage": 76.68, "elapsed_time": "7:43:29", "remaining_time": "2:20:59"} +{"current_steps": 1500, "total_steps": 1955, "loss": 1.0618, "lr": 3.124123282324503e-06, "epoch": 3.836317135549872, "percentage": 76.73, "elapsed_time": "7:43:48", "remaining_time": "2:20:41"} +{"current_steps": 1501, "total_steps": 1955, "loss": 1.0407, "lr": 3.1111660082718666e-06, "epoch": 3.8388746803069056, "percentage": 76.78, "elapsed_time": "7:44:07", "remaining_time": "2:20:22"} +{"current_steps": 1502, "total_steps": 1955, "loss": 1.0503, "lr": 3.0982307084494457e-06, "epoch": 3.8414322250639388, "percentage": 76.83, "elapsed_time": "7:44:25", "remaining_time": "2:20:04"} +{"current_steps": 1503, "total_steps": 1955, "loss": 1.0147, "lr": 3.0853174241186867e-06, "epoch": 3.843989769820972, "percentage": 76.88, "elapsed_time": "7:44:44", "remaining_time": "2:19:45"} +{"current_steps": 1504, "total_steps": 1955, "loss": 1.0612, "lr": 3.072426196470818e-06, "epoch": 3.846547314578005, "percentage": 76.93, "elapsed_time": "7:45:02", "remaining_time": "2:19:27"} +{"current_steps": 1505, "total_steps": 1955, "loss": 1.0613, "lr": 3.0595570666266996e-06, "epoch": 3.8491048593350383, "percentage": 76.98, "elapsed_time": "7:45:21", "remaining_time": "2:19:08"} +{"current_steps": 1506, "total_steps": 1955, "loss": 1.0379, "lr": 3.0467100756367064e-06, "epoch": 3.8516624040920715, "percentage": 77.03, "elapsed_time": "7:45:39", "remaining_time": "2:18:50"} +{"current_steps": 1507, "total_steps": 1955, "loss": 1.0561, "lr": 3.033885264480595e-06, "epoch": 3.8542199488491047, "percentage": 77.08, "elapsed_time": "7:45:58", "remaining_time": "2:18:31"} +{"current_steps": 1508, "total_steps": 1955, "loss": 1.0382, "lr": 3.021082674067373e-06, "epoch": 3.8567774936061383, "percentage": 77.14, "elapsed_time": "7:46:17", "remaining_time": "2:18:12"} +{"current_steps": 1509, "total_steps": 1955, "loss": 1.0667, "lr": 3.0083023452351634e-06, "epoch": 3.8593350383631715, "percentage": 77.19, "elapsed_time": "7:46:35", "remaining_time": "2:17:54"} +{"current_steps": 1510, "total_steps": 1955, "loss": 1.0435, "lr": 2.99554431875108e-06, "epoch": 3.8618925831202047, "percentage": 77.24, "elapsed_time": "7:46:54", "remaining_time": "2:17:35"} +{"current_steps": 1511, "total_steps": 1955, "loss": 1.0179, "lr": 2.982808635311104e-06, "epoch": 3.864450127877238, "percentage": 77.29, "elapsed_time": "7:47:13", "remaining_time": "2:17:17"} +{"current_steps": 1512, "total_steps": 1955, "loss": 1.0038, "lr": 2.9700953355399385e-06, "epoch": 3.867007672634271, "percentage": 77.34, "elapsed_time": "7:47:31", "remaining_time": "2:16:58"} +{"current_steps": 1513, "total_steps": 1955, "loss": 1.0327, "lr": 2.957404459990877e-06, "epoch": 3.869565217391304, "percentage": 77.39, "elapsed_time": "7:47:50", "remaining_time": "2:16:40"} +{"current_steps": 1514, "total_steps": 1955, "loss": 1.0408, "lr": 2.9447360491457034e-06, "epoch": 3.8721227621483374, "percentage": 77.44, "elapsed_time": "7:48:09", "remaining_time": "2:16:21"} +{"current_steps": 1515, "total_steps": 1955, "loss": 1.0436, "lr": 2.93209014341453e-06, "epoch": 3.874680306905371, "percentage": 77.49, "elapsed_time": "7:48:27", "remaining_time": "2:16:03"} +{"current_steps": 1516, "total_steps": 1955, "loss": 1.0549, "lr": 2.9194667831356837e-06, "epoch": 3.877237851662404, "percentage": 77.54, "elapsed_time": "7:48:46", "remaining_time": "2:15:44"} +{"current_steps": 1517, "total_steps": 1955, "loss": 1.0778, "lr": 2.9068660085755775e-06, "epoch": 3.8797953964194374, "percentage": 77.6, "elapsed_time": "7:49:04", "remaining_time": "2:15:25"} +{"current_steps": 1518, "total_steps": 1955, "loss": 1.0495, "lr": 2.8942878599285773e-06, "epoch": 3.8823529411764706, "percentage": 77.65, "elapsed_time": "7:49:22", "remaining_time": "2:15:07"} +{"current_steps": 1519, "total_steps": 1955, "loss": 1.0681, "lr": 2.8817323773168783e-06, "epoch": 3.8849104859335037, "percentage": 77.7, "elapsed_time": "7:49:41", "remaining_time": "2:14:48"} +{"current_steps": 1520, "total_steps": 1955, "loss": 1.0579, "lr": 2.869199600790369e-06, "epoch": 3.887468030690537, "percentage": 77.75, "elapsed_time": "7:50:00", "remaining_time": "2:14:30"} +{"current_steps": 1521, "total_steps": 1955, "loss": 1.0345, "lr": 2.8566895703265216e-06, "epoch": 3.89002557544757, "percentage": 77.8, "elapsed_time": "7:50:18", "remaining_time": "2:14:11"} +{"current_steps": 1522, "total_steps": 1955, "loss": 1.0031, "lr": 2.844202325830241e-06, "epoch": 3.8925831202046037, "percentage": 77.85, "elapsed_time": "7:50:37", "remaining_time": "2:13:53"} +{"current_steps": 1523, "total_steps": 1955, "loss": 1.0407, "lr": 2.8317379071337515e-06, "epoch": 3.895140664961637, "percentage": 77.9, "elapsed_time": "7:50:55", "remaining_time": "2:13:34"} +{"current_steps": 1524, "total_steps": 1955, "loss": 1.0353, "lr": 2.819296353996468e-06, "epoch": 3.89769820971867, "percentage": 77.95, "elapsed_time": "7:51:13", "remaining_time": "2:13:16"} +{"current_steps": 1525, "total_steps": 1955, "loss": 1.0584, "lr": 2.806877706104867e-06, "epoch": 3.9002557544757033, "percentage": 78.01, "elapsed_time": "7:51:32", "remaining_time": "2:12:57"} +{"current_steps": 1526, "total_steps": 1955, "loss": 1.0449, "lr": 2.794482003072364e-06, "epoch": 3.9028132992327365, "percentage": 78.06, "elapsed_time": "7:51:50", "remaining_time": "2:12:38"} +{"current_steps": 1527, "total_steps": 1955, "loss": 1.0357, "lr": 2.7821092844391762e-06, "epoch": 3.90537084398977, "percentage": 78.11, "elapsed_time": "7:52:08", "remaining_time": "2:12:20"} +{"current_steps": 1528, "total_steps": 1955, "loss": 1.0499, "lr": 2.7697595896722206e-06, "epoch": 3.907928388746803, "percentage": 78.16, "elapsed_time": "7:52:27", "remaining_time": "2:12:01"} +{"current_steps": 1529, "total_steps": 1955, "loss": 1.0142, "lr": 2.757432958164953e-06, "epoch": 3.9104859335038364, "percentage": 78.21, "elapsed_time": "7:52:45", "remaining_time": "2:11:43"} +{"current_steps": 1530, "total_steps": 1955, "loss": 1.0436, "lr": 2.745129429237269e-06, "epoch": 3.9130434782608696, "percentage": 78.26, "elapsed_time": "7:53:04", "remaining_time": "2:11:24"} +{"current_steps": 1531, "total_steps": 1955, "loss": 1.0349, "lr": 2.732849042135377e-06, "epoch": 3.915601023017903, "percentage": 78.31, "elapsed_time": "7:53:22", "remaining_time": "2:11:05"} +{"current_steps": 1532, "total_steps": 1955, "loss": 1.0487, "lr": 2.72059183603166e-06, "epoch": 3.918158567774936, "percentage": 78.36, "elapsed_time": "7:53:40", "remaining_time": "2:10:47"} +{"current_steps": 1533, "total_steps": 1955, "loss": 1.0467, "lr": 2.7083578500245567e-06, "epoch": 3.920716112531969, "percentage": 78.41, "elapsed_time": "7:53:59", "remaining_time": "2:10:28"} +{"current_steps": 1534, "total_steps": 1955, "loss": 1.0586, "lr": 2.696147123138442e-06, "epoch": 3.923273657289003, "percentage": 78.47, "elapsed_time": "7:54:17", "remaining_time": "2:10:10"} +{"current_steps": 1535, "total_steps": 1955, "loss": 1.0468, "lr": 2.6839596943234948e-06, "epoch": 3.9258312020460355, "percentage": 78.52, "elapsed_time": "7:54:35", "remaining_time": "2:09:51"} +{"current_steps": 1536, "total_steps": 1955, "loss": 1.0329, "lr": 2.671795602455578e-06, "epoch": 3.928388746803069, "percentage": 78.57, "elapsed_time": "7:54:54", "remaining_time": "2:09:32"} +{"current_steps": 1537, "total_steps": 1955, "loss": 1.0311, "lr": 2.6596548863361117e-06, "epoch": 3.9309462915601023, "percentage": 78.62, "elapsed_time": "7:55:13", "remaining_time": "2:09:14"} +{"current_steps": 1538, "total_steps": 1955, "loss": 1.0497, "lr": 2.647537584691957e-06, "epoch": 3.9335038363171355, "percentage": 78.67, "elapsed_time": "7:55:32", "remaining_time": "2:08:56"} +{"current_steps": 1539, "total_steps": 1955, "loss": 1.0308, "lr": 2.635443736175285e-06, "epoch": 3.9360613810741687, "percentage": 78.72, "elapsed_time": "7:55:50", "remaining_time": "2:08:37"} +{"current_steps": 1540, "total_steps": 1955, "loss": 1.0419, "lr": 2.6233733793634443e-06, "epoch": 3.938618925831202, "percentage": 78.77, "elapsed_time": "7:56:08", "remaining_time": "2:08:18"} +{"current_steps": 1541, "total_steps": 1955, "loss": 1.0326, "lr": 2.611326552758865e-06, "epoch": 3.9411764705882355, "percentage": 78.82, "elapsed_time": "7:56:26", "remaining_time": "2:07:59"} +{"current_steps": 1542, "total_steps": 1955, "loss": 1.0091, "lr": 2.599303294788912e-06, "epoch": 3.9437340153452687, "percentage": 78.87, "elapsed_time": "7:56:45", "remaining_time": "2:07:41"} +{"current_steps": 1543, "total_steps": 1955, "loss": 1.0337, "lr": 2.5873036438057676e-06, "epoch": 3.946291560102302, "percentage": 78.93, "elapsed_time": "7:57:03", "remaining_time": "2:07:22"} +{"current_steps": 1544, "total_steps": 1955, "loss": 1.0152, "lr": 2.5753276380863146e-06, "epoch": 3.948849104859335, "percentage": 78.98, "elapsed_time": "7:57:22", "remaining_time": "2:07:04"} +{"current_steps": 1545, "total_steps": 1955, "loss": 1.0528, "lr": 2.5633753158320187e-06, "epoch": 3.9514066496163682, "percentage": 79.03, "elapsed_time": "7:57:40", "remaining_time": "2:06:45"} +{"current_steps": 1546, "total_steps": 1955, "loss": 1.0234, "lr": 2.551446715168785e-06, "epoch": 3.9539641943734014, "percentage": 79.08, "elapsed_time": "7:57:59", "remaining_time": "2:06:27"} +{"current_steps": 1547, "total_steps": 1955, "loss": 1.0536, "lr": 2.539541874146857e-06, "epoch": 3.9565217391304346, "percentage": 79.13, "elapsed_time": "7:58:18", "remaining_time": "2:06:08"} +{"current_steps": 1548, "total_steps": 1955, "loss": 1.0312, "lr": 2.5276608307406947e-06, "epoch": 3.959079283887468, "percentage": 79.18, "elapsed_time": "7:58:36", "remaining_time": "2:05:50"} +{"current_steps": 1549, "total_steps": 1955, "loss": 1.0957, "lr": 2.5158036228488426e-06, "epoch": 3.9616368286445014, "percentage": 79.23, "elapsed_time": "7:58:54", "remaining_time": "2:05:31"} +{"current_steps": 1550, "total_steps": 1955, "loss": 1.0625, "lr": 2.503970288293811e-06, "epoch": 3.9641943734015346, "percentage": 79.28, "elapsed_time": "7:59:13", "remaining_time": "2:05:12"} +{"current_steps": 1551, "total_steps": 1955, "loss": 1.0518, "lr": 2.492160864821964e-06, "epoch": 3.9667519181585678, "percentage": 79.34, "elapsed_time": "7:59:31", "remaining_time": "2:04:54"} +{"current_steps": 1552, "total_steps": 1955, "loss": 1.0524, "lr": 2.4803753901033887e-06, "epoch": 3.969309462915601, "percentage": 79.39, "elapsed_time": "7:59:49", "remaining_time": "2:04:35"} +{"current_steps": 1553, "total_steps": 1955, "loss": 1.051, "lr": 2.4686139017317834e-06, "epoch": 3.971867007672634, "percentage": 79.44, "elapsed_time": "8:00:07", "remaining_time": "2:04:17"} +{"current_steps": 1554, "total_steps": 1955, "loss": 1.0372, "lr": 2.456876437224327e-06, "epoch": 3.9744245524296673, "percentage": 79.49, "elapsed_time": "8:00:26", "remaining_time": "2:03:58"} +{"current_steps": 1555, "total_steps": 1955, "loss": 1.0601, "lr": 2.4451630340215806e-06, "epoch": 3.976982097186701, "percentage": 79.54, "elapsed_time": "8:00:44", "remaining_time": "2:03:39"} +{"current_steps": 1556, "total_steps": 1955, "loss": 1.0738, "lr": 2.4334737294873412e-06, "epoch": 3.979539641943734, "percentage": 79.59, "elapsed_time": "8:01:03", "remaining_time": "2:03:21"} +{"current_steps": 1557, "total_steps": 1955, "loss": 1.041, "lr": 2.4218085609085317e-06, "epoch": 3.9820971867007673, "percentage": 79.64, "elapsed_time": "8:01:22", "remaining_time": "2:03:02"} +{"current_steps": 1558, "total_steps": 1955, "loss": 1.0552, "lr": 2.4101675654951007e-06, "epoch": 3.9846547314578005, "percentage": 79.69, "elapsed_time": "8:01:40", "remaining_time": "2:02:44"} +{"current_steps": 1559, "total_steps": 1955, "loss": 1.0343, "lr": 2.398550780379877e-06, "epoch": 3.9872122762148337, "percentage": 79.74, "elapsed_time": "8:01:58", "remaining_time": "2:02:25"} +{"current_steps": 1560, "total_steps": 1955, "loss": 1.0494, "lr": 2.3869582426184646e-06, "epoch": 3.9897698209718673, "percentage": 79.8, "elapsed_time": "8:02:16", "remaining_time": "2:02:06"} +{"current_steps": 1561, "total_steps": 1955, "loss": 1.0489, "lr": 2.375389989189124e-06, "epoch": 3.9923273657289, "percentage": 79.85, "elapsed_time": "8:02:35", "remaining_time": "2:01:48"} +{"current_steps": 1562, "total_steps": 1955, "loss": 1.0628, "lr": 2.3638460569926526e-06, "epoch": 3.9948849104859336, "percentage": 79.9, "elapsed_time": "8:02:53", "remaining_time": "2:01:29"} +{"current_steps": 1563, "total_steps": 1955, "loss": 1.0401, "lr": 2.3523264828522664e-06, "epoch": 3.997442455242967, "percentage": 79.95, "elapsed_time": "8:03:12", "remaining_time": "2:01:11"} +{"current_steps": 1564, "total_steps": 1955, "loss": 1.0511, "lr": 2.34083130351348e-06, "epoch": 4.0, "percentage": 80.0, "elapsed_time": "8:03:30", "remaining_time": "2:00:52"} +{"current_steps": 1565, "total_steps": 1955, "loss": 1.0239, "lr": 2.3293605556440036e-06, "epoch": 4.002557544757034, "percentage": 80.05, "elapsed_time": "8:04:00", "remaining_time": "2:00:36"} +{"current_steps": 1566, "total_steps": 1955, "loss": 1.0368, "lr": 2.3179142758336025e-06, "epoch": 4.005115089514066, "percentage": 80.1, "elapsed_time": "8:04:18", "remaining_time": "2:00:18"} +{"current_steps": 1567, "total_steps": 1955, "loss": 1.012, "lr": 2.3064925005939985e-06, "epoch": 4.0076726342711, "percentage": 80.15, "elapsed_time": "8:04:36", "remaining_time": "1:59:59"} +{"current_steps": 1568, "total_steps": 1955, "loss": 1.0293, "lr": 2.29509526635875e-06, "epoch": 4.010230179028133, "percentage": 80.2, "elapsed_time": "8:04:54", "remaining_time": "1:59:40"} +{"current_steps": 1569, "total_steps": 1955, "loss": 1.0415, "lr": 2.2837226094831277e-06, "epoch": 4.012787723785166, "percentage": 80.26, "elapsed_time": "8:05:13", "remaining_time": "1:59:22"} +{"current_steps": 1570, "total_steps": 1955, "loss": 1.0438, "lr": 2.272374566244011e-06, "epoch": 4.015345268542199, "percentage": 80.31, "elapsed_time": "8:05:31", "remaining_time": "1:59:03"} +{"current_steps": 1571, "total_steps": 1955, "loss": 1.0397, "lr": 2.261051172839759e-06, "epoch": 4.017902813299233, "percentage": 80.36, "elapsed_time": "8:05:50", "remaining_time": "1:58:45"} +{"current_steps": 1572, "total_steps": 1955, "loss": 1.0257, "lr": 2.249752465390115e-06, "epoch": 4.020460358056266, "percentage": 80.41, "elapsed_time": "8:06:09", "remaining_time": "1:58:26"} +{"current_steps": 1573, "total_steps": 1955, "loss": 1.0312, "lr": 2.2384784799360593e-06, "epoch": 4.023017902813299, "percentage": 80.46, "elapsed_time": "8:06:27", "remaining_time": "1:58:08"} +{"current_steps": 1574, "total_steps": 1955, "loss": 1.024, "lr": 2.2272292524397256e-06, "epoch": 4.025575447570333, "percentage": 80.51, "elapsed_time": "8:06:46", "remaining_time": "1:57:49"} +{"current_steps": 1575, "total_steps": 1955, "loss": 1.0285, "lr": 2.216004818784274e-06, "epoch": 4.028132992327365, "percentage": 80.56, "elapsed_time": "8:07:05", "remaining_time": "1:57:31"} +{"current_steps": 1576, "total_steps": 1955, "loss": 1.0418, "lr": 2.204805214773774e-06, "epoch": 4.030690537084399, "percentage": 80.61, "elapsed_time": "8:07:23", "remaining_time": "1:57:12"} +{"current_steps": 1577, "total_steps": 1955, "loss": 1.0129, "lr": 2.193630476133087e-06, "epoch": 4.033248081841432, "percentage": 80.66, "elapsed_time": "8:07:41", "remaining_time": "1:56:53"} +{"current_steps": 1578, "total_steps": 1955, "loss": 1.0421, "lr": 2.182480638507775e-06, "epoch": 4.035805626598465, "percentage": 80.72, "elapsed_time": "8:08:00", "remaining_time": "1:56:35"} +{"current_steps": 1579, "total_steps": 1955, "loss": 1.0416, "lr": 2.171355737463946e-06, "epoch": 4.038363171355499, "percentage": 80.77, "elapsed_time": "8:08:18", "remaining_time": "1:56:16"} +{"current_steps": 1580, "total_steps": 1955, "loss": 1.0612, "lr": 2.16025580848818e-06, "epoch": 4.040920716112532, "percentage": 80.82, "elapsed_time": "8:08:36", "remaining_time": "1:55:58"} +{"current_steps": 1581, "total_steps": 1955, "loss": 1.0341, "lr": 2.1491808869874007e-06, "epoch": 4.043478260869565, "percentage": 80.87, "elapsed_time": "8:08:54", "remaining_time": "1:55:39"} +{"current_steps": 1582, "total_steps": 1955, "loss": 1.0201, "lr": 2.1381310082887562e-06, "epoch": 4.046035805626598, "percentage": 80.92, "elapsed_time": "8:09:13", "remaining_time": "1:55:20"} +{"current_steps": 1583, "total_steps": 1955, "loss": 1.0124, "lr": 2.127106207639519e-06, "epoch": 4.048593350383632, "percentage": 80.97, "elapsed_time": "8:09:32", "remaining_time": "1:55:02"} +{"current_steps": 1584, "total_steps": 1955, "loss": 1.0319, "lr": 2.116106520206952e-06, "epoch": 4.051150895140665, "percentage": 81.02, "elapsed_time": "8:09:50", "remaining_time": "1:54:43"} +{"current_steps": 1585, "total_steps": 1955, "loss": 1.0276, "lr": 2.10513198107823e-06, "epoch": 4.053708439897698, "percentage": 81.07, "elapsed_time": "8:10:08", "remaining_time": "1:54:25"} +{"current_steps": 1586, "total_steps": 1955, "loss": 1.0699, "lr": 2.0941826252602993e-06, "epoch": 4.056265984654732, "percentage": 81.13, "elapsed_time": "8:10:27", "remaining_time": "1:54:06"} +{"current_steps": 1587, "total_steps": 1955, "loss": 0.9935, "lr": 2.083258487679772e-06, "epoch": 4.0588235294117645, "percentage": 81.18, "elapsed_time": "8:10:46", "remaining_time": "1:53:48"} +{"current_steps": 1588, "total_steps": 1955, "loss": 1.0199, "lr": 2.07235960318283e-06, "epoch": 4.061381074168798, "percentage": 81.23, "elapsed_time": "8:11:04", "remaining_time": "1:53:29"} +{"current_steps": 1589, "total_steps": 1955, "loss": 1.0648, "lr": 2.061486006535095e-06, "epoch": 4.063938618925831, "percentage": 81.28, "elapsed_time": "8:11:22", "remaining_time": "1:53:10"} +{"current_steps": 1590, "total_steps": 1955, "loss": 1.0271, "lr": 2.0506377324215153e-06, "epoch": 4.0664961636828645, "percentage": 81.33, "elapsed_time": "8:11:40", "remaining_time": "1:52:52"} +{"current_steps": 1591, "total_steps": 1955, "loss": 1.0664, "lr": 2.0398148154462826e-06, "epoch": 4.069053708439898, "percentage": 81.38, "elapsed_time": "8:11:59", "remaining_time": "1:52:33"} +{"current_steps": 1592, "total_steps": 1955, "loss": 1.0561, "lr": 2.029017290132693e-06, "epoch": 4.071611253196931, "percentage": 81.43, "elapsed_time": "8:12:18", "remaining_time": "1:52:15"} +{"current_steps": 1593, "total_steps": 1955, "loss": 1.0647, "lr": 2.0182451909230493e-06, "epoch": 4.0741687979539645, "percentage": 81.48, "elapsed_time": "8:12:36", "remaining_time": "1:51:56"} +{"current_steps": 1594, "total_steps": 1955, "loss": 1.0706, "lr": 2.0074985521785496e-06, "epoch": 4.076726342710997, "percentage": 81.53, "elapsed_time": "8:12:55", "remaining_time": "1:51:38"} +{"current_steps": 1595, "total_steps": 1955, "loss": 1.027, "lr": 1.9967774081791757e-06, "epoch": 4.079283887468031, "percentage": 81.59, "elapsed_time": "8:13:13", "remaining_time": "1:51:19"} +{"current_steps": 1596, "total_steps": 1955, "loss": 1.0564, "lr": 1.9860817931235875e-06, "epoch": 4.081841432225064, "percentage": 81.64, "elapsed_time": "8:13:32", "remaining_time": "1:51:00"} +{"current_steps": 1597, "total_steps": 1955, "loss": 1.04, "lr": 1.9754117411290098e-06, "epoch": 4.084398976982097, "percentage": 81.69, "elapsed_time": "8:13:51", "remaining_time": "1:50:42"} +{"current_steps": 1598, "total_steps": 1955, "loss": 1.0075, "lr": 1.964767286231132e-06, "epoch": 4.086956521739131, "percentage": 81.74, "elapsed_time": "8:14:09", "remaining_time": "1:50:23"} +{"current_steps": 1599, "total_steps": 1955, "loss": 1.0115, "lr": 1.9541484623839836e-06, "epoch": 4.089514066496164, "percentage": 81.79, "elapsed_time": "8:14:28", "remaining_time": "1:50:05"} +{"current_steps": 1600, "total_steps": 1955, "loss": 1.0446, "lr": 1.94355530345984e-06, "epoch": 4.092071611253197, "percentage": 81.84, "elapsed_time": "8:14:46", "remaining_time": "1:49:46"} +{"current_steps": 1601, "total_steps": 1955, "loss": 1.0408, "lr": 1.932987843249111e-06, "epoch": 4.09462915601023, "percentage": 81.89, "elapsed_time": "8:15:04", "remaining_time": "1:49:28"} +{"current_steps": 1602, "total_steps": 1955, "loss": 1.0247, "lr": 1.922446115460229e-06, "epoch": 4.0971867007672635, "percentage": 81.94, "elapsed_time": "8:15:23", "remaining_time": "1:49:09"} +{"current_steps": 1603, "total_steps": 1955, "loss": 1.0107, "lr": 1.9119301537195455e-06, "epoch": 4.099744245524296, "percentage": 81.99, "elapsed_time": "8:15:41", "remaining_time": "1:48:50"} +{"current_steps": 1604, "total_steps": 1955, "loss": 1.0528, "lr": 1.9014399915712213e-06, "epoch": 4.10230179028133, "percentage": 82.05, "elapsed_time": "8:16:00", "remaining_time": "1:48:32"} +{"current_steps": 1605, "total_steps": 1955, "loss": 1.0375, "lr": 1.8909756624771281e-06, "epoch": 4.1048593350383635, "percentage": 82.1, "elapsed_time": "8:16:18", "remaining_time": "1:48:13"} +{"current_steps": 1606, "total_steps": 1955, "loss": 1.0422, "lr": 1.880537199816722e-06, "epoch": 4.107416879795396, "percentage": 82.15, "elapsed_time": "8:16:37", "remaining_time": "1:47:55"} +{"current_steps": 1607, "total_steps": 1955, "loss": 1.0185, "lr": 1.8701246368869563e-06, "epoch": 4.10997442455243, "percentage": 82.2, "elapsed_time": "8:16:55", "remaining_time": "1:47:36"} +{"current_steps": 1608, "total_steps": 1955, "loss": 1.0249, "lr": 1.859738006902172e-06, "epoch": 4.112531969309463, "percentage": 82.25, "elapsed_time": "8:17:14", "remaining_time": "1:47:18"} +{"current_steps": 1609, "total_steps": 1955, "loss": 1.0459, "lr": 1.8493773429939822e-06, "epoch": 4.115089514066496, "percentage": 82.3, "elapsed_time": "8:17:32", "remaining_time": "1:46:59"} +{"current_steps": 1610, "total_steps": 1955, "loss": 1.0397, "lr": 1.8390426782111758e-06, "epoch": 4.117647058823529, "percentage": 82.35, "elapsed_time": "8:17:50", "remaining_time": "1:46:40"} +{"current_steps": 1611, "total_steps": 1955, "loss": 1.0306, "lr": 1.8287340455196068e-06, "epoch": 4.120204603580563, "percentage": 82.4, "elapsed_time": "8:18:09", "remaining_time": "1:46:22"} +{"current_steps": 1612, "total_steps": 1955, "loss": 1.025, "lr": 1.8184514778020935e-06, "epoch": 4.122762148337596, "percentage": 82.46, "elapsed_time": "8:18:28", "remaining_time": "1:46:03"} +{"current_steps": 1613, "total_steps": 1955, "loss": 1.0507, "lr": 1.80819500785831e-06, "epoch": 4.125319693094629, "percentage": 82.51, "elapsed_time": "8:18:46", "remaining_time": "1:45:45"} +{"current_steps": 1614, "total_steps": 1955, "loss": 1.0378, "lr": 1.7979646684046782e-06, "epoch": 4.127877237851663, "percentage": 82.56, "elapsed_time": "8:19:05", "remaining_time": "1:45:26"} +{"current_steps": 1615, "total_steps": 1955, "loss": 1.039, "lr": 1.7877604920742809e-06, "epoch": 4.130434782608695, "percentage": 82.61, "elapsed_time": "8:19:23", "remaining_time": "1:45:08"} +{"current_steps": 1616, "total_steps": 1955, "loss": 1.0381, "lr": 1.7775825114167344e-06, "epoch": 4.132992327365729, "percentage": 82.66, "elapsed_time": "8:19:42", "remaining_time": "1:44:49"} +{"current_steps": 1617, "total_steps": 1955, "loss": 1.0339, "lr": 1.767430758898092e-06, "epoch": 4.135549872122763, "percentage": 82.71, "elapsed_time": "8:20:00", "remaining_time": "1:44:30"} +{"current_steps": 1618, "total_steps": 1955, "loss": 1.0328, "lr": 1.757305266900755e-06, "epoch": 4.138107416879795, "percentage": 82.76, "elapsed_time": "8:20:19", "remaining_time": "1:44:12"} +{"current_steps": 1619, "total_steps": 1955, "loss": 1.0067, "lr": 1.7472060677233505e-06, "epoch": 4.140664961636829, "percentage": 82.81, "elapsed_time": "8:20:36", "remaining_time": "1:43:53"} +{"current_steps": 1620, "total_steps": 1955, "loss": 1.0633, "lr": 1.737133193580638e-06, "epoch": 4.143222506393862, "percentage": 82.86, "elapsed_time": "8:20:55", "remaining_time": "1:43:35"} +{"current_steps": 1621, "total_steps": 1955, "loss": 1.045, "lr": 1.7270866766034012e-06, "epoch": 4.145780051150895, "percentage": 82.92, "elapsed_time": "8:21:13", "remaining_time": "1:43:16"} +{"current_steps": 1622, "total_steps": 1955, "loss": 1.0421, "lr": 1.7170665488383598e-06, "epoch": 4.148337595907928, "percentage": 82.97, "elapsed_time": "8:21:31", "remaining_time": "1:42:57"} +{"current_steps": 1623, "total_steps": 1955, "loss": 1.0576, "lr": 1.7070728422480397e-06, "epoch": 4.150895140664962, "percentage": 83.02, "elapsed_time": "8:21:49", "remaining_time": "1:42:39"} +{"current_steps": 1624, "total_steps": 1955, "loss": 1.0206, "lr": 1.6971055887106979e-06, "epoch": 4.153452685421995, "percentage": 83.07, "elapsed_time": "8:22:08", "remaining_time": "1:42:20"} +{"current_steps": 1625, "total_steps": 1955, "loss": 1.0331, "lr": 1.687164820020213e-06, "epoch": 4.156010230179028, "percentage": 83.12, "elapsed_time": "8:22:26", "remaining_time": "1:42:02"} +{"current_steps": 1626, "total_steps": 1955, "loss": 1.0466, "lr": 1.677250567885974e-06, "epoch": 4.158567774936062, "percentage": 83.17, "elapsed_time": "8:22:45", "remaining_time": "1:41:43"} +{"current_steps": 1627, "total_steps": 1955, "loss": 1.0373, "lr": 1.667362863932792e-06, "epoch": 4.161125319693094, "percentage": 83.22, "elapsed_time": "8:23:03", "remaining_time": "1:41:24"} +{"current_steps": 1628, "total_steps": 1955, "loss": 1.0529, "lr": 1.6575017397007898e-06, "epoch": 4.163682864450128, "percentage": 83.27, "elapsed_time": "8:23:22", "remaining_time": "1:41:06"} +{"current_steps": 1629, "total_steps": 1955, "loss": 1.0568, "lr": 1.6476672266453087e-06, "epoch": 4.166240409207161, "percentage": 83.32, "elapsed_time": "8:23:41", "remaining_time": "1:40:47"} +{"current_steps": 1630, "total_steps": 1955, "loss": 1.0424, "lr": 1.6378593561368017e-06, "epoch": 4.168797953964194, "percentage": 83.38, "elapsed_time": "8:23:59", "remaining_time": "1:40:29"} +{"current_steps": 1631, "total_steps": 1955, "loss": 1.0637, "lr": 1.6280781594607364e-06, "epoch": 4.171355498721228, "percentage": 83.43, "elapsed_time": "8:24:18", "remaining_time": "1:40:10"} +{"current_steps": 1632, "total_steps": 1955, "loss": 1.0407, "lr": 1.6183236678175029e-06, "epoch": 4.173913043478261, "percentage": 83.48, "elapsed_time": "8:24:36", "remaining_time": "1:39:52"} +{"current_steps": 1633, "total_steps": 1955, "loss": 1.036, "lr": 1.6085959123222994e-06, "epoch": 4.176470588235294, "percentage": 83.53, "elapsed_time": "8:24:55", "remaining_time": "1:39:33"} +{"current_steps": 1634, "total_steps": 1955, "loss": 1.0314, "lr": 1.5988949240050344e-06, "epoch": 4.179028132992327, "percentage": 83.58, "elapsed_time": "8:25:13", "remaining_time": "1:39:15"} +{"current_steps": 1635, "total_steps": 1955, "loss": 1.0493, "lr": 1.5892207338102494e-06, "epoch": 4.181585677749361, "percentage": 83.63, "elapsed_time": "8:25:32", "remaining_time": "1:38:56"} +{"current_steps": 1636, "total_steps": 1955, "loss": 1.0836, "lr": 1.5795733725969931e-06, "epoch": 4.1841432225063935, "percentage": 83.68, "elapsed_time": "8:25:50", "remaining_time": "1:38:38"} +{"current_steps": 1637, "total_steps": 1955, "loss": 1.0341, "lr": 1.5699528711387358e-06, "epoch": 4.186700767263427, "percentage": 83.73, "elapsed_time": "8:26:09", "remaining_time": "1:38:19"} +{"current_steps": 1638, "total_steps": 1955, "loss": 1.0502, "lr": 1.560359260123272e-06, "epoch": 4.189258312020461, "percentage": 83.79, "elapsed_time": "8:26:28", "remaining_time": "1:38:00"} +{"current_steps": 1639, "total_steps": 1955, "loss": 1.0323, "lr": 1.5507925701526184e-06, "epoch": 4.1918158567774935, "percentage": 83.84, "elapsed_time": "8:26:46", "remaining_time": "1:37:42"} +{"current_steps": 1640, "total_steps": 1955, "loss": 1.0248, "lr": 1.5412528317429197e-06, "epoch": 4.194373401534527, "percentage": 83.89, "elapsed_time": "8:27:04", "remaining_time": "1:37:23"} +{"current_steps": 1641, "total_steps": 1955, "loss": 1.004, "lr": 1.5317400753243449e-06, "epoch": 4.19693094629156, "percentage": 83.94, "elapsed_time": "8:27:23", "remaining_time": "1:37:05"} +{"current_steps": 1642, "total_steps": 1955, "loss": 1.0532, "lr": 1.5222543312410042e-06, "epoch": 4.1994884910485935, "percentage": 83.99, "elapsed_time": "8:27:41", "remaining_time": "1:36:46"} +{"current_steps": 1643, "total_steps": 1955, "loss": 1.009, "lr": 1.512795629750834e-06, "epoch": 4.202046035805626, "percentage": 84.04, "elapsed_time": "8:28:00", "remaining_time": "1:36:28"} +{"current_steps": 1644, "total_steps": 1955, "loss": 1.0518, "lr": 1.5033640010255146e-06, "epoch": 4.20460358056266, "percentage": 84.09, "elapsed_time": "8:28:18", "remaining_time": "1:36:09"} +{"current_steps": 1645, "total_steps": 1955, "loss": 1.0391, "lr": 1.493959475150365e-06, "epoch": 4.207161125319693, "percentage": 84.14, "elapsed_time": "8:28:36", "remaining_time": "1:35:50"} +{"current_steps": 1646, "total_steps": 1955, "loss": 1.0183, "lr": 1.484582082124254e-06, "epoch": 4.209718670076726, "percentage": 84.19, "elapsed_time": "8:28:55", "remaining_time": "1:35:32"} +{"current_steps": 1647, "total_steps": 1955, "loss": 1.0214, "lr": 1.4752318518594988e-06, "epoch": 4.21227621483376, "percentage": 84.25, "elapsed_time": "8:29:13", "remaining_time": "1:35:13"} +{"current_steps": 1648, "total_steps": 1955, "loss": 1.0326, "lr": 1.4659088141817701e-06, "epoch": 4.2148337595907925, "percentage": 84.3, "elapsed_time": "8:29:32", "remaining_time": "1:34:55"} +{"current_steps": 1649, "total_steps": 1955, "loss": 1.0167, "lr": 1.4566129988300094e-06, "epoch": 4.217391304347826, "percentage": 84.35, "elapsed_time": "8:29:51", "remaining_time": "1:34:36"} +{"current_steps": 1650, "total_steps": 1955, "loss": 1.0201, "lr": 1.4473444354563081e-06, "epoch": 4.21994884910486, "percentage": 84.4, "elapsed_time": "8:30:09", "remaining_time": "1:34:18"} +{"current_steps": 1651, "total_steps": 1955, "loss": 1.0403, "lr": 1.438103153625835e-06, "epoch": 4.2225063938618925, "percentage": 84.45, "elapsed_time": "8:30:28", "remaining_time": "1:33:59"} +{"current_steps": 1652, "total_steps": 1955, "loss": 1.0296, "lr": 1.4288891828167428e-06, "epoch": 4.225063938618926, "percentage": 84.5, "elapsed_time": "8:30:47", "remaining_time": "1:33:41"} +{"current_steps": 1653, "total_steps": 1955, "loss": 1.0436, "lr": 1.4197025524200547e-06, "epoch": 4.227621483375959, "percentage": 84.55, "elapsed_time": "8:31:05", "remaining_time": "1:33:22"} +{"current_steps": 1654, "total_steps": 1955, "loss": 1.036, "lr": 1.4105432917395912e-06, "epoch": 4.2301790281329925, "percentage": 84.6, "elapsed_time": "8:31:23", "remaining_time": "1:33:03"} +{"current_steps": 1655, "total_steps": 1955, "loss": 1.0626, "lr": 1.4014114299918613e-06, "epoch": 4.232736572890025, "percentage": 84.65, "elapsed_time": "8:31:42", "remaining_time": "1:32:45"} +{"current_steps": 1656, "total_steps": 1955, "loss": 1.037, "lr": 1.3923069963059821e-06, "epoch": 4.235294117647059, "percentage": 84.71, "elapsed_time": "8:32:01", "remaining_time": "1:32:26"} +{"current_steps": 1657, "total_steps": 1955, "loss": 1.0224, "lr": 1.3832300197235748e-06, "epoch": 4.2378516624040925, "percentage": 84.76, "elapsed_time": "8:32:19", "remaining_time": "1:32:08"} +{"current_steps": 1658, "total_steps": 1955, "loss": 1.0588, "lr": 1.3741805291986787e-06, "epoch": 4.240409207161125, "percentage": 84.81, "elapsed_time": "8:32:37", "remaining_time": "1:31:49"} +{"current_steps": 1659, "total_steps": 1955, "loss": 1.0416, "lr": 1.3651585535976596e-06, "epoch": 4.242966751918159, "percentage": 84.86, "elapsed_time": "8:32:56", "remaining_time": "1:31:31"} +{"current_steps": 1660, "total_steps": 1955, "loss": 1.0464, "lr": 1.3561641216991162e-06, "epoch": 4.245524296675192, "percentage": 84.91, "elapsed_time": "8:33:14", "remaining_time": "1:31:12"} +{"current_steps": 1661, "total_steps": 1955, "loss": 1.0523, "lr": 1.3471972621937757e-06, "epoch": 4.248081841432225, "percentage": 84.96, "elapsed_time": "8:33:33", "remaining_time": "1:30:54"} +{"current_steps": 1662, "total_steps": 1955, "loss": 1.0461, "lr": 1.3382580036844295e-06, "epoch": 4.250639386189258, "percentage": 85.01, "elapsed_time": "8:33:52", "remaining_time": "1:30:35"} +{"current_steps": 1663, "total_steps": 1955, "loss": 1.027, "lr": 1.3293463746858182e-06, "epoch": 4.253196930946292, "percentage": 85.06, "elapsed_time": "8:34:10", "remaining_time": "1:30:16"} +{"current_steps": 1664, "total_steps": 1955, "loss": 1.0253, "lr": 1.3204624036245506e-06, "epoch": 4.255754475703325, "percentage": 85.12, "elapsed_time": "8:34:29", "remaining_time": "1:29:58"} +{"current_steps": 1665, "total_steps": 1955, "loss": 1.0299, "lr": 1.3116061188390084e-06, "epoch": 4.258312020460358, "percentage": 85.17, "elapsed_time": "8:34:47", "remaining_time": "1:29:39"} +{"current_steps": 1666, "total_steps": 1955, "loss": 1.0309, "lr": 1.3027775485792682e-06, "epoch": 4.260869565217392, "percentage": 85.22, "elapsed_time": "8:35:06", "remaining_time": "1:29:21"} +{"current_steps": 1667, "total_steps": 1955, "loss": 1.0433, "lr": 1.2939767210069876e-06, "epoch": 4.263427109974424, "percentage": 85.27, "elapsed_time": "8:35:25", "remaining_time": "1:29:02"} +{"current_steps": 1668, "total_steps": 1955, "loss": 0.9889, "lr": 1.285203664195338e-06, "epoch": 4.265984654731458, "percentage": 85.32, "elapsed_time": "8:35:43", "remaining_time": "1:28:44"} +{"current_steps": 1669, "total_steps": 1955, "loss": 1.0339, "lr": 1.2764584061289098e-06, "epoch": 4.268542199488491, "percentage": 85.37, "elapsed_time": "8:36:01", "remaining_time": "1:28:25"} +{"current_steps": 1670, "total_steps": 1955, "loss": 1.0461, "lr": 1.267740974703614e-06, "epoch": 4.271099744245524, "percentage": 85.42, "elapsed_time": "8:36:20", "remaining_time": "1:28:07"} +{"current_steps": 1671, "total_steps": 1955, "loss": 1.0214, "lr": 1.2590513977266007e-06, "epoch": 4.273657289002558, "percentage": 85.47, "elapsed_time": "8:36:39", "remaining_time": "1:27:48"} +{"current_steps": 1672, "total_steps": 1955, "loss": 1.0329, "lr": 1.2503897029161716e-06, "epoch": 4.276214833759591, "percentage": 85.52, "elapsed_time": "8:36:57", "remaining_time": "1:27:29"} +{"current_steps": 1673, "total_steps": 1955, "loss": 0.9865, "lr": 1.2417559179016835e-06, "epoch": 4.278772378516624, "percentage": 85.58, "elapsed_time": "8:37:15", "remaining_time": "1:27:11"} +{"current_steps": 1674, "total_steps": 1955, "loss": 1.0293, "lr": 1.2331500702234723e-06, "epoch": 4.281329923273657, "percentage": 85.63, "elapsed_time": "8:37:34", "remaining_time": "1:26:52"} +{"current_steps": 1675, "total_steps": 1955, "loss": 1.0458, "lr": 1.2245721873327521e-06, "epoch": 4.283887468030691, "percentage": 85.68, "elapsed_time": "8:37:53", "remaining_time": "1:26:34"} +{"current_steps": 1676, "total_steps": 1955, "loss": 1.0281, "lr": 1.2160222965915404e-06, "epoch": 4.286445012787723, "percentage": 85.73, "elapsed_time": "8:38:11", "remaining_time": "1:26:15"} +{"current_steps": 1677, "total_steps": 1955, "loss": 1.0467, "lr": 1.207500425272562e-06, "epoch": 4.289002557544757, "percentage": 85.78, "elapsed_time": "8:38:30", "remaining_time": "1:25:57"} +{"current_steps": 1678, "total_steps": 1955, "loss": 1.0456, "lr": 1.199006600559156e-06, "epoch": 4.291560102301791, "percentage": 85.83, "elapsed_time": "8:38:48", "remaining_time": "1:25:38"} +{"current_steps": 1679, "total_steps": 1955, "loss": 1.0326, "lr": 1.190540849545213e-06, "epoch": 4.294117647058823, "percentage": 85.88, "elapsed_time": "8:39:07", "remaining_time": "1:25:20"} +{"current_steps": 1680, "total_steps": 1955, "loss": 1.0506, "lr": 1.1821031992350628e-06, "epoch": 4.296675191815857, "percentage": 85.93, "elapsed_time": "8:39:25", "remaining_time": "1:25:01"} +{"current_steps": 1681, "total_steps": 1955, "loss": 1.0218, "lr": 1.1736936765434005e-06, "epoch": 4.29923273657289, "percentage": 85.98, "elapsed_time": "8:39:44", "remaining_time": "1:24:42"} +{"current_steps": 1682, "total_steps": 1955, "loss": 1.0717, "lr": 1.1653123082951968e-06, "epoch": 4.301790281329923, "percentage": 86.04, "elapsed_time": "8:40:02", "remaining_time": "1:24:24"} +{"current_steps": 1683, "total_steps": 1955, "loss": 1.0487, "lr": 1.1569591212256237e-06, "epoch": 4.304347826086957, "percentage": 86.09, "elapsed_time": "8:40:21", "remaining_time": "1:24:05"} +{"current_steps": 1684, "total_steps": 1955, "loss": 1.0735, "lr": 1.1486341419799474e-06, "epoch": 4.30690537084399, "percentage": 86.14, "elapsed_time": "8:40:39", "remaining_time": "1:23:47"} +{"current_steps": 1685, "total_steps": 1955, "loss": 1.0211, "lr": 1.1403373971134623e-06, "epoch": 4.309462915601023, "percentage": 86.19, "elapsed_time": "8:40:58", "remaining_time": "1:23:28"} +{"current_steps": 1686, "total_steps": 1955, "loss": 1.0547, "lr": 1.132068913091402e-06, "epoch": 4.312020460358056, "percentage": 86.24, "elapsed_time": "8:41:16", "remaining_time": "1:23:10"} +{"current_steps": 1687, "total_steps": 1955, "loss": 1.0141, "lr": 1.1238287162888483e-06, "epoch": 4.31457800511509, "percentage": 86.29, "elapsed_time": "8:41:35", "remaining_time": "1:22:51"} +{"current_steps": 1688, "total_steps": 1955, "loss": 1.0115, "lr": 1.1156168329906536e-06, "epoch": 4.3171355498721224, "percentage": 86.34, "elapsed_time": "8:41:54", "remaining_time": "1:22:33"} +{"current_steps": 1689, "total_steps": 1955, "loss": 1.0183, "lr": 1.1074332893913541e-06, "epoch": 4.319693094629156, "percentage": 86.39, "elapsed_time": "8:42:12", "remaining_time": "1:22:14"} +{"current_steps": 1690, "total_steps": 1955, "loss": 1.0817, "lr": 1.0992781115950867e-06, "epoch": 4.322250639386189, "percentage": 86.45, "elapsed_time": "8:42:30", "remaining_time": "1:21:55"} +{"current_steps": 1691, "total_steps": 1955, "loss": 1.0305, "lr": 1.0911513256155094e-06, "epoch": 4.324808184143222, "percentage": 86.5, "elapsed_time": "8:42:49", "remaining_time": "1:21:37"} +{"current_steps": 1692, "total_steps": 1955, "loss": 1.0271, "lr": 1.0830529573757076e-06, "epoch": 4.327365728900256, "percentage": 86.55, "elapsed_time": "8:43:07", "remaining_time": "1:21:18"} +{"current_steps": 1693, "total_steps": 1955, "loss": 1.0426, "lr": 1.074983032708129e-06, "epoch": 4.329923273657289, "percentage": 86.6, "elapsed_time": "8:43:26", "remaining_time": "1:21:00"} +{"current_steps": 1694, "total_steps": 1955, "loss": 1.0339, "lr": 1.0669415773544867e-06, "epoch": 4.332480818414322, "percentage": 86.65, "elapsed_time": "8:43:45", "remaining_time": "1:20:41"} +{"current_steps": 1695, "total_steps": 1955, "loss": 1.021, "lr": 1.0589286169656743e-06, "epoch": 4.335038363171355, "percentage": 86.7, "elapsed_time": "8:44:03", "remaining_time": "1:20:23"} +{"current_steps": 1696, "total_steps": 1955, "loss": 1.063, "lr": 1.0509441771017026e-06, "epoch": 4.337595907928389, "percentage": 86.75, "elapsed_time": "8:44:22", "remaining_time": "1:20:04"} +{"current_steps": 1697, "total_steps": 1955, "loss": 1.0279, "lr": 1.0429882832316008e-06, "epoch": 4.340153452685422, "percentage": 86.8, "elapsed_time": "8:44:40", "remaining_time": "1:19:46"} +{"current_steps": 1698, "total_steps": 1955, "loss": 1.0053, "lr": 1.0350609607333384e-06, "epoch": 4.342710997442455, "percentage": 86.85, "elapsed_time": "8:44:59", "remaining_time": "1:19:27"} +{"current_steps": 1699, "total_steps": 1955, "loss": 1.0438, "lr": 1.0271622348937581e-06, "epoch": 4.345268542199489, "percentage": 86.91, "elapsed_time": "8:45:17", "remaining_time": "1:19:09"} +{"current_steps": 1700, "total_steps": 1955, "loss": 1.0407, "lr": 1.0192921309084703e-06, "epoch": 4.3478260869565215, "percentage": 86.96, "elapsed_time": "8:45:36", "remaining_time": "1:18:50"} +{"current_steps": 1701, "total_steps": 1955, "loss": 1.0031, "lr": 1.0114506738817942e-06, "epoch": 4.350383631713555, "percentage": 87.01, "elapsed_time": "8:45:55", "remaining_time": "1:18:31"} +{"current_steps": 1702, "total_steps": 1955, "loss": 1.0547, "lr": 1.0036378888266663e-06, "epoch": 4.352941176470588, "percentage": 87.06, "elapsed_time": "8:46:13", "remaining_time": "1:18:13"} +{"current_steps": 1703, "total_steps": 1955, "loss": 1.0687, "lr": 9.958538006645702e-07, "epoch": 4.3554987212276215, "percentage": 87.11, "elapsed_time": "8:46:32", "remaining_time": "1:17:54"} +{"current_steps": 1704, "total_steps": 1955, "loss": 1.021, "lr": 9.880984342254462e-07, "epoch": 4.358056265984655, "percentage": 87.16, "elapsed_time": "8:46:51", "remaining_time": "1:17:36"} +{"current_steps": 1705, "total_steps": 1955, "loss": 1.0392, "lr": 9.803718142476182e-07, "epoch": 4.360613810741688, "percentage": 87.21, "elapsed_time": "8:47:09", "remaining_time": "1:17:17"} +{"current_steps": 1706, "total_steps": 1955, "loss": 1.0513, "lr": 9.72673965377714e-07, "epoch": 4.3631713554987215, "percentage": 87.26, "elapsed_time": "8:47:28", "remaining_time": "1:16:59"} +{"current_steps": 1707, "total_steps": 1955, "loss": 1.0409, "lr": 9.650049121705851e-07, "epoch": 4.365728900255754, "percentage": 87.31, "elapsed_time": "8:47:47", "remaining_time": "1:16:40"} +{"current_steps": 1708, "total_steps": 1955, "loss": 1.027, "lr": 9.573646790892299e-07, "epoch": 4.368286445012788, "percentage": 87.37, "elapsed_time": "8:48:05", "remaining_time": "1:16:22"} +{"current_steps": 1709, "total_steps": 1955, "loss": 1.0534, "lr": 9.497532905047202e-07, "epoch": 4.370843989769821, "percentage": 87.42, "elapsed_time": "8:48:23", "remaining_time": "1:16:03"} +{"current_steps": 1710, "total_steps": 1955, "loss": 1.0401, "lr": 9.421707706961136e-07, "epoch": 4.373401534526854, "percentage": 87.47, "elapsed_time": "8:48:41", "remaining_time": "1:15:44"} +{"current_steps": 1711, "total_steps": 1955, "loss": 1.0642, "lr": 9.346171438503781e-07, "epoch": 4.375959079283888, "percentage": 87.52, "elapsed_time": "8:49:00", "remaining_time": "1:15:26"} +{"current_steps": 1712, "total_steps": 1955, "loss": 1.0648, "lr": 9.270924340623267e-07, "epoch": 4.378516624040921, "percentage": 87.57, "elapsed_time": "8:49:19", "remaining_time": "1:15:07"} +{"current_steps": 1713, "total_steps": 1955, "loss": 1.0287, "lr": 9.195966653345256e-07, "epoch": 4.381074168797954, "percentage": 87.62, "elapsed_time": "8:49:37", "remaining_time": "1:14:49"} +{"current_steps": 1714, "total_steps": 1955, "loss": 1.0504, "lr": 9.121298615772256e-07, "epoch": 4.383631713554987, "percentage": 87.67, "elapsed_time": "8:49:56", "remaining_time": "1:14:30"} +{"current_steps": 1715, "total_steps": 1955, "loss": 1.0471, "lr": 9.046920466082809e-07, "epoch": 4.3861892583120206, "percentage": 87.72, "elapsed_time": "8:50:15", "remaining_time": "1:14:12"} +{"current_steps": 1716, "total_steps": 1955, "loss": 1.0341, "lr": 8.972832441530876e-07, "epoch": 4.388746803069053, "percentage": 87.77, "elapsed_time": "8:50:33", "remaining_time": "1:13:53"} +{"current_steps": 1717, "total_steps": 1955, "loss": 1.0735, "lr": 8.899034778444804e-07, "epoch": 4.391304347826087, "percentage": 87.83, "elapsed_time": "8:50:52", "remaining_time": "1:13:35"} +{"current_steps": 1718, "total_steps": 1955, "loss": 1.0145, "lr": 8.825527712226834e-07, "epoch": 4.3938618925831205, "percentage": 87.88, "elapsed_time": "8:51:10", "remaining_time": "1:13:16"} +{"current_steps": 1719, "total_steps": 1955, "loss": 1.0255, "lr": 8.75231147735226e-07, "epoch": 4.396419437340153, "percentage": 87.93, "elapsed_time": "8:51:28", "remaining_time": "1:12:58"} +{"current_steps": 1720, "total_steps": 1955, "loss": 1.0486, "lr": 8.679386307368631e-07, "epoch": 4.398976982097187, "percentage": 87.98, "elapsed_time": "8:51:47", "remaining_time": "1:12:39"} +{"current_steps": 1721, "total_steps": 1955, "loss": 1.0663, "lr": 8.60675243489506e-07, "epoch": 4.40153452685422, "percentage": 88.03, "elapsed_time": "8:52:05", "remaining_time": "1:12:20"} +{"current_steps": 1722, "total_steps": 1955, "loss": 1.0517, "lr": 8.534410091621481e-07, "epoch": 4.404092071611253, "percentage": 88.08, "elapsed_time": "8:52:23", "remaining_time": "1:12:02"} +{"current_steps": 1723, "total_steps": 1955, "loss": 1.0563, "lr": 8.462359508307882e-07, "epoch": 4.406649616368286, "percentage": 88.13, "elapsed_time": "8:52:42", "remaining_time": "1:11:43"} +{"current_steps": 1724, "total_steps": 1955, "loss": 1.0228, "lr": 8.390600914783597e-07, "epoch": 4.40920716112532, "percentage": 88.18, "elapsed_time": "8:53:00", "remaining_time": "1:11:25"} +{"current_steps": 1725, "total_steps": 1955, "loss": 1.0391, "lr": 8.319134539946549e-07, "epoch": 4.411764705882353, "percentage": 88.24, "elapsed_time": "8:53:19", "remaining_time": "1:11:06"} +{"current_steps": 1726, "total_steps": 1955, "loss": 1.0388, "lr": 8.247960611762563e-07, "epoch": 4.414322250639386, "percentage": 88.29, "elapsed_time": "8:53:37", "remaining_time": "1:10:47"} +{"current_steps": 1727, "total_steps": 1955, "loss": 1.0282, "lr": 8.177079357264583e-07, "epoch": 4.41687979539642, "percentage": 88.34, "elapsed_time": "8:53:55", "remaining_time": "1:10:29"} +{"current_steps": 1728, "total_steps": 1955, "loss": 1.0262, "lr": 8.106491002551941e-07, "epoch": 4.419437340153452, "percentage": 88.39, "elapsed_time": "8:54:14", "remaining_time": "1:10:10"} +{"current_steps": 1729, "total_steps": 1955, "loss": 1.0334, "lr": 8.036195772789735e-07, "epoch": 4.421994884910486, "percentage": 88.44, "elapsed_time": "8:54:33", "remaining_time": "1:09:52"} +{"current_steps": 1730, "total_steps": 1955, "loss": 1.0738, "lr": 7.966193892208007e-07, "epoch": 4.42455242966752, "percentage": 88.49, "elapsed_time": "8:54:52", "remaining_time": "1:09:33"} +{"current_steps": 1731, "total_steps": 1955, "loss": 1.0214, "lr": 7.896485584101066e-07, "epoch": 4.427109974424552, "percentage": 88.54, "elapsed_time": "8:55:10", "remaining_time": "1:09:15"} +{"current_steps": 1732, "total_steps": 1955, "loss": 1.0101, "lr": 7.827071070826775e-07, "epoch": 4.429667519181586, "percentage": 88.59, "elapsed_time": "8:55:29", "remaining_time": "1:08:56"} +{"current_steps": 1733, "total_steps": 1955, "loss": 1.0448, "lr": 7.757950573805839e-07, "epoch": 4.432225063938619, "percentage": 88.64, "elapsed_time": "8:55:47", "remaining_time": "1:08:38"} +{"current_steps": 1734, "total_steps": 1955, "loss": 1.028, "lr": 7.689124313521112e-07, "epoch": 4.434782608695652, "percentage": 88.7, "elapsed_time": "8:56:05", "remaining_time": "1:08:19"} +{"current_steps": 1735, "total_steps": 1955, "loss": 1.0508, "lr": 7.620592509516845e-07, "epoch": 4.437340153452685, "percentage": 88.75, "elapsed_time": "8:56:24", "remaining_time": "1:08:01"} +{"current_steps": 1736, "total_steps": 1955, "loss": 1.0709, "lr": 7.552355380398079e-07, "epoch": 4.439897698209719, "percentage": 88.8, "elapsed_time": "8:56:42", "remaining_time": "1:07:42"} +{"current_steps": 1737, "total_steps": 1955, "loss": 1.0328, "lr": 7.484413143829861e-07, "epoch": 4.442455242966752, "percentage": 88.85, "elapsed_time": "8:57:01", "remaining_time": "1:07:23"} +{"current_steps": 1738, "total_steps": 1955, "loss": 1.034, "lr": 7.41676601653657e-07, "epoch": 4.445012787723785, "percentage": 88.9, "elapsed_time": "8:57:19", "remaining_time": "1:07:05"} +{"current_steps": 1739, "total_steps": 1955, "loss": 1.0525, "lr": 7.349414214301242e-07, "epoch": 4.447570332480819, "percentage": 88.95, "elapsed_time": "8:57:37", "remaining_time": "1:06:46"} +{"current_steps": 1740, "total_steps": 1955, "loss": 1.006, "lr": 7.282357951964903e-07, "epoch": 4.450127877237851, "percentage": 89.0, "elapsed_time": "8:57:56", "remaining_time": "1:06:28"} +{"current_steps": 1741, "total_steps": 1955, "loss": 1.0249, "lr": 7.215597443425815e-07, "epoch": 4.452685421994885, "percentage": 89.05, "elapsed_time": "8:58:15", "remaining_time": "1:06:09"} +{"current_steps": 1742, "total_steps": 1955, "loss": 1.0712, "lr": 7.149132901638844e-07, "epoch": 4.455242966751918, "percentage": 89.1, "elapsed_time": "8:58:33", "remaining_time": "1:05:51"} +{"current_steps": 1743, "total_steps": 1955, "loss": 1.0242, "lr": 7.082964538614823e-07, "epoch": 4.457800511508951, "percentage": 89.16, "elapsed_time": "8:58:51", "remaining_time": "1:05:32"} +{"current_steps": 1744, "total_steps": 1955, "loss": 1.0311, "lr": 7.017092565419748e-07, "epoch": 4.460358056265985, "percentage": 89.21, "elapsed_time": "8:59:10", "remaining_time": "1:05:13"} +{"current_steps": 1745, "total_steps": 1955, "loss": 1.041, "lr": 6.951517192174195e-07, "epoch": 4.462915601023018, "percentage": 89.26, "elapsed_time": "8:59:28", "remaining_time": "1:04:55"} +{"current_steps": 1746, "total_steps": 1955, "loss": 1.0462, "lr": 6.886238628052711e-07, "epoch": 4.465473145780051, "percentage": 89.31, "elapsed_time": "8:59:46", "remaining_time": "1:04:36"} +{"current_steps": 1747, "total_steps": 1955, "loss": 1.0311, "lr": 6.821257081282973e-07, "epoch": 4.468030690537084, "percentage": 89.36, "elapsed_time": "9:00:05", "remaining_time": "1:04:18"} +{"current_steps": 1748, "total_steps": 1955, "loss": 1.0362, "lr": 6.756572759145285e-07, "epoch": 4.470588235294118, "percentage": 89.41, "elapsed_time": "9:00:24", "remaining_time": "1:03:59"} +{"current_steps": 1749, "total_steps": 1955, "loss": 1.0398, "lr": 6.692185867971834e-07, "epoch": 4.4731457800511505, "percentage": 89.46, "elapsed_time": "9:00:42", "remaining_time": "1:03:41"} +{"current_steps": 1750, "total_steps": 1955, "loss": 1.0647, "lr": 6.62809661314604e-07, "epoch": 4.475703324808184, "percentage": 89.51, "elapsed_time": "9:01:01", "remaining_time": "1:03:22"} +{"current_steps": 1751, "total_steps": 1955, "loss": 1.0287, "lr": 6.564305199101939e-07, "epoch": 4.478260869565218, "percentage": 89.57, "elapsed_time": "9:01:19", "remaining_time": "1:03:04"} +{"current_steps": 1752, "total_steps": 1955, "loss": 1.0455, "lr": 6.500811829323462e-07, "epoch": 4.4808184143222505, "percentage": 89.62, "elapsed_time": "9:01:38", "remaining_time": "1:02:45"} +{"current_steps": 1753, "total_steps": 1955, "loss": 1.0569, "lr": 6.437616706343896e-07, "epoch": 4.483375959079284, "percentage": 89.67, "elapsed_time": "9:01:57", "remaining_time": "1:02:26"} +{"current_steps": 1754, "total_steps": 1955, "loss": 1.058, "lr": 6.374720031745129e-07, "epoch": 4.485933503836317, "percentage": 89.72, "elapsed_time": "9:02:15", "remaining_time": "1:02:08"} +{"current_steps": 1755, "total_steps": 1955, "loss": 1.0319, "lr": 6.312122006157006e-07, "epoch": 4.4884910485933505, "percentage": 89.77, "elapsed_time": "9:02:34", "remaining_time": "1:01:49"} +{"current_steps": 1756, "total_steps": 1955, "loss": 1.0455, "lr": 6.249822829256835e-07, "epoch": 4.491048593350383, "percentage": 89.82, "elapsed_time": "9:02:52", "remaining_time": "1:01:31"} +{"current_steps": 1757, "total_steps": 1955, "loss": 1.0464, "lr": 6.187822699768553e-07, "epoch": 4.493606138107417, "percentage": 89.87, "elapsed_time": "9:03:10", "remaining_time": "1:01:12"} +{"current_steps": 1758, "total_steps": 1955, "loss": 1.0205, "lr": 6.126121815462228e-07, "epoch": 4.4961636828644505, "percentage": 89.92, "elapsed_time": "9:03:28", "remaining_time": "1:00:54"} +{"current_steps": 1759, "total_steps": 1955, "loss": 1.0062, "lr": 6.064720373153365e-07, "epoch": 4.498721227621483, "percentage": 89.97, "elapsed_time": "9:03:47", "remaining_time": "1:00:35"} +{"current_steps": 1760, "total_steps": 1955, "loss": 1.0587, "lr": 6.003618568702352e-07, "epoch": 4.501278772378517, "percentage": 90.03, "elapsed_time": "9:04:05", "remaining_time": "1:00:16"} +{"current_steps": 1761, "total_steps": 1955, "loss": 1.0516, "lr": 5.942816597013712e-07, "epoch": 4.5038363171355495, "percentage": 90.08, "elapsed_time": "9:04:23", "remaining_time": "0:59:58"} +{"current_steps": 1762, "total_steps": 1955, "loss": 1.0645, "lr": 5.882314652035582e-07, "epoch": 4.506393861892583, "percentage": 90.13, "elapsed_time": "9:04:42", "remaining_time": "0:59:39"} +{"current_steps": 1763, "total_steps": 1955, "loss": 1.0491, "lr": 5.822112926759072e-07, "epoch": 4.508951406649617, "percentage": 90.18, "elapsed_time": "9:05:00", "remaining_time": "0:59:21"} +{"current_steps": 1764, "total_steps": 1955, "loss": 1.0235, "lr": 5.762211613217649e-07, "epoch": 4.5115089514066495, "percentage": 90.23, "elapsed_time": "9:05:19", "remaining_time": "0:59:02"} +{"current_steps": 1765, "total_steps": 1955, "loss": 1.0396, "lr": 5.702610902486472e-07, "epoch": 4.514066496163683, "percentage": 90.28, "elapsed_time": "9:05:37", "remaining_time": "0:58:44"} +{"current_steps": 1766, "total_steps": 1955, "loss": 1.0253, "lr": 5.643310984681882e-07, "epoch": 4.516624040920716, "percentage": 90.33, "elapsed_time": "9:05:55", "remaining_time": "0:58:25"} +{"current_steps": 1767, "total_steps": 1955, "loss": 1.0563, "lr": 5.584312048960705e-07, "epoch": 4.5191815856777495, "percentage": 90.38, "elapsed_time": "9:06:14", "remaining_time": "0:58:07"} +{"current_steps": 1768, "total_steps": 1955, "loss": 1.054, "lr": 5.525614283519698e-07, "epoch": 4.521739130434782, "percentage": 90.43, "elapsed_time": "9:06:33", "remaining_time": "0:57:48"} +{"current_steps": 1769, "total_steps": 1955, "loss": 1.0103, "lr": 5.46721787559491e-07, "epoch": 4.524296675191816, "percentage": 90.49, "elapsed_time": "9:06:51", "remaining_time": "0:57:29"} +{"current_steps": 1770, "total_steps": 1955, "loss": 1.0371, "lr": 5.40912301146116e-07, "epoch": 4.526854219948849, "percentage": 90.54, "elapsed_time": "9:07:10", "remaining_time": "0:57:11"} +{"current_steps": 1771, "total_steps": 1955, "loss": 1.031, "lr": 5.351329876431377e-07, "epoch": 4.529411764705882, "percentage": 90.59, "elapsed_time": "9:07:29", "remaining_time": "0:56:52"} +{"current_steps": 1772, "total_steps": 1955, "loss": 1.0368, "lr": 5.293838654855965e-07, "epoch": 4.531969309462916, "percentage": 90.64, "elapsed_time": "9:07:47", "remaining_time": "0:56:34"} +{"current_steps": 1773, "total_steps": 1955, "loss": 1.0234, "lr": 5.236649530122362e-07, "epoch": 4.534526854219949, "percentage": 90.69, "elapsed_time": "9:08:06", "remaining_time": "0:56:15"} +{"current_steps": 1774, "total_steps": 1955, "loss": 1.0268, "lr": 5.17976268465431e-07, "epoch": 4.537084398976982, "percentage": 90.74, "elapsed_time": "9:08:24", "remaining_time": "0:55:57"} +{"current_steps": 1775, "total_steps": 1955, "loss": 1.0161, "lr": 5.123178299911357e-07, "epoch": 4.539641943734015, "percentage": 90.79, "elapsed_time": "9:08:43", "remaining_time": "0:55:38"} +{"current_steps": 1776, "total_steps": 1955, "loss": 1.0263, "lr": 5.066896556388223e-07, "epoch": 4.542199488491049, "percentage": 90.84, "elapsed_time": "9:09:01", "remaining_time": "0:55:20"} +{"current_steps": 1777, "total_steps": 1955, "loss": 1.0433, "lr": 5.010917633614298e-07, "epoch": 4.544757033248082, "percentage": 90.9, "elapsed_time": "9:09:20", "remaining_time": "0:55:01"} +{"current_steps": 1778, "total_steps": 1955, "loss": 1.0548, "lr": 4.95524171015298e-07, "epoch": 4.547314578005115, "percentage": 90.95, "elapsed_time": "9:09:38", "remaining_time": "0:54:43"} +{"current_steps": 1779, "total_steps": 1955, "loss": 1.0387, "lr": 4.899868963601173e-07, "epoch": 4.549872122762149, "percentage": 91.0, "elapsed_time": "9:09:57", "remaining_time": "0:54:24"} +{"current_steps": 1780, "total_steps": 1955, "loss": 1.031, "lr": 4.844799570588699e-07, "epoch": 4.552429667519181, "percentage": 91.05, "elapsed_time": "9:10:16", "remaining_time": "0:54:05"} +{"current_steps": 1781, "total_steps": 1955, "loss": 1.0628, "lr": 4.79003370677773e-07, "epoch": 4.554987212276215, "percentage": 91.1, "elapsed_time": "9:10:34", "remaining_time": "0:53:47"} +{"current_steps": 1782, "total_steps": 1955, "loss": 1.0366, "lr": 4.7355715468622164e-07, "epoch": 4.557544757033249, "percentage": 91.15, "elapsed_time": "9:10:53", "remaining_time": "0:53:28"} +{"current_steps": 1783, "total_steps": 1955, "loss": 1.0908, "lr": 4.6814132645673584e-07, "epoch": 4.560102301790281, "percentage": 91.2, "elapsed_time": "9:11:11", "remaining_time": "0:53:10"} +{"current_steps": 1784, "total_steps": 1955, "loss": 1.0109, "lr": 4.6275590326490316e-07, "epoch": 4.562659846547315, "percentage": 91.25, "elapsed_time": "9:11:30", "remaining_time": "0:52:51"} +{"current_steps": 1785, "total_steps": 1955, "loss": 1.0137, "lr": 4.5740090228932554e-07, "epoch": 4.565217391304348, "percentage": 91.3, "elapsed_time": "9:11:49", "remaining_time": "0:52:33"} +{"current_steps": 1786, "total_steps": 1955, "loss": 1.0203, "lr": 4.5207634061155913e-07, "epoch": 4.567774936061381, "percentage": 91.36, "elapsed_time": "9:12:07", "remaining_time": "0:52:14"} +{"current_steps": 1787, "total_steps": 1955, "loss": 1.0322, "lr": 4.467822352160722e-07, "epoch": 4.570332480818414, "percentage": 91.41, "elapsed_time": "9:12:26", "remaining_time": "0:51:56"} +{"current_steps": 1788, "total_steps": 1955, "loss": 1.021, "lr": 4.415186029901719e-07, "epoch": 4.572890025575448, "percentage": 91.46, "elapsed_time": "9:12:45", "remaining_time": "0:51:37"} +{"current_steps": 1789, "total_steps": 1955, "loss": 1.0747, "lr": 4.3628546072396525e-07, "epoch": 4.57544757033248, "percentage": 91.51, "elapsed_time": "9:13:03", "remaining_time": "0:51:19"} +{"current_steps": 1790, "total_steps": 1955, "loss": 1.0488, "lr": 4.310828251103072e-07, "epoch": 4.578005115089514, "percentage": 91.56, "elapsed_time": "9:13:21", "remaining_time": "0:51:00"} +{"current_steps": 1791, "total_steps": 1955, "loss": 1.0362, "lr": 4.2591071274473484e-07, "epoch": 4.580562659846548, "percentage": 91.61, "elapsed_time": "9:13:40", "remaining_time": "0:50:41"} +{"current_steps": 1792, "total_steps": 1955, "loss": 1.037, "lr": 4.2076914012542105e-07, "epoch": 4.58312020460358, "percentage": 91.66, "elapsed_time": "9:13:59", "remaining_time": "0:50:23"} +{"current_steps": 1793, "total_steps": 1955, "loss": 1.0303, "lr": 4.1565812365312654e-07, "epoch": 4.585677749360614, "percentage": 91.71, "elapsed_time": "9:14:17", "remaining_time": "0:50:04"} +{"current_steps": 1794, "total_steps": 1955, "loss": 1.0411, "lr": 4.105776796311389e-07, "epoch": 4.588235294117647, "percentage": 91.76, "elapsed_time": "9:14:36", "remaining_time": "0:49:46"} +{"current_steps": 1795, "total_steps": 1955, "loss": 1.0201, "lr": 4.055278242652272e-07, "epoch": 4.59079283887468, "percentage": 91.82, "elapsed_time": "9:14:54", "remaining_time": "0:49:27"} +{"current_steps": 1796, "total_steps": 1955, "loss": 1.0326, "lr": 4.00508573663585e-07, "epoch": 4.593350383631714, "percentage": 91.87, "elapsed_time": "9:15:12", "remaining_time": "0:49:09"} +{"current_steps": 1797, "total_steps": 1955, "loss": 1.0401, "lr": 3.955199438367874e-07, "epoch": 4.595907928388747, "percentage": 91.92, "elapsed_time": "9:15:31", "remaining_time": "0:48:50"} +{"current_steps": 1798, "total_steps": 1955, "loss": 1.0168, "lr": 3.9056195069772873e-07, "epoch": 4.59846547314578, "percentage": 91.97, "elapsed_time": "9:15:50", "remaining_time": "0:48:32"} +{"current_steps": 1799, "total_steps": 1955, "loss": 1.025, "lr": 3.85634610061576e-07, "epoch": 4.601023017902813, "percentage": 92.02, "elapsed_time": "9:16:08", "remaining_time": "0:48:13"} +{"current_steps": 1800, "total_steps": 1955, "loss": 1.0579, "lr": 3.807379376457276e-07, "epoch": 4.603580562659847, "percentage": 92.07, "elapsed_time": "9:16:27", "remaining_time": "0:47:55"} +{"current_steps": 1801, "total_steps": 1955, "loss": 1.0162, "lr": 3.758719490697493e-07, "epoch": 4.6061381074168795, "percentage": 92.12, "elapsed_time": "9:16:45", "remaining_time": "0:47:36"} +{"current_steps": 1802, "total_steps": 1955, "loss": 1.0064, "lr": 3.710366598553328e-07, "epoch": 4.608695652173913, "percentage": 92.17, "elapsed_time": "9:17:04", "remaining_time": "0:47:17"} +{"current_steps": 1803, "total_steps": 1955, "loss": 1.0317, "lr": 3.6623208542624134e-07, "epoch": 4.611253196930946, "percentage": 92.23, "elapsed_time": "9:17:22", "remaining_time": "0:46:59"} +{"current_steps": 1804, "total_steps": 1955, "loss": 0.9979, "lr": 3.61458241108269e-07, "epoch": 4.6138107416879794, "percentage": 92.28, "elapsed_time": "9:17:41", "remaining_time": "0:46:40"} +{"current_steps": 1805, "total_steps": 1955, "loss": 1.0428, "lr": 3.567151421291781e-07, "epoch": 4.616368286445013, "percentage": 92.33, "elapsed_time": "9:17:59", "remaining_time": "0:46:22"} +{"current_steps": 1806, "total_steps": 1955, "loss": 1.0055, "lr": 3.520028036186629e-07, "epoch": 4.618925831202046, "percentage": 92.38, "elapsed_time": "9:18:17", "remaining_time": "0:46:03"} +{"current_steps": 1807, "total_steps": 1955, "loss": 1.0358, "lr": 3.473212406082993e-07, "epoch": 4.621483375959079, "percentage": 92.43, "elapsed_time": "9:18:36", "remaining_time": "0:45:45"} +{"current_steps": 1808, "total_steps": 1955, "loss": 1.0818, "lr": 3.426704680314896e-07, "epoch": 4.624040920716112, "percentage": 92.48, "elapsed_time": "9:18:55", "remaining_time": "0:45:26"} +{"current_steps": 1809, "total_steps": 1955, "loss": 1.0499, "lr": 3.380505007234225e-07, "epoch": 4.626598465473146, "percentage": 92.53, "elapsed_time": "9:19:13", "remaining_time": "0:45:08"} +{"current_steps": 1810, "total_steps": 1955, "loss": 1.0368, "lr": 3.334613534210218e-07, "epoch": 4.629156010230179, "percentage": 92.58, "elapsed_time": "9:19:31", "remaining_time": "0:44:49"} +{"current_steps": 1811, "total_steps": 1955, "loss": 1.0364, "lr": 3.2890304076290124e-07, "epoch": 4.631713554987212, "percentage": 92.63, "elapsed_time": "9:19:50", "remaining_time": "0:44:30"} +{"current_steps": 1812, "total_steps": 1955, "loss": 1.0657, "lr": 3.2437557728931643e-07, "epoch": 4.634271099744246, "percentage": 92.69, "elapsed_time": "9:20:09", "remaining_time": "0:44:12"} +{"current_steps": 1813, "total_steps": 1955, "loss": 1.0011, "lr": 3.1987897744212073e-07, "epoch": 4.6368286445012785, "percentage": 92.74, "elapsed_time": "9:20:27", "remaining_time": "0:43:53"} +{"current_steps": 1814, "total_steps": 1955, "loss": 1.0713, "lr": 3.154132555647171e-07, "epoch": 4.639386189258312, "percentage": 92.79, "elapsed_time": "9:20:45", "remaining_time": "0:43:35"} +{"current_steps": 1815, "total_steps": 1955, "loss": 1.0622, "lr": 3.1097842590201433e-07, "epoch": 4.641943734015345, "percentage": 92.84, "elapsed_time": "9:21:04", "remaining_time": "0:43:16"} +{"current_steps": 1816, "total_steps": 1955, "loss": 1.0629, "lr": 3.0657450260037303e-07, "epoch": 4.6445012787723785, "percentage": 92.89, "elapsed_time": "9:21:23", "remaining_time": "0:42:58"} +{"current_steps": 1817, "total_steps": 1955, "loss": 1.0594, "lr": 3.022014997075795e-07, "epoch": 4.647058823529412, "percentage": 92.94, "elapsed_time": "9:21:41", "remaining_time": "0:42:39"} +{"current_steps": 1818, "total_steps": 1955, "loss": 1.0469, "lr": 2.9785943117277894e-07, "epoch": 4.649616368286445, "percentage": 92.99, "elapsed_time": "9:21:59", "remaining_time": "0:42:21"} +{"current_steps": 1819, "total_steps": 1955, "loss": 1.0224, "lr": 2.935483108464465e-07, "epoch": 4.6521739130434785, "percentage": 93.04, "elapsed_time": "9:22:17", "remaining_time": "0:42:02"} +{"current_steps": 1820, "total_steps": 1955, "loss": 1.0112, "lr": 2.8926815248033535e-07, "epoch": 4.654731457800511, "percentage": 93.09, "elapsed_time": "9:22:36", "remaining_time": "0:41:43"} +{"current_steps": 1821, "total_steps": 1955, "loss": 1.0105, "lr": 2.8501896972743747e-07, "epoch": 4.657289002557545, "percentage": 93.15, "elapsed_time": "9:22:54", "remaining_time": "0:41:25"} +{"current_steps": 1822, "total_steps": 1955, "loss": 1.0697, "lr": 2.808007761419351e-07, "epoch": 4.659846547314578, "percentage": 93.2, "elapsed_time": "9:23:13", "remaining_time": "0:41:06"} +{"current_steps": 1823, "total_steps": 1955, "loss": 1.0141, "lr": 2.766135851791607e-07, "epoch": 4.662404092071611, "percentage": 93.25, "elapsed_time": "9:23:31", "remaining_time": "0:40:48"} +{"current_steps": 1824, "total_steps": 1955, "loss": 1.0282, "lr": 2.7245741019555573e-07, "epoch": 4.664961636828645, "percentage": 93.3, "elapsed_time": "9:23:49", "remaining_time": "0:40:29"} +{"current_steps": 1825, "total_steps": 1955, "loss": 1.0299, "lr": 2.6833226444862526e-07, "epoch": 4.667519181585678, "percentage": 93.35, "elapsed_time": "9:24:07", "remaining_time": "0:40:11"} +{"current_steps": 1826, "total_steps": 1955, "loss": 1.0267, "lr": 2.642381610968936e-07, "epoch": 4.670076726342711, "percentage": 93.4, "elapsed_time": "9:24:25", "remaining_time": "0:39:52"} +{"current_steps": 1827, "total_steps": 1955, "loss": 1.029, "lr": 2.6017511319986753e-07, "epoch": 4.672634271099744, "percentage": 93.45, "elapsed_time": "9:24:44", "remaining_time": "0:39:33"} +{"current_steps": 1828, "total_steps": 1955, "loss": 1.0193, "lr": 2.5614313371799203e-07, "epoch": 4.675191815856778, "percentage": 93.5, "elapsed_time": "9:25:02", "remaining_time": "0:39:15"} +{"current_steps": 1829, "total_steps": 1955, "loss": 1.0267, "lr": 2.5214223551260686e-07, "epoch": 4.677749360613811, "percentage": 93.55, "elapsed_time": "9:25:21", "remaining_time": "0:38:56"} +{"current_steps": 1830, "total_steps": 1955, "loss": 1.0524, "lr": 2.4817243134591106e-07, "epoch": 4.680306905370844, "percentage": 93.61, "elapsed_time": "9:25:39", "remaining_time": "0:38:38"} +{"current_steps": 1831, "total_steps": 1955, "loss": 1.0347, "lr": 2.4423373388091755e-07, "epoch": 4.6828644501278776, "percentage": 93.66, "elapsed_time": "9:25:58", "remaining_time": "0:38:19"} +{"current_steps": 1832, "total_steps": 1955, "loss": 1.0568, "lr": 2.403261556814118e-07, "epoch": 4.68542199488491, "percentage": 93.71, "elapsed_time": "9:26:17", "remaining_time": "0:38:01"} +{"current_steps": 1833, "total_steps": 1955, "loss": 1.0266, "lr": 2.3644970921191447e-07, "epoch": 4.687979539641944, "percentage": 93.76, "elapsed_time": "9:26:35", "remaining_time": "0:37:42"} +{"current_steps": 1834, "total_steps": 1955, "loss": 1.0845, "lr": 2.3260440683764653e-07, "epoch": 4.690537084398977, "percentage": 93.81, "elapsed_time": "9:26:53", "remaining_time": "0:37:24"} +{"current_steps": 1835, "total_steps": 1955, "loss": 1.0469, "lr": 2.2879026082447743e-07, "epoch": 4.69309462915601, "percentage": 93.86, "elapsed_time": "9:27:12", "remaining_time": "0:37:05"} +{"current_steps": 1836, "total_steps": 1955, "loss": 1.0585, "lr": 2.250072833388972e-07, "epoch": 4.695652173913043, "percentage": 93.91, "elapsed_time": "9:27:30", "remaining_time": "0:36:46"} +{"current_steps": 1837, "total_steps": 1955, "loss": 1.0422, "lr": 2.2125548644797322e-07, "epoch": 4.698209718670077, "percentage": 93.96, "elapsed_time": "9:27:48", "remaining_time": "0:36:28"} +{"current_steps": 1838, "total_steps": 1955, "loss": 1.0095, "lr": 2.1753488211931017e-07, "epoch": 4.70076726342711, "percentage": 94.02, "elapsed_time": "9:28:07", "remaining_time": "0:36:09"} +{"current_steps": 1839, "total_steps": 1955, "loss": 1.0064, "lr": 2.1384548222101342e-07, "epoch": 4.703324808184143, "percentage": 94.07, "elapsed_time": "9:28:26", "remaining_time": "0:35:51"} +{"current_steps": 1840, "total_steps": 1955, "loss": 1.0682, "lr": 2.1018729852165575e-07, "epoch": 4.705882352941177, "percentage": 94.12, "elapsed_time": "9:28:44", "remaining_time": "0:35:32"} +{"current_steps": 1841, "total_steps": 1955, "loss": 1.0267, "lr": 2.0656034269022961e-07, "epoch": 4.708439897698209, "percentage": 94.17, "elapsed_time": "9:29:03", "remaining_time": "0:35:14"} +{"current_steps": 1842, "total_steps": 1955, "loss": 1.0235, "lr": 2.0296462629611934e-07, "epoch": 4.710997442455243, "percentage": 94.22, "elapsed_time": "9:29:21", "remaining_time": "0:34:55"} +{"current_steps": 1843, "total_steps": 1955, "loss": 1.0249, "lr": 1.9940016080906122e-07, "epoch": 4.713554987212277, "percentage": 94.27, "elapsed_time": "9:29:40", "remaining_time": "0:34:37"} +{"current_steps": 1844, "total_steps": 1955, "loss": 1.0477, "lr": 1.9586695759910235e-07, "epoch": 4.716112531969309, "percentage": 94.32, "elapsed_time": "9:29:58", "remaining_time": "0:34:18"} +{"current_steps": 1845, "total_steps": 1955, "loss": 1.0441, "lr": 1.9236502793657297e-07, "epoch": 4.718670076726343, "percentage": 94.37, "elapsed_time": "9:30:17", "remaining_time": "0:34:00"} +{"current_steps": 1846, "total_steps": 1955, "loss": 1.0336, "lr": 1.888943829920431e-07, "epoch": 4.721227621483376, "percentage": 94.42, "elapsed_time": "9:30:34", "remaining_time": "0:33:41"} +{"current_steps": 1847, "total_steps": 1955, "loss": 1.0376, "lr": 1.854550338362915e-07, "epoch": 4.723785166240409, "percentage": 94.48, "elapsed_time": "9:30:53", "remaining_time": "0:33:22"} +{"current_steps": 1848, "total_steps": 1955, "loss": 1.0689, "lr": 1.8204699144026894e-07, "epoch": 4.726342710997442, "percentage": 94.53, "elapsed_time": "9:31:12", "remaining_time": "0:33:04"} +{"current_steps": 1849, "total_steps": 1955, "loss": 1.0672, "lr": 1.7867026667505726e-07, "epoch": 4.728900255754476, "percentage": 94.58, "elapsed_time": "9:31:30", "remaining_time": "0:32:45"} +{"current_steps": 1850, "total_steps": 1955, "loss": 1.034, "lr": 1.753248703118482e-07, "epoch": 4.731457800511509, "percentage": 94.63, "elapsed_time": "9:31:49", "remaining_time": "0:32:27"} +{"current_steps": 1851, "total_steps": 1955, "loss": 1.0355, "lr": 1.720108130218967e-07, "epoch": 4.734015345268542, "percentage": 94.68, "elapsed_time": "9:32:08", "remaining_time": "0:32:08"} +{"current_steps": 1852, "total_steps": 1955, "loss": 1.0274, "lr": 1.6872810537649332e-07, "epoch": 4.736572890025576, "percentage": 94.73, "elapsed_time": "9:32:26", "remaining_time": "0:31:50"} +{"current_steps": 1853, "total_steps": 1955, "loss": 1.0707, "lr": 1.6547675784692517e-07, "epoch": 4.739130434782608, "percentage": 94.78, "elapsed_time": "9:32:45", "remaining_time": "0:31:31"} +{"current_steps": 1854, "total_steps": 1955, "loss": 1.0631, "lr": 1.6225678080444952e-07, "epoch": 4.741687979539642, "percentage": 94.83, "elapsed_time": "9:33:04", "remaining_time": "0:31:13"} +{"current_steps": 1855, "total_steps": 1955, "loss": 1.0394, "lr": 1.5906818452025464e-07, "epoch": 4.744245524296675, "percentage": 94.88, "elapsed_time": "9:33:22", "remaining_time": "0:30:54"} +{"current_steps": 1856, "total_steps": 1955, "loss": 1.0059, "lr": 1.559109791654301e-07, "epoch": 4.746803069053708, "percentage": 94.94, "elapsed_time": "9:33:41", "remaining_time": "0:30:36"} +{"current_steps": 1857, "total_steps": 1955, "loss": 1.0268, "lr": 1.5278517481093436e-07, "epoch": 4.749360613810742, "percentage": 94.99, "elapsed_time": "9:33:59", "remaining_time": "0:30:17"} +{"current_steps": 1858, "total_steps": 1955, "loss": 1.0565, "lr": 1.4969078142756276e-07, "epoch": 4.751918158567775, "percentage": 95.04, "elapsed_time": "9:34:18", "remaining_time": "0:29:58"} +{"current_steps": 1859, "total_steps": 1955, "loss": 1.0271, "lr": 1.4662780888591078e-07, "epoch": 4.754475703324808, "percentage": 95.09, "elapsed_time": "9:34:36", "remaining_time": "0:29:40"} +{"current_steps": 1860, "total_steps": 1955, "loss": 1.0529, "lr": 1.4359626695635176e-07, "epoch": 4.757033248081841, "percentage": 95.14, "elapsed_time": "9:34:55", "remaining_time": "0:29:21"} +{"current_steps": 1861, "total_steps": 1955, "loss": 1.047, "lr": 1.405961653089971e-07, "epoch": 4.759590792838875, "percentage": 95.19, "elapsed_time": "9:35:14", "remaining_time": "0:29:03"} +{"current_steps": 1862, "total_steps": 1955, "loss": 1.0394, "lr": 1.3762751351367066e-07, "epoch": 4.762148337595908, "percentage": 95.24, "elapsed_time": "9:35:33", "remaining_time": "0:28:44"} +{"current_steps": 1863, "total_steps": 1955, "loss": 1.0545, "lr": 1.3469032103987534e-07, "epoch": 4.764705882352941, "percentage": 95.29, "elapsed_time": "9:35:51", "remaining_time": "0:28:26"} +{"current_steps": 1864, "total_steps": 1955, "loss": 1.0633, "lr": 1.3178459725676552e-07, "epoch": 4.767263427109975, "percentage": 95.35, "elapsed_time": "9:36:10", "remaining_time": "0:28:07"} +{"current_steps": 1865, "total_steps": 1955, "loss": 1.0257, "lr": 1.289103514331147e-07, "epoch": 4.7698209718670075, "percentage": 95.4, "elapsed_time": "9:36:28", "remaining_time": "0:27:49"} +{"current_steps": 1866, "total_steps": 1955, "loss": 1.0532, "lr": 1.2606759273728563e-07, "epoch": 4.772378516624041, "percentage": 95.45, "elapsed_time": "9:36:47", "remaining_time": "0:27:30"} +{"current_steps": 1867, "total_steps": 1955, "loss": 1.0275, "lr": 1.2325633023720695e-07, "epoch": 4.774936061381074, "percentage": 95.5, "elapsed_time": "9:37:05", "remaining_time": "0:27:12"} +{"current_steps": 1868, "total_steps": 1955, "loss": 1.0286, "lr": 1.2047657290033322e-07, "epoch": 4.7774936061381075, "percentage": 95.55, "elapsed_time": "9:37:24", "remaining_time": "0:26:53"} +{"current_steps": 1869, "total_steps": 1955, "loss": 1.0433, "lr": 1.1772832959362934e-07, "epoch": 4.78005115089514, "percentage": 95.6, "elapsed_time": "9:37:42", "remaining_time": "0:26:34"} +{"current_steps": 1870, "total_steps": 1955, "loss": 1.0332, "lr": 1.150116090835307e-07, "epoch": 4.782608695652174, "percentage": 95.65, "elapsed_time": "9:38:01", "remaining_time": "0:26:16"} +{"current_steps": 1871, "total_steps": 1955, "loss": 1.0849, "lr": 1.1232642003592198e-07, "epoch": 4.7851662404092075, "percentage": 95.7, "elapsed_time": "9:38:19", "remaining_time": "0:25:57"} +{"current_steps": 1872, "total_steps": 1955, "loss": 1.0167, "lr": 1.096727710161094e-07, "epoch": 4.78772378516624, "percentage": 95.75, "elapsed_time": "9:38:38", "remaining_time": "0:25:39"} +{"current_steps": 1873, "total_steps": 1955, "loss": 1.0227, "lr": 1.0705067048878859e-07, "epoch": 4.790281329923274, "percentage": 95.81, "elapsed_time": "9:38:56", "remaining_time": "0:25:20"} +{"current_steps": 1874, "total_steps": 1955, "loss": 1.0085, "lr": 1.0446012681802342e-07, "epoch": 4.792838874680307, "percentage": 95.86, "elapsed_time": "9:39:15", "remaining_time": "0:25:02"} +{"current_steps": 1875, "total_steps": 1955, "loss": 1.0399, "lr": 1.0190114826721498e-07, "epoch": 4.79539641943734, "percentage": 95.91, "elapsed_time": "9:39:33", "remaining_time": "0:24:43"} +{"current_steps": 1876, "total_steps": 1955, "loss": 1.0065, "lr": 9.937374299907931e-08, "epoch": 4.797953964194374, "percentage": 95.96, "elapsed_time": "9:39:52", "remaining_time": "0:24:25"} +{"current_steps": 1877, "total_steps": 1955, "loss": 1.0496, "lr": 9.687791907561528e-08, "epoch": 4.8005115089514065, "percentage": 96.01, "elapsed_time": "9:40:11", "remaining_time": "0:24:06"} +{"current_steps": 1878, "total_steps": 1955, "loss": 1.0378, "lr": 9.441368445808452e-08, "epoch": 4.80306905370844, "percentage": 96.06, "elapsed_time": "9:40:29", "remaining_time": "0:23:48"} +{"current_steps": 1879, "total_steps": 1955, "loss": 1.0202, "lr": 9.198104700698596e-08, "epoch": 4.805626598465473, "percentage": 96.11, "elapsed_time": "9:40:47", "remaining_time": "0:23:29"} +{"current_steps": 1880, "total_steps": 1955, "loss": 1.0152, "lr": 8.958001448202358e-08, "epoch": 4.8081841432225065, "percentage": 96.16, "elapsed_time": "9:41:06", "remaining_time": "0:23:10"} +{"current_steps": 1881, "total_steps": 1955, "loss": 1.0485, "lr": 8.721059454209424e-08, "epoch": 4.810741687979539, "percentage": 96.21, "elapsed_time": "9:41:25", "remaining_time": "0:22:52"} +{"current_steps": 1882, "total_steps": 1955, "loss": 1.017, "lr": 8.48727947452499e-08, "epoch": 4.813299232736573, "percentage": 96.27, "elapsed_time": "9:41:43", "remaining_time": "0:22:33"} +{"current_steps": 1883, "total_steps": 1955, "loss": 1.0076, "lr": 8.256662254867986e-08, "epoch": 4.8158567774936065, "percentage": 96.32, "elapsed_time": "9:42:02", "remaining_time": "0:22:15"} +{"current_steps": 1884, "total_steps": 1955, "loss": 1.0295, "lr": 8.029208530869081e-08, "epoch": 4.818414322250639, "percentage": 96.37, "elapsed_time": "9:42:20", "remaining_time": "0:21:56"} +{"current_steps": 1885, "total_steps": 1955, "loss": 1.0307, "lr": 7.804919028067681e-08, "epoch": 4.820971867007673, "percentage": 96.42, "elapsed_time": "9:42:39", "remaining_time": "0:21:38"} +{"current_steps": 1886, "total_steps": 1955, "loss": 1.0466, "lr": 7.58379446190971e-08, "epoch": 4.823529411764706, "percentage": 96.47, "elapsed_time": "9:42:58", "remaining_time": "0:21:19"} +{"current_steps": 1887, "total_steps": 1955, "loss": 1.0442, "lr": 7.365835537745724e-08, "epoch": 4.826086956521739, "percentage": 96.52, "elapsed_time": "9:43:16", "remaining_time": "0:21:01"} +{"current_steps": 1888, "total_steps": 1955, "loss": 1.0445, "lr": 7.151042950828247e-08, "epoch": 4.828644501278772, "percentage": 96.57, "elapsed_time": "9:43:34", "remaining_time": "0:20:42"} +{"current_steps": 1889, "total_steps": 1955, "loss": 1.029, "lr": 6.939417386309765e-08, "epoch": 4.831202046035806, "percentage": 96.62, "elapsed_time": "9:43:53", "remaining_time": "0:20:24"} +{"current_steps": 1890, "total_steps": 1955, "loss": 1.0097, "lr": 6.73095951924041e-08, "epoch": 4.833759590792839, "percentage": 96.68, "elapsed_time": "9:44:11", "remaining_time": "0:20:05"} +{"current_steps": 1891, "total_steps": 1955, "loss": 1.0464, "lr": 6.525670014566166e-08, "epoch": 4.836317135549872, "percentage": 96.73, "elapsed_time": "9:44:29", "remaining_time": "0:19:46"} +{"current_steps": 1892, "total_steps": 1955, "loss": 1.0109, "lr": 6.323549527126216e-08, "epoch": 4.838874680306906, "percentage": 96.78, "elapsed_time": "9:44:48", "remaining_time": "0:19:28"} +{"current_steps": 1893, "total_steps": 1955, "loss": 0.9896, "lr": 6.124598701651052e-08, "epoch": 4.841432225063938, "percentage": 96.83, "elapsed_time": "9:45:06", "remaining_time": "0:19:09"} +{"current_steps": 1894, "total_steps": 1955, "loss": 1.0406, "lr": 5.9288181727606976e-08, "epoch": 4.843989769820972, "percentage": 96.88, "elapsed_time": "9:45:25", "remaining_time": "0:18:51"} +{"current_steps": 1895, "total_steps": 1955, "loss": 1.0292, "lr": 5.7362085649622646e-08, "epoch": 4.846547314578006, "percentage": 96.93, "elapsed_time": "9:45:44", "remaining_time": "0:18:32"} +{"current_steps": 1896, "total_steps": 1955, "loss": 1.0641, "lr": 5.546770492648401e-08, "epoch": 4.849104859335038, "percentage": 96.98, "elapsed_time": "9:46:02", "remaining_time": "0:18:14"} +{"current_steps": 1897, "total_steps": 1955, "loss": 1.0278, "lr": 5.360504560094737e-08, "epoch": 4.851662404092072, "percentage": 97.03, "elapsed_time": "9:46:21", "remaining_time": "0:17:55"} +{"current_steps": 1898, "total_steps": 1955, "loss": 1.0405, "lr": 5.1774113614586616e-08, "epoch": 4.854219948849105, "percentage": 97.08, "elapsed_time": "9:46:40", "remaining_time": "0:17:37"} +{"current_steps": 1899, "total_steps": 1955, "loss": 1.0753, "lr": 4.9974914807767724e-08, "epoch": 4.856777493606138, "percentage": 97.14, "elapsed_time": "9:46:58", "remaining_time": "0:17:18"} +{"current_steps": 1900, "total_steps": 1955, "loss": 1.0235, "lr": 4.82074549196343e-08, "epoch": 4.859335038363171, "percentage": 97.19, "elapsed_time": "9:47:17", "remaining_time": "0:17:00"} +{"current_steps": 1901, "total_steps": 1955, "loss": 1.0253, "lr": 4.6471739588089814e-08, "epoch": 4.861892583120205, "percentage": 97.24, "elapsed_time": "9:47:35", "remaining_time": "0:16:41"} +{"current_steps": 1902, "total_steps": 1955, "loss": 1.0321, "lr": 4.4767774349776527e-08, "epoch": 4.864450127877237, "percentage": 97.29, "elapsed_time": "9:47:54", "remaining_time": "0:16:22"} +{"current_steps": 1903, "total_steps": 1955, "loss": 1.043, "lr": 4.30955646400566e-08, "epoch": 4.867007672634271, "percentage": 97.34, "elapsed_time": "9:48:13", "remaining_time": "0:16:04"} +{"current_steps": 1904, "total_steps": 1955, "loss": 1.0538, "lr": 4.14551157930021e-08, "epoch": 4.869565217391305, "percentage": 97.39, "elapsed_time": "9:48:31", "remaining_time": "0:15:45"} +{"current_steps": 1905, "total_steps": 1955, "loss": 1.0476, "lr": 3.984643304136948e-08, "epoch": 4.872122762148337, "percentage": 97.44, "elapsed_time": "9:48:49", "remaining_time": "0:15:27"} +{"current_steps": 1906, "total_steps": 1955, "loss": 1.0356, "lr": 3.826952151658958e-08, "epoch": 4.874680306905371, "percentage": 97.49, "elapsed_time": "9:49:07", "remaining_time": "0:15:08"} +{"current_steps": 1907, "total_steps": 1955, "loss": 1.0402, "lr": 3.6724386248745415e-08, "epoch": 4.877237851662404, "percentage": 97.54, "elapsed_time": "9:49:25", "remaining_time": "0:14:50"} +{"current_steps": 1908, "total_steps": 1955, "loss": 1.0651, "lr": 3.5211032166561075e-08, "epoch": 4.879795396419437, "percentage": 97.6, "elapsed_time": "9:49:44", "remaining_time": "0:14:31"} +{"current_steps": 1909, "total_steps": 1955, "loss": 1.0723, "lr": 3.372946409738398e-08, "epoch": 4.882352941176471, "percentage": 97.65, "elapsed_time": "9:50:02", "remaining_time": "0:14:13"} +{"current_steps": 1910, "total_steps": 1955, "loss": 1.0322, "lr": 3.22796867671682e-08, "epoch": 4.884910485933504, "percentage": 97.7, "elapsed_time": "9:50:21", "remaining_time": "0:13:54"} +{"current_steps": 1911, "total_steps": 1955, "loss": 1.0121, "lr": 3.086170480046114e-08, "epoch": 4.887468030690537, "percentage": 97.75, "elapsed_time": "9:50:40", "remaining_time": "0:13:35"} +{"current_steps": 1912, "total_steps": 1955, "loss": 1.0571, "lr": 2.9475522720389116e-08, "epoch": 4.89002557544757, "percentage": 97.8, "elapsed_time": "9:50:58", "remaining_time": "0:13:17"} +{"current_steps": 1913, "total_steps": 1955, "loss": 1.0378, "lr": 2.8121144948641822e-08, "epoch": 4.892583120204604, "percentage": 97.85, "elapsed_time": "9:51:17", "remaining_time": "0:12:58"} +{"current_steps": 1914, "total_steps": 1955, "loss": 1.0468, "lr": 2.679857580545786e-08, "epoch": 4.8951406649616365, "percentage": 97.9, "elapsed_time": "9:51:36", "remaining_time": "0:12:40"} +{"current_steps": 1915, "total_steps": 1955, "loss": 1.0388, "lr": 2.5507819509610343e-08, "epoch": 4.89769820971867, "percentage": 97.95, "elapsed_time": "9:51:54", "remaining_time": "0:12:21"} +{"current_steps": 1916, "total_steps": 1955, "loss": 1.0633, "lr": 2.4248880178399104e-08, "epoch": 4.900255754475703, "percentage": 98.01, "elapsed_time": "9:52:13", "remaining_time": "0:12:03"} +{"current_steps": 1917, "total_steps": 1955, "loss": 1.0193, "lr": 2.3021761827628497e-08, "epoch": 4.9028132992327365, "percentage": 98.06, "elapsed_time": "9:52:32", "remaining_time": "0:11:44"} +{"current_steps": 1918, "total_steps": 1955, "loss": 0.9916, "lr": 2.182646837160185e-08, "epoch": 4.90537084398977, "percentage": 98.11, "elapsed_time": "9:52:50", "remaining_time": "0:11:26"} +{"current_steps": 1919, "total_steps": 1955, "loss": 1.0672, "lr": 2.0663003623105916e-08, "epoch": 4.907928388746803, "percentage": 98.16, "elapsed_time": "9:53:09", "remaining_time": "0:11:07"} +{"current_steps": 1920, "total_steps": 1955, "loss": 1.0298, "lr": 1.953137129339977e-08, "epoch": 4.910485933503836, "percentage": 98.21, "elapsed_time": "9:53:27", "remaining_time": "0:10:49"} +{"current_steps": 1921, "total_steps": 1955, "loss": 1.0461, "lr": 1.8431574992199276e-08, "epoch": 4.913043478260869, "percentage": 98.26, "elapsed_time": "9:53:46", "remaining_time": "0:10:30"} +{"current_steps": 1922, "total_steps": 1955, "loss": 1.0395, "lr": 1.736361822767263e-08, "epoch": 4.915601023017903, "percentage": 98.31, "elapsed_time": "9:54:04", "remaining_time": "0:10:12"} +{"current_steps": 1923, "total_steps": 1955, "loss": 1.0231, "lr": 1.632750440642261e-08, "epoch": 4.918158567774936, "percentage": 98.36, "elapsed_time": "9:54:23", "remaining_time": "0:09:53"} +{"current_steps": 1924, "total_steps": 1955, "loss": 1.0259, "lr": 1.5323236833479914e-08, "epoch": 4.920716112531969, "percentage": 98.41, "elapsed_time": "9:54:41", "remaining_time": "0:09:34"} +{"current_steps": 1925, "total_steps": 1955, "loss": 1.0613, "lr": 1.4350818712292047e-08, "epoch": 4.923273657289003, "percentage": 98.47, "elapsed_time": "9:54:59", "remaining_time": "0:09:16"} +{"current_steps": 1926, "total_steps": 1955, "loss": 1.0592, "lr": 1.3410253144707786e-08, "epoch": 4.9258312020460355, "percentage": 98.52, "elapsed_time": "9:55:18", "remaining_time": "0:08:57"} +{"current_steps": 1927, "total_steps": 1955, "loss": 1.062, "lr": 1.2501543130974958e-08, "epoch": 4.928388746803069, "percentage": 98.57, "elapsed_time": "9:55:36", "remaining_time": "0:08:39"} +{"current_steps": 1928, "total_steps": 1955, "loss": 1.0245, "lr": 1.162469156972712e-08, "epoch": 4.930946291560103, "percentage": 98.62, "elapsed_time": "9:55:55", "remaining_time": "0:08:20"} +{"current_steps": 1929, "total_steps": 1955, "loss": 1.0449, "lr": 1.0779701257974672e-08, "epoch": 4.9335038363171355, "percentage": 98.67, "elapsed_time": "9:56:13", "remaining_time": "0:08:02"} +{"current_steps": 1930, "total_steps": 1955, "loss": 1.0682, "lr": 9.96657489109487e-09, "epoch": 4.936061381074169, "percentage": 98.72, "elapsed_time": "9:56:32", "remaining_time": "0:07:43"} +{"current_steps": 1931, "total_steps": 1955, "loss": 1.0674, "lr": 9.185315062826272e-09, "epoch": 4.938618925831202, "percentage": 98.77, "elapsed_time": "9:56:50", "remaining_time": "0:07:25"} +{"current_steps": 1932, "total_steps": 1955, "loss": 1.0479, "lr": 8.435924265256523e-09, "epoch": 4.9411764705882355, "percentage": 98.82, "elapsed_time": "9:57:09", "remaining_time": "0:07:06"} +{"current_steps": 1933, "total_steps": 1955, "loss": 1.0148, "lr": 7.718404888816811e-09, "epoch": 4.943734015345268, "percentage": 98.87, "elapsed_time": "9:57:28", "remaining_time": "0:06:47"} +{"current_steps": 1934, "total_steps": 1955, "loss": 1.0408, "lr": 7.032759222274088e-09, "epoch": 4.946291560102302, "percentage": 98.93, "elapsed_time": "9:57:46", "remaining_time": "0:06:29"} +{"current_steps": 1935, "total_steps": 1955, "loss": 1.0236, "lr": 6.378989452724416e-09, "epoch": 4.948849104859335, "percentage": 98.98, "elapsed_time": "9:58:04", "remaining_time": "0:06:10"} +{"current_steps": 1936, "total_steps": 1955, "loss": 1.0335, "lr": 5.757097665584077e-09, "epoch": 4.951406649616368, "percentage": 99.03, "elapsed_time": "9:58:23", "remaining_time": "0:05:52"} +{"current_steps": 1937, "total_steps": 1955, "loss": 1.0354, "lr": 5.16708584458292e-09, "epoch": 4.953964194373402, "percentage": 99.08, "elapsed_time": "9:58:41", "remaining_time": "0:05:33"} +{"current_steps": 1938, "total_steps": 1955, "loss": 1.0077, "lr": 4.608955871761023e-09, "epoch": 4.956521739130435, "percentage": 99.13, "elapsed_time": "9:58:59", "remaining_time": "0:05:15"} +{"current_steps": 1939, "total_steps": 1955, "loss": 1.0478, "lr": 4.0827095274598165e-09, "epoch": 4.959079283887468, "percentage": 99.18, "elapsed_time": "9:59:17", "remaining_time": "0:04:56"} +{"current_steps": 1940, "total_steps": 1955, "loss": 1.0568, "lr": 3.588348490317639e-09, "epoch": 4.961636828644501, "percentage": 99.23, "elapsed_time": "9:59:35", "remaining_time": "0:04:38"} +{"current_steps": 1941, "total_steps": 1955, "loss": 1.0352, "lr": 3.1258743372619693e-09, "epoch": 4.964194373401535, "percentage": 99.28, "elapsed_time": "9:59:54", "remaining_time": "0:04:19"} +{"current_steps": 1942, "total_steps": 1955, "loss": 1.047, "lr": 2.6952885435105324e-09, "epoch": 4.966751918158568, "percentage": 99.34, "elapsed_time": "10:00:13", "remaining_time": "0:04:01"} +{"current_steps": 1943, "total_steps": 1955, "loss": 1.0417, "lr": 2.29659248255798e-09, "epoch": 4.969309462915601, "percentage": 99.39, "elapsed_time": "10:00:31", "remaining_time": "0:03:42"} +{"current_steps": 1944, "total_steps": 1955, "loss": 1.0137, "lr": 1.9297874261792194e-09, "epoch": 4.971867007672635, "percentage": 99.44, "elapsed_time": "10:00:50", "remaining_time": "0:03:23"} +{"current_steps": 1945, "total_steps": 1955, "loss": 1.0311, "lr": 1.594874544421643e-09, "epoch": 4.974424552429667, "percentage": 99.49, "elapsed_time": "10:01:08", "remaining_time": "0:03:05"} +{"current_steps": 1946, "total_steps": 1955, "loss": 1.0719, "lr": 1.2918549056006868e-09, "epoch": 4.976982097186701, "percentage": 99.54, "elapsed_time": "10:01:27", "remaining_time": "0:02:46"} +{"current_steps": 1947, "total_steps": 1955, "loss": 1.0187, "lr": 1.0207294762987208e-09, "epoch": 4.979539641943734, "percentage": 99.59, "elapsed_time": "10:01:45", "remaining_time": "0:02:28"} +{"current_steps": 1948, "total_steps": 1955, "loss": 1.0355, "lr": 7.814991213594969e-10, "epoch": 4.982097186700767, "percentage": 99.64, "elapsed_time": "10:02:04", "remaining_time": "0:02:09"} +{"current_steps": 1949, "total_steps": 1955, "loss": 1.0646, "lr": 5.741646038892601e-10, "epoch": 4.9846547314578, "percentage": 99.69, "elapsed_time": "10:02:22", "remaining_time": "0:01:51"} +{"current_steps": 1950, "total_steps": 1955, "loss": 1.0612, "lr": 3.9872658525008654e-10, "epoch": 4.987212276214834, "percentage": 99.74, "elapsed_time": "10:02:40", "remaining_time": "0:01:32"} +{"current_steps": 1951, "total_steps": 1955, "loss": 1.0244, "lr": 2.5518562505988387e-10, "epoch": 4.989769820971867, "percentage": 99.8, "elapsed_time": "10:02:59", "remaining_time": "0:01:14"} +{"current_steps": 1952, "total_steps": 1955, "loss": 1.0499, "lr": 1.435421811901705e-10, "epoch": 4.9923273657289, "percentage": 99.85, "elapsed_time": "10:03:18", "remaining_time": "0:00:55"} +{"current_steps": 1953, "total_steps": 1955, "loss": 1.0107, "lr": 6.37966097649656e-11, "epoch": 4.994884910485934, "percentage": 99.9, "elapsed_time": "10:03:36", "remaining_time": "0:00:37"} +{"current_steps": 1954, "total_steps": 1955, "loss": 1.0177, "lr": 1.5949165159678813e-11, "epoch": 4.997442455242966, "percentage": 99.95, "elapsed_time": "10:03:54", "remaining_time": "0:00:18"} +{"current_steps": 1955, "total_steps": 1955, "loss": 1.0412, "lr": 0.0, "epoch": 5.0, "percentage": 100.0, "elapsed_time": "10:04:13", "remaining_time": "0:00:00"} +{"current_steps": 1955, "total_steps": 1955, "epoch": 5.0, "percentage": 100.0, "elapsed_time": "10:04:32", "remaining_time": "0:00:00"} diff --git a/trainer_state.json b/trainer_state.json new file mode 100644 index 0000000..f3d7ff0 --- /dev/null +++ b/trainer_state.json @@ -0,0 +1,13727 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 5.0, + "eval_steps": 500, + "global_step": 1955, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0025575447570332483, + "grad_norm": 2.9562953287143814, + "learning_rate": 1.0204081632653061e-07, + "loss": 1.5213, + "step": 1 + }, + { + "epoch": 0.005115089514066497, + "grad_norm": 2.957074977095288, + "learning_rate": 2.0408163265306121e-07, + "loss": 1.4742, + "step": 2 + }, + { + "epoch": 0.0076726342710997444, + "grad_norm": 3.0195943091843875, + "learning_rate": 3.0612244897959183e-07, + "loss": 1.4948, + "step": 3 + }, + { + "epoch": 0.010230179028132993, + "grad_norm": 2.955601772064438, + "learning_rate": 4.0816326530612243e-07, + "loss": 1.4841, + "step": 4 + }, + { + "epoch": 0.01278772378516624, + "grad_norm": 2.9724831486674437, + "learning_rate": 5.102040816326531e-07, + "loss": 1.494, + "step": 5 + }, + { + "epoch": 0.015345268542199489, + "grad_norm": 2.9891106307727844, + "learning_rate": 6.122448979591837e-07, + "loss": 1.5063, + "step": 6 + }, + { + "epoch": 0.017902813299232736, + "grad_norm": 2.9117659939526934, + "learning_rate": 7.142857142857143e-07, + "loss": 1.4791, + "step": 7 + }, + { + "epoch": 0.020460358056265986, + "grad_norm": 2.90337467558492, + "learning_rate": 8.163265306122449e-07, + "loss": 1.4882, + "step": 8 + }, + { + "epoch": 0.023017902813299233, + "grad_norm": 2.872454943973889, + "learning_rate": 9.183673469387756e-07, + "loss": 1.5228, + "step": 9 + }, + { + "epoch": 0.02557544757033248, + "grad_norm": 2.8716382761013697, + "learning_rate": 1.0204081632653063e-06, + "loss": 1.5153, + "step": 10 + }, + { + "epoch": 0.028132992327365727, + "grad_norm": 2.6360453721548796, + "learning_rate": 1.122448979591837e-06, + "loss": 1.5046, + "step": 11 + }, + { + "epoch": 0.030690537084398978, + "grad_norm": 2.659598355667714, + "learning_rate": 1.2244897959183673e-06, + "loss": 1.4692, + "step": 12 + }, + { + "epoch": 0.03324808184143223, + "grad_norm": 2.5783985084733634, + "learning_rate": 1.3265306122448982e-06, + "loss": 1.5132, + "step": 13 + }, + { + "epoch": 0.03580562659846547, + "grad_norm": 2.049467948851294, + "learning_rate": 1.4285714285714286e-06, + "loss": 1.4473, + "step": 14 + }, + { + "epoch": 0.03836317135549872, + "grad_norm": 2.025687601863481, + "learning_rate": 1.5306122448979593e-06, + "loss": 1.4862, + "step": 15 + }, + { + "epoch": 0.04092071611253197, + "grad_norm": 1.9646128905085123, + "learning_rate": 1.6326530612244897e-06, + "loss": 1.4457, + "step": 16 + }, + { + "epoch": 0.043478260869565216, + "grad_norm": 1.9191066854928391, + "learning_rate": 1.7346938775510206e-06, + "loss": 1.4699, + "step": 17 + }, + { + "epoch": 0.04603580562659847, + "grad_norm": 1.8754519688248819, + "learning_rate": 1.8367346938775512e-06, + "loss": 1.4732, + "step": 18 + }, + { + "epoch": 0.04859335038363171, + "grad_norm": 1.3012814301728521, + "learning_rate": 1.938775510204082e-06, + "loss": 1.4474, + "step": 19 + }, + { + "epoch": 0.05115089514066496, + "grad_norm": 1.5591794317282675, + "learning_rate": 2.0408163265306125e-06, + "loss": 1.4366, + "step": 20 + }, + { + "epoch": 0.05370843989769821, + "grad_norm": 1.6331695295496573, + "learning_rate": 2.1428571428571427e-06, + "loss": 1.4239, + "step": 21 + }, + { + "epoch": 0.056265984654731455, + "grad_norm": 1.5966167406736418, + "learning_rate": 2.244897959183674e-06, + "loss": 1.4529, + "step": 22 + }, + { + "epoch": 0.058823529411764705, + "grad_norm": 1.4373932142776844, + "learning_rate": 2.3469387755102044e-06, + "loss": 1.4073, + "step": 23 + }, + { + "epoch": 0.061381074168797956, + "grad_norm": 1.3276581212533325, + "learning_rate": 2.4489795918367347e-06, + "loss": 1.4322, + "step": 24 + }, + { + "epoch": 0.0639386189258312, + "grad_norm": 1.0731846564216336, + "learning_rate": 2.5510204081632657e-06, + "loss": 1.4073, + "step": 25 + }, + { + "epoch": 0.06649616368286446, + "grad_norm": 1.238876393241825, + "learning_rate": 2.6530612244897964e-06, + "loss": 1.3989, + "step": 26 + }, + { + "epoch": 0.06905370843989769, + "grad_norm": 1.2629836696521368, + "learning_rate": 2.7551020408163266e-06, + "loss": 1.3804, + "step": 27 + }, + { + "epoch": 0.07161125319693094, + "grad_norm": 1.2621269804030673, + "learning_rate": 2.8571428571428573e-06, + "loss": 1.4052, + "step": 28 + }, + { + "epoch": 0.0741687979539642, + "grad_norm": 1.213554923104541, + "learning_rate": 2.959183673469388e-06, + "loss": 1.382, + "step": 29 + }, + { + "epoch": 0.07672634271099744, + "grad_norm": 1.1050135745029126, + "learning_rate": 3.0612244897959185e-06, + "loss": 1.3963, + "step": 30 + }, + { + "epoch": 0.0792838874680307, + "grad_norm": 0.9715895015186999, + "learning_rate": 3.1632653061224496e-06, + "loss": 1.3901, + "step": 31 + }, + { + "epoch": 0.08184143222506395, + "grad_norm": 0.8825544463358136, + "learning_rate": 3.2653061224489794e-06, + "loss": 1.363, + "step": 32 + }, + { + "epoch": 0.08439897698209718, + "grad_norm": 0.763331787510912, + "learning_rate": 3.3673469387755105e-06, + "loss": 1.3952, + "step": 33 + }, + { + "epoch": 0.08695652173913043, + "grad_norm": 0.6415924907748566, + "learning_rate": 3.469387755102041e-06, + "loss": 1.4103, + "step": 34 + }, + { + "epoch": 0.08951406649616368, + "grad_norm": 0.6967464439584855, + "learning_rate": 3.5714285714285718e-06, + "loss": 1.3459, + "step": 35 + }, + { + "epoch": 0.09207161125319693, + "grad_norm": 0.8908075465264809, + "learning_rate": 3.6734693877551024e-06, + "loss": 1.3561, + "step": 36 + }, + { + "epoch": 0.09462915601023018, + "grad_norm": 0.8747631888662198, + "learning_rate": 3.7755102040816327e-06, + "loss": 1.3694, + "step": 37 + }, + { + "epoch": 0.09718670076726342, + "grad_norm": 0.7588066982535068, + "learning_rate": 3.877551020408164e-06, + "loss": 1.3415, + "step": 38 + }, + { + "epoch": 0.09974424552429667, + "grad_norm": 0.6706725405381729, + "learning_rate": 3.979591836734694e-06, + "loss": 1.3571, + "step": 39 + }, + { + "epoch": 0.10230179028132992, + "grad_norm": 0.5639131487544589, + "learning_rate": 4.081632653061225e-06, + "loss": 1.2958, + "step": 40 + }, + { + "epoch": 0.10485933503836317, + "grad_norm": 0.5328548178092433, + "learning_rate": 4.183673469387755e-06, + "loss": 1.3227, + "step": 41 + }, + { + "epoch": 0.10741687979539642, + "grad_norm": 0.5772938103545907, + "learning_rate": 4.2857142857142855e-06, + "loss": 1.3212, + "step": 42 + }, + { + "epoch": 0.10997442455242967, + "grad_norm": 0.5924144289895023, + "learning_rate": 4.3877551020408165e-06, + "loss": 1.2923, + "step": 43 + }, + { + "epoch": 0.11253196930946291, + "grad_norm": 0.5904192016449847, + "learning_rate": 4.489795918367348e-06, + "loss": 1.3017, + "step": 44 + }, + { + "epoch": 0.11508951406649616, + "grad_norm": 0.5201728742501257, + "learning_rate": 4.591836734693878e-06, + "loss": 1.3013, + "step": 45 + }, + { + "epoch": 0.11764705882352941, + "grad_norm": 0.46964109225255846, + "learning_rate": 4.693877551020409e-06, + "loss": 1.3225, + "step": 46 + }, + { + "epoch": 0.12020460358056266, + "grad_norm": 0.4610564873689024, + "learning_rate": 4.795918367346939e-06, + "loss": 1.3221, + "step": 47 + }, + { + "epoch": 0.12276214833759591, + "grad_norm": 0.37401415057004755, + "learning_rate": 4.897959183673469e-06, + "loss": 1.291, + "step": 48 + }, + { + "epoch": 0.12531969309462915, + "grad_norm": 0.4024917885366837, + "learning_rate": 5e-06, + "loss": 1.304, + "step": 49 + }, + { + "epoch": 0.1278772378516624, + "grad_norm": 0.4489807884401599, + "learning_rate": 5.1020408163265315e-06, + "loss": 1.2735, + "step": 50 + }, + { + "epoch": 0.13043478260869565, + "grad_norm": 0.4407916383861822, + "learning_rate": 5.204081632653062e-06, + "loss": 1.2796, + "step": 51 + }, + { + "epoch": 0.1329923273657289, + "grad_norm": 0.3755938325328462, + "learning_rate": 5.306122448979593e-06, + "loss": 1.2591, + "step": 52 + }, + { + "epoch": 0.13554987212276215, + "grad_norm": 0.3579863910557438, + "learning_rate": 5.408163265306123e-06, + "loss": 1.2592, + "step": 53 + }, + { + "epoch": 0.13810741687979539, + "grad_norm": 0.36590626112420677, + "learning_rate": 5.510204081632653e-06, + "loss": 1.2958, + "step": 54 + }, + { + "epoch": 0.14066496163682865, + "grad_norm": 0.33001691546050266, + "learning_rate": 5.6122448979591834e-06, + "loss": 1.3075, + "step": 55 + }, + { + "epoch": 0.1432225063938619, + "grad_norm": 0.3442007478055813, + "learning_rate": 5.7142857142857145e-06, + "loss": 1.2947, + "step": 56 + }, + { + "epoch": 0.14578005115089515, + "grad_norm": 0.31990892696809964, + "learning_rate": 5.816326530612246e-06, + "loss": 1.2764, + "step": 57 + }, + { + "epoch": 0.1483375959079284, + "grad_norm": 0.29610380122523694, + "learning_rate": 5.918367346938776e-06, + "loss": 1.2911, + "step": 58 + }, + { + "epoch": 0.15089514066496162, + "grad_norm": 0.2800791150063646, + "learning_rate": 6.020408163265307e-06, + "loss": 1.2564, + "step": 59 + }, + { + "epoch": 0.1534526854219949, + "grad_norm": 0.24883718330799656, + "learning_rate": 6.122448979591837e-06, + "loss": 1.2802, + "step": 60 + }, + { + "epoch": 0.15601023017902813, + "grad_norm": 0.2689712120019464, + "learning_rate": 6.224489795918368e-06, + "loss": 1.2673, + "step": 61 + }, + { + "epoch": 0.1585677749360614, + "grad_norm": 0.274815756586181, + "learning_rate": 6.326530612244899e-06, + "loss": 1.2984, + "step": 62 + }, + { + "epoch": 0.16112531969309463, + "grad_norm": 0.24511037552821183, + "learning_rate": 6.4285714285714295e-06, + "loss": 1.2811, + "step": 63 + }, + { + "epoch": 0.1636828644501279, + "grad_norm": 0.2585673848534428, + "learning_rate": 6.530612244897959e-06, + "loss": 1.2752, + "step": 64 + }, + { + "epoch": 0.16624040920716113, + "grad_norm": 0.24726403234369912, + "learning_rate": 6.63265306122449e-06, + "loss": 1.2667, + "step": 65 + }, + { + "epoch": 0.16879795396419436, + "grad_norm": 0.2436298587258828, + "learning_rate": 6.734693877551021e-06, + "loss": 1.2279, + "step": 66 + }, + { + "epoch": 0.17135549872122763, + "grad_norm": 0.24268854362503556, + "learning_rate": 6.836734693877551e-06, + "loss": 1.2517, + "step": 67 + }, + { + "epoch": 0.17391304347826086, + "grad_norm": 0.2554114966704654, + "learning_rate": 6.938775510204082e-06, + "loss": 1.2277, + "step": 68 + }, + { + "epoch": 0.17647058823529413, + "grad_norm": 0.19396655425261772, + "learning_rate": 7.0408163265306125e-06, + "loss": 1.2119, + "step": 69 + }, + { + "epoch": 0.17902813299232737, + "grad_norm": 0.19737858451227866, + "learning_rate": 7.1428571428571436e-06, + "loss": 1.2234, + "step": 70 + }, + { + "epoch": 0.1815856777493606, + "grad_norm": 0.21921420023094632, + "learning_rate": 7.244897959183675e-06, + "loss": 1.2332, + "step": 71 + }, + { + "epoch": 0.18414322250639387, + "grad_norm": 0.20918098542181962, + "learning_rate": 7.346938775510205e-06, + "loss": 1.2476, + "step": 72 + }, + { + "epoch": 0.1867007672634271, + "grad_norm": 0.20509174967917101, + "learning_rate": 7.448979591836736e-06, + "loss": 1.2809, + "step": 73 + }, + { + "epoch": 0.18925831202046037, + "grad_norm": 0.20157359497446836, + "learning_rate": 7.551020408163265e-06, + "loss": 1.228, + "step": 74 + }, + { + "epoch": 0.1918158567774936, + "grad_norm": 0.2030800764293693, + "learning_rate": 7.653061224489796e-06, + "loss": 1.2256, + "step": 75 + }, + { + "epoch": 0.19437340153452684, + "grad_norm": 0.22457010766510804, + "learning_rate": 7.755102040816327e-06, + "loss": 1.2277, + "step": 76 + }, + { + "epoch": 0.1969309462915601, + "grad_norm": 0.192665701729149, + "learning_rate": 7.857142857142858e-06, + "loss": 1.2383, + "step": 77 + }, + { + "epoch": 0.19948849104859334, + "grad_norm": 0.19152870646216824, + "learning_rate": 7.959183673469388e-06, + "loss": 1.2176, + "step": 78 + }, + { + "epoch": 0.2020460358056266, + "grad_norm": 0.19648868962913657, + "learning_rate": 8.06122448979592e-06, + "loss": 1.2119, + "step": 79 + }, + { + "epoch": 0.20460358056265984, + "grad_norm": 0.20421924729381744, + "learning_rate": 8.16326530612245e-06, + "loss": 1.2285, + "step": 80 + }, + { + "epoch": 0.2071611253196931, + "grad_norm": 0.18789494211610624, + "learning_rate": 8.26530612244898e-06, + "loss": 1.212, + "step": 81 + }, + { + "epoch": 0.20971867007672634, + "grad_norm": 0.2092465434891838, + "learning_rate": 8.36734693877551e-06, + "loss": 1.2091, + "step": 82 + }, + { + "epoch": 0.21227621483375958, + "grad_norm": 0.20515739317387488, + "learning_rate": 8.469387755102042e-06, + "loss": 1.2235, + "step": 83 + }, + { + "epoch": 0.21483375959079284, + "grad_norm": 0.20337541481235902, + "learning_rate": 8.571428571428571e-06, + "loss": 1.2183, + "step": 84 + }, + { + "epoch": 0.21739130434782608, + "grad_norm": 0.1802747600910944, + "learning_rate": 8.673469387755103e-06, + "loss": 1.2682, + "step": 85 + }, + { + "epoch": 0.21994884910485935, + "grad_norm": 0.18621854583865133, + "learning_rate": 8.775510204081633e-06, + "loss": 1.2185, + "step": 86 + }, + { + "epoch": 0.22250639386189258, + "grad_norm": 0.1982015852510532, + "learning_rate": 8.877551020408163e-06, + "loss": 1.2255, + "step": 87 + }, + { + "epoch": 0.22506393861892582, + "grad_norm": 0.16662352110298306, + "learning_rate": 8.979591836734695e-06, + "loss": 1.2494, + "step": 88 + }, + { + "epoch": 0.22762148337595908, + "grad_norm": 0.20033046650600175, + "learning_rate": 9.081632653061225e-06, + "loss": 1.2099, + "step": 89 + }, + { + "epoch": 0.23017902813299232, + "grad_norm": 0.19772324219486356, + "learning_rate": 9.183673469387756e-06, + "loss": 1.2398, + "step": 90 + }, + { + "epoch": 0.23273657289002558, + "grad_norm": 0.17627317445341492, + "learning_rate": 9.285714285714288e-06, + "loss": 1.2359, + "step": 91 + }, + { + "epoch": 0.23529411764705882, + "grad_norm": 0.2000488853272216, + "learning_rate": 9.387755102040818e-06, + "loss": 1.2377, + "step": 92 + }, + { + "epoch": 0.23785166240409208, + "grad_norm": 0.18512197433512323, + "learning_rate": 9.489795918367348e-06, + "loss": 1.2204, + "step": 93 + }, + { + "epoch": 0.24040920716112532, + "grad_norm": 0.21513166102817025, + "learning_rate": 9.591836734693878e-06, + "loss": 1.2414, + "step": 94 + }, + { + "epoch": 0.24296675191815856, + "grad_norm": 0.18497143829689744, + "learning_rate": 9.693877551020408e-06, + "loss": 1.1886, + "step": 95 + }, + { + "epoch": 0.24552429667519182, + "grad_norm": 0.2045160941166778, + "learning_rate": 9.795918367346939e-06, + "loss": 1.2282, + "step": 96 + }, + { + "epoch": 0.24808184143222506, + "grad_norm": 0.19235390291361945, + "learning_rate": 9.89795918367347e-06, + "loss": 1.2296, + "step": 97 + }, + { + "epoch": 0.2506393861892583, + "grad_norm": 0.21361682704534116, + "learning_rate": 1e-05, + "loss": 1.2001, + "step": 98 + }, + { + "epoch": 0.2531969309462916, + "grad_norm": 0.17195039095079748, + "learning_rate": 1.0102040816326531e-05, + "loss": 1.1982, + "step": 99 + }, + { + "epoch": 0.2557544757033248, + "grad_norm": 0.19990564719956827, + "learning_rate": 1.0204081632653063e-05, + "loss": 1.1914, + "step": 100 + }, + { + "epoch": 0.25831202046035806, + "grad_norm": 0.18409790697604989, + "learning_rate": 1.0306122448979591e-05, + "loss": 1.2125, + "step": 101 + }, + { + "epoch": 0.2608695652173913, + "grad_norm": 0.2010207370883571, + "learning_rate": 1.0408163265306123e-05, + "loss": 1.2192, + "step": 102 + }, + { + "epoch": 0.26342710997442453, + "grad_norm": 0.24154477700771743, + "learning_rate": 1.0510204081632654e-05, + "loss": 1.189, + "step": 103 + }, + { + "epoch": 0.2659846547314578, + "grad_norm": 0.19539728509614046, + "learning_rate": 1.0612244897959186e-05, + "loss": 1.1791, + "step": 104 + }, + { + "epoch": 0.26854219948849106, + "grad_norm": 0.2809106862940922, + "learning_rate": 1.0714285714285714e-05, + "loss": 1.1866, + "step": 105 + }, + { + "epoch": 0.2710997442455243, + "grad_norm": 0.18139724102505636, + "learning_rate": 1.0816326530612246e-05, + "loss": 1.1772, + "step": 106 + }, + { + "epoch": 0.27365728900255754, + "grad_norm": 0.2944051456882037, + "learning_rate": 1.0918367346938776e-05, + "loss": 1.2043, + "step": 107 + }, + { + "epoch": 0.27621483375959077, + "grad_norm": 0.18837903247625115, + "learning_rate": 1.1020408163265306e-05, + "loss": 1.1831, + "step": 108 + }, + { + "epoch": 0.27877237851662406, + "grad_norm": 0.24490442547474545, + "learning_rate": 1.1122448979591838e-05, + "loss": 1.2072, + "step": 109 + }, + { + "epoch": 0.2813299232736573, + "grad_norm": 0.18434497663730703, + "learning_rate": 1.1224489795918367e-05, + "loss": 1.2077, + "step": 110 + }, + { + "epoch": 0.28388746803069054, + "grad_norm": 0.2023995573064504, + "learning_rate": 1.1326530612244899e-05, + "loss": 1.2113, + "step": 111 + }, + { + "epoch": 0.2864450127877238, + "grad_norm": 0.2232631442545948, + "learning_rate": 1.1428571428571429e-05, + "loss": 1.199, + "step": 112 + }, + { + "epoch": 0.289002557544757, + "grad_norm": 0.19538302737828642, + "learning_rate": 1.1530612244897961e-05, + "loss": 1.1707, + "step": 113 + }, + { + "epoch": 0.2915601023017903, + "grad_norm": 0.2076715375956596, + "learning_rate": 1.1632653061224491e-05, + "loss": 1.1996, + "step": 114 + }, + { + "epoch": 0.29411764705882354, + "grad_norm": 0.2291360055402482, + "learning_rate": 1.1734693877551021e-05, + "loss": 1.1818, + "step": 115 + }, + { + "epoch": 0.2966751918158568, + "grad_norm": 0.22247638512056675, + "learning_rate": 1.1836734693877552e-05, + "loss": 1.1899, + "step": 116 + }, + { + "epoch": 0.29923273657289, + "grad_norm": 0.23191223669707034, + "learning_rate": 1.1938775510204084e-05, + "loss": 1.1822, + "step": 117 + }, + { + "epoch": 0.30179028132992325, + "grad_norm": 0.2705601798192164, + "learning_rate": 1.2040816326530614e-05, + "loss": 1.2338, + "step": 118 + }, + { + "epoch": 0.30434782608695654, + "grad_norm": 0.2474348689952774, + "learning_rate": 1.2142857142857142e-05, + "loss": 1.1949, + "step": 119 + }, + { + "epoch": 0.3069053708439898, + "grad_norm": 0.20663103569329155, + "learning_rate": 1.2244897959183674e-05, + "loss": 1.2006, + "step": 120 + }, + { + "epoch": 0.309462915601023, + "grad_norm": 0.2605886951080686, + "learning_rate": 1.2346938775510204e-05, + "loss": 1.2164, + "step": 121 + }, + { + "epoch": 0.31202046035805625, + "grad_norm": 0.2181457069786774, + "learning_rate": 1.2448979591836736e-05, + "loss": 1.1664, + "step": 122 + }, + { + "epoch": 0.3145780051150895, + "grad_norm": 0.3157714554732324, + "learning_rate": 1.2551020408163267e-05, + "loss": 1.1875, + "step": 123 + }, + { + "epoch": 0.3171355498721228, + "grad_norm": 0.2775822065984946, + "learning_rate": 1.2653061224489798e-05, + "loss": 1.1822, + "step": 124 + }, + { + "epoch": 0.319693094629156, + "grad_norm": 0.25536478737195145, + "learning_rate": 1.2755102040816327e-05, + "loss": 1.1623, + "step": 125 + }, + { + "epoch": 0.32225063938618925, + "grad_norm": 0.38468633827614823, + "learning_rate": 1.2857142857142859e-05, + "loss": 1.2188, + "step": 126 + }, + { + "epoch": 0.3248081841432225, + "grad_norm": 0.287878457956867, + "learning_rate": 1.2959183673469389e-05, + "loss": 1.1605, + "step": 127 + }, + { + "epoch": 0.3273657289002558, + "grad_norm": 0.3074138761550165, + "learning_rate": 1.3061224489795918e-05, + "loss": 1.1804, + "step": 128 + }, + { + "epoch": 0.329923273657289, + "grad_norm": 0.2866479879938462, + "learning_rate": 1.316326530612245e-05, + "loss": 1.1733, + "step": 129 + }, + { + "epoch": 0.33248081841432225, + "grad_norm": 0.32148222861310927, + "learning_rate": 1.326530612244898e-05, + "loss": 1.2038, + "step": 130 + }, + { + "epoch": 0.3350383631713555, + "grad_norm": 0.22529099781853634, + "learning_rate": 1.3367346938775512e-05, + "loss": 1.1673, + "step": 131 + }, + { + "epoch": 0.3375959079283887, + "grad_norm": 0.2609715115506065, + "learning_rate": 1.3469387755102042e-05, + "loss": 1.2253, + "step": 132 + }, + { + "epoch": 0.340153452685422, + "grad_norm": 0.3063675380420315, + "learning_rate": 1.3571428571428574e-05, + "loss": 1.198, + "step": 133 + }, + { + "epoch": 0.34271099744245526, + "grad_norm": 0.22973985204151504, + "learning_rate": 1.3673469387755102e-05, + "loss": 1.1671, + "step": 134 + }, + { + "epoch": 0.3452685421994885, + "grad_norm": 0.20757849081036897, + "learning_rate": 1.3775510204081634e-05, + "loss": 1.2261, + "step": 135 + }, + { + "epoch": 0.34782608695652173, + "grad_norm": 0.239418122200512, + "learning_rate": 1.3877551020408165e-05, + "loss": 1.1854, + "step": 136 + }, + { + "epoch": 0.35038363171355497, + "grad_norm": 0.2373110786708978, + "learning_rate": 1.3979591836734696e-05, + "loss": 1.1972, + "step": 137 + }, + { + "epoch": 0.35294117647058826, + "grad_norm": 0.21952289616465737, + "learning_rate": 1.4081632653061225e-05, + "loss": 1.1841, + "step": 138 + }, + { + "epoch": 0.3554987212276215, + "grad_norm": 0.24365294125993373, + "learning_rate": 1.4183673469387755e-05, + "loss": 1.1582, + "step": 139 + }, + { + "epoch": 0.35805626598465473, + "grad_norm": 0.28057490691552545, + "learning_rate": 1.4285714285714287e-05, + "loss": 1.217, + "step": 140 + }, + { + "epoch": 0.36061381074168797, + "grad_norm": 0.2858217046456728, + "learning_rate": 1.4387755102040817e-05, + "loss": 1.1303, + "step": 141 + }, + { + "epoch": 0.3631713554987212, + "grad_norm": 0.25139449669336106, + "learning_rate": 1.448979591836735e-05, + "loss": 1.1808, + "step": 142 + }, + { + "epoch": 0.3657289002557545, + "grad_norm": 0.264472202641012, + "learning_rate": 1.4591836734693878e-05, + "loss": 1.2083, + "step": 143 + }, + { + "epoch": 0.36828644501278773, + "grad_norm": 0.3298441325989216, + "learning_rate": 1.469387755102041e-05, + "loss": 1.1761, + "step": 144 + }, + { + "epoch": 0.37084398976982097, + "grad_norm": 0.26597491650333793, + "learning_rate": 1.479591836734694e-05, + "loss": 1.1887, + "step": 145 + }, + { + "epoch": 0.3734015345268542, + "grad_norm": 0.21965241642758554, + "learning_rate": 1.4897959183673472e-05, + "loss": 1.1735, + "step": 146 + }, + { + "epoch": 0.37595907928388744, + "grad_norm": 0.25221490386812995, + "learning_rate": 1.5000000000000002e-05, + "loss": 1.2087, + "step": 147 + }, + { + "epoch": 0.37851662404092073, + "grad_norm": 0.2393195787833758, + "learning_rate": 1.510204081632653e-05, + "loss": 1.1688, + "step": 148 + }, + { + "epoch": 0.38107416879795397, + "grad_norm": 0.2753716321794055, + "learning_rate": 1.5204081632653063e-05, + "loss": 1.1553, + "step": 149 + }, + { + "epoch": 0.3836317135549872, + "grad_norm": 0.21247095213573716, + "learning_rate": 1.530612244897959e-05, + "loss": 1.1664, + "step": 150 + }, + { + "epoch": 0.38618925831202044, + "grad_norm": 0.2339506540148249, + "learning_rate": 1.5408163265306123e-05, + "loss": 1.17, + "step": 151 + }, + { + "epoch": 0.3887468030690537, + "grad_norm": 0.26795973798728384, + "learning_rate": 1.5510204081632655e-05, + "loss": 1.175, + "step": 152 + }, + { + "epoch": 0.391304347826087, + "grad_norm": 0.2583891394938216, + "learning_rate": 1.5612244897959187e-05, + "loss": 1.1646, + "step": 153 + }, + { + "epoch": 0.3938618925831202, + "grad_norm": 0.2686074219032705, + "learning_rate": 1.5714285714285715e-05, + "loss": 1.1698, + "step": 154 + }, + { + "epoch": 0.39641943734015345, + "grad_norm": 0.2286828659609244, + "learning_rate": 1.5816326530612247e-05, + "loss": 1.1821, + "step": 155 + }, + { + "epoch": 0.3989769820971867, + "grad_norm": 0.2476552598028532, + "learning_rate": 1.5918367346938776e-05, + "loss": 1.1864, + "step": 156 + }, + { + "epoch": 0.40153452685422, + "grad_norm": 0.3028181273764634, + "learning_rate": 1.6020408163265308e-05, + "loss": 1.1999, + "step": 157 + }, + { + "epoch": 0.4040920716112532, + "grad_norm": 0.3392361145642557, + "learning_rate": 1.612244897959184e-05, + "loss": 1.1928, + "step": 158 + }, + { + "epoch": 0.40664961636828645, + "grad_norm": 0.23749781222599484, + "learning_rate": 1.6224489795918368e-05, + "loss": 1.1455, + "step": 159 + }, + { + "epoch": 0.4092071611253197, + "grad_norm": 0.27301888454203366, + "learning_rate": 1.63265306122449e-05, + "loss": 1.1657, + "step": 160 + }, + { + "epoch": 0.4117647058823529, + "grad_norm": 0.3292983030079545, + "learning_rate": 1.642857142857143e-05, + "loss": 1.1598, + "step": 161 + }, + { + "epoch": 0.4143222506393862, + "grad_norm": 0.5337081556820991, + "learning_rate": 1.653061224489796e-05, + "loss": 1.1565, + "step": 162 + }, + { + "epoch": 0.41687979539641945, + "grad_norm": 0.45293962184506315, + "learning_rate": 1.6632653061224492e-05, + "loss": 1.1594, + "step": 163 + }, + { + "epoch": 0.4194373401534527, + "grad_norm": 0.4328651118135115, + "learning_rate": 1.673469387755102e-05, + "loss": 1.1699, + "step": 164 + }, + { + "epoch": 0.4219948849104859, + "grad_norm": 0.3432892281672048, + "learning_rate": 1.6836734693877553e-05, + "loss": 1.1459, + "step": 165 + }, + { + "epoch": 0.42455242966751916, + "grad_norm": 0.2730126122517006, + "learning_rate": 1.6938775510204085e-05, + "loss": 1.1667, + "step": 166 + }, + { + "epoch": 0.42710997442455245, + "grad_norm": 0.306031348484845, + "learning_rate": 1.7040816326530613e-05, + "loss": 1.1642, + "step": 167 + }, + { + "epoch": 0.4296675191815857, + "grad_norm": 0.3825044478602271, + "learning_rate": 1.7142857142857142e-05, + "loss": 1.1741, + "step": 168 + }, + { + "epoch": 0.4322250639386189, + "grad_norm": 0.3319946152258406, + "learning_rate": 1.7244897959183674e-05, + "loss": 1.1722, + "step": 169 + }, + { + "epoch": 0.43478260869565216, + "grad_norm": 0.21336378841100664, + "learning_rate": 1.7346938775510206e-05, + "loss": 1.1513, + "step": 170 + }, + { + "epoch": 0.4373401534526854, + "grad_norm": 0.2584789415828712, + "learning_rate": 1.7448979591836738e-05, + "loss": 1.1641, + "step": 171 + }, + { + "epoch": 0.4398976982097187, + "grad_norm": 0.3035587738680554, + "learning_rate": 1.7551020408163266e-05, + "loss": 1.1615, + "step": 172 + }, + { + "epoch": 0.4424552429667519, + "grad_norm": 0.3150232598756255, + "learning_rate": 1.7653061224489798e-05, + "loss": 1.1808, + "step": 173 + }, + { + "epoch": 0.44501278772378516, + "grad_norm": 0.3546603428284164, + "learning_rate": 1.7755102040816327e-05, + "loss": 1.1897, + "step": 174 + }, + { + "epoch": 0.4475703324808184, + "grad_norm": 0.34889584036889865, + "learning_rate": 1.785714285714286e-05, + "loss": 1.161, + "step": 175 + }, + { + "epoch": 0.45012787723785164, + "grad_norm": 0.29454291072360017, + "learning_rate": 1.795918367346939e-05, + "loss": 1.1912, + "step": 176 + }, + { + "epoch": 0.45268542199488493, + "grad_norm": 0.40100102719618086, + "learning_rate": 1.806122448979592e-05, + "loss": 1.1648, + "step": 177 + }, + { + "epoch": 0.45524296675191817, + "grad_norm": 0.4059968451876151, + "learning_rate": 1.816326530612245e-05, + "loss": 1.1639, + "step": 178 + }, + { + "epoch": 0.4578005115089514, + "grad_norm": 0.37064317174148603, + "learning_rate": 1.826530612244898e-05, + "loss": 1.1479, + "step": 179 + }, + { + "epoch": 0.46035805626598464, + "grad_norm": 0.37396234314183074, + "learning_rate": 1.836734693877551e-05, + "loss": 1.1568, + "step": 180 + }, + { + "epoch": 0.4629156010230179, + "grad_norm": 0.45716162446267244, + "learning_rate": 1.8469387755102043e-05, + "loss": 1.1822, + "step": 181 + }, + { + "epoch": 0.46547314578005117, + "grad_norm": 0.36679016991278174, + "learning_rate": 1.8571428571428575e-05, + "loss": 1.1579, + "step": 182 + }, + { + "epoch": 0.4680306905370844, + "grad_norm": 0.35413758345531376, + "learning_rate": 1.8673469387755104e-05, + "loss": 1.17, + "step": 183 + }, + { + "epoch": 0.47058823529411764, + "grad_norm": 0.357746233671763, + "learning_rate": 1.8775510204081636e-05, + "loss": 1.1657, + "step": 184 + }, + { + "epoch": 0.4731457800511509, + "grad_norm": 0.30987744196180955, + "learning_rate": 1.8877551020408164e-05, + "loss": 1.1727, + "step": 185 + }, + { + "epoch": 0.47570332480818417, + "grad_norm": 0.3008253966047084, + "learning_rate": 1.8979591836734696e-05, + "loss": 1.1697, + "step": 186 + }, + { + "epoch": 0.4782608695652174, + "grad_norm": 0.26780657180049305, + "learning_rate": 1.9081632653061225e-05, + "loss": 1.1504, + "step": 187 + }, + { + "epoch": 0.48081841432225064, + "grad_norm": 0.3041554719196245, + "learning_rate": 1.9183673469387756e-05, + "loss": 1.1771, + "step": 188 + }, + { + "epoch": 0.4833759590792839, + "grad_norm": 0.28069002744383614, + "learning_rate": 1.928571428571429e-05, + "loss": 1.1576, + "step": 189 + }, + { + "epoch": 0.4859335038363171, + "grad_norm": 0.2468377630882404, + "learning_rate": 1.9387755102040817e-05, + "loss": 1.1597, + "step": 190 + }, + { + "epoch": 0.4884910485933504, + "grad_norm": 0.30179882714603656, + "learning_rate": 1.948979591836735e-05, + "loss": 1.1084, + "step": 191 + }, + { + "epoch": 0.49104859335038364, + "grad_norm": 0.37503237335205775, + "learning_rate": 1.9591836734693877e-05, + "loss": 1.1505, + "step": 192 + }, + { + "epoch": 0.4936061381074169, + "grad_norm": 0.44233118769859003, + "learning_rate": 1.969387755102041e-05, + "loss": 1.1214, + "step": 193 + }, + { + "epoch": 0.4961636828644501, + "grad_norm": 0.5601563317495414, + "learning_rate": 1.979591836734694e-05, + "loss": 1.1197, + "step": 194 + }, + { + "epoch": 0.49872122762148335, + "grad_norm": 0.5687537323400834, + "learning_rate": 1.9897959183673473e-05, + "loss": 1.195, + "step": 195 + }, + { + "epoch": 0.5012787723785166, + "grad_norm": 0.4873305401442818, + "learning_rate": 2e-05, + "loss": 1.1522, + "step": 196 + }, + { + "epoch": 0.5038363171355499, + "grad_norm": 0.3556383631744906, + "learning_rate": 1.9999984050834843e-05, + "loss": 1.1393, + "step": 197 + }, + { + "epoch": 0.5063938618925832, + "grad_norm": 0.3842008338970268, + "learning_rate": 1.999993620339024e-05, + "loss": 1.172, + "step": 198 + }, + { + "epoch": 0.5089514066496164, + "grad_norm": 0.4125503709774857, + "learning_rate": 1.9999856457818812e-05, + "loss": 1.1374, + "step": 199 + }, + { + "epoch": 0.5115089514066496, + "grad_norm": 0.3795131034135739, + "learning_rate": 1.9999744814374943e-05, + "loss": 1.1511, + "step": 200 + }, + { + "epoch": 0.5140664961636828, + "grad_norm": 0.37179713266984565, + "learning_rate": 1.999960127341475e-05, + "loss": 1.1342, + "step": 201 + }, + { + "epoch": 0.5166240409207161, + "grad_norm": 0.36464040525697683, + "learning_rate": 1.9999425835396113e-05, + "loss": 1.1282, + "step": 202 + }, + { + "epoch": 0.5191815856777494, + "grad_norm": 0.2913800358603054, + "learning_rate": 1.9999218500878645e-05, + "loss": 1.1589, + "step": 203 + }, + { + "epoch": 0.5217391304347826, + "grad_norm": 0.2591052993661217, + "learning_rate": 1.9998979270523704e-05, + "loss": 1.1291, + "step": 204 + }, + { + "epoch": 0.5242966751918159, + "grad_norm": 0.337615514356955, + "learning_rate": 1.99987081450944e-05, + "loss": 1.1529, + "step": 205 + }, + { + "epoch": 0.5268542199488491, + "grad_norm": 0.28380607304098043, + "learning_rate": 1.999840512545558e-05, + "loss": 1.1418, + "step": 206 + }, + { + "epoch": 0.5294117647058824, + "grad_norm": 0.2806782620501295, + "learning_rate": 1.9998070212573823e-05, + "loss": 1.1596, + "step": 207 + }, + { + "epoch": 0.5319693094629157, + "grad_norm": 0.3717899217291509, + "learning_rate": 1.9997703407517442e-05, + "loss": 1.1457, + "step": 208 + }, + { + "epoch": 0.5345268542199488, + "grad_norm": 0.322823080295276, + "learning_rate": 1.999730471145649e-05, + "loss": 1.135, + "step": 209 + }, + { + "epoch": 0.5370843989769821, + "grad_norm": 0.45281566817756946, + "learning_rate": 1.999687412566274e-05, + "loss": 1.1286, + "step": 210 + }, + { + "epoch": 0.5396419437340153, + "grad_norm": 0.46661955735695343, + "learning_rate": 1.9996411651509683e-05, + "loss": 1.1488, + "step": 211 + }, + { + "epoch": 0.5421994884910486, + "grad_norm": 0.507809212839347, + "learning_rate": 1.999591729047254e-05, + "loss": 1.1172, + "step": 212 + }, + { + "epoch": 0.5447570332480819, + "grad_norm": 0.5943303015975859, + "learning_rate": 1.9995391044128242e-05, + "loss": 1.162, + "step": 213 + }, + { + "epoch": 0.5473145780051151, + "grad_norm": 0.6223689255513681, + "learning_rate": 1.9994832914155417e-05, + "loss": 1.1111, + "step": 214 + }, + { + "epoch": 0.5498721227621484, + "grad_norm": 0.5867365176554278, + "learning_rate": 1.9994242902334418e-05, + "loss": 1.118, + "step": 215 + }, + { + "epoch": 0.5524296675191815, + "grad_norm": 0.4772422762081861, + "learning_rate": 1.9993621010547277e-05, + "loss": 1.1515, + "step": 216 + }, + { + "epoch": 0.5549872122762148, + "grad_norm": 0.2593743678758839, + "learning_rate": 1.9992967240777727e-05, + "loss": 1.125, + "step": 217 + }, + { + "epoch": 0.5575447570332481, + "grad_norm": 0.37039422327125837, + "learning_rate": 1.9992281595111186e-05, + "loss": 1.0982, + "step": 218 + }, + { + "epoch": 0.5601023017902813, + "grad_norm": 0.43886132426547786, + "learning_rate": 1.9991564075734746e-05, + "loss": 1.121, + "step": 219 + }, + { + "epoch": 0.5626598465473146, + "grad_norm": 0.5436094329934847, + "learning_rate": 1.9990814684937177e-05, + "loss": 1.1759, + "step": 220 + }, + { + "epoch": 0.5652173913043478, + "grad_norm": 0.4981980358292973, + "learning_rate": 1.9990033425108906e-05, + "loss": 1.1183, + "step": 221 + }, + { + "epoch": 0.5677749360613811, + "grad_norm": 0.3910510290404881, + "learning_rate": 1.998922029874203e-05, + "loss": 1.1193, + "step": 222 + }, + { + "epoch": 0.5703324808184144, + "grad_norm": 0.390896851403731, + "learning_rate": 1.9988375308430275e-05, + "loss": 1.1249, + "step": 223 + }, + { + "epoch": 0.5728900255754475, + "grad_norm": 0.6251368420646798, + "learning_rate": 1.9987498456869025e-05, + "loss": 1.1553, + "step": 224 + }, + { + "epoch": 0.5754475703324808, + "grad_norm": 0.5868353015562519, + "learning_rate": 1.9986589746855295e-05, + "loss": 1.1531, + "step": 225 + }, + { + "epoch": 0.578005115089514, + "grad_norm": 0.3394930524201935, + "learning_rate": 1.998564918128771e-05, + "loss": 1.1252, + "step": 226 + }, + { + "epoch": 0.5805626598465473, + "grad_norm": 0.3795419225049313, + "learning_rate": 1.998467676316652e-05, + "loss": 1.1352, + "step": 227 + }, + { + "epoch": 0.5831202046035806, + "grad_norm": 0.5793565342648433, + "learning_rate": 1.9983672495593577e-05, + "loss": 1.1211, + "step": 228 + }, + { + "epoch": 0.5856777493606138, + "grad_norm": 0.4466950715738831, + "learning_rate": 1.9982636381772328e-05, + "loss": 1.14, + "step": 229 + }, + { + "epoch": 0.5882352941176471, + "grad_norm": 0.2812697561922094, + "learning_rate": 1.99815684250078e-05, + "loss": 1.1445, + "step": 230 + }, + { + "epoch": 0.5907928388746803, + "grad_norm": 0.29831474191094953, + "learning_rate": 1.9980468628706606e-05, + "loss": 1.1456, + "step": 231 + }, + { + "epoch": 0.5933503836317136, + "grad_norm": 0.32451111550319006, + "learning_rate": 1.9979336996376896e-05, + "loss": 1.1299, + "step": 232 + }, + { + "epoch": 0.5959079283887468, + "grad_norm": 0.2784540337914921, + "learning_rate": 1.99781735316284e-05, + "loss": 1.1344, + "step": 233 + }, + { + "epoch": 0.59846547314578, + "grad_norm": 0.2539011792274416, + "learning_rate": 1.997697823817237e-05, + "loss": 1.0922, + "step": 234 + }, + { + "epoch": 0.6010230179028133, + "grad_norm": 0.29497931492258306, + "learning_rate": 1.99757511198216e-05, + "loss": 1.1096, + "step": 235 + }, + { + "epoch": 0.6035805626598465, + "grad_norm": 0.35829886118674764, + "learning_rate": 1.997449218049039e-05, + "loss": 1.1395, + "step": 236 + }, + { + "epoch": 0.6061381074168798, + "grad_norm": 0.2898322186591816, + "learning_rate": 1.9973201424194544e-05, + "loss": 1.1245, + "step": 237 + }, + { + "epoch": 0.6086956521739131, + "grad_norm": 0.31225838552243906, + "learning_rate": 1.997187885505136e-05, + "loss": 1.1064, + "step": 238 + }, + { + "epoch": 0.6112531969309463, + "grad_norm": 0.3245053547418433, + "learning_rate": 1.997052447727961e-05, + "loss": 1.1517, + "step": 239 + }, + { + "epoch": 0.6138107416879796, + "grad_norm": 0.3043163987443152, + "learning_rate": 1.996913829519954e-05, + "loss": 1.1178, + "step": 240 + }, + { + "epoch": 0.6163682864450127, + "grad_norm": 0.354199627134913, + "learning_rate": 1.996772031323283e-05, + "loss": 1.1105, + "step": 241 + }, + { + "epoch": 0.618925831202046, + "grad_norm": 0.46016480703794116, + "learning_rate": 1.9966270535902618e-05, + "loss": 1.1373, + "step": 242 + }, + { + "epoch": 0.6214833759590793, + "grad_norm": 0.42309562067979195, + "learning_rate": 1.9964788967833438e-05, + "loss": 1.1335, + "step": 243 + }, + { + "epoch": 0.6240409207161125, + "grad_norm": 0.33036623410100835, + "learning_rate": 1.9963275613751255e-05, + "loss": 1.1791, + "step": 244 + }, + { + "epoch": 0.6265984654731458, + "grad_norm": 0.30006259434395505, + "learning_rate": 1.996173047848341e-05, + "loss": 1.138, + "step": 245 + }, + { + "epoch": 0.629156010230179, + "grad_norm": 0.32437525688350793, + "learning_rate": 1.996015356695863e-05, + "loss": 1.1343, + "step": 246 + }, + { + "epoch": 0.6317135549872123, + "grad_norm": 0.41288631390208946, + "learning_rate": 1.9958544884207e-05, + "loss": 1.127, + "step": 247 + }, + { + "epoch": 0.6342710997442456, + "grad_norm": 0.3784448720424316, + "learning_rate": 1.9956904435359945e-05, + "loss": 1.1735, + "step": 248 + }, + { + "epoch": 0.6368286445012787, + "grad_norm": 0.41447667455758297, + "learning_rate": 1.9955232225650223e-05, + "loss": 1.1535, + "step": 249 + }, + { + "epoch": 0.639386189258312, + "grad_norm": 0.43590429505717426, + "learning_rate": 1.9953528260411912e-05, + "loss": 1.1335, + "step": 250 + }, + { + "epoch": 0.6419437340153452, + "grad_norm": 0.34398920147187817, + "learning_rate": 1.9951792545080368e-05, + "loss": 1.1401, + "step": 251 + }, + { + "epoch": 0.6445012787723785, + "grad_norm": 0.4905800118095281, + "learning_rate": 1.995002508519223e-05, + "loss": 1.1454, + "step": 252 + }, + { + "epoch": 0.6470588235294118, + "grad_norm": 0.5410347720983003, + "learning_rate": 1.9948225886385415e-05, + "loss": 1.124, + "step": 253 + }, + { + "epoch": 0.649616368286445, + "grad_norm": 0.31072613789563314, + "learning_rate": 1.9946394954399055e-05, + "loss": 1.1745, + "step": 254 + }, + { + "epoch": 0.6521739130434783, + "grad_norm": 0.4693794451014107, + "learning_rate": 1.994453229507352e-05, + "loss": 1.1316, + "step": 255 + }, + { + "epoch": 0.6547314578005116, + "grad_norm": 0.5861071321893766, + "learning_rate": 1.9942637914350378e-05, + "loss": 1.1208, + "step": 256 + }, + { + "epoch": 0.6572890025575447, + "grad_norm": 0.4821820368779032, + "learning_rate": 1.9940711818272395e-05, + "loss": 1.156, + "step": 257 + }, + { + "epoch": 0.659846547314578, + "grad_norm": 0.34305622387894413, + "learning_rate": 1.993875401298349e-05, + "loss": 1.1182, + "step": 258 + }, + { + "epoch": 0.6624040920716112, + "grad_norm": 0.32785265355630766, + "learning_rate": 1.993676450472874e-05, + "loss": 1.1447, + "step": 259 + }, + { + "epoch": 0.6649616368286445, + "grad_norm": 0.31389753052349284, + "learning_rate": 1.993474329985434e-05, + "loss": 1.1273, + "step": 260 + }, + { + "epoch": 0.6675191815856778, + "grad_norm": 0.3970391405357666, + "learning_rate": 1.99326904048076e-05, + "loss": 1.0992, + "step": 261 + }, + { + "epoch": 0.670076726342711, + "grad_norm": 0.38002058368690567, + "learning_rate": 1.9930605826136904e-05, + "loss": 1.1151, + "step": 262 + }, + { + "epoch": 0.6726342710997443, + "grad_norm": 0.41881643054155515, + "learning_rate": 1.992848957049172e-05, + "loss": 1.1594, + "step": 263 + }, + { + "epoch": 0.6751918158567775, + "grad_norm": 0.4432367430176361, + "learning_rate": 1.9926341644622544e-05, + "loss": 1.1374, + "step": 264 + }, + { + "epoch": 0.6777493606138107, + "grad_norm": 0.29540032341942135, + "learning_rate": 1.9924162055380905e-05, + "loss": 1.1654, + "step": 265 + }, + { + "epoch": 0.680306905370844, + "grad_norm": 0.42262678251523295, + "learning_rate": 1.9921950809719326e-05, + "loss": 1.1324, + "step": 266 + }, + { + "epoch": 0.6828644501278772, + "grad_norm": 0.46439800650172125, + "learning_rate": 1.991970791469131e-05, + "loss": 1.0858, + "step": 267 + }, + { + "epoch": 0.6854219948849105, + "grad_norm": 0.4545225924725631, + "learning_rate": 1.991743337745132e-05, + "loss": 1.13, + "step": 268 + }, + { + "epoch": 0.6879795396419437, + "grad_norm": 0.4503197148218845, + "learning_rate": 1.991512720525475e-05, + "loss": 1.1065, + "step": 269 + }, + { + "epoch": 0.690537084398977, + "grad_norm": 0.345808256647908, + "learning_rate": 1.9912789405457907e-05, + "loss": 1.1469, + "step": 270 + }, + { + "epoch": 0.6930946291560103, + "grad_norm": 0.45309494180778404, + "learning_rate": 1.9910419985517976e-05, + "loss": 1.1456, + "step": 271 + }, + { + "epoch": 0.6956521739130435, + "grad_norm": 0.5979768056972147, + "learning_rate": 1.9908018952993017e-05, + "loss": 1.1063, + "step": 272 + }, + { + "epoch": 0.6982097186700768, + "grad_norm": 0.4235080476193215, + "learning_rate": 1.9905586315541917e-05, + "loss": 1.1221, + "step": 273 + }, + { + "epoch": 0.7007672634271099, + "grad_norm": 0.4371176178919167, + "learning_rate": 1.9903122080924385e-05, + "loss": 1.1442, + "step": 274 + }, + { + "epoch": 0.7033248081841432, + "grad_norm": 0.5421093135936572, + "learning_rate": 1.9900626257000924e-05, + "loss": 1.1211, + "step": 275 + }, + { + "epoch": 0.7058823529411765, + "grad_norm": 0.5254257134289397, + "learning_rate": 1.9898098851732787e-05, + "loss": 1.1387, + "step": 276 + }, + { + "epoch": 0.7084398976982097, + "grad_norm": 0.43013120307542063, + "learning_rate": 1.989553987318198e-05, + "loss": 1.1531, + "step": 277 + }, + { + "epoch": 0.710997442455243, + "grad_norm": 0.3719236734452485, + "learning_rate": 1.9892949329511213e-05, + "loss": 1.119, + "step": 278 + }, + { + "epoch": 0.7135549872122762, + "grad_norm": 0.4460019794088031, + "learning_rate": 1.989032722898389e-05, + "loss": 1.1492, + "step": 279 + }, + { + "epoch": 0.7161125319693095, + "grad_norm": 0.46677512285334416, + "learning_rate": 1.988767357996408e-05, + "loss": 1.1162, + "step": 280 + }, + { + "epoch": 0.7186700767263428, + "grad_norm": 0.4083748664929053, + "learning_rate": 1.988498839091647e-05, + "loss": 1.1092, + "step": 281 + }, + { + "epoch": 0.7212276214833759, + "grad_norm": 0.3564940325688949, + "learning_rate": 1.9882271670406373e-05, + "loss": 1.1357, + "step": 282 + }, + { + "epoch": 0.7237851662404092, + "grad_norm": 0.30028785539914044, + "learning_rate": 1.9879523427099668e-05, + "loss": 1.0653, + "step": 283 + }, + { + "epoch": 0.7263427109974424, + "grad_norm": 0.27869590816680806, + "learning_rate": 1.9876743669762795e-05, + "loss": 1.131, + "step": 284 + }, + { + "epoch": 0.7289002557544757, + "grad_norm": 0.2884828038008038, + "learning_rate": 1.9873932407262715e-05, + "loss": 1.1131, + "step": 285 + }, + { + "epoch": 0.731457800511509, + "grad_norm": 0.2884307017453047, + "learning_rate": 1.9871089648566887e-05, + "loss": 1.1392, + "step": 286 + }, + { + "epoch": 0.7340153452685422, + "grad_norm": 0.2783149887312821, + "learning_rate": 1.9868215402743236e-05, + "loss": 1.1369, + "step": 287 + }, + { + "epoch": 0.7365728900255755, + "grad_norm": 0.25240825307315373, + "learning_rate": 1.9865309678960125e-05, + "loss": 1.0986, + "step": 288 + }, + { + "epoch": 0.7391304347826086, + "grad_norm": 0.27833150890437497, + "learning_rate": 1.986237248648633e-05, + "loss": 1.1309, + "step": 289 + }, + { + "epoch": 0.7416879795396419, + "grad_norm": 0.3397738537561161, + "learning_rate": 1.9859403834691006e-05, + "loss": 1.1421, + "step": 290 + }, + { + "epoch": 0.7442455242966752, + "grad_norm": 0.31909191699373507, + "learning_rate": 1.985640373304365e-05, + "loss": 1.1239, + "step": 291 + }, + { + "epoch": 0.7468030690537084, + "grad_norm": 0.3160261198555746, + "learning_rate": 1.985337219111409e-05, + "loss": 1.0883, + "step": 292 + }, + { + "epoch": 0.7493606138107417, + "grad_norm": 0.31947473180563374, + "learning_rate": 1.9850309218572437e-05, + "loss": 1.1165, + "step": 293 + }, + { + "epoch": 0.7519181585677749, + "grad_norm": 0.28206583491018894, + "learning_rate": 1.984721482518907e-05, + "loss": 1.1125, + "step": 294 + }, + { + "epoch": 0.7544757033248082, + "grad_norm": 0.3087007877058229, + "learning_rate": 1.9844089020834573e-05, + "loss": 1.1041, + "step": 295 + }, + { + "epoch": 0.7570332480818415, + "grad_norm": 0.2900775892483009, + "learning_rate": 1.9840931815479747e-05, + "loss": 1.1122, + "step": 296 + }, + { + "epoch": 0.7595907928388747, + "grad_norm": 0.3049429227172959, + "learning_rate": 1.9837743219195554e-05, + "loss": 1.1508, + "step": 297 + }, + { + "epoch": 0.7621483375959079, + "grad_norm": 0.2702621520802534, + "learning_rate": 1.983452324215308e-05, + "loss": 1.1296, + "step": 298 + }, + { + "epoch": 0.7647058823529411, + "grad_norm": 0.27935338606011506, + "learning_rate": 1.983127189462351e-05, + "loss": 1.1396, + "step": 299 + }, + { + "epoch": 0.7672634271099744, + "grad_norm": 0.2806678661873222, + "learning_rate": 1.9827989186978105e-05, + "loss": 1.1025, + "step": 300 + }, + { + "epoch": 0.7698209718670077, + "grad_norm": 0.2424376386995265, + "learning_rate": 1.9824675129688152e-05, + "loss": 1.1349, + "step": 301 + }, + { + "epoch": 0.7723785166240409, + "grad_norm": 0.25294422720974147, + "learning_rate": 1.9821329733324943e-05, + "loss": 1.1333, + "step": 302 + }, + { + "epoch": 0.7749360613810742, + "grad_norm": 0.2996335354330534, + "learning_rate": 1.9817953008559733e-05, + "loss": 1.1024, + "step": 303 + }, + { + "epoch": 0.7774936061381074, + "grad_norm": 0.25368934438638874, + "learning_rate": 1.9814544966163708e-05, + "loss": 1.1046, + "step": 304 + }, + { + "epoch": 0.7800511508951407, + "grad_norm": 0.29925585101340996, + "learning_rate": 1.9811105617007958e-05, + "loss": 1.1209, + "step": 305 + }, + { + "epoch": 0.782608695652174, + "grad_norm": 0.2857142531264863, + "learning_rate": 1.980763497206343e-05, + "loss": 1.1121, + "step": 306 + }, + { + "epoch": 0.7851662404092071, + "grad_norm": 0.28949765605838534, + "learning_rate": 1.98041330424009e-05, + "loss": 1.0798, + "step": 307 + }, + { + "epoch": 0.7877237851662404, + "grad_norm": 0.39212069215270606, + "learning_rate": 1.9800599839190942e-05, + "loss": 1.1121, + "step": 308 + }, + { + "epoch": 0.7902813299232737, + "grad_norm": 0.42549878552407905, + "learning_rate": 1.979703537370388e-05, + "loss": 1.1301, + "step": 309 + }, + { + "epoch": 0.7928388746803069, + "grad_norm": 0.34373356520749754, + "learning_rate": 1.979343965730977e-05, + "loss": 1.0972, + "step": 310 + }, + { + "epoch": 0.7953964194373402, + "grad_norm": 0.28788024425928377, + "learning_rate": 1.9789812701478346e-05, + "loss": 1.1292, + "step": 311 + }, + { + "epoch": 0.7979539641943734, + "grad_norm": 0.3235972078049142, + "learning_rate": 1.9786154517778987e-05, + "loss": 1.1467, + "step": 312 + }, + { + "epoch": 0.8005115089514067, + "grad_norm": 0.4417301333695989, + "learning_rate": 1.9782465117880694e-05, + "loss": 1.109, + "step": 313 + }, + { + "epoch": 0.80306905370844, + "grad_norm": 0.3742518224113521, + "learning_rate": 1.977874451355203e-05, + "loss": 1.1127, + "step": 314 + }, + { + "epoch": 0.8056265984654731, + "grad_norm": 0.2796776192545452, + "learning_rate": 1.9774992716661106e-05, + "loss": 1.1059, + "step": 315 + }, + { + "epoch": 0.8081841432225064, + "grad_norm": 0.29748665489937487, + "learning_rate": 1.9771209739175522e-05, + "loss": 1.102, + "step": 316 + }, + { + "epoch": 0.8107416879795396, + "grad_norm": 0.3560790006624538, + "learning_rate": 1.9767395593162356e-05, + "loss": 1.131, + "step": 317 + }, + { + "epoch": 0.8132992327365729, + "grad_norm": 0.28183796528203187, + "learning_rate": 1.9763550290788086e-05, + "loss": 1.1236, + "step": 318 + }, + { + "epoch": 0.8158567774936062, + "grad_norm": 0.2767624292955607, + "learning_rate": 1.9759673844318592e-05, + "loss": 1.0826, + "step": 319 + }, + { + "epoch": 0.8184143222506394, + "grad_norm": 0.24323449746418016, + "learning_rate": 1.9755766266119085e-05, + "loss": 1.1101, + "step": 320 + }, + { + "epoch": 0.8209718670076727, + "grad_norm": 0.30635322192890063, + "learning_rate": 1.975182756865409e-05, + "loss": 1.0896, + "step": 321 + }, + { + "epoch": 0.8235294117647058, + "grad_norm": 0.31827256506593926, + "learning_rate": 1.9747857764487395e-05, + "loss": 1.119, + "step": 322 + }, + { + "epoch": 0.8260869565217391, + "grad_norm": 0.2714890499396325, + "learning_rate": 1.974385686628201e-05, + "loss": 1.139, + "step": 323 + }, + { + "epoch": 0.8286445012787724, + "grad_norm": 0.2617887796548345, + "learning_rate": 1.9739824886800134e-05, + "loss": 1.1387, + "step": 324 + }, + { + "epoch": 0.8312020460358056, + "grad_norm": 0.2799625411152913, + "learning_rate": 1.9735761838903106e-05, + "loss": 1.0992, + "step": 325 + }, + { + "epoch": 0.8337595907928389, + "grad_norm": 0.3045162427185304, + "learning_rate": 1.9731667735551376e-05, + "loss": 1.1099, + "step": 326 + }, + { + "epoch": 0.8363171355498721, + "grad_norm": 0.30841493149512844, + "learning_rate": 1.9727542589804446e-05, + "loss": 1.1504, + "step": 327 + }, + { + "epoch": 0.8388746803069054, + "grad_norm": 0.26580510867255863, + "learning_rate": 1.9723386414820844e-05, + "loss": 1.1214, + "step": 328 + }, + { + "epoch": 0.8414322250639387, + "grad_norm": 0.3102926749875895, + "learning_rate": 1.971919922385807e-05, + "loss": 1.1606, + "step": 329 + }, + { + "epoch": 0.8439897698209718, + "grad_norm": 0.23808182814042947, + "learning_rate": 1.9714981030272566e-05, + "loss": 1.1306, + "step": 330 + }, + { + "epoch": 0.8465473145780051, + "grad_norm": 0.3332758895370011, + "learning_rate": 1.9710731847519665e-05, + "loss": 1.1045, + "step": 331 + }, + { + "epoch": 0.8491048593350383, + "grad_norm": 0.3331468308085816, + "learning_rate": 1.9706451689153556e-05, + "loss": 1.0964, + "step": 332 + }, + { + "epoch": 0.8516624040920716, + "grad_norm": 0.29500865696422174, + "learning_rate": 1.9702140568827224e-05, + "loss": 1.1434, + "step": 333 + }, + { + "epoch": 0.8542199488491049, + "grad_norm": 0.28839237064630785, + "learning_rate": 1.9697798500292422e-05, + "loss": 1.0752, + "step": 334 + }, + { + "epoch": 0.8567774936061381, + "grad_norm": 0.2796744909651119, + "learning_rate": 1.969342549739963e-05, + "loss": 1.102, + "step": 335 + }, + { + "epoch": 0.8593350383631714, + "grad_norm": 0.3120222134152984, + "learning_rate": 1.9689021574097988e-05, + "loss": 1.1016, + "step": 336 + }, + { + "epoch": 0.8618925831202046, + "grad_norm": 0.3114585821970581, + "learning_rate": 1.9684586744435285e-05, + "loss": 1.125, + "step": 337 + }, + { + "epoch": 0.8644501278772379, + "grad_norm": 0.38630267959141373, + "learning_rate": 1.9680121022557882e-05, + "loss": 1.1295, + "step": 338 + }, + { + "epoch": 0.8670076726342711, + "grad_norm": 0.4245255400950163, + "learning_rate": 1.9675624422710685e-05, + "loss": 1.0954, + "step": 339 + }, + { + "epoch": 0.8695652173913043, + "grad_norm": 0.35097053556774704, + "learning_rate": 1.9671096959237102e-05, + "loss": 1.1348, + "step": 340 + }, + { + "epoch": 0.8721227621483376, + "grad_norm": 0.30986958941628523, + "learning_rate": 1.9666538646578978e-05, + "loss": 1.091, + "step": 341 + }, + { + "epoch": 0.8746803069053708, + "grad_norm": 0.3077968967678069, + "learning_rate": 1.9661949499276577e-05, + "loss": 1.1175, + "step": 342 + }, + { + "epoch": 0.8772378516624041, + "grad_norm": 0.2985995156361193, + "learning_rate": 1.965732953196851e-05, + "loss": 1.1149, + "step": 343 + }, + { + "epoch": 0.8797953964194374, + "grad_norm": 0.3167214536166744, + "learning_rate": 1.96526787593917e-05, + "loss": 1.1224, + "step": 344 + }, + { + "epoch": 0.8823529411764706, + "grad_norm": 0.318814842450484, + "learning_rate": 1.964799719638134e-05, + "loss": 1.1039, + "step": 345 + }, + { + "epoch": 0.8849104859335039, + "grad_norm": 0.35420843208500646, + "learning_rate": 1.9643284857870824e-05, + "loss": 1.1686, + "step": 346 + }, + { + "epoch": 0.887468030690537, + "grad_norm": 0.2815662324657919, + "learning_rate": 1.9638541758891733e-05, + "loss": 1.0509, + "step": 347 + }, + { + "epoch": 0.8900255754475703, + "grad_norm": 0.3284996290945278, + "learning_rate": 1.963376791457376e-05, + "loss": 1.1162, + "step": 348 + }, + { + "epoch": 0.8925831202046036, + "grad_norm": 0.33970062669533485, + "learning_rate": 1.962896334014467e-05, + "loss": 1.1, + "step": 349 + }, + { + "epoch": 0.8951406649616368, + "grad_norm": 0.32747244832244676, + "learning_rate": 1.9624128050930252e-05, + "loss": 1.1228, + "step": 350 + }, + { + "epoch": 0.8976982097186701, + "grad_norm": 0.27048623218728324, + "learning_rate": 1.9619262062354273e-05, + "loss": 1.1403, + "step": 351 + }, + { + "epoch": 0.9002557544757033, + "grad_norm": 0.2972853292863894, + "learning_rate": 1.9614365389938425e-05, + "loss": 1.0943, + "step": 352 + }, + { + "epoch": 0.9028132992327366, + "grad_norm": 0.34260939469670293, + "learning_rate": 1.9609438049302274e-05, + "loss": 1.1167, + "step": 353 + }, + { + "epoch": 0.9053708439897699, + "grad_norm": 0.38264563083677083, + "learning_rate": 1.9604480056163215e-05, + "loss": 1.1109, + "step": 354 + }, + { + "epoch": 0.907928388746803, + "grad_norm": 0.28234436041490374, + "learning_rate": 1.9599491426336415e-05, + "loss": 1.0884, + "step": 355 + }, + { + "epoch": 0.9104859335038363, + "grad_norm": 0.2708061459988167, + "learning_rate": 1.9594472175734776e-05, + "loss": 1.1013, + "step": 356 + }, + { + "epoch": 0.9130434782608695, + "grad_norm": 0.25976263313561604, + "learning_rate": 1.9589422320368863e-05, + "loss": 1.112, + "step": 357 + }, + { + "epoch": 0.9156010230179028, + "grad_norm": 0.33058002121583785, + "learning_rate": 1.9584341876346874e-05, + "loss": 1.1338, + "step": 358 + }, + { + "epoch": 0.9181585677749361, + "grad_norm": 0.32807990736066434, + "learning_rate": 1.957923085987458e-05, + "loss": 1.115, + "step": 359 + }, + { + "epoch": 0.9207161125319693, + "grad_norm": 0.3514992497926305, + "learning_rate": 1.957408928725527e-05, + "loss": 1.0819, + "step": 360 + }, + { + "epoch": 0.9232736572890026, + "grad_norm": 0.377909885361285, + "learning_rate": 1.9568917174889696e-05, + "loss": 1.1301, + "step": 361 + }, + { + "epoch": 0.9258312020460358, + "grad_norm": 0.2561753671305042, + "learning_rate": 1.9563714539276037e-05, + "loss": 1.0582, + "step": 362 + }, + { + "epoch": 0.928388746803069, + "grad_norm": 0.33150365566832235, + "learning_rate": 1.9558481397009832e-05, + "loss": 1.1201, + "step": 363 + }, + { + "epoch": 0.9309462915601023, + "grad_norm": 0.41073130363857413, + "learning_rate": 1.9553217764783928e-05, + "loss": 1.1031, + "step": 364 + }, + { + "epoch": 0.9335038363171355, + "grad_norm": 0.44385440148457334, + "learning_rate": 1.954792365938844e-05, + "loss": 1.1201, + "step": 365 + }, + { + "epoch": 0.9360613810741688, + "grad_norm": 0.3694394788747904, + "learning_rate": 1.9542599097710677e-05, + "loss": 1.1431, + "step": 366 + }, + { + "epoch": 0.9386189258312021, + "grad_norm": 0.3652537164671811, + "learning_rate": 1.9537244096735096e-05, + "loss": 1.1213, + "step": 367 + }, + { + "epoch": 0.9411764705882353, + "grad_norm": 0.3951831465406744, + "learning_rate": 1.9531858673543266e-05, + "loss": 1.1147, + "step": 368 + }, + { + "epoch": 0.9437340153452686, + "grad_norm": 0.5451604935646901, + "learning_rate": 1.952644284531378e-05, + "loss": 1.1096, + "step": 369 + }, + { + "epoch": 0.9462915601023018, + "grad_norm": 0.4328674219635996, + "learning_rate": 1.9520996629322228e-05, + "loss": 1.1056, + "step": 370 + }, + { + "epoch": 0.948849104859335, + "grad_norm": 0.34768509118734625, + "learning_rate": 1.9515520042941133e-05, + "loss": 1.1329, + "step": 371 + }, + { + "epoch": 0.9514066496163683, + "grad_norm": 0.33694594098800845, + "learning_rate": 1.9510013103639885e-05, + "loss": 1.1139, + "step": 372 + }, + { + "epoch": 0.9539641943734015, + "grad_norm": 0.2816891577865245, + "learning_rate": 1.9504475828984706e-05, + "loss": 1.1172, + "step": 373 + }, + { + "epoch": 0.9565217391304348, + "grad_norm": 0.2726181598335772, + "learning_rate": 1.9498908236638572e-05, + "loss": 1.117, + "step": 374 + }, + { + "epoch": 0.959079283887468, + "grad_norm": 0.2725400027413373, + "learning_rate": 1.949331034436118e-05, + "loss": 1.1392, + "step": 375 + }, + { + "epoch": 0.9616368286445013, + "grad_norm": 0.24708596308833097, + "learning_rate": 1.9487682170008864e-05, + "loss": 1.0888, + "step": 376 + }, + { + "epoch": 0.9641943734015346, + "grad_norm": 0.24611853998964667, + "learning_rate": 1.948202373153457e-05, + "loss": 1.1222, + "step": 377 + }, + { + "epoch": 0.9667519181585678, + "grad_norm": 0.2748730983230991, + "learning_rate": 1.9476335046987765e-05, + "loss": 1.1274, + "step": 378 + }, + { + "epoch": 0.969309462915601, + "grad_norm": 0.23212979863877406, + "learning_rate": 1.9470616134514405e-05, + "loss": 1.0955, + "step": 379 + }, + { + "epoch": 0.9718670076726342, + "grad_norm": 0.24862661491603566, + "learning_rate": 1.9464867012356866e-05, + "loss": 1.0795, + "step": 380 + }, + { + "epoch": 0.9744245524296675, + "grad_norm": 0.28519808751496534, + "learning_rate": 1.9459087698853884e-05, + "loss": 1.1092, + "step": 381 + }, + { + "epoch": 0.9769820971867008, + "grad_norm": 0.2794622526427573, + "learning_rate": 1.945327821244051e-05, + "loss": 1.113, + "step": 382 + }, + { + "epoch": 0.979539641943734, + "grad_norm": 0.27278224425786757, + "learning_rate": 1.944743857164803e-05, + "loss": 1.1169, + "step": 383 + }, + { + "epoch": 0.9820971867007673, + "grad_norm": 0.27611615092251324, + "learning_rate": 1.9441568795103932e-05, + "loss": 1.0704, + "step": 384 + }, + { + "epoch": 0.9846547314578005, + "grad_norm": 0.26697361837596756, + "learning_rate": 1.9435668901531813e-05, + "loss": 1.1517, + "step": 385 + }, + { + "epoch": 0.9872122762148338, + "grad_norm": 0.3629600651937476, + "learning_rate": 1.9429738909751355e-05, + "loss": 1.1052, + "step": 386 + }, + { + "epoch": 0.989769820971867, + "grad_norm": 0.30179358680215773, + "learning_rate": 1.9423778838678236e-05, + "loss": 1.1192, + "step": 387 + }, + { + "epoch": 0.9923273657289002, + "grad_norm": 0.3006384503184647, + "learning_rate": 1.9417788707324095e-05, + "loss": 1.133, + "step": 388 + }, + { + "epoch": 0.9948849104859335, + "grad_norm": 0.2894724527823374, + "learning_rate": 1.9411768534796445e-05, + "loss": 1.1161, + "step": 389 + }, + { + "epoch": 0.9974424552429667, + "grad_norm": 0.2807180282095206, + "learning_rate": 1.9405718340298633e-05, + "loss": 1.0955, + "step": 390 + }, + { + "epoch": 1.0, + "grad_norm": 0.348947561667452, + "learning_rate": 1.9399638143129768e-05, + "loss": 1.1251, + "step": 391 + }, + { + "epoch": 1.0025575447570332, + "grad_norm": 0.2782036837933627, + "learning_rate": 1.9393527962684665e-05, + "loss": 1.1002, + "step": 392 + }, + { + "epoch": 1.0051150895140666, + "grad_norm": 0.3519574703494421, + "learning_rate": 1.938738781845378e-05, + "loss": 1.1167, + "step": 393 + }, + { + "epoch": 1.0076726342710998, + "grad_norm": 0.4580624378113774, + "learning_rate": 1.938121773002315e-05, + "loss": 1.0882, + "step": 394 + }, + { + "epoch": 1.010230179028133, + "grad_norm": 0.40810589155384597, + "learning_rate": 1.9375017717074322e-05, + "loss": 1.0969, + "step": 395 + }, + { + "epoch": 1.0127877237851663, + "grad_norm": 0.36602089723011694, + "learning_rate": 1.9368787799384303e-05, + "loss": 1.1139, + "step": 396 + }, + { + "epoch": 1.0153452685421995, + "grad_norm": 0.35535572252914266, + "learning_rate": 1.936252799682549e-05, + "loss": 1.1153, + "step": 397 + }, + { + "epoch": 1.0179028132992327, + "grad_norm": 0.30429749075329415, + "learning_rate": 1.9356238329365614e-05, + "loss": 1.1053, + "step": 398 + }, + { + "epoch": 1.020460358056266, + "grad_norm": 0.3229547174327428, + "learning_rate": 1.9349918817067658e-05, + "loss": 1.0765, + "step": 399 + }, + { + "epoch": 1.0230179028132993, + "grad_norm": 0.41127178125569, + "learning_rate": 1.934356948008981e-05, + "loss": 1.1314, + "step": 400 + }, + { + "epoch": 1.0255754475703325, + "grad_norm": 0.45751355203161004, + "learning_rate": 1.93371903386854e-05, + "loss": 1.11, + "step": 401 + }, + { + "epoch": 1.0281329923273657, + "grad_norm": 0.3986059039543531, + "learning_rate": 1.933078141320282e-05, + "loss": 1.1286, + "step": 402 + }, + { + "epoch": 1.030690537084399, + "grad_norm": 0.3213433170159458, + "learning_rate": 1.9324342724085473e-05, + "loss": 1.0938, + "step": 403 + }, + { + "epoch": 1.0332480818414322, + "grad_norm": 0.2940898456275506, + "learning_rate": 1.9317874291871703e-05, + "loss": 1.1004, + "step": 404 + }, + { + "epoch": 1.0358056265984654, + "grad_norm": 0.32488534442637107, + "learning_rate": 1.931137613719473e-05, + "loss": 1.0987, + "step": 405 + }, + { + "epoch": 1.0383631713554988, + "grad_norm": 0.4137853980303429, + "learning_rate": 1.9304848280782582e-05, + "loss": 1.1313, + "step": 406 + }, + { + "epoch": 1.040920716112532, + "grad_norm": 0.4264194369796108, + "learning_rate": 1.9298290743458027e-05, + "loss": 1.0649, + "step": 407 + }, + { + "epoch": 1.0434782608695652, + "grad_norm": 0.45590348437434036, + "learning_rate": 1.929170354613852e-05, + "loss": 1.0967, + "step": 408 + }, + { + "epoch": 1.0460358056265984, + "grad_norm": 0.33156329123437867, + "learning_rate": 1.9285086709836115e-05, + "loss": 1.0775, + "step": 409 + }, + { + "epoch": 1.0485933503836318, + "grad_norm": 0.3704005026550043, + "learning_rate": 1.927844025565742e-05, + "loss": 1.0716, + "step": 410 + }, + { + "epoch": 1.051150895140665, + "grad_norm": 0.3169006666927623, + "learning_rate": 1.9271764204803512e-05, + "loss": 1.0984, + "step": 411 + }, + { + "epoch": 1.0537084398976981, + "grad_norm": 0.3295513943189194, + "learning_rate": 1.9265058578569878e-05, + "loss": 1.079, + "step": 412 + }, + { + "epoch": 1.0562659846547315, + "grad_norm": 0.3297315773374301, + "learning_rate": 1.9258323398346345e-05, + "loss": 1.0881, + "step": 413 + }, + { + "epoch": 1.0588235294117647, + "grad_norm": 0.3173169155920823, + "learning_rate": 1.9251558685617015e-05, + "loss": 1.1338, + "step": 414 + }, + { + "epoch": 1.061381074168798, + "grad_norm": 0.3993520424000351, + "learning_rate": 1.9244764461960193e-05, + "loss": 1.1055, + "step": 415 + }, + { + "epoch": 1.0639386189258313, + "grad_norm": 0.27163667354634313, + "learning_rate": 1.923794074904832e-05, + "loss": 1.0866, + "step": 416 + }, + { + "epoch": 1.0664961636828645, + "grad_norm": 0.3514844689560382, + "learning_rate": 1.9231087568647892e-05, + "loss": 1.0705, + "step": 417 + }, + { + "epoch": 1.0690537084398977, + "grad_norm": 0.2854882940594379, + "learning_rate": 1.9224204942619418e-05, + "loss": 1.0802, + "step": 418 + }, + { + "epoch": 1.0716112531969308, + "grad_norm": 0.33059120925775853, + "learning_rate": 1.9217292892917326e-05, + "loss": 1.1369, + "step": 419 + }, + { + "epoch": 1.0741687979539642, + "grad_norm": 0.28470692158693056, + "learning_rate": 1.9210351441589898e-05, + "loss": 1.0851, + "step": 420 + }, + { + "epoch": 1.0767263427109974, + "grad_norm": 0.27607260051782767, + "learning_rate": 1.92033806107792e-05, + "loss": 1.0983, + "step": 421 + }, + { + "epoch": 1.0792838874680306, + "grad_norm": 0.27376802007793644, + "learning_rate": 1.919638042272103e-05, + "loss": 1.0899, + "step": 422 + }, + { + "epoch": 1.081841432225064, + "grad_norm": 0.24273577112968892, + "learning_rate": 1.9189350899744807e-05, + "loss": 1.0748, + "step": 423 + }, + { + "epoch": 1.0843989769820972, + "grad_norm": 0.32941305876089433, + "learning_rate": 1.9182292064273544e-05, + "loss": 1.1209, + "step": 424 + }, + { + "epoch": 1.0869565217391304, + "grad_norm": 0.226985402664926, + "learning_rate": 1.9175203938823747e-05, + "loss": 1.0853, + "step": 425 + }, + { + "epoch": 1.0895140664961638, + "grad_norm": 0.30079073449406646, + "learning_rate": 1.9168086546005348e-05, + "loss": 1.0746, + "step": 426 + }, + { + "epoch": 1.092071611253197, + "grad_norm": 0.33309434785775865, + "learning_rate": 1.9160939908521642e-05, + "loss": 1.0991, + "step": 427 + }, + { + "epoch": 1.0946291560102301, + "grad_norm": 0.4305882959246417, + "learning_rate": 1.9153764049169213e-05, + "loss": 1.0799, + "step": 428 + }, + { + "epoch": 1.0971867007672633, + "grad_norm": 0.32261776654673774, + "learning_rate": 1.9146558990837853e-05, + "loss": 1.1098, + "step": 429 + }, + { + "epoch": 1.0997442455242967, + "grad_norm": 0.3043949510529127, + "learning_rate": 1.9139324756510496e-05, + "loss": 1.0854, + "step": 430 + }, + { + "epoch": 1.10230179028133, + "grad_norm": 0.25743020242945314, + "learning_rate": 1.913206136926314e-05, + "loss": 1.1157, + "step": 431 + }, + { + "epoch": 1.104859335038363, + "grad_norm": 0.3310331033148412, + "learning_rate": 1.9124768852264774e-05, + "loss": 1.134, + "step": 432 + }, + { + "epoch": 1.1074168797953965, + "grad_norm": 0.4154810782539844, + "learning_rate": 1.9117447228777317e-05, + "loss": 1.1111, + "step": 433 + }, + { + "epoch": 1.1099744245524297, + "grad_norm": 0.33867417586438353, + "learning_rate": 1.9110096522155525e-05, + "loss": 1.0996, + "step": 434 + }, + { + "epoch": 1.1125319693094629, + "grad_norm": 0.2859884267480265, + "learning_rate": 1.9102716755846916e-05, + "loss": 1.0916, + "step": 435 + }, + { + "epoch": 1.1150895140664963, + "grad_norm": 0.3892200070495796, + "learning_rate": 1.909530795339172e-05, + "loss": 1.0894, + "step": 436 + }, + { + "epoch": 1.1176470588235294, + "grad_norm": 0.48792801105881567, + "learning_rate": 1.9087870138422777e-05, + "loss": 1.0702, + "step": 437 + }, + { + "epoch": 1.1202046035805626, + "grad_norm": 0.4392749429122809, + "learning_rate": 1.9080403334665475e-05, + "loss": 1.0831, + "step": 438 + }, + { + "epoch": 1.1227621483375958, + "grad_norm": 0.31693271658453487, + "learning_rate": 1.9072907565937675e-05, + "loss": 1.1038, + "step": 439 + }, + { + "epoch": 1.1253196930946292, + "grad_norm": 0.3404612161405027, + "learning_rate": 1.9065382856149623e-05, + "loss": 1.0813, + "step": 440 + }, + { + "epoch": 1.1278772378516624, + "grad_norm": 0.36566368828973633, + "learning_rate": 1.905782922930389e-05, + "loss": 1.1008, + "step": 441 + }, + { + "epoch": 1.1304347826086956, + "grad_norm": 0.3310584074847857, + "learning_rate": 1.9050246709495283e-05, + "loss": 1.0811, + "step": 442 + }, + { + "epoch": 1.132992327365729, + "grad_norm": 0.2708409903975629, + "learning_rate": 1.904263532091077e-05, + "loss": 1.1057, + "step": 443 + }, + { + "epoch": 1.1355498721227621, + "grad_norm": 0.3139293999297133, + "learning_rate": 1.903499508782942e-05, + "loss": 1.0716, + "step": 444 + }, + { + "epoch": 1.1381074168797953, + "grad_norm": 0.39095800156714283, + "learning_rate": 1.902732603462229e-05, + "loss": 1.0921, + "step": 445 + }, + { + "epoch": 1.1406649616368287, + "grad_norm": 0.30908518133076435, + "learning_rate": 1.9019628185752384e-05, + "loss": 1.0648, + "step": 446 + }, + { + "epoch": 1.143222506393862, + "grad_norm": 0.3805637372734307, + "learning_rate": 1.9011901565774555e-05, + "loss": 1.0533, + "step": 447 + }, + { + "epoch": 1.145780051150895, + "grad_norm": 0.3775382301483175, + "learning_rate": 1.900414619933543e-05, + "loss": 1.0858, + "step": 448 + }, + { + "epoch": 1.1483375959079285, + "grad_norm": 0.41861102880335493, + "learning_rate": 1.8996362111173338e-05, + "loss": 1.0772, + "step": 449 + }, + { + "epoch": 1.1508951406649617, + "grad_norm": 0.3783086482998166, + "learning_rate": 1.898854932611821e-05, + "loss": 1.1276, + "step": 450 + }, + { + "epoch": 1.1534526854219949, + "grad_norm": 0.3167462811218851, + "learning_rate": 1.8980707869091534e-05, + "loss": 1.1, + "step": 451 + }, + { + "epoch": 1.156010230179028, + "grad_norm": 0.2569091030662843, + "learning_rate": 1.8972837765106244e-05, + "loss": 1.1032, + "step": 452 + }, + { + "epoch": 1.1585677749360614, + "grad_norm": 0.42403134602315007, + "learning_rate": 1.896493903926666e-05, + "loss": 1.1185, + "step": 453 + }, + { + "epoch": 1.1611253196930946, + "grad_norm": 0.4524020606510499, + "learning_rate": 1.8957011716768403e-05, + "loss": 1.0895, + "step": 454 + }, + { + "epoch": 1.1636828644501278, + "grad_norm": 0.30454573206724117, + "learning_rate": 1.89490558228983e-05, + "loss": 1.0851, + "step": 455 + }, + { + "epoch": 1.1662404092071612, + "grad_norm": 0.25991176813435285, + "learning_rate": 1.8941071383034328e-05, + "loss": 1.0773, + "step": 456 + }, + { + "epoch": 1.1687979539641944, + "grad_norm": 0.3886268706724371, + "learning_rate": 1.8933058422645516e-05, + "loss": 1.0699, + "step": 457 + }, + { + "epoch": 1.1713554987212276, + "grad_norm": 0.36471960792817715, + "learning_rate": 1.8925016967291873e-05, + "loss": 1.0742, + "step": 458 + }, + { + "epoch": 1.1739130434782608, + "grad_norm": 0.26625088784429024, + "learning_rate": 1.8916947042624294e-05, + "loss": 1.0827, + "step": 459 + }, + { + "epoch": 1.1764705882352942, + "grad_norm": 0.32264462408411615, + "learning_rate": 1.8908848674384494e-05, + "loss": 1.1101, + "step": 460 + }, + { + "epoch": 1.1790281329923273, + "grad_norm": 0.36503412806400276, + "learning_rate": 1.8900721888404917e-05, + "loss": 1.1335, + "step": 461 + }, + { + "epoch": 1.1815856777493605, + "grad_norm": 0.2719728070020309, + "learning_rate": 1.8892566710608648e-05, + "loss": 1.0758, + "step": 462 + }, + { + "epoch": 1.184143222506394, + "grad_norm": 0.2553303942221017, + "learning_rate": 1.888438316700935e-05, + "loss": 1.0658, + "step": 463 + }, + { + "epoch": 1.186700767263427, + "grad_norm": 0.295046080413778, + "learning_rate": 1.8876171283711153e-05, + "loss": 1.0937, + "step": 464 + }, + { + "epoch": 1.1892583120204603, + "grad_norm": 0.28953986695793527, + "learning_rate": 1.88679310869086e-05, + "loss": 1.1014, + "step": 465 + }, + { + "epoch": 1.1918158567774937, + "grad_norm": 0.2826951705156822, + "learning_rate": 1.885966260288654e-05, + "loss": 1.0993, + "step": 466 + }, + { + "epoch": 1.1943734015345269, + "grad_norm": 0.2944678860725613, + "learning_rate": 1.8851365858020054e-05, + "loss": 1.0761, + "step": 467 + }, + { + "epoch": 1.19693094629156, + "grad_norm": 0.27240534403446576, + "learning_rate": 1.884304087877438e-05, + "loss": 1.1276, + "step": 468 + }, + { + "epoch": 1.1994884910485935, + "grad_norm": 0.27201496505406125, + "learning_rate": 1.8834687691704807e-05, + "loss": 1.0915, + "step": 469 + }, + { + "epoch": 1.2020460358056266, + "grad_norm": 0.2627060324507912, + "learning_rate": 1.8826306323456603e-05, + "loss": 1.1133, + "step": 470 + }, + { + "epoch": 1.2046035805626598, + "grad_norm": 0.2639617835117496, + "learning_rate": 1.881789680076494e-05, + "loss": 1.085, + "step": 471 + }, + { + "epoch": 1.207161125319693, + "grad_norm": 0.2664271870627373, + "learning_rate": 1.8809459150454788e-05, + "loss": 1.0432, + "step": 472 + }, + { + "epoch": 1.2097186700767264, + "grad_norm": 0.27593650483499654, + "learning_rate": 1.8800993399440845e-05, + "loss": 1.0998, + "step": 473 + }, + { + "epoch": 1.2122762148337596, + "grad_norm": 0.24702077068855108, + "learning_rate": 1.879249957472744e-05, + "loss": 1.0828, + "step": 474 + }, + { + "epoch": 1.2148337595907928, + "grad_norm": 0.27099736060394275, + "learning_rate": 1.878397770340846e-05, + "loss": 1.0979, + "step": 475 + }, + { + "epoch": 1.2173913043478262, + "grad_norm": 0.33819417373799127, + "learning_rate": 1.877542781266725e-05, + "loss": 1.0667, + "step": 476 + }, + { + "epoch": 1.2199488491048593, + "grad_norm": 0.375994886329379, + "learning_rate": 1.876684992977653e-05, + "loss": 1.1219, + "step": 477 + }, + { + "epoch": 1.2225063938618925, + "grad_norm": 0.3006928865582519, + "learning_rate": 1.875824408209832e-05, + "loss": 1.1044, + "step": 478 + }, + { + "epoch": 1.2250639386189257, + "grad_norm": 0.2967546177260049, + "learning_rate": 1.874961029708383e-05, + "loss": 1.113, + "step": 479 + }, + { + "epoch": 1.227621483375959, + "grad_norm": 0.2721644513681839, + "learning_rate": 1.8740948602273402e-05, + "loss": 1.064, + "step": 480 + }, + { + "epoch": 1.2301790281329923, + "grad_norm": 0.2626448539398862, + "learning_rate": 1.873225902529639e-05, + "loss": 1.0728, + "step": 481 + }, + { + "epoch": 1.2327365728900257, + "grad_norm": 0.3026172920619992, + "learning_rate": 1.872354159387109e-05, + "loss": 1.0938, + "step": 482 + }, + { + "epoch": 1.2352941176470589, + "grad_norm": 0.34601754375875127, + "learning_rate": 1.8714796335804665e-05, + "loss": 1.0849, + "step": 483 + }, + { + "epoch": 1.237851662404092, + "grad_norm": 0.3327308153807525, + "learning_rate": 1.8706023278993015e-05, + "loss": 1.1004, + "step": 484 + }, + { + "epoch": 1.2404092071611252, + "grad_norm": 0.3106025965315927, + "learning_rate": 1.8697222451420735e-05, + "loss": 1.1043, + "step": 485 + }, + { + "epoch": 1.2429667519181586, + "grad_norm": 0.34912290055883255, + "learning_rate": 1.8688393881160992e-05, + "loss": 1.0987, + "step": 486 + }, + { + "epoch": 1.2455242966751918, + "grad_norm": 0.25322300465901393, + "learning_rate": 1.867953759637545e-05, + "loss": 1.0696, + "step": 487 + }, + { + "epoch": 1.248081841432225, + "grad_norm": 0.2535966365326748, + "learning_rate": 1.8670653625314184e-05, + "loss": 1.0895, + "step": 488 + }, + { + "epoch": 1.2506393861892584, + "grad_norm": 0.30478536911118187, + "learning_rate": 1.8661741996315572e-05, + "loss": 1.0975, + "step": 489 + }, + { + "epoch": 1.2531969309462916, + "grad_norm": 0.3154046265606216, + "learning_rate": 1.8652802737806227e-05, + "loss": 1.119, + "step": 490 + }, + { + "epoch": 1.2557544757033248, + "grad_norm": 0.33453327155185647, + "learning_rate": 1.864383587830089e-05, + "loss": 1.0881, + "step": 491 + }, + { + "epoch": 1.258312020460358, + "grad_norm": 0.2727533150638851, + "learning_rate": 1.8634841446402344e-05, + "loss": 1.1145, + "step": 492 + }, + { + "epoch": 1.2608695652173914, + "grad_norm": 0.36121948622829403, + "learning_rate": 1.862581947080132e-05, + "loss": 1.0836, + "step": 493 + }, + { + "epoch": 1.2634271099744245, + "grad_norm": 0.43540568550818237, + "learning_rate": 1.8616769980276427e-05, + "loss": 1.0817, + "step": 494 + }, + { + "epoch": 1.265984654731458, + "grad_norm": 0.43519257410113654, + "learning_rate": 1.860769300369402e-05, + "loss": 1.1064, + "step": 495 + }, + { + "epoch": 1.2685421994884911, + "grad_norm": 0.28730988243979366, + "learning_rate": 1.859858857000814e-05, + "loss": 1.0948, + "step": 496 + }, + { + "epoch": 1.2710997442455243, + "grad_norm": 0.3115117836172728, + "learning_rate": 1.858945670826041e-05, + "loss": 1.0867, + "step": 497 + }, + { + "epoch": 1.2736572890025575, + "grad_norm": 0.37248383339237146, + "learning_rate": 1.858029744757995e-05, + "loss": 1.0731, + "step": 498 + }, + { + "epoch": 1.2762148337595907, + "grad_norm": 0.3888532232815489, + "learning_rate": 1.857111081718326e-05, + "loss": 1.0876, + "step": 499 + }, + { + "epoch": 1.278772378516624, + "grad_norm": 0.3244399041871337, + "learning_rate": 1.8561896846374167e-05, + "loss": 1.0621, + "step": 500 + }, + { + "epoch": 1.2813299232736572, + "grad_norm": 0.2633849807180457, + "learning_rate": 1.8552655564543695e-05, + "loss": 1.0828, + "step": 501 + }, + { + "epoch": 1.2838874680306906, + "grad_norm": 0.30310812918802343, + "learning_rate": 1.8543387001169994e-05, + "loss": 1.0869, + "step": 502 + }, + { + "epoch": 1.2864450127877238, + "grad_norm": 0.38986142912075833, + "learning_rate": 1.853409118581823e-05, + "loss": 1.1069, + "step": 503 + }, + { + "epoch": 1.289002557544757, + "grad_norm": 0.2920169315779173, + "learning_rate": 1.8524768148140504e-05, + "loss": 1.0826, + "step": 504 + }, + { + "epoch": 1.2915601023017902, + "grad_norm": 0.2574557432611984, + "learning_rate": 1.851541791787575e-05, + "loss": 1.0964, + "step": 505 + }, + { + "epoch": 1.2941176470588236, + "grad_norm": 0.2828254579434041, + "learning_rate": 1.850604052484964e-05, + "loss": 1.0721, + "step": 506 + }, + { + "epoch": 1.2966751918158568, + "grad_norm": 0.3701320049597822, + "learning_rate": 1.8496635998974487e-05, + "loss": 1.1063, + "step": 507 + }, + { + "epoch": 1.29923273657289, + "grad_norm": 0.3423233955071227, + "learning_rate": 1.848720437024917e-05, + "loss": 1.0664, + "step": 508 + }, + { + "epoch": 1.3017902813299234, + "grad_norm": 0.2178713925167967, + "learning_rate": 1.8477745668758998e-05, + "loss": 1.1, + "step": 509 + }, + { + "epoch": 1.3043478260869565, + "grad_norm": 0.2826697186052642, + "learning_rate": 1.8468259924675655e-05, + "loss": 1.068, + "step": 510 + }, + { + "epoch": 1.3069053708439897, + "grad_norm": 0.30593309813937236, + "learning_rate": 1.8458747168257086e-05, + "loss": 1.0768, + "step": 511 + }, + { + "epoch": 1.309462915601023, + "grad_norm": 0.32702026057465294, + "learning_rate": 1.8449207429847383e-05, + "loss": 1.1078, + "step": 512 + }, + { + "epoch": 1.3120204603580563, + "grad_norm": 0.2947477975017397, + "learning_rate": 1.8439640739876732e-05, + "loss": 1.0634, + "step": 513 + }, + { + "epoch": 1.3145780051150895, + "grad_norm": 0.3621645781464773, + "learning_rate": 1.8430047128861267e-05, + "loss": 1.0846, + "step": 514 + }, + { + "epoch": 1.317135549872123, + "grad_norm": 0.3135575999432535, + "learning_rate": 1.8420426627403008e-05, + "loss": 1.076, + "step": 515 + }, + { + "epoch": 1.319693094629156, + "grad_norm": 0.3227803754151594, + "learning_rate": 1.8410779266189752e-05, + "loss": 1.0715, + "step": 516 + }, + { + "epoch": 1.3222506393861893, + "grad_norm": 0.2571799145689103, + "learning_rate": 1.840110507599497e-05, + "loss": 1.0676, + "step": 517 + }, + { + "epoch": 1.3248081841432224, + "grad_norm": 0.30888680030870047, + "learning_rate": 1.8391404087677704e-05, + "loss": 1.0847, + "step": 518 + }, + { + "epoch": 1.3273657289002558, + "grad_norm": 0.29428683307499354, + "learning_rate": 1.8381676332182498e-05, + "loss": 1.0749, + "step": 519 + }, + { + "epoch": 1.329923273657289, + "grad_norm": 0.28408204459805186, + "learning_rate": 1.8371921840539264e-05, + "loss": 1.1205, + "step": 520 + }, + { + "epoch": 1.3324808184143222, + "grad_norm": 0.356882249127386, + "learning_rate": 1.83621406438632e-05, + "loss": 1.1202, + "step": 521 + }, + { + "epoch": 1.3350383631713556, + "grad_norm": 0.37405650968529813, + "learning_rate": 1.8352332773354696e-05, + "loss": 1.0973, + "step": 522 + }, + { + "epoch": 1.3375959079283888, + "grad_norm": 0.39837269879026094, + "learning_rate": 1.8342498260299212e-05, + "loss": 1.0864, + "step": 523 + }, + { + "epoch": 1.340153452685422, + "grad_norm": 0.38120725373511777, + "learning_rate": 1.833263713606721e-05, + "loss": 1.1128, + "step": 524 + }, + { + "epoch": 1.3427109974424551, + "grad_norm": 0.2376913907862642, + "learning_rate": 1.8322749432114028e-05, + "loss": 1.083, + "step": 525 + }, + { + "epoch": 1.3452685421994885, + "grad_norm": 0.33071151652108544, + "learning_rate": 1.8312835179979788e-05, + "loss": 1.1089, + "step": 526 + }, + { + "epoch": 1.3478260869565217, + "grad_norm": 0.4571075110202611, + "learning_rate": 1.8302894411289303e-05, + "loss": 1.0576, + "step": 527 + }, + { + "epoch": 1.350383631713555, + "grad_norm": 0.4136171758085541, + "learning_rate": 1.8292927157751964e-05, + "loss": 1.1227, + "step": 528 + }, + { + "epoch": 1.3529411764705883, + "grad_norm": 0.32738816543873817, + "learning_rate": 1.8282933451161645e-05, + "loss": 1.0839, + "step": 529 + }, + { + "epoch": 1.3554987212276215, + "grad_norm": 0.263019612016907, + "learning_rate": 1.8272913323396598e-05, + "loss": 1.0556, + "step": 530 + }, + { + "epoch": 1.3580562659846547, + "grad_norm": 0.2915540173192062, + "learning_rate": 1.8262866806419365e-05, + "loss": 1.0941, + "step": 531 + }, + { + "epoch": 1.3606138107416879, + "grad_norm": 0.3503149189432805, + "learning_rate": 1.8252793932276652e-05, + "loss": 1.0782, + "step": 532 + }, + { + "epoch": 1.3631713554987213, + "grad_norm": 0.3642123746279616, + "learning_rate": 1.8242694733099248e-05, + "loss": 1.1175, + "step": 533 + }, + { + "epoch": 1.3657289002557544, + "grad_norm": 0.25071410695368307, + "learning_rate": 1.823256924110191e-05, + "loss": 1.0834, + "step": 534 + }, + { + "epoch": 1.3682864450127878, + "grad_norm": 0.3177321847491843, + "learning_rate": 1.822241748858327e-05, + "loss": 1.0943, + "step": 535 + }, + { + "epoch": 1.370843989769821, + "grad_norm": 0.404922841317797, + "learning_rate": 1.821223950792572e-05, + "loss": 1.1287, + "step": 536 + }, + { + "epoch": 1.3734015345268542, + "grad_norm": 0.3420279383578445, + "learning_rate": 1.8202035331595322e-05, + "loss": 1.0849, + "step": 537 + }, + { + "epoch": 1.3759590792838874, + "grad_norm": 0.2634701796758814, + "learning_rate": 1.8191804992141694e-05, + "loss": 1.0976, + "step": 538 + }, + { + "epoch": 1.3785166240409208, + "grad_norm": 0.2444475712273965, + "learning_rate": 1.818154852219791e-05, + "loss": 1.082, + "step": 539 + }, + { + "epoch": 1.381074168797954, + "grad_norm": 0.26246699977104565, + "learning_rate": 1.8171265954480396e-05, + "loss": 1.0908, + "step": 540 + }, + { + "epoch": 1.3836317135549872, + "grad_norm": 0.2779279537467054, + "learning_rate": 1.8160957321788826e-05, + "loss": 1.1171, + "step": 541 + }, + { + "epoch": 1.3861892583120206, + "grad_norm": 0.28014567822266767, + "learning_rate": 1.8150622657006017e-05, + "loss": 1.1102, + "step": 542 + }, + { + "epoch": 1.3887468030690537, + "grad_norm": 0.252626441198564, + "learning_rate": 1.814026199309783e-05, + "loss": 1.0649, + "step": 543 + }, + { + "epoch": 1.391304347826087, + "grad_norm": 0.2177931908988342, + "learning_rate": 1.8129875363113045e-05, + "loss": 1.0855, + "step": 544 + }, + { + "epoch": 1.39386189258312, + "grad_norm": 0.3025811951941802, + "learning_rate": 1.811946280018328e-05, + "loss": 1.074, + "step": 545 + }, + { + "epoch": 1.3964194373401535, + "grad_norm": 0.3295849163261702, + "learning_rate": 1.8109024337522875e-05, + "loss": 1.053, + "step": 546 + }, + { + "epoch": 1.3989769820971867, + "grad_norm": 0.24317934869172414, + "learning_rate": 1.8098560008428778e-05, + "loss": 1.0931, + "step": 547 + }, + { + "epoch": 1.40153452685422, + "grad_norm": 0.2760178217098737, + "learning_rate": 1.8088069846280456e-05, + "loss": 1.0836, + "step": 548 + }, + { + "epoch": 1.4040920716112533, + "grad_norm": 0.31057489529721605, + "learning_rate": 1.8077553884539772e-05, + "loss": 1.1068, + "step": 549 + }, + { + "epoch": 1.4066496163682864, + "grad_norm": 0.30877701785098377, + "learning_rate": 1.8067012156750892e-05, + "loss": 1.0469, + "step": 550 + }, + { + "epoch": 1.4092071611253196, + "grad_norm": 0.24875376222756798, + "learning_rate": 1.8056444696540162e-05, + "loss": 1.0579, + "step": 551 + }, + { + "epoch": 1.4117647058823528, + "grad_norm": 0.300038374330651, + "learning_rate": 1.804585153761602e-05, + "loss": 1.1032, + "step": 552 + }, + { + "epoch": 1.4143222506393862, + "grad_norm": 0.26562355653772984, + "learning_rate": 1.803523271376887e-05, + "loss": 1.087, + "step": 553 + }, + { + "epoch": 1.4168797953964194, + "grad_norm": 0.27772856419594794, + "learning_rate": 1.802458825887099e-05, + "loss": 1.0976, + "step": 554 + }, + { + "epoch": 1.4194373401534528, + "grad_norm": 0.27940847480468906, + "learning_rate": 1.8013918206876416e-05, + "loss": 1.1109, + "step": 555 + }, + { + "epoch": 1.421994884910486, + "grad_norm": 0.23839863723478302, + "learning_rate": 1.8003222591820827e-05, + "loss": 1.0955, + "step": 556 + }, + { + "epoch": 1.4245524296675192, + "grad_norm": 0.3544376729040761, + "learning_rate": 1.7992501447821454e-05, + "loss": 1.0815, + "step": 557 + }, + { + "epoch": 1.4271099744245523, + "grad_norm": 0.3190990432271867, + "learning_rate": 1.7981754809076952e-05, + "loss": 1.0743, + "step": 558 + }, + { + "epoch": 1.4296675191815857, + "grad_norm": 0.30878135684811764, + "learning_rate": 1.7970982709867307e-05, + "loss": 1.0665, + "step": 559 + }, + { + "epoch": 1.432225063938619, + "grad_norm": 0.2667660845147987, + "learning_rate": 1.7960185184553717e-05, + "loss": 1.0576, + "step": 560 + }, + { + "epoch": 1.434782608695652, + "grad_norm": 0.27959465142572, + "learning_rate": 1.7949362267578485e-05, + "loss": 1.07, + "step": 561 + }, + { + "epoch": 1.4373401534526855, + "grad_norm": 0.3088308257405165, + "learning_rate": 1.793851399346491e-05, + "loss": 1.0837, + "step": 562 + }, + { + "epoch": 1.4398976982097187, + "grad_norm": 0.23954634942810205, + "learning_rate": 1.7927640396817172e-05, + "loss": 1.1108, + "step": 563 + }, + { + "epoch": 1.4424552429667519, + "grad_norm": 0.3054367971170986, + "learning_rate": 1.7916741512320227e-05, + "loss": 1.109, + "step": 564 + }, + { + "epoch": 1.445012787723785, + "grad_norm": 0.30343643697738276, + "learning_rate": 1.7905817374739706e-05, + "loss": 1.1167, + "step": 565 + }, + { + "epoch": 1.4475703324808185, + "grad_norm": 0.2863389806141574, + "learning_rate": 1.789486801892177e-05, + "loss": 1.0784, + "step": 566 + }, + { + "epoch": 1.4501278772378516, + "grad_norm": 0.21449179573185784, + "learning_rate": 1.7883893479793052e-05, + "loss": 1.0596, + "step": 567 + }, + { + "epoch": 1.452685421994885, + "grad_norm": 0.281747796628926, + "learning_rate": 1.7872893792360488e-05, + "loss": 1.0907, + "step": 568 + }, + { + "epoch": 1.4552429667519182, + "grad_norm": 0.3606611327789867, + "learning_rate": 1.7861868991711245e-05, + "loss": 1.0833, + "step": 569 + }, + { + "epoch": 1.4578005115089514, + "grad_norm": 0.29103388553102216, + "learning_rate": 1.78508191130126e-05, + "loss": 1.0702, + "step": 570 + }, + { + "epoch": 1.4603580562659846, + "grad_norm": 0.2515965600824205, + "learning_rate": 1.783974419151182e-05, + "loss": 1.092, + "step": 571 + }, + { + "epoch": 1.4629156010230178, + "grad_norm": 0.3852115976211409, + "learning_rate": 1.782864426253606e-05, + "loss": 1.0969, + "step": 572 + }, + { + "epoch": 1.4654731457800512, + "grad_norm": 0.4829141442027968, + "learning_rate": 1.7817519361492227e-05, + "loss": 1.0826, + "step": 573 + }, + { + "epoch": 1.4680306905370843, + "grad_norm": 0.3459344778256553, + "learning_rate": 1.7806369523866914e-05, + "loss": 1.0674, + "step": 574 + }, + { + "epoch": 1.4705882352941178, + "grad_norm": 0.29132112478633115, + "learning_rate": 1.779519478522623e-05, + "loss": 1.1002, + "step": 575 + }, + { + "epoch": 1.473145780051151, + "grad_norm": 0.2968482079990474, + "learning_rate": 1.7783995181215727e-05, + "loss": 1.1042, + "step": 576 + }, + { + "epoch": 1.4757033248081841, + "grad_norm": 0.27855659630768903, + "learning_rate": 1.7772770747560276e-05, + "loss": 1.0958, + "step": 577 + }, + { + "epoch": 1.4782608695652173, + "grad_norm": 0.2800062391704218, + "learning_rate": 1.7761521520063945e-05, + "loss": 1.0857, + "step": 578 + }, + { + "epoch": 1.4808184143222507, + "grad_norm": 0.3422253981462545, + "learning_rate": 1.775024753460989e-05, + "loss": 1.0638, + "step": 579 + }, + { + "epoch": 1.4833759590792839, + "grad_norm": 0.3042841136396043, + "learning_rate": 1.773894882716024e-05, + "loss": 1.081, + "step": 580 + }, + { + "epoch": 1.485933503836317, + "grad_norm": 0.3584276848759106, + "learning_rate": 1.7727625433755992e-05, + "loss": 1.0979, + "step": 581 + }, + { + "epoch": 1.4884910485933505, + "grad_norm": 0.4132176697960573, + "learning_rate": 1.7716277390516875e-05, + "loss": 1.0872, + "step": 582 + }, + { + "epoch": 1.4910485933503836, + "grad_norm": 0.39456958648721346, + "learning_rate": 1.7704904733641255e-05, + "loss": 1.1129, + "step": 583 + }, + { + "epoch": 1.4936061381074168, + "grad_norm": 0.3047880364961936, + "learning_rate": 1.7693507499406003e-05, + "loss": 1.1183, + "step": 584 + }, + { + "epoch": 1.49616368286445, + "grad_norm": 0.27995089336581735, + "learning_rate": 1.76820857241664e-05, + "loss": 1.0564, + "step": 585 + }, + { + "epoch": 1.4987212276214834, + "grad_norm": 0.4582307602852646, + "learning_rate": 1.7670639444355998e-05, + "loss": 1.0723, + "step": 586 + }, + { + "epoch": 1.5012787723785166, + "grad_norm": 0.3974212413813853, + "learning_rate": 1.7659168696486523e-05, + "loss": 1.0807, + "step": 587 + }, + { + "epoch": 1.50383631713555, + "grad_norm": 0.26098921400780506, + "learning_rate": 1.764767351714774e-05, + "loss": 1.1072, + "step": 588 + }, + { + "epoch": 1.5063938618925832, + "grad_norm": 0.3073004608853978, + "learning_rate": 1.763615394300735e-05, + "loss": 1.048, + "step": 589 + }, + { + "epoch": 1.5089514066496164, + "grad_norm": 0.3496180311318968, + "learning_rate": 1.762461001081088e-05, + "loss": 1.0773, + "step": 590 + }, + { + "epoch": 1.5115089514066495, + "grad_norm": 0.3459451539906394, + "learning_rate": 1.7613041757381538e-05, + "loss": 1.0509, + "step": 591 + }, + { + "epoch": 1.5140664961636827, + "grad_norm": 0.2883593584409338, + "learning_rate": 1.7601449219620127e-05, + "loss": 1.1046, + "step": 592 + }, + { + "epoch": 1.5166240409207161, + "grad_norm": 0.3418272120044329, + "learning_rate": 1.7589832434504903e-05, + "loss": 1.0329, + "step": 593 + }, + { + "epoch": 1.5191815856777495, + "grad_norm": 0.3955352555839306, + "learning_rate": 1.757819143909147e-05, + "loss": 1.1, + "step": 594 + }, + { + "epoch": 1.5217391304347827, + "grad_norm": 0.28213312623631964, + "learning_rate": 1.7566526270512663e-05, + "loss": 1.0706, + "step": 595 + }, + { + "epoch": 1.5242966751918159, + "grad_norm": 0.2723059300728885, + "learning_rate": 1.755483696597842e-05, + "loss": 1.0811, + "step": 596 + }, + { + "epoch": 1.526854219948849, + "grad_norm": 0.3521368659597091, + "learning_rate": 1.754312356277567e-05, + "loss": 1.0892, + "step": 597 + }, + { + "epoch": 1.5294117647058822, + "grad_norm": 0.31456673262803825, + "learning_rate": 1.753138609826822e-05, + "loss": 1.0919, + "step": 598 + }, + { + "epoch": 1.5319693094629157, + "grad_norm": 0.2616348598492441, + "learning_rate": 1.7519624609896614e-05, + "loss": 1.0679, + "step": 599 + }, + { + "epoch": 1.5345268542199488, + "grad_norm": 0.2813410383684714, + "learning_rate": 1.750783913517804e-05, + "loss": 1.1173, + "step": 600 + }, + { + "epoch": 1.5370843989769822, + "grad_norm": 0.21204624924930726, + "learning_rate": 1.749602971170619e-05, + "loss": 1.065, + "step": 601 + }, + { + "epoch": 1.5396419437340154, + "grad_norm": 0.28741509553485445, + "learning_rate": 1.748419637715116e-05, + "loss": 1.0628, + "step": 602 + }, + { + "epoch": 1.5421994884910486, + "grad_norm": 0.38267662644256223, + "learning_rate": 1.7472339169259306e-05, + "loss": 1.0757, + "step": 603 + }, + { + "epoch": 1.5447570332480818, + "grad_norm": 0.29841393332702054, + "learning_rate": 1.7460458125853143e-05, + "loss": 1.0865, + "step": 604 + }, + { + "epoch": 1.547314578005115, + "grad_norm": 0.2530976843608534, + "learning_rate": 1.744855328483122e-05, + "loss": 1.0731, + "step": 605 + }, + { + "epoch": 1.5498721227621484, + "grad_norm": 0.32358407442808673, + "learning_rate": 1.7436624684167984e-05, + "loss": 1.0495, + "step": 606 + }, + { + "epoch": 1.5524296675191815, + "grad_norm": 0.31348665372322687, + "learning_rate": 1.7424672361913685e-05, + "loss": 1.0736, + "step": 607 + }, + { + "epoch": 1.554987212276215, + "grad_norm": 0.2418916979845008, + "learning_rate": 1.7412696356194233e-05, + "loss": 1.0983, + "step": 608 + }, + { + "epoch": 1.5575447570332481, + "grad_norm": 0.2946315527683715, + "learning_rate": 1.740069670521109e-05, + "loss": 1.0942, + "step": 609 + }, + { + "epoch": 1.5601023017902813, + "grad_norm": 0.26418886749171827, + "learning_rate": 1.738867344724114e-05, + "loss": 1.0643, + "step": 610 + }, + { + "epoch": 1.5626598465473145, + "grad_norm": 0.2375625195085471, + "learning_rate": 1.7376626620636558e-05, + "loss": 1.0981, + "step": 611 + }, + { + "epoch": 1.5652173913043477, + "grad_norm": 0.30919626032852776, + "learning_rate": 1.7364556263824718e-05, + "loss": 1.0884, + "step": 612 + }, + { + "epoch": 1.567774936061381, + "grad_norm": 0.2723698903787219, + "learning_rate": 1.7352462415308043e-05, + "loss": 1.0784, + "step": 613 + }, + { + "epoch": 1.5703324808184145, + "grad_norm": 0.22887874198225955, + "learning_rate": 1.734034511366389e-05, + "loss": 1.0824, + "step": 614 + }, + { + "epoch": 1.5728900255754477, + "grad_norm": 0.29550746740956296, + "learning_rate": 1.7328204397544426e-05, + "loss": 1.1013, + "step": 615 + }, + { + "epoch": 1.5754475703324808, + "grad_norm": 0.28314944714751183, + "learning_rate": 1.731604030567651e-05, + "loss": 1.0553, + "step": 616 + }, + { + "epoch": 1.578005115089514, + "grad_norm": 0.2504546352702014, + "learning_rate": 1.730385287686156e-05, + "loss": 1.0892, + "step": 617 + }, + { + "epoch": 1.5805626598465472, + "grad_norm": 0.2461609364517437, + "learning_rate": 1.7291642149975447e-05, + "loss": 1.0825, + "step": 618 + }, + { + "epoch": 1.5831202046035806, + "grad_norm": 0.3290342905968507, + "learning_rate": 1.7279408163968344e-05, + "loss": 1.0682, + "step": 619 + }, + { + "epoch": 1.5856777493606138, + "grad_norm": 0.35081666137815004, + "learning_rate": 1.7267150957864626e-05, + "loss": 1.0562, + "step": 620 + }, + { + "epoch": 1.5882352941176472, + "grad_norm": 0.28999486002511976, + "learning_rate": 1.7254870570762734e-05, + "loss": 1.0859, + "step": 621 + }, + { + "epoch": 1.5907928388746804, + "grad_norm": 0.20775563782106674, + "learning_rate": 1.724256704183505e-05, + "loss": 1.0983, + "step": 622 + }, + { + "epoch": 1.5933503836317136, + "grad_norm": 0.25046303669870285, + "learning_rate": 1.7230240410327784e-05, + "loss": 1.0722, + "step": 623 + }, + { + "epoch": 1.5959079283887467, + "grad_norm": 0.3120122500057057, + "learning_rate": 1.7217890715560822e-05, + "loss": 1.1117, + "step": 624 + }, + { + "epoch": 1.59846547314578, + "grad_norm": 0.23438996871875717, + "learning_rate": 1.720551799692764e-05, + "loss": 1.1324, + "step": 625 + }, + { + "epoch": 1.6010230179028133, + "grad_norm": 0.20904515546827102, + "learning_rate": 1.7193122293895137e-05, + "loss": 1.0592, + "step": 626 + }, + { + "epoch": 1.6035805626598465, + "grad_norm": 0.32914062124072535, + "learning_rate": 1.7180703646003536e-05, + "loss": 1.0697, + "step": 627 + }, + { + "epoch": 1.60613810741688, + "grad_norm": 0.37833542302745304, + "learning_rate": 1.7168262092866252e-05, + "loss": 1.1086, + "step": 628 + }, + { + "epoch": 1.608695652173913, + "grad_norm": 0.2633777304132735, + "learning_rate": 1.715579767416976e-05, + "loss": 1.0619, + "step": 629 + }, + { + "epoch": 1.6112531969309463, + "grad_norm": 0.26093223398559484, + "learning_rate": 1.714331042967348e-05, + "loss": 1.0676, + "step": 630 + }, + { + "epoch": 1.6138107416879794, + "grad_norm": 0.31095342245005936, + "learning_rate": 1.7130800399209632e-05, + "loss": 1.0759, + "step": 631 + }, + { + "epoch": 1.6163682864450126, + "grad_norm": 0.4055927660522638, + "learning_rate": 1.7118267622683123e-05, + "loss": 1.0696, + "step": 632 + }, + { + "epoch": 1.618925831202046, + "grad_norm": 0.38006329300781394, + "learning_rate": 1.7105712140071427e-05, + "loss": 1.052, + "step": 633 + }, + { + "epoch": 1.6214833759590794, + "grad_norm": 0.25696206250183334, + "learning_rate": 1.7093133991424425e-05, + "loss": 1.1003, + "step": 634 + }, + { + "epoch": 1.6240409207161126, + "grad_norm": 0.4045520870520815, + "learning_rate": 1.7080533216864318e-05, + "loss": 1.053, + "step": 635 + }, + { + "epoch": 1.6265984654731458, + "grad_norm": 0.3944457103687557, + "learning_rate": 1.7067909856585474e-05, + "loss": 1.0994, + "step": 636 + }, + { + "epoch": 1.629156010230179, + "grad_norm": 0.28734017232782005, + "learning_rate": 1.70552639508543e-05, + "loss": 1.0932, + "step": 637 + }, + { + "epoch": 1.6317135549872122, + "grad_norm": 0.30914836592137274, + "learning_rate": 1.7042595540009123e-05, + "loss": 1.0777, + "step": 638 + }, + { + "epoch": 1.6342710997442456, + "grad_norm": 0.3249860111231293, + "learning_rate": 1.7029904664460066e-05, + "loss": 1.1112, + "step": 639 + }, + { + "epoch": 1.6368286445012787, + "grad_norm": 0.31287350836163863, + "learning_rate": 1.7017191364688896e-05, + "loss": 1.0452, + "step": 640 + }, + { + "epoch": 1.6393861892583121, + "grad_norm": 0.33833362239628567, + "learning_rate": 1.7004455681248918e-05, + "loss": 1.0658, + "step": 641 + }, + { + "epoch": 1.6419437340153453, + "grad_norm": 0.242559004260857, + "learning_rate": 1.699169765476484e-05, + "loss": 1.0679, + "step": 642 + }, + { + "epoch": 1.6445012787723785, + "grad_norm": 0.2790491473058214, + "learning_rate": 1.6978917325932628e-05, + "loss": 1.0833, + "step": 643 + }, + { + "epoch": 1.6470588235294117, + "grad_norm": 0.23663508280237597, + "learning_rate": 1.6966114735519408e-05, + "loss": 1.1294, + "step": 644 + }, + { + "epoch": 1.6496163682864449, + "grad_norm": 0.19911244010343537, + "learning_rate": 1.6953289924363297e-05, + "loss": 1.0381, + "step": 645 + }, + { + "epoch": 1.6521739130434783, + "grad_norm": 0.25134030034591326, + "learning_rate": 1.6940442933373305e-05, + "loss": 1.0953, + "step": 646 + }, + { + "epoch": 1.6547314578005117, + "grad_norm": 0.2548060423300198, + "learning_rate": 1.6927573803529186e-05, + "loss": 1.0531, + "step": 647 + }, + { + "epoch": 1.6572890025575449, + "grad_norm": 0.25550208965898963, + "learning_rate": 1.6914682575881312e-05, + "loss": 1.0712, + "step": 648 + }, + { + "epoch": 1.659846547314578, + "grad_norm": 0.28172858056396866, + "learning_rate": 1.690176929155056e-05, + "loss": 1.0762, + "step": 649 + }, + { + "epoch": 1.6624040920716112, + "grad_norm": 0.22586367515717168, + "learning_rate": 1.6888833991728137e-05, + "loss": 1.0598, + "step": 650 + }, + { + "epoch": 1.6649616368286444, + "grad_norm": 0.23131122480912064, + "learning_rate": 1.68758767176755e-05, + "loss": 1.0865, + "step": 651 + }, + { + "epoch": 1.6675191815856778, + "grad_norm": 0.25447757509954316, + "learning_rate": 1.6862897510724176e-05, + "loss": 1.1109, + "step": 652 + }, + { + "epoch": 1.670076726342711, + "grad_norm": 0.2553359805367843, + "learning_rate": 1.6849896412275684e-05, + "loss": 1.086, + "step": 653 + }, + { + "epoch": 1.6726342710997444, + "grad_norm": 0.2745541400801621, + "learning_rate": 1.6836873463801348e-05, + "loss": 1.0907, + "step": 654 + }, + { + "epoch": 1.6751918158567776, + "grad_norm": 0.2730286942160579, + "learning_rate": 1.68238287068422e-05, + "loss": 1.0798, + "step": 655 + }, + { + "epoch": 1.6777493606138107, + "grad_norm": 0.26706214928876043, + "learning_rate": 1.6810762183008845e-05, + "loss": 1.0662, + "step": 656 + }, + { + "epoch": 1.680306905370844, + "grad_norm": 0.2690779370594318, + "learning_rate": 1.67976739339813e-05, + "loss": 1.0685, + "step": 657 + }, + { + "epoch": 1.682864450127877, + "grad_norm": 0.24513041409636943, + "learning_rate": 1.67845640015089e-05, + "loss": 1.0904, + "step": 658 + }, + { + "epoch": 1.6854219948849105, + "grad_norm": 0.2808465989622888, + "learning_rate": 1.6771432427410136e-05, + "loss": 1.09, + "step": 659 + }, + { + "epoch": 1.6879795396419437, + "grad_norm": 0.3142220119591379, + "learning_rate": 1.6758279253572548e-05, + "loss": 1.107, + "step": 660 + }, + { + "epoch": 1.690537084398977, + "grad_norm": 0.23175871520363883, + "learning_rate": 1.6745104521952552e-05, + "loss": 1.0859, + "step": 661 + }, + { + "epoch": 1.6930946291560103, + "grad_norm": 0.24672930460906686, + "learning_rate": 1.673190827457535e-05, + "loss": 1.078, + "step": 662 + }, + { + "epoch": 1.6956521739130435, + "grad_norm": 0.2601230359714199, + "learning_rate": 1.6718690553534766e-05, + "loss": 1.0998, + "step": 663 + }, + { + "epoch": 1.6982097186700766, + "grad_norm": 0.27044003845983056, + "learning_rate": 1.670545140099312e-05, + "loss": 1.0856, + "step": 664 + }, + { + "epoch": 1.7007672634271098, + "grad_norm": 0.2323455396605389, + "learning_rate": 1.6692190859181102e-05, + "loss": 1.0898, + "step": 665 + }, + { + "epoch": 1.7033248081841432, + "grad_norm": 0.24195265129162116, + "learning_rate": 1.6678908970397623e-05, + "loss": 1.0609, + "step": 666 + }, + { + "epoch": 1.7058823529411766, + "grad_norm": 0.2417510214052941, + "learning_rate": 1.6665605777009698e-05, + "loss": 1.0816, + "step": 667 + }, + { + "epoch": 1.7084398976982098, + "grad_norm": 0.2651395518131203, + "learning_rate": 1.665228132145228e-05, + "loss": 1.0705, + "step": 668 + }, + { + "epoch": 1.710997442455243, + "grad_norm": 0.22279503342335966, + "learning_rate": 1.6638935646228168e-05, + "loss": 1.0969, + "step": 669 + }, + { + "epoch": 1.7135549872122762, + "grad_norm": 0.27626190830784103, + "learning_rate": 1.6625568793907835e-05, + "loss": 1.0758, + "step": 670 + }, + { + "epoch": 1.7161125319693094, + "grad_norm": 0.293354770533148, + "learning_rate": 1.6612180807129305e-05, + "loss": 1.0928, + "step": 671 + }, + { + "epoch": 1.7186700767263428, + "grad_norm": 0.22212142950085934, + "learning_rate": 1.6598771728598024e-05, + "loss": 1.0943, + "step": 672 + }, + { + "epoch": 1.721227621483376, + "grad_norm": 0.23025373127686816, + "learning_rate": 1.6585341601086713e-05, + "loss": 1.0714, + "step": 673 + }, + { + "epoch": 1.7237851662404093, + "grad_norm": 0.3137233490884019, + "learning_rate": 1.6571890467435233e-05, + "loss": 1.1156, + "step": 674 + }, + { + "epoch": 1.7263427109974425, + "grad_norm": 0.2496052608065762, + "learning_rate": 1.655841837055046e-05, + "loss": 1.0854, + "step": 675 + }, + { + "epoch": 1.7289002557544757, + "grad_norm": 0.240092026635267, + "learning_rate": 1.6544925353406127e-05, + "loss": 1.0516, + "step": 676 + }, + { + "epoch": 1.7314578005115089, + "grad_norm": 0.3007943197702808, + "learning_rate": 1.653141145904271e-05, + "loss": 1.0684, + "step": 677 + }, + { + "epoch": 1.734015345268542, + "grad_norm": 0.25757031156270827, + "learning_rate": 1.651787673056728e-05, + "loss": 1.07, + "step": 678 + }, + { + "epoch": 1.7365728900255755, + "grad_norm": 0.27328963709800014, + "learning_rate": 1.6504321211153355e-05, + "loss": 1.0711, + "step": 679 + }, + { + "epoch": 1.7391304347826086, + "grad_norm": 0.33112287394739875, + "learning_rate": 1.6490744944040776e-05, + "loss": 1.0726, + "step": 680 + }, + { + "epoch": 1.741687979539642, + "grad_norm": 0.3214662959562765, + "learning_rate": 1.6477147972535577e-05, + "loss": 1.061, + "step": 681 + }, + { + "epoch": 1.7442455242966752, + "grad_norm": 0.3027544525583265, + "learning_rate": 1.6463530340009817e-05, + "loss": 1.1065, + "step": 682 + }, + { + "epoch": 1.7468030690537084, + "grad_norm": 0.2609885345467691, + "learning_rate": 1.644989208990148e-05, + "loss": 1.0993, + "step": 683 + }, + { + "epoch": 1.7493606138107416, + "grad_norm": 0.2993893219580389, + "learning_rate": 1.6436233265714297e-05, + "loss": 1.084, + "step": 684 + }, + { + "epoch": 1.7519181585677748, + "grad_norm": 0.24875924140149924, + "learning_rate": 1.6422553911017642e-05, + "loss": 1.0761, + "step": 685 + }, + { + "epoch": 1.7544757033248082, + "grad_norm": 0.2749545983405175, + "learning_rate": 1.6408854069446374e-05, + "loss": 1.1069, + "step": 686 + }, + { + "epoch": 1.7570332480818416, + "grad_norm": 0.25348934595842737, + "learning_rate": 1.6395133784700696e-05, + "loss": 1.0813, + "step": 687 + }, + { + "epoch": 1.7595907928388748, + "grad_norm": 0.29036852249839945, + "learning_rate": 1.638139310054603e-05, + "loss": 1.0692, + "step": 688 + }, + { + "epoch": 1.762148337595908, + "grad_norm": 0.2420069325944099, + "learning_rate": 1.6367632060812856e-05, + "loss": 1.0506, + "step": 689 + }, + { + "epoch": 1.7647058823529411, + "grad_norm": 0.19405618601521618, + "learning_rate": 1.6353850709396605e-05, + "loss": 1.0764, + "step": 690 + }, + { + "epoch": 1.7672634271099743, + "grad_norm": 0.22821864695269162, + "learning_rate": 1.6340049090257476e-05, + "loss": 1.0887, + "step": 691 + }, + { + "epoch": 1.7698209718670077, + "grad_norm": 0.2561864664730904, + "learning_rate": 1.6326227247420337e-05, + "loss": 1.0812, + "step": 692 + }, + { + "epoch": 1.772378516624041, + "grad_norm": 0.23401055500498363, + "learning_rate": 1.6312385224974555e-05, + "loss": 1.0971, + "step": 693 + }, + { + "epoch": 1.7749360613810743, + "grad_norm": 0.2810618748667006, + "learning_rate": 1.6298523067073872e-05, + "loss": 1.0485, + "step": 694 + }, + { + "epoch": 1.7774936061381075, + "grad_norm": 0.23900591068457366, + "learning_rate": 1.6284640817936254e-05, + "loss": 1.0629, + "step": 695 + }, + { + "epoch": 1.7800511508951407, + "grad_norm": 0.307631766044326, + "learning_rate": 1.6270738521843762e-05, + "loss": 1.097, + "step": 696 + }, + { + "epoch": 1.7826086956521738, + "grad_norm": 0.23849786656568, + "learning_rate": 1.62568162231424e-05, + "loss": 1.074, + "step": 697 + }, + { + "epoch": 1.785166240409207, + "grad_norm": 0.3363422147425774, + "learning_rate": 1.6242873966241975e-05, + "loss": 1.09, + "step": 698 + }, + { + "epoch": 1.7877237851662404, + "grad_norm": 0.2785499919723351, + "learning_rate": 1.6228911795615953e-05, + "loss": 1.0786, + "step": 699 + }, + { + "epoch": 1.7902813299232738, + "grad_norm": 0.2592472181350103, + "learning_rate": 1.6214929755801335e-05, + "loss": 1.0959, + "step": 700 + }, + { + "epoch": 1.792838874680307, + "grad_norm": 0.27675406867952207, + "learning_rate": 1.6200927891398486e-05, + "loss": 1.0617, + "step": 701 + }, + { + "epoch": 1.7953964194373402, + "grad_norm": 0.2707725171412112, + "learning_rate": 1.6186906247071026e-05, + "loss": 1.082, + "step": 702 + }, + { + "epoch": 1.7979539641943734, + "grad_norm": 0.23850993661114755, + "learning_rate": 1.6172864867545653e-05, + "loss": 1.0842, + "step": 703 + }, + { + "epoch": 1.8005115089514065, + "grad_norm": 0.26668294516379776, + "learning_rate": 1.615880379761202e-05, + "loss": 1.0643, + "step": 704 + }, + { + "epoch": 1.80306905370844, + "grad_norm": 0.22065847889215587, + "learning_rate": 1.6144723082122596e-05, + "loss": 1.0953, + "step": 705 + }, + { + "epoch": 1.8056265984654731, + "grad_norm": 0.22726046087383975, + "learning_rate": 1.613062276599251e-05, + "loss": 1.0457, + "step": 706 + }, + { + "epoch": 1.8081841432225065, + "grad_norm": 0.2592104058900014, + "learning_rate": 1.611650289419942e-05, + "loss": 1.0503, + "step": 707 + }, + { + "epoch": 1.8107416879795397, + "grad_norm": 0.2336564349302167, + "learning_rate": 1.610236351178336e-05, + "loss": 1.0763, + "step": 708 + }, + { + "epoch": 1.813299232736573, + "grad_norm": 0.2528721298539705, + "learning_rate": 1.6088204663846596e-05, + "loss": 1.0253, + "step": 709 + }, + { + "epoch": 1.815856777493606, + "grad_norm": 0.2206997256192134, + "learning_rate": 1.6074026395553487e-05, + "loss": 1.0709, + "step": 710 + }, + { + "epoch": 1.8184143222506393, + "grad_norm": 0.27638089805935434, + "learning_rate": 1.6059828752130347e-05, + "loss": 1.1258, + "step": 711 + }, + { + "epoch": 1.8209718670076727, + "grad_norm": 0.22616837000778253, + "learning_rate": 1.6045611778865282e-05, + "loss": 1.0728, + "step": 712 + }, + { + "epoch": 1.8235294117647058, + "grad_norm": 0.2494603672128, + "learning_rate": 1.6031375521108068e-05, + "loss": 1.0817, + "step": 713 + }, + { + "epoch": 1.8260869565217392, + "grad_norm": 0.20041954468789, + "learning_rate": 1.601712002426999e-05, + "loss": 1.0707, + "step": 714 + }, + { + "epoch": 1.8286445012787724, + "grad_norm": 0.2148146257062548, + "learning_rate": 1.6002845333823696e-05, + "loss": 1.0537, + "step": 715 + }, + { + "epoch": 1.8312020460358056, + "grad_norm": 0.22346033453757363, + "learning_rate": 1.5988551495303072e-05, + "loss": 1.0707, + "step": 716 + }, + { + "epoch": 1.8337595907928388, + "grad_norm": 0.2066902162333923, + "learning_rate": 1.5974238554303078e-05, + "loss": 1.041, + "step": 717 + }, + { + "epoch": 1.836317135549872, + "grad_norm": 0.2813820556225397, + "learning_rate": 1.5959906556479595e-05, + "loss": 1.0912, + "step": 718 + }, + { + "epoch": 1.8388746803069054, + "grad_norm": 0.2509331071041519, + "learning_rate": 1.5945555547549316e-05, + "loss": 1.0905, + "step": 719 + }, + { + "epoch": 1.8414322250639388, + "grad_norm": 0.26806908429404896, + "learning_rate": 1.5931185573289557e-05, + "loss": 1.0977, + "step": 720 + }, + { + "epoch": 1.843989769820972, + "grad_norm": 0.2456487155316564, + "learning_rate": 1.5916796679538137e-05, + "loss": 1.0983, + "step": 721 + }, + { + "epoch": 1.8465473145780051, + "grad_norm": 0.25480301698890395, + "learning_rate": 1.5902388912193223e-05, + "loss": 1.0538, + "step": 722 + }, + { + "epoch": 1.8491048593350383, + "grad_norm": 0.23153697687225744, + "learning_rate": 1.588796231721319e-05, + "loss": 1.0815, + "step": 723 + }, + { + "epoch": 1.8516624040920715, + "grad_norm": 0.21914721485154337, + "learning_rate": 1.5873516940616468e-05, + "loss": 1.0713, + "step": 724 + }, + { + "epoch": 1.854219948849105, + "grad_norm": 0.24748552983182054, + "learning_rate": 1.5859052828481395e-05, + "loss": 1.1034, + "step": 725 + }, + { + "epoch": 1.856777493606138, + "grad_norm": 0.24836286711397054, + "learning_rate": 1.5844570026946068e-05, + "loss": 1.0254, + "step": 726 + }, + { + "epoch": 1.8593350383631715, + "grad_norm": 0.2477358959532818, + "learning_rate": 1.5830068582208217e-05, + "loss": 1.0513, + "step": 727 + }, + { + "epoch": 1.8618925831202047, + "grad_norm": 0.22729630350906133, + "learning_rate": 1.581554854052502e-05, + "loss": 1.0559, + "step": 728 + }, + { + "epoch": 1.8644501278772379, + "grad_norm": 0.2527375012776645, + "learning_rate": 1.5801009948212992e-05, + "loss": 1.1154, + "step": 729 + }, + { + "epoch": 1.867007672634271, + "grad_norm": 0.2430437408352163, + "learning_rate": 1.578645285164781e-05, + "loss": 1.0687, + "step": 730 + }, + { + "epoch": 1.8695652173913042, + "grad_norm": 0.20727173935167562, + "learning_rate": 1.5771877297264184e-05, + "loss": 1.0398, + "step": 731 + }, + { + "epoch": 1.8721227621483376, + "grad_norm": 0.21679809470270497, + "learning_rate": 1.5757283331555698e-05, + "loss": 1.0769, + "step": 732 + }, + { + "epoch": 1.8746803069053708, + "grad_norm": 0.23176747130476263, + "learning_rate": 1.5742671001074666e-05, + "loss": 1.0889, + "step": 733 + }, + { + "epoch": 1.8772378516624042, + "grad_norm": 0.23200124479221654, + "learning_rate": 1.5728040352431983e-05, + "loss": 1.0472, + "step": 734 + }, + { + "epoch": 1.8797953964194374, + "grad_norm": 0.2227788458921205, + "learning_rate": 1.5713391432296976e-05, + "loss": 1.0766, + "step": 735 + }, + { + "epoch": 1.8823529411764706, + "grad_norm": 0.2355356583350579, + "learning_rate": 1.5698724287397254e-05, + "loss": 1.1086, + "step": 736 + }, + { + "epoch": 1.8849104859335037, + "grad_norm": 0.24042677593727868, + "learning_rate": 1.5684038964518557e-05, + "loss": 1.0814, + "step": 737 + }, + { + "epoch": 1.887468030690537, + "grad_norm": 0.19933827334366755, + "learning_rate": 1.566933551050462e-05, + "loss": 1.0886, + "step": 738 + }, + { + "epoch": 1.8900255754475703, + "grad_norm": 0.2591080445820066, + "learning_rate": 1.5654613972256997e-05, + "loss": 1.0723, + "step": 739 + }, + { + "epoch": 1.8925831202046037, + "grad_norm": 0.2774761431017259, + "learning_rate": 1.5639874396734943e-05, + "loss": 1.0432, + "step": 740 + }, + { + "epoch": 1.895140664961637, + "grad_norm": 0.22087846289480834, + "learning_rate": 1.5625116830955246e-05, + "loss": 1.0531, + "step": 741 + }, + { + "epoch": 1.89769820971867, + "grad_norm": 0.27029341996359707, + "learning_rate": 1.5610341321992068e-05, + "loss": 1.0597, + "step": 742 + }, + { + "epoch": 1.9002557544757033, + "grad_norm": 0.25764475336363896, + "learning_rate": 1.559554791697682e-05, + "loss": 1.0988, + "step": 743 + }, + { + "epoch": 1.9028132992327365, + "grad_norm": 0.2587858950265225, + "learning_rate": 1.5580736663097998e-05, + "loss": 1.0533, + "step": 744 + }, + { + "epoch": 1.9053708439897699, + "grad_norm": 0.2416978562367689, + "learning_rate": 1.5565907607601025e-05, + "loss": 1.0502, + "step": 745 + }, + { + "epoch": 1.907928388746803, + "grad_norm": 0.2772485628569736, + "learning_rate": 1.5551060797788107e-05, + "loss": 1.0742, + "step": 746 + }, + { + "epoch": 1.9104859335038364, + "grad_norm": 0.2636514324137545, + "learning_rate": 1.5536196281018096e-05, + "loss": 1.0656, + "step": 747 + }, + { + "epoch": 1.9130434782608696, + "grad_norm": 0.2488817318322834, + "learning_rate": 1.5521314104706318e-05, + "loss": 1.0813, + "step": 748 + }, + { + "epoch": 1.9156010230179028, + "grad_norm": 0.2607950822000825, + "learning_rate": 1.5506414316324426e-05, + "loss": 1.1034, + "step": 749 + }, + { + "epoch": 1.918158567774936, + "grad_norm": 0.22986661833547864, + "learning_rate": 1.549149696340026e-05, + "loss": 1.0563, + "step": 750 + }, + { + "epoch": 1.9207161125319692, + "grad_norm": 0.2531299090639568, + "learning_rate": 1.547656209351769e-05, + "loss": 1.1094, + "step": 751 + }, + { + "epoch": 1.9232736572890026, + "grad_norm": 0.2362443082311155, + "learning_rate": 1.5461609754316446e-05, + "loss": 1.0598, + "step": 752 + }, + { + "epoch": 1.9258312020460358, + "grad_norm": 0.2879220642694635, + "learning_rate": 1.5446639993492003e-05, + "loss": 1.0752, + "step": 753 + }, + { + "epoch": 1.9283887468030692, + "grad_norm": 0.3209490019435245, + "learning_rate": 1.5431652858795394e-05, + "loss": 1.0701, + "step": 754 + }, + { + "epoch": 1.9309462915601023, + "grad_norm": 0.2510875490823265, + "learning_rate": 1.5416648398033076e-05, + "loss": 1.0393, + "step": 755 + }, + { + "epoch": 1.9335038363171355, + "grad_norm": 0.31573886950529617, + "learning_rate": 1.5401626659066774e-05, + "loss": 1.0779, + "step": 756 + }, + { + "epoch": 1.9360613810741687, + "grad_norm": 0.2876023977996849, + "learning_rate": 1.5386587689813333e-05, + "loss": 1.0741, + "step": 757 + }, + { + "epoch": 1.938618925831202, + "grad_norm": 0.26378670917157787, + "learning_rate": 1.5371531538244547e-05, + "loss": 1.0554, + "step": 758 + }, + { + "epoch": 1.9411764705882353, + "grad_norm": 0.26374981252096197, + "learning_rate": 1.5356458252387025e-05, + "loss": 1.102, + "step": 759 + }, + { + "epoch": 1.9437340153452687, + "grad_norm": 0.251595744088401, + "learning_rate": 1.534136788032204e-05, + "loss": 1.1062, + "step": 760 + }, + { + "epoch": 1.9462915601023019, + "grad_norm": 0.2926779701415132, + "learning_rate": 1.5326260470185353e-05, + "loss": 1.1062, + "step": 761 + }, + { + "epoch": 1.948849104859335, + "grad_norm": 0.2568048616159297, + "learning_rate": 1.5311136070167076e-05, + "loss": 1.0834, + "step": 762 + }, + { + "epoch": 1.9514066496163682, + "grad_norm": 0.20314175224340375, + "learning_rate": 1.529599472851153e-05, + "loss": 1.0716, + "step": 763 + }, + { + "epoch": 1.9539641943734014, + "grad_norm": 0.23750489185611817, + "learning_rate": 1.5280836493517058e-05, + "loss": 1.0872, + "step": 764 + }, + { + "epoch": 1.9565217391304348, + "grad_norm": 0.21282755180586768, + "learning_rate": 1.5265661413535907e-05, + "loss": 1.0184, + "step": 765 + }, + { + "epoch": 1.959079283887468, + "grad_norm": 0.24625497205334856, + "learning_rate": 1.5250469536974043e-05, + "loss": 1.0848, + "step": 766 + }, + { + "epoch": 1.9616368286445014, + "grad_norm": 0.21996647040639863, + "learning_rate": 1.5235260912291012e-05, + "loss": 1.0531, + "step": 767 + }, + { + "epoch": 1.9641943734015346, + "grad_norm": 0.19256607977575677, + "learning_rate": 1.5220035587999795e-05, + "loss": 1.0798, + "step": 768 + }, + { + "epoch": 1.9667519181585678, + "grad_norm": 0.23024008315602099, + "learning_rate": 1.5204793612666628e-05, + "loss": 1.0884, + "step": 769 + }, + { + "epoch": 1.969309462915601, + "grad_norm": 0.2129496532170118, + "learning_rate": 1.5189535034910874e-05, + "loss": 1.07, + "step": 770 + }, + { + "epoch": 1.9718670076726341, + "grad_norm": 0.20545579206254114, + "learning_rate": 1.5174259903404844e-05, + "loss": 1.0618, + "step": 771 + }, + { + "epoch": 1.9744245524296675, + "grad_norm": 0.2576111775510631, + "learning_rate": 1.5158968266873657e-05, + "loss": 1.0774, + "step": 772 + }, + { + "epoch": 1.976982097186701, + "grad_norm": 0.22350592963669672, + "learning_rate": 1.514366017409508e-05, + "loss": 1.0865, + "step": 773 + }, + { + "epoch": 1.979539641943734, + "grad_norm": 0.2207879071877279, + "learning_rate": 1.5128335673899377e-05, + "loss": 1.0534, + "step": 774 + }, + { + "epoch": 1.9820971867007673, + "grad_norm": 0.24592252516616664, + "learning_rate": 1.5112994815169143e-05, + "loss": 1.0643, + "step": 775 + }, + { + "epoch": 1.9846547314578005, + "grad_norm": 0.24493501976628093, + "learning_rate": 1.509763764683915e-05, + "loss": 1.0827, + "step": 776 + }, + { + "epoch": 1.9872122762148337, + "grad_norm": 0.23172692279011306, + "learning_rate": 1.5082264217896208e-05, + "loss": 1.0622, + "step": 777 + }, + { + "epoch": 1.989769820971867, + "grad_norm": 0.252231827090011, + "learning_rate": 1.5066874577378987e-05, + "loss": 1.1021, + "step": 778 + }, + { + "epoch": 1.9923273657289002, + "grad_norm": 0.26229178894088223, + "learning_rate": 1.5051468774377868e-05, + "loss": 1.0833, + "step": 779 + }, + { + "epoch": 1.9948849104859336, + "grad_norm": 0.28506988516984694, + "learning_rate": 1.5036046858034796e-05, + "loss": 1.0656, + "step": 780 + }, + { + "epoch": 1.9974424552429668, + "grad_norm": 0.3078431102413717, + "learning_rate": 1.5020608877543103e-05, + "loss": 1.0718, + "step": 781 + }, + { + "epoch": 2.0, + "grad_norm": 0.21345572776509852, + "learning_rate": 1.5005154882147373e-05, + "loss": 1.0498, + "step": 782 + }, + { + "epoch": 2.002557544757033, + "grad_norm": 0.34751524748697765, + "learning_rate": 1.4989684921143268e-05, + "loss": 1.0568, + "step": 783 + }, + { + "epoch": 2.0051150895140664, + "grad_norm": 0.2632901250486642, + "learning_rate": 1.4974199043877383e-05, + "loss": 1.044, + "step": 784 + }, + { + "epoch": 2.0076726342710995, + "grad_norm": 0.2710330496616998, + "learning_rate": 1.495869729974708e-05, + "loss": 1.0467, + "step": 785 + }, + { + "epoch": 2.010230179028133, + "grad_norm": 0.23547876798507567, + "learning_rate": 1.4943179738200334e-05, + "loss": 1.1337, + "step": 786 + }, + { + "epoch": 2.0127877237851663, + "grad_norm": 0.24663671849680985, + "learning_rate": 1.4927646408735576e-05, + "loss": 1.0676, + "step": 787 + }, + { + "epoch": 2.0153452685421995, + "grad_norm": 0.21964612551270962, + "learning_rate": 1.4912097360901534e-05, + "loss": 1.0775, + "step": 788 + }, + { + "epoch": 2.0179028132992327, + "grad_norm": 0.20575727001469502, + "learning_rate": 1.489653264429707e-05, + "loss": 1.0604, + "step": 789 + }, + { + "epoch": 2.020460358056266, + "grad_norm": 0.21546616580679193, + "learning_rate": 1.4880952308571039e-05, + "loss": 1.0497, + "step": 790 + }, + { + "epoch": 2.023017902813299, + "grad_norm": 0.21519990807634928, + "learning_rate": 1.4865356403422104e-05, + "loss": 1.0481, + "step": 791 + }, + { + "epoch": 2.0255754475703327, + "grad_norm": 0.23762776400824376, + "learning_rate": 1.4849744978598604e-05, + "loss": 1.0529, + "step": 792 + }, + { + "epoch": 2.028132992327366, + "grad_norm": 0.22687488068426334, + "learning_rate": 1.4834118083898373e-05, + "loss": 1.0449, + "step": 793 + }, + { + "epoch": 2.030690537084399, + "grad_norm": 0.2712907192245646, + "learning_rate": 1.4818475769168594e-05, + "loss": 1.0448, + "step": 794 + }, + { + "epoch": 2.0332480818414322, + "grad_norm": 0.25795810211644293, + "learning_rate": 1.4802818084305646e-05, + "loss": 1.0491, + "step": 795 + }, + { + "epoch": 2.0358056265984654, + "grad_norm": 0.2276186453417444, + "learning_rate": 1.4787145079254926e-05, + "loss": 1.0631, + "step": 796 + }, + { + "epoch": 2.0383631713554986, + "grad_norm": 0.2853584063467578, + "learning_rate": 1.4771456804010703e-05, + "loss": 1.0844, + "step": 797 + }, + { + "epoch": 2.040920716112532, + "grad_norm": 0.24492230560921283, + "learning_rate": 1.4755753308615953e-05, + "loss": 1.0561, + "step": 798 + }, + { + "epoch": 2.0434782608695654, + "grad_norm": 0.23340351705875467, + "learning_rate": 1.474003464316221e-05, + "loss": 1.0572, + "step": 799 + }, + { + "epoch": 2.0460358056265986, + "grad_norm": 0.2825027758244028, + "learning_rate": 1.4724300857789387e-05, + "loss": 1.1033, + "step": 800 + }, + { + "epoch": 2.0485933503836318, + "grad_norm": 0.2499570786028914, + "learning_rate": 1.4708552002685634e-05, + "loss": 1.0545, + "step": 801 + }, + { + "epoch": 2.051150895140665, + "grad_norm": 0.24594618416181435, + "learning_rate": 1.4692788128087174e-05, + "loss": 1.06, + "step": 802 + }, + { + "epoch": 2.053708439897698, + "grad_norm": 0.20776683930789736, + "learning_rate": 1.4677009284278128e-05, + "loss": 1.0685, + "step": 803 + }, + { + "epoch": 2.0562659846547313, + "grad_norm": 0.24408797497104165, + "learning_rate": 1.4661215521590375e-05, + "loss": 1.0969, + "step": 804 + }, + { + "epoch": 2.0588235294117645, + "grad_norm": 0.2398442146389859, + "learning_rate": 1.4645406890403384e-05, + "loss": 1.0437, + "step": 805 + }, + { + "epoch": 2.061381074168798, + "grad_norm": 0.27459743419840466, + "learning_rate": 1.4629583441144044e-05, + "loss": 1.0571, + "step": 806 + }, + { + "epoch": 2.0639386189258313, + "grad_norm": 0.32419594005338426, + "learning_rate": 1.4613745224286523e-05, + "loss": 1.0992, + "step": 807 + }, + { + "epoch": 2.0664961636828645, + "grad_norm": 0.24865801570827994, + "learning_rate": 1.459789229035208e-05, + "loss": 1.071, + "step": 808 + }, + { + "epoch": 2.0690537084398977, + "grad_norm": 0.2805482334848806, + "learning_rate": 1.4582024689908933e-05, + "loss": 1.0472, + "step": 809 + }, + { + "epoch": 2.071611253196931, + "grad_norm": 0.27730449859404743, + "learning_rate": 1.456614247357208e-05, + "loss": 1.0509, + "step": 810 + }, + { + "epoch": 2.074168797953964, + "grad_norm": 0.21758995784556315, + "learning_rate": 1.455024569200313e-05, + "loss": 1.0754, + "step": 811 + }, + { + "epoch": 2.0767263427109977, + "grad_norm": 0.30576610905327345, + "learning_rate": 1.4534334395910173e-05, + "loss": 1.0487, + "step": 812 + }, + { + "epoch": 2.079283887468031, + "grad_norm": 0.2685773932095298, + "learning_rate": 1.4518408636047579e-05, + "loss": 1.049, + "step": 813 + }, + { + "epoch": 2.081841432225064, + "grad_norm": 0.2752647553950956, + "learning_rate": 1.4502468463215866e-05, + "loss": 1.111, + "step": 814 + }, + { + "epoch": 2.084398976982097, + "grad_norm": 0.2228850184143428, + "learning_rate": 1.4486513928261525e-05, + "loss": 1.0377, + "step": 815 + }, + { + "epoch": 2.0869565217391304, + "grad_norm": 0.2791891115513624, + "learning_rate": 1.4470545082076855e-05, + "loss": 1.0498, + "step": 816 + }, + { + "epoch": 2.0895140664961636, + "grad_norm": 0.24096614975523328, + "learning_rate": 1.445456197559981e-05, + "loss": 1.0612, + "step": 817 + }, + { + "epoch": 2.0920716112531967, + "grad_norm": 0.22086178812552099, + "learning_rate": 1.4438564659813833e-05, + "loss": 1.0406, + "step": 818 + }, + { + "epoch": 2.0946291560102304, + "grad_norm": 0.23208084745447222, + "learning_rate": 1.4422553185747693e-05, + "loss": 1.0436, + "step": 819 + }, + { + "epoch": 2.0971867007672635, + "grad_norm": 0.25938781713310394, + "learning_rate": 1.4406527604475309e-05, + "loss": 1.0457, + "step": 820 + }, + { + "epoch": 2.0997442455242967, + "grad_norm": 0.2298859593490135, + "learning_rate": 1.4390487967115619e-05, + "loss": 1.0576, + "step": 821 + }, + { + "epoch": 2.10230179028133, + "grad_norm": 0.2440313535649294, + "learning_rate": 1.4374434324832386e-05, + "loss": 1.0421, + "step": 822 + }, + { + "epoch": 2.104859335038363, + "grad_norm": 0.21648476955247134, + "learning_rate": 1.4358366728834043e-05, + "loss": 1.0737, + "step": 823 + }, + { + "epoch": 2.1074168797953963, + "grad_norm": 0.27334150818408187, + "learning_rate": 1.434228523037355e-05, + "loss": 1.067, + "step": 824 + }, + { + "epoch": 2.10997442455243, + "grad_norm": 0.2339298799419931, + "learning_rate": 1.4326189880748187e-05, + "loss": 1.067, + "step": 825 + }, + { + "epoch": 2.112531969309463, + "grad_norm": 0.2439620002391159, + "learning_rate": 1.4310080731299443e-05, + "loss": 1.0641, + "step": 826 + }, + { + "epoch": 2.1150895140664963, + "grad_norm": 0.19808496444753365, + "learning_rate": 1.4293957833412813e-05, + "loss": 1.0547, + "step": 827 + }, + { + "epoch": 2.1176470588235294, + "grad_norm": 0.23889835747137905, + "learning_rate": 1.4277821238517644e-05, + "loss": 1.066, + "step": 828 + }, + { + "epoch": 2.1202046035805626, + "grad_norm": 0.23925379076145686, + "learning_rate": 1.4261670998086988e-05, + "loss": 1.0691, + "step": 829 + }, + { + "epoch": 2.122762148337596, + "grad_norm": 0.22622163591866712, + "learning_rate": 1.4245507163637406e-05, + "loss": 1.0967, + "step": 830 + }, + { + "epoch": 2.125319693094629, + "grad_norm": 0.24493596074293522, + "learning_rate": 1.4229329786728839e-05, + "loss": 1.0879, + "step": 831 + }, + { + "epoch": 2.1278772378516626, + "grad_norm": 0.2952728967619172, + "learning_rate": 1.4213138918964415e-05, + "loss": 1.0847, + "step": 832 + }, + { + "epoch": 2.130434782608696, + "grad_norm": 0.2422929005577866, + "learning_rate": 1.4196934611990296e-05, + "loss": 1.0836, + "step": 833 + }, + { + "epoch": 2.132992327365729, + "grad_norm": 0.21959782576337455, + "learning_rate": 1.4180716917495521e-05, + "loss": 1.0431, + "step": 834 + }, + { + "epoch": 2.135549872122762, + "grad_norm": 0.24719640294214473, + "learning_rate": 1.4164485887211825e-05, + "loss": 1.0748, + "step": 835 + }, + { + "epoch": 2.1381074168797953, + "grad_norm": 0.2856624330721336, + "learning_rate": 1.4148241572913482e-05, + "loss": 1.0915, + "step": 836 + }, + { + "epoch": 2.1406649616368285, + "grad_norm": 0.20175687613571772, + "learning_rate": 1.4131984026417147e-05, + "loss": 1.0289, + "step": 837 + }, + { + "epoch": 2.1432225063938617, + "grad_norm": 0.24838388258268573, + "learning_rate": 1.4115713299581677e-05, + "loss": 1.0898, + "step": 838 + }, + { + "epoch": 2.1457800511508953, + "grad_norm": 0.26748186697500365, + "learning_rate": 1.4099429444307973e-05, + "loss": 1.067, + "step": 839 + }, + { + "epoch": 2.1483375959079285, + "grad_norm": 0.2586912877126989, + "learning_rate": 1.4083132512538815e-05, + "loss": 1.0905, + "step": 840 + }, + { + "epoch": 2.1508951406649617, + "grad_norm": 0.24121340420048135, + "learning_rate": 1.4066822556258694e-05, + "loss": 1.0666, + "step": 841 + }, + { + "epoch": 2.153452685421995, + "grad_norm": 0.2639059153532029, + "learning_rate": 1.4050499627493648e-05, + "loss": 1.0798, + "step": 842 + }, + { + "epoch": 2.156010230179028, + "grad_norm": 0.27343712262396147, + "learning_rate": 1.4034163778311096e-05, + "loss": 1.0529, + "step": 843 + }, + { + "epoch": 2.1585677749360612, + "grad_norm": 0.24142653515867196, + "learning_rate": 1.4017815060819667e-05, + "loss": 1.1025, + "step": 844 + }, + { + "epoch": 2.1611253196930944, + "grad_norm": 0.32920032155082907, + "learning_rate": 1.4001453527169035e-05, + "loss": 1.0612, + "step": 845 + }, + { + "epoch": 2.163682864450128, + "grad_norm": 0.3015833792100841, + "learning_rate": 1.3985079229549771e-05, + "loss": 1.0569, + "step": 846 + }, + { + "epoch": 2.166240409207161, + "grad_norm": 0.2522803636771771, + "learning_rate": 1.3968692220193145e-05, + "loss": 1.0174, + "step": 847 + }, + { + "epoch": 2.1687979539641944, + "grad_norm": 0.26162844295546517, + "learning_rate": 1.3952292551370978e-05, + "loss": 1.11, + "step": 848 + }, + { + "epoch": 2.1713554987212276, + "grad_norm": 0.27023676278734887, + "learning_rate": 1.3935880275395483e-05, + "loss": 1.0695, + "step": 849 + }, + { + "epoch": 2.1739130434782608, + "grad_norm": 0.23810673971999766, + "learning_rate": 1.3919455444619074e-05, + "loss": 1.07, + "step": 850 + }, + { + "epoch": 2.176470588235294, + "grad_norm": 0.2371489539651306, + "learning_rate": 1.390301811143422e-05, + "loss": 1.0748, + "step": 851 + }, + { + "epoch": 2.1790281329923276, + "grad_norm": 0.24704910714072092, + "learning_rate": 1.388656832827327e-05, + "loss": 1.0886, + "step": 852 + }, + { + "epoch": 2.1815856777493607, + "grad_norm": 0.25821013482512084, + "learning_rate": 1.3870106147608282e-05, + "loss": 1.026, + "step": 853 + }, + { + "epoch": 2.184143222506394, + "grad_norm": 0.2595892420977119, + "learning_rate": 1.385363162195087e-05, + "loss": 1.0095, + "step": 854 + }, + { + "epoch": 2.186700767263427, + "grad_norm": 0.2751769970647654, + "learning_rate": 1.3837144803852016e-05, + "loss": 1.0625, + "step": 855 + }, + { + "epoch": 2.1892583120204603, + "grad_norm": 0.2611393387522944, + "learning_rate": 1.3820645745901917e-05, + "loss": 1.0819, + "step": 856 + }, + { + "epoch": 2.1918158567774935, + "grad_norm": 0.2329077525103117, + "learning_rate": 1.3804134500729817e-05, + "loss": 1.0987, + "step": 857 + }, + { + "epoch": 2.1943734015345266, + "grad_norm": 0.2443911678779139, + "learning_rate": 1.3787611121003825e-05, + "loss": 1.091, + "step": 858 + }, + { + "epoch": 2.1969309462915603, + "grad_norm": 0.23687678853190128, + "learning_rate": 1.377107565943077e-05, + "loss": 1.0872, + "step": 859 + }, + { + "epoch": 2.1994884910485935, + "grad_norm": 0.2576015050852838, + "learning_rate": 1.3754528168756007e-05, + "loss": 1.0834, + "step": 860 + }, + { + "epoch": 2.2020460358056266, + "grad_norm": 0.2599012056741005, + "learning_rate": 1.3737968701763275e-05, + "loss": 1.0766, + "step": 861 + }, + { + "epoch": 2.20460358056266, + "grad_norm": 0.209435005291082, + "learning_rate": 1.3721397311274505e-05, + "loss": 1.0488, + "step": 862 + }, + { + "epoch": 2.207161125319693, + "grad_norm": 0.25527194953058974, + "learning_rate": 1.3704814050149664e-05, + "loss": 1.0714, + "step": 863 + }, + { + "epoch": 2.209718670076726, + "grad_norm": 0.2353445658736852, + "learning_rate": 1.3688218971286588e-05, + "loss": 1.0745, + "step": 864 + }, + { + "epoch": 2.21227621483376, + "grad_norm": 0.26020939155181216, + "learning_rate": 1.367161212762081e-05, + "loss": 1.0298, + "step": 865 + }, + { + "epoch": 2.214833759590793, + "grad_norm": 0.23209974209064985, + "learning_rate": 1.3654993572125385e-05, + "loss": 1.0609, + "step": 866 + }, + { + "epoch": 2.217391304347826, + "grad_norm": 0.2953990139308075, + "learning_rate": 1.3638363357810734e-05, + "loss": 1.1041, + "step": 867 + }, + { + "epoch": 2.2199488491048593, + "grad_norm": 0.30464352233125175, + "learning_rate": 1.362172153772446e-05, + "loss": 1.0437, + "step": 868 + }, + { + "epoch": 2.2225063938618925, + "grad_norm": 0.23970358266125305, + "learning_rate": 1.3605068164951195e-05, + "loss": 1.0818, + "step": 869 + }, + { + "epoch": 2.2250639386189257, + "grad_norm": 0.2974010740237724, + "learning_rate": 1.3588403292612409e-05, + "loss": 1.0667, + "step": 870 + }, + { + "epoch": 2.227621483375959, + "grad_norm": 0.2986189698714367, + "learning_rate": 1.3571726973866275e-05, + "loss": 1.0468, + "step": 871 + }, + { + "epoch": 2.2301790281329925, + "grad_norm": 0.22393515546922488, + "learning_rate": 1.3555039261907452e-05, + "loss": 1.0681, + "step": 872 + }, + { + "epoch": 2.2327365728900257, + "grad_norm": 0.26982155718376244, + "learning_rate": 1.3538340209966965e-05, + "loss": 1.0846, + "step": 873 + }, + { + "epoch": 2.235294117647059, + "grad_norm": 0.2421804429950771, + "learning_rate": 1.3521629871311997e-05, + "loss": 1.0731, + "step": 874 + }, + { + "epoch": 2.237851662404092, + "grad_norm": 0.21568056876105401, + "learning_rate": 1.3504908299245738e-05, + "loss": 1.0393, + "step": 875 + }, + { + "epoch": 2.2404092071611252, + "grad_norm": 0.2820400540340047, + "learning_rate": 1.3488175547107214e-05, + "loss": 1.0592, + "step": 876 + }, + { + "epoch": 2.2429667519181584, + "grad_norm": 0.2415975374420239, + "learning_rate": 1.3471431668271105e-05, + "loss": 1.0488, + "step": 877 + }, + { + "epoch": 2.2455242966751916, + "grad_norm": 0.2066593963187682, + "learning_rate": 1.3454676716147595e-05, + "loss": 1.0557, + "step": 878 + }, + { + "epoch": 2.2480818414322252, + "grad_norm": 0.2835186467703053, + "learning_rate": 1.343791074418218e-05, + "loss": 1.0557, + "step": 879 + }, + { + "epoch": 2.2506393861892584, + "grad_norm": 0.2038398858990616, + "learning_rate": 1.342113380585551e-05, + "loss": 1.0305, + "step": 880 + }, + { + "epoch": 2.2531969309462916, + "grad_norm": 0.2467659045058108, + "learning_rate": 1.3404345954683223e-05, + "loss": 1.0502, + "step": 881 + }, + { + "epoch": 2.2557544757033248, + "grad_norm": 0.23271887429473748, + "learning_rate": 1.3387547244215755e-05, + "loss": 1.017, + "step": 882 + }, + { + "epoch": 2.258312020460358, + "grad_norm": 0.23672669602819807, + "learning_rate": 1.337073772803819e-05, + "loss": 1.0577, + "step": 883 + }, + { + "epoch": 2.260869565217391, + "grad_norm": 0.20836338901899323, + "learning_rate": 1.3353917459770078e-05, + "loss": 1.0523, + "step": 884 + }, + { + "epoch": 2.2634271099744243, + "grad_norm": 0.2689164014770486, + "learning_rate": 1.3337086493065266e-05, + "loss": 1.0727, + "step": 885 + }, + { + "epoch": 2.265984654731458, + "grad_norm": 0.20010892395917804, + "learning_rate": 1.3320244881611728e-05, + "loss": 1.0801, + "step": 886 + }, + { + "epoch": 2.268542199488491, + "grad_norm": 0.21502634804444792, + "learning_rate": 1.3303392679131393e-05, + "loss": 1.0342, + "step": 887 + }, + { + "epoch": 2.2710997442455243, + "grad_norm": 0.22292944251420502, + "learning_rate": 1.3286529939379969e-05, + "loss": 1.0403, + "step": 888 + }, + { + "epoch": 2.2736572890025575, + "grad_norm": 0.1889567658964355, + "learning_rate": 1.3269656716146786e-05, + "loss": 1.0657, + "step": 889 + }, + { + "epoch": 2.2762148337595907, + "grad_norm": 0.22240068914596578, + "learning_rate": 1.3252773063254609e-05, + "loss": 1.0996, + "step": 890 + }, + { + "epoch": 2.2787723785166243, + "grad_norm": 0.20078276182072644, + "learning_rate": 1.3235879034559467e-05, + "loss": 1.0756, + "step": 891 + }, + { + "epoch": 2.2813299232736575, + "grad_norm": 0.20342234409609955, + "learning_rate": 1.321897468395049e-05, + "loss": 1.0592, + "step": 892 + }, + { + "epoch": 2.2838874680306906, + "grad_norm": 0.20922107317317695, + "learning_rate": 1.320206006534974e-05, + "loss": 1.0656, + "step": 893 + }, + { + "epoch": 2.286445012787724, + "grad_norm": 0.19491180577509853, + "learning_rate": 1.3185135232712024e-05, + "loss": 1.0862, + "step": 894 + }, + { + "epoch": 2.289002557544757, + "grad_norm": 0.21756827855179173, + "learning_rate": 1.316820024002473e-05, + "loss": 1.0757, + "step": 895 + }, + { + "epoch": 2.29156010230179, + "grad_norm": 0.19452656906284796, + "learning_rate": 1.3151255141307658e-05, + "loss": 1.0618, + "step": 896 + }, + { + "epoch": 2.2941176470588234, + "grad_norm": 0.23416160900137958, + "learning_rate": 1.3134299990612843e-05, + "loss": 1.0607, + "step": 897 + }, + { + "epoch": 2.296675191815857, + "grad_norm": 0.20712138927673676, + "learning_rate": 1.3117334842024386e-05, + "loss": 1.036, + "step": 898 + }, + { + "epoch": 2.29923273657289, + "grad_norm": 0.21330832703203775, + "learning_rate": 1.310035974965828e-05, + "loss": 1.0722, + "step": 899 + }, + { + "epoch": 2.3017902813299234, + "grad_norm": 0.21219344261681405, + "learning_rate": 1.308337476766223e-05, + "loss": 1.0559, + "step": 900 + }, + { + "epoch": 2.3043478260869565, + "grad_norm": 0.2258939301473968, + "learning_rate": 1.3066379950215499e-05, + "loss": 1.0473, + "step": 901 + }, + { + "epoch": 2.3069053708439897, + "grad_norm": 0.19516875187667224, + "learning_rate": 1.3049375351528713e-05, + "loss": 1.0525, + "step": 902 + }, + { + "epoch": 2.309462915601023, + "grad_norm": 0.21614015553918936, + "learning_rate": 1.3032361025843704e-05, + "loss": 1.0763, + "step": 903 + }, + { + "epoch": 2.312020460358056, + "grad_norm": 0.20862663707679718, + "learning_rate": 1.301533702743333e-05, + "loss": 0.9991, + "step": 904 + }, + { + "epoch": 2.3145780051150897, + "grad_norm": 0.2379949878093605, + "learning_rate": 1.29983034106013e-05, + "loss": 1.0369, + "step": 905 + }, + { + "epoch": 2.317135549872123, + "grad_norm": 0.2447808238713185, + "learning_rate": 1.2981260229682018e-05, + "loss": 1.0461, + "step": 906 + }, + { + "epoch": 2.319693094629156, + "grad_norm": 0.27684254724066776, + "learning_rate": 1.2964207539040373e-05, + "loss": 1.0505, + "step": 907 + }, + { + "epoch": 2.3222506393861893, + "grad_norm": 0.2078769477308904, + "learning_rate": 1.2947145393071607e-05, + "loss": 1.0912, + "step": 908 + }, + { + "epoch": 2.3248081841432224, + "grad_norm": 0.2108719484786227, + "learning_rate": 1.2930073846201116e-05, + "loss": 1.0261, + "step": 909 + }, + { + "epoch": 2.3273657289002556, + "grad_norm": 0.21718674615633884, + "learning_rate": 1.2912992952884283e-05, + "loss": 1.1043, + "step": 910 + }, + { + "epoch": 2.329923273657289, + "grad_norm": 0.22685276306972108, + "learning_rate": 1.289590276760631e-05, + "loss": 1.0806, + "step": 911 + }, + { + "epoch": 2.3324808184143224, + "grad_norm": 0.23081228405085022, + "learning_rate": 1.2878803344882029e-05, + "loss": 1.0392, + "step": 912 + }, + { + "epoch": 2.3350383631713556, + "grad_norm": 0.2011412522246678, + "learning_rate": 1.2861694739255746e-05, + "loss": 1.0713, + "step": 913 + }, + { + "epoch": 2.337595907928389, + "grad_norm": 0.23619794732179872, + "learning_rate": 1.2844577005301055e-05, + "loss": 1.0704, + "step": 914 + }, + { + "epoch": 2.340153452685422, + "grad_norm": 0.2698850856720426, + "learning_rate": 1.2827450197620673e-05, + "loss": 1.0685, + "step": 915 + }, + { + "epoch": 2.342710997442455, + "grad_norm": 0.19779154874996993, + "learning_rate": 1.2810314370846252e-05, + "loss": 1.0895, + "step": 916 + }, + { + "epoch": 2.3452685421994883, + "grad_norm": 0.2579605967239507, + "learning_rate": 1.2793169579638222e-05, + "loss": 1.0299, + "step": 917 + }, + { + "epoch": 2.3478260869565215, + "grad_norm": 0.22140608359673922, + "learning_rate": 1.2776015878685603e-05, + "loss": 1.0582, + "step": 918 + }, + { + "epoch": 2.350383631713555, + "grad_norm": 0.21371909171227776, + "learning_rate": 1.2758853322705835e-05, + "loss": 1.0451, + "step": 919 + }, + { + "epoch": 2.3529411764705883, + "grad_norm": 0.23474961790937948, + "learning_rate": 1.2741681966444608e-05, + "loss": 1.045, + "step": 920 + }, + { + "epoch": 2.3554987212276215, + "grad_norm": 0.19870860509615906, + "learning_rate": 1.2724501864675682e-05, + "loss": 1.0487, + "step": 921 + }, + { + "epoch": 2.3580562659846547, + "grad_norm": 0.20253838337881694, + "learning_rate": 1.270731307220071e-05, + "loss": 1.0481, + "step": 922 + }, + { + "epoch": 2.360613810741688, + "grad_norm": 0.23011797101382442, + "learning_rate": 1.269011564384908e-05, + "loss": 1.0515, + "step": 923 + }, + { + "epoch": 2.363171355498721, + "grad_norm": 0.22374354187112894, + "learning_rate": 1.2672909634477701e-05, + "loss": 1.0593, + "step": 924 + }, + { + "epoch": 2.3657289002557547, + "grad_norm": 0.2500406678934162, + "learning_rate": 1.2655695098970881e-05, + "loss": 1.0468, + "step": 925 + }, + { + "epoch": 2.368286445012788, + "grad_norm": 0.21282680432072176, + "learning_rate": 1.2638472092240112e-05, + "loss": 1.0224, + "step": 926 + }, + { + "epoch": 2.370843989769821, + "grad_norm": 0.2631854118749461, + "learning_rate": 1.2621240669223906e-05, + "loss": 1.0799, + "step": 927 + }, + { + "epoch": 2.373401534526854, + "grad_norm": 0.21209069180980092, + "learning_rate": 1.2604000884887636e-05, + "loss": 1.0855, + "step": 928 + }, + { + "epoch": 2.3759590792838874, + "grad_norm": 0.2433054732140672, + "learning_rate": 1.2586752794223321e-05, + "loss": 1.0541, + "step": 929 + }, + { + "epoch": 2.3785166240409206, + "grad_norm": 0.2630201541116053, + "learning_rate": 1.25694964522495e-05, + "loss": 1.0786, + "step": 930 + }, + { + "epoch": 2.381074168797954, + "grad_norm": 0.19419910941081836, + "learning_rate": 1.2552231914011017e-05, + "loss": 1.0452, + "step": 931 + }, + { + "epoch": 2.3836317135549874, + "grad_norm": 0.2699915357205121, + "learning_rate": 1.2534959234578872e-05, + "loss": 1.071, + "step": 932 + }, + { + "epoch": 2.3861892583120206, + "grad_norm": 0.29724325552356046, + "learning_rate": 1.2517678469050023e-05, + "loss": 1.053, + "step": 933 + }, + { + "epoch": 2.3887468030690537, + "grad_norm": 0.22602756795970175, + "learning_rate": 1.2500389672547233e-05, + "loss": 1.0411, + "step": 934 + }, + { + "epoch": 2.391304347826087, + "grad_norm": 0.27261078615742584, + "learning_rate": 1.2483092900218872e-05, + "loss": 1.0397, + "step": 935 + }, + { + "epoch": 2.39386189258312, + "grad_norm": 0.33769171299450684, + "learning_rate": 1.2465788207238754e-05, + "loss": 1.0579, + "step": 936 + }, + { + "epoch": 2.3964194373401533, + "grad_norm": 0.24562682326063173, + "learning_rate": 1.2448475648805965e-05, + "loss": 1.06, + "step": 937 + }, + { + "epoch": 2.398976982097187, + "grad_norm": 0.31976622618763517, + "learning_rate": 1.243115528014467e-05, + "loss": 1.0693, + "step": 938 + }, + { + "epoch": 2.40153452685422, + "grad_norm": 0.27193293795403, + "learning_rate": 1.2413827156503959e-05, + "loss": 1.0517, + "step": 939 + }, + { + "epoch": 2.4040920716112533, + "grad_norm": 0.2800053392929295, + "learning_rate": 1.2396491333157653e-05, + "loss": 1.0618, + "step": 940 + }, + { + "epoch": 2.4066496163682864, + "grad_norm": 0.3288127671391703, + "learning_rate": 1.2379147865404126e-05, + "loss": 1.0503, + "step": 941 + }, + { + "epoch": 2.4092071611253196, + "grad_norm": 0.21261108528502098, + "learning_rate": 1.2361796808566154e-05, + "loss": 1.0608, + "step": 942 + }, + { + "epoch": 2.411764705882353, + "grad_norm": 0.2940216342886679, + "learning_rate": 1.2344438217990706e-05, + "loss": 1.0632, + "step": 943 + }, + { + "epoch": 2.414322250639386, + "grad_norm": 0.2887698618246557, + "learning_rate": 1.2327072149048787e-05, + "loss": 1.0639, + "step": 944 + }, + { + "epoch": 2.4168797953964196, + "grad_norm": 0.24521223833198755, + "learning_rate": 1.2309698657135263e-05, + "loss": 1.0162, + "step": 945 + }, + { + "epoch": 2.419437340153453, + "grad_norm": 0.185509429971427, + "learning_rate": 1.2292317797668664e-05, + "loss": 1.0363, + "step": 946 + }, + { + "epoch": 2.421994884910486, + "grad_norm": 0.25872152655717207, + "learning_rate": 1.2274929626091035e-05, + "loss": 1.0881, + "step": 947 + }, + { + "epoch": 2.424552429667519, + "grad_norm": 0.18008549030963775, + "learning_rate": 1.2257534197867744e-05, + "loss": 1.0446, + "step": 948 + }, + { + "epoch": 2.4271099744245523, + "grad_norm": 0.23690086440423483, + "learning_rate": 1.2240131568487292e-05, + "loss": 1.0616, + "step": 949 + }, + { + "epoch": 2.4296675191815855, + "grad_norm": 0.1983151726827165, + "learning_rate": 1.222272179346117e-05, + "loss": 1.0419, + "step": 950 + }, + { + "epoch": 2.4322250639386187, + "grad_norm": 0.18747005388252086, + "learning_rate": 1.2205304928323649e-05, + "loss": 1.0595, + "step": 951 + }, + { + "epoch": 2.4347826086956523, + "grad_norm": 0.20660938308532778, + "learning_rate": 1.2187881028631621e-05, + "loss": 1.022, + "step": 952 + }, + { + "epoch": 2.4373401534526855, + "grad_norm": 0.20200728083626612, + "learning_rate": 1.2170450149964422e-05, + "loss": 1.0756, + "step": 953 + }, + { + "epoch": 2.4398976982097187, + "grad_norm": 0.21912464070245444, + "learning_rate": 1.2153012347923635e-05, + "loss": 1.0771, + "step": 954 + }, + { + "epoch": 2.442455242966752, + "grad_norm": 0.23866041513967645, + "learning_rate": 1.2135567678132942e-05, + "loss": 1.0523, + "step": 955 + }, + { + "epoch": 2.445012787723785, + "grad_norm": 0.21521617514557267, + "learning_rate": 1.211811619623793e-05, + "loss": 1.059, + "step": 956 + }, + { + "epoch": 2.4475703324808182, + "grad_norm": 0.24265391190201183, + "learning_rate": 1.2100657957905908e-05, + "loss": 1.0307, + "step": 957 + }, + { + "epoch": 2.4501278772378514, + "grad_norm": 0.2132548988703794, + "learning_rate": 1.2083193018825745e-05, + "loss": 1.025, + "step": 958 + }, + { + "epoch": 2.452685421994885, + "grad_norm": 0.22131174672899057, + "learning_rate": 1.2065721434707677e-05, + "loss": 1.0474, + "step": 959 + }, + { + "epoch": 2.455242966751918, + "grad_norm": 0.19904663813024445, + "learning_rate": 1.2048243261283144e-05, + "loss": 1.0226, + "step": 960 + }, + { + "epoch": 2.4578005115089514, + "grad_norm": 0.200161353355971, + "learning_rate": 1.2030758554304593e-05, + "loss": 1.0924, + "step": 961 + }, + { + "epoch": 2.4603580562659846, + "grad_norm": 0.19830084311557072, + "learning_rate": 1.2013267369545329e-05, + "loss": 1.0667, + "step": 962 + }, + { + "epoch": 2.4629156010230178, + "grad_norm": 0.22835260283972367, + "learning_rate": 1.1995769762799307e-05, + "loss": 1.0754, + "step": 963 + }, + { + "epoch": 2.4654731457800514, + "grad_norm": 0.210015027120209, + "learning_rate": 1.1978265789880974e-05, + "loss": 1.0808, + "step": 964 + }, + { + "epoch": 2.4680306905370846, + "grad_norm": 0.2697432497468849, + "learning_rate": 1.1960755506625077e-05, + "loss": 1.0307, + "step": 965 + }, + { + "epoch": 2.4705882352941178, + "grad_norm": 0.23179650613650557, + "learning_rate": 1.1943238968886492e-05, + "loss": 1.046, + "step": 966 + }, + { + "epoch": 2.473145780051151, + "grad_norm": 0.24001578621119793, + "learning_rate": 1.1925716232540062e-05, + "loss": 1.0571, + "step": 967 + }, + { + "epoch": 2.475703324808184, + "grad_norm": 0.23255368408434818, + "learning_rate": 1.1908187353480381e-05, + "loss": 1.0621, + "step": 968 + }, + { + "epoch": 2.4782608695652173, + "grad_norm": 0.48924197246918594, + "learning_rate": 1.1890652387621645e-05, + "loss": 1.0603, + "step": 969 + }, + { + "epoch": 2.4808184143222505, + "grad_norm": 0.24215100765578446, + "learning_rate": 1.1873111390897475e-05, + "loss": 1.0301, + "step": 970 + }, + { + "epoch": 2.483375959079284, + "grad_norm": 0.23579580906631162, + "learning_rate": 1.1855564419260714e-05, + "loss": 1.0531, + "step": 971 + }, + { + "epoch": 2.4859335038363173, + "grad_norm": 0.21475267684489457, + "learning_rate": 1.1838011528683279e-05, + "loss": 1.074, + "step": 972 + }, + { + "epoch": 2.4884910485933505, + "grad_norm": 0.2584450023059667, + "learning_rate": 1.1820452775155957e-05, + "loss": 1.0536, + "step": 973 + }, + { + "epoch": 2.4910485933503836, + "grad_norm": 0.1843171436312304, + "learning_rate": 1.1802888214688235e-05, + "loss": 1.0526, + "step": 974 + }, + { + "epoch": 2.493606138107417, + "grad_norm": 0.25846478231848835, + "learning_rate": 1.1785317903308138e-05, + "loss": 1.0319, + "step": 975 + }, + { + "epoch": 2.49616368286445, + "grad_norm": 0.20971255177789871, + "learning_rate": 1.1767741897062018e-05, + "loss": 1.0613, + "step": 976 + }, + { + "epoch": 2.498721227621483, + "grad_norm": 0.20077211970723166, + "learning_rate": 1.1750160252014401e-05, + "loss": 1.0498, + "step": 977 + }, + { + "epoch": 2.501278772378517, + "grad_norm": 0.22501567037645762, + "learning_rate": 1.1732573024247803e-05, + "loss": 1.0659, + "step": 978 + }, + { + "epoch": 2.50383631713555, + "grad_norm": 0.21521993533238856, + "learning_rate": 1.171498026986254e-05, + "loss": 1.0702, + "step": 979 + }, + { + "epoch": 2.506393861892583, + "grad_norm": 0.2503577565530584, + "learning_rate": 1.1697382044976563e-05, + "loss": 1.0288, + "step": 980 + }, + { + "epoch": 2.5089514066496164, + "grad_norm": 0.2047302770203722, + "learning_rate": 1.1679778405725274e-05, + "loss": 1.0376, + "step": 981 + }, + { + "epoch": 2.5115089514066495, + "grad_norm": 0.23939263890617613, + "learning_rate": 1.166216940826134e-05, + "loss": 1.0187, + "step": 982 + }, + { + "epoch": 2.5140664961636827, + "grad_norm": 0.23761678919939125, + "learning_rate": 1.1644555108754518e-05, + "loss": 1.0308, + "step": 983 + }, + { + "epoch": 2.516624040920716, + "grad_norm": 0.21248204414574057, + "learning_rate": 1.1626935563391491e-05, + "loss": 1.0618, + "step": 984 + }, + { + "epoch": 2.5191815856777495, + "grad_norm": 0.24094115122902504, + "learning_rate": 1.1609310828375662e-05, + "loss": 1.0594, + "step": 985 + }, + { + "epoch": 2.5217391304347827, + "grad_norm": 0.19223099803507016, + "learning_rate": 1.1591680959926994e-05, + "loss": 1.0322, + "step": 986 + }, + { + "epoch": 2.524296675191816, + "grad_norm": 0.20707448434011191, + "learning_rate": 1.1574046014281823e-05, + "loss": 1.044, + "step": 987 + }, + { + "epoch": 2.526854219948849, + "grad_norm": 0.24109072712135063, + "learning_rate": 1.1556406047692681e-05, + "loss": 1.0695, + "step": 988 + }, + { + "epoch": 2.5294117647058822, + "grad_norm": 0.1979569377790155, + "learning_rate": 1.153876111642812e-05, + "loss": 1.0685, + "step": 989 + }, + { + "epoch": 2.531969309462916, + "grad_norm": 0.2936427182568951, + "learning_rate": 1.1521111276772519e-05, + "loss": 1.0778, + "step": 990 + }, + { + "epoch": 2.5345268542199486, + "grad_norm": 0.2547622419807327, + "learning_rate": 1.1503456585025918e-05, + "loss": 1.0573, + "step": 991 + }, + { + "epoch": 2.5370843989769822, + "grad_norm": 0.20636327167754573, + "learning_rate": 1.1485797097503848e-05, + "loss": 1.0753, + "step": 992 + }, + { + "epoch": 2.5396419437340154, + "grad_norm": 0.23885851857779103, + "learning_rate": 1.1468132870537112e-05, + "loss": 1.0565, + "step": 993 + }, + { + "epoch": 2.5421994884910486, + "grad_norm": 0.23233207950400891, + "learning_rate": 1.1450463960471652e-05, + "loss": 1.0388, + "step": 994 + }, + { + "epoch": 2.544757033248082, + "grad_norm": 0.19749031677810552, + "learning_rate": 1.143279042366834e-05, + "loss": 1.0315, + "step": 995 + }, + { + "epoch": 2.547314578005115, + "grad_norm": 0.21579547394805096, + "learning_rate": 1.1415112316502803e-05, + "loss": 1.0494, + "step": 996 + }, + { + "epoch": 2.5498721227621486, + "grad_norm": 0.20484578816866386, + "learning_rate": 1.139742969536526e-05, + "loss": 1.0528, + "step": 997 + }, + { + "epoch": 2.5524296675191813, + "grad_norm": 0.21000639971172463, + "learning_rate": 1.137974261666031e-05, + "loss": 1.0512, + "step": 998 + }, + { + "epoch": 2.554987212276215, + "grad_norm": 0.21807550593498504, + "learning_rate": 1.1362051136806789e-05, + "loss": 1.0457, + "step": 999 + }, + { + "epoch": 2.557544757033248, + "grad_norm": 0.21057183954898145, + "learning_rate": 1.134435531223756e-05, + "loss": 1.0489, + "step": 1000 + }, + { + "epoch": 2.5601023017902813, + "grad_norm": 0.2407874905737883, + "learning_rate": 1.1326655199399345e-05, + "loss": 1.0505, + "step": 1001 + }, + { + "epoch": 2.5626598465473145, + "grad_norm": 0.18794902645115422, + "learning_rate": 1.1308950854752558e-05, + "loss": 1.0412, + "step": 1002 + }, + { + "epoch": 2.5652173913043477, + "grad_norm": 0.22621031822986007, + "learning_rate": 1.1291242334771095e-05, + "loss": 1.069, + "step": 1003 + }, + { + "epoch": 2.5677749360613813, + "grad_norm": 0.23324199109224647, + "learning_rate": 1.1273529695942182e-05, + "loss": 1.0759, + "step": 1004 + }, + { + "epoch": 2.5703324808184145, + "grad_norm": 0.18313611243360517, + "learning_rate": 1.1255812994766175e-05, + "loss": 1.0684, + "step": 1005 + }, + { + "epoch": 2.5728900255754477, + "grad_norm": 0.2573525956981243, + "learning_rate": 1.1238092287756398e-05, + "loss": 1.0048, + "step": 1006 + }, + { + "epoch": 2.575447570332481, + "grad_norm": 0.20054767287206388, + "learning_rate": 1.1220367631438942e-05, + "loss": 1.0829, + "step": 1007 + }, + { + "epoch": 2.578005115089514, + "grad_norm": 0.24044976517280556, + "learning_rate": 1.1202639082352507e-05, + "loss": 1.0709, + "step": 1008 + }, + { + "epoch": 2.580562659846547, + "grad_norm": 0.2395771076173231, + "learning_rate": 1.1184906697048202e-05, + "loss": 1.0693, + "step": 1009 + }, + { + "epoch": 2.5831202046035804, + "grad_norm": 0.20887269846123238, + "learning_rate": 1.1167170532089369e-05, + "loss": 1.0562, + "step": 1010 + }, + { + "epoch": 2.585677749360614, + "grad_norm": 0.19484085301448842, + "learning_rate": 1.1149430644051424e-05, + "loss": 1.0568, + "step": 1011 + }, + { + "epoch": 2.588235294117647, + "grad_norm": 0.21120065497048005, + "learning_rate": 1.1131687089521642e-05, + "loss": 1.0614, + "step": 1012 + }, + { + "epoch": 2.5907928388746804, + "grad_norm": 0.1815853723973056, + "learning_rate": 1.1113939925098997e-05, + "loss": 1.0446, + "step": 1013 + }, + { + "epoch": 2.5933503836317136, + "grad_norm": 0.2128237313274396, + "learning_rate": 1.1096189207393988e-05, + "loss": 1.062, + "step": 1014 + }, + { + "epoch": 2.5959079283887467, + "grad_norm": 0.18118399034984362, + "learning_rate": 1.107843499302843e-05, + "loss": 1.0439, + "step": 1015 + }, + { + "epoch": 2.59846547314578, + "grad_norm": 0.21014691974066443, + "learning_rate": 1.106067733863531e-05, + "loss": 1.044, + "step": 1016 + }, + { + "epoch": 2.601023017902813, + "grad_norm": 0.2143448693572357, + "learning_rate": 1.1042916300858584e-05, + "loss": 1.0535, + "step": 1017 + }, + { + "epoch": 2.6035805626598467, + "grad_norm": 0.22800193329799043, + "learning_rate": 1.1025151936352987e-05, + "loss": 1.0573, + "step": 1018 + }, + { + "epoch": 2.60613810741688, + "grad_norm": 0.16936959510436445, + "learning_rate": 1.1007384301783884e-05, + "loss": 1.0287, + "step": 1019 + }, + { + "epoch": 2.608695652173913, + "grad_norm": 0.23646633318638022, + "learning_rate": 1.0989613453827058e-05, + "loss": 1.0376, + "step": 1020 + }, + { + "epoch": 2.6112531969309463, + "grad_norm": 0.20474382321861417, + "learning_rate": 1.0971839449168543e-05, + "loss": 1.05, + "step": 1021 + }, + { + "epoch": 2.6138107416879794, + "grad_norm": 0.25295508871584677, + "learning_rate": 1.095406234450446e-05, + "loss": 1.0431, + "step": 1022 + }, + { + "epoch": 2.6163682864450126, + "grad_norm": 0.20284701204586733, + "learning_rate": 1.0936282196540788e-05, + "loss": 1.0668, + "step": 1023 + }, + { + "epoch": 2.618925831202046, + "grad_norm": 0.23447500011585404, + "learning_rate": 1.0918499061993241e-05, + "loss": 1.0454, + "step": 1024 + }, + { + "epoch": 2.6214833759590794, + "grad_norm": 0.218326014757165, + "learning_rate": 1.0900712997587048e-05, + "loss": 1.0405, + "step": 1025 + }, + { + "epoch": 2.6240409207161126, + "grad_norm": 0.19544037134244654, + "learning_rate": 1.0882924060056782e-05, + "loss": 1.0487, + "step": 1026 + }, + { + "epoch": 2.626598465473146, + "grad_norm": 0.24715529739963063, + "learning_rate": 1.0865132306146182e-05, + "loss": 1.0561, + "step": 1027 + }, + { + "epoch": 2.629156010230179, + "grad_norm": 0.19894449110120455, + "learning_rate": 1.0847337792607978e-05, + "loss": 1.0662, + "step": 1028 + }, + { + "epoch": 2.631713554987212, + "grad_norm": 0.21505303389916322, + "learning_rate": 1.0829540576203695e-05, + "loss": 1.069, + "step": 1029 + }, + { + "epoch": 2.634271099744246, + "grad_norm": 0.18761978697875684, + "learning_rate": 1.0811740713703476e-05, + "loss": 1.0545, + "step": 1030 + }, + { + "epoch": 2.6368286445012785, + "grad_norm": 0.20721529389957177, + "learning_rate": 1.0793938261885917e-05, + "loss": 1.0598, + "step": 1031 + }, + { + "epoch": 2.639386189258312, + "grad_norm": 0.20950006821732456, + "learning_rate": 1.0776133277537865e-05, + "loss": 1.0253, + "step": 1032 + }, + { + "epoch": 2.6419437340153453, + "grad_norm": 0.22635292623674808, + "learning_rate": 1.075832581745425e-05, + "loss": 1.0847, + "step": 1033 + }, + { + "epoch": 2.6445012787723785, + "grad_norm": 0.19656361617686485, + "learning_rate": 1.074051593843789e-05, + "loss": 1.0618, + "step": 1034 + }, + { + "epoch": 2.6470588235294117, + "grad_norm": 0.2260452430200194, + "learning_rate": 1.0722703697299328e-05, + "loss": 1.0662, + "step": 1035 + }, + { + "epoch": 2.649616368286445, + "grad_norm": 0.19750297888085683, + "learning_rate": 1.070488915085664e-05, + "loss": 1.0387, + "step": 1036 + }, + { + "epoch": 2.6521739130434785, + "grad_norm": 0.1995618558238664, + "learning_rate": 1.0687072355935257e-05, + "loss": 1.0458, + "step": 1037 + }, + { + "epoch": 2.6547314578005117, + "grad_norm": 0.19344732111610985, + "learning_rate": 1.0669253369367776e-05, + "loss": 1.0501, + "step": 1038 + }, + { + "epoch": 2.657289002557545, + "grad_norm": 0.21983414069144497, + "learning_rate": 1.0651432247993793e-05, + "loss": 1.0585, + "step": 1039 + }, + { + "epoch": 2.659846547314578, + "grad_norm": 0.20977798376089846, + "learning_rate": 1.0633609048659706e-05, + "loss": 1.0691, + "step": 1040 + }, + { + "epoch": 2.662404092071611, + "grad_norm": 0.18193153369079076, + "learning_rate": 1.0615783828218546e-05, + "loss": 1.0482, + "step": 1041 + }, + { + "epoch": 2.6649616368286444, + "grad_norm": 0.2362607838969862, + "learning_rate": 1.0597956643529789e-05, + "loss": 1.0795, + "step": 1042 + }, + { + "epoch": 2.6675191815856776, + "grad_norm": 0.18789714832551685, + "learning_rate": 1.0580127551459178e-05, + "loss": 1.0123, + "step": 1043 + }, + { + "epoch": 2.670076726342711, + "grad_norm": 0.19255681241817454, + "learning_rate": 1.0562296608878546e-05, + "loss": 1.0545, + "step": 1044 + }, + { + "epoch": 2.6726342710997444, + "grad_norm": 0.1914112200112221, + "learning_rate": 1.0544463872665612e-05, + "loss": 1.0412, + "step": 1045 + }, + { + "epoch": 2.6751918158567776, + "grad_norm": 0.1956849800060106, + "learning_rate": 1.0526629399703834e-05, + "loss": 1.0252, + "step": 1046 + }, + { + "epoch": 2.6777493606138107, + "grad_norm": 0.2100982102976319, + "learning_rate": 1.0508793246882202e-05, + "loss": 1.0589, + "step": 1047 + }, + { + "epoch": 2.680306905370844, + "grad_norm": 0.1915658983950445, + "learning_rate": 1.0490955471095062e-05, + "loss": 1.0584, + "step": 1048 + }, + { + "epoch": 2.682864450127877, + "grad_norm": 0.23563078056985684, + "learning_rate": 1.0473116129241944e-05, + "loss": 1.0329, + "step": 1049 + }, + { + "epoch": 2.6854219948849103, + "grad_norm": 0.19242917127297468, + "learning_rate": 1.045527527822737e-05, + "loss": 1.06, + "step": 1050 + }, + { + "epoch": 2.687979539641944, + "grad_norm": 0.22869967853846762, + "learning_rate": 1.0437432974960676e-05, + "loss": 1.0564, + "step": 1051 + }, + { + "epoch": 2.690537084398977, + "grad_norm": 0.2133654830434205, + "learning_rate": 1.0419589276355826e-05, + "loss": 1.0475, + "step": 1052 + }, + { + "epoch": 2.6930946291560103, + "grad_norm": 0.2118696215206806, + "learning_rate": 1.0401744239331244e-05, + "loss": 1.0614, + "step": 1053 + }, + { + "epoch": 2.6956521739130435, + "grad_norm": 0.19733955046640414, + "learning_rate": 1.0383897920809618e-05, + "loss": 1.0241, + "step": 1054 + }, + { + "epoch": 2.6982097186700766, + "grad_norm": 0.2155680013638521, + "learning_rate": 1.0366050377717723e-05, + "loss": 1.0386, + "step": 1055 + }, + { + "epoch": 2.70076726342711, + "grad_norm": 0.21857115127975912, + "learning_rate": 1.0348201666986241e-05, + "loss": 1.0587, + "step": 1056 + }, + { + "epoch": 2.703324808184143, + "grad_norm": 0.21312799905294016, + "learning_rate": 1.0330351845549577e-05, + "loss": 1.0586, + "step": 1057 + }, + { + "epoch": 2.7058823529411766, + "grad_norm": 0.21137475767635053, + "learning_rate": 1.031250097034569e-05, + "loss": 1.0476, + "step": 1058 + }, + { + "epoch": 2.70843989769821, + "grad_norm": 0.1778309692168584, + "learning_rate": 1.029464909831588e-05, + "loss": 1.0758, + "step": 1059 + }, + { + "epoch": 2.710997442455243, + "grad_norm": 0.20578943102646913, + "learning_rate": 1.0276796286404644e-05, + "loss": 1.0346, + "step": 1060 + }, + { + "epoch": 2.713554987212276, + "grad_norm": 0.20714284355574877, + "learning_rate": 1.0258942591559475e-05, + "loss": 1.0741, + "step": 1061 + }, + { + "epoch": 2.7161125319693094, + "grad_norm": 0.18109264555313395, + "learning_rate": 1.024108807073067e-05, + "loss": 1.062, + "step": 1062 + }, + { + "epoch": 2.718670076726343, + "grad_norm": 0.19624851311668945, + "learning_rate": 1.0223232780871173e-05, + "loss": 1.0958, + "step": 1063 + }, + { + "epoch": 2.7212276214833757, + "grad_norm": 0.19392717064910245, + "learning_rate": 1.020537677893638e-05, + "loss": 1.0572, + "step": 1064 + }, + { + "epoch": 2.7237851662404093, + "grad_norm": 0.18070548523837038, + "learning_rate": 1.0187520121883952e-05, + "loss": 1.0266, + "step": 1065 + }, + { + "epoch": 2.7263427109974425, + "grad_norm": 0.1940777698141172, + "learning_rate": 1.0169662866673647e-05, + "loss": 1.0456, + "step": 1066 + }, + { + "epoch": 2.7289002557544757, + "grad_norm": 0.19806191898621253, + "learning_rate": 1.0151805070267123e-05, + "loss": 1.0525, + "step": 1067 + }, + { + "epoch": 2.731457800511509, + "grad_norm": 0.19834077877996045, + "learning_rate": 1.0133946789627773e-05, + "loss": 1.0714, + "step": 1068 + }, + { + "epoch": 2.734015345268542, + "grad_norm": 0.22704784534870529, + "learning_rate": 1.0116088081720528e-05, + "loss": 1.0476, + "step": 1069 + }, + { + "epoch": 2.7365728900255757, + "grad_norm": 0.1981045516344043, + "learning_rate": 1.0098229003511684e-05, + "loss": 1.0935, + "step": 1070 + }, + { + "epoch": 2.7391304347826084, + "grad_norm": 0.21563650872651463, + "learning_rate": 1.0080369611968724e-05, + "loss": 1.042, + "step": 1071 + }, + { + "epoch": 2.741687979539642, + "grad_norm": 0.21593748692546988, + "learning_rate": 1.0062509964060119e-05, + "loss": 1.0744, + "step": 1072 + }, + { + "epoch": 2.7442455242966752, + "grad_norm": 0.22642223122270672, + "learning_rate": 1.0044650116755166e-05, + "loss": 1.056, + "step": 1073 + }, + { + "epoch": 2.7468030690537084, + "grad_norm": 0.19475366026833285, + "learning_rate": 1.0026790127023793e-05, + "loss": 1.0406, + "step": 1074 + }, + { + "epoch": 2.7493606138107416, + "grad_norm": 0.24030640546026497, + "learning_rate": 1.0008930051836389e-05, + "loss": 1.0957, + "step": 1075 + }, + { + "epoch": 2.7519181585677748, + "grad_norm": 0.22630901508829493, + "learning_rate": 9.991069948163616e-06, + "loss": 1.0509, + "step": 1076 + }, + { + "epoch": 2.7544757033248084, + "grad_norm": 0.18042335767327197, + "learning_rate": 9.97320987297621e-06, + "loss": 1.0727, + "step": 1077 + }, + { + "epoch": 2.7570332480818416, + "grad_norm": 0.22313730186227937, + "learning_rate": 9.955349883244837e-06, + "loss": 1.0343, + "step": 1078 + }, + { + "epoch": 2.7595907928388748, + "grad_norm": 0.1797662118649205, + "learning_rate": 9.937490035939884e-06, + "loss": 1.0119, + "step": 1079 + }, + { + "epoch": 2.762148337595908, + "grad_norm": 0.19520318025917605, + "learning_rate": 9.919630388031278e-06, + "loss": 1.0598, + "step": 1080 + }, + { + "epoch": 2.764705882352941, + "grad_norm": 0.18513277316186183, + "learning_rate": 9.901770996488316e-06, + "loss": 1.0224, + "step": 1081 + }, + { + "epoch": 2.7672634271099743, + "grad_norm": 0.2448129514637694, + "learning_rate": 9.883911918279476e-06, + "loss": 1.0609, + "step": 1082 + }, + { + "epoch": 2.7698209718670075, + "grad_norm": 0.17314061842855177, + "learning_rate": 9.86605321037223e-06, + "loss": 1.0578, + "step": 1083 + }, + { + "epoch": 2.772378516624041, + "grad_norm": 0.24279893745727923, + "learning_rate": 9.84819492973288e-06, + "loss": 1.0606, + "step": 1084 + }, + { + "epoch": 2.7749360613810743, + "grad_norm": 0.20431457711634385, + "learning_rate": 9.830337133326355e-06, + "loss": 1.0351, + "step": 1085 + }, + { + "epoch": 2.7774936061381075, + "grad_norm": 0.19458070922054774, + "learning_rate": 9.81247987811605e-06, + "loss": 1.0897, + "step": 1086 + }, + { + "epoch": 2.7800511508951407, + "grad_norm": 0.2100381247728313, + "learning_rate": 9.794623221063625e-06, + "loss": 1.0146, + "step": 1087 + }, + { + "epoch": 2.782608695652174, + "grad_norm": 0.2143570049157338, + "learning_rate": 9.776767219128829e-06, + "loss": 1.042, + "step": 1088 + }, + { + "epoch": 2.785166240409207, + "grad_norm": 0.20143887728169868, + "learning_rate": 9.758911929269334e-06, + "loss": 1.0633, + "step": 1089 + }, + { + "epoch": 2.78772378516624, + "grad_norm": 0.26056402255926914, + "learning_rate": 9.741057408440527e-06, + "loss": 1.0285, + "step": 1090 + }, + { + "epoch": 2.790281329923274, + "grad_norm": 0.18543103389813437, + "learning_rate": 9.723203713595356e-06, + "loss": 1.004, + "step": 1091 + }, + { + "epoch": 2.792838874680307, + "grad_norm": 0.2601236122874053, + "learning_rate": 9.70535090168412e-06, + "loss": 1.0651, + "step": 1092 + }, + { + "epoch": 2.79539641943734, + "grad_norm": 0.19784218292302086, + "learning_rate": 9.687499029654314e-06, + "loss": 1.068, + "step": 1093 + }, + { + "epoch": 2.7979539641943734, + "grad_norm": 0.23796716838458418, + "learning_rate": 9.669648154450425e-06, + "loss": 1.0499, + "step": 1094 + }, + { + "epoch": 2.8005115089514065, + "grad_norm": 0.2120185185123368, + "learning_rate": 9.651798333013762e-06, + "loss": 1.0455, + "step": 1095 + }, + { + "epoch": 2.80306905370844, + "grad_norm": 0.2105859374042988, + "learning_rate": 9.63394962228228e-06, + "loss": 1.0956, + "step": 1096 + }, + { + "epoch": 2.805626598465473, + "grad_norm": 0.2191672276731036, + "learning_rate": 9.616102079190382e-06, + "loss": 1.062, + "step": 1097 + }, + { + "epoch": 2.8081841432225065, + "grad_norm": 0.2119528196350952, + "learning_rate": 9.59825576066876e-06, + "loss": 1.0769, + "step": 1098 + }, + { + "epoch": 2.8107416879795397, + "grad_norm": 0.21455177436347436, + "learning_rate": 9.580410723644177e-06, + "loss": 1.0665, + "step": 1099 + }, + { + "epoch": 2.813299232736573, + "grad_norm": 0.18662737672691734, + "learning_rate": 9.562567025039326e-06, + "loss": 1.0403, + "step": 1100 + }, + { + "epoch": 2.815856777493606, + "grad_norm": 0.22100028069203664, + "learning_rate": 9.544724721772632e-06, + "loss": 1.0679, + "step": 1101 + }, + { + "epoch": 2.8184143222506393, + "grad_norm": 0.1935799574613574, + "learning_rate": 9.526883870758056e-06, + "loss": 1.0499, + "step": 1102 + }, + { + "epoch": 2.820971867007673, + "grad_norm": 0.19527712864881602, + "learning_rate": 9.50904452890494e-06, + "loss": 1.0652, + "step": 1103 + }, + { + "epoch": 2.8235294117647056, + "grad_norm": 0.20509388285010824, + "learning_rate": 9.491206753117803e-06, + "loss": 1.0413, + "step": 1104 + }, + { + "epoch": 2.8260869565217392, + "grad_norm": 0.1992258084768072, + "learning_rate": 9.47337060029617e-06, + "loss": 1.0316, + "step": 1105 + }, + { + "epoch": 2.8286445012787724, + "grad_norm": 0.22625008837517527, + "learning_rate": 9.45553612733439e-06, + "loss": 1.0618, + "step": 1106 + }, + { + "epoch": 2.8312020460358056, + "grad_norm": 0.22423324668003744, + "learning_rate": 9.437703391121457e-06, + "loss": 1.0577, + "step": 1107 + }, + { + "epoch": 2.833759590792839, + "grad_norm": 0.16927725611399697, + "learning_rate": 9.419872448540821e-06, + "loss": 1.069, + "step": 1108 + }, + { + "epoch": 2.836317135549872, + "grad_norm": 0.20438283854954223, + "learning_rate": 9.402043356470214e-06, + "loss": 1.073, + "step": 1109 + }, + { + "epoch": 2.8388746803069056, + "grad_norm": 0.18438606363143578, + "learning_rate": 9.384216171781458e-06, + "loss": 1.0535, + "step": 1110 + }, + { + "epoch": 2.8414322250639388, + "grad_norm": 0.20553714025032518, + "learning_rate": 9.366390951340298e-06, + "loss": 1.053, + "step": 1111 + }, + { + "epoch": 2.843989769820972, + "grad_norm": 0.214363275474718, + "learning_rate": 9.348567752006208e-06, + "loss": 1.0618, + "step": 1112 + }, + { + "epoch": 2.846547314578005, + "grad_norm": 0.20301805224099836, + "learning_rate": 9.330746630632224e-06, + "loss": 1.0511, + "step": 1113 + }, + { + "epoch": 2.8491048593350383, + "grad_norm": 0.2579128437030396, + "learning_rate": 9.312927644064741e-06, + "loss": 1.0518, + "step": 1114 + }, + { + "epoch": 2.8516624040920715, + "grad_norm": 0.2181960904834944, + "learning_rate": 9.295110849143361e-06, + "loss": 1.0568, + "step": 1115 + }, + { + "epoch": 2.8542199488491047, + "grad_norm": 0.19693379707125938, + "learning_rate": 9.277296302700675e-06, + "loss": 1.0484, + "step": 1116 + }, + { + "epoch": 2.8567774936061383, + "grad_norm": 0.2157310477319499, + "learning_rate": 9.259484061562114e-06, + "loss": 1.0334, + "step": 1117 + }, + { + "epoch": 2.8593350383631715, + "grad_norm": 0.18163692565811201, + "learning_rate": 9.241674182545755e-06, + "loss": 1.0649, + "step": 1118 + }, + { + "epoch": 2.8618925831202047, + "grad_norm": 0.22334516764144044, + "learning_rate": 9.223866722462134e-06, + "loss": 1.0379, + "step": 1119 + }, + { + "epoch": 2.864450127877238, + "grad_norm": 0.18179305778409852, + "learning_rate": 9.206061738114087e-06, + "loss": 1.0941, + "step": 1120 + }, + { + "epoch": 2.867007672634271, + "grad_norm": 0.22000901646758675, + "learning_rate": 9.188259286296527e-06, + "loss": 1.062, + "step": 1121 + }, + { + "epoch": 2.869565217391304, + "grad_norm": 0.20132285430691718, + "learning_rate": 9.170459423796309e-06, + "loss": 1.0418, + "step": 1122 + }, + { + "epoch": 2.8721227621483374, + "grad_norm": 0.18151007124149462, + "learning_rate": 9.152662207392025e-06, + "loss": 1.0117, + "step": 1123 + }, + { + "epoch": 2.874680306905371, + "grad_norm": 0.20958529291718883, + "learning_rate": 9.134867693853818e-06, + "loss": 1.0953, + "step": 1124 + }, + { + "epoch": 2.877237851662404, + "grad_norm": 0.22158861933579652, + "learning_rate": 9.11707593994322e-06, + "loss": 1.0714, + "step": 1125 + }, + { + "epoch": 2.8797953964194374, + "grad_norm": 0.18234147869625544, + "learning_rate": 9.099287002412957e-06, + "loss": 1.0621, + "step": 1126 + }, + { + "epoch": 2.8823529411764706, + "grad_norm": 0.21232832072702237, + "learning_rate": 9.08150093800676e-06, + "loss": 1.0292, + "step": 1127 + }, + { + "epoch": 2.8849104859335037, + "grad_norm": 0.22176137811022292, + "learning_rate": 9.063717803459215e-06, + "loss": 1.0336, + "step": 1128 + }, + { + "epoch": 2.887468030690537, + "grad_norm": 0.2318369191230884, + "learning_rate": 9.045937655495544e-06, + "loss": 1.0408, + "step": 1129 + }, + { + "epoch": 2.89002557544757, + "grad_norm": 0.21500699732164258, + "learning_rate": 9.028160550831458e-06, + "loss": 1.0667, + "step": 1130 + }, + { + "epoch": 2.8925831202046037, + "grad_norm": 0.2380668338506343, + "learning_rate": 9.01038654617295e-06, + "loss": 1.0792, + "step": 1131 + }, + { + "epoch": 2.895140664961637, + "grad_norm": 0.19481533312148408, + "learning_rate": 8.992615698216121e-06, + "loss": 1.015, + "step": 1132 + }, + { + "epoch": 2.89769820971867, + "grad_norm": 0.21841251935433692, + "learning_rate": 8.974848063647017e-06, + "loss": 1.0342, + "step": 1133 + }, + { + "epoch": 2.9002557544757033, + "grad_norm": 0.21137856495367685, + "learning_rate": 8.95708369914142e-06, + "loss": 1.0361, + "step": 1134 + }, + { + "epoch": 2.9028132992327365, + "grad_norm": 0.1946101170447674, + "learning_rate": 8.93932266136469e-06, + "loss": 1.0608, + "step": 1135 + }, + { + "epoch": 2.90537084398977, + "grad_norm": 0.23628212061281142, + "learning_rate": 8.92156500697157e-06, + "loss": 1.0324, + "step": 1136 + }, + { + "epoch": 2.907928388746803, + "grad_norm": 0.19457336047499793, + "learning_rate": 8.903810792606019e-06, + "loss": 1.0974, + "step": 1137 + }, + { + "epoch": 2.9104859335038364, + "grad_norm": 0.23806632883818715, + "learning_rate": 8.886060074901006e-06, + "loss": 1.0339, + "step": 1138 + }, + { + "epoch": 2.9130434782608696, + "grad_norm": 0.20401432332115726, + "learning_rate": 8.868312910478362e-06, + "loss": 1.0478, + "step": 1139 + }, + { + "epoch": 2.915601023017903, + "grad_norm": 0.20265994858679084, + "learning_rate": 8.850569355948579e-06, + "loss": 1.0411, + "step": 1140 + }, + { + "epoch": 2.918158567774936, + "grad_norm": 0.20101667064863665, + "learning_rate": 8.832829467910631e-06, + "loss": 1.0576, + "step": 1141 + }, + { + "epoch": 2.920716112531969, + "grad_norm": 0.20307554858476298, + "learning_rate": 8.815093302951805e-06, + "loss": 1.0676, + "step": 1142 + }, + { + "epoch": 2.923273657289003, + "grad_norm": 0.18733836795641187, + "learning_rate": 8.797360917647498e-06, + "loss": 1.0635, + "step": 1143 + }, + { + "epoch": 2.9258312020460355, + "grad_norm": 0.19358622617848392, + "learning_rate": 8.77963236856106e-06, + "loss": 1.0418, + "step": 1144 + }, + { + "epoch": 2.928388746803069, + "grad_norm": 0.18883350362824358, + "learning_rate": 8.761907712243606e-06, + "loss": 1.0491, + "step": 1145 + }, + { + "epoch": 2.9309462915601023, + "grad_norm": 0.16898484877424805, + "learning_rate": 8.744187005233826e-06, + "loss": 1.0361, + "step": 1146 + }, + { + "epoch": 2.9335038363171355, + "grad_norm": 0.16398622387807313, + "learning_rate": 8.726470304057821e-06, + "loss": 1.0541, + "step": 1147 + }, + { + "epoch": 2.9360613810741687, + "grad_norm": 0.16164350986788764, + "learning_rate": 8.708757665228909e-06, + "loss": 1.0523, + "step": 1148 + }, + { + "epoch": 2.938618925831202, + "grad_norm": 0.16164661766952496, + "learning_rate": 8.691049145247446e-06, + "loss": 1.0681, + "step": 1149 + }, + { + "epoch": 2.9411764705882355, + "grad_norm": 0.15749749181969402, + "learning_rate": 8.673344800600657e-06, + "loss": 1.0573, + "step": 1150 + }, + { + "epoch": 2.9437340153452687, + "grad_norm": 0.15525122241929215, + "learning_rate": 8.655644687762443e-06, + "loss": 1.0722, + "step": 1151 + }, + { + "epoch": 2.946291560102302, + "grad_norm": 0.15709889969206967, + "learning_rate": 8.637948863193215e-06, + "loss": 1.0431, + "step": 1152 + }, + { + "epoch": 2.948849104859335, + "grad_norm": 0.16989598488301447, + "learning_rate": 8.620257383339694e-06, + "loss": 1.0658, + "step": 1153 + }, + { + "epoch": 2.9514066496163682, + "grad_norm": 0.1595964424160409, + "learning_rate": 8.602570304634745e-06, + "loss": 1.0701, + "step": 1154 + }, + { + "epoch": 2.9539641943734014, + "grad_norm": 0.16986735433032052, + "learning_rate": 8.5848876834972e-06, + "loss": 1.086, + "step": 1155 + }, + { + "epoch": 2.9565217391304346, + "grad_norm": 0.19325690151423974, + "learning_rate": 8.567209576331662e-06, + "loss": 1.059, + "step": 1156 + }, + { + "epoch": 2.959079283887468, + "grad_norm": 0.16112687651565402, + "learning_rate": 8.549536039528351e-06, + "loss": 1.0505, + "step": 1157 + }, + { + "epoch": 2.9616368286445014, + "grad_norm": 0.18938106017855888, + "learning_rate": 8.53186712946289e-06, + "loss": 1.0448, + "step": 1158 + }, + { + "epoch": 2.9641943734015346, + "grad_norm": 0.17573092327281262, + "learning_rate": 8.514202902496157e-06, + "loss": 1.045, + "step": 1159 + }, + { + "epoch": 2.9667519181585678, + "grad_norm": 0.16472675003522239, + "learning_rate": 8.496543414974084e-06, + "loss": 1.0672, + "step": 1160 + }, + { + "epoch": 2.969309462915601, + "grad_norm": 0.1748756285533725, + "learning_rate": 8.478888723227486e-06, + "loss": 1.0364, + "step": 1161 + }, + { + "epoch": 2.971867007672634, + "grad_norm": 0.1767205601325944, + "learning_rate": 8.461238883571885e-06, + "loss": 1.0413, + "step": 1162 + }, + { + "epoch": 2.9744245524296673, + "grad_norm": 0.15740765778773358, + "learning_rate": 8.443593952307319e-06, + "loss": 1.0833, + "step": 1163 + }, + { + "epoch": 2.976982097186701, + "grad_norm": 0.1896056695064984, + "learning_rate": 8.425953985718182e-06, + "loss": 1.036, + "step": 1164 + }, + { + "epoch": 2.979539641943734, + "grad_norm": 0.17711517435630209, + "learning_rate": 8.408319040073011e-06, + "loss": 1.0447, + "step": 1165 + }, + { + "epoch": 2.9820971867007673, + "grad_norm": 0.16635113314658131, + "learning_rate": 8.390689171624341e-06, + "loss": 1.0771, + "step": 1166 + }, + { + "epoch": 2.9846547314578005, + "grad_norm": 0.17437808910596117, + "learning_rate": 8.373064436608512e-06, + "loss": 1.053, + "step": 1167 + }, + { + "epoch": 2.9872122762148337, + "grad_norm": 0.17619213607284423, + "learning_rate": 8.355444891245482e-06, + "loss": 1.0684, + "step": 1168 + }, + { + "epoch": 2.9897698209718673, + "grad_norm": 0.18483592261300663, + "learning_rate": 8.337830591738664e-06, + "loss": 1.0437, + "step": 1169 + }, + { + "epoch": 2.9923273657289, + "grad_norm": 0.18442193916145447, + "learning_rate": 8.32022159427473e-06, + "loss": 1.0422, + "step": 1170 + }, + { + "epoch": 2.9948849104859336, + "grad_norm": 0.2119894606858352, + "learning_rate": 8.302617955023438e-06, + "loss": 1.0383, + "step": 1171 + }, + { + "epoch": 2.997442455242967, + "grad_norm": 0.2113360486461104, + "learning_rate": 8.285019730137462e-06, + "loss": 1.0814, + "step": 1172 + }, + { + "epoch": 3.0, + "grad_norm": 0.2111827709122195, + "learning_rate": 8.2674269757522e-06, + "loss": 1.0749, + "step": 1173 + }, + { + "epoch": 3.002557544757033, + "grad_norm": 0.17743968023223414, + "learning_rate": 8.2498397479856e-06, + "loss": 1.0689, + "step": 1174 + }, + { + "epoch": 3.0051150895140664, + "grad_norm": 0.21230668911328027, + "learning_rate": 8.232258102937987e-06, + "loss": 1.0685, + "step": 1175 + }, + { + "epoch": 3.0076726342710995, + "grad_norm": 0.18796325619774717, + "learning_rate": 8.214682096691866e-06, + "loss": 1.0474, + "step": 1176 + }, + { + "epoch": 3.010230179028133, + "grad_norm": 0.1863635190742179, + "learning_rate": 8.197111785311767e-06, + "loss": 1.0858, + "step": 1177 + }, + { + "epoch": 3.0127877237851663, + "grad_norm": 0.1838798652920452, + "learning_rate": 8.179547224844046e-06, + "loss": 1.0743, + "step": 1178 + }, + { + "epoch": 3.0153452685421995, + "grad_norm": 0.17485799326659002, + "learning_rate": 8.161988471316723e-06, + "loss": 1.0228, + "step": 1179 + }, + { + "epoch": 3.0179028132992327, + "grad_norm": 0.21207868945480415, + "learning_rate": 8.144435580739285e-06, + "loss": 1.0336, + "step": 1180 + }, + { + "epoch": 3.020460358056266, + "grad_norm": 0.16449752076147575, + "learning_rate": 8.12688860910253e-06, + "loss": 1.0455, + "step": 1181 + }, + { + "epoch": 3.023017902813299, + "grad_norm": 0.19947300226681555, + "learning_rate": 8.109347612378357e-06, + "loss": 1.0395, + "step": 1182 + }, + { + "epoch": 3.0255754475703327, + "grad_norm": 0.18190648941101212, + "learning_rate": 8.091812646519624e-06, + "loss": 1.0623, + "step": 1183 + }, + { + "epoch": 3.028132992327366, + "grad_norm": 0.1646190514225652, + "learning_rate": 8.074283767459941e-06, + "loss": 1.0264, + "step": 1184 + }, + { + "epoch": 3.030690537084399, + "grad_norm": 0.19900332810490962, + "learning_rate": 8.056761031113506e-06, + "loss": 1.0419, + "step": 1185 + }, + { + "epoch": 3.0332480818414322, + "grad_norm": 0.19790261143418011, + "learning_rate": 8.03924449337493e-06, + "loss": 1.0239, + "step": 1186 + }, + { + "epoch": 3.0358056265984654, + "grad_norm": 0.1735899464590682, + "learning_rate": 8.021734210119031e-06, + "loss": 1.0482, + "step": 1187 + }, + { + "epoch": 3.0383631713554986, + "grad_norm": 0.24811292812795957, + "learning_rate": 8.004230237200695e-06, + "loss": 1.0117, + "step": 1188 + }, + { + "epoch": 3.040920716112532, + "grad_norm": 0.17073008104062196, + "learning_rate": 7.986732630454673e-06, + "loss": 1.0469, + "step": 1189 + }, + { + "epoch": 3.0434782608695654, + "grad_norm": 0.2528836646880793, + "learning_rate": 7.969241445695407e-06, + "loss": 1.0341, + "step": 1190 + }, + { + "epoch": 3.0460358056265986, + "grad_norm": 0.19007874013606105, + "learning_rate": 7.951756738716861e-06, + "loss": 1.0147, + "step": 1191 + }, + { + "epoch": 3.0485933503836318, + "grad_norm": 0.20187206991127415, + "learning_rate": 7.934278565292328e-06, + "loss": 1.0535, + "step": 1192 + }, + { + "epoch": 3.051150895140665, + "grad_norm": 0.20526762415260827, + "learning_rate": 7.916806981174258e-06, + "loss": 1.0605, + "step": 1193 + }, + { + "epoch": 3.053708439897698, + "grad_norm": 0.20055119376497793, + "learning_rate": 7.899342042094094e-06, + "loss": 1.0316, + "step": 1194 + }, + { + "epoch": 3.0562659846547313, + "grad_norm": 0.1968308279791684, + "learning_rate": 7.881883803762071e-06, + "loss": 1.0367, + "step": 1195 + }, + { + "epoch": 3.0588235294117645, + "grad_norm": 0.18030703210119325, + "learning_rate": 7.864432321867056e-06, + "loss": 1.0504, + "step": 1196 + }, + { + "epoch": 3.061381074168798, + "grad_norm": 0.18935954626691623, + "learning_rate": 7.846987652076372e-06, + "loss": 1.0511, + "step": 1197 + }, + { + "epoch": 3.0639386189258313, + "grad_norm": 0.17290346534417522, + "learning_rate": 7.829549850035585e-06, + "loss": 1.0484, + "step": 1198 + }, + { + "epoch": 3.0664961636828645, + "grad_norm": 0.18722490812221246, + "learning_rate": 7.812118971368384e-06, + "loss": 1.0661, + "step": 1199 + }, + { + "epoch": 3.0690537084398977, + "grad_norm": 0.20012871337269098, + "learning_rate": 7.794695071676355e-06, + "loss": 1.0245, + "step": 1200 + }, + { + "epoch": 3.071611253196931, + "grad_norm": 0.21712726425678322, + "learning_rate": 7.777278206538832e-06, + "loss": 1.0368, + "step": 1201 + }, + { + "epoch": 3.074168797953964, + "grad_norm": 0.18764265039969955, + "learning_rate": 7.75986843151271e-06, + "loss": 1.0603, + "step": 1202 + }, + { + "epoch": 3.0767263427109977, + "grad_norm": 0.2284378574380906, + "learning_rate": 7.742465802132261e-06, + "loss": 1.0666, + "step": 1203 + }, + { + "epoch": 3.079283887468031, + "grad_norm": 0.18138326397527504, + "learning_rate": 7.725070373908967e-06, + "loss": 1.0653, + "step": 1204 + }, + { + "epoch": 3.081841432225064, + "grad_norm": 0.20238266442132807, + "learning_rate": 7.707682202331338e-06, + "loss": 1.0659, + "step": 1205 + }, + { + "epoch": 3.084398976982097, + "grad_norm": 0.16899939143822162, + "learning_rate": 7.690301342864738e-06, + "loss": 1.0361, + "step": 1206 + }, + { + "epoch": 3.0869565217391304, + "grad_norm": 0.21399879876961514, + "learning_rate": 7.672927850951213e-06, + "loss": 1.0579, + "step": 1207 + }, + { + "epoch": 3.0895140664961636, + "grad_norm": 0.16140650557730965, + "learning_rate": 7.6555617820093e-06, + "loss": 1.0867, + "step": 1208 + }, + { + "epoch": 3.0920716112531967, + "grad_norm": 0.2030576541987284, + "learning_rate": 7.638203191433848e-06, + "loss": 1.0312, + "step": 1209 + }, + { + "epoch": 3.0946291560102304, + "grad_norm": 0.1653674904692065, + "learning_rate": 7.620852134595875e-06, + "loss": 1.0553, + "step": 1210 + }, + { + "epoch": 3.0971867007672635, + "grad_norm": 0.21004837372358645, + "learning_rate": 7.60350866684235e-06, + "loss": 1.0566, + "step": 1211 + }, + { + "epoch": 3.0997442455242967, + "grad_norm": 0.18582555924313174, + "learning_rate": 7.586172843496042e-06, + "loss": 1.0194, + "step": 1212 + }, + { + "epoch": 3.10230179028133, + "grad_norm": 0.2350752210692893, + "learning_rate": 7.568844719855328e-06, + "loss": 1.0887, + "step": 1213 + }, + { + "epoch": 3.104859335038363, + "grad_norm": 0.1752787960126917, + "learning_rate": 7.5515243511940395e-06, + "loss": 1.0367, + "step": 1214 + }, + { + "epoch": 3.1074168797953963, + "grad_norm": 0.20184036652559692, + "learning_rate": 7.534211792761249e-06, + "loss": 1.0784, + "step": 1215 + }, + { + "epoch": 3.10997442455243, + "grad_norm": 0.19595647838718663, + "learning_rate": 7.516907099781131e-06, + "loss": 1.0298, + "step": 1216 + }, + { + "epoch": 3.112531969309463, + "grad_norm": 0.20222428308830126, + "learning_rate": 7.49961032745277e-06, + "loss": 1.0678, + "step": 1217 + }, + { + "epoch": 3.1150895140664963, + "grad_norm": 0.19294563663829822, + "learning_rate": 7.4823215309499766e-06, + "loss": 1.0397, + "step": 1218 + }, + { + "epoch": 3.1176470588235294, + "grad_norm": 0.19397261170806676, + "learning_rate": 7.465040765421132e-06, + "loss": 1.058, + "step": 1219 + }, + { + "epoch": 3.1202046035805626, + "grad_norm": 0.22587828052192943, + "learning_rate": 7.447768085988987e-06, + "loss": 1.0461, + "step": 1220 + }, + { + "epoch": 3.122762148337596, + "grad_norm": 0.2140035097820999, + "learning_rate": 7.430503547750505e-06, + "loss": 1.0425, + "step": 1221 + }, + { + "epoch": 3.125319693094629, + "grad_norm": 0.20328335993216895, + "learning_rate": 7.413247205776684e-06, + "loss": 0.9899, + "step": 1222 + }, + { + "epoch": 3.1278772378516626, + "grad_norm": 0.17357867143063074, + "learning_rate": 7.3959991151123685e-06, + "loss": 1.0532, + "step": 1223 + }, + { + "epoch": 3.130434782608696, + "grad_norm": 0.22322271848597508, + "learning_rate": 7.378759330776093e-06, + "loss": 1.0212, + "step": 1224 + }, + { + "epoch": 3.132992327365729, + "grad_norm": 0.16810780127137187, + "learning_rate": 7.361527907759893e-06, + "loss": 1.0279, + "step": 1225 + }, + { + "epoch": 3.135549872122762, + "grad_norm": 0.23876493089254128, + "learning_rate": 7.344304901029121e-06, + "loss": 1.0474, + "step": 1226 + }, + { + "epoch": 3.1381074168797953, + "grad_norm": 0.1666563839380977, + "learning_rate": 7.327090365522303e-06, + "loss": 1.0457, + "step": 1227 + }, + { + "epoch": 3.1406649616368285, + "grad_norm": 0.233785304749979, + "learning_rate": 7.309884356150925e-06, + "loss": 1.0644, + "step": 1228 + }, + { + "epoch": 3.1432225063938617, + "grad_norm": 0.16224260381430458, + "learning_rate": 7.2926869277992884e-06, + "loss": 1.0704, + "step": 1229 + }, + { + "epoch": 3.1457800511508953, + "grad_norm": 0.213314385450332, + "learning_rate": 7.275498135324322e-06, + "loss": 1.0518, + "step": 1230 + }, + { + "epoch": 3.1483375959079285, + "grad_norm": 0.1676105524904316, + "learning_rate": 7.258318033555394e-06, + "loss": 1.0622, + "step": 1231 + }, + { + "epoch": 3.1508951406649617, + "grad_norm": 0.20581044358420533, + "learning_rate": 7.2411466772941684e-06, + "loss": 1.0389, + "step": 1232 + }, + { + "epoch": 3.153452685421995, + "grad_norm": 0.15515519926002075, + "learning_rate": 7.2239841213144e-06, + "loss": 1.0802, + "step": 1233 + }, + { + "epoch": 3.156010230179028, + "grad_norm": 0.1937872427855819, + "learning_rate": 7.20683042036178e-06, + "loss": 1.0616, + "step": 1234 + }, + { + "epoch": 3.1585677749360612, + "grad_norm": 0.16287563642941633, + "learning_rate": 7.189685629153749e-06, + "loss": 1.0518, + "step": 1235 + }, + { + "epoch": 3.1611253196930944, + "grad_norm": 0.16444450384166215, + "learning_rate": 7.1725498023793305e-06, + "loss": 1.055, + "step": 1236 + }, + { + "epoch": 3.163682864450128, + "grad_norm": 0.15483746878996799, + "learning_rate": 7.155422994698948e-06, + "loss": 1.044, + "step": 1237 + }, + { + "epoch": 3.166240409207161, + "grad_norm": 0.16555800371038418, + "learning_rate": 7.138305260744256e-06, + "loss": 1.0489, + "step": 1238 + }, + { + "epoch": 3.1687979539641944, + "grad_norm": 0.16421185384625095, + "learning_rate": 7.121196655117975e-06, + "loss": 1.0133, + "step": 1239 + }, + { + "epoch": 3.1713554987212276, + "grad_norm": 0.15905768015523744, + "learning_rate": 7.104097232393692e-06, + "loss": 1.039, + "step": 1240 + }, + { + "epoch": 3.1739130434782608, + "grad_norm": 0.15268576953010804, + "learning_rate": 7.087007047115719e-06, + "loss": 1.0275, + "step": 1241 + }, + { + "epoch": 3.176470588235294, + "grad_norm": 0.1710081046719881, + "learning_rate": 7.069926153798889e-06, + "loss": 1.0343, + "step": 1242 + }, + { + "epoch": 3.1790281329923276, + "grad_norm": 0.16935470183977597, + "learning_rate": 7.052854606928396e-06, + "loss": 1.0442, + "step": 1243 + }, + { + "epoch": 3.1815856777493607, + "grad_norm": 0.17381989218653407, + "learning_rate": 7.0357924609596305e-06, + "loss": 1.0272, + "step": 1244 + }, + { + "epoch": 3.184143222506394, + "grad_norm": 0.1809090074415964, + "learning_rate": 7.018739770317985e-06, + "loss": 1.0039, + "step": 1245 + }, + { + "epoch": 3.186700767263427, + "grad_norm": 0.16405848889955926, + "learning_rate": 7.001696589398699e-06, + "loss": 1.0569, + "step": 1246 + }, + { + "epoch": 3.1892583120204603, + "grad_norm": 0.19485294563253486, + "learning_rate": 6.984662972566674e-06, + "loss": 1.0636, + "step": 1247 + }, + { + "epoch": 3.1918158567774935, + "grad_norm": 0.1603872937860603, + "learning_rate": 6.967638974156299e-06, + "loss": 1.0439, + "step": 1248 + }, + { + "epoch": 3.1943734015345266, + "grad_norm": 0.19115402938555065, + "learning_rate": 6.950624648471289e-06, + "loss": 1.033, + "step": 1249 + }, + { + "epoch": 3.1969309462915603, + "grad_norm": 0.15897058089861524, + "learning_rate": 6.933620049784502e-06, + "loss": 1.0437, + "step": 1250 + }, + { + "epoch": 3.1994884910485935, + "grad_norm": 0.1865666397196807, + "learning_rate": 6.9166252323377705e-06, + "loss": 1.0201, + "step": 1251 + }, + { + "epoch": 3.2020460358056266, + "grad_norm": 0.18005383040199222, + "learning_rate": 6.899640250341726e-06, + "loss": 1.0607, + "step": 1252 + }, + { + "epoch": 3.20460358056266, + "grad_norm": 0.18667666493469226, + "learning_rate": 6.882665157975617e-06, + "loss": 1.0151, + "step": 1253 + }, + { + "epoch": 3.207161125319693, + "grad_norm": 0.16755268817909583, + "learning_rate": 6.865700009387161e-06, + "loss": 1.0398, + "step": 1254 + }, + { + "epoch": 3.209718670076726, + "grad_norm": 0.17489599503934009, + "learning_rate": 6.8487448586923445e-06, + "loss": 1.0411, + "step": 1255 + }, + { + "epoch": 3.21227621483376, + "grad_norm": 0.19955868168103402, + "learning_rate": 6.831799759975273e-06, + "loss": 1.0312, + "step": 1256 + }, + { + "epoch": 3.214833759590793, + "grad_norm": 0.17120578080679916, + "learning_rate": 6.814864767287977e-06, + "loss": 1.0364, + "step": 1257 + }, + { + "epoch": 3.217391304347826, + "grad_norm": 0.20773510702967912, + "learning_rate": 6.797939934650263e-06, + "loss": 1.0324, + "step": 1258 + }, + { + "epoch": 3.2199488491048593, + "grad_norm": 0.1586496816689849, + "learning_rate": 6.781025316049512e-06, + "loss": 1.0794, + "step": 1259 + }, + { + "epoch": 3.2225063938618925, + "grad_norm": 0.20969996522100806, + "learning_rate": 6.7641209654405375e-06, + "loss": 1.0463, + "step": 1260 + }, + { + "epoch": 3.2250639386189257, + "grad_norm": 0.16911339560919933, + "learning_rate": 6.747226936745395e-06, + "loss": 1.0292, + "step": 1261 + }, + { + "epoch": 3.227621483375959, + "grad_norm": 0.17406389220265206, + "learning_rate": 6.730343283853213e-06, + "loss": 1.0104, + "step": 1262 + }, + { + "epoch": 3.2301790281329925, + "grad_norm": 0.19086270430612148, + "learning_rate": 6.713470060620033e-06, + "loss": 1.0333, + "step": 1263 + }, + { + "epoch": 3.2327365728900257, + "grad_norm": 0.15722213977712743, + "learning_rate": 6.6966073208686135e-06, + "loss": 1.0392, + "step": 1264 + }, + { + "epoch": 3.235294117647059, + "grad_norm": 0.18709369627876488, + "learning_rate": 6.679755118388275e-06, + "loss": 1.0333, + "step": 1265 + }, + { + "epoch": 3.237851662404092, + "grad_norm": 0.1699678534444773, + "learning_rate": 6.662913506934737e-06, + "loss": 1.0404, + "step": 1266 + }, + { + "epoch": 3.2404092071611252, + "grad_norm": 0.18697964330748695, + "learning_rate": 6.646082540229923e-06, + "loss": 1.0341, + "step": 1267 + }, + { + "epoch": 3.2429667519181584, + "grad_norm": 0.16867872295801503, + "learning_rate": 6.629262271961811e-06, + "loss": 0.9955, + "step": 1268 + }, + { + "epoch": 3.2455242966751916, + "grad_norm": 0.1757388936159618, + "learning_rate": 6.61245275578425e-06, + "loss": 1.0588, + "step": 1269 + }, + { + "epoch": 3.2480818414322252, + "grad_norm": 0.179354789438552, + "learning_rate": 6.595654045316781e-06, + "loss": 1.0342, + "step": 1270 + }, + { + "epoch": 3.2506393861892584, + "grad_norm": 0.16261099723737935, + "learning_rate": 6.5788661941444934e-06, + "loss": 1.0512, + "step": 1271 + }, + { + "epoch": 3.2531969309462916, + "grad_norm": 0.17256463607415848, + "learning_rate": 6.562089255817823e-06, + "loss": 1.0344, + "step": 1272 + }, + { + "epoch": 3.2557544757033248, + "grad_norm": 0.17074946791571327, + "learning_rate": 6.545323283852407e-06, + "loss": 1.0343, + "step": 1273 + }, + { + "epoch": 3.258312020460358, + "grad_norm": 0.16871272465633477, + "learning_rate": 6.528568331728895e-06, + "loss": 1.0476, + "step": 1274 + }, + { + "epoch": 3.260869565217391, + "grad_norm": 0.1578043264468705, + "learning_rate": 6.511824452892791e-06, + "loss": 1.0539, + "step": 1275 + }, + { + "epoch": 3.2634271099744243, + "grad_norm": 0.16975041042828792, + "learning_rate": 6.495091700754266e-06, + "loss": 1.009, + "step": 1276 + }, + { + "epoch": 3.265984654731458, + "grad_norm": 0.16784506398603977, + "learning_rate": 6.478370128688005e-06, + "loss": 1.0654, + "step": 1277 + }, + { + "epoch": 3.268542199488491, + "grad_norm": 0.1573687001595116, + "learning_rate": 6.461659790033038e-06, + "loss": 1.0333, + "step": 1278 + }, + { + "epoch": 3.2710997442455243, + "grad_norm": 0.15498306536956233, + "learning_rate": 6.444960738092548e-06, + "loss": 1.0848, + "step": 1279 + }, + { + "epoch": 3.2736572890025575, + "grad_norm": 0.1895685081315579, + "learning_rate": 6.428273026133731e-06, + "loss": 1.0533, + "step": 1280 + }, + { + "epoch": 3.2762148337595907, + "grad_norm": 0.19242571512113535, + "learning_rate": 6.411596707387594e-06, + "loss": 1.0621, + "step": 1281 + }, + { + "epoch": 3.2787723785166243, + "grad_norm": 0.15700921828708964, + "learning_rate": 6.39493183504881e-06, + "loss": 1.0477, + "step": 1282 + }, + { + "epoch": 3.2813299232736575, + "grad_norm": 0.21159038732742347, + "learning_rate": 6.378278462275542e-06, + "loss": 1.0468, + "step": 1283 + }, + { + "epoch": 3.2838874680306906, + "grad_norm": 0.19962601553872036, + "learning_rate": 6.361636642189269e-06, + "loss": 1.0426, + "step": 1284 + }, + { + "epoch": 3.286445012787724, + "grad_norm": 0.1763686863110666, + "learning_rate": 6.345006427874616e-06, + "loss": 1.0453, + "step": 1285 + }, + { + "epoch": 3.289002557544757, + "grad_norm": 0.1776125396661667, + "learning_rate": 6.328387872379193e-06, + "loss": 1.0424, + "step": 1286 + }, + { + "epoch": 3.29156010230179, + "grad_norm": 0.18658506842064834, + "learning_rate": 6.311781028713414e-06, + "loss": 1.0337, + "step": 1287 + }, + { + "epoch": 3.2941176470588234, + "grad_norm": 0.15532653749572625, + "learning_rate": 6.29518594985034e-06, + "loss": 1.0185, + "step": 1288 + }, + { + "epoch": 3.296675191815857, + "grad_norm": 0.20040506008569664, + "learning_rate": 6.278602688725498e-06, + "loss": 1.019, + "step": 1289 + }, + { + "epoch": 3.29923273657289, + "grad_norm": 0.15024078882430822, + "learning_rate": 6.262031298236728e-06, + "loss": 1.0163, + "step": 1290 + }, + { + "epoch": 3.3017902813299234, + "grad_norm": 0.187370700502857, + "learning_rate": 6.245471831243997e-06, + "loss": 1.0294, + "step": 1291 + }, + { + "epoch": 3.3043478260869565, + "grad_norm": 0.1836564928922123, + "learning_rate": 6.228924340569234e-06, + "loss": 1.0576, + "step": 1292 + }, + { + "epoch": 3.3069053708439897, + "grad_norm": 0.1680090110613116, + "learning_rate": 6.212388878996177e-06, + "loss": 0.9984, + "step": 1293 + }, + { + "epoch": 3.309462915601023, + "grad_norm": 0.18336870531248498, + "learning_rate": 6.195865499270186e-06, + "loss": 1.0298, + "step": 1294 + }, + { + "epoch": 3.312020460358056, + "grad_norm": 0.15787805699206167, + "learning_rate": 6.179354254098085e-06, + "loss": 1.0431, + "step": 1295 + }, + { + "epoch": 3.3145780051150897, + "grad_norm": 0.16543341313363566, + "learning_rate": 6.1628551961479856e-06, + "loss": 1.0544, + "step": 1296 + }, + { + "epoch": 3.317135549872123, + "grad_norm": 0.17482194497586678, + "learning_rate": 6.146368378049134e-06, + "loss": 1.0098, + "step": 1297 + }, + { + "epoch": 3.319693094629156, + "grad_norm": 0.15937828046028765, + "learning_rate": 6.129893852391722e-06, + "loss": 1.0294, + "step": 1298 + }, + { + "epoch": 3.3222506393861893, + "grad_norm": 0.19027583920574043, + "learning_rate": 6.113431671726735e-06, + "loss": 1.0549, + "step": 1299 + }, + { + "epoch": 3.3248081841432224, + "grad_norm": 0.16857302745550032, + "learning_rate": 6.096981888565784e-06, + "loss": 1.0772, + "step": 1300 + }, + { + "epoch": 3.3273657289002556, + "grad_norm": 0.16700609829050267, + "learning_rate": 6.080544555380927e-06, + "loss": 1.0184, + "step": 1301 + }, + { + "epoch": 3.329923273657289, + "grad_norm": 0.1829806527515392, + "learning_rate": 6.06411972460452e-06, + "loss": 1.0503, + "step": 1302 + }, + { + "epoch": 3.3324808184143224, + "grad_norm": 0.16756778358359, + "learning_rate": 6.047707448629023e-06, + "loss": 1.0294, + "step": 1303 + }, + { + "epoch": 3.3350383631713556, + "grad_norm": 0.17523646623624167, + "learning_rate": 6.031307779806858e-06, + "loss": 1.045, + "step": 1304 + }, + { + "epoch": 3.337595907928389, + "grad_norm": 0.16450304941716934, + "learning_rate": 6.014920770450232e-06, + "loss": 1.0194, + "step": 1305 + }, + { + "epoch": 3.340153452685422, + "grad_norm": 0.1805469762569222, + "learning_rate": 5.998546472830965e-06, + "loss": 1.0541, + "step": 1306 + }, + { + "epoch": 3.342710997442455, + "grad_norm": 0.1419264397187406, + "learning_rate": 5.982184939180337e-06, + "loss": 1.007, + "step": 1307 + }, + { + "epoch": 3.3452685421994883, + "grad_norm": 0.1604620222757409, + "learning_rate": 5.96583622168891e-06, + "loss": 1.0498, + "step": 1308 + }, + { + "epoch": 3.3478260869565215, + "grad_norm": 0.15382052852134748, + "learning_rate": 5.949500372506354e-06, + "loss": 1.0339, + "step": 1309 + }, + { + "epoch": 3.350383631713555, + "grad_norm": 0.1492360356034545, + "learning_rate": 5.9331774437413095e-06, + "loss": 1.0739, + "step": 1310 + }, + { + "epoch": 3.3529411764705883, + "grad_norm": 0.16403001997456812, + "learning_rate": 5.916867487461186e-06, + "loss": 1.0416, + "step": 1311 + }, + { + "epoch": 3.3554987212276215, + "grad_norm": 0.1614357839239018, + "learning_rate": 5.900570555692029e-06, + "loss": 1.0338, + "step": 1312 + }, + { + "epoch": 3.3580562659846547, + "grad_norm": 0.16872463368760673, + "learning_rate": 5.884286700418328e-06, + "loss": 1.0369, + "step": 1313 + }, + { + "epoch": 3.360613810741688, + "grad_norm": 0.18005310056772503, + "learning_rate": 5.868015973582856e-06, + "loss": 1.0474, + "step": 1314 + }, + { + "epoch": 3.363171355498721, + "grad_norm": 0.18096998826427219, + "learning_rate": 5.85175842708652e-06, + "loss": 1.0168, + "step": 1315 + }, + { + "epoch": 3.3657289002557547, + "grad_norm": 0.15348669875527965, + "learning_rate": 5.835514112788178e-06, + "loss": 1.0404, + "step": 1316 + }, + { + "epoch": 3.368286445012788, + "grad_norm": 0.18543116172669075, + "learning_rate": 5.819283082504482e-06, + "loss": 1.0501, + "step": 1317 + }, + { + "epoch": 3.370843989769821, + "grad_norm": 0.19694837114442768, + "learning_rate": 5.803065388009707e-06, + "loss": 1.047, + "step": 1318 + }, + { + "epoch": 3.373401534526854, + "grad_norm": 0.1550062877849358, + "learning_rate": 5.78686108103559e-06, + "loss": 1.0459, + "step": 1319 + }, + { + "epoch": 3.3759590792838874, + "grad_norm": 0.18130690935498406, + "learning_rate": 5.770670213271166e-06, + "loss": 1.0203, + "step": 1320 + }, + { + "epoch": 3.3785166240409206, + "grad_norm": 0.1791636691198155, + "learning_rate": 5.754492836362597e-06, + "loss": 1.0211, + "step": 1321 + }, + { + "epoch": 3.381074168797954, + "grad_norm": 0.1648503792031205, + "learning_rate": 5.7383290019130145e-06, + "loss": 1.0312, + "step": 1322 + }, + { + "epoch": 3.3836317135549874, + "grad_norm": 0.1796907978935755, + "learning_rate": 5.722178761482357e-06, + "loss": 1.0233, + "step": 1323 + }, + { + "epoch": 3.3861892583120206, + "grad_norm": 0.18399008293027794, + "learning_rate": 5.706042166587194e-06, + "loss": 1.038, + "step": 1324 + }, + { + "epoch": 3.3887468030690537, + "grad_norm": 0.14911113859818034, + "learning_rate": 5.689919268700559e-06, + "loss": 1.0401, + "step": 1325 + }, + { + "epoch": 3.391304347826087, + "grad_norm": 0.1721196020218365, + "learning_rate": 5.673810119251815e-06, + "loss": 1.042, + "step": 1326 + }, + { + "epoch": 3.39386189258312, + "grad_norm": 0.16271756082297875, + "learning_rate": 5.657714769626456e-06, + "loss": 1.028, + "step": 1327 + }, + { + "epoch": 3.3964194373401533, + "grad_norm": 0.16838385908481096, + "learning_rate": 5.641633271165955e-06, + "loss": 1.0164, + "step": 1328 + }, + { + "epoch": 3.398976982097187, + "grad_norm": 0.16460071206345933, + "learning_rate": 5.625565675167614e-06, + "loss": 1.0785, + "step": 1329 + }, + { + "epoch": 3.40153452685422, + "grad_norm": 0.15708748073278123, + "learning_rate": 5.609512032884385e-06, + "loss": 1.0622, + "step": 1330 + }, + { + "epoch": 3.4040920716112533, + "grad_norm": 0.16538861056024953, + "learning_rate": 5.593472395524692e-06, + "loss": 1.0335, + "step": 1331 + }, + { + "epoch": 3.4066496163682864, + "grad_norm": 0.17518337046372068, + "learning_rate": 5.577446814252311e-06, + "loss": 1.0527, + "step": 1332 + }, + { + "epoch": 3.4092071611253196, + "grad_norm": 0.15882089668156443, + "learning_rate": 5.561435340186167e-06, + "loss": 1.0281, + "step": 1333 + }, + { + "epoch": 3.411764705882353, + "grad_norm": 0.15563095727972648, + "learning_rate": 5.545438024400193e-06, + "loss": 1.0581, + "step": 1334 + }, + { + "epoch": 3.414322250639386, + "grad_norm": 0.17128623415715957, + "learning_rate": 5.529454917923149e-06, + "loss": 1.0419, + "step": 1335 + }, + { + "epoch": 3.4168797953964196, + "grad_norm": 0.15167801306485404, + "learning_rate": 5.513486071738481e-06, + "loss": 1.0404, + "step": 1336 + }, + { + "epoch": 3.419437340153453, + "grad_norm": 0.16952883948633865, + "learning_rate": 5.497531536784138e-06, + "loss": 1.0375, + "step": 1337 + }, + { + "epoch": 3.421994884910486, + "grad_norm": 0.1630572687589045, + "learning_rate": 5.481591363952421e-06, + "loss": 1.0262, + "step": 1338 + }, + { + "epoch": 3.424552429667519, + "grad_norm": 0.16085228272090568, + "learning_rate": 5.465665604089829e-06, + "loss": 1.0317, + "step": 1339 + }, + { + "epoch": 3.4271099744245523, + "grad_norm": 0.19118486517294114, + "learning_rate": 5.449754307996871e-06, + "loss": 1.0867, + "step": 1340 + }, + { + "epoch": 3.4296675191815855, + "grad_norm": 0.1691680997069369, + "learning_rate": 5.433857526427924e-06, + "loss": 1.0755, + "step": 1341 + }, + { + "epoch": 3.4322250639386187, + "grad_norm": 0.17086025750717415, + "learning_rate": 5.417975310091068e-06, + "loss": 1.038, + "step": 1342 + }, + { + "epoch": 3.4347826086956523, + "grad_norm": 0.1547439630691515, + "learning_rate": 5.402107709647921e-06, + "loss": 1.0512, + "step": 1343 + }, + { + "epoch": 3.4373401534526855, + "grad_norm": 0.18002318829806357, + "learning_rate": 5.386254775713482e-06, + "loss": 1.0333, + "step": 1344 + }, + { + "epoch": 3.4398976982097187, + "grad_norm": 0.15474642805276234, + "learning_rate": 5.370416558855955e-06, + "loss": 1.0215, + "step": 1345 + }, + { + "epoch": 3.442455242966752, + "grad_norm": 0.16033707255186253, + "learning_rate": 5.354593109596621e-06, + "loss": 1.0628, + "step": 1346 + }, + { + "epoch": 3.445012787723785, + "grad_norm": 0.17670977631385593, + "learning_rate": 5.338784478409629e-06, + "loss": 1.0398, + "step": 1347 + }, + { + "epoch": 3.4475703324808182, + "grad_norm": 0.1659664475684347, + "learning_rate": 5.3229907157218745e-06, + "loss": 1.0603, + "step": 1348 + }, + { + "epoch": 3.4501278772378514, + "grad_norm": 0.15593100830308165, + "learning_rate": 5.307211871912828e-06, + "loss": 1.0651, + "step": 1349 + }, + { + "epoch": 3.452685421994885, + "grad_norm": 0.17222844205737484, + "learning_rate": 5.291447997314367e-06, + "loss": 1.0524, + "step": 1350 + }, + { + "epoch": 3.455242966751918, + "grad_norm": 0.18445018135377744, + "learning_rate": 5.275699142210615e-06, + "loss": 1.0483, + "step": 1351 + }, + { + "epoch": 3.4578005115089514, + "grad_norm": 0.15254595235934099, + "learning_rate": 5.259965356837795e-06, + "loss": 1.056, + "step": 1352 + }, + { + "epoch": 3.4603580562659846, + "grad_norm": 0.1681256587998916, + "learning_rate": 5.2442466913840504e-06, + "loss": 1.0201, + "step": 1353 + }, + { + "epoch": 3.4629156010230178, + "grad_norm": 0.20513309582498962, + "learning_rate": 5.228543195989303e-06, + "loss": 1.042, + "step": 1354 + }, + { + "epoch": 3.4654731457800514, + "grad_norm": 0.1651579107715655, + "learning_rate": 5.212854920745075e-06, + "loss": 1.0335, + "step": 1355 + }, + { + "epoch": 3.4680306905370846, + "grad_norm": 0.1672316644360442, + "learning_rate": 5.197181915694355e-06, + "loss": 1.0402, + "step": 1356 + }, + { + "epoch": 3.4705882352941178, + "grad_norm": 0.18835532310561898, + "learning_rate": 5.18152423083141e-06, + "loss": 1.0441, + "step": 1357 + }, + { + "epoch": 3.473145780051151, + "grad_norm": 0.15869844025048313, + "learning_rate": 5.1658819161016296e-06, + "loss": 1.0468, + "step": 1358 + }, + { + "epoch": 3.475703324808184, + "grad_norm": 0.16464692549899385, + "learning_rate": 5.1502550214013994e-06, + "loss": 1.0664, + "step": 1359 + }, + { + "epoch": 3.4782608695652173, + "grad_norm": 0.15694449028311525, + "learning_rate": 5.134643596577898e-06, + "loss": 1.0536, + "step": 1360 + }, + { + "epoch": 3.4808184143222505, + "grad_norm": 0.15197955681788783, + "learning_rate": 5.1190476914289645e-06, + "loss": 1.0357, + "step": 1361 + }, + { + "epoch": 3.483375959079284, + "grad_norm": 0.1666030618291994, + "learning_rate": 5.103467355702928e-06, + "loss": 1.0428, + "step": 1362 + }, + { + "epoch": 3.4859335038363173, + "grad_norm": 0.15565303487254922, + "learning_rate": 5.087902639098472e-06, + "loss": 1.0586, + "step": 1363 + }, + { + "epoch": 3.4884910485933505, + "grad_norm": 0.159004503149546, + "learning_rate": 5.072353591264429e-06, + "loss": 1.0271, + "step": 1364 + }, + { + "epoch": 3.4910485933503836, + "grad_norm": 0.19059651465127486, + "learning_rate": 5.056820261799668e-06, + "loss": 1.0668, + "step": 1365 + }, + { + "epoch": 3.493606138107417, + "grad_norm": 0.15760115088433216, + "learning_rate": 5.041302700252921e-06, + "loss": 1.0853, + "step": 1366 + }, + { + "epoch": 3.49616368286445, + "grad_norm": 0.1634157254255831, + "learning_rate": 5.02580095612262e-06, + "loss": 1.0664, + "step": 1367 + }, + { + "epoch": 3.498721227621483, + "grad_norm": 0.18070651331042822, + "learning_rate": 5.010315078856733e-06, + "loss": 1.0043, + "step": 1368 + }, + { + "epoch": 3.501278772378517, + "grad_norm": 0.1517363485498429, + "learning_rate": 4.99484511785263e-06, + "loss": 1.0535, + "step": 1369 + }, + { + "epoch": 3.50383631713555, + "grad_norm": 0.15908324436429314, + "learning_rate": 4.979391122456899e-06, + "loss": 1.0565, + "step": 1370 + }, + { + "epoch": 3.506393861892583, + "grad_norm": 0.16295413916211288, + "learning_rate": 4.963953141965208e-06, + "loss": 1.0531, + "step": 1371 + }, + { + "epoch": 3.5089514066496164, + "grad_norm": 0.15384339113595663, + "learning_rate": 4.94853122562213e-06, + "loss": 1.0707, + "step": 1372 + }, + { + "epoch": 3.5115089514066495, + "grad_norm": 0.1555405750624787, + "learning_rate": 4.9331254226210136e-06, + "loss": 1.0288, + "step": 1373 + }, + { + "epoch": 3.5140664961636827, + "grad_norm": 0.1574920923229238, + "learning_rate": 4.917735782103796e-06, + "loss": 1.0516, + "step": 1374 + }, + { + "epoch": 3.516624040920716, + "grad_norm": 0.1685616882842499, + "learning_rate": 4.902362353160851e-06, + "loss": 1.0232, + "step": 1375 + }, + { + "epoch": 3.5191815856777495, + "grad_norm": 0.14391567303608666, + "learning_rate": 4.887005184830861e-06, + "loss": 1.0296, + "step": 1376 + }, + { + "epoch": 3.5217391304347827, + "grad_norm": 0.14995128976992425, + "learning_rate": 4.871664326100625e-06, + "loss": 1.0571, + "step": 1377 + }, + { + "epoch": 3.524296675191816, + "grad_norm": 0.15919702291165838, + "learning_rate": 4.856339825904921e-06, + "loss": 1.0221, + "step": 1378 + }, + { + "epoch": 3.526854219948849, + "grad_norm": 0.15107002812900702, + "learning_rate": 4.841031733126346e-06, + "loss": 1.0872, + "step": 1379 + }, + { + "epoch": 3.5294117647058822, + "grad_norm": 0.14144172181733294, + "learning_rate": 4.825740096595159e-06, + "loss": 1.0272, + "step": 1380 + }, + { + "epoch": 3.531969309462916, + "grad_norm": 0.17206998247154082, + "learning_rate": 4.81046496508913e-06, + "loss": 1.0833, + "step": 1381 + }, + { + "epoch": 3.5345268542199486, + "grad_norm": 0.1387566110901943, + "learning_rate": 4.795206387333371e-06, + "loss": 1.0531, + "step": 1382 + }, + { + "epoch": 3.5370843989769822, + "grad_norm": 0.14298771692753376, + "learning_rate": 4.779964412000207e-06, + "loss": 1.0378, + "step": 1383 + }, + { + "epoch": 3.5396419437340154, + "grad_norm": 0.14816320170562203, + "learning_rate": 4.764739087708988e-06, + "loss": 1.0668, + "step": 1384 + }, + { + "epoch": 3.5421994884910486, + "grad_norm": 0.1476382719637328, + "learning_rate": 4.749530463025961e-06, + "loss": 1.0448, + "step": 1385 + }, + { + "epoch": 3.544757033248082, + "grad_norm": 0.14204568545092844, + "learning_rate": 4.734338586464096e-06, + "loss": 1.0647, + "step": 1386 + }, + { + "epoch": 3.547314578005115, + "grad_norm": 0.13917278007082326, + "learning_rate": 4.719163506482942e-06, + "loss": 1.0165, + "step": 1387 + }, + { + "epoch": 3.5498721227621486, + "grad_norm": 0.15529645602703832, + "learning_rate": 4.704005271488472e-06, + "loss": 1.0381, + "step": 1388 + }, + { + "epoch": 3.5524296675191813, + "grad_norm": 0.14796765919906726, + "learning_rate": 4.688863929832922e-06, + "loss": 1.0659, + "step": 1389 + }, + { + "epoch": 3.554987212276215, + "grad_norm": 0.14135671867370417, + "learning_rate": 4.673739529814653e-06, + "loss": 1.0741, + "step": 1390 + }, + { + "epoch": 3.557544757033248, + "grad_norm": 0.1387525451398238, + "learning_rate": 4.658632119677965e-06, + "loss": 1.0079, + "step": 1391 + }, + { + "epoch": 3.5601023017902813, + "grad_norm": 0.15154983652739518, + "learning_rate": 4.6435417476129745e-06, + "loss": 1.0333, + "step": 1392 + }, + { + "epoch": 3.5626598465473145, + "grad_norm": 0.15835829501275145, + "learning_rate": 4.628468461755456e-06, + "loss": 1.0531, + "step": 1393 + }, + { + "epoch": 3.5652173913043477, + "grad_norm": 0.14664150937490233, + "learning_rate": 4.613412310186669e-06, + "loss": 1.0541, + "step": 1394 + }, + { + "epoch": 3.5677749360613813, + "grad_norm": 0.15043461437274602, + "learning_rate": 4.598373340933226e-06, + "loss": 1.0608, + "step": 1395 + }, + { + "epoch": 3.5703324808184145, + "grad_norm": 0.1452832340949103, + "learning_rate": 4.583351601966928e-06, + "loss": 1.0385, + "step": 1396 + }, + { + "epoch": 3.5728900255754477, + "grad_norm": 0.13933252151544814, + "learning_rate": 4.568347141204611e-06, + "loss": 1.019, + "step": 1397 + }, + { + "epoch": 3.575447570332481, + "grad_norm": 0.1522016997768107, + "learning_rate": 4.553360006508003e-06, + "loss": 1.0592, + "step": 1398 + }, + { + "epoch": 3.578005115089514, + "grad_norm": 0.14375660032866683, + "learning_rate": 4.538390245683555e-06, + "loss": 1.0616, + "step": 1399 + }, + { + "epoch": 3.580562659846547, + "grad_norm": 0.15304841832573549, + "learning_rate": 4.523437906482313e-06, + "loss": 1.0209, + "step": 1400 + }, + { + "epoch": 3.5831202046035804, + "grad_norm": 0.16188117769084115, + "learning_rate": 4.508503036599743e-06, + "loss": 1.0308, + "step": 1401 + }, + { + "epoch": 3.585677749360614, + "grad_norm": 0.14884489221085204, + "learning_rate": 4.493585683675575e-06, + "loss": 1.0512, + "step": 1402 + }, + { + "epoch": 3.588235294117647, + "grad_norm": 0.13887978241681936, + "learning_rate": 4.478685895293685e-06, + "loss": 1.0459, + "step": 1403 + }, + { + "epoch": 3.5907928388746804, + "grad_norm": 0.16378567976403613, + "learning_rate": 4.463803718981905e-06, + "loss": 1.0521, + "step": 1404 + }, + { + "epoch": 3.5933503836317136, + "grad_norm": 0.151640894218935, + "learning_rate": 4.448939202211896e-06, + "loss": 1.0426, + "step": 1405 + }, + { + "epoch": 3.5959079283887467, + "grad_norm": 0.14699983592365332, + "learning_rate": 4.434092392398978e-06, + "loss": 1.0446, + "step": 1406 + }, + { + "epoch": 3.59846547314578, + "grad_norm": 0.15099144797792488, + "learning_rate": 4.419263336902006e-06, + "loss": 1.0536, + "step": 1407 + }, + { + "epoch": 3.601023017902813, + "grad_norm": 0.15653391024407642, + "learning_rate": 4.404452083023183e-06, + "loss": 1.056, + "step": 1408 + }, + { + "epoch": 3.6035805626598467, + "grad_norm": 0.16221666231166285, + "learning_rate": 4.389658678007933e-06, + "loss": 1.0326, + "step": 1409 + }, + { + "epoch": 3.60613810741688, + "grad_norm": 0.15454011824050923, + "learning_rate": 4.374883169044757e-06, + "loss": 1.0065, + "step": 1410 + }, + { + "epoch": 3.608695652173913, + "grad_norm": 0.14534327322247192, + "learning_rate": 4.3601256032650576e-06, + "loss": 1.0487, + "step": 1411 + }, + { + "epoch": 3.6112531969309463, + "grad_norm": 0.1417228468510155, + "learning_rate": 4.345386027743005e-06, + "loss": 1.067, + "step": 1412 + }, + { + "epoch": 3.6138107416879794, + "grad_norm": 0.1472202179946865, + "learning_rate": 4.330664489495386e-06, + "loss": 1.0233, + "step": 1413 + }, + { + "epoch": 3.6163682864450126, + "grad_norm": 0.152838033504761, + "learning_rate": 4.315961035481445e-06, + "loss": 1.0429, + "step": 1414 + }, + { + "epoch": 3.618925831202046, + "grad_norm": 0.14202501165868497, + "learning_rate": 4.30127571260275e-06, + "loss": 1.0221, + "step": 1415 + }, + { + "epoch": 3.6214833759590794, + "grad_norm": 0.15558682308936234, + "learning_rate": 4.286608567703025e-06, + "loss": 1.0454, + "step": 1416 + }, + { + "epoch": 3.6240409207161126, + "grad_norm": 0.15926194693025708, + "learning_rate": 4.271959647568017e-06, + "loss": 1.0208, + "step": 1417 + }, + { + "epoch": 3.626598465473146, + "grad_norm": 0.1563509007491471, + "learning_rate": 4.257328998925339e-06, + "loss": 1.0575, + "step": 1418 + }, + { + "epoch": 3.629156010230179, + "grad_norm": 0.14556455028498103, + "learning_rate": 4.242716668444304e-06, + "loss": 1.0486, + "step": 1419 + }, + { + "epoch": 3.631713554987212, + "grad_norm": 0.1870985092309066, + "learning_rate": 4.228122702735818e-06, + "loss": 1.0511, + "step": 1420 + }, + { + "epoch": 3.634271099744246, + "grad_norm": 0.1437574000454239, + "learning_rate": 4.213547148352193e-06, + "loss": 1.0664, + "step": 1421 + }, + { + "epoch": 3.6368286445012785, + "grad_norm": 0.1464730506258114, + "learning_rate": 4.198990051787012e-06, + "loss": 1.0185, + "step": 1422 + }, + { + "epoch": 3.639386189258312, + "grad_norm": 0.1480513669086599, + "learning_rate": 4.184451459474983e-06, + "loss": 1.0481, + "step": 1423 + }, + { + "epoch": 3.6419437340153453, + "grad_norm": 0.15929462236223393, + "learning_rate": 4.169931417791788e-06, + "loss": 1.0392, + "step": 1424 + }, + { + "epoch": 3.6445012787723785, + "grad_norm": 0.14283038176368648, + "learning_rate": 4.155429973053935e-06, + "loss": 1.0682, + "step": 1425 + }, + { + "epoch": 3.6470588235294117, + "grad_norm": 0.1587622512950815, + "learning_rate": 4.140947171518609e-06, + "loss": 1.0256, + "step": 1426 + }, + { + "epoch": 3.649616368286445, + "grad_norm": 0.1521708022719752, + "learning_rate": 4.126483059383534e-06, + "loss": 1.0385, + "step": 1427 + }, + { + "epoch": 3.6521739130434785, + "grad_norm": 0.15273099328180773, + "learning_rate": 4.112037682786811e-06, + "loss": 1.0318, + "step": 1428 + }, + { + "epoch": 3.6547314578005117, + "grad_norm": 0.15588241294914806, + "learning_rate": 4.097611087806779e-06, + "loss": 1.0216, + "step": 1429 + }, + { + "epoch": 3.657289002557545, + "grad_norm": 0.13906215524951357, + "learning_rate": 4.083203320461867e-06, + "loss": 1.0534, + "step": 1430 + }, + { + "epoch": 3.659846547314578, + "grad_norm": 0.13375858888458342, + "learning_rate": 4.068814426710447e-06, + "loss": 1.0466, + "step": 1431 + }, + { + "epoch": 3.662404092071611, + "grad_norm": 0.14715897307378298, + "learning_rate": 4.054444452450688e-06, + "loss": 1.0511, + "step": 1432 + }, + { + "epoch": 3.6649616368286444, + "grad_norm": 0.15947927453300095, + "learning_rate": 4.040093443520404e-06, + "loss": 1.0249, + "step": 1433 + }, + { + "epoch": 3.6675191815856776, + "grad_norm": 0.1387655661713122, + "learning_rate": 4.02576144569693e-06, + "loss": 1.0316, + "step": 1434 + }, + { + "epoch": 3.670076726342711, + "grad_norm": 0.14307072067989993, + "learning_rate": 4.011448504696933e-06, + "loss": 1.0269, + "step": 1435 + }, + { + "epoch": 3.6726342710997444, + "grad_norm": 0.15772626616588967, + "learning_rate": 3.997154666176306e-06, + "loss": 1.0582, + "step": 1436 + }, + { + "epoch": 3.6751918158567776, + "grad_norm": 0.14282292553538378, + "learning_rate": 3.982879975730015e-06, + "loss": 1.0323, + "step": 1437 + }, + { + "epoch": 3.6777493606138107, + "grad_norm": 0.1534255051225497, + "learning_rate": 3.968624478891935e-06, + "loss": 1.0925, + "step": 1438 + }, + { + "epoch": 3.680306905370844, + "grad_norm": 0.14321832602712206, + "learning_rate": 3.954388221134721e-06, + "loss": 1.0265, + "step": 1439 + }, + { + "epoch": 3.682864450127877, + "grad_norm": 0.144382148106992, + "learning_rate": 3.9401712478696584e-06, + "loss": 1.0263, + "step": 1440 + }, + { + "epoch": 3.6854219948849103, + "grad_norm": 0.15170436155022662, + "learning_rate": 3.9259736044465165e-06, + "loss": 1.0271, + "step": 1441 + }, + { + "epoch": 3.687979539641944, + "grad_norm": 0.14290531170670692, + "learning_rate": 3.91179533615341e-06, + "loss": 1.0233, + "step": 1442 + }, + { + "epoch": 3.690537084398977, + "grad_norm": 0.13889175521979244, + "learning_rate": 3.897636488216641e-06, + "loss": 1.0072, + "step": 1443 + }, + { + "epoch": 3.6930946291560103, + "grad_norm": 0.14619106107662108, + "learning_rate": 3.88349710580058e-06, + "loss": 1.0349, + "step": 1444 + }, + { + "epoch": 3.6956521739130435, + "grad_norm": 0.14842632075321688, + "learning_rate": 3.8693772340074945e-06, + "loss": 1.057, + "step": 1445 + }, + { + "epoch": 3.6982097186700766, + "grad_norm": 0.13344432430502018, + "learning_rate": 3.855276917877408e-06, + "loss": 1.0182, + "step": 1446 + }, + { + "epoch": 3.70076726342711, + "grad_norm": 0.13677433922377288, + "learning_rate": 3.8411962023879844e-06, + "loss": 1.0478, + "step": 1447 + }, + { + "epoch": 3.703324808184143, + "grad_norm": 0.1519820293465984, + "learning_rate": 3.827135132454351e-06, + "loss": 1.0695, + "step": 1448 + }, + { + "epoch": 3.7058823529411766, + "grad_norm": 0.1380667164839018, + "learning_rate": 3.8130937529289734e-06, + "loss": 1.0203, + "step": 1449 + }, + { + "epoch": 3.70843989769821, + "grad_norm": 0.1346633518238781, + "learning_rate": 3.7990721086015114e-06, + "loss": 1.0344, + "step": 1450 + }, + { + "epoch": 3.710997442455243, + "grad_norm": 0.14686955698606644, + "learning_rate": 3.78507024419867e-06, + "loss": 1.0135, + "step": 1451 + }, + { + "epoch": 3.713554987212276, + "grad_norm": 0.15250547269977222, + "learning_rate": 3.7710882043840513e-06, + "loss": 1.0443, + "step": 1452 + }, + { + "epoch": 3.7161125319693094, + "grad_norm": 0.14183857654080673, + "learning_rate": 3.7571260337580283e-06, + "loss": 1.0276, + "step": 1453 + }, + { + "epoch": 3.718670076726343, + "grad_norm": 0.14830544734060966, + "learning_rate": 3.743183776857602e-06, + "loss": 1.0716, + "step": 1454 + }, + { + "epoch": 3.7212276214833757, + "grad_norm": 0.1456400419210818, + "learning_rate": 3.7292614781562384e-06, + "loss": 1.026, + "step": 1455 + }, + { + "epoch": 3.7237851662404093, + "grad_norm": 0.15092175218127304, + "learning_rate": 3.715359182063748e-06, + "loss": 1.0232, + "step": 1456 + }, + { + "epoch": 3.7263427109974425, + "grad_norm": 0.1699845240726765, + "learning_rate": 3.7014769329261325e-06, + "loss": 1.068, + "step": 1457 + }, + { + "epoch": 3.7289002557544757, + "grad_norm": 0.15520588635807145, + "learning_rate": 3.68761477502545e-06, + "loss": 1.0256, + "step": 1458 + }, + { + "epoch": 3.731457800511509, + "grad_norm": 0.14562325712512364, + "learning_rate": 3.673772752579665e-06, + "loss": 1.0607, + "step": 1459 + }, + { + "epoch": 3.734015345268542, + "grad_norm": 0.15541923560195023, + "learning_rate": 3.6599509097425246e-06, + "loss": 1.0436, + "step": 1460 + }, + { + "epoch": 3.7365728900255757, + "grad_norm": 0.1607476965680906, + "learning_rate": 3.646149290603398e-06, + "loss": 1.0202, + "step": 1461 + }, + { + "epoch": 3.7391304347826084, + "grad_norm": 0.14491441420503393, + "learning_rate": 3.6323679391871447e-06, + "loss": 1.0263, + "step": 1462 + }, + { + "epoch": 3.741687979539642, + "grad_norm": 0.17277986367023201, + "learning_rate": 3.6186068994539746e-06, + "loss": 1.0356, + "step": 1463 + }, + { + "epoch": 3.7442455242966752, + "grad_norm": 0.16679682898908768, + "learning_rate": 3.6048662152993062e-06, + "loss": 1.0554, + "step": 1464 + }, + { + "epoch": 3.7468030690537084, + "grad_norm": 0.14727808773917042, + "learning_rate": 3.5911459305536302e-06, + "loss": 1.0428, + "step": 1465 + }, + { + "epoch": 3.7493606138107416, + "grad_norm": 0.161921935441449, + "learning_rate": 3.577446088982357e-06, + "loss": 1.0384, + "step": 1466 + }, + { + "epoch": 3.7519181585677748, + "grad_norm": 0.16271933317209, + "learning_rate": 3.563766734285704e-06, + "loss": 1.0403, + "step": 1467 + }, + { + "epoch": 3.7544757033248084, + "grad_norm": 0.16870928574249397, + "learning_rate": 3.550107910098526e-06, + "loss": 1.036, + "step": 1468 + }, + { + "epoch": 3.7570332480818416, + "grad_norm": 0.1553381789909288, + "learning_rate": 3.5364696599901837e-06, + "loss": 1.0601, + "step": 1469 + }, + { + "epoch": 3.7595907928388748, + "grad_norm": 0.14997588844332052, + "learning_rate": 3.5228520274644264e-06, + "loss": 1.016, + "step": 1470 + }, + { + "epoch": 3.762148337595908, + "grad_norm": 0.17329631324712091, + "learning_rate": 3.5092550559592243e-06, + "loss": 1.0493, + "step": 1471 + }, + { + "epoch": 3.764705882352941, + "grad_norm": 0.1569662893495077, + "learning_rate": 3.495678788846648e-06, + "loss": 1.0433, + "step": 1472 + }, + { + "epoch": 3.7672634271099743, + "grad_norm": 0.15139032350205295, + "learning_rate": 3.482123269432722e-06, + "loss": 1.0257, + "step": 1473 + }, + { + "epoch": 3.7698209718670075, + "grad_norm": 0.15007562912485958, + "learning_rate": 3.4685885409572896e-06, + "loss": 1.0329, + "step": 1474 + }, + { + "epoch": 3.772378516624041, + "grad_norm": 0.16102630257285133, + "learning_rate": 3.4550746465938754e-06, + "loss": 1.0491, + "step": 1475 + }, + { + "epoch": 3.7749360613810743, + "grad_norm": 0.16752120082373823, + "learning_rate": 3.441581629449542e-06, + "loss": 1.041, + "step": 1476 + }, + { + "epoch": 3.7774936061381075, + "grad_norm": 0.1424221688932342, + "learning_rate": 3.4281095325647682e-06, + "loss": 1.009, + "step": 1477 + }, + { + "epoch": 3.7800511508951407, + "grad_norm": 0.14586125645039055, + "learning_rate": 3.4146583989132897e-06, + "loss": 1.0347, + "step": 1478 + }, + { + "epoch": 3.782608695652174, + "grad_norm": 0.15136208820123082, + "learning_rate": 3.4012282714019784e-06, + "loss": 1.0205, + "step": 1479 + }, + { + "epoch": 3.785166240409207, + "grad_norm": 0.1502892464390208, + "learning_rate": 3.387819192870697e-06, + "loss": 1.0289, + "step": 1480 + }, + { + "epoch": 3.78772378516624, + "grad_norm": 0.1412412642364067, + "learning_rate": 3.374431206092168e-06, + "loss": 1.0305, + "step": 1481 + }, + { + "epoch": 3.790281329923274, + "grad_norm": 0.13814570417279723, + "learning_rate": 3.3610643537718347e-06, + "loss": 1.0485, + "step": 1482 + }, + { + "epoch": 3.792838874680307, + "grad_norm": 0.13640331630857278, + "learning_rate": 3.347718678547719e-06, + "loss": 1.1023, + "step": 1483 + }, + { + "epoch": 3.79539641943734, + "grad_norm": 0.14994268119891938, + "learning_rate": 3.334394222990307e-06, + "loss": 1.0457, + "step": 1484 + }, + { + "epoch": 3.7979539641943734, + "grad_norm": 0.15575803797953353, + "learning_rate": 3.321091029602378e-06, + "loss": 1.0816, + "step": 1485 + }, + { + "epoch": 3.8005115089514065, + "grad_norm": 0.13866906616126037, + "learning_rate": 3.3078091408188984e-06, + "loss": 1.0702, + "step": 1486 + }, + { + "epoch": 3.80306905370844, + "grad_norm": 0.12980796709609282, + "learning_rate": 3.29454859900688e-06, + "loss": 1.0553, + "step": 1487 + }, + { + "epoch": 3.805626598465473, + "grad_norm": 0.1383223434086824, + "learning_rate": 3.2813094464652362e-06, + "loss": 1.0366, + "step": 1488 + }, + { + "epoch": 3.8081841432225065, + "grad_norm": 0.1435484790801586, + "learning_rate": 3.2680917254246515e-06, + "loss": 1.0434, + "step": 1489 + }, + { + "epoch": 3.8107416879795397, + "grad_norm": 0.1429077430292127, + "learning_rate": 3.2548954780474485e-06, + "loss": 1.0465, + "step": 1490 + }, + { + "epoch": 3.813299232736573, + "grad_norm": 0.1426703166662691, + "learning_rate": 3.241720746427456e-06, + "loss": 1.0458, + "step": 1491 + }, + { + "epoch": 3.815856777493606, + "grad_norm": 0.14480062983873204, + "learning_rate": 3.228567572589865e-06, + "loss": 1.0407, + "step": 1492 + }, + { + "epoch": 3.8184143222506393, + "grad_norm": 0.1439275121559339, + "learning_rate": 3.215435998491102e-06, + "loss": 1.0487, + "step": 1493 + }, + { + "epoch": 3.820971867007673, + "grad_norm": 0.1559869804995835, + "learning_rate": 3.2023260660187018e-06, + "loss": 1.0253, + "step": 1494 + }, + { + "epoch": 3.8235294117647056, + "grad_norm": 0.13239520801102936, + "learning_rate": 3.189237816991161e-06, + "loss": 1.0735, + "step": 1495 + }, + { + "epoch": 3.8260869565217392, + "grad_norm": 0.13906138820650482, + "learning_rate": 3.176171293157798e-06, + "loss": 1.0434, + "step": 1496 + }, + { + "epoch": 3.8286445012787724, + "grad_norm": 0.14386997194374815, + "learning_rate": 3.163126536198653e-06, + "loss": 1.0568, + "step": 1497 + }, + { + "epoch": 3.8312020460358056, + "grad_norm": 0.13652465945815134, + "learning_rate": 3.150103587724318e-06, + "loss": 1.0244, + "step": 1498 + }, + { + "epoch": 3.833759590792839, + "grad_norm": 0.14696915009976544, + "learning_rate": 3.137102489275824e-06, + "loss": 1.0333, + "step": 1499 + }, + { + "epoch": 3.836317135549872, + "grad_norm": 0.14077378579589328, + "learning_rate": 3.124123282324503e-06, + "loss": 1.0618, + "step": 1500 + }, + { + "epoch": 3.8388746803069056, + "grad_norm": 0.13614318341580398, + "learning_rate": 3.1111660082718666e-06, + "loss": 1.0407, + "step": 1501 + }, + { + "epoch": 3.8414322250639388, + "grad_norm": 0.13328361531738175, + "learning_rate": 3.0982307084494457e-06, + "loss": 1.0503, + "step": 1502 + }, + { + "epoch": 3.843989769820972, + "grad_norm": 0.13500227612563093, + "learning_rate": 3.0853174241186867e-06, + "loss": 1.0147, + "step": 1503 + }, + { + "epoch": 3.846547314578005, + "grad_norm": 0.15319897471724173, + "learning_rate": 3.072426196470818e-06, + "loss": 1.0612, + "step": 1504 + }, + { + "epoch": 3.8491048593350383, + "grad_norm": 0.13414516138399563, + "learning_rate": 3.0595570666266996e-06, + "loss": 1.0613, + "step": 1505 + }, + { + "epoch": 3.8516624040920715, + "grad_norm": 0.13525360876497583, + "learning_rate": 3.0467100756367064e-06, + "loss": 1.0379, + "step": 1506 + }, + { + "epoch": 3.8542199488491047, + "grad_norm": 0.1429209705032719, + "learning_rate": 3.033885264480595e-06, + "loss": 1.0561, + "step": 1507 + }, + { + "epoch": 3.8567774936061383, + "grad_norm": 0.14899220326181728, + "learning_rate": 3.021082674067373e-06, + "loss": 1.0382, + "step": 1508 + }, + { + "epoch": 3.8593350383631715, + "grad_norm": 0.1390523791967841, + "learning_rate": 3.0083023452351634e-06, + "loss": 1.0667, + "step": 1509 + }, + { + "epoch": 3.8618925831202047, + "grad_norm": 0.139515567894694, + "learning_rate": 2.99554431875108e-06, + "loss": 1.0435, + "step": 1510 + }, + { + "epoch": 3.864450127877238, + "grad_norm": 0.14313433937785855, + "learning_rate": 2.982808635311104e-06, + "loss": 1.0179, + "step": 1511 + }, + { + "epoch": 3.867007672634271, + "grad_norm": 0.1308603519857462, + "learning_rate": 2.9700953355399385e-06, + "loss": 1.0038, + "step": 1512 + }, + { + "epoch": 3.869565217391304, + "grad_norm": 0.14459372229041947, + "learning_rate": 2.957404459990877e-06, + "loss": 1.0327, + "step": 1513 + }, + { + "epoch": 3.8721227621483374, + "grad_norm": 0.1418328836444824, + "learning_rate": 2.9447360491457034e-06, + "loss": 1.0408, + "step": 1514 + }, + { + "epoch": 3.874680306905371, + "grad_norm": 0.13508377583914918, + "learning_rate": 2.93209014341453e-06, + "loss": 1.0436, + "step": 1515 + }, + { + "epoch": 3.877237851662404, + "grad_norm": 0.13732362700671252, + "learning_rate": 2.9194667831356837e-06, + "loss": 1.0549, + "step": 1516 + }, + { + "epoch": 3.8797953964194374, + "grad_norm": 0.13395895878948533, + "learning_rate": 2.9068660085755775e-06, + "loss": 1.0778, + "step": 1517 + }, + { + "epoch": 3.8823529411764706, + "grad_norm": 0.13299070727246437, + "learning_rate": 2.8942878599285773e-06, + "loss": 1.0495, + "step": 1518 + }, + { + "epoch": 3.8849104859335037, + "grad_norm": 0.14423958774637266, + "learning_rate": 2.8817323773168783e-06, + "loss": 1.0681, + "step": 1519 + }, + { + "epoch": 3.887468030690537, + "grad_norm": 0.13441811948188004, + "learning_rate": 2.869199600790369e-06, + "loss": 1.0579, + "step": 1520 + }, + { + "epoch": 3.89002557544757, + "grad_norm": 0.15663102989889852, + "learning_rate": 2.8566895703265216e-06, + "loss": 1.0345, + "step": 1521 + }, + { + "epoch": 3.8925831202046037, + "grad_norm": 0.14208764333909965, + "learning_rate": 2.844202325830241e-06, + "loss": 1.0031, + "step": 1522 + }, + { + "epoch": 3.895140664961637, + "grad_norm": 0.14573616081300186, + "learning_rate": 2.8317379071337515e-06, + "loss": 1.0407, + "step": 1523 + }, + { + "epoch": 3.89769820971867, + "grad_norm": 0.15711176247504646, + "learning_rate": 2.819296353996468e-06, + "loss": 1.0353, + "step": 1524 + }, + { + "epoch": 3.9002557544757033, + "grad_norm": 0.13860891537498624, + "learning_rate": 2.806877706104867e-06, + "loss": 1.0584, + "step": 1525 + }, + { + "epoch": 3.9028132992327365, + "grad_norm": 0.14228002085549987, + "learning_rate": 2.794482003072364e-06, + "loss": 1.0449, + "step": 1526 + }, + { + "epoch": 3.90537084398977, + "grad_norm": 0.1375210322551571, + "learning_rate": 2.7821092844391762e-06, + "loss": 1.0357, + "step": 1527 + }, + { + "epoch": 3.907928388746803, + "grad_norm": 0.13889206339574903, + "learning_rate": 2.7697595896722206e-06, + "loss": 1.0499, + "step": 1528 + }, + { + "epoch": 3.9104859335038364, + "grad_norm": 0.15252492951738833, + "learning_rate": 2.757432958164953e-06, + "loss": 1.0142, + "step": 1529 + }, + { + "epoch": 3.9130434782608696, + "grad_norm": 0.13347562424578444, + "learning_rate": 2.745129429237269e-06, + "loss": 1.0436, + "step": 1530 + }, + { + "epoch": 3.915601023017903, + "grad_norm": 0.13176945767052098, + "learning_rate": 2.732849042135377e-06, + "loss": 1.0349, + "step": 1531 + }, + { + "epoch": 3.918158567774936, + "grad_norm": 0.14363585761020053, + "learning_rate": 2.72059183603166e-06, + "loss": 1.0487, + "step": 1532 + }, + { + "epoch": 3.920716112531969, + "grad_norm": 0.14363411928958503, + "learning_rate": 2.7083578500245567e-06, + "loss": 1.0467, + "step": 1533 + }, + { + "epoch": 3.923273657289003, + "grad_norm": 0.13936136441370217, + "learning_rate": 2.696147123138442e-06, + "loss": 1.0586, + "step": 1534 + }, + { + "epoch": 3.9258312020460355, + "grad_norm": 0.14655776610626794, + "learning_rate": 2.6839596943234948e-06, + "loss": 1.0468, + "step": 1535 + }, + { + "epoch": 3.928388746803069, + "grad_norm": 0.15051847712859898, + "learning_rate": 2.671795602455578e-06, + "loss": 1.0329, + "step": 1536 + }, + { + "epoch": 3.9309462915601023, + "grad_norm": 0.13505484364512485, + "learning_rate": 2.6596548863361117e-06, + "loss": 1.0311, + "step": 1537 + }, + { + "epoch": 3.9335038363171355, + "grad_norm": 0.16238346606896784, + "learning_rate": 2.647537584691957e-06, + "loss": 1.0497, + "step": 1538 + }, + { + "epoch": 3.9360613810741687, + "grad_norm": 0.14175763037908903, + "learning_rate": 2.635443736175285e-06, + "loss": 1.0308, + "step": 1539 + }, + { + "epoch": 3.938618925831202, + "grad_norm": 0.13596964536942902, + "learning_rate": 2.6233733793634443e-06, + "loss": 1.0419, + "step": 1540 + }, + { + "epoch": 3.9411764705882355, + "grad_norm": 0.153459320504519, + "learning_rate": 2.611326552758865e-06, + "loss": 1.0326, + "step": 1541 + }, + { + "epoch": 3.9437340153452687, + "grad_norm": 0.13805486780739087, + "learning_rate": 2.599303294788912e-06, + "loss": 1.0091, + "step": 1542 + }, + { + "epoch": 3.946291560102302, + "grad_norm": 0.15471538183539127, + "learning_rate": 2.5873036438057676e-06, + "loss": 1.0337, + "step": 1543 + }, + { + "epoch": 3.948849104859335, + "grad_norm": 0.13381926520485934, + "learning_rate": 2.5753276380863146e-06, + "loss": 1.0152, + "step": 1544 + }, + { + "epoch": 3.9514066496163682, + "grad_norm": 0.1526645485398575, + "learning_rate": 2.5633753158320187e-06, + "loss": 1.0528, + "step": 1545 + }, + { + "epoch": 3.9539641943734014, + "grad_norm": 0.1345563700018159, + "learning_rate": 2.551446715168785e-06, + "loss": 1.0234, + "step": 1546 + }, + { + "epoch": 3.9565217391304346, + "grad_norm": 0.14109365045502603, + "learning_rate": 2.539541874146857e-06, + "loss": 1.0536, + "step": 1547 + }, + { + "epoch": 3.959079283887468, + "grad_norm": 0.13079972583954438, + "learning_rate": 2.5276608307406947e-06, + "loss": 1.0312, + "step": 1548 + }, + { + "epoch": 3.9616368286445014, + "grad_norm": 0.14344519725256097, + "learning_rate": 2.5158036228488426e-06, + "loss": 1.0957, + "step": 1549 + }, + { + "epoch": 3.9641943734015346, + "grad_norm": 0.1373250552318116, + "learning_rate": 2.503970288293811e-06, + "loss": 1.0625, + "step": 1550 + }, + { + "epoch": 3.9667519181585678, + "grad_norm": 0.1272077247026377, + "learning_rate": 2.492160864821964e-06, + "loss": 1.0518, + "step": 1551 + }, + { + "epoch": 3.969309462915601, + "grad_norm": 0.14387184269881323, + "learning_rate": 2.4803753901033887e-06, + "loss": 1.0524, + "step": 1552 + }, + { + "epoch": 3.971867007672634, + "grad_norm": 0.14549905203239785, + "learning_rate": 2.4686139017317834e-06, + "loss": 1.051, + "step": 1553 + }, + { + "epoch": 3.9744245524296673, + "grad_norm": 0.1365288629952198, + "learning_rate": 2.456876437224327e-06, + "loss": 1.0372, + "step": 1554 + }, + { + "epoch": 3.976982097186701, + "grad_norm": 0.13426973718739119, + "learning_rate": 2.4451630340215806e-06, + "loss": 1.0601, + "step": 1555 + }, + { + "epoch": 3.979539641943734, + "grad_norm": 0.14101312603970495, + "learning_rate": 2.4334737294873412e-06, + "loss": 1.0738, + "step": 1556 + }, + { + "epoch": 3.9820971867007673, + "grad_norm": 0.13956525094149905, + "learning_rate": 2.4218085609085317e-06, + "loss": 1.041, + "step": 1557 + }, + { + "epoch": 3.9846547314578005, + "grad_norm": 0.137559865762302, + "learning_rate": 2.4101675654951007e-06, + "loss": 1.0552, + "step": 1558 + }, + { + "epoch": 3.9872122762148337, + "grad_norm": 0.1451836876243983, + "learning_rate": 2.398550780379877e-06, + "loss": 1.0343, + "step": 1559 + }, + { + "epoch": 3.9897698209718673, + "grad_norm": 0.13670823609416857, + "learning_rate": 2.3869582426184646e-06, + "loss": 1.0494, + "step": 1560 + }, + { + "epoch": 3.9923273657289, + "grad_norm": 0.1291970541444472, + "learning_rate": 2.375389989189124e-06, + "loss": 1.0489, + "step": 1561 + }, + { + "epoch": 3.9948849104859336, + "grad_norm": 0.13910494832994932, + "learning_rate": 2.3638460569926526e-06, + "loss": 1.0628, + "step": 1562 + }, + { + "epoch": 3.997442455242967, + "grad_norm": 0.14089850988803235, + "learning_rate": 2.3523264828522664e-06, + "loss": 1.0401, + "step": 1563 + }, + { + "epoch": 4.0, + "grad_norm": 0.13798191201750412, + "learning_rate": 2.34083130351348e-06, + "loss": 1.0511, + "step": 1564 + }, + { + "epoch": 4.002557544757034, + "grad_norm": 0.1460757819403398, + "learning_rate": 2.3293605556440036e-06, + "loss": 1.0239, + "step": 1565 + }, + { + "epoch": 4.005115089514066, + "grad_norm": 0.1388523372795974, + "learning_rate": 2.3179142758336025e-06, + "loss": 1.0368, + "step": 1566 + }, + { + "epoch": 4.0076726342711, + "grad_norm": 0.14222700155961276, + "learning_rate": 2.3064925005939985e-06, + "loss": 1.012, + "step": 1567 + }, + { + "epoch": 4.010230179028133, + "grad_norm": 0.13819903902418096, + "learning_rate": 2.29509526635875e-06, + "loss": 1.0293, + "step": 1568 + }, + { + "epoch": 4.012787723785166, + "grad_norm": 0.13498705400952052, + "learning_rate": 2.2837226094831277e-06, + "loss": 1.0415, + "step": 1569 + }, + { + "epoch": 4.015345268542199, + "grad_norm": 0.13346320464672834, + "learning_rate": 2.272374566244011e-06, + "loss": 1.0438, + "step": 1570 + }, + { + "epoch": 4.017902813299233, + "grad_norm": 0.13390383794077007, + "learning_rate": 2.261051172839759e-06, + "loss": 1.0397, + "step": 1571 + }, + { + "epoch": 4.020460358056266, + "grad_norm": 0.1370898771780535, + "learning_rate": 2.249752465390115e-06, + "loss": 1.0257, + "step": 1572 + }, + { + "epoch": 4.023017902813299, + "grad_norm": 0.13111044024506377, + "learning_rate": 2.2384784799360593e-06, + "loss": 1.0312, + "step": 1573 + }, + { + "epoch": 4.025575447570333, + "grad_norm": 0.13928353777385094, + "learning_rate": 2.2272292524397256e-06, + "loss": 1.024, + "step": 1574 + }, + { + "epoch": 4.028132992327365, + "grad_norm": 0.13970299818718698, + "learning_rate": 2.216004818784274e-06, + "loss": 1.0285, + "step": 1575 + }, + { + "epoch": 4.030690537084399, + "grad_norm": 0.12461772368830429, + "learning_rate": 2.204805214773774e-06, + "loss": 1.0418, + "step": 1576 + }, + { + "epoch": 4.033248081841432, + "grad_norm": 0.13051534501054377, + "learning_rate": 2.193630476133087e-06, + "loss": 1.0129, + "step": 1577 + }, + { + "epoch": 4.035805626598465, + "grad_norm": 0.15306722598206782, + "learning_rate": 2.182480638507775e-06, + "loss": 1.0421, + "step": 1578 + }, + { + "epoch": 4.038363171355499, + "grad_norm": 0.14455268449480663, + "learning_rate": 2.171355737463946e-06, + "loss": 1.0416, + "step": 1579 + }, + { + "epoch": 4.040920716112532, + "grad_norm": 0.13063652782059365, + "learning_rate": 2.16025580848818e-06, + "loss": 1.0612, + "step": 1580 + }, + { + "epoch": 4.043478260869565, + "grad_norm": 0.14360502497244268, + "learning_rate": 2.1491808869874007e-06, + "loss": 1.0341, + "step": 1581 + }, + { + "epoch": 4.046035805626598, + "grad_norm": 0.13201827533910473, + "learning_rate": 2.1381310082887562e-06, + "loss": 1.0201, + "step": 1582 + }, + { + "epoch": 4.048593350383632, + "grad_norm": 0.13153522938184467, + "learning_rate": 2.127106207639519e-06, + "loss": 1.0124, + "step": 1583 + }, + { + "epoch": 4.051150895140665, + "grad_norm": 0.13613076423821696, + "learning_rate": 2.116106520206952e-06, + "loss": 1.0319, + "step": 1584 + }, + { + "epoch": 4.053708439897698, + "grad_norm": 0.13691012590975757, + "learning_rate": 2.10513198107823e-06, + "loss": 1.0276, + "step": 1585 + }, + { + "epoch": 4.056265984654732, + "grad_norm": 0.12697375062593624, + "learning_rate": 2.0941826252602993e-06, + "loss": 1.0699, + "step": 1586 + }, + { + "epoch": 4.0588235294117645, + "grad_norm": 0.12622299253237756, + "learning_rate": 2.083258487679772e-06, + "loss": 0.9935, + "step": 1587 + }, + { + "epoch": 4.061381074168798, + "grad_norm": 0.12883881312565362, + "learning_rate": 2.07235960318283e-06, + "loss": 1.0199, + "step": 1588 + }, + { + "epoch": 4.063938618925831, + "grad_norm": 0.1433790683750576, + "learning_rate": 2.061486006535095e-06, + "loss": 1.0648, + "step": 1589 + }, + { + "epoch": 4.0664961636828645, + "grad_norm": 0.1334070559519316, + "learning_rate": 2.0506377324215153e-06, + "loss": 1.0271, + "step": 1590 + }, + { + "epoch": 4.069053708439898, + "grad_norm": 0.13107810691262306, + "learning_rate": 2.0398148154462826e-06, + "loss": 1.0664, + "step": 1591 + }, + { + "epoch": 4.071611253196931, + "grad_norm": 0.13064607210418142, + "learning_rate": 2.029017290132693e-06, + "loss": 1.0561, + "step": 1592 + }, + { + "epoch": 4.0741687979539645, + "grad_norm": 0.13192450653543714, + "learning_rate": 2.0182451909230493e-06, + "loss": 1.0647, + "step": 1593 + }, + { + "epoch": 4.076726342710997, + "grad_norm": 0.13409898938813986, + "learning_rate": 2.0074985521785496e-06, + "loss": 1.0706, + "step": 1594 + }, + { + "epoch": 4.079283887468031, + "grad_norm": 0.12098414017807911, + "learning_rate": 1.9967774081791757e-06, + "loss": 1.027, + "step": 1595 + }, + { + "epoch": 4.081841432225064, + "grad_norm": 0.12932021861152668, + "learning_rate": 1.9860817931235875e-06, + "loss": 1.0564, + "step": 1596 + }, + { + "epoch": 4.084398976982097, + "grad_norm": 0.13119851497814697, + "learning_rate": 1.9754117411290098e-06, + "loss": 1.04, + "step": 1597 + }, + { + "epoch": 4.086956521739131, + "grad_norm": 0.12723835341568177, + "learning_rate": 1.964767286231132e-06, + "loss": 1.0075, + "step": 1598 + }, + { + "epoch": 4.089514066496164, + "grad_norm": 0.1331440706566852, + "learning_rate": 1.9541484623839836e-06, + "loss": 1.0115, + "step": 1599 + }, + { + "epoch": 4.092071611253197, + "grad_norm": 0.13004369154123793, + "learning_rate": 1.94355530345984e-06, + "loss": 1.0446, + "step": 1600 + }, + { + "epoch": 4.09462915601023, + "grad_norm": 0.13439541656021142, + "learning_rate": 1.932987843249111e-06, + "loss": 1.0408, + "step": 1601 + }, + { + "epoch": 4.0971867007672635, + "grad_norm": 0.12617676641867243, + "learning_rate": 1.922446115460229e-06, + "loss": 1.0247, + "step": 1602 + }, + { + "epoch": 4.099744245524296, + "grad_norm": 0.12373854222060104, + "learning_rate": 1.9119301537195455e-06, + "loss": 1.0107, + "step": 1603 + }, + { + "epoch": 4.10230179028133, + "grad_norm": 0.12497624014354403, + "learning_rate": 1.9014399915712213e-06, + "loss": 1.0528, + "step": 1604 + }, + { + "epoch": 4.1048593350383635, + "grad_norm": 0.12786730473206026, + "learning_rate": 1.8909756624771281e-06, + "loss": 1.0375, + "step": 1605 + }, + { + "epoch": 4.107416879795396, + "grad_norm": 0.12984834445815097, + "learning_rate": 1.880537199816722e-06, + "loss": 1.0422, + "step": 1606 + }, + { + "epoch": 4.10997442455243, + "grad_norm": 0.1251770868811918, + "learning_rate": 1.8701246368869563e-06, + "loss": 1.0185, + "step": 1607 + }, + { + "epoch": 4.112531969309463, + "grad_norm": 0.1361892397850472, + "learning_rate": 1.859738006902172e-06, + "loss": 1.0249, + "step": 1608 + }, + { + "epoch": 4.115089514066496, + "grad_norm": 0.12533127135022198, + "learning_rate": 1.8493773429939822e-06, + "loss": 1.0459, + "step": 1609 + }, + { + "epoch": 4.117647058823529, + "grad_norm": 0.13786171148092025, + "learning_rate": 1.8390426782111758e-06, + "loss": 1.0397, + "step": 1610 + }, + { + "epoch": 4.120204603580563, + "grad_norm": 0.12675000210328083, + "learning_rate": 1.8287340455196068e-06, + "loss": 1.0306, + "step": 1611 + }, + { + "epoch": 4.122762148337596, + "grad_norm": 0.13402518451365739, + "learning_rate": 1.8184514778020935e-06, + "loss": 1.025, + "step": 1612 + }, + { + "epoch": 4.125319693094629, + "grad_norm": 0.1290845343245023, + "learning_rate": 1.80819500785831e-06, + "loss": 1.0507, + "step": 1613 + }, + { + "epoch": 4.127877237851663, + "grad_norm": 0.13511534437927844, + "learning_rate": 1.7979646684046782e-06, + "loss": 1.0378, + "step": 1614 + }, + { + "epoch": 4.130434782608695, + "grad_norm": 0.12639858921905558, + "learning_rate": 1.7877604920742809e-06, + "loss": 1.039, + "step": 1615 + }, + { + "epoch": 4.132992327365729, + "grad_norm": 0.12601178623729056, + "learning_rate": 1.7775825114167344e-06, + "loss": 1.0381, + "step": 1616 + }, + { + "epoch": 4.135549872122763, + "grad_norm": 0.12901338885710367, + "learning_rate": 1.767430758898092e-06, + "loss": 1.0339, + "step": 1617 + }, + { + "epoch": 4.138107416879795, + "grad_norm": 0.13798456507222667, + "learning_rate": 1.757305266900755e-06, + "loss": 1.0328, + "step": 1618 + }, + { + "epoch": 4.140664961636829, + "grad_norm": 0.12951208934796613, + "learning_rate": 1.7472060677233505e-06, + "loss": 1.0067, + "step": 1619 + }, + { + "epoch": 4.143222506393862, + "grad_norm": 0.133587178193347, + "learning_rate": 1.737133193580638e-06, + "loss": 1.0633, + "step": 1620 + }, + { + "epoch": 4.145780051150895, + "grad_norm": 0.14209051304513082, + "learning_rate": 1.7270866766034012e-06, + "loss": 1.045, + "step": 1621 + }, + { + "epoch": 4.148337595907928, + "grad_norm": 0.12839529485772747, + "learning_rate": 1.7170665488383598e-06, + "loss": 1.0421, + "step": 1622 + }, + { + "epoch": 4.150895140664962, + "grad_norm": 0.12930109704079273, + "learning_rate": 1.7070728422480397e-06, + "loss": 1.0576, + "step": 1623 + }, + { + "epoch": 4.153452685421995, + "grad_norm": 0.12789118434710597, + "learning_rate": 1.6971055887106979e-06, + "loss": 1.0206, + "step": 1624 + }, + { + "epoch": 4.156010230179028, + "grad_norm": 0.13625076201484126, + "learning_rate": 1.687164820020213e-06, + "loss": 1.0331, + "step": 1625 + }, + { + "epoch": 4.158567774936062, + "grad_norm": 0.12958940476816713, + "learning_rate": 1.677250567885974e-06, + "loss": 1.0466, + "step": 1626 + }, + { + "epoch": 4.161125319693094, + "grad_norm": 0.13211702039553985, + "learning_rate": 1.667362863932792e-06, + "loss": 1.0373, + "step": 1627 + }, + { + "epoch": 4.163682864450128, + "grad_norm": 0.13148536889339357, + "learning_rate": 1.6575017397007898e-06, + "loss": 1.0529, + "step": 1628 + }, + { + "epoch": 4.166240409207161, + "grad_norm": 0.14114126711572433, + "learning_rate": 1.6476672266453087e-06, + "loss": 1.0568, + "step": 1629 + }, + { + "epoch": 4.168797953964194, + "grad_norm": 0.13504256421222302, + "learning_rate": 1.6378593561368017e-06, + "loss": 1.0424, + "step": 1630 + }, + { + "epoch": 4.171355498721228, + "grad_norm": 0.12899095401998986, + "learning_rate": 1.6280781594607364e-06, + "loss": 1.0637, + "step": 1631 + }, + { + "epoch": 4.173913043478261, + "grad_norm": 0.1264131779917514, + "learning_rate": 1.6183236678175029e-06, + "loss": 1.0407, + "step": 1632 + }, + { + "epoch": 4.176470588235294, + "grad_norm": 0.13782869294761288, + "learning_rate": 1.6085959123222994e-06, + "loss": 1.036, + "step": 1633 + }, + { + "epoch": 4.179028132992327, + "grad_norm": 0.140891816042728, + "learning_rate": 1.5988949240050344e-06, + "loss": 1.0314, + "step": 1634 + }, + { + "epoch": 4.181585677749361, + "grad_norm": 0.12917479657781017, + "learning_rate": 1.5892207338102494e-06, + "loss": 1.0493, + "step": 1635 + }, + { + "epoch": 4.1841432225063935, + "grad_norm": 0.1334411939557938, + "learning_rate": 1.5795733725969931e-06, + "loss": 1.0836, + "step": 1636 + }, + { + "epoch": 4.186700767263427, + "grad_norm": 0.142993534386663, + "learning_rate": 1.5699528711387358e-06, + "loss": 1.0341, + "step": 1637 + }, + { + "epoch": 4.189258312020461, + "grad_norm": 0.12912047709199104, + "learning_rate": 1.560359260123272e-06, + "loss": 1.0502, + "step": 1638 + }, + { + "epoch": 4.1918158567774935, + "grad_norm": 0.12884960069552234, + "learning_rate": 1.5507925701526184e-06, + "loss": 1.0323, + "step": 1639 + }, + { + "epoch": 4.194373401534527, + "grad_norm": 0.1289892440406385, + "learning_rate": 1.5412528317429197e-06, + "loss": 1.0248, + "step": 1640 + }, + { + "epoch": 4.19693094629156, + "grad_norm": 0.12221838744224985, + "learning_rate": 1.5317400753243449e-06, + "loss": 1.004, + "step": 1641 + }, + { + "epoch": 4.1994884910485935, + "grad_norm": 0.13486097471794437, + "learning_rate": 1.5222543312410042e-06, + "loss": 1.0532, + "step": 1642 + }, + { + "epoch": 4.202046035805626, + "grad_norm": 0.12829008561367278, + "learning_rate": 1.512795629750834e-06, + "loss": 1.009, + "step": 1643 + }, + { + "epoch": 4.20460358056266, + "grad_norm": 0.13061127802497013, + "learning_rate": 1.5033640010255146e-06, + "loss": 1.0518, + "step": 1644 + }, + { + "epoch": 4.207161125319693, + "grad_norm": 0.1357231506544221, + "learning_rate": 1.493959475150365e-06, + "loss": 1.0391, + "step": 1645 + }, + { + "epoch": 4.209718670076726, + "grad_norm": 0.12492335548008021, + "learning_rate": 1.484582082124254e-06, + "loss": 1.0183, + "step": 1646 + }, + { + "epoch": 4.21227621483376, + "grad_norm": 0.13080607014236745, + "learning_rate": 1.4752318518594988e-06, + "loss": 1.0214, + "step": 1647 + }, + { + "epoch": 4.2148337595907925, + "grad_norm": 0.13032141745435707, + "learning_rate": 1.4659088141817701e-06, + "loss": 1.0326, + "step": 1648 + }, + { + "epoch": 4.217391304347826, + "grad_norm": 0.1410760353481451, + "learning_rate": 1.4566129988300094e-06, + "loss": 1.0167, + "step": 1649 + }, + { + "epoch": 4.21994884910486, + "grad_norm": 0.12878198988086415, + "learning_rate": 1.4473444354563081e-06, + "loss": 1.0201, + "step": 1650 + }, + { + "epoch": 4.2225063938618925, + "grad_norm": 0.13730461890440382, + "learning_rate": 1.438103153625835e-06, + "loss": 1.0403, + "step": 1651 + }, + { + "epoch": 4.225063938618926, + "grad_norm": 0.1283690268065539, + "learning_rate": 1.4288891828167428e-06, + "loss": 1.0296, + "step": 1652 + }, + { + "epoch": 4.227621483375959, + "grad_norm": 0.13230399753509858, + "learning_rate": 1.4197025524200547e-06, + "loss": 1.0436, + "step": 1653 + }, + { + "epoch": 4.2301790281329925, + "grad_norm": 0.13610939627100485, + "learning_rate": 1.4105432917395912e-06, + "loss": 1.036, + "step": 1654 + }, + { + "epoch": 4.232736572890025, + "grad_norm": 0.1411030873308251, + "learning_rate": 1.4014114299918613e-06, + "loss": 1.0626, + "step": 1655 + }, + { + "epoch": 4.235294117647059, + "grad_norm": 0.1360722157833826, + "learning_rate": 1.3923069963059821e-06, + "loss": 1.037, + "step": 1656 + }, + { + "epoch": 4.2378516624040925, + "grad_norm": 0.13086138342111206, + "learning_rate": 1.3832300197235748e-06, + "loss": 1.0224, + "step": 1657 + }, + { + "epoch": 4.240409207161125, + "grad_norm": 0.13316912474292783, + "learning_rate": 1.3741805291986787e-06, + "loss": 1.0588, + "step": 1658 + }, + { + "epoch": 4.242966751918159, + "grad_norm": 0.12776448396795964, + "learning_rate": 1.3651585535976596e-06, + "loss": 1.0416, + "step": 1659 + }, + { + "epoch": 4.245524296675192, + "grad_norm": 0.13394602718641618, + "learning_rate": 1.3561641216991162e-06, + "loss": 1.0464, + "step": 1660 + }, + { + "epoch": 4.248081841432225, + "grad_norm": 0.12411528913176928, + "learning_rate": 1.3471972621937757e-06, + "loss": 1.0523, + "step": 1661 + }, + { + "epoch": 4.250639386189258, + "grad_norm": 0.13483004817710278, + "learning_rate": 1.3382580036844295e-06, + "loss": 1.0461, + "step": 1662 + }, + { + "epoch": 4.253196930946292, + "grad_norm": 0.13250722022034925, + "learning_rate": 1.3293463746858182e-06, + "loss": 1.027, + "step": 1663 + }, + { + "epoch": 4.255754475703325, + "grad_norm": 0.13098173123869175, + "learning_rate": 1.3204624036245506e-06, + "loss": 1.0253, + "step": 1664 + }, + { + "epoch": 4.258312020460358, + "grad_norm": 0.13628427441357657, + "learning_rate": 1.3116061188390084e-06, + "loss": 1.0299, + "step": 1665 + }, + { + "epoch": 4.260869565217392, + "grad_norm": 0.12506601980421145, + "learning_rate": 1.3027775485792682e-06, + "loss": 1.0309, + "step": 1666 + }, + { + "epoch": 4.263427109974424, + "grad_norm": 0.12809650463562688, + "learning_rate": 1.2939767210069876e-06, + "loss": 1.0433, + "step": 1667 + }, + { + "epoch": 4.265984654731458, + "grad_norm": 0.12820756645898626, + "learning_rate": 1.285203664195338e-06, + "loss": 0.9889, + "step": 1668 + }, + { + "epoch": 4.268542199488491, + "grad_norm": 0.13676696388260404, + "learning_rate": 1.2764584061289098e-06, + "loss": 1.0339, + "step": 1669 + }, + { + "epoch": 4.271099744245524, + "grad_norm": 0.13840454125887916, + "learning_rate": 1.267740974703614e-06, + "loss": 1.0461, + "step": 1670 + }, + { + "epoch": 4.273657289002558, + "grad_norm": 0.1154351492718522, + "learning_rate": 1.2590513977266007e-06, + "loss": 1.0214, + "step": 1671 + }, + { + "epoch": 4.276214833759591, + "grad_norm": 0.13166991117354376, + "learning_rate": 1.2503897029161716e-06, + "loss": 1.0329, + "step": 1672 + }, + { + "epoch": 4.278772378516624, + "grad_norm": 0.1321979634403043, + "learning_rate": 1.2417559179016835e-06, + "loss": 0.9865, + "step": 1673 + }, + { + "epoch": 4.281329923273657, + "grad_norm": 0.13877049369386268, + "learning_rate": 1.2331500702234723e-06, + "loss": 1.0293, + "step": 1674 + }, + { + "epoch": 4.283887468030691, + "grad_norm": 0.12752281245986202, + "learning_rate": 1.2245721873327521e-06, + "loss": 1.0458, + "step": 1675 + }, + { + "epoch": 4.286445012787723, + "grad_norm": 0.12343044689847779, + "learning_rate": 1.2160222965915404e-06, + "loss": 1.0281, + "step": 1676 + }, + { + "epoch": 4.289002557544757, + "grad_norm": 0.12611866524321558, + "learning_rate": 1.207500425272562e-06, + "loss": 1.0467, + "step": 1677 + }, + { + "epoch": 4.291560102301791, + "grad_norm": 0.12003173247581543, + "learning_rate": 1.199006600559156e-06, + "loss": 1.0456, + "step": 1678 + }, + { + "epoch": 4.294117647058823, + "grad_norm": 0.1307538379654529, + "learning_rate": 1.190540849545213e-06, + "loss": 1.0326, + "step": 1679 + }, + { + "epoch": 4.296675191815857, + "grad_norm": 0.12722242311205958, + "learning_rate": 1.1821031992350628e-06, + "loss": 1.0506, + "step": 1680 + }, + { + "epoch": 4.29923273657289, + "grad_norm": 0.12736209379546135, + "learning_rate": 1.1736936765434005e-06, + "loss": 1.0218, + "step": 1681 + }, + { + "epoch": 4.301790281329923, + "grad_norm": 0.12638559752999246, + "learning_rate": 1.1653123082951968e-06, + "loss": 1.0717, + "step": 1682 + }, + { + "epoch": 4.304347826086957, + "grad_norm": 0.1383277702988731, + "learning_rate": 1.1569591212256237e-06, + "loss": 1.0487, + "step": 1683 + }, + { + "epoch": 4.30690537084399, + "grad_norm": 0.12792509279371292, + "learning_rate": 1.1486341419799474e-06, + "loss": 1.0735, + "step": 1684 + }, + { + "epoch": 4.309462915601023, + "grad_norm": 0.11817435751394731, + "learning_rate": 1.1403373971134623e-06, + "loss": 1.0211, + "step": 1685 + }, + { + "epoch": 4.312020460358056, + "grad_norm": 0.12075851868919982, + "learning_rate": 1.132068913091402e-06, + "loss": 1.0547, + "step": 1686 + }, + { + "epoch": 4.31457800511509, + "grad_norm": 0.12419038252191225, + "learning_rate": 1.1238287162888483e-06, + "loss": 1.0141, + "step": 1687 + }, + { + "epoch": 4.3171355498721224, + "grad_norm": 0.12503779001905366, + "learning_rate": 1.1156168329906536e-06, + "loss": 1.0115, + "step": 1688 + }, + { + "epoch": 4.319693094629156, + "grad_norm": 0.12357405258717082, + "learning_rate": 1.1074332893913541e-06, + "loss": 1.0183, + "step": 1689 + }, + { + "epoch": 4.322250639386189, + "grad_norm": 0.12336501389642063, + "learning_rate": 1.0992781115950867e-06, + "loss": 1.0817, + "step": 1690 + }, + { + "epoch": 4.324808184143222, + "grad_norm": 0.1237037130184548, + "learning_rate": 1.0911513256155094e-06, + "loss": 1.0305, + "step": 1691 + }, + { + "epoch": 4.327365728900256, + "grad_norm": 0.12122352134815065, + "learning_rate": 1.0830529573757076e-06, + "loss": 1.0271, + "step": 1692 + }, + { + "epoch": 4.329923273657289, + "grad_norm": 0.125450807160178, + "learning_rate": 1.074983032708129e-06, + "loss": 1.0426, + "step": 1693 + }, + { + "epoch": 4.332480818414322, + "grad_norm": 0.1331817263523311, + "learning_rate": 1.0669415773544867e-06, + "loss": 1.0339, + "step": 1694 + }, + { + "epoch": 4.335038363171355, + "grad_norm": 0.12379967651287854, + "learning_rate": 1.0589286169656743e-06, + "loss": 1.021, + "step": 1695 + }, + { + "epoch": 4.337595907928389, + "grad_norm": 0.1256760216863504, + "learning_rate": 1.0509441771017026e-06, + "loss": 1.063, + "step": 1696 + }, + { + "epoch": 4.340153452685422, + "grad_norm": 0.12701844597546846, + "learning_rate": 1.0429882832316008e-06, + "loss": 1.0279, + "step": 1697 + }, + { + "epoch": 4.342710997442455, + "grad_norm": 0.11989925917845673, + "learning_rate": 1.0350609607333384e-06, + "loss": 1.0053, + "step": 1698 + }, + { + "epoch": 4.345268542199489, + "grad_norm": 0.12375987829879317, + "learning_rate": 1.0271622348937581e-06, + "loss": 1.0438, + "step": 1699 + }, + { + "epoch": 4.3478260869565215, + "grad_norm": 0.1270476044771331, + "learning_rate": 1.0192921309084703e-06, + "loss": 1.0407, + "step": 1700 + }, + { + "epoch": 4.350383631713555, + "grad_norm": 0.12555015007408019, + "learning_rate": 1.0114506738817942e-06, + "loss": 1.0031, + "step": 1701 + }, + { + "epoch": 4.352941176470588, + "grad_norm": 0.12622543787693985, + "learning_rate": 1.0036378888266663e-06, + "loss": 1.0547, + "step": 1702 + }, + { + "epoch": 4.3554987212276215, + "grad_norm": 0.12762154218606617, + "learning_rate": 9.958538006645702e-07, + "loss": 1.0687, + "step": 1703 + }, + { + "epoch": 4.358056265984655, + "grad_norm": 0.12475219304030633, + "learning_rate": 9.880984342254462e-07, + "loss": 1.021, + "step": 1704 + }, + { + "epoch": 4.360613810741688, + "grad_norm": 0.1298726344994412, + "learning_rate": 9.803718142476182e-07, + "loss": 1.0392, + "step": 1705 + }, + { + "epoch": 4.3631713554987215, + "grad_norm": 0.12364003006997323, + "learning_rate": 9.72673965377714e-07, + "loss": 1.0513, + "step": 1706 + }, + { + "epoch": 4.365728900255754, + "grad_norm": 0.1285635408520734, + "learning_rate": 9.650049121705851e-07, + "loss": 1.0409, + "step": 1707 + }, + { + "epoch": 4.368286445012788, + "grad_norm": 0.1258885386792609, + "learning_rate": 9.573646790892299e-07, + "loss": 1.027, + "step": 1708 + }, + { + "epoch": 4.370843989769821, + "grad_norm": 0.1187303423804672, + "learning_rate": 9.497532905047202e-07, + "loss": 1.0534, + "step": 1709 + }, + { + "epoch": 4.373401534526854, + "grad_norm": 0.12469445611611187, + "learning_rate": 9.421707706961136e-07, + "loss": 1.0401, + "step": 1710 + }, + { + "epoch": 4.375959079283888, + "grad_norm": 0.12581258682849217, + "learning_rate": 9.346171438503781e-07, + "loss": 1.0642, + "step": 1711 + }, + { + "epoch": 4.378516624040921, + "grad_norm": 0.12336801659179744, + "learning_rate": 9.270924340623267e-07, + "loss": 1.0648, + "step": 1712 + }, + { + "epoch": 4.381074168797954, + "grad_norm": 0.12013777856733562, + "learning_rate": 9.195966653345256e-07, + "loss": 1.0287, + "step": 1713 + }, + { + "epoch": 4.383631713554987, + "grad_norm": 0.1240880814390023, + "learning_rate": 9.121298615772256e-07, + "loss": 1.0504, + "step": 1714 + }, + { + "epoch": 4.3861892583120206, + "grad_norm": 0.11807776342242135, + "learning_rate": 9.046920466082809e-07, + "loss": 1.0471, + "step": 1715 + }, + { + "epoch": 4.388746803069053, + "grad_norm": 0.11769299130070347, + "learning_rate": 8.972832441530876e-07, + "loss": 1.0341, + "step": 1716 + }, + { + "epoch": 4.391304347826087, + "grad_norm": 0.12193499241391831, + "learning_rate": 8.899034778444804e-07, + "loss": 1.0735, + "step": 1717 + }, + { + "epoch": 4.3938618925831205, + "grad_norm": 0.12571448070016797, + "learning_rate": 8.825527712226834e-07, + "loss": 1.0145, + "step": 1718 + }, + { + "epoch": 4.396419437340153, + "grad_norm": 0.11812287728798011, + "learning_rate": 8.75231147735226e-07, + "loss": 1.0255, + "step": 1719 + }, + { + "epoch": 4.398976982097187, + "grad_norm": 0.11941179582628143, + "learning_rate": 8.679386307368631e-07, + "loss": 1.0486, + "step": 1720 + }, + { + "epoch": 4.40153452685422, + "grad_norm": 0.1243184329514753, + "learning_rate": 8.60675243489506e-07, + "loss": 1.0663, + "step": 1721 + }, + { + "epoch": 4.404092071611253, + "grad_norm": 0.11711236888037205, + "learning_rate": 8.534410091621481e-07, + "loss": 1.0517, + "step": 1722 + }, + { + "epoch": 4.406649616368286, + "grad_norm": 0.12398332201899291, + "learning_rate": 8.462359508307882e-07, + "loss": 1.0563, + "step": 1723 + }, + { + "epoch": 4.40920716112532, + "grad_norm": 0.12262204617331557, + "learning_rate": 8.390600914783597e-07, + "loss": 1.0228, + "step": 1724 + }, + { + "epoch": 4.411764705882353, + "grad_norm": 0.1182059276608206, + "learning_rate": 8.319134539946549e-07, + "loss": 1.0391, + "step": 1725 + }, + { + "epoch": 4.414322250639386, + "grad_norm": 0.12345372785960933, + "learning_rate": 8.247960611762563e-07, + "loss": 1.0388, + "step": 1726 + }, + { + "epoch": 4.41687979539642, + "grad_norm": 0.12839333147920715, + "learning_rate": 8.177079357264583e-07, + "loss": 1.0282, + "step": 1727 + }, + { + "epoch": 4.419437340153452, + "grad_norm": 0.12729581387520422, + "learning_rate": 8.106491002551941e-07, + "loss": 1.0262, + "step": 1728 + }, + { + "epoch": 4.421994884910486, + "grad_norm": 0.12208859639468621, + "learning_rate": 8.036195772789735e-07, + "loss": 1.0334, + "step": 1729 + }, + { + "epoch": 4.42455242966752, + "grad_norm": 0.12356081440190549, + "learning_rate": 7.966193892208007e-07, + "loss": 1.0738, + "step": 1730 + }, + { + "epoch": 4.427109974424552, + "grad_norm": 0.1179560690186426, + "learning_rate": 7.896485584101066e-07, + "loss": 1.0214, + "step": 1731 + }, + { + "epoch": 4.429667519181586, + "grad_norm": 0.12396501408128464, + "learning_rate": 7.827071070826775e-07, + "loss": 1.0101, + "step": 1732 + }, + { + "epoch": 4.432225063938619, + "grad_norm": 0.120435484140843, + "learning_rate": 7.757950573805839e-07, + "loss": 1.0448, + "step": 1733 + }, + { + "epoch": 4.434782608695652, + "grad_norm": 0.1288542576587824, + "learning_rate": 7.689124313521112e-07, + "loss": 1.028, + "step": 1734 + }, + { + "epoch": 4.437340153452685, + "grad_norm": 0.11947326921621014, + "learning_rate": 7.620592509516845e-07, + "loss": 1.0508, + "step": 1735 + }, + { + "epoch": 4.439897698209719, + "grad_norm": 0.12201218897189074, + "learning_rate": 7.552355380398079e-07, + "loss": 1.0709, + "step": 1736 + }, + { + "epoch": 4.442455242966752, + "grad_norm": 0.12421373397429394, + "learning_rate": 7.484413143829861e-07, + "loss": 1.0328, + "step": 1737 + }, + { + "epoch": 4.445012787723785, + "grad_norm": 0.12410163367364695, + "learning_rate": 7.41676601653657e-07, + "loss": 1.034, + "step": 1738 + }, + { + "epoch": 4.447570332480819, + "grad_norm": 0.1219420965833109, + "learning_rate": 7.349414214301242e-07, + "loss": 1.0525, + "step": 1739 + }, + { + "epoch": 4.450127877237851, + "grad_norm": 0.11735248664674708, + "learning_rate": 7.282357951964903e-07, + "loss": 1.006, + "step": 1740 + }, + { + "epoch": 4.452685421994885, + "grad_norm": 0.11694083667726769, + "learning_rate": 7.215597443425815e-07, + "loss": 1.0249, + "step": 1741 + }, + { + "epoch": 4.455242966751918, + "grad_norm": 0.11713103311193784, + "learning_rate": 7.149132901638844e-07, + "loss": 1.0712, + "step": 1742 + }, + { + "epoch": 4.457800511508951, + "grad_norm": 0.11740763151594824, + "learning_rate": 7.082964538614823e-07, + "loss": 1.0242, + "step": 1743 + }, + { + "epoch": 4.460358056265985, + "grad_norm": 0.11802743122673677, + "learning_rate": 7.017092565419748e-07, + "loss": 1.0311, + "step": 1744 + }, + { + "epoch": 4.462915601023018, + "grad_norm": 0.11497098419518056, + "learning_rate": 6.951517192174195e-07, + "loss": 1.041, + "step": 1745 + }, + { + "epoch": 4.465473145780051, + "grad_norm": 0.11725056660448066, + "learning_rate": 6.886238628052711e-07, + "loss": 1.0462, + "step": 1746 + }, + { + "epoch": 4.468030690537084, + "grad_norm": 0.11844411179617265, + "learning_rate": 6.821257081282973e-07, + "loss": 1.0311, + "step": 1747 + }, + { + "epoch": 4.470588235294118, + "grad_norm": 0.11969749025311724, + "learning_rate": 6.756572759145285e-07, + "loss": 1.0362, + "step": 1748 + }, + { + "epoch": 4.4731457800511505, + "grad_norm": 0.12100274315376988, + "learning_rate": 6.692185867971834e-07, + "loss": 1.0398, + "step": 1749 + }, + { + "epoch": 4.475703324808184, + "grad_norm": 0.1189550665367411, + "learning_rate": 6.62809661314604e-07, + "loss": 1.0647, + "step": 1750 + }, + { + "epoch": 4.478260869565218, + "grad_norm": 0.12364355910534239, + "learning_rate": 6.564305199101939e-07, + "loss": 1.0287, + "step": 1751 + }, + { + "epoch": 4.4808184143222505, + "grad_norm": 0.11726566582294938, + "learning_rate": 6.500811829323462e-07, + "loss": 1.0455, + "step": 1752 + }, + { + "epoch": 4.483375959079284, + "grad_norm": 0.11446881967048579, + "learning_rate": 6.437616706343896e-07, + "loss": 1.0569, + "step": 1753 + }, + { + "epoch": 4.485933503836317, + "grad_norm": 0.12374756246454661, + "learning_rate": 6.374720031745129e-07, + "loss": 1.058, + "step": 1754 + }, + { + "epoch": 4.4884910485933505, + "grad_norm": 0.1233951395541208, + "learning_rate": 6.312122006157006e-07, + "loss": 1.0319, + "step": 1755 + }, + { + "epoch": 4.491048593350383, + "grad_norm": 0.11976730576289767, + "learning_rate": 6.249822829256835e-07, + "loss": 1.0455, + "step": 1756 + }, + { + "epoch": 4.493606138107417, + "grad_norm": 0.11824907431040846, + "learning_rate": 6.187822699768553e-07, + "loss": 1.0464, + "step": 1757 + }, + { + "epoch": 4.4961636828644505, + "grad_norm": 0.11525969916785428, + "learning_rate": 6.126121815462228e-07, + "loss": 1.0205, + "step": 1758 + }, + { + "epoch": 4.498721227621483, + "grad_norm": 0.13175731720889955, + "learning_rate": 6.064720373153365e-07, + "loss": 1.0062, + "step": 1759 + }, + { + "epoch": 4.501278772378517, + "grad_norm": 0.12453743339788283, + "learning_rate": 6.003618568702352e-07, + "loss": 1.0587, + "step": 1760 + }, + { + "epoch": 4.5038363171355495, + "grad_norm": 0.11732654328622537, + "learning_rate": 5.942816597013712e-07, + "loss": 1.0516, + "step": 1761 + }, + { + "epoch": 4.506393861892583, + "grad_norm": 0.12027453268886372, + "learning_rate": 5.882314652035582e-07, + "loss": 1.0645, + "step": 1762 + }, + { + "epoch": 4.508951406649617, + "grad_norm": 0.1233241454074132, + "learning_rate": 5.822112926759072e-07, + "loss": 1.0491, + "step": 1763 + }, + { + "epoch": 4.5115089514066495, + "grad_norm": 0.11543180230907084, + "learning_rate": 5.762211613217649e-07, + "loss": 1.0235, + "step": 1764 + }, + { + "epoch": 4.514066496163683, + "grad_norm": 0.1161090396637108, + "learning_rate": 5.702610902486472e-07, + "loss": 1.0396, + "step": 1765 + }, + { + "epoch": 4.516624040920716, + "grad_norm": 0.11434106483636258, + "learning_rate": 5.643310984681882e-07, + "loss": 1.0253, + "step": 1766 + }, + { + "epoch": 4.5191815856777495, + "grad_norm": 0.12772671626242307, + "learning_rate": 5.584312048960705e-07, + "loss": 1.0563, + "step": 1767 + }, + { + "epoch": 4.521739130434782, + "grad_norm": 0.12357938851666896, + "learning_rate": 5.525614283519698e-07, + "loss": 1.054, + "step": 1768 + }, + { + "epoch": 4.524296675191816, + "grad_norm": 0.12191915599526636, + "learning_rate": 5.46721787559491e-07, + "loss": 1.0103, + "step": 1769 + }, + { + "epoch": 4.526854219948849, + "grad_norm": 0.11728270071784068, + "learning_rate": 5.40912301146116e-07, + "loss": 1.0371, + "step": 1770 + }, + { + "epoch": 4.529411764705882, + "grad_norm": 0.1171100966224759, + "learning_rate": 5.351329876431377e-07, + "loss": 1.031, + "step": 1771 + }, + { + "epoch": 4.531969309462916, + "grad_norm": 0.1205942913894916, + "learning_rate": 5.293838654855965e-07, + "loss": 1.0368, + "step": 1772 + }, + { + "epoch": 4.534526854219949, + "grad_norm": 0.11538796588212798, + "learning_rate": 5.236649530122362e-07, + "loss": 1.0234, + "step": 1773 + }, + { + "epoch": 4.537084398976982, + "grad_norm": 0.11552252830049363, + "learning_rate": 5.17976268465431e-07, + "loss": 1.0268, + "step": 1774 + }, + { + "epoch": 4.539641943734015, + "grad_norm": 0.12030123767608153, + "learning_rate": 5.123178299911357e-07, + "loss": 1.0161, + "step": 1775 + }, + { + "epoch": 4.542199488491049, + "grad_norm": 0.11733012254657782, + "learning_rate": 5.066896556388223e-07, + "loss": 1.0263, + "step": 1776 + }, + { + "epoch": 4.544757033248082, + "grad_norm": 0.12261186958150777, + "learning_rate": 5.010917633614298e-07, + "loss": 1.0433, + "step": 1777 + }, + { + "epoch": 4.547314578005115, + "grad_norm": 0.11969607011170212, + "learning_rate": 4.95524171015298e-07, + "loss": 1.0548, + "step": 1778 + }, + { + "epoch": 4.549872122762149, + "grad_norm": 0.12029153180895684, + "learning_rate": 4.899868963601173e-07, + "loss": 1.0387, + "step": 1779 + }, + { + "epoch": 4.552429667519181, + "grad_norm": 0.11614133716710796, + "learning_rate": 4.844799570588699e-07, + "loss": 1.031, + "step": 1780 + }, + { + "epoch": 4.554987212276215, + "grad_norm": 0.11916764474488396, + "learning_rate": 4.79003370677773e-07, + "loss": 1.0628, + "step": 1781 + }, + { + "epoch": 4.557544757033249, + "grad_norm": 0.11805084554309808, + "learning_rate": 4.7355715468622164e-07, + "loss": 1.0366, + "step": 1782 + }, + { + "epoch": 4.560102301790281, + "grad_norm": 0.12003861008176886, + "learning_rate": 4.6814132645673584e-07, + "loss": 1.0908, + "step": 1783 + }, + { + "epoch": 4.562659846547315, + "grad_norm": 0.1148544867733163, + "learning_rate": 4.6275590326490316e-07, + "loss": 1.0109, + "step": 1784 + }, + { + "epoch": 4.565217391304348, + "grad_norm": 0.11850485924591762, + "learning_rate": 4.5740090228932554e-07, + "loss": 1.0137, + "step": 1785 + }, + { + "epoch": 4.567774936061381, + "grad_norm": 0.12073176991107863, + "learning_rate": 4.5207634061155913e-07, + "loss": 1.0203, + "step": 1786 + }, + { + "epoch": 4.570332480818414, + "grad_norm": 0.11565168377707126, + "learning_rate": 4.467822352160722e-07, + "loss": 1.0322, + "step": 1787 + }, + { + "epoch": 4.572890025575448, + "grad_norm": 0.11767773542643316, + "learning_rate": 4.415186029901719e-07, + "loss": 1.021, + "step": 1788 + }, + { + "epoch": 4.57544757033248, + "grad_norm": 0.11903289285879093, + "learning_rate": 4.3628546072396525e-07, + "loss": 1.0747, + "step": 1789 + }, + { + "epoch": 4.578005115089514, + "grad_norm": 0.12099267853278806, + "learning_rate": 4.310828251103072e-07, + "loss": 1.0488, + "step": 1790 + }, + { + "epoch": 4.580562659846548, + "grad_norm": 0.11812503754866224, + "learning_rate": 4.2591071274473484e-07, + "loss": 1.0362, + "step": 1791 + }, + { + "epoch": 4.58312020460358, + "grad_norm": 0.11817672051919005, + "learning_rate": 4.2076914012542105e-07, + "loss": 1.037, + "step": 1792 + }, + { + "epoch": 4.585677749360614, + "grad_norm": 0.11967377201350358, + "learning_rate": 4.1565812365312654e-07, + "loss": 1.0303, + "step": 1793 + }, + { + "epoch": 4.588235294117647, + "grad_norm": 0.12091871976659212, + "learning_rate": 4.105776796311389e-07, + "loss": 1.0411, + "step": 1794 + }, + { + "epoch": 4.59079283887468, + "grad_norm": 0.11889200883802002, + "learning_rate": 4.055278242652272e-07, + "loss": 1.0201, + "step": 1795 + }, + { + "epoch": 4.593350383631714, + "grad_norm": 0.11404037707431111, + "learning_rate": 4.00508573663585e-07, + "loss": 1.0326, + "step": 1796 + }, + { + "epoch": 4.595907928388747, + "grad_norm": 0.11895441005803606, + "learning_rate": 3.955199438367874e-07, + "loss": 1.0401, + "step": 1797 + }, + { + "epoch": 4.59846547314578, + "grad_norm": 0.11452304569258381, + "learning_rate": 3.9056195069772873e-07, + "loss": 1.0168, + "step": 1798 + }, + { + "epoch": 4.601023017902813, + "grad_norm": 0.11818309537389297, + "learning_rate": 3.85634610061576e-07, + "loss": 1.025, + "step": 1799 + }, + { + "epoch": 4.603580562659847, + "grad_norm": 0.1177738878198216, + "learning_rate": 3.807379376457276e-07, + "loss": 1.0579, + "step": 1800 + }, + { + "epoch": 4.6061381074168795, + "grad_norm": 0.11760050731022088, + "learning_rate": 3.758719490697493e-07, + "loss": 1.0162, + "step": 1801 + }, + { + "epoch": 4.608695652173913, + "grad_norm": 0.11692046035536714, + "learning_rate": 3.710366598553328e-07, + "loss": 1.0064, + "step": 1802 + }, + { + "epoch": 4.611253196930946, + "grad_norm": 0.11817659321984543, + "learning_rate": 3.6623208542624134e-07, + "loss": 1.0317, + "step": 1803 + }, + { + "epoch": 4.6138107416879794, + "grad_norm": 0.11416484517518209, + "learning_rate": 3.61458241108269e-07, + "loss": 0.9979, + "step": 1804 + }, + { + "epoch": 4.616368286445013, + "grad_norm": 0.11317634968030488, + "learning_rate": 3.567151421291781e-07, + "loss": 1.0428, + "step": 1805 + }, + { + "epoch": 4.618925831202046, + "grad_norm": 0.12015756336365767, + "learning_rate": 3.520028036186629e-07, + "loss": 1.0055, + "step": 1806 + }, + { + "epoch": 4.621483375959079, + "grad_norm": 0.11708694204425783, + "learning_rate": 3.473212406082993e-07, + "loss": 1.0358, + "step": 1807 + }, + { + "epoch": 4.624040920716112, + "grad_norm": 0.12291270689470975, + "learning_rate": 3.426704680314896e-07, + "loss": 1.0818, + "step": 1808 + }, + { + "epoch": 4.626598465473146, + "grad_norm": 0.12218447911853864, + "learning_rate": 3.380505007234225e-07, + "loss": 1.0499, + "step": 1809 + }, + { + "epoch": 4.629156010230179, + "grad_norm": 0.1197028476607302, + "learning_rate": 3.334613534210218e-07, + "loss": 1.0368, + "step": 1810 + }, + { + "epoch": 4.631713554987212, + "grad_norm": 0.11611382742011438, + "learning_rate": 3.2890304076290124e-07, + "loss": 1.0364, + "step": 1811 + }, + { + "epoch": 4.634271099744246, + "grad_norm": 0.11714111776328456, + "learning_rate": 3.2437557728931643e-07, + "loss": 1.0657, + "step": 1812 + }, + { + "epoch": 4.6368286445012785, + "grad_norm": 0.11376983368069393, + "learning_rate": 3.1987897744212073e-07, + "loss": 1.0011, + "step": 1813 + }, + { + "epoch": 4.639386189258312, + "grad_norm": 0.11405249072962266, + "learning_rate": 3.154132555647171e-07, + "loss": 1.0713, + "step": 1814 + }, + { + "epoch": 4.641943734015345, + "grad_norm": 0.1199549787481166, + "learning_rate": 3.1097842590201433e-07, + "loss": 1.0622, + "step": 1815 + }, + { + "epoch": 4.6445012787723785, + "grad_norm": 0.11926673117091068, + "learning_rate": 3.0657450260037303e-07, + "loss": 1.0629, + "step": 1816 + }, + { + "epoch": 4.647058823529412, + "grad_norm": 0.12139769893175593, + "learning_rate": 3.022014997075795e-07, + "loss": 1.0594, + "step": 1817 + }, + { + "epoch": 4.649616368286445, + "grad_norm": 0.11681143235855584, + "learning_rate": 2.9785943117277894e-07, + "loss": 1.0469, + "step": 1818 + }, + { + "epoch": 4.6521739130434785, + "grad_norm": 0.12421776159811096, + "learning_rate": 2.935483108464465e-07, + "loss": 1.0224, + "step": 1819 + }, + { + "epoch": 4.654731457800511, + "grad_norm": 0.11231911224982111, + "learning_rate": 2.8926815248033535e-07, + "loss": 1.0112, + "step": 1820 + }, + { + "epoch": 4.657289002557545, + "grad_norm": 0.11708199653029536, + "learning_rate": 2.8501896972743747e-07, + "loss": 1.0105, + "step": 1821 + }, + { + "epoch": 4.659846547314578, + "grad_norm": 0.11649971834696357, + "learning_rate": 2.808007761419351e-07, + "loss": 1.0697, + "step": 1822 + }, + { + "epoch": 4.662404092071611, + "grad_norm": 0.11071167386647211, + "learning_rate": 2.766135851791607e-07, + "loss": 1.0141, + "step": 1823 + }, + { + "epoch": 4.664961636828645, + "grad_norm": 0.12010263976141954, + "learning_rate": 2.7245741019555573e-07, + "loss": 1.0282, + "step": 1824 + }, + { + "epoch": 4.667519181585678, + "grad_norm": 0.1151865695779186, + "learning_rate": 2.6833226444862526e-07, + "loss": 1.0299, + "step": 1825 + }, + { + "epoch": 4.670076726342711, + "grad_norm": 0.11255028443601578, + "learning_rate": 2.642381610968936e-07, + "loss": 1.0267, + "step": 1826 + }, + { + "epoch": 4.672634271099744, + "grad_norm": 0.11904179423196315, + "learning_rate": 2.6017511319986753e-07, + "loss": 1.029, + "step": 1827 + }, + { + "epoch": 4.675191815856778, + "grad_norm": 0.11156796619875065, + "learning_rate": 2.5614313371799203e-07, + "loss": 1.0193, + "step": 1828 + }, + { + "epoch": 4.677749360613811, + "grad_norm": 0.12152103248098621, + "learning_rate": 2.5214223551260686e-07, + "loss": 1.0267, + "step": 1829 + }, + { + "epoch": 4.680306905370844, + "grad_norm": 0.12044476016060106, + "learning_rate": 2.4817243134591106e-07, + "loss": 1.0524, + "step": 1830 + }, + { + "epoch": 4.6828644501278776, + "grad_norm": 0.1165747082957651, + "learning_rate": 2.4423373388091755e-07, + "loss": 1.0347, + "step": 1831 + }, + { + "epoch": 4.68542199488491, + "grad_norm": 0.11850587011434784, + "learning_rate": 2.403261556814118e-07, + "loss": 1.0568, + "step": 1832 + }, + { + "epoch": 4.687979539641944, + "grad_norm": 0.1162877587424683, + "learning_rate": 2.3644970921191447e-07, + "loss": 1.0266, + "step": 1833 + }, + { + "epoch": 4.690537084398977, + "grad_norm": 0.12535561464481632, + "learning_rate": 2.3260440683764653e-07, + "loss": 1.0845, + "step": 1834 + }, + { + "epoch": 4.69309462915601, + "grad_norm": 0.1218176284645347, + "learning_rate": 2.2879026082447743e-07, + "loss": 1.0469, + "step": 1835 + }, + { + "epoch": 4.695652173913043, + "grad_norm": 0.11820198842847074, + "learning_rate": 2.250072833388972e-07, + "loss": 1.0585, + "step": 1836 + }, + { + "epoch": 4.698209718670077, + "grad_norm": 0.12029732293646825, + "learning_rate": 2.2125548644797322e-07, + "loss": 1.0422, + "step": 1837 + }, + { + "epoch": 4.70076726342711, + "grad_norm": 0.11483889783410742, + "learning_rate": 2.1753488211931017e-07, + "loss": 1.0095, + "step": 1838 + }, + { + "epoch": 4.703324808184143, + "grad_norm": 0.11736543180370013, + "learning_rate": 2.1384548222101342e-07, + "loss": 1.0064, + "step": 1839 + }, + { + "epoch": 4.705882352941177, + "grad_norm": 0.11729827431422608, + "learning_rate": 2.1018729852165575e-07, + "loss": 1.0682, + "step": 1840 + }, + { + "epoch": 4.708439897698209, + "grad_norm": 0.11626920918028295, + "learning_rate": 2.0656034269022961e-07, + "loss": 1.0267, + "step": 1841 + }, + { + "epoch": 4.710997442455243, + "grad_norm": 0.11821893661532155, + "learning_rate": 2.0296462629611934e-07, + "loss": 1.0235, + "step": 1842 + }, + { + "epoch": 4.713554987212277, + "grad_norm": 0.11027628288597482, + "learning_rate": 1.9940016080906122e-07, + "loss": 1.0249, + "step": 1843 + }, + { + "epoch": 4.716112531969309, + "grad_norm": 0.11583827201970737, + "learning_rate": 1.9586695759910235e-07, + "loss": 1.0477, + "step": 1844 + }, + { + "epoch": 4.718670076726343, + "grad_norm": 0.11248702023779261, + "learning_rate": 1.9236502793657297e-07, + "loss": 1.0441, + "step": 1845 + }, + { + "epoch": 4.721227621483376, + "grad_norm": 0.1240044200990586, + "learning_rate": 1.888943829920431e-07, + "loss": 1.0336, + "step": 1846 + }, + { + "epoch": 4.723785166240409, + "grad_norm": 0.11690696902623407, + "learning_rate": 1.854550338362915e-07, + "loss": 1.0376, + "step": 1847 + }, + { + "epoch": 4.726342710997442, + "grad_norm": 0.11486713464937727, + "learning_rate": 1.8204699144026894e-07, + "loss": 1.0689, + "step": 1848 + }, + { + "epoch": 4.728900255754476, + "grad_norm": 0.11581909664334852, + "learning_rate": 1.7867026667505726e-07, + "loss": 1.0672, + "step": 1849 + }, + { + "epoch": 4.731457800511509, + "grad_norm": 0.11440636945394728, + "learning_rate": 1.753248703118482e-07, + "loss": 1.034, + "step": 1850 + }, + { + "epoch": 4.734015345268542, + "grad_norm": 0.1145247227720534, + "learning_rate": 1.720108130218967e-07, + "loss": 1.0355, + "step": 1851 + }, + { + "epoch": 4.736572890025576, + "grad_norm": 0.11936327466725484, + "learning_rate": 1.6872810537649332e-07, + "loss": 1.0274, + "step": 1852 + }, + { + "epoch": 4.739130434782608, + "grad_norm": 0.1191110933377237, + "learning_rate": 1.6547675784692517e-07, + "loss": 1.0707, + "step": 1853 + }, + { + "epoch": 4.741687979539642, + "grad_norm": 0.11898800646585839, + "learning_rate": 1.6225678080444952e-07, + "loss": 1.0631, + "step": 1854 + }, + { + "epoch": 4.744245524296675, + "grad_norm": 0.11617957331470609, + "learning_rate": 1.5906818452025464e-07, + "loss": 1.0394, + "step": 1855 + }, + { + "epoch": 4.746803069053708, + "grad_norm": 0.11575268445319527, + "learning_rate": 1.559109791654301e-07, + "loss": 1.0059, + "step": 1856 + }, + { + "epoch": 4.749360613810742, + "grad_norm": 0.11400146363398818, + "learning_rate": 1.5278517481093436e-07, + "loss": 1.0268, + "step": 1857 + }, + { + "epoch": 4.751918158567775, + "grad_norm": 0.11361459986982794, + "learning_rate": 1.4969078142756276e-07, + "loss": 1.0565, + "step": 1858 + }, + { + "epoch": 4.754475703324808, + "grad_norm": 0.11718329814715675, + "learning_rate": 1.4662780888591078e-07, + "loss": 1.0271, + "step": 1859 + }, + { + "epoch": 4.757033248081841, + "grad_norm": 0.11412559030858926, + "learning_rate": 1.4359626695635176e-07, + "loss": 1.0529, + "step": 1860 + }, + { + "epoch": 4.759590792838875, + "grad_norm": 0.1149335211731899, + "learning_rate": 1.405961653089971e-07, + "loss": 1.047, + "step": 1861 + }, + { + "epoch": 4.762148337595908, + "grad_norm": 0.12700071915120056, + "learning_rate": 1.3762751351367066e-07, + "loss": 1.0394, + "step": 1862 + }, + { + "epoch": 4.764705882352941, + "grad_norm": 0.11612798778465196, + "learning_rate": 1.3469032103987534e-07, + "loss": 1.0545, + "step": 1863 + }, + { + "epoch": 4.767263427109975, + "grad_norm": 0.12026346999851885, + "learning_rate": 1.3178459725676552e-07, + "loss": 1.0633, + "step": 1864 + }, + { + "epoch": 4.7698209718670075, + "grad_norm": 0.1154062718622743, + "learning_rate": 1.289103514331147e-07, + "loss": 1.0257, + "step": 1865 + }, + { + "epoch": 4.772378516624041, + "grad_norm": 0.1196571096618875, + "learning_rate": 1.2606759273728563e-07, + "loss": 1.0532, + "step": 1866 + }, + { + "epoch": 4.774936061381074, + "grad_norm": 0.11924573106561724, + "learning_rate": 1.2325633023720695e-07, + "loss": 1.0275, + "step": 1867 + }, + { + "epoch": 4.7774936061381075, + "grad_norm": 0.11270811144881691, + "learning_rate": 1.2047657290033322e-07, + "loss": 1.0286, + "step": 1868 + }, + { + "epoch": 4.78005115089514, + "grad_norm": 0.1161070504851422, + "learning_rate": 1.1772832959362934e-07, + "loss": 1.0433, + "step": 1869 + }, + { + "epoch": 4.782608695652174, + "grad_norm": 0.11545585799058032, + "learning_rate": 1.150116090835307e-07, + "loss": 1.0332, + "step": 1870 + }, + { + "epoch": 4.7851662404092075, + "grad_norm": 0.1152683133795509, + "learning_rate": 1.1232642003592198e-07, + "loss": 1.0849, + "step": 1871 + }, + { + "epoch": 4.78772378516624, + "grad_norm": 0.11412468149661283, + "learning_rate": 1.096727710161094e-07, + "loss": 1.0167, + "step": 1872 + }, + { + "epoch": 4.790281329923274, + "grad_norm": 0.1180180215345883, + "learning_rate": 1.0705067048878859e-07, + "loss": 1.0227, + "step": 1873 + }, + { + "epoch": 4.792838874680307, + "grad_norm": 0.11718614653864518, + "learning_rate": 1.0446012681802342e-07, + "loss": 1.0085, + "step": 1874 + }, + { + "epoch": 4.79539641943734, + "grad_norm": 0.11364381432368652, + "learning_rate": 1.0190114826721498e-07, + "loss": 1.0399, + "step": 1875 + }, + { + "epoch": 4.797953964194374, + "grad_norm": 0.11712192627428837, + "learning_rate": 9.937374299907931e-08, + "loss": 1.0065, + "step": 1876 + }, + { + "epoch": 4.8005115089514065, + "grad_norm": 0.11451658146251059, + "learning_rate": 9.687791907561528e-08, + "loss": 1.0496, + "step": 1877 + }, + { + "epoch": 4.80306905370844, + "grad_norm": 0.11508044552253341, + "learning_rate": 9.441368445808452e-08, + "loss": 1.0378, + "step": 1878 + }, + { + "epoch": 4.805626598465473, + "grad_norm": 0.11151643216416461, + "learning_rate": 9.198104700698596e-08, + "loss": 1.0202, + "step": 1879 + }, + { + "epoch": 4.8081841432225065, + "grad_norm": 0.11464186324553732, + "learning_rate": 8.958001448202358e-08, + "loss": 1.0152, + "step": 1880 + }, + { + "epoch": 4.810741687979539, + "grad_norm": 0.11662463464886089, + "learning_rate": 8.721059454209424e-08, + "loss": 1.0485, + "step": 1881 + }, + { + "epoch": 4.813299232736573, + "grad_norm": 0.12464254809498643, + "learning_rate": 8.48727947452499e-08, + "loss": 1.017, + "step": 1882 + }, + { + "epoch": 4.8158567774936065, + "grad_norm": 0.11127980935947186, + "learning_rate": 8.256662254867986e-08, + "loss": 1.0076, + "step": 1883 + }, + { + "epoch": 4.818414322250639, + "grad_norm": 0.11536372225589502, + "learning_rate": 8.029208530869081e-08, + "loss": 1.0295, + "step": 1884 + }, + { + "epoch": 4.820971867007673, + "grad_norm": 0.11600907835953211, + "learning_rate": 7.804919028067681e-08, + "loss": 1.0307, + "step": 1885 + }, + { + "epoch": 4.823529411764706, + "grad_norm": 0.11464611599367469, + "learning_rate": 7.58379446190971e-08, + "loss": 1.0466, + "step": 1886 + }, + { + "epoch": 4.826086956521739, + "grad_norm": 0.11686971960574333, + "learning_rate": 7.365835537745724e-08, + "loss": 1.0442, + "step": 1887 + }, + { + "epoch": 4.828644501278772, + "grad_norm": 0.11487887116386886, + "learning_rate": 7.151042950828247e-08, + "loss": 1.0445, + "step": 1888 + }, + { + "epoch": 4.831202046035806, + "grad_norm": 0.11340523651243395, + "learning_rate": 6.939417386309765e-08, + "loss": 1.029, + "step": 1889 + }, + { + "epoch": 4.833759590792839, + "grad_norm": 0.1150493275878801, + "learning_rate": 6.73095951924041e-08, + "loss": 1.0097, + "step": 1890 + }, + { + "epoch": 4.836317135549872, + "grad_norm": 0.11903827021172662, + "learning_rate": 6.525670014566166e-08, + "loss": 1.0464, + "step": 1891 + }, + { + "epoch": 4.838874680306906, + "grad_norm": 0.10980997002331325, + "learning_rate": 6.323549527126216e-08, + "loss": 1.0109, + "step": 1892 + }, + { + "epoch": 4.841432225063938, + "grad_norm": 0.1150321871820009, + "learning_rate": 6.124598701651052e-08, + "loss": 0.9896, + "step": 1893 + }, + { + "epoch": 4.843989769820972, + "grad_norm": 0.11826635724952957, + "learning_rate": 5.9288181727606976e-08, + "loss": 1.0406, + "step": 1894 + }, + { + "epoch": 4.846547314578006, + "grad_norm": 0.12341344246437233, + "learning_rate": 5.7362085649622646e-08, + "loss": 1.0292, + "step": 1895 + }, + { + "epoch": 4.849104859335038, + "grad_norm": 0.11768310013022554, + "learning_rate": 5.546770492648401e-08, + "loss": 1.0641, + "step": 1896 + }, + { + "epoch": 4.851662404092072, + "grad_norm": 0.1161461545199052, + "learning_rate": 5.360504560094737e-08, + "loss": 1.0278, + "step": 1897 + }, + { + "epoch": 4.854219948849105, + "grad_norm": 0.1139025404375565, + "learning_rate": 5.1774113614586616e-08, + "loss": 1.0405, + "step": 1898 + }, + { + "epoch": 4.856777493606138, + "grad_norm": 0.1215131651232801, + "learning_rate": 4.9974914807767724e-08, + "loss": 1.0753, + "step": 1899 + }, + { + "epoch": 4.859335038363171, + "grad_norm": 0.11755019743640818, + "learning_rate": 4.82074549196343e-08, + "loss": 1.0235, + "step": 1900 + }, + { + "epoch": 4.861892583120205, + "grad_norm": 0.11408088672302473, + "learning_rate": 4.6471739588089814e-08, + "loss": 1.0253, + "step": 1901 + }, + { + "epoch": 4.864450127877237, + "grad_norm": 0.11289577037034211, + "learning_rate": 4.4767774349776527e-08, + "loss": 1.0321, + "step": 1902 + }, + { + "epoch": 4.867007672634271, + "grad_norm": 0.11895648456750522, + "learning_rate": 4.30955646400566e-08, + "loss": 1.043, + "step": 1903 + }, + { + "epoch": 4.869565217391305, + "grad_norm": 0.1179294889008452, + "learning_rate": 4.14551157930021e-08, + "loss": 1.0538, + "step": 1904 + }, + { + "epoch": 4.872122762148337, + "grad_norm": 0.11618208912614107, + "learning_rate": 3.984643304136948e-08, + "loss": 1.0476, + "step": 1905 + }, + { + "epoch": 4.874680306905371, + "grad_norm": 0.11598189775025243, + "learning_rate": 3.826952151658958e-08, + "loss": 1.0356, + "step": 1906 + }, + { + "epoch": 4.877237851662404, + "grad_norm": 0.11586014435065378, + "learning_rate": 3.6724386248745415e-08, + "loss": 1.0402, + "step": 1907 + }, + { + "epoch": 4.879795396419437, + "grad_norm": 0.11273964633582521, + "learning_rate": 3.5211032166561075e-08, + "loss": 1.0651, + "step": 1908 + }, + { + "epoch": 4.882352941176471, + "grad_norm": 0.11553305320282162, + "learning_rate": 3.372946409738398e-08, + "loss": 1.0723, + "step": 1909 + }, + { + "epoch": 4.884910485933504, + "grad_norm": 0.1164335776440849, + "learning_rate": 3.22796867671682e-08, + "loss": 1.0322, + "step": 1910 + }, + { + "epoch": 4.887468030690537, + "grad_norm": 0.11551854002367276, + "learning_rate": 3.086170480046114e-08, + "loss": 1.0121, + "step": 1911 + }, + { + "epoch": 4.89002557544757, + "grad_norm": 0.11890262672457762, + "learning_rate": 2.9475522720389116e-08, + "loss": 1.0571, + "step": 1912 + }, + { + "epoch": 4.892583120204604, + "grad_norm": 0.11559520376004895, + "learning_rate": 2.8121144948641822e-08, + "loss": 1.0378, + "step": 1913 + }, + { + "epoch": 4.8951406649616365, + "grad_norm": 0.11021123166347066, + "learning_rate": 2.679857580545786e-08, + "loss": 1.0468, + "step": 1914 + }, + { + "epoch": 4.89769820971867, + "grad_norm": 0.11314952605469081, + "learning_rate": 2.5507819509610343e-08, + "loss": 1.0388, + "step": 1915 + }, + { + "epoch": 4.900255754475703, + "grad_norm": 0.11452758417807123, + "learning_rate": 2.4248880178399104e-08, + "loss": 1.0633, + "step": 1916 + }, + { + "epoch": 4.9028132992327365, + "grad_norm": 0.11870376108345816, + "learning_rate": 2.3021761827628497e-08, + "loss": 1.0193, + "step": 1917 + }, + { + "epoch": 4.90537084398977, + "grad_norm": 0.11658933357067867, + "learning_rate": 2.182646837160185e-08, + "loss": 0.9916, + "step": 1918 + }, + { + "epoch": 4.907928388746803, + "grad_norm": 0.11670304567415841, + "learning_rate": 2.0663003623105916e-08, + "loss": 1.0672, + "step": 1919 + }, + { + "epoch": 4.910485933503836, + "grad_norm": 0.1102254208498854, + "learning_rate": 1.953137129339977e-08, + "loss": 1.0298, + "step": 1920 + }, + { + "epoch": 4.913043478260869, + "grad_norm": 0.11803357978542595, + "learning_rate": 1.8431574992199276e-08, + "loss": 1.0461, + "step": 1921 + }, + { + "epoch": 4.915601023017903, + "grad_norm": 0.11020661745660215, + "learning_rate": 1.736361822767263e-08, + "loss": 1.0395, + "step": 1922 + }, + { + "epoch": 4.918158567774936, + "grad_norm": 0.11305349069484284, + "learning_rate": 1.632750440642261e-08, + "loss": 1.0231, + "step": 1923 + }, + { + "epoch": 4.920716112531969, + "grad_norm": 0.11510462052686142, + "learning_rate": 1.5323236833479914e-08, + "loss": 1.0259, + "step": 1924 + }, + { + "epoch": 4.923273657289003, + "grad_norm": 0.11702355777739097, + "learning_rate": 1.4350818712292047e-08, + "loss": 1.0613, + "step": 1925 + }, + { + "epoch": 4.9258312020460355, + "grad_norm": 0.11292851087049081, + "learning_rate": 1.3410253144707786e-08, + "loss": 1.0592, + "step": 1926 + }, + { + "epoch": 4.928388746803069, + "grad_norm": 0.11400163077833202, + "learning_rate": 1.2501543130974958e-08, + "loss": 1.062, + "step": 1927 + }, + { + "epoch": 4.930946291560103, + "grad_norm": 0.1183936881159329, + "learning_rate": 1.162469156972712e-08, + "loss": 1.0245, + "step": 1928 + }, + { + "epoch": 4.9335038363171355, + "grad_norm": 0.11337241506632771, + "learning_rate": 1.0779701257974672e-08, + "loss": 1.0449, + "step": 1929 + }, + { + "epoch": 4.936061381074169, + "grad_norm": 0.11376144202297905, + "learning_rate": 9.96657489109487e-09, + "loss": 1.0682, + "step": 1930 + }, + { + "epoch": 4.938618925831202, + "grad_norm": 0.11795763495218076, + "learning_rate": 9.185315062826272e-09, + "loss": 1.0674, + "step": 1931 + }, + { + "epoch": 4.9411764705882355, + "grad_norm": 0.11749004694568776, + "learning_rate": 8.435924265256523e-09, + "loss": 1.0479, + "step": 1932 + }, + { + "epoch": 4.943734015345268, + "grad_norm": 0.10996587345538478, + "learning_rate": 7.718404888816811e-09, + "loss": 1.0148, + "step": 1933 + }, + { + "epoch": 4.946291560102302, + "grad_norm": 0.11480984252460183, + "learning_rate": 7.032759222274088e-09, + "loss": 1.0408, + "step": 1934 + }, + { + "epoch": 4.948849104859335, + "grad_norm": 0.11630744934298476, + "learning_rate": 6.378989452724416e-09, + "loss": 1.0236, + "step": 1935 + }, + { + "epoch": 4.951406649616368, + "grad_norm": 0.1124895154235232, + "learning_rate": 5.757097665584077e-09, + "loss": 1.0335, + "step": 1936 + }, + { + "epoch": 4.953964194373402, + "grad_norm": 0.11425081997455588, + "learning_rate": 5.16708584458292e-09, + "loss": 1.0354, + "step": 1937 + }, + { + "epoch": 4.956521739130435, + "grad_norm": 0.11330328401845938, + "learning_rate": 4.608955871761023e-09, + "loss": 1.0077, + "step": 1938 + }, + { + "epoch": 4.959079283887468, + "grad_norm": 0.11496557714114423, + "learning_rate": 4.0827095274598165e-09, + "loss": 1.0478, + "step": 1939 + }, + { + "epoch": 4.961636828644501, + "grad_norm": 0.12009777724541458, + "learning_rate": 3.588348490317639e-09, + "loss": 1.0568, + "step": 1940 + }, + { + "epoch": 4.964194373401535, + "grad_norm": 0.1165369144605732, + "learning_rate": 3.1258743372619693e-09, + "loss": 1.0352, + "step": 1941 + }, + { + "epoch": 4.966751918158568, + "grad_norm": 0.11826068216824383, + "learning_rate": 2.6952885435105324e-09, + "loss": 1.047, + "step": 1942 + }, + { + "epoch": 4.969309462915601, + "grad_norm": 0.11246869085468798, + "learning_rate": 2.29659248255798e-09, + "loss": 1.0417, + "step": 1943 + }, + { + "epoch": 4.971867007672635, + "grad_norm": 0.11023466743258699, + "learning_rate": 1.9297874261792194e-09, + "loss": 1.0137, + "step": 1944 + }, + { + "epoch": 4.974424552429667, + "grad_norm": 0.11511965541940995, + "learning_rate": 1.594874544421643e-09, + "loss": 1.0311, + "step": 1945 + }, + { + "epoch": 4.976982097186701, + "grad_norm": 0.124147218545505, + "learning_rate": 1.2918549056006868e-09, + "loss": 1.0719, + "step": 1946 + }, + { + "epoch": 4.979539641943734, + "grad_norm": 0.11557957987866309, + "learning_rate": 1.0207294762987208e-09, + "loss": 1.0187, + "step": 1947 + }, + { + "epoch": 4.982097186700767, + "grad_norm": 0.11775087960856384, + "learning_rate": 7.814991213594969e-10, + "loss": 1.0355, + "step": 1948 + }, + { + "epoch": 4.9846547314578, + "grad_norm": 0.11570647330926523, + "learning_rate": 5.741646038892601e-10, + "loss": 1.0646, + "step": 1949 + }, + { + "epoch": 4.987212276214834, + "grad_norm": 0.1150012391819302, + "learning_rate": 3.9872658525008654e-10, + "loss": 1.0612, + "step": 1950 + }, + { + "epoch": 4.989769820971867, + "grad_norm": 0.10963077899170262, + "learning_rate": 2.5518562505988387e-10, + "loss": 1.0244, + "step": 1951 + }, + { + "epoch": 4.9923273657289, + "grad_norm": 0.1137791798614674, + "learning_rate": 1.435421811901705e-10, + "loss": 1.0499, + "step": 1952 + }, + { + "epoch": 4.994884910485934, + "grad_norm": 0.11415395430572649, + "learning_rate": 6.37966097649656e-11, + "loss": 1.0107, + "step": 1953 + }, + { + "epoch": 4.997442455242966, + "grad_norm": 0.11255930720124045, + "learning_rate": 1.5949165159678813e-11, + "loss": 1.0177, + "step": 1954 + }, + { + "epoch": 5.0, + "grad_norm": 0.11539135868139018, + "learning_rate": 0.0, + "loss": 1.0412, + "step": 1955 + }, + { + "epoch": 5.0, + "step": 1955, + "total_flos": 7122204608430080.0, + "train_loss": 1.082116018475779, + "train_runtime": 36273.766, + "train_samples_per_second": 13.784, + "train_steps_per_second": 0.054 + } + ], + "logging_steps": 1.0, + "max_steps": 1955, + "num_input_tokens_seen": 0, + "num_train_epochs": 5, + "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": 7122204608430080.0, + "train_batch_size": 4, + "trial_name": null, + "trial_params": null +} diff --git a/training_args.bin b/training_args.bin new file mode 100644 index 0000000..cde5fe2 --- /dev/null +++ b/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c5c06f9e620eb2c59ec7149070c38f4af2d1d5a02b6474c17bbf6283c824342e +size 7224 diff --git a/training_loss.png b/training_loss.png new file mode 100644 index 0000000..b131b26 Binary files /dev/null and b/training_loss.png differ diff --git a/vocab.json b/vocab.json new file mode 100644 index 0000000..6c49fc6 --- /dev/null +++ b/vocab.json @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ca10d7e9fb3ed18575dd1e277a2579c16d108e32f27439684afa0e10b1440910 +size 2776833