commit 28a5293549829bc6ae814f3d05fe50bd6e2b6e8f Author: ModelHub XC Date: Fri Jun 26 19:31:13 2026 +0800 初始化项目,由ModelHub XC社区提供模型 Model: mlfoundations-dev/openthoughts3_100k_qwen25_1b_bsz256_lr8e5_epochs5 Source: Original Platform diff --git a/.gitattributes b/.gitattributes new file mode 100644 index 0000000..fd072cd --- /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 + +training_args.bin filter=lfs diff=lfs merge=lfs -text +vocab.json filter=lfs diff=lfs merge=lfs -text +model.safetensors filter=lfs diff=lfs merge=lfs -text +tokenizer.json filter=lfs diff=lfs merge=lfs -text +merges.txt 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..de8e844 --- /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_lr8e5_epochs5 + results: [] +--- + + + +# openthoughts3_100k_qwen25_1b_bsz256_lr8e5_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: 8e-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..7f428bb --- /dev/null +++ b/all_results.json @@ -0,0 +1,8 @@ +{ + "epoch": 5.0, + "total_flos": 7122204608430080.0, + "train_loss": 1.0287959775656386, + "train_runtime": 35452.5316, + "train_samples_per_second": 14.103, + "train_steps_per_second": 0.055 +} \ 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..eb3caad --- /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_lr8e5_epochs5 +learning_rate: 8e-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_lr8e5_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..198b39f --- /dev/null +++ b/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b01d400078c8069bc557377efa95bf40fbe651b6c95b6137c7b8bbdbf1a41e0c +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..7f428bb --- /dev/null +++ b/train_results.json @@ -0,0 +1,8 @@ +{ + "epoch": 5.0, + "total_flos": 7122204608430080.0, + "train_loss": 1.0287959775656386, + "train_runtime": 35452.5316, + "train_samples_per_second": 14.103, + "train_steps_per_second": 0.055 +} \ No newline at end of file diff --git a/trainer_log.jsonl b/trainer_log.jsonl new file mode 100644 index 0000000..b291d87 --- /dev/null +++ b/trainer_log.jsonl @@ -0,0 +1,1956 @@ +{"current_steps": 1, "total_steps": 1955, "loss": 1.5213, "lr": 4.0816326530612243e-07, "epoch": 0.0025575447570332483, "percentage": 0.05, "elapsed_time": "0:00:25", "remaining_time": "13:50:13"} +{"current_steps": 2, "total_steps": 1955, "loss": 1.4742, "lr": 8.163265306122449e-07, "epoch": 0.005115089514066497, "percentage": 0.1, "elapsed_time": "0:00:43", "remaining_time": "11:54:45"} +{"current_steps": 3, "total_steps": 1955, "loss": 1.4946, "lr": 1.2244897959183673e-06, "epoch": 0.0076726342710997444, "percentage": 0.15, "elapsed_time": "0:01:02", "remaining_time": "11:16:09"} +{"current_steps": 4, "total_steps": 1955, "loss": 1.4837, "lr": 1.6326530612244897e-06, "epoch": 0.010230179028132993, "percentage": 0.2, "elapsed_time": "0:01:20", "remaining_time": "10:55:23"} +{"current_steps": 5, "total_steps": 1955, "loss": 1.4922, "lr": 2.0408163265306125e-06, "epoch": 0.01278772378516624, "percentage": 0.26, "elapsed_time": "0:01:38", "remaining_time": "10:42:40"} +{"current_steps": 6, "total_steps": 1955, "loss": 1.4997, "lr": 2.4489795918367347e-06, "epoch": 0.015345268542199489, "percentage": 0.31, "elapsed_time": "0:01:57", "remaining_time": "10:37:39"} +{"current_steps": 7, "total_steps": 1955, "loss": 1.4709, "lr": 2.8571428571428573e-06, "epoch": 0.017902813299232736, "percentage": 0.36, "elapsed_time": "0:02:15", "remaining_time": "10:29:49"} +{"current_steps": 8, "total_steps": 1955, "loss": 1.4673, "lr": 3.2653061224489794e-06, "epoch": 0.020460358056265986, "percentage": 0.41, "elapsed_time": "0:02:34", "remaining_time": "10:28:30"} +{"current_steps": 9, "total_steps": 1955, "loss": 1.5002, "lr": 3.6734693877551024e-06, "epoch": 0.023017902813299233, "percentage": 0.46, "elapsed_time": "0:02:52", "remaining_time": "10:23:19"} +{"current_steps": 10, "total_steps": 1955, "loss": 1.4897, "lr": 4.081632653061225e-06, "epoch": 0.02557544757033248, "percentage": 0.51, "elapsed_time": "0:03:10", "remaining_time": "10:19:07"} +{"current_steps": 11, "total_steps": 1955, "loss": 1.4669, "lr": 4.489795918367348e-06, "epoch": 0.028132992327365727, "percentage": 0.56, "elapsed_time": "0:03:29", "remaining_time": "10:15:40"} +{"current_steps": 12, "total_steps": 1955, "loss": 1.4282, "lr": 4.897959183673469e-06, "epoch": 0.030690537084398978, "percentage": 0.61, "elapsed_time": "0:03:47", "remaining_time": "10:12:43"} +{"current_steps": 13, "total_steps": 1955, "loss": 1.4664, "lr": 5.306122448979593e-06, "epoch": 0.03324808184143223, "percentage": 0.66, "elapsed_time": "0:04:05", "remaining_time": "10:10:09"} +{"current_steps": 14, "total_steps": 1955, "loss": 1.3861, "lr": 5.7142857142857145e-06, "epoch": 0.03580562659846547, "percentage": 0.72, "elapsed_time": "0:04:23", "remaining_time": "10:07:55"} +{"current_steps": 15, "total_steps": 1955, "loss": 1.4201, "lr": 6.122448979591837e-06, "epoch": 0.03836317135549872, "percentage": 0.77, "elapsed_time": "0:04:41", "remaining_time": "10:06:06"} +{"current_steps": 16, "total_steps": 1955, "loss": 1.3764, "lr": 6.530612244897959e-06, "epoch": 0.04092071611253197, "percentage": 0.82, "elapsed_time": "0:05:00", "remaining_time": "10:06:15"} +{"current_steps": 17, "total_steps": 1955, "loss": 1.3926, "lr": 6.938775510204082e-06, "epoch": 0.043478260869565216, "percentage": 0.87, "elapsed_time": "0:05:18", "remaining_time": "10:04:33"} +{"current_steps": 18, "total_steps": 1955, "loss": 1.3886, "lr": 7.346938775510205e-06, "epoch": 0.04603580562659847, "percentage": 0.92, "elapsed_time": "0:05:35", "remaining_time": "10:02:19"} +{"current_steps": 19, "total_steps": 1955, "loss": 1.3666, "lr": 7.755102040816327e-06, "epoch": 0.04859335038363171, "percentage": 0.97, "elapsed_time": "0:05:53", "remaining_time": "10:00:56"} +{"current_steps": 20, "total_steps": 1955, "loss": 1.3479, "lr": 8.16326530612245e-06, "epoch": 0.05115089514066496, "percentage": 1.02, "elapsed_time": "0:06:13", "remaining_time": "10:01:36"} +{"current_steps": 21, "total_steps": 1955, "loss": 1.3275, "lr": 8.571428571428571e-06, "epoch": 0.05370843989769821, "percentage": 1.07, "elapsed_time": "0:06:31", "remaining_time": "10:00:18"} +{"current_steps": 22, "total_steps": 1955, "loss": 1.3511, "lr": 8.979591836734695e-06, "epoch": 0.056265984654731455, "percentage": 1.13, "elapsed_time": "0:06:49", "remaining_time": "9:59:04"} +{"current_steps": 23, "total_steps": 1955, "loss": 1.3045, "lr": 9.387755102040818e-06, "epoch": 0.058823529411764705, "percentage": 1.18, "elapsed_time": "0:07:07", "remaining_time": "9:57:58"} +{"current_steps": 24, "total_steps": 1955, "loss": 1.3239, "lr": 9.795918367346939e-06, "epoch": 0.061381074168797956, "percentage": 1.23, "elapsed_time": "0:07:25", "remaining_time": "9:56:55"} +{"current_steps": 25, "total_steps": 1955, "loss": 1.2989, "lr": 1.0204081632653063e-05, "epoch": 0.0639386189258312, "percentage": 1.28, "elapsed_time": "0:07:43", "remaining_time": "9:56:42"} +{"current_steps": 26, "total_steps": 1955, "loss": 1.3023, "lr": 1.0612244897959186e-05, "epoch": 0.06649616368286446, "percentage": 1.33, "elapsed_time": "0:08:01", "remaining_time": "9:55:45"} +{"current_steps": 27, "total_steps": 1955, "loss": 1.2843, "lr": 1.1020408163265306e-05, "epoch": 0.06905370843989769, "percentage": 1.38, "elapsed_time": "0:08:19", "remaining_time": "9:54:58"} +{"current_steps": 28, "total_steps": 1955, "loss": 1.3075, "lr": 1.1428571428571429e-05, "epoch": 0.07161125319693094, "percentage": 1.43, "elapsed_time": "0:08:37", "remaining_time": "9:54:06"} +{"current_steps": 29, "total_steps": 1955, "loss": 1.2835, "lr": 1.1836734693877552e-05, "epoch": 0.0741687979539642, "percentage": 1.48, "elapsed_time": "0:08:56", "remaining_time": "9:53:19"} +{"current_steps": 30, "total_steps": 1955, "loss": 1.2971, "lr": 1.2244897959183674e-05, "epoch": 0.07672634271099744, "percentage": 1.53, "elapsed_time": "0:09:14", "remaining_time": "9:52:36"} +{"current_steps": 31, "total_steps": 1955, "loss": 1.2911, "lr": 1.2653061224489798e-05, "epoch": 0.0792838874680307, "percentage": 1.59, "elapsed_time": "0:09:32", "remaining_time": "9:51:52"} +{"current_steps": 32, "total_steps": 1955, "loss": 1.2638, "lr": 1.3061224489795918e-05, "epoch": 0.08184143222506395, "percentage": 1.64, "elapsed_time": "0:09:50", "remaining_time": "9:51:08"} +{"current_steps": 33, "total_steps": 1955, "loss": 1.2964, "lr": 1.3469387755102042e-05, "epoch": 0.08439897698209718, "percentage": 1.69, "elapsed_time": "0:10:09", "remaining_time": "9:51:12"} +{"current_steps": 34, "total_steps": 1955, "loss": 1.3138, "lr": 1.3877551020408165e-05, "epoch": 0.08695652173913043, "percentage": 1.74, "elapsed_time": "0:10:27", "remaining_time": "9:50:28"} +{"current_steps": 35, "total_steps": 1955, "loss": 1.2555, "lr": 1.4285714285714287e-05, "epoch": 0.08951406649616368, "percentage": 1.79, "elapsed_time": "0:10:45", "remaining_time": "9:49:47"} +{"current_steps": 36, "total_steps": 1955, "loss": 1.2684, "lr": 1.469387755102041e-05, "epoch": 0.09207161125319693, "percentage": 1.84, "elapsed_time": "0:11:02", "remaining_time": "9:48:24"} +{"current_steps": 37, "total_steps": 1955, "loss": 1.2828, "lr": 1.510204081632653e-05, "epoch": 0.09462915601023018, "percentage": 1.89, "elapsed_time": "0:11:20", "remaining_time": "9:47:48"} +{"current_steps": 38, "total_steps": 1955, "loss": 1.2585, "lr": 1.5510204081632655e-05, "epoch": 0.09718670076726342, "percentage": 1.94, "elapsed_time": "0:11:39", "remaining_time": "9:48:07"} +{"current_steps": 39, "total_steps": 1955, "loss": 1.2739, "lr": 1.5918367346938776e-05, "epoch": 0.09974424552429667, "percentage": 1.99, "elapsed_time": "0:11:57", "remaining_time": "9:47:31"} +{"current_steps": 40, "total_steps": 1955, "loss": 1.2148, "lr": 1.63265306122449e-05, "epoch": 0.10230179028132992, "percentage": 2.05, "elapsed_time": "0:12:15", "remaining_time": "9:46:56"} +{"current_steps": 41, "total_steps": 1955, "loss": 1.242, "lr": 1.673469387755102e-05, "epoch": 0.10485933503836317, "percentage": 2.1, "elapsed_time": "0:12:33", "remaining_time": "9:46:19"} +{"current_steps": 42, "total_steps": 1955, "loss": 1.2409, "lr": 1.7142857142857142e-05, "epoch": 0.10741687979539642, "percentage": 2.15, "elapsed_time": "0:12:51", "remaining_time": "9:45:44"} +{"current_steps": 43, "total_steps": 1955, "loss": 1.2136, "lr": 1.7551020408163266e-05, "epoch": 0.10997442455242967, "percentage": 2.2, "elapsed_time": "0:13:09", "remaining_time": "9:44:44"} +{"current_steps": 44, "total_steps": 1955, "loss": 1.2245, "lr": 1.795918367346939e-05, "epoch": 0.11253196930946291, "percentage": 2.25, "elapsed_time": "0:13:27", "remaining_time": "9:44:24"} +{"current_steps": 45, "total_steps": 1955, "loss": 1.2252, "lr": 1.836734693877551e-05, "epoch": 0.11508951406649616, "percentage": 2.3, "elapsed_time": "0:13:45", "remaining_time": "9:43:52"} +{"current_steps": 46, "total_steps": 1955, "loss": 1.2464, "lr": 1.8775510204081636e-05, "epoch": 0.11764705882352941, "percentage": 2.35, "elapsed_time": "0:14:03", "remaining_time": "9:43:21"} +{"current_steps": 47, "total_steps": 1955, "loss": 1.2466, "lr": 1.9183673469387756e-05, "epoch": 0.12020460358056266, "percentage": 2.4, "elapsed_time": "0:14:22", "remaining_time": "9:43:35"} +{"current_steps": 48, "total_steps": 1955, "loss": 1.2171, "lr": 1.9591836734693877e-05, "epoch": 0.12276214833759591, "percentage": 2.46, "elapsed_time": "0:14:40", "remaining_time": "9:43:04"} +{"current_steps": 49, "total_steps": 1955, "loss": 1.2317, "lr": 2e-05, "epoch": 0.12531969309462915, "percentage": 2.51, "elapsed_time": "0:14:58", "remaining_time": "9:42:13"} +{"current_steps": 50, "total_steps": 1955, "loss": 1.201, "lr": 2.0408163265306126e-05, "epoch": 0.1278772378516624, "percentage": 2.56, "elapsed_time": "0:15:16", "remaining_time": "9:41:45"} +{"current_steps": 51, "total_steps": 1955, "loss": 1.2083, "lr": 2.0816326530612247e-05, "epoch": 0.13043478260869565, "percentage": 2.61, "elapsed_time": "0:15:34", "remaining_time": "9:41:14"} +{"current_steps": 52, "total_steps": 1955, "loss": 1.189, "lr": 2.122448979591837e-05, "epoch": 0.1329923273657289, "percentage": 2.66, "elapsed_time": "0:15:52", "remaining_time": "9:41:14"} +{"current_steps": 53, "total_steps": 1955, "loss": 1.1904, "lr": 2.1632653061224492e-05, "epoch": 0.13554987212276215, "percentage": 2.71, "elapsed_time": "0:16:11", "remaining_time": "9:40:46"} +{"current_steps": 54, "total_steps": 1955, "loss": 1.2268, "lr": 2.2040816326530613e-05, "epoch": 0.13810741687979539, "percentage": 2.76, "elapsed_time": "0:16:29", "remaining_time": "9:40:28"} +{"current_steps": 55, "total_steps": 1955, "loss": 1.2391, "lr": 2.2448979591836734e-05, "epoch": 0.14066496163682865, "percentage": 2.81, "elapsed_time": "0:16:47", "remaining_time": "9:39:57"} +{"current_steps": 56, "total_steps": 1955, "loss": 1.227, "lr": 2.2857142857142858e-05, "epoch": 0.1432225063938619, "percentage": 2.86, "elapsed_time": "0:17:06", "remaining_time": "9:39:53"} +{"current_steps": 57, "total_steps": 1955, "loss": 1.2093, "lr": 2.3265306122448982e-05, "epoch": 0.14578005115089515, "percentage": 2.92, "elapsed_time": "0:17:24", "remaining_time": "9:39:26"} +{"current_steps": 58, "total_steps": 1955, "loss": 1.2253, "lr": 2.3673469387755103e-05, "epoch": 0.1483375959079284, "percentage": 2.97, "elapsed_time": "0:17:42", "remaining_time": "9:38:57"} +{"current_steps": 59, "total_steps": 1955, "loss": 1.191, "lr": 2.4081632653061227e-05, "epoch": 0.15089514066496162, "percentage": 3.02, "elapsed_time": "0:18:00", "remaining_time": "9:38:30"} +{"current_steps": 60, "total_steps": 1955, "loss": 1.2151, "lr": 2.448979591836735e-05, "epoch": 0.1534526854219949, "percentage": 3.07, "elapsed_time": "0:18:17", "remaining_time": "9:37:45"} +{"current_steps": 61, "total_steps": 1955, "loss": 1.2032, "lr": 2.4897959183673473e-05, "epoch": 0.15601023017902813, "percentage": 3.12, "elapsed_time": "0:18:35", "remaining_time": "9:37:18"} +{"current_steps": 62, "total_steps": 1955, "loss": 1.2352, "lr": 2.5306122448979597e-05, "epoch": 0.1585677749360614, "percentage": 3.17, "elapsed_time": "0:18:53", "remaining_time": "9:36:51"} +{"current_steps": 63, "total_steps": 1955, "loss": 1.2182, "lr": 2.5714285714285718e-05, "epoch": 0.16112531969309463, "percentage": 3.22, "elapsed_time": "0:19:11", "remaining_time": "9:36:26"} +{"current_steps": 64, "total_steps": 1955, "loss": 1.2128, "lr": 2.6122448979591835e-05, "epoch": 0.1636828644501279, "percentage": 3.27, "elapsed_time": "0:19:29", "remaining_time": "9:35:49"} +{"current_steps": 65, "total_steps": 1955, "loss": 1.2052, "lr": 2.653061224489796e-05, "epoch": 0.16624040920716113, "percentage": 3.32, "elapsed_time": "0:19:48", "remaining_time": "9:35:46"} +{"current_steps": 66, "total_steps": 1955, "loss": 1.1671, "lr": 2.6938775510204084e-05, "epoch": 0.16879795396419436, "percentage": 3.38, "elapsed_time": "0:20:06", "remaining_time": "9:35:36"} +{"current_steps": 67, "total_steps": 1955, "loss": 1.1915, "lr": 2.7346938775510205e-05, "epoch": 0.17135549872122763, "percentage": 3.43, "elapsed_time": "0:20:24", "remaining_time": "9:35:12"} +{"current_steps": 68, "total_steps": 1955, "loss": 1.1683, "lr": 2.775510204081633e-05, "epoch": 0.17391304347826086, "percentage": 3.48, "elapsed_time": "0:20:43", "remaining_time": "9:35:05"} +{"current_steps": 69, "total_steps": 1955, "loss": 1.152, "lr": 2.816326530612245e-05, "epoch": 0.17647058823529413, "percentage": 3.53, "elapsed_time": "0:21:01", "remaining_time": "9:34:40"} +{"current_steps": 70, "total_steps": 1955, "loss": 1.1638, "lr": 2.8571428571428574e-05, "epoch": 0.17902813299232737, "percentage": 3.58, "elapsed_time": "0:21:19", "remaining_time": "9:34:15"} +{"current_steps": 71, "total_steps": 1955, "loss": 1.1747, "lr": 2.89795918367347e-05, "epoch": 0.1815856777493606, "percentage": 3.63, "elapsed_time": "0:21:37", "remaining_time": "9:33:52"} +{"current_steps": 72, "total_steps": 1955, "loss": 1.1894, "lr": 2.938775510204082e-05, "epoch": 0.18414322250639387, "percentage": 3.68, "elapsed_time": "0:21:55", "remaining_time": "9:33:28"} +{"current_steps": 73, "total_steps": 1955, "loss": 1.2237, "lr": 2.9795918367346944e-05, "epoch": 0.1867007672634271, "percentage": 3.73, "elapsed_time": "0:22:13", "remaining_time": "9:33:03"} +{"current_steps": 74, "total_steps": 1955, "loss": 1.1695, "lr": 3.020408163265306e-05, "epoch": 0.18925831202046037, "percentage": 3.79, "elapsed_time": "0:22:31", "remaining_time": "9:32:39"} +{"current_steps": 75, "total_steps": 1955, "loss": 1.1671, "lr": 3.061224489795918e-05, "epoch": 0.1918158567774936, "percentage": 3.84, "elapsed_time": "0:22:49", "remaining_time": "9:32:15"} +{"current_steps": 76, "total_steps": 1955, "loss": 1.1705, "lr": 3.102040816326531e-05, "epoch": 0.19437340153452684, "percentage": 3.89, "elapsed_time": "0:23:07", "remaining_time": "9:31:52"} +{"current_steps": 77, "total_steps": 1955, "loss": 1.1808, "lr": 3.142857142857143e-05, "epoch": 0.1969309462915601, "percentage": 3.94, "elapsed_time": "0:23:25", "remaining_time": "9:31:29"} +{"current_steps": 78, "total_steps": 1955, "loss": 1.1616, "lr": 3.183673469387755e-05, "epoch": 0.19948849104859334, "percentage": 3.99, "elapsed_time": "0:23:43", "remaining_time": "9:31:05"} +{"current_steps": 79, "total_steps": 1955, "loss": 1.1554, "lr": 3.224489795918368e-05, "epoch": 0.2020460358056266, "percentage": 4.04, "elapsed_time": "0:24:01", "remaining_time": "9:30:41"} +{"current_steps": 80, "total_steps": 1955, "loss": 1.1725, "lr": 3.26530612244898e-05, "epoch": 0.20460358056265984, "percentage": 4.09, "elapsed_time": "0:24:19", "remaining_time": "9:30:05"} +{"current_steps": 81, "total_steps": 1955, "loss": 1.1561, "lr": 3.306122448979592e-05, "epoch": 0.2071611253196931, "percentage": 4.14, "elapsed_time": "0:24:38", "remaining_time": "9:29:57"} +{"current_steps": 82, "total_steps": 1955, "loss": 1.153, "lr": 3.346938775510204e-05, "epoch": 0.20971867007672634, "percentage": 4.19, "elapsed_time": "0:24:56", "remaining_time": "9:29:35"} +{"current_steps": 83, "total_steps": 1955, "loss": 1.1677, "lr": 3.387755102040817e-05, "epoch": 0.21227621483375958, "percentage": 4.25, "elapsed_time": "0:25:14", "remaining_time": "9:29:13"} +{"current_steps": 84, "total_steps": 1955, "loss": 1.1634, "lr": 3.4285714285714284e-05, "epoch": 0.21483375959079284, "percentage": 4.3, "elapsed_time": "0:25:32", "remaining_time": "9:28:51"} +{"current_steps": 85, "total_steps": 1955, "loss": 1.2139, "lr": 3.469387755102041e-05, "epoch": 0.21739130434782608, "percentage": 4.35, "elapsed_time": "0:25:50", "remaining_time": "9:28:28"} +{"current_steps": 86, "total_steps": 1955, "loss": 1.1666, "lr": 3.510204081632653e-05, "epoch": 0.21994884910485935, "percentage": 4.4, "elapsed_time": "0:26:08", "remaining_time": "9:28:05"} +{"current_steps": 87, "total_steps": 1955, "loss": 1.1727, "lr": 3.551020408163265e-05, "epoch": 0.22250639386189258, "percentage": 4.45, "elapsed_time": "0:26:26", "remaining_time": "9:27:42"} +{"current_steps": 88, "total_steps": 1955, "loss": 1.1958, "lr": 3.591836734693878e-05, "epoch": 0.22506393861892582, "percentage": 4.5, "elapsed_time": "0:26:44", "remaining_time": "9:27:19"} +{"current_steps": 89, "total_steps": 1955, "loss": 1.1561, "lr": 3.63265306122449e-05, "epoch": 0.22762148337595908, "percentage": 4.55, "elapsed_time": "0:27:02", "remaining_time": "9:26:57"} +{"current_steps": 90, "total_steps": 1955, "loss": 1.1866, "lr": 3.673469387755102e-05, "epoch": 0.23017902813299232, "percentage": 4.6, "elapsed_time": "0:27:20", "remaining_time": "9:26:34"} +{"current_steps": 91, "total_steps": 1955, "loss": 1.1831, "lr": 3.714285714285715e-05, "epoch": 0.23273657289002558, "percentage": 4.65, "elapsed_time": "0:27:38", "remaining_time": "9:26:12"} +{"current_steps": 92, "total_steps": 1955, "loss": 1.1865, "lr": 3.755102040816327e-05, "epoch": 0.23529411764705882, "percentage": 4.71, "elapsed_time": "0:27:56", "remaining_time": "9:25:58"} +{"current_steps": 93, "total_steps": 1955, "loss": 1.1676, "lr": 3.795918367346939e-05, "epoch": 0.23785166240409208, "percentage": 4.76, "elapsed_time": "0:28:15", "remaining_time": "9:25:36"} +{"current_steps": 94, "total_steps": 1955, "loss": 1.1884, "lr": 3.836734693877551e-05, "epoch": 0.24040920716112532, "percentage": 4.81, "elapsed_time": "0:28:33", "remaining_time": "9:25:14"} +{"current_steps": 95, "total_steps": 1955, "loss": 1.1359, "lr": 3.8775510204081634e-05, "epoch": 0.24296675191815856, "percentage": 4.86, "elapsed_time": "0:28:51", "remaining_time": "9:24:52"} +{"current_steps": 96, "total_steps": 1955, "loss": 1.1769, "lr": 3.9183673469387755e-05, "epoch": 0.24552429667519182, "percentage": 4.91, "elapsed_time": "0:29:08", "remaining_time": "9:24:27"} +{"current_steps": 97, "total_steps": 1955, "loss": 1.18, "lr": 3.959183673469388e-05, "epoch": 0.24808184143222506, "percentage": 4.96, "elapsed_time": "0:29:26", "remaining_time": "9:24:05"} +{"current_steps": 98, "total_steps": 1955, "loss": 1.1508, "lr": 4e-05, "epoch": 0.2506393861892583, "percentage": 5.01, "elapsed_time": "0:29:45", "remaining_time": "9:23:44"} +{"current_steps": 99, "total_steps": 1955, "loss": 1.1473, "lr": 4.0408163265306124e-05, "epoch": 0.2531969309462916, "percentage": 5.06, "elapsed_time": "0:30:03", "remaining_time": "9:23:22"} +{"current_steps": 100, "total_steps": 1955, "loss": 1.1413, "lr": 4.081632653061225e-05, "epoch": 0.2557544757033248, "percentage": 5.12, "elapsed_time": "0:30:21", "remaining_time": "9:23:00"} +{"current_steps": 101, "total_steps": 1955, "loss": 1.1647, "lr": 4.1224489795918366e-05, "epoch": 0.25831202046035806, "percentage": 5.17, "elapsed_time": "0:30:39", "remaining_time": "9:22:53"} +{"current_steps": 102, "total_steps": 1955, "loss": 1.1698, "lr": 4.1632653061224494e-05, "epoch": 0.2608695652173913, "percentage": 5.22, "elapsed_time": "0:30:57", "remaining_time": "9:22:32"} +{"current_steps": 103, "total_steps": 1955, "loss": 1.1374, "lr": 4.2040816326530615e-05, "epoch": 0.26342710997442453, "percentage": 5.27, "elapsed_time": "0:31:15", "remaining_time": "9:22:10"} +{"current_steps": 104, "total_steps": 1955, "loss": 1.1295, "lr": 4.244897959183674e-05, "epoch": 0.2659846547314578, "percentage": 5.32, "elapsed_time": "0:31:33", "remaining_time": "9:21:48"} +{"current_steps": 105, "total_steps": 1955, "loss": 1.1397, "lr": 4.2857142857142856e-05, "epoch": 0.26854219948849106, "percentage": 5.37, "elapsed_time": "0:31:51", "remaining_time": "9:21:26"} +{"current_steps": 106, "total_steps": 1955, "loss": 1.1273, "lr": 4.3265306122448984e-05, "epoch": 0.2710997442455243, "percentage": 5.42, "elapsed_time": "0:32:09", "remaining_time": "9:21:04"} +{"current_steps": 107, "total_steps": 1955, "loss": 1.1544, "lr": 4.3673469387755105e-05, "epoch": 0.27365728900255754, "percentage": 5.47, "elapsed_time": "0:32:27", "remaining_time": "9:20:43"} +{"current_steps": 108, "total_steps": 1955, "loss": 1.134, "lr": 4.4081632653061226e-05, "epoch": 0.27621483375959077, "percentage": 5.52, "elapsed_time": "0:32:46", "remaining_time": "9:20:22"} +{"current_steps": 109, "total_steps": 1955, "loss": 1.1582, "lr": 4.4489795918367353e-05, "epoch": 0.27877237851662406, "percentage": 5.58, "elapsed_time": "0:33:04", "remaining_time": "9:20:01"} +{"current_steps": 110, "total_steps": 1955, "loss": 1.1614, "lr": 4.489795918367347e-05, "epoch": 0.2813299232736573, "percentage": 5.63, "elapsed_time": "0:33:22", "remaining_time": "9:19:40"} +{"current_steps": 111, "total_steps": 1955, "loss": 1.162, "lr": 4.5306122448979595e-05, "epoch": 0.28388746803069054, "percentage": 5.68, "elapsed_time": "0:33:40", "remaining_time": "9:19:25"} +{"current_steps": 112, "total_steps": 1955, "loss": 1.1508, "lr": 4.5714285714285716e-05, "epoch": 0.2864450127877238, "percentage": 5.73, "elapsed_time": "0:33:58", "remaining_time": "9:19:00"} +{"current_steps": 113, "total_steps": 1955, "loss": 1.1246, "lr": 4.6122448979591844e-05, "epoch": 0.289002557544757, "percentage": 5.78, "elapsed_time": "0:34:17", "remaining_time": "9:19:00"} +{"current_steps": 114, "total_steps": 1955, "loss": 1.155, "lr": 4.6530612244897965e-05, "epoch": 0.2915601023017903, "percentage": 5.83, "elapsed_time": "0:34:35", "remaining_time": "9:18:39"} +{"current_steps": 115, "total_steps": 1955, "loss": 1.1396, "lr": 4.6938775510204086e-05, "epoch": 0.29411764705882354, "percentage": 5.88, "elapsed_time": "0:34:53", "remaining_time": "9:18:18"} +{"current_steps": 116, "total_steps": 1955, "loss": 1.146, "lr": 4.7346938775510206e-05, "epoch": 0.2966751918158568, "percentage": 5.93, "elapsed_time": "0:35:11", "remaining_time": "9:17:58"} +{"current_steps": 117, "total_steps": 1955, "loss": 1.1366, "lr": 4.7755102040816334e-05, "epoch": 0.29923273657289, "percentage": 5.98, "elapsed_time": "0:35:29", "remaining_time": "9:17:30"} +{"current_steps": 118, "total_steps": 1955, "loss": 1.1863, "lr": 4.8163265306122455e-05, "epoch": 0.30179028132992325, "percentage": 6.04, "elapsed_time": "0:35:47", "remaining_time": "9:17:09"} +{"current_steps": 119, "total_steps": 1955, "loss": 1.1481, "lr": 4.857142857142857e-05, "epoch": 0.30434782608695654, "percentage": 6.09, "elapsed_time": "0:36:05", "remaining_time": "9:16:48"} +{"current_steps": 120, "total_steps": 1955, "loss": 1.1574, "lr": 4.89795918367347e-05, "epoch": 0.3069053708439898, "percentage": 6.14, "elapsed_time": "0:36:22", "remaining_time": "9:16:19"} +{"current_steps": 121, "total_steps": 1955, "loss": 1.1701, "lr": 4.938775510204082e-05, "epoch": 0.309462915601023, "percentage": 6.19, "elapsed_time": "0:36:40", "remaining_time": "9:15:59"} +{"current_steps": 122, "total_steps": 1955, "loss": 1.1204, "lr": 4.9795918367346945e-05, "epoch": 0.31202046035805625, "percentage": 6.24, "elapsed_time": "0:36:58", "remaining_time": "9:15:39"} +{"current_steps": 123, "total_steps": 1955, "loss": 1.1424, "lr": 5.0204081632653066e-05, "epoch": 0.3145780051150895, "percentage": 6.29, "elapsed_time": "0:37:17", "remaining_time": "9:15:18"} +{"current_steps": 124, "total_steps": 1955, "loss": 1.1374, "lr": 5.0612244897959194e-05, "epoch": 0.3171355498721228, "percentage": 6.34, "elapsed_time": "0:37:35", "remaining_time": "9:14:57"} +{"current_steps": 125, "total_steps": 1955, "loss": 1.1181, "lr": 5.102040816326531e-05, "epoch": 0.319693094629156, "percentage": 6.39, "elapsed_time": "0:37:53", "remaining_time": "9:14:37"} +{"current_steps": 126, "total_steps": 1955, "loss": 1.1755, "lr": 5.1428571428571436e-05, "epoch": 0.32225063938618925, "percentage": 6.45, "elapsed_time": "0:38:12", "remaining_time": "9:14:38"} +{"current_steps": 127, "total_steps": 1955, "loss": 1.12, "lr": 5.1836734693877557e-05, "epoch": 0.3248081841432225, "percentage": 6.5, "elapsed_time": "0:38:30", "remaining_time": "9:14:18"} +{"current_steps": 128, "total_steps": 1955, "loss": 1.1371, "lr": 5.224489795918367e-05, "epoch": 0.3273657289002558, "percentage": 6.55, "elapsed_time": "0:38:48", "remaining_time": "9:13:58"} +{"current_steps": 129, "total_steps": 1955, "loss": 1.1289, "lr": 5.26530612244898e-05, "epoch": 0.329923273657289, "percentage": 6.6, "elapsed_time": "0:39:06", "remaining_time": "9:13:37"} +{"current_steps": 130, "total_steps": 1955, "loss": 1.1588, "lr": 5.306122448979592e-05, "epoch": 0.33248081841432225, "percentage": 6.65, "elapsed_time": "0:39:24", "remaining_time": "9:13:17"} +{"current_steps": 131, "total_steps": 1955, "loss": 1.1252, "lr": 5.346938775510205e-05, "epoch": 0.3350383631713555, "percentage": 6.7, "elapsed_time": "0:39:42", "remaining_time": "9:12:56"} +{"current_steps": 132, "total_steps": 1955, "loss": 1.1789, "lr": 5.387755102040817e-05, "epoch": 0.3375959079283887, "percentage": 6.75, "elapsed_time": "0:40:00", "remaining_time": "9:12:36"} +{"current_steps": 133, "total_steps": 1955, "loss": 1.1548, "lr": 5.4285714285714295e-05, "epoch": 0.340153452685422, "percentage": 6.8, "elapsed_time": "0:40:19", "remaining_time": "9:12:28"} +{"current_steps": 134, "total_steps": 1955, "loss": 1.1259, "lr": 5.469387755102041e-05, "epoch": 0.34271099744245526, "percentage": 6.85, "elapsed_time": "0:40:37", "remaining_time": "9:12:08"} +{"current_steps": 135, "total_steps": 1955, "loss": 1.1825, "lr": 5.510204081632654e-05, "epoch": 0.3452685421994885, "percentage": 6.91, "elapsed_time": "0:40:56", "remaining_time": "9:11:51"} +{"current_steps": 136, "total_steps": 1955, "loss": 1.1453, "lr": 5.551020408163266e-05, "epoch": 0.34782608695652173, "percentage": 6.96, "elapsed_time": "0:41:14", "remaining_time": "9:11:31"} +{"current_steps": 137, "total_steps": 1955, "loss": 1.1554, "lr": 5.5918367346938786e-05, "epoch": 0.35038363171355497, "percentage": 7.01, "elapsed_time": "0:41:32", "remaining_time": "9:11:11"} +{"current_steps": 138, "total_steps": 1955, "loss": 1.1429, "lr": 5.63265306122449e-05, "epoch": 0.35294117647058826, "percentage": 7.06, "elapsed_time": "0:41:51", "remaining_time": "9:11:05"} +{"current_steps": 139, "total_steps": 1955, "loss": 1.116, "lr": 5.673469387755102e-05, "epoch": 0.3554987212276215, "percentage": 7.11, "elapsed_time": "0:42:09", "remaining_time": "9:10:45"} +{"current_steps": 140, "total_steps": 1955, "loss": 1.176, "lr": 5.714285714285715e-05, "epoch": 0.35805626598465473, "percentage": 7.16, "elapsed_time": "0:42:27", "remaining_time": "9:10:31"} +{"current_steps": 141, "total_steps": 1955, "loss": 1.0897, "lr": 5.755102040816327e-05, "epoch": 0.36061381074168797, "percentage": 7.21, "elapsed_time": "0:42:45", "remaining_time": "9:10:10"} +{"current_steps": 142, "total_steps": 1955, "loss": 1.1411, "lr": 5.79591836734694e-05, "epoch": 0.3631713554987212, "percentage": 7.26, "elapsed_time": "0:43:03", "remaining_time": "9:09:49"} +{"current_steps": 143, "total_steps": 1955, "loss": 1.1689, "lr": 5.836734693877551e-05, "epoch": 0.3657289002557545, "percentage": 7.31, "elapsed_time": "0:43:22", "remaining_time": "9:09:43"} +{"current_steps": 144, "total_steps": 1955, "loss": 1.1345, "lr": 5.877551020408164e-05, "epoch": 0.36828644501278773, "percentage": 7.37, "elapsed_time": "0:43:40", "remaining_time": "9:09:21"} +{"current_steps": 145, "total_steps": 1955, "loss": 1.1466, "lr": 5.918367346938776e-05, "epoch": 0.37084398976982097, "percentage": 7.42, "elapsed_time": "0:43:58", "remaining_time": "9:09:01"} +{"current_steps": 146, "total_steps": 1955, "loss": 1.1303, "lr": 5.959183673469389e-05, "epoch": 0.3734015345268542, "percentage": 7.47, "elapsed_time": "0:44:17", "remaining_time": "9:08:41"} +{"current_steps": 147, "total_steps": 1955, "loss": 1.1675, "lr": 6.000000000000001e-05, "epoch": 0.37595907928388744, "percentage": 7.52, "elapsed_time": "0:44:34", "remaining_time": "9:08:15"} +{"current_steps": 148, "total_steps": 1955, "loss": 1.13, "lr": 6.040816326530612e-05, "epoch": 0.37851662404092073, "percentage": 7.57, "elapsed_time": "0:44:52", "remaining_time": "9:07:56"} +{"current_steps": 149, "total_steps": 1955, "loss": 1.1145, "lr": 6.081632653061225e-05, "epoch": 0.38107416879795397, "percentage": 7.62, "elapsed_time": "0:45:10", "remaining_time": "9:07:36"} +{"current_steps": 150, "total_steps": 1955, "loss": 1.1266, "lr": 6.122448979591836e-05, "epoch": 0.3836317135549872, "percentage": 7.67, "elapsed_time": "0:45:28", "remaining_time": "9:07:17"} +{"current_steps": 151, "total_steps": 1955, "loss": 1.129, "lr": 6.163265306122449e-05, "epoch": 0.38618925831202044, "percentage": 7.72, "elapsed_time": "0:45:47", "remaining_time": "9:07:03"} +{"current_steps": 152, "total_steps": 1955, "loss": 1.1361, "lr": 6.204081632653062e-05, "epoch": 0.3887468030690537, "percentage": 7.77, "elapsed_time": "0:46:05", "remaining_time": "9:06:43"} +{"current_steps": 153, "total_steps": 1955, "loss": 1.1266, "lr": 6.244897959183675e-05, "epoch": 0.391304347826087, "percentage": 7.83, "elapsed_time": "0:46:23", "remaining_time": "9:06:23"} +{"current_steps": 154, "total_steps": 1955, "loss": 1.1343, "lr": 6.285714285714286e-05, "epoch": 0.3938618925831202, "percentage": 7.88, "elapsed_time": "0:46:41", "remaining_time": "9:05:59"} +{"current_steps": 155, "total_steps": 1955, "loss": 1.1463, "lr": 6.326530612244899e-05, "epoch": 0.39641943734015345, "percentage": 7.93, "elapsed_time": "0:46:59", "remaining_time": "9:05:40"} +{"current_steps": 156, "total_steps": 1955, "loss": 1.1454, "lr": 6.36734693877551e-05, "epoch": 0.3989769820971867, "percentage": 7.98, "elapsed_time": "0:47:17", "remaining_time": "9:05:20"} +{"current_steps": 157, "total_steps": 1955, "loss": 1.1616, "lr": 6.408163265306123e-05, "epoch": 0.40153452685422, "percentage": 8.03, "elapsed_time": "0:47:35", "remaining_time": "9:05:05"} +{"current_steps": 158, "total_steps": 1955, "loss": 1.1545, "lr": 6.448979591836736e-05, "epoch": 0.4040920716112532, "percentage": 8.08, "elapsed_time": "0:47:53", "remaining_time": "9:04:43"} +{"current_steps": 159, "total_steps": 1955, "loss": 1.1052, "lr": 6.489795918367347e-05, "epoch": 0.40664961636828645, "percentage": 8.13, "elapsed_time": "0:48:11", "remaining_time": "9:04:23"} +{"current_steps": 160, "total_steps": 1955, "loss": 1.1264, "lr": 6.53061224489796e-05, "epoch": 0.4092071611253197, "percentage": 8.18, "elapsed_time": "0:48:29", "remaining_time": "9:04:03"} +{"current_steps": 161, "total_steps": 1955, "loss": 1.12, "lr": 6.571428571428571e-05, "epoch": 0.4117647058823529, "percentage": 8.24, "elapsed_time": "0:48:47", "remaining_time": "9:03:43"} +{"current_steps": 162, "total_steps": 1955, "loss": 1.1171, "lr": 6.612244897959184e-05, "epoch": 0.4143222506393862, "percentage": 8.29, "elapsed_time": "0:49:06", "remaining_time": "9:03:26"} +{"current_steps": 163, "total_steps": 1955, "loss": 1.1196, "lr": 6.653061224489797e-05, "epoch": 0.41687979539641945, "percentage": 8.34, "elapsed_time": "0:49:24", "remaining_time": "9:03:06"} +{"current_steps": 164, "total_steps": 1955, "loss": 1.1316, "lr": 6.693877551020408e-05, "epoch": 0.4194373401534527, "percentage": 8.39, "elapsed_time": "0:49:42", "remaining_time": "9:02:47"} +{"current_steps": 165, "total_steps": 1955, "loss": 1.107, "lr": 6.734693877551021e-05, "epoch": 0.4219948849104859, "percentage": 8.44, "elapsed_time": "0:50:00", "remaining_time": "9:02:27"} +{"current_steps": 166, "total_steps": 1955, "loss": 1.1294, "lr": 6.775510204081634e-05, "epoch": 0.42455242966751916, "percentage": 8.49, "elapsed_time": "0:50:18", "remaining_time": "9:02:07"} +{"current_steps": 167, "total_steps": 1955, "loss": 1.1245, "lr": 6.816326530612245e-05, "epoch": 0.42710997442455245, "percentage": 8.54, "elapsed_time": "0:50:36", "remaining_time": "9:01:50"} +{"current_steps": 168, "total_steps": 1955, "loss": 1.1338, "lr": 6.857142857142857e-05, "epoch": 0.4296675191815857, "percentage": 8.59, "elapsed_time": "0:50:54", "remaining_time": "9:01:34"} +{"current_steps": 169, "total_steps": 1955, "loss": 1.1332, "lr": 6.89795918367347e-05, "epoch": 0.4322250639386189, "percentage": 8.64, "elapsed_time": "0:51:13", "remaining_time": "9:01:21"} +{"current_steps": 170, "total_steps": 1955, "loss": 1.1127, "lr": 6.938775510204082e-05, "epoch": 0.43478260869565216, "percentage": 8.7, "elapsed_time": "0:51:31", "remaining_time": "9:01:01"} +{"current_steps": 171, "total_steps": 1955, "loss": 1.1285, "lr": 6.979591836734695e-05, "epoch": 0.4373401534526854, "percentage": 8.75, "elapsed_time": "0:51:49", "remaining_time": "9:00:44"} +{"current_steps": 172, "total_steps": 1955, "loss": 1.1223, "lr": 7.020408163265306e-05, "epoch": 0.4398976982097187, "percentage": 8.8, "elapsed_time": "0:52:07", "remaining_time": "9:00:25"} +{"current_steps": 173, "total_steps": 1955, "loss": 1.142, "lr": 7.061224489795919e-05, "epoch": 0.4424552429667519, "percentage": 8.85, "elapsed_time": "0:52:26", "remaining_time": "9:00:05"} +{"current_steps": 174, "total_steps": 1955, "loss": 1.151, "lr": 7.10204081632653e-05, "epoch": 0.44501278772378516, "percentage": 8.9, "elapsed_time": "0:52:44", "remaining_time": "8:59:46"} +{"current_steps": 175, "total_steps": 1955, "loss": 1.1235, "lr": 7.142857142857143e-05, "epoch": 0.4475703324808184, "percentage": 8.95, "elapsed_time": "0:53:02", "remaining_time": "8:59:27"} +{"current_steps": 176, "total_steps": 1955, "loss": 1.1553, "lr": 7.183673469387756e-05, "epoch": 0.45012787723785164, "percentage": 9.0, "elapsed_time": "0:53:20", "remaining_time": "8:59:07"} +{"current_steps": 177, "total_steps": 1955, "loss": 1.1281, "lr": 7.224489795918368e-05, "epoch": 0.45268542199488493, "percentage": 9.05, "elapsed_time": "0:53:38", "remaining_time": "8:58:48"} +{"current_steps": 178, "total_steps": 1955, "loss": 1.1267, "lr": 7.26530612244898e-05, "epoch": 0.45524296675191817, "percentage": 9.1, "elapsed_time": "0:53:56", "remaining_time": "8:58:25"} +{"current_steps": 179, "total_steps": 1955, "loss": 1.1147, "lr": 7.306122448979592e-05, "epoch": 0.4578005115089514, "percentage": 9.16, "elapsed_time": "0:54:14", "remaining_time": "8:58:05"} +{"current_steps": 180, "total_steps": 1955, "loss": 1.1197, "lr": 7.346938775510205e-05, "epoch": 0.46035805626598464, "percentage": 9.21, "elapsed_time": "0:54:31", "remaining_time": "8:57:38"} +{"current_steps": 181, "total_steps": 1955, "loss": 1.1482, "lr": 7.387755102040817e-05, "epoch": 0.4629156010230179, "percentage": 9.26, "elapsed_time": "0:54:49", "remaining_time": "8:57:18"} +{"current_steps": 182, "total_steps": 1955, "loss": 1.1206, "lr": 7.42857142857143e-05, "epoch": 0.46547314578005117, "percentage": 9.31, "elapsed_time": "0:55:07", "remaining_time": "8:56:58"} +{"current_steps": 183, "total_steps": 1955, "loss": 1.1366, "lr": 7.469387755102041e-05, "epoch": 0.4680306905370844, "percentage": 9.36, "elapsed_time": "0:55:26", "remaining_time": "8:56:46"} +{"current_steps": 184, "total_steps": 1955, "loss": 1.1294, "lr": 7.510204081632654e-05, "epoch": 0.47058823529411764, "percentage": 9.41, "elapsed_time": "0:55:44", "remaining_time": "8:56:26"} +{"current_steps": 185, "total_steps": 1955, "loss": 1.1395, "lr": 7.551020408163266e-05, "epoch": 0.4731457800511509, "percentage": 9.46, "elapsed_time": "0:56:02", "remaining_time": "8:56:06"} +{"current_steps": 186, "total_steps": 1955, "loss": 1.1332, "lr": 7.591836734693878e-05, "epoch": 0.47570332480818417, "percentage": 9.51, "elapsed_time": "0:56:20", "remaining_time": "8:55:47"} +{"current_steps": 187, "total_steps": 1955, "loss": 1.1187, "lr": 7.63265306122449e-05, "epoch": 0.4782608695652174, "percentage": 9.57, "elapsed_time": "0:56:38", "remaining_time": "8:55:28"} +{"current_steps": 188, "total_steps": 1955, "loss": 1.1428, "lr": 7.673469387755103e-05, "epoch": 0.48081841432225064, "percentage": 9.62, "elapsed_time": "0:56:56", "remaining_time": "8:55:09"} +{"current_steps": 189, "total_steps": 1955, "loss": 1.126, "lr": 7.714285714285715e-05, "epoch": 0.4833759590792839, "percentage": 9.67, "elapsed_time": "0:57:14", "remaining_time": "8:54:50"} +{"current_steps": 190, "total_steps": 1955, "loss": 1.1262, "lr": 7.755102040816327e-05, "epoch": 0.4859335038363171, "percentage": 9.72, "elapsed_time": "0:57:32", "remaining_time": "8:54:30"} +{"current_steps": 191, "total_steps": 1955, "loss": 1.0769, "lr": 7.79591836734694e-05, "epoch": 0.4884910485933504, "percentage": 9.77, "elapsed_time": "0:57:50", "remaining_time": "8:54:11"} +{"current_steps": 192, "total_steps": 1955, "loss": 1.1177, "lr": 7.836734693877551e-05, "epoch": 0.49104859335038364, "percentage": 9.82, "elapsed_time": "0:58:08", "remaining_time": "8:53:51"} +{"current_steps": 193, "total_steps": 1955, "loss": 1.0895, "lr": 7.877551020408164e-05, "epoch": 0.4936061381074169, "percentage": 9.87, "elapsed_time": "0:58:27", "remaining_time": "8:53:44"} +{"current_steps": 194, "total_steps": 1955, "loss": 1.0864, "lr": 7.918367346938776e-05, "epoch": 0.4961636828644501, "percentage": 9.92, "elapsed_time": "0:58:46", "remaining_time": "8:53:29"} +{"current_steps": 195, "total_steps": 1955, "loss": 1.1628, "lr": 7.959183673469389e-05, "epoch": 0.49872122762148335, "percentage": 9.97, "elapsed_time": "0:59:04", "remaining_time": "8:53:10"} +{"current_steps": 196, "total_steps": 1955, "loss": 1.1215, "lr": 8e-05, "epoch": 0.5012787723785166, "percentage": 10.03, "elapsed_time": "0:59:22", "remaining_time": "8:52:50"} +{"current_steps": 197, "total_steps": 1955, "loss": 1.1072, "lr": 7.999993620333937e-05, "epoch": 0.5038363171355499, "percentage": 10.08, "elapsed_time": "0:59:41", "remaining_time": "8:52:39"} +{"current_steps": 198, "total_steps": 1955, "loss": 1.1389, "lr": 7.999974481356096e-05, "epoch": 0.5063938618925832, "percentage": 10.13, "elapsed_time": "0:59:59", "remaining_time": "8:52:19"} +{"current_steps": 199, "total_steps": 1955, "loss": 1.1039, "lr": 7.999942583127525e-05, "epoch": 0.5089514066496164, "percentage": 10.18, "elapsed_time": "1:00:17", "remaining_time": "8:51:59"} +{"current_steps": 200, "total_steps": 1955, "loss": 1.1205, "lr": 7.999897925749977e-05, "epoch": 0.5115089514066496, "percentage": 10.23, "elapsed_time": "1:00:35", "remaining_time": "8:51:40"} +{"current_steps": 201, "total_steps": 1955, "loss": 1.0991, "lr": 7.9998405093659e-05, "epoch": 0.5140664961636828, "percentage": 10.28, "elapsed_time": "1:00:53", "remaining_time": "8:51:22"} +{"current_steps": 202, "total_steps": 1955, "loss": 1.0939, "lr": 7.999770334158445e-05, "epoch": 0.5166240409207161, "percentage": 10.33, "elapsed_time": "1:01:11", "remaining_time": "8:51:00"} +{"current_steps": 203, "total_steps": 1955, "loss": 1.1263, "lr": 7.999687400351458e-05, "epoch": 0.5191815856777494, "percentage": 10.38, "elapsed_time": "1:01:29", "remaining_time": "8:50:41"} +{"current_steps": 204, "total_steps": 1955, "loss": 1.0945, "lr": 7.999591708209482e-05, "epoch": 0.5217391304347826, "percentage": 10.43, "elapsed_time": "1:01:47", "remaining_time": "8:50:21"} +{"current_steps": 205, "total_steps": 1955, "loss": 1.1189, "lr": 7.99948325803776e-05, "epoch": 0.5242966751918159, "percentage": 10.49, "elapsed_time": "1:02:05", "remaining_time": "8:50:02"} +{"current_steps": 206, "total_steps": 1955, "loss": 1.1079, "lr": 7.999362050182232e-05, "epoch": 0.5268542199488491, "percentage": 10.54, "elapsed_time": "1:02:22", "remaining_time": "8:49:36"} +{"current_steps": 207, "total_steps": 1955, "loss": 1.1252, "lr": 7.999228085029529e-05, "epoch": 0.5294117647058824, "percentage": 10.59, "elapsed_time": "1:02:40", "remaining_time": "8:49:17"} +{"current_steps": 208, "total_steps": 1955, "loss": 1.1119, "lr": 7.999081363006977e-05, "epoch": 0.5319693094629157, "percentage": 10.64, "elapsed_time": "1:02:59", "remaining_time": "8:49:00"} +{"current_steps": 209, "total_steps": 1955, "loss": 1.0998, "lr": 7.998921884582596e-05, "epoch": 0.5345268542199488, "percentage": 10.69, "elapsed_time": "1:03:16", "remaining_time": "8:48:35"} +{"current_steps": 210, "total_steps": 1955, "loss": 1.0953, "lr": 7.998749650265096e-05, "epoch": 0.5370843989769821, "percentage": 10.74, "elapsed_time": "1:03:34", "remaining_time": "8:48:15"} +{"current_steps": 211, "total_steps": 1955, "loss": 1.1146, "lr": 7.998564660603873e-05, "epoch": 0.5396419437340153, "percentage": 10.79, "elapsed_time": "1:03:52", "remaining_time": "8:47:56"} +{"current_steps": 212, "total_steps": 1955, "loss": 1.0821, "lr": 7.998366916189016e-05, "epoch": 0.5421994884910486, "percentage": 10.84, "elapsed_time": "1:04:10", "remaining_time": "8:47:37"} +{"current_steps": 213, "total_steps": 1955, "loss": 1.1258, "lr": 7.998156417651297e-05, "epoch": 0.5447570332480819, "percentage": 10.9, "elapsed_time": "1:04:28", "remaining_time": "8:47:18"} +{"current_steps": 214, "total_steps": 1955, "loss": 1.0766, "lr": 7.997933165662167e-05, "epoch": 0.5473145780051151, "percentage": 10.95, "elapsed_time": "1:04:46", "remaining_time": "8:46:59"} +{"current_steps": 215, "total_steps": 1955, "loss": 1.0825, "lr": 7.997697160933767e-05, "epoch": 0.5498721227621484, "percentage": 11.0, "elapsed_time": "1:05:04", "remaining_time": "8:46:40"} +{"current_steps": 216, "total_steps": 1955, "loss": 1.1183, "lr": 7.99744840421891e-05, "epoch": 0.5524296675191815, "percentage": 11.05, "elapsed_time": "1:05:22", "remaining_time": "8:46:20"} +{"current_steps": 217, "total_steps": 1955, "loss": 1.0898, "lr": 7.997186896311091e-05, "epoch": 0.5549872122762148, "percentage": 11.1, "elapsed_time": "1:05:40", "remaining_time": "8:46:01"} +{"current_steps": 218, "total_steps": 1955, "loss": 1.0641, "lr": 7.996912638044474e-05, "epoch": 0.5575447570332481, "percentage": 11.15, "elapsed_time": "1:05:58", "remaining_time": "8:45:42"} +{"current_steps": 219, "total_steps": 1955, "loss": 1.087, "lr": 7.996625630293898e-05, "epoch": 0.5601023017902813, "percentage": 11.2, "elapsed_time": "1:06:16", "remaining_time": "8:45:23"} +{"current_steps": 220, "total_steps": 1955, "loss": 1.1414, "lr": 7.996325873974871e-05, "epoch": 0.5626598465473146, "percentage": 11.25, "elapsed_time": "1:06:34", "remaining_time": "8:45:05"} +{"current_steps": 221, "total_steps": 1955, "loss": 1.0817, "lr": 7.996013370043562e-05, "epoch": 0.5652173913043478, "percentage": 11.3, "elapsed_time": "1:06:52", "remaining_time": "8:44:46"} +{"current_steps": 222, "total_steps": 1955, "loss": 1.0851, "lr": 7.995688119496811e-05, "epoch": 0.5677749360613811, "percentage": 11.36, "elapsed_time": "1:07:10", "remaining_time": "8:44:27"} +{"current_steps": 223, "total_steps": 1955, "loss": 1.0933, "lr": 7.99535012337211e-05, "epoch": 0.5703324808184144, "percentage": 11.41, "elapsed_time": "1:07:29", "remaining_time": "8:44:08"} +{"current_steps": 224, "total_steps": 1955, "loss": 1.1196, "lr": 7.99499938274761e-05, "epoch": 0.5728900255754475, "percentage": 11.46, "elapsed_time": "1:07:47", "remaining_time": "8:43:53"} +{"current_steps": 225, "total_steps": 1955, "loss": 1.1199, "lr": 7.994635898742118e-05, "epoch": 0.5754475703324808, "percentage": 11.51, "elapsed_time": "1:08:05", "remaining_time": "8:43:33"} +{"current_steps": 226, "total_steps": 1955, "loss": 1.0922, "lr": 7.994259672515083e-05, "epoch": 0.578005115089514, "percentage": 11.56, "elapsed_time": "1:08:23", "remaining_time": "8:43:14"} +{"current_steps": 227, "total_steps": 1955, "loss": 1.1012, "lr": 7.993870705266608e-05, "epoch": 0.5805626598465473, "percentage": 11.61, "elapsed_time": "1:08:41", "remaining_time": "8:42:56"} +{"current_steps": 228, "total_steps": 1955, "loss": 1.086, "lr": 7.99346899823743e-05, "epoch": 0.5831202046035806, "percentage": 11.66, "elapsed_time": "1:08:59", "remaining_time": "8:42:37"} +{"current_steps": 229, "total_steps": 1955, "loss": 1.1069, "lr": 7.993054552708931e-05, "epoch": 0.5856777493606138, "percentage": 11.71, "elapsed_time": "1:09:17", "remaining_time": "8:42:18"} +{"current_steps": 230, "total_steps": 1955, "loss": 1.1105, "lr": 7.99262737000312e-05, "epoch": 0.5882352941176471, "percentage": 11.76, "elapsed_time": "1:09:35", "remaining_time": "8:41:58"} +{"current_steps": 231, "total_steps": 1955, "loss": 1.1132, "lr": 7.992187451482642e-05, "epoch": 0.5907928388746803, "percentage": 11.82, "elapsed_time": "1:09:53", "remaining_time": "8:41:39"} +{"current_steps": 232, "total_steps": 1955, "loss": 1.0944, "lr": 7.991734798550758e-05, "epoch": 0.5933503836317136, "percentage": 11.87, "elapsed_time": "1:10:13", "remaining_time": "8:41:29"} +{"current_steps": 233, "total_steps": 1955, "loss": 1.1008, "lr": 7.99126941265136e-05, "epoch": 0.5959079283887468, "percentage": 11.92, "elapsed_time": "1:10:31", "remaining_time": "8:41:09"} +{"current_steps": 234, "total_steps": 1955, "loss": 1.0579, "lr": 7.990791295268949e-05, "epoch": 0.59846547314578, "percentage": 11.97, "elapsed_time": "1:10:49", "remaining_time": "8:40:50"} +{"current_steps": 235, "total_steps": 1955, "loss": 1.0766, "lr": 7.99030044792864e-05, "epoch": 0.6010230179028133, "percentage": 12.02, "elapsed_time": "1:11:07", "remaining_time": "8:40:31"} +{"current_steps": 236, "total_steps": 1955, "loss": 1.1069, "lr": 7.989796872196156e-05, "epoch": 0.6035805626598465, "percentage": 12.07, "elapsed_time": "1:11:25", "remaining_time": "8:40:13"} +{"current_steps": 237, "total_steps": 1955, "loss": 1.0935, "lr": 7.989280569677817e-05, "epoch": 0.6061381074168798, "percentage": 12.12, "elapsed_time": "1:11:44", "remaining_time": "8:40:04"} +{"current_steps": 238, "total_steps": 1955, "loss": 1.0755, "lr": 7.988751542020543e-05, "epoch": 0.6086956521739131, "percentage": 12.17, "elapsed_time": "1:12:02", "remaining_time": "8:39:44"} +{"current_steps": 239, "total_steps": 1955, "loss": 1.1183, "lr": 7.988209790911844e-05, "epoch": 0.6112531969309463, "percentage": 12.23, "elapsed_time": "1:12:20", "remaining_time": "8:39:25"} +{"current_steps": 240, "total_steps": 1955, "loss": 1.084, "lr": 7.987655318079816e-05, "epoch": 0.6138107416879796, "percentage": 12.28, "elapsed_time": "1:12:38", "remaining_time": "8:39:05"} +{"current_steps": 241, "total_steps": 1955, "loss": 1.0765, "lr": 7.987088125293133e-05, "epoch": 0.6163682864450127, "percentage": 12.33, "elapsed_time": "1:12:56", "remaining_time": "8:38:47"} +{"current_steps": 242, "total_steps": 1955, "loss": 1.1047, "lr": 7.986508214361047e-05, "epoch": 0.618925831202046, "percentage": 12.38, "elapsed_time": "1:13:14", "remaining_time": "8:38:27"} +{"current_steps": 243, "total_steps": 1955, "loss": 1.0975, "lr": 7.985915587133375e-05, "epoch": 0.6214833759590793, "percentage": 12.43, "elapsed_time": "1:13:32", "remaining_time": "8:38:09"} +{"current_steps": 244, "total_steps": 1955, "loss": 1.1448, "lr": 7.985310245500502e-05, "epoch": 0.6240409207161125, "percentage": 12.48, "elapsed_time": "1:13:50", "remaining_time": "8:37:48"} +{"current_steps": 245, "total_steps": 1955, "loss": 1.1053, "lr": 7.984692191393364e-05, "epoch": 0.6265984654731458, "percentage": 12.53, "elapsed_time": "1:14:08", "remaining_time": "8:37:29"} +{"current_steps": 246, "total_steps": 1955, "loss": 1.1014, "lr": 7.984061426783452e-05, "epoch": 0.629156010230179, "percentage": 12.58, "elapsed_time": "1:14:27", "remaining_time": "8:37:14"} +{"current_steps": 247, "total_steps": 1955, "loss": 1.0916, "lr": 7.9834179536828e-05, "epoch": 0.6317135549872123, "percentage": 12.63, "elapsed_time": "1:14:45", "remaining_time": "8:36:55"} +{"current_steps": 248, "total_steps": 1955, "loss": 1.1379, "lr": 7.982761774143978e-05, "epoch": 0.6342710997442456, "percentage": 12.69, "elapsed_time": "1:15:03", "remaining_time": "8:36:40"} +{"current_steps": 249, "total_steps": 1955, "loss": 1.1201, "lr": 7.98209289026009e-05, "epoch": 0.6368286445012787, "percentage": 12.74, "elapsed_time": "1:15:21", "remaining_time": "8:36:21"} +{"current_steps": 250, "total_steps": 1955, "loss": 1.0988, "lr": 7.981411304164765e-05, "epoch": 0.639386189258312, "percentage": 12.79, "elapsed_time": "1:15:39", "remaining_time": "8:36:02"} +{"current_steps": 251, "total_steps": 1955, "loss": 1.1052, "lr": 7.980717018032147e-05, "epoch": 0.6419437340153452, "percentage": 12.84, "elapsed_time": "1:15:57", "remaining_time": "8:35:42"} +{"current_steps": 252, "total_steps": 1955, "loss": 1.1121, "lr": 7.980010034076893e-05, "epoch": 0.6445012787723785, "percentage": 12.89, "elapsed_time": "1:16:15", "remaining_time": "8:35:23"} +{"current_steps": 253, "total_steps": 1955, "loss": 1.089, "lr": 7.979290354554166e-05, "epoch": 0.6470588235294118, "percentage": 12.94, "elapsed_time": "1:16:34", "remaining_time": "8:35:05"} +{"current_steps": 254, "total_steps": 1955, "loss": 1.1395, "lr": 7.978557981759622e-05, "epoch": 0.649616368286445, "percentage": 12.99, "elapsed_time": "1:16:52", "remaining_time": "8:34:46"} +{"current_steps": 255, "total_steps": 1955, "loss": 1.0977, "lr": 7.977812918029407e-05, "epoch": 0.6521739130434783, "percentage": 13.04, "elapsed_time": "1:17:10", "remaining_time": "8:34:27"} +{"current_steps": 256, "total_steps": 1955, "loss": 1.0859, "lr": 7.977055165740151e-05, "epoch": 0.6547314578005116, "percentage": 13.09, "elapsed_time": "1:17:27", "remaining_time": "8:34:03"} +{"current_steps": 257, "total_steps": 1955, "loss": 1.1208, "lr": 7.976284727308958e-05, "epoch": 0.6572890025575447, "percentage": 13.15, "elapsed_time": "1:17:45", "remaining_time": "8:33:45"} +{"current_steps": 258, "total_steps": 1955, "loss": 1.0834, "lr": 7.975501605193396e-05, "epoch": 0.659846547314578, "percentage": 13.2, "elapsed_time": "1:18:03", "remaining_time": "8:33:26"} +{"current_steps": 259, "total_steps": 1955, "loss": 1.1104, "lr": 7.974705801891496e-05, "epoch": 0.6624040920716112, "percentage": 13.25, "elapsed_time": "1:18:21", "remaining_time": "8:33:06"} +{"current_steps": 260, "total_steps": 1955, "loss": 1.0931, "lr": 7.973897319941736e-05, "epoch": 0.6649616368286445, "percentage": 13.3, "elapsed_time": "1:18:39", "remaining_time": "8:32:47"} +{"current_steps": 261, "total_steps": 1955, "loss": 1.0654, "lr": 7.97307616192304e-05, "epoch": 0.6675191815856778, "percentage": 13.35, "elapsed_time": "1:18:57", "remaining_time": "8:32:29"} +{"current_steps": 262, "total_steps": 1955, "loss": 1.0806, "lr": 7.972242330454762e-05, "epoch": 0.670076726342711, "percentage": 13.4, "elapsed_time": "1:19:15", "remaining_time": "8:32:09"} +{"current_steps": 263, "total_steps": 1955, "loss": 1.1243, "lr": 7.971395828196688e-05, "epoch": 0.6726342710997443, "percentage": 13.45, "elapsed_time": "1:19:33", "remaining_time": "8:31:50"} +{"current_steps": 264, "total_steps": 1955, "loss": 1.102, "lr": 7.970536657849018e-05, "epoch": 0.6751918158567775, "percentage": 13.5, "elapsed_time": "1:19:51", "remaining_time": "8:31:31"} +{"current_steps": 265, "total_steps": 1955, "loss": 1.1313, "lr": 7.969664822152362e-05, "epoch": 0.6777493606138107, "percentage": 13.55, "elapsed_time": "1:20:09", "remaining_time": "8:31:12"} +{"current_steps": 266, "total_steps": 1955, "loss": 1.0976, "lr": 7.96878032388773e-05, "epoch": 0.680306905370844, "percentage": 13.61, "elapsed_time": "1:20:27", "remaining_time": "8:30:54"} +{"current_steps": 267, "total_steps": 1955, "loss": 1.0522, "lr": 7.967883165876524e-05, "epoch": 0.6828644501278772, "percentage": 13.66, "elapsed_time": "1:20:45", "remaining_time": "8:30:35"} +{"current_steps": 268, "total_steps": 1955, "loss": 1.0948, "lr": 7.966973350980528e-05, "epoch": 0.6854219948849105, "percentage": 13.71, "elapsed_time": "1:21:04", "remaining_time": "8:30:20"} +{"current_steps": 269, "total_steps": 1955, "loss": 1.0707, "lr": 7.9660508821019e-05, "epoch": 0.6879795396419437, "percentage": 13.76, "elapsed_time": "1:21:22", "remaining_time": "8:30:01"} +{"current_steps": 270, "total_steps": 1955, "loss": 1.1105, "lr": 7.965115762183163e-05, "epoch": 0.690537084398977, "percentage": 13.81, "elapsed_time": "1:21:40", "remaining_time": "8:29:41"} +{"current_steps": 271, "total_steps": 1955, "loss": 1.1096, "lr": 7.96416799420719e-05, "epoch": 0.6930946291560103, "percentage": 13.86, "elapsed_time": "1:21:58", "remaining_time": "8:29:22"} +{"current_steps": 272, "total_steps": 1955, "loss": 1.0721, "lr": 7.963207581197207e-05, "epoch": 0.6956521739130435, "percentage": 13.91, "elapsed_time": "1:22:16", "remaining_time": "8:29:03"} +{"current_steps": 273, "total_steps": 1955, "loss": 1.0879, "lr": 7.962234526216767e-05, "epoch": 0.6982097186700768, "percentage": 13.96, "elapsed_time": "1:22:34", "remaining_time": "8:28:43"} +{"current_steps": 274, "total_steps": 1955, "loss": 1.1085, "lr": 7.961248832369754e-05, "epoch": 0.7007672634271099, "percentage": 14.02, "elapsed_time": "1:22:52", "remaining_time": "8:28:24"} +{"current_steps": 275, "total_steps": 1955, "loss": 1.0864, "lr": 7.96025050280037e-05, "epoch": 0.7033248081841432, "percentage": 14.07, "elapsed_time": "1:23:10", "remaining_time": "8:28:04"} +{"current_steps": 276, "total_steps": 1955, "loss": 1.1034, "lr": 7.959239540693115e-05, "epoch": 0.7058823529411765, "percentage": 14.12, "elapsed_time": "1:23:28", "remaining_time": "8:27:46"} +{"current_steps": 277, "total_steps": 1955, "loss": 1.1176, "lr": 7.958215949272791e-05, "epoch": 0.7084398976982097, "percentage": 14.17, "elapsed_time": "1:23:46", "remaining_time": "8:27:27"} +{"current_steps": 278, "total_steps": 1955, "loss": 1.0847, "lr": 7.957179731804485e-05, "epoch": 0.710997442455243, "percentage": 14.22, "elapsed_time": "1:24:04", "remaining_time": "8:27:08"} +{"current_steps": 279, "total_steps": 1955, "loss": 1.114, "lr": 7.956130891593556e-05, "epoch": 0.7135549872122762, "percentage": 14.27, "elapsed_time": "1:24:23", "remaining_time": "8:26:56"} +{"current_steps": 280, "total_steps": 1955, "loss": 1.0802, "lr": 7.955069431985632e-05, "epoch": 0.7161125319693095, "percentage": 14.32, "elapsed_time": "1:24:41", "remaining_time": "8:26:37"} +{"current_steps": 281, "total_steps": 1955, "loss": 1.0739, "lr": 7.953995356366588e-05, "epoch": 0.7186700767263428, "percentage": 14.37, "elapsed_time": "1:25:00", "remaining_time": "8:26:22"} +{"current_steps": 282, "total_steps": 1955, "loss": 1.1003, "lr": 7.952908668162549e-05, "epoch": 0.7212276214833759, "percentage": 14.42, "elapsed_time": "1:25:18", "remaining_time": "8:26:03"} +{"current_steps": 283, "total_steps": 1955, "loss": 1.0299, "lr": 7.951809370839867e-05, "epoch": 0.7237851662404092, "percentage": 14.48, "elapsed_time": "1:25:36", "remaining_time": "8:25:45"} +{"current_steps": 284, "total_steps": 1955, "loss": 1.0951, "lr": 7.950697467905118e-05, "epoch": 0.7263427109974424, "percentage": 14.53, "elapsed_time": "1:25:54", "remaining_time": "8:25:26"} +{"current_steps": 285, "total_steps": 1955, "loss": 1.0772, "lr": 7.949572962905086e-05, "epoch": 0.7289002557544757, "percentage": 14.58, "elapsed_time": "1:26:12", "remaining_time": "8:25:07"} +{"current_steps": 286, "total_steps": 1955, "loss": 1.1039, "lr": 7.948435859426755e-05, "epoch": 0.731457800511509, "percentage": 14.63, "elapsed_time": "1:26:30", "remaining_time": "8:24:48"} +{"current_steps": 287, "total_steps": 1955, "loss": 1.1019, "lr": 7.947286161097294e-05, "epoch": 0.7340153452685422, "percentage": 14.68, "elapsed_time": "1:26:48", "remaining_time": "8:24:29"} +{"current_steps": 288, "total_steps": 1955, "loss": 1.0641, "lr": 7.94612387158405e-05, "epoch": 0.7365728900255755, "percentage": 14.73, "elapsed_time": "1:27:07", "remaining_time": "8:24:16"} +{"current_steps": 289, "total_steps": 1955, "loss": 1.0976, "lr": 7.944948994594532e-05, "epoch": 0.7391304347826086, "percentage": 14.78, "elapsed_time": "1:27:25", "remaining_time": "8:23:57"} +{"current_steps": 290, "total_steps": 1955, "loss": 1.1066, "lr": 7.943761533876402e-05, "epoch": 0.7416879795396419, "percentage": 14.83, "elapsed_time": "1:27:43", "remaining_time": "8:23:38"} +{"current_steps": 291, "total_steps": 1955, "loss": 1.0889, "lr": 7.94256149321746e-05, "epoch": 0.7442455242966752, "percentage": 14.88, "elapsed_time": "1:28:01", "remaining_time": "8:23:19"} +{"current_steps": 292, "total_steps": 1955, "loss": 1.0538, "lr": 7.941348876445636e-05, "epoch": 0.7468030690537084, "percentage": 14.94, "elapsed_time": "1:28:19", "remaining_time": "8:23:00"} +{"current_steps": 293, "total_steps": 1955, "loss": 1.0817, "lr": 7.940123687428975e-05, "epoch": 0.7493606138107417, "percentage": 14.99, "elapsed_time": "1:28:37", "remaining_time": "8:22:41"} +{"current_steps": 294, "total_steps": 1955, "loss": 1.0781, "lr": 7.938885930075628e-05, "epoch": 0.7519181585677749, "percentage": 15.04, "elapsed_time": "1:28:55", "remaining_time": "8:22:22"} +{"current_steps": 295, "total_steps": 1955, "loss": 1.0709, "lr": 7.937635608333829e-05, "epoch": 0.7544757033248082, "percentage": 15.09, "elapsed_time": "1:29:13", "remaining_time": "8:22:03"} +{"current_steps": 296, "total_steps": 1955, "loss": 1.0792, "lr": 7.936372726191899e-05, "epoch": 0.7570332480818415, "percentage": 15.14, "elapsed_time": "1:29:31", "remaining_time": "8:21:44"} +{"current_steps": 297, "total_steps": 1955, "loss": 1.1203, "lr": 7.935097287678222e-05, "epoch": 0.7595907928388747, "percentage": 15.19, "elapsed_time": "1:29:49", "remaining_time": "8:21:25"} +{"current_steps": 298, "total_steps": 1955, "loss": 1.0951, "lr": 7.933809296861232e-05, "epoch": 0.7621483375959079, "percentage": 15.24, "elapsed_time": "1:30:07", "remaining_time": "8:21:08"} +{"current_steps": 299, "total_steps": 1955, "loss": 1.1087, "lr": 7.932508757849403e-05, "epoch": 0.7647058823529411, "percentage": 15.29, "elapsed_time": "1:30:25", "remaining_time": "8:20:49"} +{"current_steps": 300, "total_steps": 1955, "loss": 1.0713, "lr": 7.931195674791242e-05, "epoch": 0.7672634271099744, "percentage": 15.35, "elapsed_time": "1:30:43", "remaining_time": "8:20:31"} +{"current_steps": 301, "total_steps": 1955, "loss": 1.1025, "lr": 7.929870051875261e-05, "epoch": 0.7698209718670077, "percentage": 15.4, "elapsed_time": "1:31:01", "remaining_time": "8:20:12"} +{"current_steps": 302, "total_steps": 1955, "loss": 1.1099, "lr": 7.928531893329977e-05, "epoch": 0.7723785166240409, "percentage": 15.45, "elapsed_time": "1:31:20", "remaining_time": "8:19:58"} +{"current_steps": 303, "total_steps": 1955, "loss": 1.0761, "lr": 7.927181203423893e-05, "epoch": 0.7749360613810742, "percentage": 15.5, "elapsed_time": "1:31:38", "remaining_time": "8:19:39"} +{"current_steps": 304, "total_steps": 1955, "loss": 1.0964, "lr": 7.925817986465483e-05, "epoch": 0.7774936061381074, "percentage": 15.55, "elapsed_time": "1:31:57", "remaining_time": "8:19:26"} +{"current_steps": 305, "total_steps": 1955, "loss": 1.1114, "lr": 7.924442246803183e-05, "epoch": 0.7800511508951407, "percentage": 15.6, "elapsed_time": "1:32:15", "remaining_time": "8:19:07"} +{"current_steps": 306, "total_steps": 1955, "loss": 1.0863, "lr": 7.923053988825372e-05, "epoch": 0.782608695652174, "percentage": 15.65, "elapsed_time": "1:32:33", "remaining_time": "8:18:48"} +{"current_steps": 307, "total_steps": 1955, "loss": 1.0609, "lr": 7.92165321696036e-05, "epoch": 0.7851662404092071, "percentage": 15.7, "elapsed_time": "1:32:51", "remaining_time": "8:18:29"} +{"current_steps": 308, "total_steps": 1955, "loss": 1.0832, "lr": 7.920239935676377e-05, "epoch": 0.7877237851662404, "percentage": 15.75, "elapsed_time": "1:33:09", "remaining_time": "8:18:10"} +{"current_steps": 309, "total_steps": 1955, "loss": 1.1062, "lr": 7.918814149481553e-05, "epoch": 0.7902813299232737, "percentage": 15.81, "elapsed_time": "1:33:27", "remaining_time": "8:17:53"} +{"current_steps": 310, "total_steps": 1955, "loss": 1.0699, "lr": 7.917375862923908e-05, "epoch": 0.7928388746803069, "percentage": 15.86, "elapsed_time": "1:33:46", "remaining_time": "8:17:34"} +{"current_steps": 311, "total_steps": 1955, "loss": 1.0999, "lr": 7.915925080591338e-05, "epoch": 0.7953964194373402, "percentage": 15.91, "elapsed_time": "1:34:04", "remaining_time": "8:17:15"} +{"current_steps": 312, "total_steps": 1955, "loss": 1.1169, "lr": 7.914461807111595e-05, "epoch": 0.7979539641943734, "percentage": 15.96, "elapsed_time": "1:34:21", "remaining_time": "8:16:54"} +{"current_steps": 313, "total_steps": 1955, "loss": 1.0783, "lr": 7.912986047152278e-05, "epoch": 0.8005115089514067, "percentage": 16.01, "elapsed_time": "1:34:39", "remaining_time": "8:16:32"} +{"current_steps": 314, "total_steps": 1955, "loss": 1.0823, "lr": 7.911497805420811e-05, "epoch": 0.80306905370844, "percentage": 16.06, "elapsed_time": "1:34:56", "remaining_time": "8:16:10"} +{"current_steps": 315, "total_steps": 1955, "loss": 1.0748, "lr": 7.909997086664443e-05, "epoch": 0.8056265984654731, "percentage": 16.11, "elapsed_time": "1:35:14", "remaining_time": "8:15:51"} +{"current_steps": 316, "total_steps": 1955, "loss": 1.071, "lr": 7.908483895670209e-05, "epoch": 0.8081841432225064, "percentage": 16.16, "elapsed_time": "1:35:32", "remaining_time": "8:15:32"} +{"current_steps": 317, "total_steps": 1955, "loss": 1.0984, "lr": 7.906958237264942e-05, "epoch": 0.8107416879795396, "percentage": 16.21, "elapsed_time": "1:35:50", "remaining_time": "8:15:13"} +{"current_steps": 318, "total_steps": 1955, "loss": 1.0908, "lr": 7.905420116315234e-05, "epoch": 0.8132992327365729, "percentage": 16.27, "elapsed_time": "1:36:08", "remaining_time": "8:14:54"} +{"current_steps": 319, "total_steps": 1955, "loss": 1.0503, "lr": 7.903869537727437e-05, "epoch": 0.8158567774936062, "percentage": 16.32, "elapsed_time": "1:36:26", "remaining_time": "8:14:36"} +{"current_steps": 320, "total_steps": 1955, "loss": 1.0783, "lr": 7.902306506447634e-05, "epoch": 0.8184143222506394, "percentage": 16.37, "elapsed_time": "1:36:44", "remaining_time": "8:14:17"} +{"current_steps": 321, "total_steps": 1955, "loss": 1.057, "lr": 7.900731027461635e-05, "epoch": 0.8209718670076727, "percentage": 16.42, "elapsed_time": "1:37:02", "remaining_time": "8:13:59"} +{"current_steps": 322, "total_steps": 1955, "loss": 1.087, "lr": 7.899143105794958e-05, "epoch": 0.8235294117647058, "percentage": 16.47, "elapsed_time": "1:37:20", "remaining_time": "8:13:40"} +{"current_steps": 323, "total_steps": 1955, "loss": 1.1064, "lr": 7.897542746512804e-05, "epoch": 0.8260869565217391, "percentage": 16.52, "elapsed_time": "1:37:38", "remaining_time": "8:13:21"} +{"current_steps": 324, "total_steps": 1955, "loss": 1.1046, "lr": 7.895929954720053e-05, "epoch": 0.8286445012787724, "percentage": 16.57, "elapsed_time": "1:37:56", "remaining_time": "8:13:03"} +{"current_steps": 325, "total_steps": 1955, "loss": 1.0652, "lr": 7.894304735561243e-05, "epoch": 0.8312020460358056, "percentage": 16.62, "elapsed_time": "1:38:14", "remaining_time": "8:12:44"} +{"current_steps": 326, "total_steps": 1955, "loss": 1.0756, "lr": 7.89266709422055e-05, "epoch": 0.8337595907928389, "percentage": 16.68, "elapsed_time": "1:38:32", "remaining_time": "8:12:25"} +{"current_steps": 327, "total_steps": 1955, "loss": 1.1155, "lr": 7.891017035921778e-05, "epoch": 0.8363171355498721, "percentage": 16.73, "elapsed_time": "1:38:50", "remaining_time": "8:12:07"} +{"current_steps": 328, "total_steps": 1955, "loss": 1.0879, "lr": 7.889354565928337e-05, "epoch": 0.8388746803069054, "percentage": 16.78, "elapsed_time": "1:39:08", "remaining_time": "8:11:48"} +{"current_steps": 329, "total_steps": 1955, "loss": 1.1265, "lr": 7.887679689543227e-05, "epoch": 0.8414322250639387, "percentage": 16.83, "elapsed_time": "1:39:26", "remaining_time": "8:11:29"} +{"current_steps": 330, "total_steps": 1955, "loss": 1.0971, "lr": 7.885992412109026e-05, "epoch": 0.8439897698209718, "percentage": 16.88, "elapsed_time": "1:39:45", "remaining_time": "8:11:12"} +{"current_steps": 331, "total_steps": 1955, "loss": 1.0698, "lr": 7.884292739007866e-05, "epoch": 0.8465473145780051, "percentage": 16.93, "elapsed_time": "1:40:02", "remaining_time": "8:10:51"} +{"current_steps": 332, "total_steps": 1955, "loss": 1.0612, "lr": 7.882580675661422e-05, "epoch": 0.8491048593350383, "percentage": 16.98, "elapsed_time": "1:40:21", "remaining_time": "8:10:37"} +{"current_steps": 333, "total_steps": 1955, "loss": 1.1092, "lr": 7.88085622753089e-05, "epoch": 0.8516624040920716, "percentage": 17.03, "elapsed_time": "1:40:39", "remaining_time": "8:10:18"} +{"current_steps": 334, "total_steps": 1955, "loss": 1.0411, "lr": 7.879119400116969e-05, "epoch": 0.8542199488491049, "percentage": 17.08, "elapsed_time": "1:40:57", "remaining_time": "8:10:00"} +{"current_steps": 335, "total_steps": 1955, "loss": 1.0671, "lr": 7.877370198959852e-05, "epoch": 0.8567774936061381, "percentage": 17.14, "elapsed_time": "1:41:15", "remaining_time": "8:09:39"} +{"current_steps": 336, "total_steps": 1955, "loss": 1.067, "lr": 7.875608629639195e-05, "epoch": 0.8593350383631714, "percentage": 17.19, "elapsed_time": "1:41:33", "remaining_time": "8:09:20"} +{"current_steps": 337, "total_steps": 1955, "loss": 1.0896, "lr": 7.873834697774114e-05, "epoch": 0.8618925831202046, "percentage": 17.24, "elapsed_time": "1:41:50", "remaining_time": "8:08:58"} +{"current_steps": 338, "total_steps": 1955, "loss": 1.0945, "lr": 7.872048409023153e-05, "epoch": 0.8644501278772379, "percentage": 17.29, "elapsed_time": "1:42:08", "remaining_time": "8:08:40"} +{"current_steps": 339, "total_steps": 1955, "loss": 1.0599, "lr": 7.870249769084274e-05, "epoch": 0.8670076726342711, "percentage": 17.34, "elapsed_time": "1:42:27", "remaining_time": "8:08:23"} +{"current_steps": 340, "total_steps": 1955, "loss": 1.0984, "lr": 7.868438783694841e-05, "epoch": 0.8695652173913043, "percentage": 17.39, "elapsed_time": "1:42:45", "remaining_time": "8:08:04"} +{"current_steps": 341, "total_steps": 1955, "loss": 1.0551, "lr": 7.866615458631591e-05, "epoch": 0.8721227621483376, "percentage": 17.44, "elapsed_time": "1:43:03", "remaining_time": "8:07:46"} +{"current_steps": 342, "total_steps": 1955, "loss": 1.0821, "lr": 7.864779799710631e-05, "epoch": 0.8746803069053708, "percentage": 17.49, "elapsed_time": "1:43:21", "remaining_time": "8:07:29"} +{"current_steps": 343, "total_steps": 1955, "loss": 1.0802, "lr": 7.862931812787404e-05, "epoch": 0.8772378516624041, "percentage": 17.54, "elapsed_time": "1:43:39", "remaining_time": "8:07:10"} +{"current_steps": 344, "total_steps": 1955, "loss": 1.0876, "lr": 7.86107150375668e-05, "epoch": 0.8797953964194374, "percentage": 17.6, "elapsed_time": "1:43:56", "remaining_time": "8:06:48"} +{"current_steps": 345, "total_steps": 1955, "loss": 1.068, "lr": 7.859198878552536e-05, "epoch": 0.8823529411764706, "percentage": 17.65, "elapsed_time": "1:44:15", "remaining_time": "8:06:31"} +{"current_steps": 346, "total_steps": 1955, "loss": 1.1331, "lr": 7.85731394314833e-05, "epoch": 0.8849104859335039, "percentage": 17.7, "elapsed_time": "1:44:33", "remaining_time": "8:06:12"} +{"current_steps": 347, "total_steps": 1955, "loss": 1.0161, "lr": 7.855416703556693e-05, "epoch": 0.887468030690537, "percentage": 17.75, "elapsed_time": "1:44:51", "remaining_time": "8:05:53"} +{"current_steps": 348, "total_steps": 1955, "loss": 1.0811, "lr": 7.853507165829504e-05, "epoch": 0.8900255754475703, "percentage": 17.8, "elapsed_time": "1:45:10", "remaining_time": "8:05:38"} +{"current_steps": 349, "total_steps": 1955, "loss": 1.0653, "lr": 7.851585336057868e-05, "epoch": 0.8925831202046036, "percentage": 17.85, "elapsed_time": "1:45:28", "remaining_time": "8:05:20"} +{"current_steps": 350, "total_steps": 1955, "loss": 1.0867, "lr": 7.849651220372101e-05, "epoch": 0.8951406649616368, "percentage": 17.9, "elapsed_time": "1:45:46", "remaining_time": "8:05:02"} +{"current_steps": 351, "total_steps": 1955, "loss": 1.105, "lr": 7.847704824941709e-05, "epoch": 0.8976982097186701, "percentage": 17.95, "elapsed_time": "1:46:04", "remaining_time": "8:04:43"} +{"current_steps": 352, "total_steps": 1955, "loss": 1.0596, "lr": 7.84574615597537e-05, "epoch": 0.9002557544757033, "percentage": 18.01, "elapsed_time": "1:46:22", "remaining_time": "8:04:24"} +{"current_steps": 353, "total_steps": 1955, "loss": 1.0823, "lr": 7.84377521972091e-05, "epoch": 0.9028132992327366, "percentage": 18.06, "elapsed_time": "1:46:40", "remaining_time": "8:04:06"} +{"current_steps": 354, "total_steps": 1955, "loss": 1.0763, "lr": 7.841792022465286e-05, "epoch": 0.9053708439897699, "percentage": 18.11, "elapsed_time": "1:46:58", "remaining_time": "8:03:47"} +{"current_steps": 355, "total_steps": 1955, "loss": 1.0542, "lr": 7.839796570534566e-05, "epoch": 0.907928388746803, "percentage": 18.16, "elapsed_time": "1:47:17", "remaining_time": "8:03:32"} +{"current_steps": 356, "total_steps": 1955, "loss": 1.0668, "lr": 7.83778887029391e-05, "epoch": 0.9104859335038363, "percentage": 18.21, "elapsed_time": "1:47:36", "remaining_time": "8:03:18"} +{"current_steps": 357, "total_steps": 1955, "loss": 1.0759, "lr": 7.835768928147545e-05, "epoch": 0.9130434782608695, "percentage": 18.26, "elapsed_time": "1:47:55", "remaining_time": "8:03:03"} +{"current_steps": 358, "total_steps": 1955, "loss": 1.0976, "lr": 7.83373675053875e-05, "epoch": 0.9156010230179028, "percentage": 18.31, "elapsed_time": "1:48:13", "remaining_time": "8:02:45"} +{"current_steps": 359, "total_steps": 1955, "loss": 1.0792, "lr": 7.831692343949832e-05, "epoch": 0.9181585677749361, "percentage": 18.36, "elapsed_time": "1:48:31", "remaining_time": "8:02:26"} +{"current_steps": 360, "total_steps": 1955, "loss": 1.0461, "lr": 7.829635714902108e-05, "epoch": 0.9207161125319693, "percentage": 18.41, "elapsed_time": "1:48:49", "remaining_time": "8:02:07"} +{"current_steps": 361, "total_steps": 1955, "loss": 1.0957, "lr": 7.827566869955878e-05, "epoch": 0.9232736572890026, "percentage": 18.47, "elapsed_time": "1:49:07", "remaining_time": "8:01:49"} +{"current_steps": 362, "total_steps": 1955, "loss": 1.0246, "lr": 7.825485815710415e-05, "epoch": 0.9258312020460358, "percentage": 18.52, "elapsed_time": "1:49:25", "remaining_time": "8:01:30"} +{"current_steps": 363, "total_steps": 1955, "loss": 1.0859, "lr": 7.823392558803933e-05, "epoch": 0.928388746803069, "percentage": 18.57, "elapsed_time": "1:49:43", "remaining_time": "8:01:11"} +{"current_steps": 364, "total_steps": 1955, "loss": 1.069, "lr": 7.821287105913571e-05, "epoch": 0.9309462915601023, "percentage": 18.62, "elapsed_time": "1:50:01", "remaining_time": "8:00:53"} +{"current_steps": 365, "total_steps": 1955, "loss": 1.0857, "lr": 7.819169463755376e-05, "epoch": 0.9335038363171355, "percentage": 18.67, "elapsed_time": "1:50:19", "remaining_time": "8:00:34"} +{"current_steps": 366, "total_steps": 1955, "loss": 1.1084, "lr": 7.817039639084271e-05, "epoch": 0.9360613810741688, "percentage": 18.72, "elapsed_time": "1:50:37", "remaining_time": "8:00:16"} +{"current_steps": 367, "total_steps": 1955, "loss": 1.0864, "lr": 7.814897638694039e-05, "epoch": 0.9386189258312021, "percentage": 18.77, "elapsed_time": "1:50:55", "remaining_time": "7:59:57"} +{"current_steps": 368, "total_steps": 1955, "loss": 1.0797, "lr": 7.812743469417306e-05, "epoch": 0.9411764705882353, "percentage": 18.82, "elapsed_time": "1:51:13", "remaining_time": "7:59:39"} +{"current_steps": 369, "total_steps": 1955, "loss": 1.074, "lr": 7.810577138125512e-05, "epoch": 0.9437340153452686, "percentage": 18.87, "elapsed_time": "1:51:31", "remaining_time": "7:59:20"} +{"current_steps": 370, "total_steps": 1955, "loss": 1.0704, "lr": 7.808398651728891e-05, "epoch": 0.9462915601023018, "percentage": 18.93, "elapsed_time": "1:51:49", "remaining_time": "7:59:01"} +{"current_steps": 371, "total_steps": 1955, "loss": 1.0978, "lr": 7.806208017176453e-05, "epoch": 0.948849104859335, "percentage": 18.98, "elapsed_time": "1:52:07", "remaining_time": "7:58:41"} +{"current_steps": 372, "total_steps": 1955, "loss": 1.0799, "lr": 7.804005241455954e-05, "epoch": 0.9514066496163683, "percentage": 19.03, "elapsed_time": "1:52:25", "remaining_time": "7:58:22"} +{"current_steps": 373, "total_steps": 1955, "loss": 1.0847, "lr": 7.801790331593882e-05, "epoch": 0.9539641943734015, "percentage": 19.08, "elapsed_time": "1:52:43", "remaining_time": "7:58:04"} +{"current_steps": 374, "total_steps": 1955, "loss": 1.0843, "lr": 7.799563294655429e-05, "epoch": 0.9565217391304348, "percentage": 19.13, "elapsed_time": "1:53:01", "remaining_time": "7:57:45"} +{"current_steps": 375, "total_steps": 1955, "loss": 1.1051, "lr": 7.797324137744472e-05, "epoch": 0.959079283887468, "percentage": 19.18, "elapsed_time": "1:53:19", "remaining_time": "7:57:27"} +{"current_steps": 376, "total_steps": 1955, "loss": 1.0541, "lr": 7.795072868003546e-05, "epoch": 0.9616368286445013, "percentage": 19.23, "elapsed_time": "1:53:37", "remaining_time": "7:57:08"} +{"current_steps": 377, "total_steps": 1955, "loss": 1.0867, "lr": 7.792809492613828e-05, "epoch": 0.9641943734015346, "percentage": 19.28, "elapsed_time": "1:53:55", "remaining_time": "7:56:50"} +{"current_steps": 378, "total_steps": 1955, "loss": 1.0925, "lr": 7.790534018795106e-05, "epoch": 0.9667519181585678, "percentage": 19.34, "elapsed_time": "1:54:13", "remaining_time": "7:56:32"} +{"current_steps": 379, "total_steps": 1955, "loss": 1.0609, "lr": 7.788246453805762e-05, "epoch": 0.969309462915601, "percentage": 19.39, "elapsed_time": "1:54:31", "remaining_time": "7:56:13"} +{"current_steps": 380, "total_steps": 1955, "loss": 1.0443, "lr": 7.785946804942746e-05, "epoch": 0.9718670076726342, "percentage": 19.44, "elapsed_time": "1:54:49", "remaining_time": "7:55:55"} +{"current_steps": 381, "total_steps": 1955, "loss": 1.0742, "lr": 7.783635079541553e-05, "epoch": 0.9744245524296675, "percentage": 19.49, "elapsed_time": "1:55:08", "remaining_time": "7:55:39"} +{"current_steps": 382, "total_steps": 1955, "loss": 1.0784, "lr": 7.781311284976204e-05, "epoch": 0.9769820971867008, "percentage": 19.54, "elapsed_time": "1:55:26", "remaining_time": "7:55:20"} +{"current_steps": 383, "total_steps": 1955, "loss": 1.0828, "lr": 7.778975428659213e-05, "epoch": 0.979539641943734, "percentage": 19.59, "elapsed_time": "1:55:44", "remaining_time": "7:55:02"} +{"current_steps": 384, "total_steps": 1955, "loss": 1.0359, "lr": 7.776627518041573e-05, "epoch": 0.9820971867007673, "percentage": 19.64, "elapsed_time": "1:56:02", "remaining_time": "7:54:44"} +{"current_steps": 385, "total_steps": 1955, "loss": 1.1162, "lr": 7.774267560612725e-05, "epoch": 0.9846547314578005, "percentage": 19.69, "elapsed_time": "1:56:21", "remaining_time": "7:54:28"} +{"current_steps": 386, "total_steps": 1955, "loss": 1.0707, "lr": 7.771895563900542e-05, "epoch": 0.9872122762148338, "percentage": 19.74, "elapsed_time": "1:56:39", "remaining_time": "7:54:10"} +{"current_steps": 387, "total_steps": 1955, "loss": 1.0835, "lr": 7.769511535471294e-05, "epoch": 0.989769820971867, "percentage": 19.8, "elapsed_time": "1:56:57", "remaining_time": "7:53:51"} +{"current_steps": 388, "total_steps": 1955, "loss": 1.0974, "lr": 7.767115482929638e-05, "epoch": 0.9923273657289002, "percentage": 19.85, "elapsed_time": "1:57:15", "remaining_time": "7:53:33"} +{"current_steps": 389, "total_steps": 1955, "loss": 1.0806, "lr": 7.764707413918578e-05, "epoch": 0.9948849104859335, "percentage": 19.9, "elapsed_time": "1:57:33", "remaining_time": "7:53:14"} +{"current_steps": 390, "total_steps": 1955, "loss": 1.0609, "lr": 7.762287336119453e-05, "epoch": 0.9974424552429667, "percentage": 19.95, "elapsed_time": "1:57:52", "remaining_time": "7:52:59"} +{"current_steps": 391, "total_steps": 1955, "loss": 1.0896, "lr": 7.759855257251907e-05, "epoch": 1.0, "percentage": 20.0, "elapsed_time": "1:58:10", "remaining_time": "7:52:40"} +{"current_steps": 392, "total_steps": 1955, "loss": 1.0565, "lr": 7.757411185073866e-05, "epoch": 1.0025575447570332, "percentage": 20.05, "elapsed_time": "1:58:40", "remaining_time": "7:53:09"} +{"current_steps": 393, "total_steps": 1955, "loss": 1.0719, "lr": 7.754955127381512e-05, "epoch": 1.0051150895140666, "percentage": 20.1, "elapsed_time": "1:58:58", "remaining_time": "7:52:51"} +{"current_steps": 394, "total_steps": 1955, "loss": 1.044, "lr": 7.75248709200926e-05, "epoch": 1.0076726342710998, "percentage": 20.15, "elapsed_time": "1:59:16", "remaining_time": "7:52:33"} +{"current_steps": 395, "total_steps": 1955, "loss": 1.0534, "lr": 7.750007086829729e-05, "epoch": 1.010230179028133, "percentage": 20.2, "elapsed_time": "1:59:35", "remaining_time": "7:52:16"} +{"current_steps": 396, "total_steps": 1955, "loss": 1.0731, "lr": 7.747515119753721e-05, "epoch": 1.0127877237851663, "percentage": 20.26, "elapsed_time": "1:59:53", "remaining_time": "7:51:58"} +{"current_steps": 397, "total_steps": 1955, "loss": 1.0712, "lr": 7.745011198730196e-05, "epoch": 1.0153452685421995, "percentage": 20.31, "elapsed_time": "2:00:11", "remaining_time": "7:51:39"} +{"current_steps": 398, "total_steps": 1955, "loss": 1.0622, "lr": 7.742495331746246e-05, "epoch": 1.0179028132992327, "percentage": 20.36, "elapsed_time": "2:00:29", "remaining_time": "7:51:23"} +{"current_steps": 399, "total_steps": 1955, "loss": 1.0357, "lr": 7.739967526827063e-05, "epoch": 1.020460358056266, "percentage": 20.41, "elapsed_time": "2:00:48", "remaining_time": "7:51:05"} +{"current_steps": 400, "total_steps": 1955, "loss": 1.0885, "lr": 7.737427792035923e-05, "epoch": 1.0230179028132993, "percentage": 20.46, "elapsed_time": "2:01:06", "remaining_time": "7:50:48"} +{"current_steps": 401, "total_steps": 1955, "loss": 1.0677, "lr": 7.73487613547416e-05, "epoch": 1.0255754475703325, "percentage": 20.51, "elapsed_time": "2:01:24", "remaining_time": "7:50:30"} +{"current_steps": 402, "total_steps": 1955, "loss": 1.0844, "lr": 7.732312565281128e-05, "epoch": 1.0281329923273657, "percentage": 20.56, "elapsed_time": "2:01:42", "remaining_time": "7:50:09"} +{"current_steps": 403, "total_steps": 1955, "loss": 1.0497, "lr": 7.72973708963419e-05, "epoch": 1.030690537084399, "percentage": 20.61, "elapsed_time": "2:02:00", "remaining_time": "7:49:51"} +{"current_steps": 404, "total_steps": 1955, "loss": 1.0572, "lr": 7.727149716748681e-05, "epoch": 1.0332480818414322, "percentage": 20.66, "elapsed_time": "2:02:18", "remaining_time": "7:49:32"} +{"current_steps": 405, "total_steps": 1955, "loss": 1.0549, "lr": 7.724550454877892e-05, "epoch": 1.0358056265984654, "percentage": 20.72, "elapsed_time": "2:02:36", "remaining_time": "7:49:13"} +{"current_steps": 406, "total_steps": 1955, "loss": 1.0869, "lr": 7.721939312313033e-05, "epoch": 1.0383631713554988, "percentage": 20.77, "elapsed_time": "2:02:54", "remaining_time": "7:48:54"} +{"current_steps": 407, "total_steps": 1955, "loss": 1.022, "lr": 7.71931629738321e-05, "epoch": 1.040920716112532, "percentage": 20.82, "elapsed_time": "2:03:12", "remaining_time": "7:48:35"} +{"current_steps": 408, "total_steps": 1955, "loss": 1.0529, "lr": 7.716681418455409e-05, "epoch": 1.0434782608695652, "percentage": 20.87, "elapsed_time": "2:03:30", "remaining_time": "7:48:18"} +{"current_steps": 409, "total_steps": 1955, "loss": 1.034, "lr": 7.714034683934446e-05, "epoch": 1.0460358056265984, "percentage": 20.92, "elapsed_time": "2:03:48", "remaining_time": "7:48:00"} +{"current_steps": 410, "total_steps": 1955, "loss": 1.0282, "lr": 7.711376102262968e-05, "epoch": 1.0485933503836318, "percentage": 20.97, "elapsed_time": "2:04:07", "remaining_time": "7:47:42"} +{"current_steps": 411, "total_steps": 1955, "loss": 1.0553, "lr": 7.708705681921405e-05, "epoch": 1.051150895140665, "percentage": 21.02, "elapsed_time": "2:04:25", "remaining_time": "7:47:25"} +{"current_steps": 412, "total_steps": 1955, "loss": 1.0365, "lr": 7.706023431427951e-05, "epoch": 1.0537084398976981, "percentage": 21.07, "elapsed_time": "2:04:43", "remaining_time": "7:47:06"} +{"current_steps": 413, "total_steps": 1955, "loss": 1.0437, "lr": 7.703329359338538e-05, "epoch": 1.0562659846547315, "percentage": 21.13, "elapsed_time": "2:05:01", "remaining_time": "7:46:48"} +{"current_steps": 414, "total_steps": 1955, "loss": 1.0898, "lr": 7.700623474246806e-05, "epoch": 1.0588235294117647, "percentage": 21.18, "elapsed_time": "2:05:20", "remaining_time": "7:46:31"} +{"current_steps": 415, "total_steps": 1955, "loss": 1.061, "lr": 7.697905784784077e-05, "epoch": 1.061381074168798, "percentage": 21.23, "elapsed_time": "2:05:38", "remaining_time": "7:46:15"} +{"current_steps": 416, "total_steps": 1955, "loss": 1.0433, "lr": 7.695176299619327e-05, "epoch": 1.0639386189258313, "percentage": 21.28, "elapsed_time": "2:05:56", "remaining_time": "7:45:56"} +{"current_steps": 417, "total_steps": 1955, "loss": 1.0268, "lr": 7.692435027459157e-05, "epoch": 1.0664961636828645, "percentage": 21.33, "elapsed_time": "2:06:14", "remaining_time": "7:45:37"} +{"current_steps": 418, "total_steps": 1955, "loss": 1.0372, "lr": 7.689681977047767e-05, "epoch": 1.0690537084398977, "percentage": 21.38, "elapsed_time": "2:06:32", "remaining_time": "7:45:19"} +{"current_steps": 419, "total_steps": 1955, "loss": 1.0924, "lr": 7.68691715716693e-05, "epoch": 1.0716112531969308, "percentage": 21.43, "elapsed_time": "2:06:50", "remaining_time": "7:45:00"} +{"current_steps": 420, "total_steps": 1955, "loss": 1.0415, "lr": 7.684140576635959e-05, "epoch": 1.0741687979539642, "percentage": 21.48, "elapsed_time": "2:07:09", "remaining_time": "7:44:42"} +{"current_steps": 421, "total_steps": 1955, "loss": 1.0556, "lr": 7.68135224431168e-05, "epoch": 1.0767263427109974, "percentage": 21.53, "elapsed_time": "2:07:27", "remaining_time": "7:44:25"} +{"current_steps": 422, "total_steps": 1955, "loss": 1.0458, "lr": 7.678552169088411e-05, "epoch": 1.0792838874680306, "percentage": 21.59, "elapsed_time": "2:07:45", "remaining_time": "7:44:06"} +{"current_steps": 423, "total_steps": 1955, "loss": 1.0314, "lr": 7.675740359897923e-05, "epoch": 1.081841432225064, "percentage": 21.64, "elapsed_time": "2:08:03", "remaining_time": "7:43:48"} +{"current_steps": 424, "total_steps": 1955, "loss": 1.0779, "lr": 7.672916825709417e-05, "epoch": 1.0843989769820972, "percentage": 21.69, "elapsed_time": "2:08:21", "remaining_time": "7:43:30"} +{"current_steps": 425, "total_steps": 1955, "loss": 1.0412, "lr": 7.670081575529499e-05, "epoch": 1.0869565217391304, "percentage": 21.74, "elapsed_time": "2:08:40", "remaining_time": "7:43:12"} +{"current_steps": 426, "total_steps": 1955, "loss": 1.0312, "lr": 7.667234618402139e-05, "epoch": 1.0895140664961638, "percentage": 21.79, "elapsed_time": "2:08:58", "remaining_time": "7:42:54"} +{"current_steps": 427, "total_steps": 1955, "loss": 1.0559, "lr": 7.664375963408657e-05, "epoch": 1.092071611253197, "percentage": 21.84, "elapsed_time": "2:09:16", "remaining_time": "7:42:37"} +{"current_steps": 428, "total_steps": 1955, "loss": 1.0367, "lr": 7.661505619667685e-05, "epoch": 1.0946291560102301, "percentage": 21.89, "elapsed_time": "2:09:34", "remaining_time": "7:42:19"} +{"current_steps": 429, "total_steps": 1955, "loss": 1.0662, "lr": 7.658623596335141e-05, "epoch": 1.0971867007672633, "percentage": 21.94, "elapsed_time": "2:09:52", "remaining_time": "7:42:00"} +{"current_steps": 430, "total_steps": 1955, "loss": 1.0426, "lr": 7.655729902604198e-05, "epoch": 1.0997442455242967, "percentage": 21.99, "elapsed_time": "2:10:10", "remaining_time": "7:41:41"} +{"current_steps": 431, "total_steps": 1955, "loss": 1.0724, "lr": 7.652824547705255e-05, "epoch": 1.10230179028133, "percentage": 22.05, "elapsed_time": "2:10:29", "remaining_time": "7:41:26"} +{"current_steps": 432, "total_steps": 1955, "loss": 1.0899, "lr": 7.64990754090591e-05, "epoch": 1.104859335038363, "percentage": 22.1, "elapsed_time": "2:10:47", "remaining_time": "7:41:07"} +{"current_steps": 433, "total_steps": 1955, "loss": 1.0671, "lr": 7.646978891510927e-05, "epoch": 1.1074168797953965, "percentage": 22.15, "elapsed_time": "2:11:05", "remaining_time": "7:40:48"} +{"current_steps": 434, "total_steps": 1955, "loss": 1.0558, "lr": 7.64403860886221e-05, "epoch": 1.1099744245524297, "percentage": 22.2, "elapsed_time": "2:11:24", "remaining_time": "7:40:31"} +{"current_steps": 435, "total_steps": 1955, "loss": 1.0477, "lr": 7.641086702338766e-05, "epoch": 1.1125319693094629, "percentage": 22.25, "elapsed_time": "2:11:42", "remaining_time": "7:40:14"} +{"current_steps": 436, "total_steps": 1955, "loss": 1.046, "lr": 7.638123181356688e-05, "epoch": 1.1150895140664963, "percentage": 22.3, "elapsed_time": "2:12:00", "remaining_time": "7:39:55"} +{"current_steps": 437, "total_steps": 1955, "loss": 1.0268, "lr": 7.635148055369111e-05, "epoch": 1.1176470588235294, "percentage": 22.35, "elapsed_time": "2:12:18", "remaining_time": "7:39:36"} +{"current_steps": 438, "total_steps": 1955, "loss": 1.0407, "lr": 7.63216133386619e-05, "epoch": 1.1202046035805626, "percentage": 22.4, "elapsed_time": "2:12:37", "remaining_time": "7:39:20"} +{"current_steps": 439, "total_steps": 1955, "loss": 1.0613, "lr": 7.62916302637507e-05, "epoch": 1.1227621483375958, "percentage": 22.46, "elapsed_time": "2:12:55", "remaining_time": "7:39:02"} +{"current_steps": 440, "total_steps": 1955, "loss": 1.0389, "lr": 7.62615314245985e-05, "epoch": 1.1253196930946292, "percentage": 22.51, "elapsed_time": "2:13:13", "remaining_time": "7:38:43"} +{"current_steps": 441, "total_steps": 1955, "loss": 1.0604, "lr": 7.623131691721556e-05, "epoch": 1.1278772378516624, "percentage": 22.56, "elapsed_time": "2:13:32", "remaining_time": "7:38:26"} +{"current_steps": 442, "total_steps": 1955, "loss": 1.0396, "lr": 7.620098683798113e-05, "epoch": 1.1304347826086956, "percentage": 22.61, "elapsed_time": "2:13:50", "remaining_time": "7:38:08"} +{"current_steps": 443, "total_steps": 1955, "loss": 1.064, "lr": 7.617054128364308e-05, "epoch": 1.132992327365729, "percentage": 22.66, "elapsed_time": "2:14:09", "remaining_time": "7:37:51"} +{"current_steps": 444, "total_steps": 1955, "loss": 1.0302, "lr": 7.613998035131767e-05, "epoch": 1.1355498721227621, "percentage": 22.71, "elapsed_time": "2:14:27", "remaining_time": "7:37:33"} +{"current_steps": 445, "total_steps": 1955, "loss": 1.0499, "lr": 7.610930413848916e-05, "epoch": 1.1381074168797953, "percentage": 22.76, "elapsed_time": "2:14:45", "remaining_time": "7:37:16"} +{"current_steps": 446, "total_steps": 1955, "loss": 1.0241, "lr": 7.607851274300954e-05, "epoch": 1.1406649616368287, "percentage": 22.81, "elapsed_time": "2:15:03", "remaining_time": "7:36:56"} +{"current_steps": 447, "total_steps": 1955, "loss": 1.0136, "lr": 7.604760626309822e-05, "epoch": 1.143222506393862, "percentage": 22.86, "elapsed_time": "2:15:21", "remaining_time": "7:36:38"} +{"current_steps": 448, "total_steps": 1955, "loss": 1.0478, "lr": 7.601658479734172e-05, "epoch": 1.145780051150895, "percentage": 22.92, "elapsed_time": "2:15:39", "remaining_time": "7:36:19"} +{"current_steps": 449, "total_steps": 1955, "loss": 1.0371, "lr": 7.598544844469335e-05, "epoch": 1.1483375959079285, "percentage": 22.97, "elapsed_time": "2:15:57", "remaining_time": "7:36:01"} +{"current_steps": 450, "total_steps": 1955, "loss": 1.0871, "lr": 7.595419730447283e-05, "epoch": 1.1508951406649617, "percentage": 23.02, "elapsed_time": "2:16:15", "remaining_time": "7:35:42"} +{"current_steps": 451, "total_steps": 1955, "loss": 1.0584, "lr": 7.592283147636613e-05, "epoch": 1.1534526854219949, "percentage": 23.07, "elapsed_time": "2:16:33", "remaining_time": "7:35:23"} +{"current_steps": 452, "total_steps": 1955, "loss": 1.061, "lr": 7.589135106042497e-05, "epoch": 1.156010230179028, "percentage": 23.12, "elapsed_time": "2:16:51", "remaining_time": "7:35:05"} +{"current_steps": 453, "total_steps": 1955, "loss": 1.0782, "lr": 7.585975615706664e-05, "epoch": 1.1585677749360614, "percentage": 23.17, "elapsed_time": "2:17:10", "remaining_time": "7:34:48"} +{"current_steps": 454, "total_steps": 1955, "loss": 1.0487, "lr": 7.582804686707361e-05, "epoch": 1.1611253196930946, "percentage": 23.22, "elapsed_time": "2:17:28", "remaining_time": "7:34:29"} +{"current_steps": 455, "total_steps": 1955, "loss": 1.0443, "lr": 7.57962232915932e-05, "epoch": 1.1636828644501278, "percentage": 23.27, "elapsed_time": "2:17:46", "remaining_time": "7:34:11"} +{"current_steps": 456, "total_steps": 1955, "loss": 1.0366, "lr": 7.576428553213731e-05, "epoch": 1.1662404092071612, "percentage": 23.32, "elapsed_time": "2:18:04", "remaining_time": "7:33:53"} +{"current_steps": 457, "total_steps": 1955, "loss": 1.0295, "lr": 7.573223369058206e-05, "epoch": 1.1687979539641944, "percentage": 23.38, "elapsed_time": "2:18:22", "remaining_time": "7:33:34"} +{"current_steps": 458, "total_steps": 1955, "loss": 1.0329, "lr": 7.570006786916749e-05, "epoch": 1.1713554987212276, "percentage": 23.43, "elapsed_time": "2:18:40", "remaining_time": "7:33:16"} +{"current_steps": 459, "total_steps": 1955, "loss": 1.0402, "lr": 7.566778817049718e-05, "epoch": 1.1739130434782608, "percentage": 23.48, "elapsed_time": "2:18:58", "remaining_time": "7:32:57"} +{"current_steps": 460, "total_steps": 1955, "loss": 1.069, "lr": 7.563539469753798e-05, "epoch": 1.1764705882352942, "percentage": 23.53, "elapsed_time": "2:19:16", "remaining_time": "7:32:39"} +{"current_steps": 461, "total_steps": 1955, "loss": 1.0908, "lr": 7.560288755361967e-05, "epoch": 1.1790281329923273, "percentage": 23.58, "elapsed_time": "2:19:34", "remaining_time": "7:32:20"} +{"current_steps": 462, "total_steps": 1955, "loss": 1.0348, "lr": 7.557026684243459e-05, "epoch": 1.1815856777493605, "percentage": 23.63, "elapsed_time": "2:19:53", "remaining_time": "7:32:03"} +{"current_steps": 463, "total_steps": 1955, "loss": 1.0241, "lr": 7.55375326680374e-05, "epoch": 1.184143222506394, "percentage": 23.68, "elapsed_time": "2:20:11", "remaining_time": "7:31:44"} +{"current_steps": 464, "total_steps": 1955, "loss": 1.0514, "lr": 7.550468513484461e-05, "epoch": 1.186700767263427, "percentage": 23.73, "elapsed_time": "2:20:29", "remaining_time": "7:31:26"} +{"current_steps": 465, "total_steps": 1955, "loss": 1.0606, "lr": 7.54717243476344e-05, "epoch": 1.1892583120204603, "percentage": 23.79, "elapsed_time": "2:20:47", "remaining_time": "7:31:07"} +{"current_steps": 466, "total_steps": 1955, "loss": 1.0563, "lr": 7.543865041154616e-05, "epoch": 1.1918158567774937, "percentage": 23.84, "elapsed_time": "2:21:05", "remaining_time": "7:30:49"} +{"current_steps": 467, "total_steps": 1955, "loss": 1.0336, "lr": 7.540546343208021e-05, "epoch": 1.1943734015345269, "percentage": 23.89, "elapsed_time": "2:21:23", "remaining_time": "7:30:31"} +{"current_steps": 468, "total_steps": 1955, "loss": 1.0844, "lr": 7.537216351509752e-05, "epoch": 1.19693094629156, "percentage": 23.94, "elapsed_time": "2:21:41", "remaining_time": "7:30:13"} +{"current_steps": 469, "total_steps": 1955, "loss": 1.0492, "lr": 7.533875076681923e-05, "epoch": 1.1994884910485935, "percentage": 23.99, "elapsed_time": "2:21:59", "remaining_time": "7:29:54"} +{"current_steps": 470, "total_steps": 1955, "loss": 1.071, "lr": 7.530522529382641e-05, "epoch": 1.2020460358056266, "percentage": 24.04, "elapsed_time": "2:22:18", "remaining_time": "7:29:37"} +{"current_steps": 471, "total_steps": 1955, "loss": 1.0429, "lr": 7.527158720305975e-05, "epoch": 1.2046035805626598, "percentage": 24.09, "elapsed_time": "2:22:36", "remaining_time": "7:29:18"} +{"current_steps": 472, "total_steps": 1955, "loss": 1.0021, "lr": 7.523783660181915e-05, "epoch": 1.207161125319693, "percentage": 24.14, "elapsed_time": "2:22:54", "remaining_time": "7:28:59"} +{"current_steps": 473, "total_steps": 1955, "loss": 1.0574, "lr": 7.520397359776338e-05, "epoch": 1.2097186700767264, "percentage": 24.19, "elapsed_time": "2:23:12", "remaining_time": "7:28:41"} +{"current_steps": 474, "total_steps": 1955, "loss": 1.0409, "lr": 7.516999829890976e-05, "epoch": 1.2122762148337596, "percentage": 24.25, "elapsed_time": "2:23:30", "remaining_time": "7:28:22"} +{"current_steps": 475, "total_steps": 1955, "loss": 1.0555, "lr": 7.513591081363384e-05, "epoch": 1.2148337595907928, "percentage": 24.3, "elapsed_time": "2:23:48", "remaining_time": "7:28:04"} +{"current_steps": 476, "total_steps": 1955, "loss": 1.0253, "lr": 7.5101711250669e-05, "epoch": 1.2173913043478262, "percentage": 24.35, "elapsed_time": "2:24:05", "remaining_time": "7:27:43"} +{"current_steps": 477, "total_steps": 1955, "loss": 1.0792, "lr": 7.506739971910613e-05, "epoch": 1.2199488491048593, "percentage": 24.4, "elapsed_time": "2:24:23", "remaining_time": "7:27:25"} +{"current_steps": 478, "total_steps": 1955, "loss": 1.0618, "lr": 7.503297632839328e-05, "epoch": 1.2225063938618925, "percentage": 24.45, "elapsed_time": "2:24:41", "remaining_time": "7:27:06"} +{"current_steps": 479, "total_steps": 1955, "loss": 1.0711, "lr": 7.499844118833533e-05, "epoch": 1.2250639386189257, "percentage": 24.5, "elapsed_time": "2:25:00", "remaining_time": "7:26:50"} +{"current_steps": 480, "total_steps": 1955, "loss": 1.0225, "lr": 7.496379440909361e-05, "epoch": 1.227621483375959, "percentage": 24.55, "elapsed_time": "2:25:18", "remaining_time": "7:26:31"} +{"current_steps": 481, "total_steps": 1955, "loss": 1.0322, "lr": 7.492903610118556e-05, "epoch": 1.2301790281329923, "percentage": 24.6, "elapsed_time": "2:25:36", "remaining_time": "7:26:13"} +{"current_steps": 482, "total_steps": 1955, "loss": 1.0511, "lr": 7.489416637548436e-05, "epoch": 1.2327365728900257, "percentage": 24.65, "elapsed_time": "2:25:54", "remaining_time": "7:25:54"} +{"current_steps": 483, "total_steps": 1955, "loss": 1.0439, "lr": 7.485918534321866e-05, "epoch": 1.2352941176470589, "percentage": 24.71, "elapsed_time": "2:26:12", "remaining_time": "7:25:36"} +{"current_steps": 484, "total_steps": 1955, "loss": 1.0597, "lr": 7.482409311597206e-05, "epoch": 1.237851662404092, "percentage": 24.76, "elapsed_time": "2:26:30", "remaining_time": "7:25:17"} +{"current_steps": 485, "total_steps": 1955, "loss": 1.0622, "lr": 7.478888980568294e-05, "epoch": 1.2404092071611252, "percentage": 24.81, "elapsed_time": "2:26:49", "remaining_time": "7:24:59"} +{"current_steps": 486, "total_steps": 1955, "loss": 1.0559, "lr": 7.475357552464397e-05, "epoch": 1.2429667519181586, "percentage": 24.86, "elapsed_time": "2:27:07", "remaining_time": "7:24:41"} +{"current_steps": 487, "total_steps": 1955, "loss": 1.0283, "lr": 7.47181503855018e-05, "epoch": 1.2455242966751918, "percentage": 24.91, "elapsed_time": "2:27:25", "remaining_time": "7:24:23"} +{"current_steps": 488, "total_steps": 1955, "loss": 1.0463, "lr": 7.468261450125674e-05, "epoch": 1.248081841432225, "percentage": 24.96, "elapsed_time": "2:27:43", "remaining_time": "7:24:04"} +{"current_steps": 489, "total_steps": 1955, "loss": 1.0553, "lr": 7.464696798526229e-05, "epoch": 1.2506393861892584, "percentage": 25.01, "elapsed_time": "2:28:01", "remaining_time": "7:23:46"} +{"current_steps": 490, "total_steps": 1955, "loss": 1.0754, "lr": 7.461121095122491e-05, "epoch": 1.2531969309462916, "percentage": 25.06, "elapsed_time": "2:28:19", "remaining_time": "7:23:27"} +{"current_steps": 491, "total_steps": 1955, "loss": 1.0475, "lr": 7.457534351320356e-05, "epoch": 1.2557544757033248, "percentage": 25.12, "elapsed_time": "2:28:37", "remaining_time": "7:23:09"} +{"current_steps": 492, "total_steps": 1955, "loss": 1.0724, "lr": 7.453936578560937e-05, "epoch": 1.258312020460358, "percentage": 25.17, "elapsed_time": "2:28:55", "remaining_time": "7:22:50"} +{"current_steps": 493, "total_steps": 1955, "loss": 1.0411, "lr": 7.450327788320529e-05, "epoch": 1.2608695652173914, "percentage": 25.22, "elapsed_time": "2:29:13", "remaining_time": "7:22:32"} +{"current_steps": 494, "total_steps": 1955, "loss": 1.0391, "lr": 7.446707992110571e-05, "epoch": 1.2634271099744245, "percentage": 25.27, "elapsed_time": "2:29:32", "remaining_time": "7:22:14"} +{"current_steps": 495, "total_steps": 1955, "loss": 1.0631, "lr": 7.443077201477609e-05, "epoch": 1.265984654731458, "percentage": 25.32, "elapsed_time": "2:29:50", "remaining_time": "7:21:57"} +{"current_steps": 496, "total_steps": 1955, "loss": 1.0523, "lr": 7.439435428003256e-05, "epoch": 1.2685421994884911, "percentage": 25.37, "elapsed_time": "2:30:09", "remaining_time": "7:21:40"} +{"current_steps": 497, "total_steps": 1955, "loss": 1.0446, "lr": 7.435782683304164e-05, "epoch": 1.2710997442455243, "percentage": 25.42, "elapsed_time": "2:30:27", "remaining_time": "7:21:22"} +{"current_steps": 498, "total_steps": 1955, "loss": 1.0306, "lr": 7.43211897903198e-05, "epoch": 1.2736572890025575, "percentage": 25.47, "elapsed_time": "2:30:45", "remaining_time": "7:21:04"} +{"current_steps": 499, "total_steps": 1955, "loss": 1.0448, "lr": 7.428444326873303e-05, "epoch": 1.2762148337595907, "percentage": 25.52, "elapsed_time": "2:31:03", "remaining_time": "7:20:45"} +{"current_steps": 500, "total_steps": 1955, "loss": 1.0201, "lr": 7.424758738549667e-05, "epoch": 1.278772378516624, "percentage": 25.58, "elapsed_time": "2:31:21", "remaining_time": "7:20:27"} +{"current_steps": 501, "total_steps": 1955, "loss": 1.0397, "lr": 7.421062225817478e-05, "epoch": 1.2813299232736572, "percentage": 25.63, "elapsed_time": "2:31:39", "remaining_time": "7:20:09"} +{"current_steps": 502, "total_steps": 1955, "loss": 1.0449, "lr": 7.417354800467997e-05, "epoch": 1.2838874680306906, "percentage": 25.68, "elapsed_time": "2:31:58", "remaining_time": "7:19:53"} +{"current_steps": 503, "total_steps": 1955, "loss": 1.0642, "lr": 7.413636474327292e-05, "epoch": 1.2864450127877238, "percentage": 25.73, "elapsed_time": "2:32:16", "remaining_time": "7:19:34"} +{"current_steps": 504, "total_steps": 1955, "loss": 1.0397, "lr": 7.409907259256202e-05, "epoch": 1.289002557544757, "percentage": 25.78, "elapsed_time": "2:32:34", "remaining_time": "7:19:16"} +{"current_steps": 505, "total_steps": 1955, "loss": 1.0542, "lr": 7.4061671671503e-05, "epoch": 1.2915601023017902, "percentage": 25.83, "elapsed_time": "2:32:53", "remaining_time": "7:18:59"} +{"current_steps": 506, "total_steps": 1955, "loss": 1.0304, "lr": 7.402416209939856e-05, "epoch": 1.2941176470588236, "percentage": 25.88, "elapsed_time": "2:33:11", "remaining_time": "7:18:40"} +{"current_steps": 507, "total_steps": 1955, "loss": 1.0632, "lr": 7.398654399589795e-05, "epoch": 1.2966751918158568, "percentage": 25.93, "elapsed_time": "2:33:29", "remaining_time": "7:18:21"} +{"current_steps": 508, "total_steps": 1955, "loss": 1.026, "lr": 7.394881748099668e-05, "epoch": 1.29923273657289, "percentage": 25.98, "elapsed_time": "2:33:47", "remaining_time": "7:18:03"} +{"current_steps": 509, "total_steps": 1955, "loss": 1.0584, "lr": 7.391098267503599e-05, "epoch": 1.3017902813299234, "percentage": 26.04, "elapsed_time": "2:34:06", "remaining_time": "7:17:47"} +{"current_steps": 510, "total_steps": 1955, "loss": 1.0265, "lr": 7.387303969870262e-05, "epoch": 1.3043478260869565, "percentage": 26.09, "elapsed_time": "2:34:24", "remaining_time": "7:17:28"} +{"current_steps": 511, "total_steps": 1955, "loss": 1.0362, "lr": 7.383498867302835e-05, "epoch": 1.3069053708439897, "percentage": 26.14, "elapsed_time": "2:34:42", "remaining_time": "7:17:10"} +{"current_steps": 512, "total_steps": 1955, "loss": 1.0658, "lr": 7.379682971938953e-05, "epoch": 1.309462915601023, "percentage": 26.19, "elapsed_time": "2:35:00", "remaining_time": "7:16:51"} +{"current_steps": 513, "total_steps": 1955, "loss": 1.0222, "lr": 7.375856295950693e-05, "epoch": 1.3120204603580563, "percentage": 26.24, "elapsed_time": "2:35:18", "remaining_time": "7:16:33"} +{"current_steps": 514, "total_steps": 1955, "loss": 1.0431, "lr": 7.372018851544507e-05, "epoch": 1.3145780051150895, "percentage": 26.29, "elapsed_time": "2:35:36", "remaining_time": "7:16:15"} +{"current_steps": 515, "total_steps": 1955, "loss": 1.0349, "lr": 7.368170650961203e-05, "epoch": 1.317135549872123, "percentage": 26.34, "elapsed_time": "2:35:55", "remaining_time": "7:15:58"} +{"current_steps": 516, "total_steps": 1955, "loss": 1.0304, "lr": 7.364311706475901e-05, "epoch": 1.319693094629156, "percentage": 26.39, "elapsed_time": "2:36:13", "remaining_time": "7:15:40"} +{"current_steps": 517, "total_steps": 1955, "loss": 1.0267, "lr": 7.360442030397987e-05, "epoch": 1.3222506393861893, "percentage": 26.45, "elapsed_time": "2:36:31", "remaining_time": "7:15:21"} +{"current_steps": 518, "total_steps": 1955, "loss": 1.0437, "lr": 7.356561635071082e-05, "epoch": 1.3248081841432224, "percentage": 26.5, "elapsed_time": "2:36:49", "remaining_time": "7:15:04"} +{"current_steps": 519, "total_steps": 1955, "loss": 1.0342, "lr": 7.352670532872999e-05, "epoch": 1.3273657289002558, "percentage": 26.55, "elapsed_time": "2:37:07", "remaining_time": "7:14:45"} +{"current_steps": 520, "total_steps": 1955, "loss": 1.0784, "lr": 7.348768736215706e-05, "epoch": 1.329923273657289, "percentage": 26.6, "elapsed_time": "2:37:25", "remaining_time": "7:14:27"} +{"current_steps": 521, "total_steps": 1955, "loss": 1.0785, "lr": 7.34485625754528e-05, "epoch": 1.3324808184143222, "percentage": 26.65, "elapsed_time": "2:37:43", "remaining_time": "7:14:07"} +{"current_steps": 522, "total_steps": 1955, "loss": 1.0554, "lr": 7.340933109341878e-05, "epoch": 1.3350383631713556, "percentage": 26.7, "elapsed_time": "2:38:01", "remaining_time": "7:13:48"} +{"current_steps": 523, "total_steps": 1955, "loss": 1.0461, "lr": 7.336999304119685e-05, "epoch": 1.3375959079283888, "percentage": 26.75, "elapsed_time": "2:38:20", "remaining_time": "7:13:31"} +{"current_steps": 524, "total_steps": 1955, "loss": 1.0714, "lr": 7.333054854426884e-05, "epoch": 1.340153452685422, "percentage": 26.8, "elapsed_time": "2:38:38", "remaining_time": "7:13:13"} +{"current_steps": 525, "total_steps": 1955, "loss": 1.0427, "lr": 7.329099772845611e-05, "epoch": 1.3427109974424551, "percentage": 26.85, "elapsed_time": "2:38:56", "remaining_time": "7:12:55"} +{"current_steps": 526, "total_steps": 1955, "loss": 1.0673, "lr": 7.325134071991915e-05, "epoch": 1.3452685421994885, "percentage": 26.91, "elapsed_time": "2:39:15", "remaining_time": "7:12:40"} +{"current_steps": 527, "total_steps": 1955, "loss": 1.0172, "lr": 7.321157764515721e-05, "epoch": 1.3478260869565217, "percentage": 26.96, "elapsed_time": "2:39:33", "remaining_time": "7:12:22"} +{"current_steps": 528, "total_steps": 1955, "loss": 1.0814, "lr": 7.317170863100786e-05, "epoch": 1.350383631713555, "percentage": 27.01, "elapsed_time": "2:39:51", "remaining_time": "7:12:03"} +{"current_steps": 529, "total_steps": 1955, "loss": 1.0427, "lr": 7.313173380464658e-05, "epoch": 1.3529411764705883, "percentage": 27.06, "elapsed_time": "2:40:09", "remaining_time": "7:11:45"} +{"current_steps": 530, "total_steps": 1955, "loss": 1.0141, "lr": 7.309165329358639e-05, "epoch": 1.3554987212276215, "percentage": 27.11, "elapsed_time": "2:40:27", "remaining_time": "7:11:26"} +{"current_steps": 531, "total_steps": 1955, "loss": 1.0518, "lr": 7.305146722567746e-05, "epoch": 1.3580562659846547, "percentage": 27.16, "elapsed_time": "2:40:46", "remaining_time": "7:11:08"} +{"current_steps": 532, "total_steps": 1955, "loss": 1.0374, "lr": 7.301117572910661e-05, "epoch": 1.3606138107416879, "percentage": 27.21, "elapsed_time": "2:41:04", "remaining_time": "7:10:49"} +{"current_steps": 533, "total_steps": 1955, "loss": 1.0751, "lr": 7.297077893239699e-05, "epoch": 1.3631713554987213, "percentage": 27.26, "elapsed_time": "2:41:22", "remaining_time": "7:10:30"} +{"current_steps": 534, "total_steps": 1955, "loss": 1.0429, "lr": 7.293027696440763e-05, "epoch": 1.3657289002557544, "percentage": 27.31, "elapsed_time": "2:41:40", "remaining_time": "7:10:12"} +{"current_steps": 535, "total_steps": 1955, "loss": 1.0531, "lr": 7.288966995433309e-05, "epoch": 1.3682864450127878, "percentage": 27.37, "elapsed_time": "2:41:58", "remaining_time": "7:09:53"} +{"current_steps": 536, "total_steps": 1955, "loss": 1.0875, "lr": 7.284895803170288e-05, "epoch": 1.370843989769821, "percentage": 27.42, "elapsed_time": "2:42:16", "remaining_time": "7:09:37"} +{"current_steps": 537, "total_steps": 1955, "loss": 1.0439, "lr": 7.280814132638129e-05, "epoch": 1.3734015345268542, "percentage": 27.47, "elapsed_time": "2:42:34", "remaining_time": "7:09:18"} +{"current_steps": 538, "total_steps": 1955, "loss": 1.0554, "lr": 7.276721996856677e-05, "epoch": 1.3759590792838874, "percentage": 27.52, "elapsed_time": "2:42:53", "remaining_time": "7:09:00"} +{"current_steps": 539, "total_steps": 1955, "loss": 1.0404, "lr": 7.272619408879164e-05, "epoch": 1.3785166240409208, "percentage": 27.57, "elapsed_time": "2:43:11", "remaining_time": "7:08:42"} +{"current_steps": 540, "total_steps": 1955, "loss": 1.049, "lr": 7.268506381792158e-05, "epoch": 1.381074168797954, "percentage": 27.62, "elapsed_time": "2:43:29", "remaining_time": "7:08:23"} +{"current_steps": 541, "total_steps": 1955, "loss": 1.076, "lr": 7.26438292871553e-05, "epoch": 1.3836317135549872, "percentage": 27.67, "elapsed_time": "2:43:47", "remaining_time": "7:08:05"} +{"current_steps": 542, "total_steps": 1955, "loss": 1.0677, "lr": 7.260249062802407e-05, "epoch": 1.3861892583120206, "percentage": 27.72, "elapsed_time": "2:44:05", "remaining_time": "7:07:46"} +{"current_steps": 543, "total_steps": 1955, "loss": 1.0227, "lr": 7.256104797239132e-05, "epoch": 1.3887468030690537, "percentage": 27.77, "elapsed_time": "2:44:23", "remaining_time": "7:07:28"} +{"current_steps": 544, "total_steps": 1955, "loss": 1.044, "lr": 7.251950145245218e-05, "epoch": 1.391304347826087, "percentage": 27.83, "elapsed_time": "2:44:41", "remaining_time": "7:07:09"} +{"current_steps": 545, "total_steps": 1955, "loss": 1.0333, "lr": 7.247785120073313e-05, "epoch": 1.39386189258312, "percentage": 27.88, "elapsed_time": "2:44:59", "remaining_time": "7:06:51"} +{"current_steps": 546, "total_steps": 1955, "loss": 1.0128, "lr": 7.24360973500915e-05, "epoch": 1.3964194373401535, "percentage": 27.93, "elapsed_time": "2:45:17", "remaining_time": "7:06:33"} +{"current_steps": 547, "total_steps": 1955, "loss": 1.0519, "lr": 7.239424003371511e-05, "epoch": 1.3989769820971867, "percentage": 27.98, "elapsed_time": "2:45:35", "remaining_time": "7:06:15"} +{"current_steps": 548, "total_steps": 1955, "loss": 1.0421, "lr": 7.235227938512182e-05, "epoch": 1.40153452685422, "percentage": 28.03, "elapsed_time": "2:45:53", "remaining_time": "7:05:54"} +{"current_steps": 549, "total_steps": 1955, "loss": 1.0647, "lr": 7.231021553815909e-05, "epoch": 1.4040920716112533, "percentage": 28.08, "elapsed_time": "2:46:11", "remaining_time": "7:05:36"} +{"current_steps": 550, "total_steps": 1955, "loss": 1.006, "lr": 7.226804862700357e-05, "epoch": 1.4066496163682864, "percentage": 28.13, "elapsed_time": "2:46:29", "remaining_time": "7:05:17"} +{"current_steps": 551, "total_steps": 1955, "loss": 1.018, "lr": 7.222577878616065e-05, "epoch": 1.4092071611253196, "percentage": 28.18, "elapsed_time": "2:46:47", "remaining_time": "7:05:01"} +{"current_steps": 552, "total_steps": 1955, "loss": 1.0609, "lr": 7.218340615046407e-05, "epoch": 1.4117647058823528, "percentage": 28.24, "elapsed_time": "2:47:06", "remaining_time": "7:04:42"} +{"current_steps": 553, "total_steps": 1955, "loss": 1.0455, "lr": 7.214093085507549e-05, "epoch": 1.4143222506393862, "percentage": 28.29, "elapsed_time": "2:47:24", "remaining_time": "7:04:24"} +{"current_steps": 554, "total_steps": 1955, "loss": 1.0558, "lr": 7.209835303548396e-05, "epoch": 1.4168797953964194, "percentage": 28.34, "elapsed_time": "2:47:41", "remaining_time": "7:04:04"} +{"current_steps": 555, "total_steps": 1955, "loss": 1.069, "lr": 7.205567282750567e-05, "epoch": 1.4194373401534528, "percentage": 28.39, "elapsed_time": "2:47:59", "remaining_time": "7:03:46"} +{"current_steps": 556, "total_steps": 1955, "loss": 1.0536, "lr": 7.201289036728331e-05, "epoch": 1.421994884910486, "percentage": 28.44, "elapsed_time": "2:48:17", "remaining_time": "7:03:27"} +{"current_steps": 557, "total_steps": 1955, "loss": 1.0403, "lr": 7.197000579128582e-05, "epoch": 1.4245524296675192, "percentage": 28.49, "elapsed_time": "2:48:35", "remaining_time": "7:03:09"} +{"current_steps": 558, "total_steps": 1955, "loss": 1.0328, "lr": 7.192701923630781e-05, "epoch": 1.4271099744245523, "percentage": 28.54, "elapsed_time": "2:48:53", "remaining_time": "7:02:50"} +{"current_steps": 559, "total_steps": 1955, "loss": 1.0258, "lr": 7.188393083946923e-05, "epoch": 1.4296675191815857, "percentage": 28.59, "elapsed_time": "2:49:11", "remaining_time": "7:02:32"} +{"current_steps": 560, "total_steps": 1955, "loss": 1.0174, "lr": 7.184074073821487e-05, "epoch": 1.432225063938619, "percentage": 28.64, "elapsed_time": "2:49:29", "remaining_time": "7:02:13"} +{"current_steps": 561, "total_steps": 1955, "loss": 1.0287, "lr": 7.179744907031394e-05, "epoch": 1.434782608695652, "percentage": 28.7, "elapsed_time": "2:49:48", "remaining_time": "7:01:57"} +{"current_steps": 562, "total_steps": 1955, "loss": 1.0412, "lr": 7.175405597385964e-05, "epoch": 1.4373401534526855, "percentage": 28.75, "elapsed_time": "2:50:06", "remaining_time": "7:01:38"} +{"current_steps": 563, "total_steps": 1955, "loss": 1.0688, "lr": 7.171056158726869e-05, "epoch": 1.4398976982097187, "percentage": 28.8, "elapsed_time": "2:50:24", "remaining_time": "7:01:20"} +{"current_steps": 564, "total_steps": 1955, "loss": 1.0668, "lr": 7.166696604928091e-05, "epoch": 1.4424552429667519, "percentage": 28.85, "elapsed_time": "2:50:42", "remaining_time": "7:01:01"} +{"current_steps": 565, "total_steps": 1955, "loss": 1.0752, "lr": 7.162326949895882e-05, "epoch": 1.445012787723785, "percentage": 28.9, "elapsed_time": "2:51:00", "remaining_time": "7:00:42"} +{"current_steps": 566, "total_steps": 1955, "loss": 1.038, "lr": 7.157947207568708e-05, "epoch": 1.4475703324808185, "percentage": 28.95, "elapsed_time": "2:51:18", "remaining_time": "7:00:23"} +{"current_steps": 567, "total_steps": 1955, "loss": 1.0186, "lr": 7.153557391917221e-05, "epoch": 1.4501278772378516, "percentage": 29.0, "elapsed_time": "2:51:36", "remaining_time": "7:00:05"} +{"current_steps": 568, "total_steps": 1955, "loss": 1.0496, "lr": 7.149157516944195e-05, "epoch": 1.452685421994885, "percentage": 29.05, "elapsed_time": "2:51:54", "remaining_time": "6:59:46"} +{"current_steps": 569, "total_steps": 1955, "loss": 1.0401, "lr": 7.144747596684498e-05, "epoch": 1.4552429667519182, "percentage": 29.1, "elapsed_time": "2:52:12", "remaining_time": "6:59:28"} +{"current_steps": 570, "total_steps": 1955, "loss": 1.0298, "lr": 7.14032764520504e-05, "epoch": 1.4578005115089514, "percentage": 29.16, "elapsed_time": "2:52:30", "remaining_time": "6:59:10"} +{"current_steps": 571, "total_steps": 1955, "loss": 1.0504, "lr": 7.135897676604728e-05, "epoch": 1.4603580562659846, "percentage": 29.21, "elapsed_time": "2:52:48", "remaining_time": "6:58:51"} +{"current_steps": 572, "total_steps": 1955, "loss": 1.0553, "lr": 7.131457705014424e-05, "epoch": 1.4629156010230178, "percentage": 29.26, "elapsed_time": "2:53:07", "remaining_time": "6:58:35"} +{"current_steps": 573, "total_steps": 1955, "loss": 1.0402, "lr": 7.127007744596891e-05, "epoch": 1.4654731457800512, "percentage": 29.31, "elapsed_time": "2:53:25", "remaining_time": "6:58:16"} +{"current_steps": 574, "total_steps": 1955, "loss": 1.0266, "lr": 7.122547809546766e-05, "epoch": 1.4680306905370843, "percentage": 29.36, "elapsed_time": "2:53:43", "remaining_time": "6:57:58"} +{"current_steps": 575, "total_steps": 1955, "loss": 1.0593, "lr": 7.118077914090492e-05, "epoch": 1.4705882352941178, "percentage": 29.41, "elapsed_time": "2:54:01", "remaining_time": "6:57:39"} +{"current_steps": 576, "total_steps": 1955, "loss": 1.0628, "lr": 7.113598072486291e-05, "epoch": 1.473145780051151, "percentage": 29.46, "elapsed_time": "2:54:20", "remaining_time": "6:57:23"} +{"current_steps": 577, "total_steps": 1955, "loss": 1.0545, "lr": 7.10910829902411e-05, "epoch": 1.4757033248081841, "percentage": 29.51, "elapsed_time": "2:54:37", "remaining_time": "6:57:03"} +{"current_steps": 578, "total_steps": 1955, "loss": 1.0441, "lr": 7.104608608025578e-05, "epoch": 1.4782608695652173, "percentage": 29.57, "elapsed_time": "2:54:55", "remaining_time": "6:56:44"} +{"current_steps": 579, "total_steps": 1955, "loss": 1.0233, "lr": 7.100099013843956e-05, "epoch": 1.4808184143222507, "percentage": 29.62, "elapsed_time": "2:55:13", "remaining_time": "6:56:26"} +{"current_steps": 580, "total_steps": 1955, "loss": 1.0393, "lr": 7.095579530864097e-05, "epoch": 1.4833759590792839, "percentage": 29.67, "elapsed_time": "2:55:31", "remaining_time": "6:56:07"} +{"current_steps": 581, "total_steps": 1955, "loss": 1.0578, "lr": 7.091050173502397e-05, "epoch": 1.485933503836317, "percentage": 29.72, "elapsed_time": "2:55:49", "remaining_time": "6:55:49"} +{"current_steps": 582, "total_steps": 1955, "loss": 1.0463, "lr": 7.08651095620675e-05, "epoch": 1.4884910485933505, "percentage": 29.77, "elapsed_time": "2:56:08", "remaining_time": "6:55:31"} +{"current_steps": 583, "total_steps": 1955, "loss": 1.0715, "lr": 7.081961893456502e-05, "epoch": 1.4910485933503836, "percentage": 29.82, "elapsed_time": "2:56:26", "remaining_time": "6:55:12"} +{"current_steps": 584, "total_steps": 1955, "loss": 1.0774, "lr": 7.077402999762401e-05, "epoch": 1.4936061381074168, "percentage": 29.87, "elapsed_time": "2:56:44", "remaining_time": "6:54:54"} +{"current_steps": 585, "total_steps": 1955, "loss": 1.0167, "lr": 7.07283428966656e-05, "epoch": 1.49616368286445, "percentage": 29.92, "elapsed_time": "2:57:02", "remaining_time": "6:54:35"} +{"current_steps": 586, "total_steps": 1955, "loss": 1.0325, "lr": 7.068255777742399e-05, "epoch": 1.4987212276214834, "percentage": 29.97, "elapsed_time": "2:57:20", "remaining_time": "6:54:17"} +{"current_steps": 587, "total_steps": 1955, "loss": 1.0408, "lr": 7.063667478594609e-05, "epoch": 1.5012787723785166, "percentage": 30.03, "elapsed_time": "2:57:38", "remaining_time": "6:53:58"} +{"current_steps": 588, "total_steps": 1955, "loss": 1.0671, "lr": 7.059069406859096e-05, "epoch": 1.50383631713555, "percentage": 30.08, "elapsed_time": "2:57:56", "remaining_time": "6:53:40"} +{"current_steps": 589, "total_steps": 1955, "loss": 1.0071, "lr": 7.05446157720294e-05, "epoch": 1.5063938618925832, "percentage": 30.13, "elapsed_time": "2:58:14", "remaining_time": "6:53:22"} +{"current_steps": 590, "total_steps": 1955, "loss": 1.0364, "lr": 7.049844004324352e-05, "epoch": 1.5089514066496164, "percentage": 30.18, "elapsed_time": "2:58:31", "remaining_time": "6:53:02"} +{"current_steps": 591, "total_steps": 1955, "loss": 1.0099, "lr": 7.045216702952615e-05, "epoch": 1.5115089514066495, "percentage": 30.23, "elapsed_time": "2:58:50", "remaining_time": "6:52:44"} +{"current_steps": 592, "total_steps": 1955, "loss": 1.0629, "lr": 7.040579687848051e-05, "epoch": 1.5140664961636827, "percentage": 30.28, "elapsed_time": "2:59:08", "remaining_time": "6:52:25"} +{"current_steps": 593, "total_steps": 1955, "loss": 0.9933, "lr": 7.035932973801961e-05, "epoch": 1.5166240409207161, "percentage": 30.33, "elapsed_time": "2:59:26", "remaining_time": "6:52:07"} +{"current_steps": 594, "total_steps": 1955, "loss": 1.0581, "lr": 7.031276575636588e-05, "epoch": 1.5191815856777495, "percentage": 30.38, "elapsed_time": "2:59:44", "remaining_time": "6:51:48"} +{"current_steps": 595, "total_steps": 1955, "loss": 1.03, "lr": 7.026610508205065e-05, "epoch": 1.5217391304347827, "percentage": 30.43, "elapsed_time": "3:00:02", "remaining_time": "6:51:30"} +{"current_steps": 596, "total_steps": 1955, "loss": 1.0404, "lr": 7.021934786391369e-05, "epoch": 1.5242966751918159, "percentage": 30.49, "elapsed_time": "3:00:20", "remaining_time": "6:51:12"} +{"current_steps": 597, "total_steps": 1955, "loss": 1.0477, "lr": 7.017249425110268e-05, "epoch": 1.526854219948849, "percentage": 30.54, "elapsed_time": "3:00:38", "remaining_time": "6:50:53"} +{"current_steps": 598, "total_steps": 1955, "loss": 1.0496, "lr": 7.012554439307288e-05, "epoch": 1.5294117647058822, "percentage": 30.59, "elapsed_time": "3:00:56", "remaining_time": "6:50:35"} +{"current_steps": 599, "total_steps": 1955, "loss": 1.027, "lr": 7.007849843958646e-05, "epoch": 1.5319693094629157, "percentage": 30.64, "elapsed_time": "3:01:14", "remaining_time": "6:50:17"} +{"current_steps": 600, "total_steps": 1955, "loss": 1.0771, "lr": 7.003135654071216e-05, "epoch": 1.5345268542199488, "percentage": 30.69, "elapsed_time": "3:01:32", "remaining_time": "6:49:58"} +{"current_steps": 601, "total_steps": 1955, "loss": 1.0245, "lr": 6.998411884682476e-05, "epoch": 1.5370843989769822, "percentage": 30.74, "elapsed_time": "3:01:50", "remaining_time": "6:49:40"} +{"current_steps": 602, "total_steps": 1955, "loss": 1.0208, "lr": 6.993678550860464e-05, "epoch": 1.5396419437340154, "percentage": 30.79, "elapsed_time": "3:02:08", "remaining_time": "6:49:21"} +{"current_steps": 603, "total_steps": 1955, "loss": 1.0341, "lr": 6.988935667703722e-05, "epoch": 1.5421994884910486, "percentage": 30.84, "elapsed_time": "3:02:27", "remaining_time": "6:49:05"} +{"current_steps": 604, "total_steps": 1955, "loss": 1.0454, "lr": 6.984183250341257e-05, "epoch": 1.5447570332480818, "percentage": 30.9, "elapsed_time": "3:02:45", "remaining_time": "6:48:46"} +{"current_steps": 605, "total_steps": 1955, "loss": 1.0315, "lr": 6.979421313932488e-05, "epoch": 1.547314578005115, "percentage": 30.95, "elapsed_time": "3:03:03", "remaining_time": "6:48:28"} +{"current_steps": 606, "total_steps": 1955, "loss": 1.0086, "lr": 6.974649873667194e-05, "epoch": 1.5498721227621484, "percentage": 31.0, "elapsed_time": "3:03:21", "remaining_time": "6:48:09"} +{"current_steps": 607, "total_steps": 1955, "loss": 1.0323, "lr": 6.969868944765474e-05, "epoch": 1.5524296675191815, "percentage": 31.05, "elapsed_time": "3:03:39", "remaining_time": "6:47:51"} +{"current_steps": 608, "total_steps": 1955, "loss": 1.0575, "lr": 6.965078542477693e-05, "epoch": 1.554987212276215, "percentage": 31.1, "elapsed_time": "3:03:57", "remaining_time": "6:47:33"} +{"current_steps": 609, "total_steps": 1955, "loss": 1.0537, "lr": 6.960278682084436e-05, "epoch": 1.5575447570332481, "percentage": 31.15, "elapsed_time": "3:04:15", "remaining_time": "6:47:14"} +{"current_steps": 610, "total_steps": 1955, "loss": 1.0231, "lr": 6.955469378896456e-05, "epoch": 1.5601023017902813, "percentage": 31.2, "elapsed_time": "3:04:34", "remaining_time": "6:46:58"} +{"current_steps": 611, "total_steps": 1955, "loss": 1.0572, "lr": 6.950650648254623e-05, "epoch": 1.5626598465473145, "percentage": 31.25, "elapsed_time": "3:04:52", "remaining_time": "6:46:39"} +{"current_steps": 612, "total_steps": 1955, "loss": 1.0466, "lr": 6.945822505529887e-05, "epoch": 1.5652173913043477, "percentage": 31.3, "elapsed_time": "3:05:10", "remaining_time": "6:46:21"} +{"current_steps": 613, "total_steps": 1955, "loss": 1.0366, "lr": 6.940984966123217e-05, "epoch": 1.567774936061381, "percentage": 31.36, "elapsed_time": "3:05:28", "remaining_time": "6:46:03"} +{"current_steps": 614, "total_steps": 1955, "loss": 1.0415, "lr": 6.936138045465556e-05, "epoch": 1.5703324808184145, "percentage": 31.41, "elapsed_time": "3:05:46", "remaining_time": "6:45:44"} +{"current_steps": 615, "total_steps": 1955, "loss": 1.0591, "lr": 6.93128175901777e-05, "epoch": 1.5728900255754477, "percentage": 31.46, "elapsed_time": "3:06:04", "remaining_time": "6:45:26"} +{"current_steps": 616, "total_steps": 1955, "loss": 1.0144, "lr": 6.926416122270604e-05, "epoch": 1.5754475703324808, "percentage": 31.51, "elapsed_time": "3:06:22", "remaining_time": "6:45:07"} +{"current_steps": 617, "total_steps": 1955, "loss": 1.0473, "lr": 6.921541150744624e-05, "epoch": 1.578005115089514, "percentage": 31.56, "elapsed_time": "3:06:40", "remaining_time": "6:44:49"} +{"current_steps": 618, "total_steps": 1955, "loss": 1.0409, "lr": 6.916656859990179e-05, "epoch": 1.5805626598465472, "percentage": 31.61, "elapsed_time": "3:06:58", "remaining_time": "6:44:31"} +{"current_steps": 619, "total_steps": 1955, "loss": 1.0274, "lr": 6.911763265587338e-05, "epoch": 1.5831202046035806, "percentage": 31.66, "elapsed_time": "3:07:16", "remaining_time": "6:44:12"} +{"current_steps": 620, "total_steps": 1955, "loss": 1.015, "lr": 6.90686038314585e-05, "epoch": 1.5856777493606138, "percentage": 31.71, "elapsed_time": "3:07:35", "remaining_time": "6:43:54"} +{"current_steps": 621, "total_steps": 1955, "loss": 1.0446, "lr": 6.901948228305094e-05, "epoch": 1.5882352941176472, "percentage": 31.76, "elapsed_time": "3:07:53", "remaining_time": "6:43:36"} +{"current_steps": 622, "total_steps": 1955, "loss": 1.0583, "lr": 6.89702681673402e-05, "epoch": 1.5907928388746804, "percentage": 31.82, "elapsed_time": "3:08:11", "remaining_time": "6:43:18"} +{"current_steps": 623, "total_steps": 1955, "loss": 1.031, "lr": 6.892096164131113e-05, "epoch": 1.5933503836317136, "percentage": 31.87, "elapsed_time": "3:08:29", "remaining_time": "6:43:00"} +{"current_steps": 624, "total_steps": 1955, "loss": 1.0698, "lr": 6.887156286224329e-05, "epoch": 1.5959079283887467, "percentage": 31.92, "elapsed_time": "3:08:47", "remaining_time": "6:42:42"} +{"current_steps": 625, "total_steps": 1955, "loss": 1.0904, "lr": 6.882207198771056e-05, "epoch": 1.59846547314578, "percentage": 31.97, "elapsed_time": "3:09:05", "remaining_time": "6:42:24"} +{"current_steps": 626, "total_steps": 1955, "loss": 1.0178, "lr": 6.877248917558055e-05, "epoch": 1.6010230179028133, "percentage": 32.02, "elapsed_time": "3:09:24", "remaining_time": "6:42:07"} +{"current_steps": 627, "total_steps": 1955, "loss": 1.028, "lr": 6.872281458401414e-05, "epoch": 1.6035805626598465, "percentage": 32.07, "elapsed_time": "3:09:42", "remaining_time": "6:41:49"} +{"current_steps": 628, "total_steps": 1955, "loss": 1.0669, "lr": 6.867304837146501e-05, "epoch": 1.60613810741688, "percentage": 32.12, "elapsed_time": "3:10:00", "remaining_time": "6:41:30"} +{"current_steps": 629, "total_steps": 1955, "loss": 1.0215, "lr": 6.862319069667904e-05, "epoch": 1.608695652173913, "percentage": 32.17, "elapsed_time": "3:10:18", "remaining_time": "6:41:12"} +{"current_steps": 630, "total_steps": 1955, "loss": 1.0268, "lr": 6.857324171869393e-05, "epoch": 1.6112531969309463, "percentage": 32.23, "elapsed_time": "3:10:37", "remaining_time": "6:40:54"} +{"current_steps": 631, "total_steps": 1955, "loss": 1.0342, "lr": 6.852320159683853e-05, "epoch": 1.6138107416879794, "percentage": 32.28, "elapsed_time": "3:10:55", "remaining_time": "6:40:35"} +{"current_steps": 632, "total_steps": 1955, "loss": 1.0283, "lr": 6.84730704907325e-05, "epoch": 1.6163682864450126, "percentage": 32.33, "elapsed_time": "3:11:13", "remaining_time": "6:40:17"} +{"current_steps": 633, "total_steps": 1955, "loss": 1.0114, "lr": 6.842284856028571e-05, "epoch": 1.618925831202046, "percentage": 32.38, "elapsed_time": "3:11:31", "remaining_time": "6:39:58"} +{"current_steps": 634, "total_steps": 1955, "loss": 1.0589, "lr": 6.83725359656977e-05, "epoch": 1.6214833759590794, "percentage": 32.43, "elapsed_time": "3:11:49", "remaining_time": "6:39:40"} +{"current_steps": 635, "total_steps": 1955, "loss": 1.0119, "lr": 6.832213286745727e-05, "epoch": 1.6240409207161126, "percentage": 32.48, "elapsed_time": "3:12:07", "remaining_time": "6:39:22"} +{"current_steps": 636, "total_steps": 1955, "loss": 1.058, "lr": 6.82716394263419e-05, "epoch": 1.6265984654731458, "percentage": 32.53, "elapsed_time": "3:12:25", "remaining_time": "6:39:05"} +{"current_steps": 637, "total_steps": 1955, "loss": 1.0517, "lr": 6.82210558034172e-05, "epoch": 1.629156010230179, "percentage": 32.58, "elapsed_time": "3:12:43", "remaining_time": "6:38:46"} +{"current_steps": 638, "total_steps": 1955, "loss": 1.0361, "lr": 6.817038216003649e-05, "epoch": 1.6317135549872122, "percentage": 32.63, "elapsed_time": "3:13:02", "remaining_time": "6:38:28"} +{"current_steps": 639, "total_steps": 1955, "loss": 1.0691, "lr": 6.811961865784026e-05, "epoch": 1.6342710997442456, "percentage": 32.69, "elapsed_time": "3:13:20", "remaining_time": "6:38:09"} +{"current_steps": 640, "total_steps": 1955, "loss": 1.004, "lr": 6.806876545875558e-05, "epoch": 1.6368286445012787, "percentage": 32.74, "elapsed_time": "3:13:38", "remaining_time": "6:37:51"} +{"current_steps": 641, "total_steps": 1955, "loss": 1.0249, "lr": 6.801782272499567e-05, "epoch": 1.6393861892583121, "percentage": 32.79, "elapsed_time": "3:13:56", "remaining_time": "6:37:32"} +{"current_steps": 642, "total_steps": 1955, "loss": 1.0275, "lr": 6.796679061905936e-05, "epoch": 1.6419437340153453, "percentage": 32.84, "elapsed_time": "3:14:14", "remaining_time": "6:37:14"} +{"current_steps": 643, "total_steps": 1955, "loss": 1.0431, "lr": 6.791566930373051e-05, "epoch": 1.6445012787723785, "percentage": 32.89, "elapsed_time": "3:14:32", "remaining_time": "6:36:56"} +{"current_steps": 644, "total_steps": 1955, "loss": 1.088, "lr": 6.786445894207763e-05, "epoch": 1.6470588235294117, "percentage": 32.94, "elapsed_time": "3:14:50", "remaining_time": "6:36:37"} +{"current_steps": 645, "total_steps": 1955, "loss": 0.9978, "lr": 6.781315969745319e-05, "epoch": 1.6496163682864449, "percentage": 32.99, "elapsed_time": "3:15:08", "remaining_time": "6:36:19"} +{"current_steps": 646, "total_steps": 1955, "loss": 1.0542, "lr": 6.776177173349322e-05, "epoch": 1.6521739130434783, "percentage": 33.04, "elapsed_time": "3:15:26", "remaining_time": "6:36:00"} +{"current_steps": 647, "total_steps": 1955, "loss": 1.0127, "lr": 6.771029521411674e-05, "epoch": 1.6547314578005117, "percentage": 33.09, "elapsed_time": "3:15:44", "remaining_time": "6:35:42"} +{"current_steps": 648, "total_steps": 1955, "loss": 1.0307, "lr": 6.765873030352525e-05, "epoch": 1.6572890025575449, "percentage": 33.15, "elapsed_time": "3:16:01", "remaining_time": "6:35:23"} +{"current_steps": 649, "total_steps": 1955, "loss": 1.0364, "lr": 6.760707716620224e-05, "epoch": 1.659846547314578, "percentage": 33.2, "elapsed_time": "3:16:20", "remaining_time": "6:35:05"} +{"current_steps": 650, "total_steps": 1955, "loss": 1.0197, "lr": 6.755533596691255e-05, "epoch": 1.6624040920716112, "percentage": 33.25, "elapsed_time": "3:16:38", "remaining_time": "6:34:46"} +{"current_steps": 651, "total_steps": 1955, "loss": 1.0456, "lr": 6.7503506870702e-05, "epoch": 1.6649616368286444, "percentage": 33.3, "elapsed_time": "3:16:56", "remaining_time": "6:34:28"} +{"current_steps": 652, "total_steps": 1955, "loss": 1.0698, "lr": 6.74515900428967e-05, "epoch": 1.6675191815856778, "percentage": 33.35, "elapsed_time": "3:17:14", "remaining_time": "6:34:10"} +{"current_steps": 653, "total_steps": 1955, "loss": 1.0444, "lr": 6.739958564910274e-05, "epoch": 1.670076726342711, "percentage": 33.4, "elapsed_time": "3:17:32", "remaining_time": "6:33:52"} +{"current_steps": 654, "total_steps": 1955, "loss": 1.0494, "lr": 6.734749385520539e-05, "epoch": 1.6726342710997444, "percentage": 33.45, "elapsed_time": "3:17:50", "remaining_time": "6:33:34"} +{"current_steps": 655, "total_steps": 1955, "loss": 1.0384, "lr": 6.72953148273688e-05, "epoch": 1.6751918158567776, "percentage": 33.5, "elapsed_time": "3:18:08", "remaining_time": "6:33:15"} +{"current_steps": 656, "total_steps": 1955, "loss": 1.0252, "lr": 6.724304873203538e-05, "epoch": 1.6777493606138107, "percentage": 33.55, "elapsed_time": "3:18:26", "remaining_time": "6:32:57"} +{"current_steps": 657, "total_steps": 1955, "loss": 1.0279, "lr": 6.71906957359252e-05, "epoch": 1.680306905370844, "percentage": 33.61, "elapsed_time": "3:18:44", "remaining_time": "6:32:38"} +{"current_steps": 658, "total_steps": 1955, "loss": 1.0492, "lr": 6.71382560060356e-05, "epoch": 1.682864450127877, "percentage": 33.66, "elapsed_time": "3:19:02", "remaining_time": "6:32:20"} +{"current_steps": 659, "total_steps": 1955, "loss": 1.0496, "lr": 6.708572970964054e-05, "epoch": 1.6854219948849105, "percentage": 33.71, "elapsed_time": "3:19:20", "remaining_time": "6:32:02"} +{"current_steps": 660, "total_steps": 1955, "loss": 1.0657, "lr": 6.703311701429019e-05, "epoch": 1.6879795396419437, "percentage": 33.76, "elapsed_time": "3:19:38", "remaining_time": "6:31:43"} +{"current_steps": 661, "total_steps": 1955, "loss": 1.045, "lr": 6.698041808781021e-05, "epoch": 1.690537084398977, "percentage": 33.81, "elapsed_time": "3:19:56", "remaining_time": "6:31:25"} +{"current_steps": 662, "total_steps": 1955, "loss": 1.0372, "lr": 6.69276330983014e-05, "epoch": 1.6930946291560103, "percentage": 33.86, "elapsed_time": "3:20:14", "remaining_time": "6:31:07"} +{"current_steps": 663, "total_steps": 1955, "loss": 1.0591, "lr": 6.687476221413907e-05, "epoch": 1.6956521739130435, "percentage": 33.91, "elapsed_time": "3:20:32", "remaining_time": "6:30:48"} +{"current_steps": 664, "total_steps": 1955, "loss": 1.045, "lr": 6.682180560397248e-05, "epoch": 1.6982097186700766, "percentage": 33.96, "elapsed_time": "3:20:50", "remaining_time": "6:30:30"} +{"current_steps": 665, "total_steps": 1955, "loss": 1.0487, "lr": 6.676876343672441e-05, "epoch": 1.7007672634271098, "percentage": 34.02, "elapsed_time": "3:21:09", "remaining_time": "6:30:12"} +{"current_steps": 666, "total_steps": 1955, "loss": 1.0198, "lr": 6.67156358815905e-05, "epoch": 1.7033248081841432, "percentage": 34.07, "elapsed_time": "3:21:27", "remaining_time": "6:29:53"} +{"current_steps": 667, "total_steps": 1955, "loss": 1.0403, "lr": 6.666242310803879e-05, "epoch": 1.7058823529411766, "percentage": 34.12, "elapsed_time": "3:21:44", "remaining_time": "6:29:35"} +{"current_steps": 668, "total_steps": 1955, "loss": 1.0297, "lr": 6.660912528580912e-05, "epoch": 1.7084398976982098, "percentage": 34.17, "elapsed_time": "3:22:03", "remaining_time": "6:29:18"} +{"current_steps": 669, "total_steps": 1955, "loss": 1.0554, "lr": 6.655574258491267e-05, "epoch": 1.710997442455243, "percentage": 34.22, "elapsed_time": "3:22:21", "remaining_time": "6:29:00"} +{"current_steps": 670, "total_steps": 1955, "loss": 1.0345, "lr": 6.650227517563134e-05, "epoch": 1.7135549872122762, "percentage": 34.27, "elapsed_time": "3:22:39", "remaining_time": "6:28:41"} +{"current_steps": 671, "total_steps": 1955, "loss": 1.0512, "lr": 6.644872322851722e-05, "epoch": 1.7161125319693094, "percentage": 34.32, "elapsed_time": "3:22:58", "remaining_time": "6:28:25"} +{"current_steps": 672, "total_steps": 1955, "loss": 1.0531, "lr": 6.63950869143921e-05, "epoch": 1.7186700767263428, "percentage": 34.37, "elapsed_time": "3:23:16", "remaining_time": "6:28:06"} +{"current_steps": 673, "total_steps": 1955, "loss": 1.0305, "lr": 6.634136640434685e-05, "epoch": 1.721227621483376, "percentage": 34.42, "elapsed_time": "3:23:34", "remaining_time": "6:27:48"} +{"current_steps": 674, "total_steps": 1955, "loss": 1.0734, "lr": 6.628756186974093e-05, "epoch": 1.7237851662404093, "percentage": 34.48, "elapsed_time": "3:23:53", "remaining_time": "6:27:30"} +{"current_steps": 675, "total_steps": 1955, "loss": 1.044, "lr": 6.623367348220184e-05, "epoch": 1.7263427109974425, "percentage": 34.53, "elapsed_time": "3:24:11", "remaining_time": "6:27:12"} +{"current_steps": 676, "total_steps": 1955, "loss": 1.01, "lr": 6.61797014136245e-05, "epoch": 1.7289002557544757, "percentage": 34.58, "elapsed_time": "3:24:29", "remaining_time": "6:26:54"} +{"current_steps": 677, "total_steps": 1955, "loss": 1.0271, "lr": 6.612564583617085e-05, "epoch": 1.7314578005115089, "percentage": 34.63, "elapsed_time": "3:24:47", "remaining_time": "6:26:35"} +{"current_steps": 678, "total_steps": 1955, "loss": 1.0289, "lr": 6.607150692226912e-05, "epoch": 1.734015345268542, "percentage": 34.68, "elapsed_time": "3:25:05", "remaining_time": "6:26:17"} +{"current_steps": 679, "total_steps": 1955, "loss": 1.0301, "lr": 6.601728484461342e-05, "epoch": 1.7365728900255755, "percentage": 34.73, "elapsed_time": "3:25:23", "remaining_time": "6:25:58"} +{"current_steps": 680, "total_steps": 1955, "loss": 1.0309, "lr": 6.59629797761631e-05, "epoch": 1.7391304347826086, "percentage": 34.78, "elapsed_time": "3:25:42", "remaining_time": "6:25:42"} +{"current_steps": 681, "total_steps": 1955, "loss": 1.0192, "lr": 6.59085918901423e-05, "epoch": 1.741687979539642, "percentage": 34.83, "elapsed_time": "3:26:00", "remaining_time": "6:25:23"} +{"current_steps": 682, "total_steps": 1955, "loss": 1.0649, "lr": 6.585412136003927e-05, "epoch": 1.7442455242966752, "percentage": 34.88, "elapsed_time": "3:26:18", "remaining_time": "6:25:05"} +{"current_steps": 683, "total_steps": 1955, "loss": 1.0573, "lr": 6.579956835960592e-05, "epoch": 1.7468030690537084, "percentage": 34.94, "elapsed_time": "3:26:36", "remaining_time": "6:24:47"} +{"current_steps": 684, "total_steps": 1955, "loss": 1.0425, "lr": 6.574493306285719e-05, "epoch": 1.7493606138107416, "percentage": 34.99, "elapsed_time": "3:26:54", "remaining_time": "6:24:28"} +{"current_steps": 685, "total_steps": 1955, "loss": 1.0349, "lr": 6.569021564407057e-05, "epoch": 1.7519181585677748, "percentage": 35.04, "elapsed_time": "3:27:12", "remaining_time": "6:24:10"} +{"current_steps": 686, "total_steps": 1955, "loss": 1.0648, "lr": 6.56354162777855e-05, "epoch": 1.7544757033248082, "percentage": 35.09, "elapsed_time": "3:27:30", "remaining_time": "6:23:52"} +{"current_steps": 687, "total_steps": 1955, "loss": 1.0399, "lr": 6.558053513880278e-05, "epoch": 1.7570332480818416, "percentage": 35.14, "elapsed_time": "3:27:48", "remaining_time": "6:23:33"} +{"current_steps": 688, "total_steps": 1955, "loss": 1.0274, "lr": 6.552557240218412e-05, "epoch": 1.7595907928388748, "percentage": 35.19, "elapsed_time": "3:28:06", "remaining_time": "6:23:15"} +{"current_steps": 689, "total_steps": 1955, "loss": 1.0105, "lr": 6.547052824325142e-05, "epoch": 1.762148337595908, "percentage": 35.24, "elapsed_time": "3:28:24", "remaining_time": "6:22:56"} +{"current_steps": 690, "total_steps": 1955, "loss": 1.0363, "lr": 6.541540283758642e-05, "epoch": 1.7647058823529411, "percentage": 35.29, "elapsed_time": "3:28:42", "remaining_time": "6:22:38"} +{"current_steps": 691, "total_steps": 1955, "loss": 1.0469, "lr": 6.53601963610299e-05, "epoch": 1.7672634271099743, "percentage": 35.35, "elapsed_time": "3:29:00", "remaining_time": "6:22:20"} +{"current_steps": 692, "total_steps": 1955, "loss": 1.0405, "lr": 6.530490898968135e-05, "epoch": 1.7698209718670077, "percentage": 35.4, "elapsed_time": "3:29:18", "remaining_time": "6:22:01"} +{"current_steps": 693, "total_steps": 1955, "loss": 1.0554, "lr": 6.524954089989822e-05, "epoch": 1.772378516624041, "percentage": 35.45, "elapsed_time": "3:29:36", "remaining_time": "6:21:43"} +{"current_steps": 694, "total_steps": 1955, "loss": 1.0071, "lr": 6.519409226829549e-05, "epoch": 1.7749360613810743, "percentage": 35.5, "elapsed_time": "3:29:54", "remaining_time": "6:21:25"} +{"current_steps": 695, "total_steps": 1955, "loss": 1.0219, "lr": 6.513856327174501e-05, "epoch": 1.7774936061381075, "percentage": 35.55, "elapsed_time": "3:30:12", "remaining_time": "6:21:06"} +{"current_steps": 696, "total_steps": 1955, "loss": 1.0553, "lr": 6.508295408737505e-05, "epoch": 1.7800511508951407, "percentage": 35.6, "elapsed_time": "3:30:30", "remaining_time": "6:20:48"} +{"current_steps": 697, "total_steps": 1955, "loss": 1.0331, "lr": 6.50272648925696e-05, "epoch": 1.7826086956521738, "percentage": 35.65, "elapsed_time": "3:30:48", "remaining_time": "6:20:29"} +{"current_steps": 698, "total_steps": 1955, "loss": 1.0474, "lr": 6.49714958649679e-05, "epoch": 1.785166240409207, "percentage": 35.7, "elapsed_time": "3:31:06", "remaining_time": "6:20:11"} +{"current_steps": 699, "total_steps": 1955, "loss": 1.0372, "lr": 6.491564718246381e-05, "epoch": 1.7877237851662404, "percentage": 35.75, "elapsed_time": "3:31:24", "remaining_time": "6:19:52"} +{"current_steps": 700, "total_steps": 1955, "loss": 1.0545, "lr": 6.485971902320534e-05, "epoch": 1.7902813299232738, "percentage": 35.81, "elapsed_time": "3:31:42", "remaining_time": "6:19:34"} +{"current_steps": 701, "total_steps": 1955, "loss": 1.0207, "lr": 6.480371156559395e-05, "epoch": 1.792838874680307, "percentage": 35.86, "elapsed_time": "3:32:00", "remaining_time": "6:19:15"} +{"current_steps": 702, "total_steps": 1955, "loss": 1.0395, "lr": 6.47476249882841e-05, "epoch": 1.7953964194373402, "percentage": 35.91, "elapsed_time": "3:32:18", "remaining_time": "6:18:57"} +{"current_steps": 703, "total_steps": 1955, "loss": 1.0423, "lr": 6.469145947018261e-05, "epoch": 1.7979539641943734, "percentage": 35.96, "elapsed_time": "3:32:36", "remaining_time": "6:18:39"} +{"current_steps": 704, "total_steps": 1955, "loss": 1.0228, "lr": 6.463521519044808e-05, "epoch": 1.8005115089514065, "percentage": 36.01, "elapsed_time": "3:32:54", "remaining_time": "6:18:20"} +{"current_steps": 705, "total_steps": 1955, "loss": 1.0534, "lr": 6.457889232849039e-05, "epoch": 1.80306905370844, "percentage": 36.06, "elapsed_time": "3:33:13", "remaining_time": "6:18:03"} +{"current_steps": 706, "total_steps": 1955, "loss": 1.0043, "lr": 6.452249106397004e-05, "epoch": 1.8056265984654731, "percentage": 36.11, "elapsed_time": "3:33:31", "remaining_time": "6:17:44"} +{"current_steps": 707, "total_steps": 1955, "loss": 1.0086, "lr": 6.446601157679768e-05, "epoch": 1.8081841432225065, "percentage": 36.16, "elapsed_time": "3:33:49", "remaining_time": "6:17:26"} +{"current_steps": 708, "total_steps": 1955, "loss": 1.0346, "lr": 6.440945404713344e-05, "epoch": 1.8107416879795397, "percentage": 36.21, "elapsed_time": "3:34:07", "remaining_time": "6:17:08"} +{"current_steps": 709, "total_steps": 1955, "loss": 0.9851, "lr": 6.435281865538638e-05, "epoch": 1.813299232736573, "percentage": 36.27, "elapsed_time": "3:34:25", "remaining_time": "6:16:49"} +{"current_steps": 710, "total_steps": 1955, "loss": 1.0297, "lr": 6.429610558221395e-05, "epoch": 1.815856777493606, "percentage": 36.32, "elapsed_time": "3:34:43", "remaining_time": "6:16:31"} +{"current_steps": 711, "total_steps": 1955, "loss": 1.0827, "lr": 6.423931500852139e-05, "epoch": 1.8184143222506393, "percentage": 36.37, "elapsed_time": "3:35:01", "remaining_time": "6:16:13"} +{"current_steps": 712, "total_steps": 1955, "loss": 1.0311, "lr": 6.418244711546113e-05, "epoch": 1.8209718670076727, "percentage": 36.42, "elapsed_time": "3:35:19", "remaining_time": "6:15:54"} +{"current_steps": 713, "total_steps": 1955, "loss": 1.0402, "lr": 6.412550208443227e-05, "epoch": 1.8235294117647058, "percentage": 36.47, "elapsed_time": "3:35:37", "remaining_time": "6:15:36"} +{"current_steps": 714, "total_steps": 1955, "loss": 1.0283, "lr": 6.406848009707995e-05, "epoch": 1.8260869565217392, "percentage": 36.52, "elapsed_time": "3:35:55", "remaining_time": "6:15:17"} +{"current_steps": 715, "total_steps": 1955, "loss": 1.0125, "lr": 6.401138133529478e-05, "epoch": 1.8286445012787724, "percentage": 36.57, "elapsed_time": "3:36:13", "remaining_time": "6:14:59"} +{"current_steps": 716, "total_steps": 1955, "loss": 1.0293, "lr": 6.395420598121229e-05, "epoch": 1.8312020460358056, "percentage": 36.62, "elapsed_time": "3:36:31", "remaining_time": "6:14:41"} +{"current_steps": 717, "total_steps": 1955, "loss": 0.9997, "lr": 6.389695421721231e-05, "epoch": 1.8337595907928388, "percentage": 36.68, "elapsed_time": "3:36:49", "remaining_time": "6:14:22"} +{"current_steps": 718, "total_steps": 1955, "loss": 1.0498, "lr": 6.383962622591838e-05, "epoch": 1.836317135549872, "percentage": 36.73, "elapsed_time": "3:37:07", "remaining_time": "6:14:04"} +{"current_steps": 719, "total_steps": 1955, "loss": 1.0491, "lr": 6.378222219019726e-05, "epoch": 1.8388746803069054, "percentage": 36.78, "elapsed_time": "3:37:25", "remaining_time": "6:13:45"} +{"current_steps": 720, "total_steps": 1955, "loss": 1.0556, "lr": 6.372474229315823e-05, "epoch": 1.8414322250639388, "percentage": 36.83, "elapsed_time": "3:37:43", "remaining_time": "6:13:27"} +{"current_steps": 721, "total_steps": 1955, "loss": 1.0566, "lr": 6.366718671815255e-05, "epoch": 1.843989769820972, "percentage": 36.88, "elapsed_time": "3:38:01", "remaining_time": "6:13:09"} +{"current_steps": 722, "total_steps": 1955, "loss": 1.0128, "lr": 6.360955564877289e-05, "epoch": 1.8465473145780051, "percentage": 36.93, "elapsed_time": "3:38:19", "remaining_time": "6:12:50"} +{"current_steps": 723, "total_steps": 1955, "loss": 1.0398, "lr": 6.355184926885275e-05, "epoch": 1.8491048593350383, "percentage": 36.98, "elapsed_time": "3:38:37", "remaining_time": "6:12:32"} +{"current_steps": 724, "total_steps": 1955, "loss": 1.0303, "lr": 6.349406776246587e-05, "epoch": 1.8516624040920715, "percentage": 37.03, "elapsed_time": "3:38:55", "remaining_time": "6:12:14"} +{"current_steps": 725, "total_steps": 1955, "loss": 1.0613, "lr": 6.343621131392558e-05, "epoch": 1.854219948849105, "percentage": 37.08, "elapsed_time": "3:39:14", "remaining_time": "6:11:57"} +{"current_steps": 726, "total_steps": 1955, "loss": 0.9846, "lr": 6.337828010778427e-05, "epoch": 1.856777493606138, "percentage": 37.14, "elapsed_time": "3:39:32", "remaining_time": "6:11:38"} +{"current_steps": 727, "total_steps": 1955, "loss": 1.01, "lr": 6.332027432883287e-05, "epoch": 1.8593350383631715, "percentage": 37.19, "elapsed_time": "3:39:49", "remaining_time": "6:11:19"} +{"current_steps": 728, "total_steps": 1955, "loss": 1.015, "lr": 6.326219416210009e-05, "epoch": 1.8618925831202047, "percentage": 37.24, "elapsed_time": "3:40:07", "remaining_time": "6:11:00"} +{"current_steps": 729, "total_steps": 1955, "loss": 1.0721, "lr": 6.320403979285197e-05, "epoch": 1.8644501278772379, "percentage": 37.29, "elapsed_time": "3:40:25", "remaining_time": "6:10:42"} +{"current_steps": 730, "total_steps": 1955, "loss": 1.027, "lr": 6.314581140659124e-05, "epoch": 1.867007672634271, "percentage": 37.34, "elapsed_time": "3:40:43", "remaining_time": "6:10:24"} +{"current_steps": 731, "total_steps": 1955, "loss": 0.9984, "lr": 6.308750918905674e-05, "epoch": 1.8695652173913042, "percentage": 37.39, "elapsed_time": "3:41:01", "remaining_time": "6:10:05"} +{"current_steps": 732, "total_steps": 1955, "loss": 1.0354, "lr": 6.302913332622279e-05, "epoch": 1.8721227621483376, "percentage": 37.44, "elapsed_time": "3:41:19", "remaining_time": "6:09:47"} +{"current_steps": 733, "total_steps": 1955, "loss": 1.0471, "lr": 6.297068400429866e-05, "epoch": 1.8746803069053708, "percentage": 37.49, "elapsed_time": "3:41:37", "remaining_time": "6:09:28"} +{"current_steps": 734, "total_steps": 1955, "loss": 1.0069, "lr": 6.291216140972793e-05, "epoch": 1.8772378516624042, "percentage": 37.54, "elapsed_time": "3:41:55", "remaining_time": "6:09:10"} +{"current_steps": 735, "total_steps": 1955, "loss": 1.0364, "lr": 6.28535657291879e-05, "epoch": 1.8797953964194374, "percentage": 37.6, "elapsed_time": "3:42:13", "remaining_time": "6:08:52"} +{"current_steps": 736, "total_steps": 1955, "loss": 1.0685, "lr": 6.279489714958902e-05, "epoch": 1.8823529411764706, "percentage": 37.65, "elapsed_time": "3:42:32", "remaining_time": "6:08:35"} +{"current_steps": 737, "total_steps": 1955, "loss": 1.0413, "lr": 6.273615585807423e-05, "epoch": 1.8849104859335037, "percentage": 37.7, "elapsed_time": "3:42:50", "remaining_time": "6:08:16"} +{"current_steps": 738, "total_steps": 1955, "loss": 1.0468, "lr": 6.267734204201848e-05, "epoch": 1.887468030690537, "percentage": 37.75, "elapsed_time": "3:43:08", "remaining_time": "6:07:58"} +{"current_steps": 739, "total_steps": 1955, "loss": 1.0311, "lr": 6.261845588902799e-05, "epoch": 1.8900255754475703, "percentage": 37.8, "elapsed_time": "3:43:26", "remaining_time": "6:07:40"} +{"current_steps": 740, "total_steps": 1955, "loss": 1.0032, "lr": 6.255949758693977e-05, "epoch": 1.8925831202046037, "percentage": 37.85, "elapsed_time": "3:43:44", "remaining_time": "6:07:21"} +{"current_steps": 741, "total_steps": 1955, "loss": 1.0131, "lr": 6.250046732382099e-05, "epoch": 1.895140664961637, "percentage": 37.9, "elapsed_time": "3:44:02", "remaining_time": "6:07:03"} +{"current_steps": 742, "total_steps": 1955, "loss": 1.0191, "lr": 6.244136528796827e-05, "epoch": 1.89769820971867, "percentage": 37.95, "elapsed_time": "3:44:20", "remaining_time": "6:06:45"} +{"current_steps": 743, "total_steps": 1955, "loss": 1.0563, "lr": 6.238219166790728e-05, "epoch": 1.9002557544757033, "percentage": 38.01, "elapsed_time": "3:44:38", "remaining_time": "6:06:26"} +{"current_steps": 744, "total_steps": 1955, "loss": 1.012, "lr": 6.232294665239199e-05, "epoch": 1.9028132992327365, "percentage": 38.06, "elapsed_time": "3:44:56", "remaining_time": "6:06:08"} +{"current_steps": 745, "total_steps": 1955, "loss": 1.0094, "lr": 6.22636304304041e-05, "epoch": 1.9053708439897699, "percentage": 38.11, "elapsed_time": "3:45:14", "remaining_time": "6:05:50"} +{"current_steps": 746, "total_steps": 1955, "loss": 1.0333, "lr": 6.220424319115243e-05, "epoch": 1.907928388746803, "percentage": 38.16, "elapsed_time": "3:45:32", "remaining_time": "6:05:32"} +{"current_steps": 747, "total_steps": 1955, "loss": 1.025, "lr": 6.214478512407238e-05, "epoch": 1.9104859335038364, "percentage": 38.21, "elapsed_time": "3:45:50", "remaining_time": "6:05:13"} +{"current_steps": 748, "total_steps": 1955, "loss": 1.0401, "lr": 6.208525641882527e-05, "epoch": 1.9130434782608696, "percentage": 38.26, "elapsed_time": "3:46:08", "remaining_time": "6:04:55"} +{"current_steps": 749, "total_steps": 1955, "loss": 1.0626, "lr": 6.20256572652977e-05, "epoch": 1.9156010230179028, "percentage": 38.31, "elapsed_time": "3:46:26", "remaining_time": "6:04:37"} +{"current_steps": 750, "total_steps": 1955, "loss": 1.0152, "lr": 6.196598785360104e-05, "epoch": 1.918158567774936, "percentage": 38.36, "elapsed_time": "3:46:44", "remaining_time": "6:04:18"} +{"current_steps": 751, "total_steps": 1955, "loss": 1.0711, "lr": 6.190624837407075e-05, "epoch": 1.9207161125319692, "percentage": 38.41, "elapsed_time": "3:47:02", "remaining_time": "6:03:59"} +{"current_steps": 752, "total_steps": 1955, "loss": 1.0184, "lr": 6.184643901726578e-05, "epoch": 1.9232736572890026, "percentage": 38.47, "elapsed_time": "3:47:20", "remaining_time": "6:03:41"} +{"current_steps": 753, "total_steps": 1955, "loss": 1.0341, "lr": 6.178655997396801e-05, "epoch": 1.9258312020460358, "percentage": 38.52, "elapsed_time": "3:47:38", "remaining_time": "6:03:23"} +{"current_steps": 754, "total_steps": 1955, "loss": 1.0288, "lr": 6.172661143518158e-05, "epoch": 1.9283887468030692, "percentage": 38.57, "elapsed_time": "3:47:56", "remaining_time": "6:03:04"} +{"current_steps": 755, "total_steps": 1955, "loss": 0.9989, "lr": 6.16665935921323e-05, "epoch": 1.9309462915601023, "percentage": 38.62, "elapsed_time": "3:48:15", "remaining_time": "6:02:46"} +{"current_steps": 756, "total_steps": 1955, "loss": 1.0358, "lr": 6.16065066362671e-05, "epoch": 1.9335038363171355, "percentage": 38.67, "elapsed_time": "3:48:33", "remaining_time": "6:02:28"} +{"current_steps": 757, "total_steps": 1955, "loss": 1.0328, "lr": 6.154635075925333e-05, "epoch": 1.9360613810741687, "percentage": 38.72, "elapsed_time": "3:48:51", "remaining_time": "6:02:10"} +{"current_steps": 758, "total_steps": 1955, "loss": 1.0143, "lr": 6.148612615297819e-05, "epoch": 1.938618925831202, "percentage": 38.77, "elapsed_time": "3:49:09", "remaining_time": "6:01:52"} +{"current_steps": 759, "total_steps": 1955, "loss": 1.0603, "lr": 6.14258330095481e-05, "epoch": 1.9411764705882353, "percentage": 38.82, "elapsed_time": "3:49:27", "remaining_time": "6:01:33"} +{"current_steps": 760, "total_steps": 1955, "loss": 1.0642, "lr": 6.136547152128817e-05, "epoch": 1.9437340153452687, "percentage": 38.87, "elapsed_time": "3:49:45", "remaining_time": "6:01:15"} +{"current_steps": 761, "total_steps": 1955, "loss": 1.0637, "lr": 6.130504188074141e-05, "epoch": 1.9462915601023019, "percentage": 38.93, "elapsed_time": "3:50:03", "remaining_time": "6:00:57"} +{"current_steps": 762, "total_steps": 1955, "loss": 1.0411, "lr": 6.12445442806683e-05, "epoch": 1.948849104859335, "percentage": 38.98, "elapsed_time": "3:50:21", "remaining_time": "6:00:39"} +{"current_steps": 763, "total_steps": 1955, "loss": 1.0311, "lr": 6.118397891404612e-05, "epoch": 1.9514066496163682, "percentage": 39.03, "elapsed_time": "3:50:40", "remaining_time": "6:00:21"} +{"current_steps": 764, "total_steps": 1955, "loss": 1.0455, "lr": 6.112334597406823e-05, "epoch": 1.9539641943734014, "percentage": 39.08, "elapsed_time": "3:50:58", "remaining_time": "6:00:03"} +{"current_steps": 765, "total_steps": 1955, "loss": 0.978, "lr": 6.106264565414363e-05, "epoch": 1.9565217391304348, "percentage": 39.13, "elapsed_time": "3:51:16", "remaining_time": "5:59:45"} +{"current_steps": 766, "total_steps": 1955, "loss": 1.0441, "lr": 6.100187814789617e-05, "epoch": 1.959079283887468, "percentage": 39.18, "elapsed_time": "3:51:34", "remaining_time": "5:59:26"} +{"current_steps": 767, "total_steps": 1955, "loss": 1.0126, "lr": 6.094104364916405e-05, "epoch": 1.9616368286445014, "percentage": 39.23, "elapsed_time": "3:51:52", "remaining_time": "5:59:08"} +{"current_steps": 768, "total_steps": 1955, "loss": 1.0374, "lr": 6.088014235199918e-05, "epoch": 1.9641943734015346, "percentage": 39.28, "elapsed_time": "3:52:10", "remaining_time": "5:58:50"} +{"current_steps": 769, "total_steps": 1955, "loss": 1.0469, "lr": 6.081917445066651e-05, "epoch": 1.9667519181585678, "percentage": 39.34, "elapsed_time": "3:52:28", "remaining_time": "5:58:31"} +{"current_steps": 770, "total_steps": 1955, "loss": 1.0298, "lr": 6.07581401396435e-05, "epoch": 1.969309462915601, "percentage": 39.39, "elapsed_time": "3:52:46", "remaining_time": "5:58:13"} +{"current_steps": 771, "total_steps": 1955, "loss": 1.0209, "lr": 6.0697039613619376e-05, "epoch": 1.9718670076726341, "percentage": 39.44, "elapsed_time": "3:53:04", "remaining_time": "5:57:54"} +{"current_steps": 772, "total_steps": 1955, "loss": 1.0364, "lr": 6.063587306749463e-05, "epoch": 1.9744245524296675, "percentage": 39.49, "elapsed_time": "3:53:22", "remaining_time": "5:57:36"} +{"current_steps": 773, "total_steps": 1955, "loss": 1.0454, "lr": 6.057464069638032e-05, "epoch": 1.976982097186701, "percentage": 39.54, "elapsed_time": "3:53:41", "remaining_time": "5:57:19"} +{"current_steps": 774, "total_steps": 1955, "loss": 1.0121, "lr": 6.051334269559751e-05, "epoch": 1.979539641943734, "percentage": 39.59, "elapsed_time": "3:53:59", "remaining_time": "5:57:01"} +{"current_steps": 775, "total_steps": 1955, "loss": 1.0229, "lr": 6.045197926067657e-05, "epoch": 1.9820971867007673, "percentage": 39.64, "elapsed_time": "3:54:17", "remaining_time": "5:56:42"} +{"current_steps": 776, "total_steps": 1955, "loss": 1.04, "lr": 6.03905505873566e-05, "epoch": 1.9846547314578005, "percentage": 39.69, "elapsed_time": "3:54:35", "remaining_time": "5:56:24"} +{"current_steps": 777, "total_steps": 1955, "loss": 1.0201, "lr": 6.032905687158483e-05, "epoch": 1.9872122762148337, "percentage": 39.74, "elapsed_time": "3:54:53", "remaining_time": "5:56:06"} +{"current_steps": 778, "total_steps": 1955, "loss": 1.0604, "lr": 6.026749830951595e-05, "epoch": 1.989769820971867, "percentage": 39.8, "elapsed_time": "3:55:11", "remaining_time": "5:55:47"} +{"current_steps": 779, "total_steps": 1955, "loss": 1.042, "lr": 6.0205875097511474e-05, "epoch": 1.9923273657289002, "percentage": 39.85, "elapsed_time": "3:55:29", "remaining_time": "5:55:29"} +{"current_steps": 780, "total_steps": 1955, "loss": 1.024, "lr": 6.0144187432139185e-05, "epoch": 1.9948849104859336, "percentage": 39.9, "elapsed_time": "3:55:47", "remaining_time": "5:55:11"} +{"current_steps": 781, "total_steps": 1955, "loss": 1.0303, "lr": 6.008243551017241e-05, "epoch": 1.9974424552429668, "percentage": 39.95, "elapsed_time": "3:56:05", "remaining_time": "5:54:53"} +{"current_steps": 782, "total_steps": 1955, "loss": 1.0061, "lr": 6.002061952858949e-05, "epoch": 2.0, "percentage": 40.0, "elapsed_time": "3:56:23", "remaining_time": "5:54:35"} +{"current_steps": 783, "total_steps": 1955, "loss": 1.0042, "lr": 5.995873968457307e-05, "epoch": 2.002557544757033, "percentage": 40.05, "elapsed_time": "3:56:53", "remaining_time": "5:54:35"} +{"current_steps": 784, "total_steps": 1955, "loss": 0.9913, "lr": 5.989679617550953e-05, "epoch": 2.0051150895140664, "percentage": 40.1, "elapsed_time": "3:57:11", "remaining_time": "5:54:16"} +{"current_steps": 785, "total_steps": 1955, "loss": 0.9942, "lr": 5.983478919898832e-05, "epoch": 2.0076726342710995, "percentage": 40.15, "elapsed_time": "3:57:29", "remaining_time": "5:53:58"} +{"current_steps": 786, "total_steps": 1955, "loss": 1.077, "lr": 5.9772718952801335e-05, "epoch": 2.010230179028133, "percentage": 40.2, "elapsed_time": "3:57:47", "remaining_time": "5:53:40"} +{"current_steps": 787, "total_steps": 1955, "loss": 1.0141, "lr": 5.9710585634942304e-05, "epoch": 2.0127877237851663, "percentage": 40.26, "elapsed_time": "3:58:05", "remaining_time": "5:53:22"} +{"current_steps": 788, "total_steps": 1955, "loss": 1.024, "lr": 5.9648389443606136e-05, "epoch": 2.0153452685421995, "percentage": 40.31, "elapsed_time": "3:58:23", "remaining_time": "5:53:03"} +{"current_steps": 789, "total_steps": 1955, "loss": 1.008, "lr": 5.958613057718828e-05, "epoch": 2.0179028132992327, "percentage": 40.36, "elapsed_time": "3:58:41", "remaining_time": "5:52:44"} +{"current_steps": 790, "total_steps": 1955, "loss": 0.9975, "lr": 5.9523809234284155e-05, "epoch": 2.020460358056266, "percentage": 40.41, "elapsed_time": "3:59:00", "remaining_time": "5:52:27"} +{"current_steps": 791, "total_steps": 1955, "loss": 0.995, "lr": 5.9461425613688416e-05, "epoch": 2.023017902813299, "percentage": 40.46, "elapsed_time": "3:59:18", "remaining_time": "5:52:09"} +{"current_steps": 792, "total_steps": 1955, "loss": 0.9994, "lr": 5.9398979914394416e-05, "epoch": 2.0255754475703327, "percentage": 40.51, "elapsed_time": "3:59:36", "remaining_time": "5:51:50"} +{"current_steps": 793, "total_steps": 1955, "loss": 0.9927, "lr": 5.933647233559349e-05, "epoch": 2.028132992327366, "percentage": 40.56, "elapsed_time": "3:59:54", "remaining_time": "5:51:32"} +{"current_steps": 794, "total_steps": 1955, "loss": 0.9919, "lr": 5.9273903076674374e-05, "epoch": 2.030690537084399, "percentage": 40.61, "elapsed_time": "4:00:12", "remaining_time": "5:51:14"} +{"current_steps": 795, "total_steps": 1955, "loss": 0.9973, "lr": 5.921127233722258e-05, "epoch": 2.0332480818414322, "percentage": 40.66, "elapsed_time": "4:00:30", "remaining_time": "5:50:56"} +{"current_steps": 796, "total_steps": 1955, "loss": 1.0086, "lr": 5.91485803170197e-05, "epoch": 2.0358056265984654, "percentage": 40.72, "elapsed_time": "4:00:48", "remaining_time": "5:50:37"} +{"current_steps": 797, "total_steps": 1955, "loss": 1.0301, "lr": 5.908582721604281e-05, "epoch": 2.0383631713554986, "percentage": 40.77, "elapsed_time": "4:01:06", "remaining_time": "5:50:19"} +{"current_steps": 798, "total_steps": 1955, "loss": 1.0039, "lr": 5.902301323446381e-05, "epoch": 2.040920716112532, "percentage": 40.82, "elapsed_time": "4:01:24", "remaining_time": "5:50:01"} +{"current_steps": 799, "total_steps": 1955, "loss": 1.0043, "lr": 5.896013857264884e-05, "epoch": 2.0434782608695654, "percentage": 40.87, "elapsed_time": "4:01:42", "remaining_time": "5:49:42"} +{"current_steps": 800, "total_steps": 1955, "loss": 1.0487, "lr": 5.889720343115755e-05, "epoch": 2.0460358056265986, "percentage": 40.92, "elapsed_time": "4:02:00", "remaining_time": "5:49:24"} +{"current_steps": 801, "total_steps": 1955, "loss": 1.0009, "lr": 5.8834208010742534e-05, "epoch": 2.0485933503836318, "percentage": 40.97, "elapsed_time": "4:02:18", "remaining_time": "5:49:06"} +{"current_steps": 802, "total_steps": 1955, "loss": 1.0075, "lr": 5.8771152512348694e-05, "epoch": 2.051150895140665, "percentage": 41.02, "elapsed_time": "4:02:36", "remaining_time": "5:48:47"} +{"current_steps": 803, "total_steps": 1955, "loss": 1.0153, "lr": 5.870803713711251e-05, "epoch": 2.053708439897698, "percentage": 41.07, "elapsed_time": "4:02:54", "remaining_time": "5:48:29"} +{"current_steps": 804, "total_steps": 1955, "loss": 1.0436, "lr": 5.86448620863615e-05, "epoch": 2.0562659846547313, "percentage": 41.13, "elapsed_time": "4:03:12", "remaining_time": "5:48:11"} +{"current_steps": 805, "total_steps": 1955, "loss": 0.9909, "lr": 5.8581627561613536e-05, "epoch": 2.0588235294117645, "percentage": 41.18, "elapsed_time": "4:03:30", "remaining_time": "5:47:52"} +{"current_steps": 806, "total_steps": 1955, "loss": 1.0042, "lr": 5.8518333764576176e-05, "epoch": 2.061381074168798, "percentage": 41.23, "elapsed_time": "4:03:49", "remaining_time": "5:47:34"} +{"current_steps": 807, "total_steps": 1955, "loss": 1.0454, "lr": 5.8454980897146094e-05, "epoch": 2.0639386189258313, "percentage": 41.28, "elapsed_time": "4:04:07", "remaining_time": "5:47:16"} +{"current_steps": 808, "total_steps": 1955, "loss": 1.0176, "lr": 5.839156916140832e-05, "epoch": 2.0664961636828645, "percentage": 41.33, "elapsed_time": "4:04:25", "remaining_time": "5:46:57"} +{"current_steps": 809, "total_steps": 1955, "loss": 0.9945, "lr": 5.832809875963573e-05, "epoch": 2.0690537084398977, "percentage": 41.38, "elapsed_time": "4:04:42", "remaining_time": "5:46:39"} +{"current_steps": 810, "total_steps": 1955, "loss": 0.9978, "lr": 5.826456989428832e-05, "epoch": 2.071611253196931, "percentage": 41.43, "elapsed_time": "4:05:00", "remaining_time": "5:46:20"} +{"current_steps": 811, "total_steps": 1955, "loss": 1.0212, "lr": 5.820098276801252e-05, "epoch": 2.074168797953964, "percentage": 41.48, "elapsed_time": "4:05:18", "remaining_time": "5:46:02"} +{"current_steps": 812, "total_steps": 1955, "loss": 0.9955, "lr": 5.813733758364069e-05, "epoch": 2.0767263427109977, "percentage": 41.53, "elapsed_time": "4:05:37", "remaining_time": "5:45:44"} +{"current_steps": 813, "total_steps": 1955, "loss": 0.9961, "lr": 5.8073634544190315e-05, "epoch": 2.079283887468031, "percentage": 41.59, "elapsed_time": "4:05:55", "remaining_time": "5:45:26"} +{"current_steps": 814, "total_steps": 1955, "loss": 1.0569, "lr": 5.800987385286346e-05, "epoch": 2.081841432225064, "percentage": 41.64, "elapsed_time": "4:06:13", "remaining_time": "5:45:07"} +{"current_steps": 815, "total_steps": 1955, "loss": 0.9854, "lr": 5.79460557130461e-05, "epoch": 2.084398976982097, "percentage": 41.69, "elapsed_time": "4:06:31", "remaining_time": "5:44:49"} +{"current_steps": 816, "total_steps": 1955, "loss": 0.9973, "lr": 5.788218032830742e-05, "epoch": 2.0869565217391304, "percentage": 41.74, "elapsed_time": "4:06:49", "remaining_time": "5:44:31"} +{"current_steps": 817, "total_steps": 1955, "loss": 1.0084, "lr": 5.781824790239924e-05, "epoch": 2.0895140664961636, "percentage": 41.79, "elapsed_time": "4:07:07", "remaining_time": "5:44:12"} +{"current_steps": 818, "total_steps": 1955, "loss": 0.9879, "lr": 5.7754258639255334e-05, "epoch": 2.0920716112531967, "percentage": 41.84, "elapsed_time": "4:07:25", "remaining_time": "5:43:54"} +{"current_steps": 819, "total_steps": 1955, "loss": 0.9916, "lr": 5.769021274299077e-05, "epoch": 2.0946291560102304, "percentage": 41.89, "elapsed_time": "4:07:42", "remaining_time": "5:43:35"} +{"current_steps": 820, "total_steps": 1955, "loss": 0.9926, "lr": 5.7626110417901236e-05, "epoch": 2.0971867007672635, "percentage": 41.94, "elapsed_time": "4:08:00", "remaining_time": "5:43:17"} +{"current_steps": 821, "total_steps": 1955, "loss": 1.0049, "lr": 5.7561951868462476e-05, "epoch": 2.0997442455242967, "percentage": 41.99, "elapsed_time": "4:08:19", "remaining_time": "5:43:00"} +{"current_steps": 822, "total_steps": 1955, "loss": 0.99, "lr": 5.749773729932954e-05, "epoch": 2.10230179028133, "percentage": 42.05, "elapsed_time": "4:08:37", "remaining_time": "5:42:41"} +{"current_steps": 823, "total_steps": 1955, "loss": 1.0209, "lr": 5.743346691533617e-05, "epoch": 2.104859335038363, "percentage": 42.1, "elapsed_time": "4:08:55", "remaining_time": "5:42:23"} +{"current_steps": 824, "total_steps": 1955, "loss": 1.0133, "lr": 5.73691409214942e-05, "epoch": 2.1074168797953963, "percentage": 42.15, "elapsed_time": "4:09:13", "remaining_time": "5:42:05"} +{"current_steps": 825, "total_steps": 1955, "loss": 1.0144, "lr": 5.730475952299275e-05, "epoch": 2.10997442455243, "percentage": 42.2, "elapsed_time": "4:09:31", "remaining_time": "5:41:47"} +{"current_steps": 826, "total_steps": 1955, "loss": 1.0104, "lr": 5.724032292519777e-05, "epoch": 2.112531969309463, "percentage": 42.25, "elapsed_time": "4:09:49", "remaining_time": "5:41:28"} +{"current_steps": 827, "total_steps": 1955, "loss": 1.0029, "lr": 5.717583133365125e-05, "epoch": 2.1150895140664963, "percentage": 42.3, "elapsed_time": "4:10:07", "remaining_time": "5:41:09"} +{"current_steps": 828, "total_steps": 1955, "loss": 1.0131, "lr": 5.7111284954070576e-05, "epoch": 2.1176470588235294, "percentage": 42.35, "elapsed_time": "4:10:25", "remaining_time": "5:40:51"} +{"current_steps": 829, "total_steps": 1955, "loss": 1.017, "lr": 5.704668399234795e-05, "epoch": 2.1202046035805626, "percentage": 42.4, "elapsed_time": "4:10:43", "remaining_time": "5:40:32"} +{"current_steps": 830, "total_steps": 1955, "loss": 1.0435, "lr": 5.6982028654549624e-05, "epoch": 2.122762148337596, "percentage": 42.46, "elapsed_time": "4:11:01", "remaining_time": "5:40:14"} +{"current_steps": 831, "total_steps": 1955, "loss": 1.0343, "lr": 5.6917319146915354e-05, "epoch": 2.125319693094629, "percentage": 42.51, "elapsed_time": "4:11:19", "remaining_time": "5:39:56"} +{"current_steps": 832, "total_steps": 1955, "loss": 1.0317, "lr": 5.685255567585766e-05, "epoch": 2.1278772378516626, "percentage": 42.56, "elapsed_time": "4:11:37", "remaining_time": "5:39:38"} +{"current_steps": 833, "total_steps": 1955, "loss": 1.0305, "lr": 5.678773844796118e-05, "epoch": 2.130434782608696, "percentage": 42.61, "elapsed_time": "4:11:55", "remaining_time": "5:39:19"} +{"current_steps": 834, "total_steps": 1955, "loss": 0.9911, "lr": 5.6722867669982084e-05, "epoch": 2.132992327365729, "percentage": 42.66, "elapsed_time": "4:12:14", "remaining_time": "5:39:02"} +{"current_steps": 835, "total_steps": 1955, "loss": 1.0223, "lr": 5.66579435488473e-05, "epoch": 2.135549872122762, "percentage": 42.71, "elapsed_time": "4:12:32", "remaining_time": "5:38:44"} +{"current_steps": 836, "total_steps": 1955, "loss": 1.0383, "lr": 5.659296629165393e-05, "epoch": 2.1381074168797953, "percentage": 42.76, "elapsed_time": "4:12:51", "remaining_time": "5:38:26"} +{"current_steps": 837, "total_steps": 1955, "loss": 0.9768, "lr": 5.652793610566859e-05, "epoch": 2.1406649616368285, "percentage": 42.81, "elapsed_time": "4:13:09", "remaining_time": "5:38:08"} +{"current_steps": 838, "total_steps": 1955, "loss": 1.0356, "lr": 5.646285319832671e-05, "epoch": 2.1432225063938617, "percentage": 42.86, "elapsed_time": "4:13:27", "remaining_time": "5:37:50"} +{"current_steps": 839, "total_steps": 1955, "loss": 1.0125, "lr": 5.639771777723189e-05, "epoch": 2.1457800511508953, "percentage": 42.92, "elapsed_time": "4:13:45", "remaining_time": "5:37:31"} +{"current_steps": 840, "total_steps": 1955, "loss": 1.0369, "lr": 5.633253005015526e-05, "epoch": 2.1483375959079285, "percentage": 42.97, "elapsed_time": "4:14:03", "remaining_time": "5:37:13"} +{"current_steps": 841, "total_steps": 1955, "loss": 1.0142, "lr": 5.6267290225034775e-05, "epoch": 2.1508951406649617, "percentage": 43.02, "elapsed_time": "4:14:21", "remaining_time": "5:36:55"} +{"current_steps": 842, "total_steps": 1955, "loss": 1.0262, "lr": 5.620199850997459e-05, "epoch": 2.153452685421995, "percentage": 43.07, "elapsed_time": "4:14:39", "remaining_time": "5:36:36"} +{"current_steps": 843, "total_steps": 1955, "loss": 1.0015, "lr": 5.6136655113244386e-05, "epoch": 2.156010230179028, "percentage": 43.12, "elapsed_time": "4:14:57", "remaining_time": "5:36:18"} +{"current_steps": 844, "total_steps": 1955, "loss": 1.0493, "lr": 5.607126024327867e-05, "epoch": 2.1585677749360612, "percentage": 43.17, "elapsed_time": "4:15:15", "remaining_time": "5:36:00"} +{"current_steps": 845, "total_steps": 1955, "loss": 1.0086, "lr": 5.600581410867614e-05, "epoch": 2.1611253196930944, "percentage": 43.22, "elapsed_time": "4:15:33", "remaining_time": "5:35:41"} +{"current_steps": 846, "total_steps": 1955, "loss": 1.0048, "lr": 5.5940316918199084e-05, "epoch": 2.163682864450128, "percentage": 43.27, "elapsed_time": "4:15:51", "remaining_time": "5:35:23"} +{"current_steps": 847, "total_steps": 1955, "loss": 0.9661, "lr": 5.587476888077258e-05, "epoch": 2.166240409207161, "percentage": 43.32, "elapsed_time": "4:16:09", "remaining_time": "5:35:04"} +{"current_steps": 848, "total_steps": 1955, "loss": 1.0578, "lr": 5.580917020548391e-05, "epoch": 2.1687979539641944, "percentage": 43.38, "elapsed_time": "4:16:27", "remaining_time": "5:34:46"} +{"current_steps": 849, "total_steps": 1955, "loss": 1.0163, "lr": 5.5743521101581934e-05, "epoch": 2.1713554987212276, "percentage": 43.43, "elapsed_time": "4:16:45", "remaining_time": "5:34:28"} +{"current_steps": 850, "total_steps": 1955, "loss": 1.0171, "lr": 5.56778217784763e-05, "epoch": 2.1739130434782608, "percentage": 43.48, "elapsed_time": "4:17:03", "remaining_time": "5:34:10"} +{"current_steps": 851, "total_steps": 1955, "loss": 1.0219, "lr": 5.561207244573688e-05, "epoch": 2.176470588235294, "percentage": 43.53, "elapsed_time": "4:17:21", "remaining_time": "5:33:52"} +{"current_steps": 852, "total_steps": 1955, "loss": 1.0355, "lr": 5.554627331309308e-05, "epoch": 2.1790281329923276, "percentage": 43.58, "elapsed_time": "4:17:39", "remaining_time": "5:33:33"} +{"current_steps": 853, "total_steps": 1955, "loss": 0.9744, "lr": 5.548042459043313e-05, "epoch": 2.1815856777493607, "percentage": 43.63, "elapsed_time": "4:17:57", "remaining_time": "5:33:15"} +{"current_steps": 854, "total_steps": 1955, "loss": 0.9585, "lr": 5.541452648780348e-05, "epoch": 2.184143222506394, "percentage": 43.68, "elapsed_time": "4:18:15", "remaining_time": "5:32:56"} +{"current_steps": 855, "total_steps": 1955, "loss": 1.0107, "lr": 5.5348579215408064e-05, "epoch": 2.186700767263427, "percentage": 43.73, "elapsed_time": "4:18:33", "remaining_time": "5:32:38"} +{"current_steps": 856, "total_steps": 1955, "loss": 1.029, "lr": 5.528258298360767e-05, "epoch": 2.1892583120204603, "percentage": 43.79, "elapsed_time": "4:18:51", "remaining_time": "5:32:20"} +{"current_steps": 857, "total_steps": 1955, "loss": 1.0448, "lr": 5.521653800291927e-05, "epoch": 2.1918158567774935, "percentage": 43.84, "elapsed_time": "4:19:09", "remaining_time": "5:32:02"} +{"current_steps": 858, "total_steps": 1955, "loss": 1.0376, "lr": 5.51504444840153e-05, "epoch": 2.1943734015345266, "percentage": 43.89, "elapsed_time": "4:19:27", "remaining_time": "5:31:43"} +{"current_steps": 859, "total_steps": 1955, "loss": 1.0336, "lr": 5.508430263772308e-05, "epoch": 2.1969309462915603, "percentage": 43.94, "elapsed_time": "4:19:45", "remaining_time": "5:31:26"} +{"current_steps": 860, "total_steps": 1955, "loss": 1.0302, "lr": 5.501811267502403e-05, "epoch": 2.1994884910485935, "percentage": 43.99, "elapsed_time": "4:20:03", "remaining_time": "5:31:07"} +{"current_steps": 861, "total_steps": 1955, "loss": 1.023, "lr": 5.49518748070531e-05, "epoch": 2.2020460358056266, "percentage": 44.04, "elapsed_time": "4:20:21", "remaining_time": "5:30:49"} +{"current_steps": 862, "total_steps": 1955, "loss": 0.9957, "lr": 5.488558924509802e-05, "epoch": 2.20460358056266, "percentage": 44.09, "elapsed_time": "4:20:40", "remaining_time": "5:30:32"} +{"current_steps": 863, "total_steps": 1955, "loss": 1.0192, "lr": 5.481925620059866e-05, "epoch": 2.207161125319693, "percentage": 44.14, "elapsed_time": "4:20:59", "remaining_time": "5:30:15"} +{"current_steps": 864, "total_steps": 1955, "loss": 1.0212, "lr": 5.4752875885146354e-05, "epoch": 2.209718670076726, "percentage": 44.19, "elapsed_time": "4:21:17", "remaining_time": "5:29:56"} +{"current_steps": 865, "total_steps": 1955, "loss": 0.9785, "lr": 5.468644851048324e-05, "epoch": 2.21227621483376, "percentage": 44.25, "elapsed_time": "4:21:35", "remaining_time": "5:29:38"} +{"current_steps": 866, "total_steps": 1955, "loss": 1.0092, "lr": 5.461997428850154e-05, "epoch": 2.214833759590793, "percentage": 44.3, "elapsed_time": "4:21:53", "remaining_time": "5:29:20"} +{"current_steps": 867, "total_steps": 1955, "loss": 1.0514, "lr": 5.4553453431242936e-05, "epoch": 2.217391304347826, "percentage": 44.35, "elapsed_time": "4:22:12", "remaining_time": "5:29:02"} +{"current_steps": 868, "total_steps": 1955, "loss": 0.9922, "lr": 5.448688615089784e-05, "epoch": 2.2199488491048593, "percentage": 44.4, "elapsed_time": "4:22:30", "remaining_time": "5:28:44"} +{"current_steps": 869, "total_steps": 1955, "loss": 1.0292, "lr": 5.442027265980478e-05, "epoch": 2.2225063938618925, "percentage": 44.45, "elapsed_time": "4:22:48", "remaining_time": "5:28:26"} +{"current_steps": 870, "total_steps": 1955, "loss": 1.0144, "lr": 5.4353613170449636e-05, "epoch": 2.2250639386189257, "percentage": 44.5, "elapsed_time": "4:23:06", "remaining_time": "5:28:08"} +{"current_steps": 871, "total_steps": 1955, "loss": 0.995, "lr": 5.42869078954651e-05, "epoch": 2.227621483375959, "percentage": 44.55, "elapsed_time": "4:23:24", "remaining_time": "5:27:49"} +{"current_steps": 872, "total_steps": 1955, "loss": 1.016, "lr": 5.422015704762981e-05, "epoch": 2.2301790281329925, "percentage": 44.6, "elapsed_time": "4:23:42", "remaining_time": "5:27:31"} +{"current_steps": 873, "total_steps": 1955, "loss": 1.0321, "lr": 5.415336083986786e-05, "epoch": 2.2327365728900257, "percentage": 44.65, "elapsed_time": "4:24:00", "remaining_time": "5:27:13"} +{"current_steps": 874, "total_steps": 1955, "loss": 1.0206, "lr": 5.408651948524799e-05, "epoch": 2.235294117647059, "percentage": 44.71, "elapsed_time": "4:24:18", "remaining_time": "5:26:54"} +{"current_steps": 875, "total_steps": 1955, "loss": 0.9878, "lr": 5.401963319698295e-05, "epoch": 2.237851662404092, "percentage": 44.76, "elapsed_time": "4:24:36", "remaining_time": "5:26:36"} +{"current_steps": 876, "total_steps": 1955, "loss": 1.0073, "lr": 5.395270218842886e-05, "epoch": 2.2404092071611252, "percentage": 44.81, "elapsed_time": "4:24:54", "remaining_time": "5:26:18"} +{"current_steps": 877, "total_steps": 1955, "loss": 0.9968, "lr": 5.388572667308442e-05, "epoch": 2.2429667519181584, "percentage": 44.86, "elapsed_time": "4:25:12", "remaining_time": "5:25:59"} +{"current_steps": 878, "total_steps": 1955, "loss": 1.0038, "lr": 5.381870686459038e-05, "epoch": 2.2455242966751916, "percentage": 44.91, "elapsed_time": "4:25:30", "remaining_time": "5:25:41"} +{"current_steps": 879, "total_steps": 1955, "loss": 1.0037, "lr": 5.375164297672872e-05, "epoch": 2.2480818414322252, "percentage": 44.96, "elapsed_time": "4:25:48", "remaining_time": "5:25:23"} +{"current_steps": 880, "total_steps": 1955, "loss": 0.9793, "lr": 5.368453522342204e-05, "epoch": 2.2506393861892584, "percentage": 45.01, "elapsed_time": "4:26:07", "remaining_time": "5:25:06"} +{"current_steps": 881, "total_steps": 1955, "loss": 0.9997, "lr": 5.361738381873289e-05, "epoch": 2.2531969309462916, "percentage": 45.06, "elapsed_time": "4:26:25", "remaining_time": "5:24:47"} +{"current_steps": 882, "total_steps": 1955, "loss": 0.9663, "lr": 5.355018897686302e-05, "epoch": 2.2557544757033248, "percentage": 45.12, "elapsed_time": "4:26:43", "remaining_time": "5:24:29"} +{"current_steps": 883, "total_steps": 1955, "loss": 1.0061, "lr": 5.348295091215276e-05, "epoch": 2.258312020460358, "percentage": 45.17, "elapsed_time": "4:27:01", "remaining_time": "5:24:11"} +{"current_steps": 884, "total_steps": 1955, "loss": 1.0006, "lr": 5.341566983908031e-05, "epoch": 2.260869565217391, "percentage": 45.22, "elapsed_time": "4:27:19", "remaining_time": "5:23:52"} +{"current_steps": 885, "total_steps": 1955, "loss": 1.0199, "lr": 5.3348345972261065e-05, "epoch": 2.2634271099744243, "percentage": 45.27, "elapsed_time": "4:27:38", "remaining_time": "5:23:34"} +{"current_steps": 886, "total_steps": 1955, "loss": 1.028, "lr": 5.328097952644691e-05, "epoch": 2.265984654731458, "percentage": 45.32, "elapsed_time": "4:27:56", "remaining_time": "5:23:16"} +{"current_steps": 887, "total_steps": 1955, "loss": 0.9823, "lr": 5.321357071652557e-05, "epoch": 2.268542199488491, "percentage": 45.37, "elapsed_time": "4:28:14", "remaining_time": "5:22:58"} +{"current_steps": 888, "total_steps": 1955, "loss": 0.9887, "lr": 5.3146119757519875e-05, "epoch": 2.2710997442455243, "percentage": 45.42, "elapsed_time": "4:28:32", "remaining_time": "5:22:40"} +{"current_steps": 889, "total_steps": 1955, "loss": 1.0133, "lr": 5.3078626864587144e-05, "epoch": 2.2736572890025575, "percentage": 45.47, "elapsed_time": "4:28:50", "remaining_time": "5:22:22"} +{"current_steps": 890, "total_steps": 1955, "loss": 1.0467, "lr": 5.3011092253018434e-05, "epoch": 2.2762148337595907, "percentage": 45.52, "elapsed_time": "4:29:08", "remaining_time": "5:22:04"} +{"current_steps": 891, "total_steps": 1955, "loss": 1.0236, "lr": 5.294351613823787e-05, "epoch": 2.2787723785166243, "percentage": 45.58, "elapsed_time": "4:29:26", "remaining_time": "5:21:46"} +{"current_steps": 892, "total_steps": 1955, "loss": 1.0073, "lr": 5.287589873580196e-05, "epoch": 2.2813299232736575, "percentage": 45.63, "elapsed_time": "4:29:44", "remaining_time": "5:21:26"} +{"current_steps": 893, "total_steps": 1955, "loss": 1.0137, "lr": 5.280824026139896e-05, "epoch": 2.2838874680306906, "percentage": 45.68, "elapsed_time": "4:30:02", "remaining_time": "5:21:08"} +{"current_steps": 894, "total_steps": 1955, "loss": 1.0335, "lr": 5.2740540930848094e-05, "epoch": 2.286445012787724, "percentage": 45.73, "elapsed_time": "4:30:20", "remaining_time": "5:20:50"} +{"current_steps": 895, "total_steps": 1955, "loss": 1.0228, "lr": 5.267280096009892e-05, "epoch": 2.289002557544757, "percentage": 45.78, "elapsed_time": "4:30:38", "remaining_time": "5:20:31"} +{"current_steps": 896, "total_steps": 1955, "loss": 1.0098, "lr": 5.260502056523063e-05, "epoch": 2.29156010230179, "percentage": 45.83, "elapsed_time": "4:30:56", "remaining_time": "5:20:13"} +{"current_steps": 897, "total_steps": 1955, "loss": 1.0088, "lr": 5.253719996245137e-05, "epoch": 2.2941176470588234, "percentage": 45.88, "elapsed_time": "4:31:14", "remaining_time": "5:19:55"} +{"current_steps": 898, "total_steps": 1955, "loss": 0.9849, "lr": 5.246933936809754e-05, "epoch": 2.296675191815857, "percentage": 45.93, "elapsed_time": "4:31:32", "remaining_time": "5:19:37"} +{"current_steps": 899, "total_steps": 1955, "loss": 1.0194, "lr": 5.240143899863312e-05, "epoch": 2.29923273657289, "percentage": 45.98, "elapsed_time": "4:31:50", "remaining_time": "5:19:18"} +{"current_steps": 900, "total_steps": 1955, "loss": 1.0043, "lr": 5.233349907064892e-05, "epoch": 2.3017902813299234, "percentage": 46.04, "elapsed_time": "4:32:08", "remaining_time": "5:19:00"} +{"current_steps": 901, "total_steps": 1955, "loss": 0.9957, "lr": 5.2265519800861996e-05, "epoch": 2.3043478260869565, "percentage": 46.09, "elapsed_time": "4:32:26", "remaining_time": "5:18:42"} +{"current_steps": 902, "total_steps": 1955, "loss": 1.0013, "lr": 5.219750140611485e-05, "epoch": 2.3069053708439897, "percentage": 46.14, "elapsed_time": "4:32:44", "remaining_time": "5:18:23"} +{"current_steps": 903, "total_steps": 1955, "loss": 1.0235, "lr": 5.212944410337482e-05, "epoch": 2.309462915601023, "percentage": 46.19, "elapsed_time": "4:33:03", "remaining_time": "5:18:06"} +{"current_steps": 904, "total_steps": 1955, "loss": 0.9498, "lr": 5.206134810973332e-05, "epoch": 2.312020460358056, "percentage": 46.24, "elapsed_time": "4:33:21", "remaining_time": "5:17:49"} +{"current_steps": 905, "total_steps": 1955, "loss": 0.9856, "lr": 5.19932136424052e-05, "epoch": 2.3145780051150897, "percentage": 46.29, "elapsed_time": "4:33:39", "remaining_time": "5:17:30"} +{"current_steps": 906, "total_steps": 1955, "loss": 0.9942, "lr": 5.192504091872807e-05, "epoch": 2.317135549872123, "percentage": 46.34, "elapsed_time": "4:33:57", "remaining_time": "5:17:12"} +{"current_steps": 907, "total_steps": 1955, "loss": 0.9985, "lr": 5.185683015616149e-05, "epoch": 2.319693094629156, "percentage": 46.39, "elapsed_time": "4:34:16", "remaining_time": "5:16:54"} +{"current_steps": 908, "total_steps": 1955, "loss": 1.0386, "lr": 5.178858157228643e-05, "epoch": 2.3222506393861893, "percentage": 46.45, "elapsed_time": "4:34:34", "remaining_time": "5:16:36"} +{"current_steps": 909, "total_steps": 1955, "loss": 0.9756, "lr": 5.1720295384804466e-05, "epoch": 2.3248081841432224, "percentage": 46.5, "elapsed_time": "4:34:52", "remaining_time": "5:16:17"} +{"current_steps": 910, "total_steps": 1955, "loss": 1.0509, "lr": 5.165197181153713e-05, "epoch": 2.3273657289002556, "percentage": 46.55, "elapsed_time": "4:35:10", "remaining_time": "5:16:00"} +{"current_steps": 911, "total_steps": 1955, "loss": 1.0279, "lr": 5.158361107042524e-05, "epoch": 2.329923273657289, "percentage": 46.6, "elapsed_time": "4:35:28", "remaining_time": "5:15:41"} +{"current_steps": 912, "total_steps": 1955, "loss": 0.9884, "lr": 5.1515213379528115e-05, "epoch": 2.3324808184143224, "percentage": 46.65, "elapsed_time": "4:35:46", "remaining_time": "5:15:23"} +{"current_steps": 913, "total_steps": 1955, "loss": 1.0194, "lr": 5.1446778957022984e-05, "epoch": 2.3350383631713556, "percentage": 46.7, "elapsed_time": "4:36:04", "remaining_time": "5:15:05"} +{"current_steps": 914, "total_steps": 1955, "loss": 1.0192, "lr": 5.137830802120422e-05, "epoch": 2.337595907928389, "percentage": 46.75, "elapsed_time": "4:36:22", "remaining_time": "5:14:46"} +{"current_steps": 915, "total_steps": 1955, "loss": 1.0164, "lr": 5.130980079048269e-05, "epoch": 2.340153452685422, "percentage": 46.8, "elapsed_time": "4:36:40", "remaining_time": "5:14:28"} +{"current_steps": 916, "total_steps": 1955, "loss": 1.0374, "lr": 5.124125748338501e-05, "epoch": 2.342710997442455, "percentage": 46.85, "elapsed_time": "4:36:58", "remaining_time": "5:14:10"} +{"current_steps": 917, "total_steps": 1955, "loss": 0.9791, "lr": 5.117267831855289e-05, "epoch": 2.3452685421994883, "percentage": 46.91, "elapsed_time": "4:37:17", "remaining_time": "5:13:52"} +{"current_steps": 918, "total_steps": 1955, "loss": 1.0071, "lr": 5.110406351474241e-05, "epoch": 2.3478260869565215, "percentage": 46.96, "elapsed_time": "4:37:35", "remaining_time": "5:13:34"} +{"current_steps": 919, "total_steps": 1955, "loss": 0.9929, "lr": 5.103541329082334e-05, "epoch": 2.350383631713555, "percentage": 47.01, "elapsed_time": "4:37:53", "remaining_time": "5:13:16"} +{"current_steps": 920, "total_steps": 1955, "loss": 0.9936, "lr": 5.096672786577843e-05, "epoch": 2.3529411764705883, "percentage": 47.06, "elapsed_time": "4:38:12", "remaining_time": "5:12:59"} +{"current_steps": 921, "total_steps": 1955, "loss": 0.9976, "lr": 5.089800745870273e-05, "epoch": 2.3554987212276215, "percentage": 47.11, "elapsed_time": "4:38:30", "remaining_time": "5:12:40"} +{"current_steps": 922, "total_steps": 1955, "loss": 0.9972, "lr": 5.082925228880284e-05, "epoch": 2.3580562659846547, "percentage": 47.16, "elapsed_time": "4:38:48", "remaining_time": "5:12:22"} +{"current_steps": 923, "total_steps": 1955, "loss": 0.9994, "lr": 5.076046257539632e-05, "epoch": 2.360613810741688, "percentage": 47.21, "elapsed_time": "4:39:06", "remaining_time": "5:12:04"} +{"current_steps": 924, "total_steps": 1955, "loss": 1.0077, "lr": 5.0691638537910806e-05, "epoch": 2.363171355498721, "percentage": 47.26, "elapsed_time": "4:39:24", "remaining_time": "5:11:45"} +{"current_steps": 925, "total_steps": 1955, "loss": 0.9952, "lr": 5.0622780395883526e-05, "epoch": 2.3657289002557547, "percentage": 47.31, "elapsed_time": "4:39:42", "remaining_time": "5:11:27"} +{"current_steps": 926, "total_steps": 1955, "loss": 0.9719, "lr": 5.055388836896045e-05, "epoch": 2.368286445012788, "percentage": 47.37, "elapsed_time": "4:40:00", "remaining_time": "5:11:09"} +{"current_steps": 927, "total_steps": 1955, "loss": 1.0277, "lr": 5.0484962676895624e-05, "epoch": 2.370843989769821, "percentage": 47.42, "elapsed_time": "4:40:18", "remaining_time": "5:10:50"} +{"current_steps": 928, "total_steps": 1955, "loss": 1.0331, "lr": 5.041600353955054e-05, "epoch": 2.373401534526854, "percentage": 47.47, "elapsed_time": "4:40:36", "remaining_time": "5:10:32"} +{"current_steps": 929, "total_steps": 1955, "loss": 1.003, "lr": 5.0347011176893284e-05, "epoch": 2.3759590792838874, "percentage": 47.52, "elapsed_time": "4:40:54", "remaining_time": "5:10:14"} +{"current_steps": 930, "total_steps": 1955, "loss": 1.0268, "lr": 5.0277985808998e-05, "epoch": 2.3785166240409206, "percentage": 47.57, "elapsed_time": "4:41:12", "remaining_time": "5:09:56"} +{"current_steps": 931, "total_steps": 1955, "loss": 0.9939, "lr": 5.020892765604407e-05, "epoch": 2.381074168797954, "percentage": 47.62, "elapsed_time": "4:41:30", "remaining_time": "5:09:37"} +{"current_steps": 932, "total_steps": 1955, "loss": 1.0188, "lr": 5.013983693831549e-05, "epoch": 2.3836317135549874, "percentage": 47.67, "elapsed_time": "4:41:48", "remaining_time": "5:09:19"} +{"current_steps": 933, "total_steps": 1955, "loss": 1.0014, "lr": 5.007071387620009e-05, "epoch": 2.3861892583120206, "percentage": 47.72, "elapsed_time": "4:42:07", "remaining_time": "5:09:02"} +{"current_steps": 934, "total_steps": 1955, "loss": 0.9896, "lr": 5.000155869018893e-05, "epoch": 2.3887468030690537, "percentage": 47.77, "elapsed_time": "4:42:25", "remaining_time": "5:08:44"} +{"current_steps": 935, "total_steps": 1955, "loss": 0.9879, "lr": 4.993237160087549e-05, "epoch": 2.391304347826087, "percentage": 47.83, "elapsed_time": "4:42:43", "remaining_time": "5:08:25"} +{"current_steps": 936, "total_steps": 1955, "loss": 1.0062, "lr": 4.986315282895502e-05, "epoch": 2.39386189258312, "percentage": 47.88, "elapsed_time": "4:43:01", "remaining_time": "5:08:07"} +{"current_steps": 937, "total_steps": 1955, "loss": 1.0087, "lr": 4.979390259522386e-05, "epoch": 2.3964194373401533, "percentage": 47.93, "elapsed_time": "4:43:19", "remaining_time": "5:07:49"} +{"current_steps": 938, "total_steps": 1955, "loss": 1.0173, "lr": 4.972462112057868e-05, "epoch": 2.398976982097187, "percentage": 47.98, "elapsed_time": "4:43:38", "remaining_time": "5:07:31"} +{"current_steps": 939, "total_steps": 1955, "loss": 1.0014, "lr": 4.9655308626015835e-05, "epoch": 2.40153452685422, "percentage": 48.03, "elapsed_time": "4:43:56", "remaining_time": "5:07:13"} +{"current_steps": 940, "total_steps": 1955, "loss": 1.0102, "lr": 4.958596533263061e-05, "epoch": 2.4040920716112533, "percentage": 48.08, "elapsed_time": "4:44:15", "remaining_time": "5:06:55"} +{"current_steps": 941, "total_steps": 1955, "loss": 0.999, "lr": 4.9516591461616505e-05, "epoch": 2.4066496163682864, "percentage": 48.13, "elapsed_time": "4:44:33", "remaining_time": "5:06:37"} +{"current_steps": 942, "total_steps": 1955, "loss": 1.0091, "lr": 4.9447187234264615e-05, "epoch": 2.4092071611253196, "percentage": 48.18, "elapsed_time": "4:44:51", "remaining_time": "5:06:19"} +{"current_steps": 943, "total_steps": 1955, "loss": 1.0122, "lr": 4.937775287196282e-05, "epoch": 2.411764705882353, "percentage": 48.24, "elapsed_time": "4:45:09", "remaining_time": "5:06:01"} +{"current_steps": 944, "total_steps": 1955, "loss": 1.0119, "lr": 4.930828859619515e-05, "epoch": 2.414322250639386, "percentage": 48.29, "elapsed_time": "4:45:27", "remaining_time": "5:05:43"} +{"current_steps": 945, "total_steps": 1955, "loss": 0.9659, "lr": 4.923879462854105e-05, "epoch": 2.4168797953964196, "percentage": 48.34, "elapsed_time": "4:45:46", "remaining_time": "5:05:25"} +{"current_steps": 946, "total_steps": 1955, "loss": 0.9856, "lr": 4.9169271190674656e-05, "epoch": 2.419437340153453, "percentage": 48.39, "elapsed_time": "4:46:04", "remaining_time": "5:05:07"} +{"current_steps": 947, "total_steps": 1955, "loss": 1.0361, "lr": 4.909971850436414e-05, "epoch": 2.421994884910486, "percentage": 48.44, "elapsed_time": "4:46:22", "remaining_time": "5:04:49"} +{"current_steps": 948, "total_steps": 1955, "loss": 0.9944, "lr": 4.9030136791470976e-05, "epoch": 2.424552429667519, "percentage": 48.49, "elapsed_time": "4:46:40", "remaining_time": "5:04:31"} +{"current_steps": 949, "total_steps": 1955, "loss": 1.0093, "lr": 4.896052627394917e-05, "epoch": 2.4271099744245523, "percentage": 48.54, "elapsed_time": "4:46:58", "remaining_time": "5:04:12"} +{"current_steps": 950, "total_steps": 1955, "loss": 0.9912, "lr": 4.889088717384468e-05, "epoch": 2.4296675191815855, "percentage": 48.59, "elapsed_time": "4:47:16", "remaining_time": "5:03:54"} +{"current_steps": 951, "total_steps": 1955, "loss": 1.0071, "lr": 4.8821219713294595e-05, "epoch": 2.4322250639386187, "percentage": 48.64, "elapsed_time": "4:47:34", "remaining_time": "5:03:36"} +{"current_steps": 952, "total_steps": 1955, "loss": 0.9719, "lr": 4.8751524114526486e-05, "epoch": 2.4347826086956523, "percentage": 48.7, "elapsed_time": "4:47:52", "remaining_time": "5:03:18"} +{"current_steps": 953, "total_steps": 1955, "loss": 1.0234, "lr": 4.8681800599857687e-05, "epoch": 2.4373401534526855, "percentage": 48.75, "elapsed_time": "4:48:10", "remaining_time": "5:03:00"} +{"current_steps": 954, "total_steps": 1955, "loss": 1.0258, "lr": 4.861204939169454e-05, "epoch": 2.4398976982097187, "percentage": 48.8, "elapsed_time": "4:48:29", "remaining_time": "5:02:41"} +{"current_steps": 955, "total_steps": 1955, "loss": 1.0007, "lr": 4.854227071253177e-05, "epoch": 2.442455242966752, "percentage": 48.85, "elapsed_time": "4:48:47", "remaining_time": "5:02:23"} +{"current_steps": 956, "total_steps": 1955, "loss": 1.0077, "lr": 4.847246478495172e-05, "epoch": 2.445012787723785, "percentage": 48.9, "elapsed_time": "4:49:05", "remaining_time": "5:02:05"} +{"current_steps": 957, "total_steps": 1955, "loss": 0.9793, "lr": 4.840263183162363e-05, "epoch": 2.4475703324808182, "percentage": 48.95, "elapsed_time": "4:49:23", "remaining_time": "5:01:47"} +{"current_steps": 958, "total_steps": 1955, "loss": 0.9742, "lr": 4.833277207530298e-05, "epoch": 2.4501278772378514, "percentage": 49.0, "elapsed_time": "4:49:41", "remaining_time": "5:01:29"} +{"current_steps": 959, "total_steps": 1955, "loss": 0.9968, "lr": 4.826288573883071e-05, "epoch": 2.452685421994885, "percentage": 49.05, "elapsed_time": "4:49:59", "remaining_time": "5:01:10"} +{"current_steps": 960, "total_steps": 1955, "loss": 0.9728, "lr": 4.8192973045132575e-05, "epoch": 2.455242966751918, "percentage": 49.1, "elapsed_time": "4:50:17", "remaining_time": "5:00:52"} +{"current_steps": 961, "total_steps": 1955, "loss": 1.0399, "lr": 4.812303421721837e-05, "epoch": 2.4578005115089514, "percentage": 49.16, "elapsed_time": "4:50:35", "remaining_time": "5:00:34"} +{"current_steps": 962, "total_steps": 1955, "loss": 1.0156, "lr": 4.8053069478181315e-05, "epoch": 2.4603580562659846, "percentage": 49.21, "elapsed_time": "4:50:53", "remaining_time": "5:00:16"} +{"current_steps": 963, "total_steps": 1955, "loss": 1.023, "lr": 4.798307905119723e-05, "epoch": 2.4629156010230178, "percentage": 49.26, "elapsed_time": "4:51:11", "remaining_time": "4:59:57"} +{"current_steps": 964, "total_steps": 1955, "loss": 1.0285, "lr": 4.7913063159523896e-05, "epoch": 2.4654731457800514, "percentage": 49.31, "elapsed_time": "4:51:29", "remaining_time": "4:59:39"} +{"current_steps": 965, "total_steps": 1955, "loss": 0.9802, "lr": 4.784302202650031e-05, "epoch": 2.4680306905370846, "percentage": 49.36, "elapsed_time": "4:51:47", "remaining_time": "4:59:21"} +{"current_steps": 966, "total_steps": 1955, "loss": 0.9958, "lr": 4.777295587554597e-05, "epoch": 2.4705882352941178, "percentage": 49.41, "elapsed_time": "4:52:05", "remaining_time": "4:59:03"} +{"current_steps": 967, "total_steps": 1955, "loss": 1.0056, "lr": 4.770286493016025e-05, "epoch": 2.473145780051151, "percentage": 49.46, "elapsed_time": "4:52:24", "remaining_time": "4:58:45"} +{"current_steps": 968, "total_steps": 1955, "loss": 1.0109, "lr": 4.7632749413921524e-05, "epoch": 2.475703324808184, "percentage": 49.51, "elapsed_time": "4:52:42", "remaining_time": "4:58:26"} +{"current_steps": 969, "total_steps": 1955, "loss": 1.0193, "lr": 4.756260955048658e-05, "epoch": 2.4782608695652173, "percentage": 49.57, "elapsed_time": "4:53:00", "remaining_time": "4:58:08"} +{"current_steps": 970, "total_steps": 1955, "loss": 0.9803, "lr": 4.74924455635899e-05, "epoch": 2.4808184143222505, "percentage": 49.62, "elapsed_time": "4:53:18", "remaining_time": "4:57:50"} +{"current_steps": 971, "total_steps": 1955, "loss": 1.0025, "lr": 4.742225767704286e-05, "epoch": 2.483375959079284, "percentage": 49.67, "elapsed_time": "4:53:36", "remaining_time": "4:57:32"} +{"current_steps": 972, "total_steps": 1955, "loss": 1.0224, "lr": 4.7352046114733116e-05, "epoch": 2.4859335038363173, "percentage": 49.72, "elapsed_time": "4:53:54", "remaining_time": "4:57:14"} +{"current_steps": 973, "total_steps": 1955, "loss": 1.0031, "lr": 4.728181110062383e-05, "epoch": 2.4884910485933505, "percentage": 49.77, "elapsed_time": "4:54:12", "remaining_time": "4:56:55"} +{"current_steps": 974, "total_steps": 1955, "loss": 1.0016, "lr": 4.721155285875294e-05, "epoch": 2.4910485933503836, "percentage": 49.82, "elapsed_time": "4:54:30", "remaining_time": "4:56:37"} +{"current_steps": 975, "total_steps": 1955, "loss": 0.9816, "lr": 4.714127161323255e-05, "epoch": 2.493606138107417, "percentage": 49.87, "elapsed_time": "4:54:48", "remaining_time": "4:56:19"} +{"current_steps": 976, "total_steps": 1955, "loss": 1.0099, "lr": 4.707096758824807e-05, "epoch": 2.49616368286445, "percentage": 49.92, "elapsed_time": "4:55:06", "remaining_time": "4:56:01"} +{"current_steps": 977, "total_steps": 1955, "loss": 0.999, "lr": 4.7000641008057605e-05, "epoch": 2.498721227621483, "percentage": 49.97, "elapsed_time": "4:55:24", "remaining_time": "4:55:42"} +{"current_steps": 978, "total_steps": 1955, "loss": 1.0151, "lr": 4.6930292096991214e-05, "epoch": 2.501278772378517, "percentage": 50.03, "elapsed_time": "4:55:42", "remaining_time": "4:55:24"} +{"current_steps": 979, "total_steps": 1955, "loss": 1.0189, "lr": 4.685992107945016e-05, "epoch": 2.50383631713555, "percentage": 50.08, "elapsed_time": "4:56:00", "remaining_time": "4:55:06"} +{"current_steps": 980, "total_steps": 1955, "loss": 0.9779, "lr": 4.6789528179906254e-05, "epoch": 2.506393861892583, "percentage": 50.13, "elapsed_time": "4:56:18", "remaining_time": "4:54:48"} +{"current_steps": 981, "total_steps": 1955, "loss": 0.987, "lr": 4.67191136229011e-05, "epoch": 2.5089514066496164, "percentage": 50.18, "elapsed_time": "4:56:36", "remaining_time": "4:54:29"} +{"current_steps": 982, "total_steps": 1955, "loss": 0.9693, "lr": 4.664867763304536e-05, "epoch": 2.5115089514066495, "percentage": 50.23, "elapsed_time": "4:56:54", "remaining_time": "4:54:11"} +{"current_steps": 983, "total_steps": 1955, "loss": 0.98, "lr": 4.657822043501807e-05, "epoch": 2.5140664961636827, "percentage": 50.28, "elapsed_time": "4:57:12", "remaining_time": "4:53:53"} +{"current_steps": 984, "total_steps": 1955, "loss": 1.0107, "lr": 4.6507742253565964e-05, "epoch": 2.516624040920716, "percentage": 50.33, "elapsed_time": "4:57:31", "remaining_time": "4:53:35"} +{"current_steps": 985, "total_steps": 1955, "loss": 1.0085, "lr": 4.643724331350265e-05, "epoch": 2.5191815856777495, "percentage": 50.38, "elapsed_time": "4:57:50", "remaining_time": "4:53:18"} +{"current_steps": 986, "total_steps": 1955, "loss": 0.9817, "lr": 4.6366723839707976e-05, "epoch": 2.5217391304347827, "percentage": 50.43, "elapsed_time": "4:58:08", "remaining_time": "4:52:59"} +{"current_steps": 987, "total_steps": 1955, "loss": 0.9931, "lr": 4.629618405712729e-05, "epoch": 2.524296675191816, "percentage": 50.49, "elapsed_time": "4:58:26", "remaining_time": "4:52:41"} +{"current_steps": 988, "total_steps": 1955, "loss": 1.0179, "lr": 4.6225624190770725e-05, "epoch": 2.526854219948849, "percentage": 50.54, "elapsed_time": "4:58:44", "remaining_time": "4:52:23"} +{"current_steps": 989, "total_steps": 1955, "loss": 1.0176, "lr": 4.615504446571248e-05, "epoch": 2.5294117647058822, "percentage": 50.59, "elapsed_time": "4:59:02", "remaining_time": "4:52:05"} +{"current_steps": 990, "total_steps": 1955, "loss": 1.0257, "lr": 4.6084445107090076e-05, "epoch": 2.531969309462916, "percentage": 50.64, "elapsed_time": "4:59:20", "remaining_time": "4:51:46"} +{"current_steps": 991, "total_steps": 1955, "loss": 1.0067, "lr": 4.601382634010367e-05, "epoch": 2.5345268542199486, "percentage": 50.69, "elapsed_time": "4:59:38", "remaining_time": "4:51:28"} +{"current_steps": 992, "total_steps": 1955, "loss": 1.0238, "lr": 4.594318839001539e-05, "epoch": 2.5370843989769822, "percentage": 50.74, "elapsed_time": "4:59:56", "remaining_time": "4:51:10"} +{"current_steps": 993, "total_steps": 1955, "loss": 1.006, "lr": 4.587253148214845e-05, "epoch": 2.5396419437340154, "percentage": 50.79, "elapsed_time": "5:00:14", "remaining_time": "4:50:52"} +{"current_steps": 994, "total_steps": 1955, "loss": 0.9882, "lr": 4.580185584188661e-05, "epoch": 2.5421994884910486, "percentage": 50.84, "elapsed_time": "5:00:32", "remaining_time": "4:50:34"} +{"current_steps": 995, "total_steps": 1955, "loss": 0.9815, "lr": 4.573116169467336e-05, "epoch": 2.544757033248082, "percentage": 50.9, "elapsed_time": "5:00:50", "remaining_time": "4:50:15"} +{"current_steps": 996, "total_steps": 1955, "loss": 0.9984, "lr": 4.5660449266011214e-05, "epoch": 2.547314578005115, "percentage": 50.95, "elapsed_time": "5:01:09", "remaining_time": "4:49:57"} +{"current_steps": 997, "total_steps": 1955, "loss": 1.0026, "lr": 4.558971878146104e-05, "epoch": 2.5498721227621486, "percentage": 51.0, "elapsed_time": "5:01:27", "remaining_time": "4:49:39"} +{"current_steps": 998, "total_steps": 1955, "loss": 0.9993, "lr": 4.551897046664124e-05, "epoch": 2.5524296675191813, "percentage": 51.05, "elapsed_time": "5:01:45", "remaining_time": "4:49:21"} +{"current_steps": 999, "total_steps": 1955, "loss": 0.9949, "lr": 4.5448204547227154e-05, "epoch": 2.554987212276215, "percentage": 51.1, "elapsed_time": "5:02:03", "remaining_time": "4:49:03"} +{"current_steps": 1000, "total_steps": 1955, "loss": 0.9982, "lr": 4.537742124895024e-05, "epoch": 2.557544757033248, "percentage": 51.15, "elapsed_time": "5:02:21", "remaining_time": "4:48:44"} +{"current_steps": 1001, "total_steps": 1955, "loss": 0.9997, "lr": 4.530662079759738e-05, "epoch": 2.5601023017902813, "percentage": 51.2, "elapsed_time": "5:02:39", "remaining_time": "4:48:26"} +{"current_steps": 1002, "total_steps": 1955, "loss": 0.9904, "lr": 4.523580341901023e-05, "epoch": 2.5626598465473145, "percentage": 51.25, "elapsed_time": "5:02:57", "remaining_time": "4:48:08"} +{"current_steps": 1003, "total_steps": 1955, "loss": 1.0181, "lr": 4.516496933908438e-05, "epoch": 2.5652173913043477, "percentage": 51.3, "elapsed_time": "5:03:15", "remaining_time": "4:47:50"} +{"current_steps": 1004, "total_steps": 1955, "loss": 1.0241, "lr": 4.509411878376873e-05, "epoch": 2.5677749360613813, "percentage": 51.36, "elapsed_time": "5:03:33", "remaining_time": "4:47:32"} +{"current_steps": 1005, "total_steps": 1955, "loss": 1.0167, "lr": 4.50232519790647e-05, "epoch": 2.5703324808184145, "percentage": 51.41, "elapsed_time": "5:03:52", "remaining_time": "4:47:14"} +{"current_steps": 1006, "total_steps": 1955, "loss": 0.9547, "lr": 4.495236915102559e-05, "epoch": 2.5728900255754477, "percentage": 51.46, "elapsed_time": "5:04:10", "remaining_time": "4:46:56"} +{"current_steps": 1007, "total_steps": 1955, "loss": 1.0314, "lr": 4.488147052575577e-05, "epoch": 2.575447570332481, "percentage": 51.51, "elapsed_time": "5:04:28", "remaining_time": "4:46:38"} +{"current_steps": 1008, "total_steps": 1955, "loss": 1.0195, "lr": 4.481055632941003e-05, "epoch": 2.578005115089514, "percentage": 51.56, "elapsed_time": "5:04:46", "remaining_time": "4:46:20"} +{"current_steps": 1009, "total_steps": 1955, "loss": 1.0179, "lr": 4.4739626788192806e-05, "epoch": 2.580562659846547, "percentage": 51.61, "elapsed_time": "5:05:04", "remaining_time": "4:46:01"} +{"current_steps": 1010, "total_steps": 1955, "loss": 1.0057, "lr": 4.4668682128357474e-05, "epoch": 2.5831202046035804, "percentage": 51.66, "elapsed_time": "5:05:22", "remaining_time": "4:45:43"} +{"current_steps": 1011, "total_steps": 1955, "loss": 1.0057, "lr": 4.45977225762057e-05, "epoch": 2.585677749360614, "percentage": 51.71, "elapsed_time": "5:05:41", "remaining_time": "4:45:25"} +{"current_steps": 1012, "total_steps": 1955, "loss": 1.0112, "lr": 4.4526748358086566e-05, "epoch": 2.588235294117647, "percentage": 51.76, "elapsed_time": "5:05:59", "remaining_time": "4:45:07"} +{"current_steps": 1013, "total_steps": 1955, "loss": 0.9937, "lr": 4.445575970039599e-05, "epoch": 2.5907928388746804, "percentage": 51.82, "elapsed_time": "5:06:17", "remaining_time": "4:44:49"} +{"current_steps": 1014, "total_steps": 1955, "loss": 1.0112, "lr": 4.438475682957595e-05, "epoch": 2.5933503836317136, "percentage": 51.87, "elapsed_time": "5:06:35", "remaining_time": "4:44:31"} +{"current_steps": 1015, "total_steps": 1955, "loss": 0.9941, "lr": 4.431373997211372e-05, "epoch": 2.5959079283887467, "percentage": 51.92, "elapsed_time": "5:06:53", "remaining_time": "4:44:12"} +{"current_steps": 1016, "total_steps": 1955, "loss": 0.9932, "lr": 4.424270935454124e-05, "epoch": 2.59846547314578, "percentage": 51.97, "elapsed_time": "5:07:11", "remaining_time": "4:43:54"} +{"current_steps": 1017, "total_steps": 1955, "loss": 1.0022, "lr": 4.4171665203434336e-05, "epoch": 2.601023017902813, "percentage": 52.02, "elapsed_time": "5:07:29", "remaining_time": "4:43:36"} +{"current_steps": 1018, "total_steps": 1955, "loss": 1.0067, "lr": 4.4100607745411947e-05, "epoch": 2.6035805626598467, "percentage": 52.07, "elapsed_time": "5:07:47", "remaining_time": "4:43:18"} +{"current_steps": 1019, "total_steps": 1955, "loss": 0.9788, "lr": 4.4029537207135535e-05, "epoch": 2.60613810741688, "percentage": 52.12, "elapsed_time": "5:08:05", "remaining_time": "4:42:59"} +{"current_steps": 1020, "total_steps": 1955, "loss": 0.9874, "lr": 4.395845381530823e-05, "epoch": 2.608695652173913, "percentage": 52.17, "elapsed_time": "5:08:23", "remaining_time": "4:42:41"} +{"current_steps": 1021, "total_steps": 1955, "loss": 0.9998, "lr": 4.3887357796674174e-05, "epoch": 2.6112531969309463, "percentage": 52.23, "elapsed_time": "5:08:40", "remaining_time": "4:42:22"} +{"current_steps": 1022, "total_steps": 1955, "loss": 0.9925, "lr": 4.381624937801784e-05, "epoch": 2.6138107416879794, "percentage": 52.28, "elapsed_time": "5:08:58", "remaining_time": "4:42:04"} +{"current_steps": 1023, "total_steps": 1955, "loss": 1.0151, "lr": 4.374512878616315e-05, "epoch": 2.6163682864450126, "percentage": 52.33, "elapsed_time": "5:09:17", "remaining_time": "4:41:46"} +{"current_steps": 1024, "total_steps": 1955, "loss": 0.9945, "lr": 4.3673996247972966e-05, "epoch": 2.618925831202046, "percentage": 52.38, "elapsed_time": "5:09:35", "remaining_time": "4:41:28"} +{"current_steps": 1025, "total_steps": 1955, "loss": 0.9899, "lr": 4.360285199034819e-05, "epoch": 2.6214833759590794, "percentage": 52.43, "elapsed_time": "5:09:53", "remaining_time": "4:41:09"} +{"current_steps": 1026, "total_steps": 1955, "loss": 0.9982, "lr": 4.353169624022713e-05, "epoch": 2.6240409207161126, "percentage": 52.48, "elapsed_time": "5:10:11", "remaining_time": "4:40:51"} +{"current_steps": 1027, "total_steps": 1955, "loss": 1.005, "lr": 4.346052922458473e-05, "epoch": 2.626598465473146, "percentage": 52.53, "elapsed_time": "5:10:29", "remaining_time": "4:40:33"} +{"current_steps": 1028, "total_steps": 1955, "loss": 1.0143, "lr": 4.3389351170431914e-05, "epoch": 2.629156010230179, "percentage": 52.58, "elapsed_time": "5:10:47", "remaining_time": "4:40:15"} +{"current_steps": 1029, "total_steps": 1955, "loss": 1.0175, "lr": 4.331816230481478e-05, "epoch": 2.631713554987212, "percentage": 52.63, "elapsed_time": "5:11:05", "remaining_time": "4:39:57"} +{"current_steps": 1030, "total_steps": 1955, "loss": 1.0032, "lr": 4.3246962854813904e-05, "epoch": 2.634271099744246, "percentage": 52.69, "elapsed_time": "5:11:23", "remaining_time": "4:39:39"} +{"current_steps": 1031, "total_steps": 1955, "loss": 1.0079, "lr": 4.317575304754367e-05, "epoch": 2.6368286445012785, "percentage": 52.74, "elapsed_time": "5:11:41", "remaining_time": "4:39:21"} +{"current_steps": 1032, "total_steps": 1955, "loss": 0.9751, "lr": 4.310453311015146e-05, "epoch": 2.639386189258312, "percentage": 52.79, "elapsed_time": "5:11:59", "remaining_time": "4:39:02"} +{"current_steps": 1033, "total_steps": 1955, "loss": 1.0328, "lr": 4.3033303269817e-05, "epoch": 2.6419437340153453, "percentage": 52.84, "elapsed_time": "5:12:18", "remaining_time": "4:38:44"} +{"current_steps": 1034, "total_steps": 1955, "loss": 1.011, "lr": 4.296206375375156e-05, "epoch": 2.6445012787723785, "percentage": 52.89, "elapsed_time": "5:12:36", "remaining_time": "4:38:26"} +{"current_steps": 1035, "total_steps": 1955, "loss": 1.0149, "lr": 4.289081478919731e-05, "epoch": 2.6470588235294117, "percentage": 52.94, "elapsed_time": "5:12:54", "remaining_time": "4:38:08"} +{"current_steps": 1036, "total_steps": 1955, "loss": 0.9874, "lr": 4.281955660342656e-05, "epoch": 2.649616368286445, "percentage": 52.99, "elapsed_time": "5:13:12", "remaining_time": "4:37:50"} +{"current_steps": 1037, "total_steps": 1955, "loss": 0.9949, "lr": 4.274828942374103e-05, "epoch": 2.6521739130434785, "percentage": 53.04, "elapsed_time": "5:13:31", "remaining_time": "4:37:32"} +{"current_steps": 1038, "total_steps": 1955, "loss": 0.9997, "lr": 4.2677013477471104e-05, "epoch": 2.6547314578005117, "percentage": 53.09, "elapsed_time": "5:13:49", "remaining_time": "4:37:14"} +{"current_steps": 1039, "total_steps": 1955, "loss": 1.0078, "lr": 4.260572899197517e-05, "epoch": 2.657289002557545, "percentage": 53.15, "elapsed_time": "5:14:07", "remaining_time": "4:36:56"} +{"current_steps": 1040, "total_steps": 1955, "loss": 1.0185, "lr": 4.253443619463882e-05, "epoch": 2.659846547314578, "percentage": 53.2, "elapsed_time": "5:14:25", "remaining_time": "4:36:37"} +{"current_steps": 1041, "total_steps": 1955, "loss": 0.9965, "lr": 4.246313531287418e-05, "epoch": 2.662404092071611, "percentage": 53.25, "elapsed_time": "5:14:43", "remaining_time": "4:36:19"} +{"current_steps": 1042, "total_steps": 1955, "loss": 1.0276, "lr": 4.2391826574119156e-05, "epoch": 2.6649616368286444, "percentage": 53.3, "elapsed_time": "5:15:01", "remaining_time": "4:36:01"} +{"current_steps": 1043, "total_steps": 1955, "loss": 0.9618, "lr": 4.2320510205836714e-05, "epoch": 2.6675191815856776, "percentage": 53.35, "elapsed_time": "5:15:19", "remaining_time": "4:35:43"} +{"current_steps": 1044, "total_steps": 1955, "loss": 1.0035, "lr": 4.2249186435514184e-05, "epoch": 2.670076726342711, "percentage": 53.4, "elapsed_time": "5:15:37", "remaining_time": "4:35:24"} +{"current_steps": 1045, "total_steps": 1955, "loss": 0.9896, "lr": 4.217785549066245e-05, "epoch": 2.6726342710997444, "percentage": 53.45, "elapsed_time": "5:15:55", "remaining_time": "4:35:06"} +{"current_steps": 1046, "total_steps": 1955, "loss": 0.9743, "lr": 4.2106517598815337e-05, "epoch": 2.6751918158567776, "percentage": 53.5, "elapsed_time": "5:16:13", "remaining_time": "4:34:48"} +{"current_steps": 1047, "total_steps": 1955, "loss": 1.0077, "lr": 4.203517298752881e-05, "epoch": 2.6777493606138107, "percentage": 53.55, "elapsed_time": "5:16:31", "remaining_time": "4:34:29"} +{"current_steps": 1048, "total_steps": 1955, "loss": 1.008, "lr": 4.1963821884380246e-05, "epoch": 2.680306905370844, "percentage": 53.61, "elapsed_time": "5:16:49", "remaining_time": "4:34:11"} +{"current_steps": 1049, "total_steps": 1955, "loss": 0.9824, "lr": 4.1892464516967776e-05, "epoch": 2.682864450127877, "percentage": 53.66, "elapsed_time": "5:17:07", "remaining_time": "4:33:53"} +{"current_steps": 1050, "total_steps": 1955, "loss": 1.0083, "lr": 4.182110111290948e-05, "epoch": 2.6854219948849103, "percentage": 53.71, "elapsed_time": "5:17:25", "remaining_time": "4:33:35"} +{"current_steps": 1051, "total_steps": 1955, "loss": 1.0048, "lr": 4.17497318998427e-05, "epoch": 2.687979539641944, "percentage": 53.76, "elapsed_time": "5:17:44", "remaining_time": "4:33:17"} +{"current_steps": 1052, "total_steps": 1955, "loss": 0.9973, "lr": 4.1678357105423306e-05, "epoch": 2.690537084398977, "percentage": 53.81, "elapsed_time": "5:18:02", "remaining_time": "4:32:59"} +{"current_steps": 1053, "total_steps": 1955, "loss": 1.01, "lr": 4.1606976957324976e-05, "epoch": 2.6930946291560103, "percentage": 53.86, "elapsed_time": "5:18:20", "remaining_time": "4:32:41"} +{"current_steps": 1054, "total_steps": 1955, "loss": 0.9737, "lr": 4.153559168323847e-05, "epoch": 2.6956521739130435, "percentage": 53.91, "elapsed_time": "5:18:38", "remaining_time": "4:32:22"} +{"current_steps": 1055, "total_steps": 1955, "loss": 0.9875, "lr": 4.146420151087089e-05, "epoch": 2.6982097186700766, "percentage": 53.96, "elapsed_time": "5:18:56", "remaining_time": "4:32:04"} +{"current_steps": 1056, "total_steps": 1955, "loss": 1.0072, "lr": 4.1392806667944965e-05, "epoch": 2.70076726342711, "percentage": 54.02, "elapsed_time": "5:19:14", "remaining_time": "4:31:46"} +{"current_steps": 1057, "total_steps": 1955, "loss": 1.0074, "lr": 4.132140738219831e-05, "epoch": 2.703324808184143, "percentage": 54.07, "elapsed_time": "5:19:32", "remaining_time": "4:31:28"} +{"current_steps": 1058, "total_steps": 1955, "loss": 0.9964, "lr": 4.125000388138276e-05, "epoch": 2.7058823529411766, "percentage": 54.12, "elapsed_time": "5:19:50", "remaining_time": "4:31:10"} +{"current_steps": 1059, "total_steps": 1955, "loss": 1.0246, "lr": 4.117859639326352e-05, "epoch": 2.70843989769821, "percentage": 54.17, "elapsed_time": "5:20:09", "remaining_time": "4:30:52"} +{"current_steps": 1060, "total_steps": 1955, "loss": 0.9832, "lr": 4.1107185145618574e-05, "epoch": 2.710997442455243, "percentage": 54.22, "elapsed_time": "5:20:27", "remaining_time": "4:30:34"} +{"current_steps": 1061, "total_steps": 1955, "loss": 1.0324, "lr": 4.10357703662379e-05, "epoch": 2.713554987212276, "percentage": 54.27, "elapsed_time": "5:20:45", "remaining_time": "4:30:15"} +{"current_steps": 1062, "total_steps": 1955, "loss": 1.0105, "lr": 4.096435228292268e-05, "epoch": 2.7161125319693094, "percentage": 54.32, "elapsed_time": "5:21:03", "remaining_time": "4:29:57"} +{"current_steps": 1063, "total_steps": 1955, "loss": 1.043, "lr": 4.089293112348469e-05, "epoch": 2.718670076726343, "percentage": 54.37, "elapsed_time": "5:21:21", "remaining_time": "4:29:39"} +{"current_steps": 1064, "total_steps": 1955, "loss": 1.007, "lr": 4.082150711574552e-05, "epoch": 2.7212276214833757, "percentage": 54.42, "elapsed_time": "5:21:39", "remaining_time": "4:29:21"} +{"current_steps": 1065, "total_steps": 1955, "loss": 0.9762, "lr": 4.0750080487535807e-05, "epoch": 2.7237851662404093, "percentage": 54.48, "elapsed_time": "5:21:57", "remaining_time": "4:29:03"} +{"current_steps": 1066, "total_steps": 1955, "loss": 0.9949, "lr": 4.067865146669459e-05, "epoch": 2.7263427109974425, "percentage": 54.53, "elapsed_time": "5:22:15", "remaining_time": "4:28:44"} +{"current_steps": 1067, "total_steps": 1955, "loss": 1.0013, "lr": 4.060722028106849e-05, "epoch": 2.7289002557544757, "percentage": 54.58, "elapsed_time": "5:22:33", "remaining_time": "4:28:26"} +{"current_steps": 1068, "total_steps": 1955, "loss": 1.0213, "lr": 4.053578715851109e-05, "epoch": 2.731457800511509, "percentage": 54.63, "elapsed_time": "5:22:51", "remaining_time": "4:28:08"} +{"current_steps": 1069, "total_steps": 1955, "loss": 0.9954, "lr": 4.046435232688211e-05, "epoch": 2.734015345268542, "percentage": 54.68, "elapsed_time": "5:23:09", "remaining_time": "4:27:50"} +{"current_steps": 1070, "total_steps": 1955, "loss": 1.0414, "lr": 4.0392916014046735e-05, "epoch": 2.7365728900255757, "percentage": 54.73, "elapsed_time": "5:23:27", "remaining_time": "4:27:32"} +{"current_steps": 1071, "total_steps": 1955, "loss": 0.9911, "lr": 4.0321478447874895e-05, "epoch": 2.7391304347826084, "percentage": 54.78, "elapsed_time": "5:23:45", "remaining_time": "4:27:14"} +{"current_steps": 1072, "total_steps": 1955, "loss": 1.0219, "lr": 4.0250039856240476e-05, "epoch": 2.741687979539642, "percentage": 54.83, "elapsed_time": "5:24:03", "remaining_time": "4:26:55"} +{"current_steps": 1073, "total_steps": 1955, "loss": 1.0048, "lr": 4.0178600467020664e-05, "epoch": 2.7442455242966752, "percentage": 54.88, "elapsed_time": "5:24:22", "remaining_time": "4:26:38"} +{"current_steps": 1074, "total_steps": 1955, "loss": 0.9903, "lr": 4.010716050809517e-05, "epoch": 2.7468030690537084, "percentage": 54.94, "elapsed_time": "5:24:40", "remaining_time": "4:26:19"} +{"current_steps": 1075, "total_steps": 1955, "loss": 1.0435, "lr": 4.0035720207345556e-05, "epoch": 2.7493606138107416, "percentage": 54.99, "elapsed_time": "5:24:58", "remaining_time": "4:26:01"} +{"current_steps": 1076, "total_steps": 1955, "loss": 1.0001, "lr": 3.9964279792654464e-05, "epoch": 2.7519181585677748, "percentage": 55.04, "elapsed_time": "5:25:16", "remaining_time": "4:25:43"} +{"current_steps": 1077, "total_steps": 1955, "loss": 1.0211, "lr": 3.989283949190484e-05, "epoch": 2.7544757033248084, "percentage": 55.09, "elapsed_time": "5:25:34", "remaining_time": "4:25:25"} +{"current_steps": 1078, "total_steps": 1955, "loss": 0.9836, "lr": 3.982139953297935e-05, "epoch": 2.7570332480818416, "percentage": 55.14, "elapsed_time": "5:25:52", "remaining_time": "4:25:06"} +{"current_steps": 1079, "total_steps": 1955, "loss": 0.9624, "lr": 3.974996014375954e-05, "epoch": 2.7595907928388748, "percentage": 55.19, "elapsed_time": "5:26:10", "remaining_time": "4:24:48"} +{"current_steps": 1080, "total_steps": 1955, "loss": 1.0087, "lr": 3.967852155212511e-05, "epoch": 2.762148337595908, "percentage": 55.24, "elapsed_time": "5:26:28", "remaining_time": "4:24:30"} +{"current_steps": 1081, "total_steps": 1955, "loss": 0.9716, "lr": 3.9607083985953265e-05, "epoch": 2.764705882352941, "percentage": 55.29, "elapsed_time": "5:26:46", "remaining_time": "4:24:12"} +{"current_steps": 1082, "total_steps": 1955, "loss": 1.009, "lr": 3.95356476731179e-05, "epoch": 2.7672634271099743, "percentage": 55.35, "elapsed_time": "5:27:04", "remaining_time": "4:23:54"} +{"current_steps": 1083, "total_steps": 1955, "loss": 1.0071, "lr": 3.946421284148892e-05, "epoch": 2.7698209718670075, "percentage": 55.4, "elapsed_time": "5:27:22", "remaining_time": "4:23:35"} +{"current_steps": 1084, "total_steps": 1955, "loss": 1.0099, "lr": 3.939277971893152e-05, "epoch": 2.772378516624041, "percentage": 55.45, "elapsed_time": "5:27:40", "remaining_time": "4:23:17"} +{"current_steps": 1085, "total_steps": 1955, "loss": 0.9844, "lr": 3.932134853330542e-05, "epoch": 2.7749360613810743, "percentage": 55.5, "elapsed_time": "5:27:58", "remaining_time": "4:22:59"} +{"current_steps": 1086, "total_steps": 1955, "loss": 1.0384, "lr": 3.92499195124642e-05, "epoch": 2.7774936061381075, "percentage": 55.55, "elapsed_time": "5:28:16", "remaining_time": "4:22:41"} +{"current_steps": 1087, "total_steps": 1955, "loss": 0.9644, "lr": 3.91784928842545e-05, "epoch": 2.7800511508951407, "percentage": 55.6, "elapsed_time": "5:28:34", "remaining_time": "4:22:22"} +{"current_steps": 1088, "total_steps": 1955, "loss": 0.9907, "lr": 3.9107068876515314e-05, "epoch": 2.782608695652174, "percentage": 55.65, "elapsed_time": "5:28:52", "remaining_time": "4:22:04"} +{"current_steps": 1089, "total_steps": 1955, "loss": 1.0117, "lr": 3.9035647717077335e-05, "epoch": 2.785166240409207, "percentage": 55.7, "elapsed_time": "5:29:10", "remaining_time": "4:21:46"} +{"current_steps": 1090, "total_steps": 1955, "loss": 0.9782, "lr": 3.896422963376211e-05, "epoch": 2.78772378516624, "percentage": 55.75, "elapsed_time": "5:29:28", "remaining_time": "4:21:28"} +{"current_steps": 1091, "total_steps": 1955, "loss": 0.9543, "lr": 3.8892814854381425e-05, "epoch": 2.790281329923274, "percentage": 55.81, "elapsed_time": "5:29:46", "remaining_time": "4:21:09"} +{"current_steps": 1092, "total_steps": 1955, "loss": 1.0135, "lr": 3.882140360673648e-05, "epoch": 2.792838874680307, "percentage": 55.86, "elapsed_time": "5:30:04", "remaining_time": "4:20:51"} +{"current_steps": 1093, "total_steps": 1955, "loss": 1.0158, "lr": 3.8749996118617256e-05, "epoch": 2.79539641943734, "percentage": 55.91, "elapsed_time": "5:30:22", "remaining_time": "4:20:33"} +{"current_steps": 1094, "total_steps": 1955, "loss": 0.9991, "lr": 3.86785926178017e-05, "epoch": 2.7979539641943734, "percentage": 55.96, "elapsed_time": "5:30:41", "remaining_time": "4:20:15"} +{"current_steps": 1095, "total_steps": 1955, "loss": 0.9949, "lr": 3.860719333205505e-05, "epoch": 2.8005115089514065, "percentage": 56.01, "elapsed_time": "5:30:59", "remaining_time": "4:19:57"} +{"current_steps": 1096, "total_steps": 1955, "loss": 1.043, "lr": 3.853579848912912e-05, "epoch": 2.80306905370844, "percentage": 56.06, "elapsed_time": "5:31:17", "remaining_time": "4:19:39"} +{"current_steps": 1097, "total_steps": 1955, "loss": 1.0109, "lr": 3.846440831676153e-05, "epoch": 2.805626598465473, "percentage": 56.11, "elapsed_time": "5:31:35", "remaining_time": "4:19:20"} +{"current_steps": 1098, "total_steps": 1955, "loss": 1.0243, "lr": 3.839302304267504e-05, "epoch": 2.8081841432225065, "percentage": 56.16, "elapsed_time": "5:31:53", "remaining_time": "4:19:02"} +{"current_steps": 1099, "total_steps": 1955, "loss": 1.0148, "lr": 3.832164289457671e-05, "epoch": 2.8107416879795397, "percentage": 56.21, "elapsed_time": "5:32:11", "remaining_time": "4:18:44"} +{"current_steps": 1100, "total_steps": 1955, "loss": 0.9899, "lr": 3.8250268100157304e-05, "epoch": 2.813299232736573, "percentage": 56.27, "elapsed_time": "5:32:29", "remaining_time": "4:18:26"} +{"current_steps": 1101, "total_steps": 1955, "loss": 1.0172, "lr": 3.817889888709053e-05, "epoch": 2.815856777493606, "percentage": 56.32, "elapsed_time": "5:32:47", "remaining_time": "4:18:08"} +{"current_steps": 1102, "total_steps": 1955, "loss": 0.999, "lr": 3.8107535483032224e-05, "epoch": 2.8184143222506393, "percentage": 56.37, "elapsed_time": "5:33:06", "remaining_time": "4:17:50"} +{"current_steps": 1103, "total_steps": 1955, "loss": 1.0137, "lr": 3.803617811561976e-05, "epoch": 2.820971867007673, "percentage": 56.42, "elapsed_time": "5:33:24", "remaining_time": "4:17:32"} +{"current_steps": 1104, "total_steps": 1955, "loss": 0.9906, "lr": 3.796482701247121e-05, "epoch": 2.8235294117647056, "percentage": 56.47, "elapsed_time": "5:33:42", "remaining_time": "4:17:14"} +{"current_steps": 1105, "total_steps": 1955, "loss": 0.9807, "lr": 3.789348240118468e-05, "epoch": 2.8260869565217392, "percentage": 56.52, "elapsed_time": "5:34:00", "remaining_time": "4:16:56"} +{"current_steps": 1106, "total_steps": 1955, "loss": 1.0106, "lr": 3.782214450933756e-05, "epoch": 2.8286445012787724, "percentage": 56.57, "elapsed_time": "5:34:18", "remaining_time": "4:16:37"} +{"current_steps": 1107, "total_steps": 1955, "loss": 1.0069, "lr": 3.775081356448583e-05, "epoch": 2.8312020460358056, "percentage": 56.62, "elapsed_time": "5:34:36", "remaining_time": "4:16:19"} +{"current_steps": 1108, "total_steps": 1955, "loss": 1.017, "lr": 3.7679489794163286e-05, "epoch": 2.833759590792839, "percentage": 56.68, "elapsed_time": "5:34:55", "remaining_time": "4:16:01"} +{"current_steps": 1109, "total_steps": 1955, "loss": 1.0209, "lr": 3.760817342588086e-05, "epoch": 2.836317135549872, "percentage": 56.73, "elapsed_time": "5:35:13", "remaining_time": "4:15:43"} +{"current_steps": 1110, "total_steps": 1955, "loss": 1.002, "lr": 3.753686468712583e-05, "epoch": 2.8388746803069056, "percentage": 56.78, "elapsed_time": "5:35:31", "remaining_time": "4:15:25"} +{"current_steps": 1111, "total_steps": 1955, "loss": 1.0018, "lr": 3.746556380536119e-05, "epoch": 2.8414322250639388, "percentage": 56.83, "elapsed_time": "5:35:49", "remaining_time": "4:15:07"} +{"current_steps": 1112, "total_steps": 1955, "loss": 1.0102, "lr": 3.7394271008024834e-05, "epoch": 2.843989769820972, "percentage": 56.88, "elapsed_time": "5:36:07", "remaining_time": "4:14:48"} +{"current_steps": 1113, "total_steps": 1955, "loss": 0.9998, "lr": 3.7322986522528896e-05, "epoch": 2.846547314578005, "percentage": 56.93, "elapsed_time": "5:36:26", "remaining_time": "4:14:31"} +{"current_steps": 1114, "total_steps": 1955, "loss": 1.0001, "lr": 3.7251710576258966e-05, "epoch": 2.8491048593350383, "percentage": 56.98, "elapsed_time": "5:36:44", "remaining_time": "4:14:12"} +{"current_steps": 1115, "total_steps": 1955, "loss": 1.0051, "lr": 3.7180443396573445e-05, "epoch": 2.8516624040920715, "percentage": 57.03, "elapsed_time": "5:37:02", "remaining_time": "4:13:54"} +{"current_steps": 1116, "total_steps": 1955, "loss": 0.997, "lr": 3.71091852108027e-05, "epoch": 2.8542199488491047, "percentage": 57.08, "elapsed_time": "5:37:20", "remaining_time": "4:13:36"} +{"current_steps": 1117, "total_steps": 1955, "loss": 0.9823, "lr": 3.7037936246248455e-05, "epoch": 2.8567774936061383, "percentage": 57.14, "elapsed_time": "5:37:38", "remaining_time": "4:13:18"} +{"current_steps": 1118, "total_steps": 1955, "loss": 1.014, "lr": 3.696669673018302e-05, "epoch": 2.8593350383631715, "percentage": 57.19, "elapsed_time": "5:37:56", "remaining_time": "4:12:59"} +{"current_steps": 1119, "total_steps": 1955, "loss": 0.9867, "lr": 3.689546688984854e-05, "epoch": 2.8618925831202047, "percentage": 57.24, "elapsed_time": "5:38:14", "remaining_time": "4:12:41"} +{"current_steps": 1120, "total_steps": 1955, "loss": 1.0419, "lr": 3.6824246952456346e-05, "epoch": 2.864450127877238, "percentage": 57.29, "elapsed_time": "5:38:32", "remaining_time": "4:12:23"} +{"current_steps": 1121, "total_steps": 1955, "loss": 1.0103, "lr": 3.675303714518611e-05, "epoch": 2.867007672634271, "percentage": 57.34, "elapsed_time": "5:38:50", "remaining_time": "4:12:05"} +{"current_steps": 1122, "total_steps": 1955, "loss": 0.9924, "lr": 3.6681837695185235e-05, "epoch": 2.869565217391304, "percentage": 57.39, "elapsed_time": "5:39:08", "remaining_time": "4:11:47"} +{"current_steps": 1123, "total_steps": 1955, "loss": 0.9621, "lr": 3.66106488295681e-05, "epoch": 2.8721227621483374, "percentage": 57.44, "elapsed_time": "5:39:26", "remaining_time": "4:11:28"} +{"current_steps": 1124, "total_steps": 1955, "loss": 1.0429, "lr": 3.653947077541527e-05, "epoch": 2.874680306905371, "percentage": 57.49, "elapsed_time": "5:39:44", "remaining_time": "4:11:10"} +{"current_steps": 1125, "total_steps": 1955, "loss": 1.0197, "lr": 3.646830375977288e-05, "epoch": 2.877237851662404, "percentage": 57.54, "elapsed_time": "5:40:02", "remaining_time": "4:10:52"} +{"current_steps": 1126, "total_steps": 1955, "loss": 1.0112, "lr": 3.639714800965183e-05, "epoch": 2.8797953964194374, "percentage": 57.6, "elapsed_time": "5:40:20", "remaining_time": "4:10:34"} +{"current_steps": 1127, "total_steps": 1955, "loss": 0.9783, "lr": 3.632600375202704e-05, "epoch": 2.8823529411764706, "percentage": 57.65, "elapsed_time": "5:40:38", "remaining_time": "4:10:16"} +{"current_steps": 1128, "total_steps": 1955, "loss": 0.9839, "lr": 3.625487121383686e-05, "epoch": 2.8849104859335037, "percentage": 57.7, "elapsed_time": "5:40:56", "remaining_time": "4:09:57"} +{"current_steps": 1129, "total_steps": 1955, "loss": 0.9899, "lr": 3.6183750621982176e-05, "epoch": 2.887468030690537, "percentage": 57.75, "elapsed_time": "5:41:14", "remaining_time": "4:09:39"} +{"current_steps": 1130, "total_steps": 1955, "loss": 1.0155, "lr": 3.611264220332583e-05, "epoch": 2.89002557544757, "percentage": 57.8, "elapsed_time": "5:41:32", "remaining_time": "4:09:21"} +{"current_steps": 1131, "total_steps": 1955, "loss": 1.0281, "lr": 3.60415461846918e-05, "epoch": 2.8925831202046037, "percentage": 57.85, "elapsed_time": "5:41:50", "remaining_time": "4:09:03"} +{"current_steps": 1132, "total_steps": 1955, "loss": 0.9653, "lr": 3.5970462792864486e-05, "epoch": 2.895140664961637, "percentage": 57.9, "elapsed_time": "5:42:08", "remaining_time": "4:08:44"} +{"current_steps": 1133, "total_steps": 1955, "loss": 0.9844, "lr": 3.589939225458807e-05, "epoch": 2.89769820971867, "percentage": 57.95, "elapsed_time": "5:42:26", "remaining_time": "4:08:26"} +{"current_steps": 1134, "total_steps": 1955, "loss": 0.9857, "lr": 3.582833479656568e-05, "epoch": 2.9002557544757033, "percentage": 58.01, "elapsed_time": "5:42:44", "remaining_time": "4:08:08"} +{"current_steps": 1135, "total_steps": 1955, "loss": 1.01, "lr": 3.575729064545876e-05, "epoch": 2.9028132992327365, "percentage": 58.06, "elapsed_time": "5:43:02", "remaining_time": "4:07:50"} +{"current_steps": 1136, "total_steps": 1955, "loss": 0.9812, "lr": 3.568626002788628e-05, "epoch": 2.90537084398977, "percentage": 58.11, "elapsed_time": "5:43:20", "remaining_time": "4:07:32"} +{"current_steps": 1137, "total_steps": 1955, "loss": 1.0438, "lr": 3.5615243170424074e-05, "epoch": 2.907928388746803, "percentage": 58.16, "elapsed_time": "5:43:38", "remaining_time": "4:07:13"} +{"current_steps": 1138, "total_steps": 1955, "loss": 0.9827, "lr": 3.5544240299604024e-05, "epoch": 2.9104859335038364, "percentage": 58.21, "elapsed_time": "5:43:57", "remaining_time": "4:06:56"} +{"current_steps": 1139, "total_steps": 1955, "loss": 0.9976, "lr": 3.547325164191345e-05, "epoch": 2.9130434782608696, "percentage": 58.26, "elapsed_time": "5:44:15", "remaining_time": "4:06:37"} +{"current_steps": 1140, "total_steps": 1955, "loss": 0.9897, "lr": 3.5402277423794316e-05, "epoch": 2.915601023017903, "percentage": 58.31, "elapsed_time": "5:44:33", "remaining_time": "4:06:19"} +{"current_steps": 1141, "total_steps": 1955, "loss": 1.0057, "lr": 3.5331317871642525e-05, "epoch": 2.918158567774936, "percentage": 58.36, "elapsed_time": "5:44:52", "remaining_time": "4:06:01"} +{"current_steps": 1142, "total_steps": 1955, "loss": 1.0152, "lr": 3.526037321180722e-05, "epoch": 2.920716112531969, "percentage": 58.41, "elapsed_time": "5:45:10", "remaining_time": "4:05:43"} +{"current_steps": 1143, "total_steps": 1955, "loss": 1.0117, "lr": 3.518944367058999e-05, "epoch": 2.923273657289003, "percentage": 58.47, "elapsed_time": "5:45:28", "remaining_time": "4:05:25"} +{"current_steps": 1144, "total_steps": 1955, "loss": 0.99, "lr": 3.511852947424424e-05, "epoch": 2.9258312020460355, "percentage": 58.52, "elapsed_time": "5:45:46", "remaining_time": "4:05:07"} +{"current_steps": 1145, "total_steps": 1955, "loss": 0.9982, "lr": 3.504763084897442e-05, "epoch": 2.928388746803069, "percentage": 58.57, "elapsed_time": "5:46:04", "remaining_time": "4:04:49"} +{"current_steps": 1146, "total_steps": 1955, "loss": 0.9852, "lr": 3.4976748020935305e-05, "epoch": 2.9309462915601023, "percentage": 58.62, "elapsed_time": "5:46:22", "remaining_time": "4:04:30"} +{"current_steps": 1147, "total_steps": 1955, "loss": 1.0029, "lr": 3.4905881216231284e-05, "epoch": 2.9335038363171355, "percentage": 58.67, "elapsed_time": "5:46:40", "remaining_time": "4:04:12"} +{"current_steps": 1148, "total_steps": 1955, "loss": 1.0008, "lr": 3.4835030660915634e-05, "epoch": 2.9360613810741687, "percentage": 58.72, "elapsed_time": "5:46:59", "remaining_time": "4:03:55"} +{"current_steps": 1149, "total_steps": 1955, "loss": 1.0166, "lr": 3.476419658098978e-05, "epoch": 2.938618925831202, "percentage": 58.77, "elapsed_time": "5:47:17", "remaining_time": "4:03:36"} +{"current_steps": 1150, "total_steps": 1955, "loss": 1.0058, "lr": 3.469337920240263e-05, "epoch": 2.9411764705882355, "percentage": 58.82, "elapsed_time": "5:47:36", "remaining_time": "4:03:19"} +{"current_steps": 1151, "total_steps": 1955, "loss": 1.0211, "lr": 3.462257875104977e-05, "epoch": 2.9437340153452687, "percentage": 58.87, "elapsed_time": "5:47:53", "remaining_time": "4:03:00"} +{"current_steps": 1152, "total_steps": 1955, "loss": 0.9922, "lr": 3.455179545277286e-05, "epoch": 2.946291560102302, "percentage": 58.93, "elapsed_time": "5:48:11", "remaining_time": "4:02:42"} +{"current_steps": 1153, "total_steps": 1955, "loss": 1.0137, "lr": 3.4481029533358775e-05, "epoch": 2.948849104859335, "percentage": 58.98, "elapsed_time": "5:48:29", "remaining_time": "4:02:24"} +{"current_steps": 1154, "total_steps": 1955, "loss": 1.0183, "lr": 3.441028121853898e-05, "epoch": 2.9514066496163682, "percentage": 59.03, "elapsed_time": "5:48:47", "remaining_time": "4:02:06"} +{"current_steps": 1155, "total_steps": 1955, "loss": 1.0343, "lr": 3.43395507339888e-05, "epoch": 2.9539641943734014, "percentage": 59.08, "elapsed_time": "5:49:05", "remaining_time": "4:01:47"} +{"current_steps": 1156, "total_steps": 1955, "loss": 1.0078, "lr": 3.426883830532665e-05, "epoch": 2.9565217391304346, "percentage": 59.13, "elapsed_time": "5:49:23", "remaining_time": "4:01:29"} +{"current_steps": 1157, "total_steps": 1955, "loss": 0.9989, "lr": 3.4198144158113405e-05, "epoch": 2.959079283887468, "percentage": 59.18, "elapsed_time": "5:49:41", "remaining_time": "4:01:11"} +{"current_steps": 1158, "total_steps": 1955, "loss": 0.9938, "lr": 3.412746851785156e-05, "epoch": 2.9616368286445014, "percentage": 59.23, "elapsed_time": "5:49:59", "remaining_time": "4:00:53"} +{"current_steps": 1159, "total_steps": 1955, "loss": 0.9934, "lr": 3.405681160998463e-05, "epoch": 2.9641943734015346, "percentage": 59.28, "elapsed_time": "5:50:18", "remaining_time": "4:00:35"} +{"current_steps": 1160, "total_steps": 1955, "loss": 1.015, "lr": 3.3986173659896335e-05, "epoch": 2.9667519181585678, "percentage": 59.34, "elapsed_time": "5:50:36", "remaining_time": "4:00:17"} +{"current_steps": 1161, "total_steps": 1955, "loss": 0.985, "lr": 3.3915554892909944e-05, "epoch": 2.969309462915601, "percentage": 59.39, "elapsed_time": "5:50:54", "remaining_time": "3:59:59"} +{"current_steps": 1162, "total_steps": 1955, "loss": 0.9894, "lr": 3.384495553428754e-05, "epoch": 2.971867007672634, "percentage": 59.44, "elapsed_time": "5:51:12", "remaining_time": "3:59:40"} +{"current_steps": 1163, "total_steps": 1955, "loss": 1.0311, "lr": 3.3774375809229275e-05, "epoch": 2.9744245524296673, "percentage": 59.49, "elapsed_time": "5:51:29", "remaining_time": "3:59:22"} +{"current_steps": 1164, "total_steps": 1955, "loss": 0.985, "lr": 3.370381594287273e-05, "epoch": 2.976982097186701, "percentage": 59.54, "elapsed_time": "5:51:47", "remaining_time": "3:59:03"} +{"current_steps": 1165, "total_steps": 1955, "loss": 0.9935, "lr": 3.3633276160292044e-05, "epoch": 2.979539641943734, "percentage": 59.59, "elapsed_time": "5:52:05", "remaining_time": "3:58:45"} +{"current_steps": 1166, "total_steps": 1955, "loss": 1.0254, "lr": 3.3562756686497364e-05, "epoch": 2.9820971867007673, "percentage": 59.64, "elapsed_time": "5:52:23", "remaining_time": "3:58:27"} +{"current_steps": 1167, "total_steps": 1955, "loss": 1.0012, "lr": 3.349225774643405e-05, "epoch": 2.9846547314578005, "percentage": 59.69, "elapsed_time": "5:52:41", "remaining_time": "3:58:09"} +{"current_steps": 1168, "total_steps": 1955, "loss": 1.0156, "lr": 3.342177956498193e-05, "epoch": 2.9872122762148337, "percentage": 59.74, "elapsed_time": "5:53:00", "remaining_time": "3:57:51"} +{"current_steps": 1169, "total_steps": 1955, "loss": 0.9923, "lr": 3.3351322366954656e-05, "epoch": 2.9897698209718673, "percentage": 59.8, "elapsed_time": "5:53:18", "remaining_time": "3:57:32"} +{"current_steps": 1170, "total_steps": 1955, "loss": 0.9903, "lr": 3.328088637709892e-05, "epoch": 2.9923273657289, "percentage": 59.85, "elapsed_time": "5:53:36", "remaining_time": "3:57:14"} +{"current_steps": 1171, "total_steps": 1955, "loss": 0.9872, "lr": 3.321047182009375e-05, "epoch": 2.9948849104859336, "percentage": 59.9, "elapsed_time": "5:53:54", "remaining_time": "3:56:56"} +{"current_steps": 1172, "total_steps": 1955, "loss": 1.0294, "lr": 3.314007892054985e-05, "epoch": 2.997442455242967, "percentage": 59.95, "elapsed_time": "5:54:13", "remaining_time": "3:56:39"} +{"current_steps": 1173, "total_steps": 1955, "loss": 1.0199, "lr": 3.30697079030088e-05, "epoch": 3.0, "percentage": 60.0, "elapsed_time": "5:54:31", "remaining_time": "3:56:20"} +{"current_steps": 1174, "total_steps": 1955, "loss": 1.0065, "lr": 3.29993589919424e-05, "epoch": 3.002557544757033, "percentage": 60.05, "elapsed_time": "5:55:00", "remaining_time": "3:56:10"} +{"current_steps": 1175, "total_steps": 1955, "loss": 1.0053, "lr": 3.292903241175195e-05, "epoch": 3.0051150895140664, "percentage": 60.1, "elapsed_time": "5:55:18", "remaining_time": "3:55:52"} +{"current_steps": 1176, "total_steps": 1955, "loss": 0.9849, "lr": 3.285872838676746e-05, "epoch": 3.0076726342710995, "percentage": 60.15, "elapsed_time": "5:55:36", "remaining_time": "3:55:33"} +{"current_steps": 1177, "total_steps": 1955, "loss": 1.0211, "lr": 3.278844714124707e-05, "epoch": 3.010230179028133, "percentage": 60.2, "elapsed_time": "5:55:54", "remaining_time": "3:55:15"} +{"current_steps": 1178, "total_steps": 1955, "loss": 1.0096, "lr": 3.2718188899376183e-05, "epoch": 3.0127877237851663, "percentage": 60.26, "elapsed_time": "5:56:12", "remaining_time": "3:54:57"} +{"current_steps": 1179, "total_steps": 1955, "loss": 0.9608, "lr": 3.264795388526689e-05, "epoch": 3.0153452685421995, "percentage": 60.31, "elapsed_time": "5:56:31", "remaining_time": "3:54:39"} +{"current_steps": 1180, "total_steps": 1955, "loss": 0.9707, "lr": 3.257774232295714e-05, "epoch": 3.0179028132992327, "percentage": 60.36, "elapsed_time": "5:56:48", "remaining_time": "3:54:20"} +{"current_steps": 1181, "total_steps": 1955, "loss": 0.9833, "lr": 3.250755443641012e-05, "epoch": 3.020460358056266, "percentage": 60.41, "elapsed_time": "5:57:06", "remaining_time": "3:54:02"} +{"current_steps": 1182, "total_steps": 1955, "loss": 0.9764, "lr": 3.243739044951343e-05, "epoch": 3.023017902813299, "percentage": 60.46, "elapsed_time": "5:57:24", "remaining_time": "3:53:44"} +{"current_steps": 1183, "total_steps": 1955, "loss": 0.9979, "lr": 3.2367250586078496e-05, "epoch": 3.0255754475703327, "percentage": 60.51, "elapsed_time": "5:57:42", "remaining_time": "3:53:25"} +{"current_steps": 1184, "total_steps": 1955, "loss": 0.9648, "lr": 3.2297135069839766e-05, "epoch": 3.028132992327366, "percentage": 60.56, "elapsed_time": "5:58:00", "remaining_time": "3:53:07"} +{"current_steps": 1185, "total_steps": 1955, "loss": 0.979, "lr": 3.2227044124454024e-05, "epoch": 3.030690537084399, "percentage": 60.61, "elapsed_time": "5:58:19", "remaining_time": "3:52:49"} +{"current_steps": 1186, "total_steps": 1955, "loss": 0.9618, "lr": 3.215697797349972e-05, "epoch": 3.0332480818414322, "percentage": 60.66, "elapsed_time": "5:58:37", "remaining_time": "3:52:31"} +{"current_steps": 1187, "total_steps": 1955, "loss": 0.9852, "lr": 3.2086936840476124e-05, "epoch": 3.0358056265984654, "percentage": 60.72, "elapsed_time": "5:58:55", "remaining_time": "3:52:13"} +{"current_steps": 1188, "total_steps": 1955, "loss": 0.9512, "lr": 3.201692094880278e-05, "epoch": 3.0383631713554986, "percentage": 60.77, "elapsed_time": "5:59:13", "remaining_time": "3:51:55"} +{"current_steps": 1189, "total_steps": 1955, "loss": 0.9841, "lr": 3.194693052181869e-05, "epoch": 3.040920716112532, "percentage": 60.82, "elapsed_time": "5:59:31", "remaining_time": "3:51:36"} +{"current_steps": 1190, "total_steps": 1955, "loss": 0.9725, "lr": 3.1876965782781626e-05, "epoch": 3.0434782608695654, "percentage": 60.87, "elapsed_time": "5:59:49", "remaining_time": "3:51:18"} +{"current_steps": 1191, "total_steps": 1955, "loss": 0.9523, "lr": 3.1807026954867445e-05, "epoch": 3.0460358056265986, "percentage": 60.92, "elapsed_time": "6:00:06", "remaining_time": "3:51:00"} +{"current_steps": 1192, "total_steps": 1955, "loss": 0.9898, "lr": 3.173711426116931e-05, "epoch": 3.0485933503836318, "percentage": 60.97, "elapsed_time": "6:00:24", "remaining_time": "3:50:42"} +{"current_steps": 1193, "total_steps": 1955, "loss": 0.9975, "lr": 3.1667227924697033e-05, "epoch": 3.051150895140665, "percentage": 61.02, "elapsed_time": "6:00:42", "remaining_time": "3:50:23"} +{"current_steps": 1194, "total_steps": 1955, "loss": 0.9695, "lr": 3.1597368168376376e-05, "epoch": 3.053708439897698, "percentage": 61.07, "elapsed_time": "6:01:01", "remaining_time": "3:50:05"} +{"current_steps": 1195, "total_steps": 1955, "loss": 0.9757, "lr": 3.1527535215048284e-05, "epoch": 3.0562659846547313, "percentage": 61.13, "elapsed_time": "6:01:19", "remaining_time": "3:49:47"} +{"current_steps": 1196, "total_steps": 1955, "loss": 0.9874, "lr": 3.1457729287468225e-05, "epoch": 3.0588235294117645, "percentage": 61.18, "elapsed_time": "6:01:37", "remaining_time": "3:49:29"} +{"current_steps": 1197, "total_steps": 1955, "loss": 0.9883, "lr": 3.138795060830549e-05, "epoch": 3.061381074168798, "percentage": 61.23, "elapsed_time": "6:01:55", "remaining_time": "3:49:11"} +{"current_steps": 1198, "total_steps": 1955, "loss": 0.9853, "lr": 3.131819940014234e-05, "epoch": 3.0639386189258313, "percentage": 61.28, "elapsed_time": "6:02:13", "remaining_time": "3:48:53"} +{"current_steps": 1199, "total_steps": 1955, "loss": 1.0029, "lr": 3.1248475885473534e-05, "epoch": 3.0664961636828645, "percentage": 61.33, "elapsed_time": "6:02:31", "remaining_time": "3:48:35"} +{"current_steps": 1200, "total_steps": 1955, "loss": 0.9627, "lr": 3.117878028670542e-05, "epoch": 3.0690537084398977, "percentage": 61.38, "elapsed_time": "6:02:49", "remaining_time": "3:48:16"} +{"current_steps": 1201, "total_steps": 1955, "loss": 0.9749, "lr": 3.110911282615533e-05, "epoch": 3.071611253196931, "percentage": 61.43, "elapsed_time": "6:03:07", "remaining_time": "3:47:58"} +{"current_steps": 1202, "total_steps": 1955, "loss": 0.9971, "lr": 3.103947372605084e-05, "epoch": 3.074168797953964, "percentage": 61.48, "elapsed_time": "6:03:25", "remaining_time": "3:47:40"} +{"current_steps": 1203, "total_steps": 1955, "loss": 1.0029, "lr": 3.0969863208529044e-05, "epoch": 3.0767263427109977, "percentage": 61.53, "elapsed_time": "6:03:43", "remaining_time": "3:47:22"} +{"current_steps": 1204, "total_steps": 1955, "loss": 1.0025, "lr": 3.090028149563587e-05, "epoch": 3.079283887468031, "percentage": 61.59, "elapsed_time": "6:04:01", "remaining_time": "3:47:03"} +{"current_steps": 1205, "total_steps": 1955, "loss": 1.0031, "lr": 3.083072880932535e-05, "epoch": 3.081841432225064, "percentage": 61.64, "elapsed_time": "6:04:19", "remaining_time": "3:46:45"} +{"current_steps": 1206, "total_steps": 1955, "loss": 0.974, "lr": 3.0761205371458953e-05, "epoch": 3.084398976982097, "percentage": 61.69, "elapsed_time": "6:04:38", "remaining_time": "3:46:27"} +{"current_steps": 1207, "total_steps": 1955, "loss": 0.9952, "lr": 3.069171140380485e-05, "epoch": 3.0869565217391304, "percentage": 61.74, "elapsed_time": "6:04:55", "remaining_time": "3:46:09"} +{"current_steps": 1208, "total_steps": 1955, "loss": 1.0226, "lr": 3.06222471280372e-05, "epoch": 3.0895140664961636, "percentage": 61.79, "elapsed_time": "6:05:14", "remaining_time": "3:45:51"} +{"current_steps": 1209, "total_steps": 1955, "loss": 0.9692, "lr": 3.055281276573539e-05, "epoch": 3.0920716112531967, "percentage": 61.84, "elapsed_time": "6:05:32", "remaining_time": "3:45:33"} +{"current_steps": 1210, "total_steps": 1955, "loss": 0.9928, "lr": 3.04834085383835e-05, "epoch": 3.0946291560102304, "percentage": 61.89, "elapsed_time": "6:05:51", "remaining_time": "3:45:15"} +{"current_steps": 1211, "total_steps": 1955, "loss": 0.9933, "lr": 3.04140346673694e-05, "epoch": 3.0971867007672635, "percentage": 61.94, "elapsed_time": "6:06:09", "remaining_time": "3:44:57"} +{"current_steps": 1212, "total_steps": 1955, "loss": 0.9588, "lr": 3.034469137398417e-05, "epoch": 3.0997442455242967, "percentage": 61.99, "elapsed_time": "6:06:27", "remaining_time": "3:44:38"} +{"current_steps": 1213, "total_steps": 1955, "loss": 1.0255, "lr": 3.0275378879421313e-05, "epoch": 3.10230179028133, "percentage": 62.05, "elapsed_time": "6:06:45", "remaining_time": "3:44:20"} +{"current_steps": 1214, "total_steps": 1955, "loss": 0.9745, "lr": 3.0206097404776158e-05, "epoch": 3.104859335038363, "percentage": 62.1, "elapsed_time": "6:07:03", "remaining_time": "3:44:02"} +{"current_steps": 1215, "total_steps": 1955, "loss": 1.0162, "lr": 3.0136847171044996e-05, "epoch": 3.1074168797953963, "percentage": 62.15, "elapsed_time": "6:07:21", "remaining_time": "3:43:44"} +{"current_steps": 1216, "total_steps": 1955, "loss": 0.9678, "lr": 3.0067628399124523e-05, "epoch": 3.10997442455243, "percentage": 62.2, "elapsed_time": "6:07:40", "remaining_time": "3:43:26"} +{"current_steps": 1217, "total_steps": 1955, "loss": 1.0046, "lr": 2.999844130981108e-05, "epoch": 3.112531969309463, "percentage": 62.25, "elapsed_time": "6:07:58", "remaining_time": "3:43:08"} +{"current_steps": 1218, "total_steps": 1955, "loss": 0.9786, "lr": 2.9929286123799906e-05, "epoch": 3.1150895140664963, "percentage": 62.3, "elapsed_time": "6:08:16", "remaining_time": "3:42:50"} +{"current_steps": 1219, "total_steps": 1955, "loss": 0.9952, "lr": 2.986016306168453e-05, "epoch": 3.1176470588235294, "percentage": 62.35, "elapsed_time": "6:08:34", "remaining_time": "3:42:31"} +{"current_steps": 1220, "total_steps": 1955, "loss": 0.9832, "lr": 2.9791072343955946e-05, "epoch": 3.1202046035805626, "percentage": 62.4, "elapsed_time": "6:08:52", "remaining_time": "3:42:13"} +{"current_steps": 1221, "total_steps": 1955, "loss": 0.9803, "lr": 2.972201419100202e-05, "epoch": 3.122762148337596, "percentage": 62.46, "elapsed_time": "6:09:10", "remaining_time": "3:41:55"} +{"current_steps": 1222, "total_steps": 1955, "loss": 0.9296, "lr": 2.9652988823106736e-05, "epoch": 3.125319693094629, "percentage": 62.51, "elapsed_time": "6:09:28", "remaining_time": "3:41:37"} +{"current_steps": 1223, "total_steps": 1955, "loss": 0.9911, "lr": 2.9583996460449474e-05, "epoch": 3.1278772378516626, "percentage": 62.56, "elapsed_time": "6:09:45", "remaining_time": "3:41:18"} +{"current_steps": 1224, "total_steps": 1955, "loss": 0.9603, "lr": 2.9515037323104373e-05, "epoch": 3.130434782608696, "percentage": 62.61, "elapsed_time": "6:10:04", "remaining_time": "3:41:00"} +{"current_steps": 1225, "total_steps": 1955, "loss": 0.9667, "lr": 2.9446111631039572e-05, "epoch": 3.132992327365729, "percentage": 62.66, "elapsed_time": "6:10:21", "remaining_time": "3:40:42"} +{"current_steps": 1226, "total_steps": 1955, "loss": 0.9842, "lr": 2.9377219604116484e-05, "epoch": 3.135549872122762, "percentage": 62.71, "elapsed_time": "6:10:39", "remaining_time": "3:40:24"} +{"current_steps": 1227, "total_steps": 1955, "loss": 0.9838, "lr": 2.930836146208921e-05, "epoch": 3.1381074168797953, "percentage": 62.76, "elapsed_time": "6:10:58", "remaining_time": "3:40:06"} +{"current_steps": 1228, "total_steps": 1955, "loss": 1.0012, "lr": 2.92395374246037e-05, "epoch": 3.1406649616368285, "percentage": 62.81, "elapsed_time": "6:11:16", "remaining_time": "3:39:47"} +{"current_steps": 1229, "total_steps": 1955, "loss": 1.0068, "lr": 2.9170747711197154e-05, "epoch": 3.1432225063938617, "percentage": 62.86, "elapsed_time": "6:11:34", "remaining_time": "3:39:29"} +{"current_steps": 1230, "total_steps": 1955, "loss": 0.9905, "lr": 2.9101992541297288e-05, "epoch": 3.1457800511508953, "percentage": 62.92, "elapsed_time": "6:11:52", "remaining_time": "3:39:11"} +{"current_steps": 1231, "total_steps": 1955, "loss": 0.999, "lr": 2.9033272134221575e-05, "epoch": 3.1483375959079285, "percentage": 62.97, "elapsed_time": "6:12:10", "remaining_time": "3:38:53"} +{"current_steps": 1232, "total_steps": 1955, "loss": 0.9777, "lr": 2.8964586709176674e-05, "epoch": 3.1508951406649617, "percentage": 63.02, "elapsed_time": "6:12:28", "remaining_time": "3:38:35"} +{"current_steps": 1233, "total_steps": 1955, "loss": 1.0168, "lr": 2.88959364852576e-05, "epoch": 3.153452685421995, "percentage": 63.07, "elapsed_time": "6:12:46", "remaining_time": "3:38:16"} +{"current_steps": 1234, "total_steps": 1955, "loss": 0.9994, "lr": 2.882732168144712e-05, "epoch": 3.156010230179028, "percentage": 63.12, "elapsed_time": "6:13:04", "remaining_time": "3:37:58"} +{"current_steps": 1235, "total_steps": 1955, "loss": 0.9893, "lr": 2.8758742516614994e-05, "epoch": 3.1585677749360612, "percentage": 63.17, "elapsed_time": "6:13:22", "remaining_time": "3:37:40"} +{"current_steps": 1236, "total_steps": 1955, "loss": 0.9934, "lr": 2.8690199209517322e-05, "epoch": 3.1611253196930944, "percentage": 63.22, "elapsed_time": "6:13:40", "remaining_time": "3:37:22"} +{"current_steps": 1237, "total_steps": 1955, "loss": 0.9822, "lr": 2.8621691978795793e-05, "epoch": 3.163682864450128, "percentage": 63.27, "elapsed_time": "6:13:58", "remaining_time": "3:37:03"} +{"current_steps": 1238, "total_steps": 1955, "loss": 0.9868, "lr": 2.8553221042977022e-05, "epoch": 3.166240409207161, "percentage": 63.32, "elapsed_time": "6:14:16", "remaining_time": "3:36:45"} +{"current_steps": 1239, "total_steps": 1955, "loss": 0.9566, "lr": 2.84847866204719e-05, "epoch": 3.1687979539641944, "percentage": 63.38, "elapsed_time": "6:14:34", "remaining_time": "3:36:27"} +{"current_steps": 1240, "total_steps": 1955, "loss": 0.9772, "lr": 2.8416388929574768e-05, "epoch": 3.1713554987212276, "percentage": 63.43, "elapsed_time": "6:14:52", "remaining_time": "3:36:09"} +{"current_steps": 1241, "total_steps": 1955, "loss": 0.9663, "lr": 2.8348028188462876e-05, "epoch": 3.1739130434782608, "percentage": 63.48, "elapsed_time": "6:15:10", "remaining_time": "3:35:51"} +{"current_steps": 1242, "total_steps": 1955, "loss": 0.9727, "lr": 2.8279704615195554e-05, "epoch": 3.176470588235294, "percentage": 63.53, "elapsed_time": "6:15:28", "remaining_time": "3:35:32"} +{"current_steps": 1243, "total_steps": 1955, "loss": 0.9812, "lr": 2.8211418427713584e-05, "epoch": 3.1790281329923276, "percentage": 63.58, "elapsed_time": "6:15:46", "remaining_time": "3:35:14"} +{"current_steps": 1244, "total_steps": 1955, "loss": 0.9663, "lr": 2.8143169843838522e-05, "epoch": 3.1815856777493607, "percentage": 63.63, "elapsed_time": "6:16:04", "remaining_time": "3:34:56"} +{"current_steps": 1245, "total_steps": 1955, "loss": 0.944, "lr": 2.807495908127194e-05, "epoch": 3.184143222506394, "percentage": 63.68, "elapsed_time": "6:16:22", "remaining_time": "3:34:38"} +{"current_steps": 1246, "total_steps": 1955, "loss": 0.9938, "lr": 2.8006786357594795e-05, "epoch": 3.186700767263427, "percentage": 63.73, "elapsed_time": "6:16:40", "remaining_time": "3:34:20"} +{"current_steps": 1247, "total_steps": 1955, "loss": 1.0008, "lr": 2.7938651890266695e-05, "epoch": 3.1892583120204603, "percentage": 63.79, "elapsed_time": "6:16:58", "remaining_time": "3:34:02"} +{"current_steps": 1248, "total_steps": 1955, "loss": 0.9826, "lr": 2.7870555896625196e-05, "epoch": 3.1918158567774935, "percentage": 63.84, "elapsed_time": "6:17:16", "remaining_time": "3:33:43"} +{"current_steps": 1249, "total_steps": 1955, "loss": 0.9715, "lr": 2.7802498593885155e-05, "epoch": 3.1943734015345266, "percentage": 63.89, "elapsed_time": "6:17:34", "remaining_time": "3:33:25"} +{"current_steps": 1250, "total_steps": 1955, "loss": 0.981, "lr": 2.7734480199138007e-05, "epoch": 3.1969309462915603, "percentage": 63.94, "elapsed_time": "6:17:52", "remaining_time": "3:33:07"} +{"current_steps": 1251, "total_steps": 1955, "loss": 0.9591, "lr": 2.7666500929351082e-05, "epoch": 3.1994884910485935, "percentage": 63.99, "elapsed_time": "6:18:10", "remaining_time": "3:32:49"} +{"current_steps": 1252, "total_steps": 1955, "loss": 0.9972, "lr": 2.7598561001366905e-05, "epoch": 3.2020460358056266, "percentage": 64.04, "elapsed_time": "6:18:28", "remaining_time": "3:32:30"} +{"current_steps": 1253, "total_steps": 1955, "loss": 0.9543, "lr": 2.7530660631902467e-05, "epoch": 3.20460358056266, "percentage": 64.09, "elapsed_time": "6:18:46", "remaining_time": "3:32:12"} +{"current_steps": 1254, "total_steps": 1955, "loss": 0.9787, "lr": 2.7462800037548643e-05, "epoch": 3.207161125319693, "percentage": 64.14, "elapsed_time": "6:19:04", "remaining_time": "3:31:54"} +{"current_steps": 1255, "total_steps": 1955, "loss": 0.9786, "lr": 2.7394979434769378e-05, "epoch": 3.209718670076726, "percentage": 64.19, "elapsed_time": "6:19:22", "remaining_time": "3:31:36"} +{"current_steps": 1256, "total_steps": 1955, "loss": 0.9703, "lr": 2.7327199039901092e-05, "epoch": 3.21227621483376, "percentage": 64.25, "elapsed_time": "6:19:40", "remaining_time": "3:31:18"} +{"current_steps": 1257, "total_steps": 1955, "loss": 0.9748, "lr": 2.725945906915191e-05, "epoch": 3.214833759590793, "percentage": 64.3, "elapsed_time": "6:19:58", "remaining_time": "3:30:59"} +{"current_steps": 1258, "total_steps": 1955, "loss": 0.9708, "lr": 2.7191759738601052e-05, "epoch": 3.217391304347826, "percentage": 64.35, "elapsed_time": "6:20:16", "remaining_time": "3:30:41"} +{"current_steps": 1259, "total_steps": 1955, "loss": 1.0167, "lr": 2.712410126419805e-05, "epoch": 3.2199488491048593, "percentage": 64.4, "elapsed_time": "6:20:34", "remaining_time": "3:30:23"} +{"current_steps": 1260, "total_steps": 1955, "loss": 0.9843, "lr": 2.705648386176215e-05, "epoch": 3.2225063938618925, "percentage": 64.45, "elapsed_time": "6:20:52", "remaining_time": "3:30:05"} +{"current_steps": 1261, "total_steps": 1955, "loss": 0.9679, "lr": 2.698890774698158e-05, "epoch": 3.2250639386189257, "percentage": 64.5, "elapsed_time": "6:21:10", "remaining_time": "3:29:46"} +{"current_steps": 1262, "total_steps": 1955, "loss": 0.9504, "lr": 2.6921373135412852e-05, "epoch": 3.227621483375959, "percentage": 64.55, "elapsed_time": "6:21:28", "remaining_time": "3:29:28"} +{"current_steps": 1263, "total_steps": 1955, "loss": 0.9719, "lr": 2.685388024248013e-05, "epoch": 3.2301790281329925, "percentage": 64.6, "elapsed_time": "6:21:46", "remaining_time": "3:29:10"} +{"current_steps": 1264, "total_steps": 1955, "loss": 0.9774, "lr": 2.6786429283474454e-05, "epoch": 3.2327365728900257, "percentage": 64.65, "elapsed_time": "6:22:05", "remaining_time": "3:28:52"} +{"current_steps": 1265, "total_steps": 1955, "loss": 0.9719, "lr": 2.67190204735531e-05, "epoch": 3.235294117647059, "percentage": 64.71, "elapsed_time": "6:22:23", "remaining_time": "3:28:34"} +{"current_steps": 1266, "total_steps": 1955, "loss": 0.9793, "lr": 2.665165402773895e-05, "epoch": 3.237851662404092, "percentage": 64.76, "elapsed_time": "6:22:41", "remaining_time": "3:28:16"} +{"current_steps": 1267, "total_steps": 1955, "loss": 0.9725, "lr": 2.6584330160919693e-05, "epoch": 3.2404092071611252, "percentage": 64.81, "elapsed_time": "6:22:59", "remaining_time": "3:27:58"} +{"current_steps": 1268, "total_steps": 1955, "loss": 0.9351, "lr": 2.6517049087847245e-05, "epoch": 3.2429667519181584, "percentage": 64.86, "elapsed_time": "6:23:17", "remaining_time": "3:27:39"} +{"current_steps": 1269, "total_steps": 1955, "loss": 0.9962, "lr": 2.6449811023137e-05, "epoch": 3.2455242966751916, "percentage": 64.91, "elapsed_time": "6:23:35", "remaining_time": "3:27:21"} +{"current_steps": 1270, "total_steps": 1955, "loss": 0.973, "lr": 2.6382616181267125e-05, "epoch": 3.2480818414322252, "percentage": 64.96, "elapsed_time": "6:23:53", "remaining_time": "3:27:03"} +{"current_steps": 1271, "total_steps": 1955, "loss": 0.9895, "lr": 2.6315464776577974e-05, "epoch": 3.2506393861892584, "percentage": 65.01, "elapsed_time": "6:24:11", "remaining_time": "3:26:45"} +{"current_steps": 1272, "total_steps": 1955, "loss": 0.9731, "lr": 2.6248357023271292e-05, "epoch": 3.2531969309462916, "percentage": 65.06, "elapsed_time": "6:24:30", "remaining_time": "3:26:27"} +{"current_steps": 1273, "total_steps": 1955, "loss": 0.9727, "lr": 2.6181293135409628e-05, "epoch": 3.2557544757033248, "percentage": 65.12, "elapsed_time": "6:24:47", "remaining_time": "3:26:09"} +{"current_steps": 1274, "total_steps": 1955, "loss": 0.9847, "lr": 2.611427332691558e-05, "epoch": 3.258312020460358, "percentage": 65.17, "elapsed_time": "6:25:05", "remaining_time": "3:25:50"} +{"current_steps": 1275, "total_steps": 1955, "loss": 0.9929, "lr": 2.6047297811571163e-05, "epoch": 3.260869565217391, "percentage": 65.22, "elapsed_time": "6:25:23", "remaining_time": "3:25:32"} +{"current_steps": 1276, "total_steps": 1955, "loss": 0.9486, "lr": 2.5980366803017063e-05, "epoch": 3.2634271099744243, "percentage": 65.27, "elapsed_time": "6:25:41", "remaining_time": "3:25:14"} +{"current_steps": 1277, "total_steps": 1955, "loss": 1.0034, "lr": 2.591348051475202e-05, "epoch": 3.265984654731458, "percentage": 65.32, "elapsed_time": "6:25:59", "remaining_time": "3:24:56"} +{"current_steps": 1278, "total_steps": 1955, "loss": 0.9725, "lr": 2.5846639160132152e-05, "epoch": 3.268542199488491, "percentage": 65.37, "elapsed_time": "6:26:17", "remaining_time": "3:24:38"} +{"current_steps": 1279, "total_steps": 1955, "loss": 1.0223, "lr": 2.5779842952370193e-05, "epoch": 3.2710997442455243, "percentage": 65.42, "elapsed_time": "6:26:35", "remaining_time": "3:24:19"} +{"current_steps": 1280, "total_steps": 1955, "loss": 0.9905, "lr": 2.5713092104534926e-05, "epoch": 3.2736572890025575, "percentage": 65.47, "elapsed_time": "6:26:54", "remaining_time": "3:24:01"} +{"current_steps": 1281, "total_steps": 1955, "loss": 0.9996, "lr": 2.5646386829550378e-05, "epoch": 3.2762148337595907, "percentage": 65.52, "elapsed_time": "6:27:12", "remaining_time": "3:23:43"} +{"current_steps": 1282, "total_steps": 1955, "loss": 0.9859, "lr": 2.557972734019524e-05, "epoch": 3.2787723785166243, "percentage": 65.58, "elapsed_time": "6:27:30", "remaining_time": "3:23:25"} +{"current_steps": 1283, "total_steps": 1955, "loss": 0.9847, "lr": 2.5513113849102168e-05, "epoch": 3.2813299232736575, "percentage": 65.63, "elapsed_time": "6:27:48", "remaining_time": "3:23:07"} +{"current_steps": 1284, "total_steps": 1955, "loss": 0.9804, "lr": 2.5446546568757074e-05, "epoch": 3.2838874680306906, "percentage": 65.68, "elapsed_time": "6:28:06", "remaining_time": "3:22:48"} +{"current_steps": 1285, "total_steps": 1955, "loss": 0.9838, "lr": 2.5380025711498463e-05, "epoch": 3.286445012787724, "percentage": 65.73, "elapsed_time": "6:28:24", "remaining_time": "3:22:30"} +{"current_steps": 1286, "total_steps": 1955, "loss": 0.9803, "lr": 2.5313551489516773e-05, "epoch": 3.289002557544757, "percentage": 65.78, "elapsed_time": "6:28:42", "remaining_time": "3:22:12"} +{"current_steps": 1287, "total_steps": 1955, "loss": 0.972, "lr": 2.5247124114853656e-05, "epoch": 3.29156010230179, "percentage": 65.83, "elapsed_time": "6:29:00", "remaining_time": "3:21:54"} +{"current_steps": 1288, "total_steps": 1955, "loss": 0.9585, "lr": 2.518074379940136e-05, "epoch": 3.2941176470588234, "percentage": 65.88, "elapsed_time": "6:29:18", "remaining_time": "3:21:36"} +{"current_steps": 1289, "total_steps": 1955, "loss": 0.9584, "lr": 2.5114410754901992e-05, "epoch": 3.296675191815857, "percentage": 65.93, "elapsed_time": "6:29:36", "remaining_time": "3:21:18"} +{"current_steps": 1290, "total_steps": 1955, "loss": 0.9567, "lr": 2.5048125192946912e-05, "epoch": 3.29923273657289, "percentage": 65.98, "elapsed_time": "6:29:54", "remaining_time": "3:20:59"} +{"current_steps": 1291, "total_steps": 1955, "loss": 0.969, "lr": 2.4981887324975988e-05, "epoch": 3.3017902813299234, "percentage": 66.04, "elapsed_time": "6:30:12", "remaining_time": "3:20:41"} +{"current_steps": 1292, "total_steps": 1955, "loss": 0.9958, "lr": 2.4915697362276935e-05, "epoch": 3.3043478260869565, "percentage": 66.09, "elapsed_time": "6:30:30", "remaining_time": "3:20:23"} +{"current_steps": 1293, "total_steps": 1955, "loss": 0.9392, "lr": 2.484955551598471e-05, "epoch": 3.3069053708439897, "percentage": 66.14, "elapsed_time": "6:30:48", "remaining_time": "3:20:05"} +{"current_steps": 1294, "total_steps": 1955, "loss": 0.9697, "lr": 2.4783461997080745e-05, "epoch": 3.309462915601023, "percentage": 66.19, "elapsed_time": "6:31:06", "remaining_time": "3:19:47"} +{"current_steps": 1295, "total_steps": 1955, "loss": 0.9814, "lr": 2.471741701639234e-05, "epoch": 3.312020460358056, "percentage": 66.24, "elapsed_time": "6:31:24", "remaining_time": "3:19:29"} +{"current_steps": 1296, "total_steps": 1955, "loss": 0.9924, "lr": 2.4651420784591942e-05, "epoch": 3.3145780051150897, "percentage": 66.29, "elapsed_time": "6:31:42", "remaining_time": "3:19:10"} +{"current_steps": 1297, "total_steps": 1955, "loss": 0.9493, "lr": 2.4585473512196537e-05, "epoch": 3.317135549872123, "percentage": 66.34, "elapsed_time": "6:32:00", "remaining_time": "3:18:52"} +{"current_steps": 1298, "total_steps": 1955, "loss": 0.9686, "lr": 2.4519575409566888e-05, "epoch": 3.319693094629156, "percentage": 66.39, "elapsed_time": "6:32:18", "remaining_time": "3:18:34"} +{"current_steps": 1299, "total_steps": 1955, "loss": 0.9934, "lr": 2.445372668690694e-05, "epoch": 3.3222506393861893, "percentage": 66.45, "elapsed_time": "6:32:37", "remaining_time": "3:18:16"} +{"current_steps": 1300, "total_steps": 1955, "loss": 1.0145, "lr": 2.4387927554263136e-05, "epoch": 3.3248081841432224, "percentage": 66.5, "elapsed_time": "6:32:55", "remaining_time": "3:17:58"} +{"current_steps": 1301, "total_steps": 1955, "loss": 0.9585, "lr": 2.432217822152371e-05, "epoch": 3.3273657289002556, "percentage": 66.55, "elapsed_time": "6:33:13", "remaining_time": "3:17:40"} +{"current_steps": 1302, "total_steps": 1955, "loss": 0.9883, "lr": 2.425647889841808e-05, "epoch": 3.329923273657289, "percentage": 66.6, "elapsed_time": "6:33:31", "remaining_time": "3:17:21"} +{"current_steps": 1303, "total_steps": 1955, "loss": 0.9681, "lr": 2.4190829794516093e-05, "epoch": 3.3324808184143224, "percentage": 66.65, "elapsed_time": "6:33:49", "remaining_time": "3:17:03"} +{"current_steps": 1304, "total_steps": 1955, "loss": 0.9839, "lr": 2.412523111922743e-05, "epoch": 3.3350383631713556, "percentage": 66.7, "elapsed_time": "6:34:07", "remaining_time": "3:16:45"} +{"current_steps": 1305, "total_steps": 1955, "loss": 0.9592, "lr": 2.405968308180093e-05, "epoch": 3.337595907928389, "percentage": 66.75, "elapsed_time": "6:34:25", "remaining_time": "3:16:27"} +{"current_steps": 1306, "total_steps": 1955, "loss": 0.9928, "lr": 2.399418589132386e-05, "epoch": 3.340153452685422, "percentage": 66.8, "elapsed_time": "6:34:43", "remaining_time": "3:16:09"} +{"current_steps": 1307, "total_steps": 1955, "loss": 0.9465, "lr": 2.392873975672135e-05, "epoch": 3.342710997442455, "percentage": 66.85, "elapsed_time": "6:35:01", "remaining_time": "3:15:50"} +{"current_steps": 1308, "total_steps": 1955, "loss": 0.9885, "lr": 2.386334488675564e-05, "epoch": 3.3452685421994883, "percentage": 66.91, "elapsed_time": "6:35:19", "remaining_time": "3:15:32"} +{"current_steps": 1309, "total_steps": 1955, "loss": 0.9736, "lr": 2.3798001490025417e-05, "epoch": 3.3478260869565215, "percentage": 66.96, "elapsed_time": "6:35:37", "remaining_time": "3:15:14"} +{"current_steps": 1310, "total_steps": 1955, "loss": 1.0126, "lr": 2.3732709774965238e-05, "epoch": 3.350383631713555, "percentage": 67.01, "elapsed_time": "6:35:55", "remaining_time": "3:14:56"} +{"current_steps": 1311, "total_steps": 1955, "loss": 0.9801, "lr": 2.3667469949844744e-05, "epoch": 3.3529411764705883, "percentage": 67.06, "elapsed_time": "6:36:13", "remaining_time": "3:14:38"} +{"current_steps": 1312, "total_steps": 1955, "loss": 0.9735, "lr": 2.3602282222768117e-05, "epoch": 3.3554987212276215, "percentage": 67.11, "elapsed_time": "6:36:31", "remaining_time": "3:14:20"} +{"current_steps": 1313, "total_steps": 1955, "loss": 0.9759, "lr": 2.353714680167331e-05, "epoch": 3.3580562659846547, "percentage": 67.16, "elapsed_time": "6:36:49", "remaining_time": "3:14:01"} +{"current_steps": 1314, "total_steps": 1955, "loss": 0.9853, "lr": 2.3472063894331423e-05, "epoch": 3.360613810741688, "percentage": 67.21, "elapsed_time": "6:37:07", "remaining_time": "3:13:43"} +{"current_steps": 1315, "total_steps": 1955, "loss": 0.956, "lr": 2.340703370834608e-05, "epoch": 3.363171355498721, "percentage": 67.26, "elapsed_time": "6:37:25", "remaining_time": "3:13:25"} +{"current_steps": 1316, "total_steps": 1955, "loss": 0.9786, "lr": 2.3342056451152713e-05, "epoch": 3.3657289002557547, "percentage": 67.31, "elapsed_time": "6:37:43", "remaining_time": "3:13:07"} +{"current_steps": 1317, "total_steps": 1955, "loss": 0.9896, "lr": 2.327713233001793e-05, "epoch": 3.368286445012788, "percentage": 67.37, "elapsed_time": "6:38:01", "remaining_time": "3:12:49"} +{"current_steps": 1318, "total_steps": 1955, "loss": 0.9856, "lr": 2.3212261552038827e-05, "epoch": 3.370843989769821, "percentage": 67.42, "elapsed_time": "6:38:19", "remaining_time": "3:12:30"} +{"current_steps": 1319, "total_steps": 1955, "loss": 0.9846, "lr": 2.314744432414236e-05, "epoch": 3.373401534526854, "percentage": 67.47, "elapsed_time": "6:38:37", "remaining_time": "3:12:12"} +{"current_steps": 1320, "total_steps": 1955, "loss": 0.96, "lr": 2.3082680853084663e-05, "epoch": 3.3759590792838874, "percentage": 67.52, "elapsed_time": "6:38:55", "remaining_time": "3:11:54"} +{"current_steps": 1321, "total_steps": 1955, "loss": 0.9602, "lr": 2.301797134545039e-05, "epoch": 3.3785166240409206, "percentage": 67.57, "elapsed_time": "6:39:13", "remaining_time": "3:11:36"} +{"current_steps": 1322, "total_steps": 1955, "loss": 0.9701, "lr": 2.2953316007652058e-05, "epoch": 3.381074168797954, "percentage": 67.62, "elapsed_time": "6:39:32", "remaining_time": "3:11:18"} +{"current_steps": 1323, "total_steps": 1955, "loss": 0.9633, "lr": 2.2888715045929427e-05, "epoch": 3.3836317135549874, "percentage": 67.67, "elapsed_time": "6:39:50", "remaining_time": "3:11:00"} +{"current_steps": 1324, "total_steps": 1955, "loss": 0.9771, "lr": 2.2824168666348775e-05, "epoch": 3.3861892583120206, "percentage": 67.72, "elapsed_time": "6:40:08", "remaining_time": "3:10:42"} +{"current_steps": 1325, "total_steps": 1955, "loss": 0.9786, "lr": 2.2759677074802237e-05, "epoch": 3.3887468030690537, "percentage": 67.77, "elapsed_time": "6:40:26", "remaining_time": "3:10:24"} +{"current_steps": 1326, "total_steps": 1955, "loss": 0.981, "lr": 2.269524047700726e-05, "epoch": 3.391304347826087, "percentage": 67.83, "elapsed_time": "6:40:44", "remaining_time": "3:10:05"} +{"current_steps": 1327, "total_steps": 1955, "loss": 0.9674, "lr": 2.2630859078505824e-05, "epoch": 3.39386189258312, "percentage": 67.88, "elapsed_time": "6:41:02", "remaining_time": "3:09:47"} +{"current_steps": 1328, "total_steps": 1955, "loss": 0.9556, "lr": 2.256653308466382e-05, "epoch": 3.3964194373401533, "percentage": 67.93, "elapsed_time": "6:41:20", "remaining_time": "3:09:29"} +{"current_steps": 1329, "total_steps": 1955, "loss": 1.0158, "lr": 2.2502262700670457e-05, "epoch": 3.398976982097187, "percentage": 67.98, "elapsed_time": "6:41:38", "remaining_time": "3:09:11"} +{"current_steps": 1330, "total_steps": 1955, "loss": 1.0003, "lr": 2.243804813153754e-05, "epoch": 3.40153452685422, "percentage": 68.03, "elapsed_time": "6:41:56", "remaining_time": "3:08:53"} +{"current_steps": 1331, "total_steps": 1955, "loss": 0.9725, "lr": 2.2373889582098767e-05, "epoch": 3.4040920716112533, "percentage": 68.08, "elapsed_time": "6:42:14", "remaining_time": "3:08:34"} +{"current_steps": 1332, "total_steps": 1955, "loss": 0.9907, "lr": 2.2309787257009243e-05, "epoch": 3.4066496163682864, "percentage": 68.13, "elapsed_time": "6:42:32", "remaining_time": "3:08:16"} +{"current_steps": 1333, "total_steps": 1955, "loss": 0.968, "lr": 2.224574136074467e-05, "epoch": 3.4092071611253196, "percentage": 68.18, "elapsed_time": "6:42:51", "remaining_time": "3:07:58"} +{"current_steps": 1334, "total_steps": 1955, "loss": 0.9957, "lr": 2.218175209760077e-05, "epoch": 3.411764705882353, "percentage": 68.24, "elapsed_time": "6:43:08", "remaining_time": "3:07:40"} +{"current_steps": 1335, "total_steps": 1955, "loss": 0.9802, "lr": 2.2117819671692596e-05, "epoch": 3.414322250639386, "percentage": 68.29, "elapsed_time": "6:43:26", "remaining_time": "3:07:22"} +{"current_steps": 1336, "total_steps": 1955, "loss": 0.9802, "lr": 2.2053944286953925e-05, "epoch": 3.4168797953964196, "percentage": 68.34, "elapsed_time": "6:43:44", "remaining_time": "3:07:03"} +{"current_steps": 1337, "total_steps": 1955, "loss": 0.9756, "lr": 2.199012614713655e-05, "epoch": 3.419437340153453, "percentage": 68.39, "elapsed_time": "6:44:03", "remaining_time": "3:06:45"} +{"current_steps": 1338, "total_steps": 1955, "loss": 0.9646, "lr": 2.1926365455809684e-05, "epoch": 3.421994884910486, "percentage": 68.44, "elapsed_time": "6:44:21", "remaining_time": "3:06:27"} +{"current_steps": 1339, "total_steps": 1955, "loss": 0.9712, "lr": 2.1862662416359316e-05, "epoch": 3.424552429667519, "percentage": 68.49, "elapsed_time": "6:44:39", "remaining_time": "3:06:09"} +{"current_steps": 1340, "total_steps": 1955, "loss": 1.0229, "lr": 2.1799017231987483e-05, "epoch": 3.4271099744245523, "percentage": 68.54, "elapsed_time": "6:44:57", "remaining_time": "3:05:51"} +{"current_steps": 1341, "total_steps": 1955, "loss": 1.012, "lr": 2.1735430105711695e-05, "epoch": 3.4296675191815855, "percentage": 68.59, "elapsed_time": "6:45:15", "remaining_time": "3:05:33"} +{"current_steps": 1342, "total_steps": 1955, "loss": 0.9772, "lr": 2.1671901240364272e-05, "epoch": 3.4322250639386187, "percentage": 68.64, "elapsed_time": "6:45:33", "remaining_time": "3:05:14"} +{"current_steps": 1343, "total_steps": 1955, "loss": 0.9893, "lr": 2.1608430838591685e-05, "epoch": 3.4347826086956523, "percentage": 68.7, "elapsed_time": "6:45:51", "remaining_time": "3:04:56"} +{"current_steps": 1344, "total_steps": 1955, "loss": 0.9717, "lr": 2.1545019102853927e-05, "epoch": 3.4373401534526855, "percentage": 68.75, "elapsed_time": "6:46:09", "remaining_time": "3:04:38"} +{"current_steps": 1345, "total_steps": 1955, "loss": 0.9616, "lr": 2.148166623542382e-05, "epoch": 3.4398976982097187, "percentage": 68.8, "elapsed_time": "6:46:28", "remaining_time": "3:04:20"} +{"current_steps": 1346, "total_steps": 1955, "loss": 1.0011, "lr": 2.1418372438386484e-05, "epoch": 3.442455242966752, "percentage": 68.85, "elapsed_time": "6:46:46", "remaining_time": "3:04:02"} +{"current_steps": 1347, "total_steps": 1955, "loss": 0.9772, "lr": 2.1355137913638515e-05, "epoch": 3.445012787723785, "percentage": 68.9, "elapsed_time": "6:47:04", "remaining_time": "3:03:44"} +{"current_steps": 1348, "total_steps": 1955, "loss": 0.9979, "lr": 2.1291962862887498e-05, "epoch": 3.4475703324808182, "percentage": 68.95, "elapsed_time": "6:47:22", "remaining_time": "3:03:26"} +{"current_steps": 1349, "total_steps": 1955, "loss": 1.0035, "lr": 2.1228847487651312e-05, "epoch": 3.4501278772378514, "percentage": 69.0, "elapsed_time": "6:47:40", "remaining_time": "3:03:08"} +{"current_steps": 1350, "total_steps": 1955, "loss": 0.9917, "lr": 2.116579198925747e-05, "epoch": 3.452685421994885, "percentage": 69.05, "elapsed_time": "6:47:58", "remaining_time": "3:02:49"} +{"current_steps": 1351, "total_steps": 1955, "loss": 0.9865, "lr": 2.110279656884246e-05, "epoch": 3.455242966751918, "percentage": 69.1, "elapsed_time": "6:48:16", "remaining_time": "3:02:32"} +{"current_steps": 1352, "total_steps": 1955, "loss": 0.9941, "lr": 2.103986142735118e-05, "epoch": 3.4578005115089514, "percentage": 69.16, "elapsed_time": "6:48:35", "remaining_time": "3:02:14"} +{"current_steps": 1353, "total_steps": 1955, "loss": 0.9584, "lr": 2.0976986765536202e-05, "epoch": 3.4603580562659846, "percentage": 69.21, "elapsed_time": "6:48:53", "remaining_time": "3:01:55"} +{"current_steps": 1354, "total_steps": 1955, "loss": 0.9804, "lr": 2.0914172783957213e-05, "epoch": 3.4629156010230178, "percentage": 69.26, "elapsed_time": "6:49:11", "remaining_time": "3:01:37"} +{"current_steps": 1355, "total_steps": 1955, "loss": 0.9722, "lr": 2.08514196829803e-05, "epoch": 3.4654731457800514, "percentage": 69.31, "elapsed_time": "6:49:30", "remaining_time": "3:01:19"} +{"current_steps": 1356, "total_steps": 1955, "loss": 0.9788, "lr": 2.078872766277742e-05, "epoch": 3.4680306905370846, "percentage": 69.36, "elapsed_time": "6:49:48", "remaining_time": "3:01:01"} +{"current_steps": 1357, "total_steps": 1955, "loss": 0.9819, "lr": 2.072609692332564e-05, "epoch": 3.4705882352941178, "percentage": 69.41, "elapsed_time": "6:50:06", "remaining_time": "3:00:43"} +{"current_steps": 1358, "total_steps": 1955, "loss": 0.9845, "lr": 2.0663527664406518e-05, "epoch": 3.473145780051151, "percentage": 69.46, "elapsed_time": "6:50:24", "remaining_time": "3:00:25"} +{"current_steps": 1359, "total_steps": 1955, "loss": 1.0034, "lr": 2.0601020085605598e-05, "epoch": 3.475703324808184, "percentage": 69.51, "elapsed_time": "6:50:42", "remaining_time": "3:00:07"} +{"current_steps": 1360, "total_steps": 1955, "loss": 0.992, "lr": 2.053857438631159e-05, "epoch": 3.4782608695652173, "percentage": 69.57, "elapsed_time": "6:51:00", "remaining_time": "2:59:48"} +{"current_steps": 1361, "total_steps": 1955, "loss": 0.9748, "lr": 2.0476190765715858e-05, "epoch": 3.4808184143222505, "percentage": 69.62, "elapsed_time": "6:51:19", "remaining_time": "2:59:31"} +{"current_steps": 1362, "total_steps": 1955, "loss": 0.9809, "lr": 2.041386942281171e-05, "epoch": 3.483375959079284, "percentage": 69.67, "elapsed_time": "6:51:37", "remaining_time": "2:59:13"} +{"current_steps": 1363, "total_steps": 1955, "loss": 0.9971, "lr": 2.0351610556393887e-05, "epoch": 3.4859335038363173, "percentage": 69.72, "elapsed_time": "6:51:55", "remaining_time": "2:58:54"} +{"current_steps": 1364, "total_steps": 1955, "loss": 0.9669, "lr": 2.0289414365057716e-05, "epoch": 3.4884910485933505, "percentage": 69.77, "elapsed_time": "6:52:13", "remaining_time": "2:58:36"} +{"current_steps": 1365, "total_steps": 1955, "loss": 1.004, "lr": 2.022728104719867e-05, "epoch": 3.4910485933503836, "percentage": 69.82, "elapsed_time": "6:52:31", "remaining_time": "2:58:18"} +{"current_steps": 1366, "total_steps": 1955, "loss": 1.0225, "lr": 2.0165210801011684e-05, "epoch": 3.493606138107417, "percentage": 69.87, "elapsed_time": "6:52:49", "remaining_time": "2:58:00"} +{"current_steps": 1367, "total_steps": 1955, "loss": 1.0037, "lr": 2.010320382449048e-05, "epoch": 3.49616368286445, "percentage": 69.92, "elapsed_time": "6:53:07", "remaining_time": "2:57:42"} +{"current_steps": 1368, "total_steps": 1955, "loss": 0.9444, "lr": 2.0041260315426933e-05, "epoch": 3.498721227621483, "percentage": 69.97, "elapsed_time": "6:53:25", "remaining_time": "2:57:23"} +{"current_steps": 1369, "total_steps": 1955, "loss": 0.9926, "lr": 1.997938047141052e-05, "epoch": 3.501278772378517, "percentage": 70.03, "elapsed_time": "6:53:43", "remaining_time": "2:57:05"} +{"current_steps": 1370, "total_steps": 1955, "loss": 0.9941, "lr": 1.9917564489827595e-05, "epoch": 3.50383631713555, "percentage": 70.08, "elapsed_time": "6:54:01", "remaining_time": "2:56:47"} +{"current_steps": 1371, "total_steps": 1955, "loss": 0.9904, "lr": 1.9855812567860832e-05, "epoch": 3.506393861892583, "percentage": 70.13, "elapsed_time": "6:54:19", "remaining_time": "2:56:29"} +{"current_steps": 1372, "total_steps": 1955, "loss": 1.0092, "lr": 1.979412490248852e-05, "epoch": 3.5089514066496164, "percentage": 70.18, "elapsed_time": "6:54:37", "remaining_time": "2:56:11"} +{"current_steps": 1373, "total_steps": 1955, "loss": 0.9686, "lr": 1.9732501690484054e-05, "epoch": 3.5115089514066495, "percentage": 70.23, "elapsed_time": "6:54:55", "remaining_time": "2:55:53"} +{"current_steps": 1374, "total_steps": 1955, "loss": 0.9905, "lr": 1.9670943128415185e-05, "epoch": 3.5140664961636827, "percentage": 70.28, "elapsed_time": "6:55:13", "remaining_time": "2:55:34"} +{"current_steps": 1375, "total_steps": 1955, "loss": 0.9622, "lr": 1.9609449412643403e-05, "epoch": 3.516624040920716, "percentage": 70.33, "elapsed_time": "6:55:31", "remaining_time": "2:55:16"} +{"current_steps": 1376, "total_steps": 1955, "loss": 0.9684, "lr": 1.9548020739323443e-05, "epoch": 3.5191815856777495, "percentage": 70.38, "elapsed_time": "6:55:49", "remaining_time": "2:54:58"} +{"current_steps": 1377, "total_steps": 1955, "loss": 0.9961, "lr": 1.94866573044025e-05, "epoch": 3.5217391304347827, "percentage": 70.43, "elapsed_time": "6:56:07", "remaining_time": "2:54:40"} +{"current_steps": 1378, "total_steps": 1955, "loss": 0.9624, "lr": 1.9425359303619685e-05, "epoch": 3.524296675191816, "percentage": 70.49, "elapsed_time": "6:56:25", "remaining_time": "2:54:22"} +{"current_steps": 1379, "total_steps": 1955, "loss": 1.0249, "lr": 1.9364126932505383e-05, "epoch": 3.526854219948849, "percentage": 70.54, "elapsed_time": "6:56:43", "remaining_time": "2:54:04"} +{"current_steps": 1380, "total_steps": 1955, "loss": 0.9665, "lr": 1.9302960386380637e-05, "epoch": 3.5294117647058822, "percentage": 70.59, "elapsed_time": "6:57:01", "remaining_time": "2:53:45"} +{"current_steps": 1381, "total_steps": 1955, "loss": 1.021, "lr": 1.924185986035652e-05, "epoch": 3.531969309462916, "percentage": 70.64, "elapsed_time": "6:57:20", "remaining_time": "2:53:27"} +{"current_steps": 1382, "total_steps": 1955, "loss": 0.9914, "lr": 1.9180825549333483e-05, "epoch": 3.5345268542199486, "percentage": 70.69, "elapsed_time": "6:57:38", "remaining_time": "2:53:09"} +{"current_steps": 1383, "total_steps": 1955, "loss": 0.9762, "lr": 1.9119857648000827e-05, "epoch": 3.5370843989769822, "percentage": 70.74, "elapsed_time": "6:57:56", "remaining_time": "2:52:51"} +{"current_steps": 1384, "total_steps": 1955, "loss": 1.0047, "lr": 1.9058956350835952e-05, "epoch": 3.5396419437340154, "percentage": 70.79, "elapsed_time": "6:58:14", "remaining_time": "2:52:33"} +{"current_steps": 1385, "total_steps": 1955, "loss": 0.9844, "lr": 1.8998121852103843e-05, "epoch": 3.5421994884910486, "percentage": 70.84, "elapsed_time": "6:58:32", "remaining_time": "2:52:14"} +{"current_steps": 1386, "total_steps": 1955, "loss": 1.0024, "lr": 1.8937354345856383e-05, "epoch": 3.544757033248082, "percentage": 70.9, "elapsed_time": "6:58:51", "remaining_time": "2:51:57"} +{"current_steps": 1387, "total_steps": 1955, "loss": 0.9568, "lr": 1.8876654025931767e-05, "epoch": 3.547314578005115, "percentage": 70.95, "elapsed_time": "6:59:09", "remaining_time": "2:51:38"} +{"current_steps": 1388, "total_steps": 1955, "loss": 0.9762, "lr": 1.881602108595389e-05, "epoch": 3.5498721227621486, "percentage": 71.0, "elapsed_time": "6:59:27", "remaining_time": "2:51:20"} +{"current_steps": 1389, "total_steps": 1955, "loss": 1.0031, "lr": 1.875545571933169e-05, "epoch": 3.5524296675191813, "percentage": 71.05, "elapsed_time": "6:59:45", "remaining_time": "2:51:02"} +{"current_steps": 1390, "total_steps": 1955, "loss": 1.0111, "lr": 1.8694958119258612e-05, "epoch": 3.554987212276215, "percentage": 71.1, "elapsed_time": "7:00:03", "remaining_time": "2:50:44"} +{"current_steps": 1391, "total_steps": 1955, "loss": 0.948, "lr": 1.863452847871186e-05, "epoch": 3.557544757033248, "percentage": 71.15, "elapsed_time": "7:00:21", "remaining_time": "2:50:26"} +{"current_steps": 1392, "total_steps": 1955, "loss": 0.9727, "lr": 1.8574166990451898e-05, "epoch": 3.5601023017902813, "percentage": 71.2, "elapsed_time": "7:00:39", "remaining_time": "2:50:08"} +{"current_steps": 1393, "total_steps": 1955, "loss": 0.9908, "lr": 1.8513873847021823e-05, "epoch": 3.5626598465473145, "percentage": 71.25, "elapsed_time": "7:00:57", "remaining_time": "2:49:50"} +{"current_steps": 1394, "total_steps": 1955, "loss": 0.9932, "lr": 1.8453649240746675e-05, "epoch": 3.5652173913043477, "percentage": 71.3, "elapsed_time": "7:01:15", "remaining_time": "2:49:31"} +{"current_steps": 1395, "total_steps": 1955, "loss": 0.9989, "lr": 1.8393493363732905e-05, "epoch": 3.5677749360613813, "percentage": 71.36, "elapsed_time": "7:01:33", "remaining_time": "2:49:13"} +{"current_steps": 1396, "total_steps": 1955, "loss": 0.9777, "lr": 1.8333406407867712e-05, "epoch": 3.5703324808184145, "percentage": 71.41, "elapsed_time": "7:01:51", "remaining_time": "2:48:55"} +{"current_steps": 1397, "total_steps": 1955, "loss": 0.9587, "lr": 1.8273388564818443e-05, "epoch": 3.5728900255754477, "percentage": 71.46, "elapsed_time": "7:02:09", "remaining_time": "2:48:37"} +{"current_steps": 1398, "total_steps": 1955, "loss": 0.9979, "lr": 1.8213440026032013e-05, "epoch": 3.575447570332481, "percentage": 71.51, "elapsed_time": "7:02:27", "remaining_time": "2:48:19"} +{"current_steps": 1399, "total_steps": 1955, "loss": 0.9993, "lr": 1.815356098273422e-05, "epoch": 3.578005115089514, "percentage": 71.56, "elapsed_time": "7:02:45", "remaining_time": "2:48:01"} +{"current_steps": 1400, "total_steps": 1955, "loss": 0.9605, "lr": 1.8093751625929252e-05, "epoch": 3.580562659846547, "percentage": 71.61, "elapsed_time": "7:03:03", "remaining_time": "2:47:42"} +{"current_steps": 1401, "total_steps": 1955, "loss": 0.9697, "lr": 1.8034012146398972e-05, "epoch": 3.5831202046035804, "percentage": 71.66, "elapsed_time": "7:03:22", "remaining_time": "2:47:24"} +{"current_steps": 1402, "total_steps": 1955, "loss": 0.9891, "lr": 1.79743427347023e-05, "epoch": 3.585677749360614, "percentage": 71.71, "elapsed_time": "7:03:40", "remaining_time": "2:47:06"} +{"current_steps": 1403, "total_steps": 1955, "loss": 0.9841, "lr": 1.791474358117474e-05, "epoch": 3.588235294117647, "percentage": 71.76, "elapsed_time": "7:03:58", "remaining_time": "2:46:48"} +{"current_steps": 1404, "total_steps": 1955, "loss": 0.9899, "lr": 1.785521487592762e-05, "epoch": 3.5907928388746804, "percentage": 71.82, "elapsed_time": "7:04:16", "remaining_time": "2:46:30"} +{"current_steps": 1405, "total_steps": 1955, "loss": 0.9807, "lr": 1.7795756808847583e-05, "epoch": 3.5933503836317136, "percentage": 71.87, "elapsed_time": "7:04:34", "remaining_time": "2:46:12"} +{"current_steps": 1406, "total_steps": 1955, "loss": 0.9837, "lr": 1.773636956959591e-05, "epoch": 3.5959079283887467, "percentage": 71.92, "elapsed_time": "7:04:52", "remaining_time": "2:45:54"} +{"current_steps": 1407, "total_steps": 1955, "loss": 0.9929, "lr": 1.7677053347608026e-05, "epoch": 3.59846547314578, "percentage": 71.97, "elapsed_time": "7:05:10", "remaining_time": "2:45:36"} +{"current_steps": 1408, "total_steps": 1955, "loss": 0.9937, "lr": 1.7617808332092733e-05, "epoch": 3.601023017902813, "percentage": 72.02, "elapsed_time": "7:05:28", "remaining_time": "2:45:17"} +{"current_steps": 1409, "total_steps": 1955, "loss": 0.9716, "lr": 1.7558634712031733e-05, "epoch": 3.6035805626598467, "percentage": 72.07, "elapsed_time": "7:05:47", "remaining_time": "2:45:00"} +{"current_steps": 1410, "total_steps": 1955, "loss": 0.9463, "lr": 1.7499532676179027e-05, "epoch": 3.60613810741688, "percentage": 72.12, "elapsed_time": "7:06:05", "remaining_time": "2:44:41"} +{"current_steps": 1411, "total_steps": 1955, "loss": 0.9873, "lr": 1.744050241306023e-05, "epoch": 3.608695652173913, "percentage": 72.17, "elapsed_time": "7:06:23", "remaining_time": "2:44:23"} +{"current_steps": 1412, "total_steps": 1955, "loss": 1.0049, "lr": 1.738154411097202e-05, "epoch": 3.6112531969309463, "percentage": 72.23, "elapsed_time": "7:06:41", "remaining_time": "2:44:05"} +{"current_steps": 1413, "total_steps": 1955, "loss": 0.9629, "lr": 1.7322657957981543e-05, "epoch": 3.6138107416879794, "percentage": 72.28, "elapsed_time": "7:07:00", "remaining_time": "2:43:47"} +{"current_steps": 1414, "total_steps": 1955, "loss": 0.9817, "lr": 1.726384414192578e-05, "epoch": 3.6163682864450126, "percentage": 72.33, "elapsed_time": "7:07:18", "remaining_time": "2:43:29"} +{"current_steps": 1415, "total_steps": 1955, "loss": 0.9608, "lr": 1.7205102850411e-05, "epoch": 3.618925831202046, "percentage": 72.38, "elapsed_time": "7:07:36", "remaining_time": "2:43:11"} +{"current_steps": 1416, "total_steps": 1955, "loss": 0.9834, "lr": 1.71464342708121e-05, "epoch": 3.6214833759590794, "percentage": 72.43, "elapsed_time": "7:07:54", "remaining_time": "2:42:52"} +{"current_steps": 1417, "total_steps": 1955, "loss": 0.9599, "lr": 1.7087838590272067e-05, "epoch": 3.6240409207161126, "percentage": 72.48, "elapsed_time": "7:08:12", "remaining_time": "2:42:34"} +{"current_steps": 1418, "total_steps": 1955, "loss": 0.9952, "lr": 1.7029315995701355e-05, "epoch": 3.626598465473146, "percentage": 72.53, "elapsed_time": "7:08:29", "remaining_time": "2:42:16"} +{"current_steps": 1419, "total_steps": 1955, "loss": 0.9879, "lr": 1.6970866673777217e-05, "epoch": 3.629156010230179, "percentage": 72.58, "elapsed_time": "7:08:47", "remaining_time": "2:41:58"} +{"current_steps": 1420, "total_steps": 1955, "loss": 0.9886, "lr": 1.6912490810943273e-05, "epoch": 3.631713554987212, "percentage": 72.63, "elapsed_time": "7:09:05", "remaining_time": "2:41:39"} +{"current_steps": 1421, "total_steps": 1955, "loss": 1.004, "lr": 1.685418859340877e-05, "epoch": 3.634271099744246, "percentage": 72.69, "elapsed_time": "7:09:23", "remaining_time": "2:41:21"} +{"current_steps": 1422, "total_steps": 1955, "loss": 0.9585, "lr": 1.6795960207148047e-05, "epoch": 3.6368286445012785, "percentage": 72.74, "elapsed_time": "7:09:41", "remaining_time": "2:41:03"} +{"current_steps": 1423, "total_steps": 1955, "loss": 0.9864, "lr": 1.673780583789993e-05, "epoch": 3.639386189258312, "percentage": 72.79, "elapsed_time": "7:09:59", "remaining_time": "2:40:45"} +{"current_steps": 1424, "total_steps": 1955, "loss": 0.9778, "lr": 1.6679725671167153e-05, "epoch": 3.6419437340153453, "percentage": 72.84, "elapsed_time": "7:10:18", "remaining_time": "2:40:27"} +{"current_steps": 1425, "total_steps": 1955, "loss": 1.0061, "lr": 1.662171989221574e-05, "epoch": 3.6445012787723785, "percentage": 72.89, "elapsed_time": "7:10:36", "remaining_time": "2:40:09"} +{"current_steps": 1426, "total_steps": 1955, "loss": 0.9656, "lr": 1.6563788686074437e-05, "epoch": 3.6470588235294117, "percentage": 72.94, "elapsed_time": "7:10:54", "remaining_time": "2:39:51"} +{"current_steps": 1427, "total_steps": 1955, "loss": 0.9766, "lr": 1.6505932237534135e-05, "epoch": 3.649616368286445, "percentage": 72.99, "elapsed_time": "7:11:12", "remaining_time": "2:39:32"} +{"current_steps": 1428, "total_steps": 1955, "loss": 0.971, "lr": 1.6448150731147245e-05, "epoch": 3.6521739130434785, "percentage": 73.04, "elapsed_time": "7:11:30", "remaining_time": "2:39:14"} +{"current_steps": 1429, "total_steps": 1955, "loss": 0.9615, "lr": 1.6390444351227116e-05, "epoch": 3.6547314578005117, "percentage": 73.09, "elapsed_time": "7:11:48", "remaining_time": "2:38:56"} +{"current_steps": 1430, "total_steps": 1955, "loss": 0.9918, "lr": 1.6332813281847468e-05, "epoch": 3.657289002557545, "percentage": 73.15, "elapsed_time": "7:12:06", "remaining_time": "2:38:38"} +{"current_steps": 1431, "total_steps": 1955, "loss": 0.9853, "lr": 1.6275257706841787e-05, "epoch": 3.659846547314578, "percentage": 73.2, "elapsed_time": "7:12:24", "remaining_time": "2:38:20"} +{"current_steps": 1432, "total_steps": 1955, "loss": 0.9889, "lr": 1.6217777809802752e-05, "epoch": 3.662404092071611, "percentage": 73.25, "elapsed_time": "7:12:42", "remaining_time": "2:38:02"} +{"current_steps": 1433, "total_steps": 1955, "loss": 0.9645, "lr": 1.6160373774081616e-05, "epoch": 3.6649616368286444, "percentage": 73.3, "elapsed_time": "7:13:00", "remaining_time": "2:37:43"} +{"current_steps": 1434, "total_steps": 1955, "loss": 0.9704, "lr": 1.610304578278772e-05, "epoch": 3.6675191815856776, "percentage": 73.35, "elapsed_time": "7:13:18", "remaining_time": "2:37:25"} +{"current_steps": 1435, "total_steps": 1955, "loss": 0.9661, "lr": 1.604579401878773e-05, "epoch": 3.670076726342711, "percentage": 73.4, "elapsed_time": "7:13:37", "remaining_time": "2:37:07"} +{"current_steps": 1436, "total_steps": 1955, "loss": 0.9958, "lr": 1.5988618664705223e-05, "epoch": 3.6726342710997444, "percentage": 73.45, "elapsed_time": "7:13:55", "remaining_time": "2:36:49"} +{"current_steps": 1437, "total_steps": 1955, "loss": 0.9723, "lr": 1.593151990292006e-05, "epoch": 3.6751918158567776, "percentage": 73.5, "elapsed_time": "7:14:13", "remaining_time": "2:36:31"} +{"current_steps": 1438, "total_steps": 1955, "loss": 1.0298, "lr": 1.587449791556774e-05, "epoch": 3.6777493606138107, "percentage": 73.55, "elapsed_time": "7:14:31", "remaining_time": "2:36:13"} +{"current_steps": 1439, "total_steps": 1955, "loss": 0.9652, "lr": 1.5817552884538883e-05, "epoch": 3.680306905370844, "percentage": 73.61, "elapsed_time": "7:14:49", "remaining_time": "2:35:55"} +{"current_steps": 1440, "total_steps": 1955, "loss": 0.9658, "lr": 1.5760684991478634e-05, "epoch": 3.682864450127877, "percentage": 73.66, "elapsed_time": "7:15:08", "remaining_time": "2:35:37"} +{"current_steps": 1441, "total_steps": 1955, "loss": 0.9659, "lr": 1.5703894417786066e-05, "epoch": 3.6854219948849103, "percentage": 73.71, "elapsed_time": "7:15:26", "remaining_time": "2:35:19"} +{"current_steps": 1442, "total_steps": 1955, "loss": 0.9634, "lr": 1.564718134461364e-05, "epoch": 3.687979539641944, "percentage": 73.76, "elapsed_time": "7:15:45", "remaining_time": "2:35:01"} +{"current_steps": 1443, "total_steps": 1955, "loss": 0.9473, "lr": 1.5590545952866565e-05, "epoch": 3.690537084398977, "percentage": 73.81, "elapsed_time": "7:16:03", "remaining_time": "2:34:43"} +{"current_steps": 1444, "total_steps": 1955, "loss": 0.974, "lr": 1.553398842320232e-05, "epoch": 3.6930946291560103, "percentage": 73.86, "elapsed_time": "7:16:21", "remaining_time": "2:34:24"} +{"current_steps": 1445, "total_steps": 1955, "loss": 0.9945, "lr": 1.5477508936029978e-05, "epoch": 3.6956521739130435, "percentage": 73.91, "elapsed_time": "7:16:39", "remaining_time": "2:34:07"} +{"current_steps": 1446, "total_steps": 1955, "loss": 0.9586, "lr": 1.542110767150963e-05, "epoch": 3.6982097186700766, "percentage": 73.96, "elapsed_time": "7:16:58", "remaining_time": "2:33:48"} +{"current_steps": 1447, "total_steps": 1955, "loss": 0.9853, "lr": 1.5364784809551938e-05, "epoch": 3.70076726342711, "percentage": 74.02, "elapsed_time": "7:17:16", "remaining_time": "2:33:30"} +{"current_steps": 1448, "total_steps": 1955, "loss": 1.0062, "lr": 1.5308540529817405e-05, "epoch": 3.703324808184143, "percentage": 74.07, "elapsed_time": "7:17:34", "remaining_time": "2:33:12"} +{"current_steps": 1449, "total_steps": 1955, "loss": 0.9602, "lr": 1.5252375011715894e-05, "epoch": 3.7058823529411766, "percentage": 74.12, "elapsed_time": "7:17:52", "remaining_time": "2:32:54"} +{"current_steps": 1450, "total_steps": 1955, "loss": 0.9729, "lr": 1.5196288434406045e-05, "epoch": 3.70843989769821, "percentage": 74.17, "elapsed_time": "7:18:10", "remaining_time": "2:32:36"} +{"current_steps": 1451, "total_steps": 1955, "loss": 0.9534, "lr": 1.514028097679468e-05, "epoch": 3.710997442455243, "percentage": 74.22, "elapsed_time": "7:18:28", "remaining_time": "2:32:18"} +{"current_steps": 1452, "total_steps": 1955, "loss": 0.9829, "lr": 1.5084352817536205e-05, "epoch": 3.713554987212276, "percentage": 74.27, "elapsed_time": "7:18:46", "remaining_time": "2:32:00"} +{"current_steps": 1453, "total_steps": 1955, "loss": 0.9666, "lr": 1.5028504135032113e-05, "epoch": 3.7161125319693094, "percentage": 74.32, "elapsed_time": "7:19:04", "remaining_time": "2:31:41"} +{"current_steps": 1454, "total_steps": 1955, "loss": 1.0091, "lr": 1.4972735107430407e-05, "epoch": 3.718670076726343, "percentage": 74.37, "elapsed_time": "7:19:22", "remaining_time": "2:31:23"} +{"current_steps": 1455, "total_steps": 1955, "loss": 0.9656, "lr": 1.4917045912624954e-05, "epoch": 3.7212276214833757, "percentage": 74.42, "elapsed_time": "7:19:40", "remaining_time": "2:31:05"} +{"current_steps": 1456, "total_steps": 1955, "loss": 0.9626, "lr": 1.4861436728254992e-05, "epoch": 3.7237851662404093, "percentage": 74.48, "elapsed_time": "7:19:58", "remaining_time": "2:30:47"} +{"current_steps": 1457, "total_steps": 1955, "loss": 1.0058, "lr": 1.480590773170453e-05, "epoch": 3.7263427109974425, "percentage": 74.53, "elapsed_time": "7:20:17", "remaining_time": "2:30:29"} +{"current_steps": 1458, "total_steps": 1955, "loss": 0.965, "lr": 1.47504591001018e-05, "epoch": 3.7289002557544757, "percentage": 74.58, "elapsed_time": "7:20:35", "remaining_time": "2:30:11"} +{"current_steps": 1459, "total_steps": 1955, "loss": 0.9992, "lr": 1.469509101031866e-05, "epoch": 3.731457800511509, "percentage": 74.63, "elapsed_time": "7:20:53", "remaining_time": "2:29:52"} +{"current_steps": 1460, "total_steps": 1955, "loss": 0.9818, "lr": 1.4639803638970098e-05, "epoch": 3.734015345268542, "percentage": 74.68, "elapsed_time": "7:21:11", "remaining_time": "2:29:34"} +{"current_steps": 1461, "total_steps": 1955, "loss": 0.9596, "lr": 1.4584597162413592e-05, "epoch": 3.7365728900255757, "percentage": 74.73, "elapsed_time": "7:21:29", "remaining_time": "2:29:16"} +{"current_steps": 1462, "total_steps": 1955, "loss": 0.9654, "lr": 1.4529471756748579e-05, "epoch": 3.7391304347826084, "percentage": 74.78, "elapsed_time": "7:21:46", "remaining_time": "2:28:58"} +{"current_steps": 1463, "total_steps": 1955, "loss": 0.9749, "lr": 1.4474427597815898e-05, "epoch": 3.741687979539642, "percentage": 74.83, "elapsed_time": "7:22:04", "remaining_time": "2:28:40"} +{"current_steps": 1464, "total_steps": 1955, "loss": 0.9936, "lr": 1.4419464861197225e-05, "epoch": 3.7442455242966752, "percentage": 74.88, "elapsed_time": "7:22:22", "remaining_time": "2:28:21"} +{"current_steps": 1465, "total_steps": 1955, "loss": 0.9816, "lr": 1.4364583722214521e-05, "epoch": 3.7468030690537084, "percentage": 74.94, "elapsed_time": "7:22:40", "remaining_time": "2:28:03"} +{"current_steps": 1466, "total_steps": 1955, "loss": 0.9784, "lr": 1.4309784355929429e-05, "epoch": 3.7493606138107416, "percentage": 74.99, "elapsed_time": "7:22:58", "remaining_time": "2:27:45"} +{"current_steps": 1467, "total_steps": 1955, "loss": 0.9801, "lr": 1.4255066937142816e-05, "epoch": 3.7519181585677748, "percentage": 75.04, "elapsed_time": "7:23:16", "remaining_time": "2:27:27"} +{"current_steps": 1468, "total_steps": 1955, "loss": 0.9749, "lr": 1.4200431640394103e-05, "epoch": 3.7544757033248084, "percentage": 75.09, "elapsed_time": "7:23:35", "remaining_time": "2:27:09"} +{"current_steps": 1469, "total_steps": 1955, "loss": 0.9978, "lr": 1.4145878639960735e-05, "epoch": 3.7570332480818416, "percentage": 75.14, "elapsed_time": "7:23:53", "remaining_time": "2:26:51"} +{"current_steps": 1470, "total_steps": 1955, "loss": 0.9557, "lr": 1.4091408109857706e-05, "epoch": 3.7595907928388748, "percentage": 75.19, "elapsed_time": "7:24:11", "remaining_time": "2:26:33"} +{"current_steps": 1471, "total_steps": 1955, "loss": 0.9881, "lr": 1.4037020223836897e-05, "epoch": 3.762148337595908, "percentage": 75.24, "elapsed_time": "7:24:29", "remaining_time": "2:26:15"} +{"current_steps": 1472, "total_steps": 1955, "loss": 0.982, "lr": 1.3982715155386592e-05, "epoch": 3.764705882352941, "percentage": 75.29, "elapsed_time": "7:24:47", "remaining_time": "2:25:56"} +{"current_steps": 1473, "total_steps": 1955, "loss": 0.9652, "lr": 1.3928493077730889e-05, "epoch": 3.7672634271099743, "percentage": 75.35, "elapsed_time": "7:25:05", "remaining_time": "2:25:38"} +{"current_steps": 1474, "total_steps": 1955, "loss": 0.9742, "lr": 1.3874354163829158e-05, "epoch": 3.7698209718670075, "percentage": 75.4, "elapsed_time": "7:25:23", "remaining_time": "2:25:20"} +{"current_steps": 1475, "total_steps": 1955, "loss": 0.9873, "lr": 1.3820298586375502e-05, "epoch": 3.772378516624041, "percentage": 75.45, "elapsed_time": "7:25:41", "remaining_time": "2:25:02"} +{"current_steps": 1476, "total_steps": 1955, "loss": 0.9799, "lr": 1.3766326517798168e-05, "epoch": 3.7749360613810743, "percentage": 75.5, "elapsed_time": "7:25:59", "remaining_time": "2:24:44"} +{"current_steps": 1477, "total_steps": 1955, "loss": 0.9489, "lr": 1.3712438130259073e-05, "epoch": 3.7774936061381075, "percentage": 75.55, "elapsed_time": "7:26:17", "remaining_time": "2:24:26"} +{"current_steps": 1478, "total_steps": 1955, "loss": 0.9741, "lr": 1.3658633595653159e-05, "epoch": 3.7800511508951407, "percentage": 75.6, "elapsed_time": "7:26:35", "remaining_time": "2:24:07"} +{"current_steps": 1479, "total_steps": 1955, "loss": 0.959, "lr": 1.3604913085607914e-05, "epoch": 3.782608695652174, "percentage": 75.65, "elapsed_time": "7:26:54", "remaining_time": "2:23:49"} +{"current_steps": 1480, "total_steps": 1955, "loss": 0.9692, "lr": 1.3551276771482788e-05, "epoch": 3.785166240409207, "percentage": 75.7, "elapsed_time": "7:27:12", "remaining_time": "2:23:31"} +{"current_steps": 1481, "total_steps": 1955, "loss": 0.9696, "lr": 1.3497724824368672e-05, "epoch": 3.78772378516624, "percentage": 75.75, "elapsed_time": "7:27:30", "remaining_time": "2:23:13"} +{"current_steps": 1482, "total_steps": 1955, "loss": 0.9879, "lr": 1.3444257415087339e-05, "epoch": 3.790281329923274, "percentage": 75.81, "elapsed_time": "7:27:48", "remaining_time": "2:22:55"} +{"current_steps": 1483, "total_steps": 1955, "loss": 1.0394, "lr": 1.3390874714190876e-05, "epoch": 3.792838874680307, "percentage": 75.86, "elapsed_time": "7:28:06", "remaining_time": "2:22:37"} +{"current_steps": 1484, "total_steps": 1955, "loss": 0.9854, "lr": 1.3337576891961228e-05, "epoch": 3.79539641943734, "percentage": 75.91, "elapsed_time": "7:28:24", "remaining_time": "2:22:19"} +{"current_steps": 1485, "total_steps": 1955, "loss": 1.0181, "lr": 1.3284364118409512e-05, "epoch": 3.7979539641943734, "percentage": 75.96, "elapsed_time": "7:28:42", "remaining_time": "2:22:00"} +{"current_steps": 1486, "total_steps": 1955, "loss": 1.0083, "lr": 1.3231236563275593e-05, "epoch": 3.8005115089514065, "percentage": 76.01, "elapsed_time": "7:29:00", "remaining_time": "2:21:42"} +{"current_steps": 1487, "total_steps": 1955, "loss": 0.993, "lr": 1.317819439602752e-05, "epoch": 3.80306905370844, "percentage": 76.06, "elapsed_time": "7:29:18", "remaining_time": "2:21:24"} +{"current_steps": 1488, "total_steps": 1955, "loss": 0.9749, "lr": 1.3125237785860945e-05, "epoch": 3.805626598465473, "percentage": 76.11, "elapsed_time": "7:29:36", "remaining_time": "2:21:06"} +{"current_steps": 1489, "total_steps": 1955, "loss": 0.9819, "lr": 1.3072366901698606e-05, "epoch": 3.8081841432225065, "percentage": 76.16, "elapsed_time": "7:29:54", "remaining_time": "2:20:48"} +{"current_steps": 1490, "total_steps": 1955, "loss": 0.9852, "lr": 1.3019581912189794e-05, "epoch": 3.8107416879795397, "percentage": 76.21, "elapsed_time": "7:30:12", "remaining_time": "2:20:30"} +{"current_steps": 1491, "total_steps": 1955, "loss": 0.9839, "lr": 1.2966882985709824e-05, "epoch": 3.813299232736573, "percentage": 76.27, "elapsed_time": "7:30:30", "remaining_time": "2:20:11"} +{"current_steps": 1492, "total_steps": 1955, "loss": 0.9797, "lr": 1.291427029035946e-05, "epoch": 3.815856777493606, "percentage": 76.32, "elapsed_time": "7:30:48", "remaining_time": "2:19:53"} +{"current_steps": 1493, "total_steps": 1955, "loss": 0.9868, "lr": 1.2861743993964409e-05, "epoch": 3.8184143222506393, "percentage": 76.37, "elapsed_time": "7:31:06", "remaining_time": "2:19:35"} +{"current_steps": 1494, "total_steps": 1955, "loss": 0.9641, "lr": 1.2809304264074807e-05, "epoch": 3.820971867007673, "percentage": 76.42, "elapsed_time": "7:31:24", "remaining_time": "2:19:17"} +{"current_steps": 1495, "total_steps": 1955, "loss": 1.0115, "lr": 1.2756951267964644e-05, "epoch": 3.8235294117647056, "percentage": 76.47, "elapsed_time": "7:31:43", "remaining_time": "2:18:59"} +{"current_steps": 1496, "total_steps": 1955, "loss": 0.9817, "lr": 1.2704685172631193e-05, "epoch": 3.8260869565217392, "percentage": 76.52, "elapsed_time": "7:32:01", "remaining_time": "2:18:41"} +{"current_steps": 1497, "total_steps": 1955, "loss": 0.9941, "lr": 1.2652506144794612e-05, "epoch": 3.8286445012787724, "percentage": 76.57, "elapsed_time": "7:32:19", "remaining_time": "2:18:23"} +{"current_steps": 1498, "total_steps": 1955, "loss": 0.9637, "lr": 1.2600414350897272e-05, "epoch": 3.8312020460358056, "percentage": 76.62, "elapsed_time": "7:32:37", "remaining_time": "2:18:05"} +{"current_steps": 1499, "total_steps": 1955, "loss": 0.9718, "lr": 1.2548409957103296e-05, "epoch": 3.833759590792839, "percentage": 76.68, "elapsed_time": "7:32:55", "remaining_time": "2:17:46"} +{"current_steps": 1500, "total_steps": 1955, "loss": 0.9997, "lr": 1.2496493129298011e-05, "epoch": 3.836317135549872, "percentage": 76.73, "elapsed_time": "7:33:13", "remaining_time": "2:17:28"} +{"current_steps": 1501, "total_steps": 1955, "loss": 0.9797, "lr": 1.2444664033087466e-05, "epoch": 3.8388746803069056, "percentage": 76.78, "elapsed_time": "7:33:31", "remaining_time": "2:17:10"} +{"current_steps": 1502, "total_steps": 1955, "loss": 0.9881, "lr": 1.2392922833797783e-05, "epoch": 3.8414322250639388, "percentage": 76.83, "elapsed_time": "7:33:48", "remaining_time": "2:16:52"} +{"current_steps": 1503, "total_steps": 1955, "loss": 0.9535, "lr": 1.2341269696474747e-05, "epoch": 3.843989769820972, "percentage": 76.88, "elapsed_time": "7:34:06", "remaining_time": "2:16:33"} +{"current_steps": 1504, "total_steps": 1955, "loss": 0.9979, "lr": 1.2289704785883272e-05, "epoch": 3.846547314578005, "percentage": 76.93, "elapsed_time": "7:34:24", "remaining_time": "2:16:15"} +{"current_steps": 1505, "total_steps": 1955, "loss": 0.9988, "lr": 1.2238228266506798e-05, "epoch": 3.8491048593350383, "percentage": 76.98, "elapsed_time": "7:34:42", "remaining_time": "2:15:57"} +{"current_steps": 1506, "total_steps": 1955, "loss": 0.9756, "lr": 1.2186840302546826e-05, "epoch": 3.8516624040920715, "percentage": 77.03, "elapsed_time": "7:35:01", "remaining_time": "2:15:39"} +{"current_steps": 1507, "total_steps": 1955, "loss": 0.9945, "lr": 1.213554105792238e-05, "epoch": 3.8542199488491047, "percentage": 77.08, "elapsed_time": "7:35:19", "remaining_time": "2:15:21"} +{"current_steps": 1508, "total_steps": 1955, "loss": 0.9766, "lr": 1.2084330696269491e-05, "epoch": 3.8567774936061383, "percentage": 77.14, "elapsed_time": "7:35:37", "remaining_time": "2:15:03"} +{"current_steps": 1509, "total_steps": 1955, "loss": 1.004, "lr": 1.2033209380940654e-05, "epoch": 3.8593350383631715, "percentage": 77.19, "elapsed_time": "7:35:55", "remaining_time": "2:14:45"} +{"current_steps": 1510, "total_steps": 1955, "loss": 0.982, "lr": 1.198217727500432e-05, "epoch": 3.8618925831202047, "percentage": 77.24, "elapsed_time": "7:36:13", "remaining_time": "2:14:26"} +{"current_steps": 1511, "total_steps": 1955, "loss": 0.9572, "lr": 1.1931234541244416e-05, "epoch": 3.864450127877238, "percentage": 77.29, "elapsed_time": "7:36:31", "remaining_time": "2:14:08"} +{"current_steps": 1512, "total_steps": 1955, "loss": 0.9429, "lr": 1.1880381342159754e-05, "epoch": 3.867007672634271, "percentage": 77.34, "elapsed_time": "7:36:49", "remaining_time": "2:13:50"} +{"current_steps": 1513, "total_steps": 1955, "loss": 0.9706, "lr": 1.1829617839963508e-05, "epoch": 3.869565217391304, "percentage": 77.39, "elapsed_time": "7:37:07", "remaining_time": "2:13:32"} +{"current_steps": 1514, "total_steps": 1955, "loss": 0.9799, "lr": 1.1778944196582814e-05, "epoch": 3.8721227621483374, "percentage": 77.44, "elapsed_time": "7:37:25", "remaining_time": "2:13:14"} +{"current_steps": 1515, "total_steps": 1955, "loss": 0.9826, "lr": 1.172836057365812e-05, "epoch": 3.874680306905371, "percentage": 77.49, "elapsed_time": "7:37:43", "remaining_time": "2:12:56"} +{"current_steps": 1516, "total_steps": 1955, "loss": 0.993, "lr": 1.1677867132542735e-05, "epoch": 3.877237851662404, "percentage": 77.54, "elapsed_time": "7:38:01", "remaining_time": "2:12:38"} +{"current_steps": 1517, "total_steps": 1955, "loss": 1.0152, "lr": 1.162746403430231e-05, "epoch": 3.8797953964194374, "percentage": 77.6, "elapsed_time": "7:38:19", "remaining_time": "2:12:19"} +{"current_steps": 1518, "total_steps": 1955, "loss": 0.9881, "lr": 1.157715143971431e-05, "epoch": 3.8823529411764706, "percentage": 77.65, "elapsed_time": "7:38:37", "remaining_time": "2:12:01"} +{"current_steps": 1519, "total_steps": 1955, "loss": 1.0059, "lr": 1.1526929509267513e-05, "epoch": 3.8849104859335037, "percentage": 77.7, "elapsed_time": "7:38:55", "remaining_time": "2:11:43"} +{"current_steps": 1520, "total_steps": 1955, "loss": 0.9961, "lr": 1.1476798403161475e-05, "epoch": 3.887468030690537, "percentage": 77.75, "elapsed_time": "7:39:13", "remaining_time": "2:11:25"} +{"current_steps": 1521, "total_steps": 1955, "loss": 0.9726, "lr": 1.1426758281306087e-05, "epoch": 3.89002557544757, "percentage": 77.8, "elapsed_time": "7:39:31", "remaining_time": "2:11:07"} +{"current_steps": 1522, "total_steps": 1955, "loss": 0.943, "lr": 1.1376809303320964e-05, "epoch": 3.8925831202046037, "percentage": 77.85, "elapsed_time": "7:39:50", "remaining_time": "2:10:49"} +{"current_steps": 1523, "total_steps": 1955, "loss": 0.9789, "lr": 1.1326951628535006e-05, "epoch": 3.895140664961637, "percentage": 77.9, "elapsed_time": "7:40:08", "remaining_time": "2:10:31"} +{"current_steps": 1524, "total_steps": 1955, "loss": 0.9748, "lr": 1.1277185415985871e-05, "epoch": 3.89769820971867, "percentage": 77.95, "elapsed_time": "7:40:26", "remaining_time": "2:10:13"} +{"current_steps": 1525, "total_steps": 1955, "loss": 0.9964, "lr": 1.1227510824419468e-05, "epoch": 3.9002557544757033, "percentage": 78.01, "elapsed_time": "7:40:44", "remaining_time": "2:09:54"} +{"current_steps": 1526, "total_steps": 1955, "loss": 0.9832, "lr": 1.1177928012289455e-05, "epoch": 3.9028132992327365, "percentage": 78.06, "elapsed_time": "7:41:02", "remaining_time": "2:09:36"} +{"current_steps": 1527, "total_steps": 1955, "loss": 0.9744, "lr": 1.1128437137756705e-05, "epoch": 3.90537084398977, "percentage": 78.11, "elapsed_time": "7:41:20", "remaining_time": "2:09:18"} +{"current_steps": 1528, "total_steps": 1955, "loss": 0.9882, "lr": 1.1079038358688882e-05, "epoch": 3.907928388746803, "percentage": 78.16, "elapsed_time": "7:41:39", "remaining_time": "2:09:00"} +{"current_steps": 1529, "total_steps": 1955, "loss": 0.9538, "lr": 1.1029731832659811e-05, "epoch": 3.9104859335038364, "percentage": 78.21, "elapsed_time": "7:41:57", "remaining_time": "2:08:42"} +{"current_steps": 1530, "total_steps": 1955, "loss": 0.9822, "lr": 1.0980517716949075e-05, "epoch": 3.9130434782608696, "percentage": 78.26, "elapsed_time": "7:42:15", "remaining_time": "2:08:24"} +{"current_steps": 1531, "total_steps": 1955, "loss": 0.9734, "lr": 1.0931396168541509e-05, "epoch": 3.915601023017903, "percentage": 78.31, "elapsed_time": "7:42:33", "remaining_time": "2:08:06"} +{"current_steps": 1532, "total_steps": 1955, "loss": 0.9877, "lr": 1.088236734412664e-05, "epoch": 3.918158567774936, "percentage": 78.36, "elapsed_time": "7:42:51", "remaining_time": "2:07:48"} +{"current_steps": 1533, "total_steps": 1955, "loss": 0.9846, "lr": 1.0833431400098227e-05, "epoch": 3.920716112531969, "percentage": 78.41, "elapsed_time": "7:43:10", "remaining_time": "2:07:29"} +{"current_steps": 1534, "total_steps": 1955, "loss": 0.9964, "lr": 1.0784588492553768e-05, "epoch": 3.923273657289003, "percentage": 78.47, "elapsed_time": "7:43:28", "remaining_time": "2:07:11"} +{"current_steps": 1535, "total_steps": 1955, "loss": 0.9858, "lr": 1.0735838777293979e-05, "epoch": 3.9258312020460355, "percentage": 78.52, "elapsed_time": "7:43:46", "remaining_time": "2:06:53"} +{"current_steps": 1536, "total_steps": 1955, "loss": 0.9716, "lr": 1.0687182409822311e-05, "epoch": 3.928388746803069, "percentage": 78.57, "elapsed_time": "7:44:04", "remaining_time": "2:06:35"} +{"current_steps": 1537, "total_steps": 1955, "loss": 0.9699, "lr": 1.0638619545344447e-05, "epoch": 3.9309462915601023, "percentage": 78.62, "elapsed_time": "7:44:21", "remaining_time": "2:06:17"} +{"current_steps": 1538, "total_steps": 1955, "loss": 0.9877, "lr": 1.0590150338767828e-05, "epoch": 3.9335038363171355, "percentage": 78.67, "elapsed_time": "7:44:40", "remaining_time": "2:05:59"} +{"current_steps": 1539, "total_steps": 1955, "loss": 0.97, "lr": 1.054177494470114e-05, "epoch": 3.9360613810741687, "percentage": 78.72, "elapsed_time": "7:44:58", "remaining_time": "2:05:41"} +{"current_steps": 1540, "total_steps": 1955, "loss": 0.98, "lr": 1.0493493517453777e-05, "epoch": 3.938618925831202, "percentage": 78.77, "elapsed_time": "7:45:16", "remaining_time": "2:05:22"} +{"current_steps": 1541, "total_steps": 1955, "loss": 0.9714, "lr": 1.044530621103546e-05, "epoch": 3.9411764705882355, "percentage": 78.82, "elapsed_time": "7:45:34", "remaining_time": "2:05:04"} +{"current_steps": 1542, "total_steps": 1955, "loss": 0.9482, "lr": 1.0397213179155647e-05, "epoch": 3.9437340153452687, "percentage": 78.87, "elapsed_time": "7:45:52", "remaining_time": "2:04:46"} +{"current_steps": 1543, "total_steps": 1955, "loss": 0.9727, "lr": 1.034921457522307e-05, "epoch": 3.946291560102302, "percentage": 78.93, "elapsed_time": "7:46:10", "remaining_time": "2:04:28"} +{"current_steps": 1544, "total_steps": 1955, "loss": 0.9548, "lr": 1.0301310552345258e-05, "epoch": 3.948849104859335, "percentage": 78.98, "elapsed_time": "7:46:28", "remaining_time": "2:04:10"} +{"current_steps": 1545, "total_steps": 1955, "loss": 0.9902, "lr": 1.0253501263328075e-05, "epoch": 3.9514066496163682, "percentage": 79.03, "elapsed_time": "7:46:46", "remaining_time": "2:03:52"} +{"current_steps": 1546, "total_steps": 1955, "loss": 0.9627, "lr": 1.020578686067514e-05, "epoch": 3.9539641943734014, "percentage": 79.08, "elapsed_time": "7:47:05", "remaining_time": "2:03:34"} +{"current_steps": 1547, "total_steps": 1955, "loss": 0.9909, "lr": 1.0158167496587428e-05, "epoch": 3.9565217391304346, "percentage": 79.13, "elapsed_time": "7:47:23", "remaining_time": "2:03:16"} +{"current_steps": 1548, "total_steps": 1955, "loss": 0.9693, "lr": 1.0110643322962779e-05, "epoch": 3.959079283887468, "percentage": 79.18, "elapsed_time": "7:47:41", "remaining_time": "2:02:58"} +{"current_steps": 1549, "total_steps": 1955, "loss": 1.0323, "lr": 1.006321449139537e-05, "epoch": 3.9616368286445014, "percentage": 79.23, "elapsed_time": "7:48:00", "remaining_time": "2:02:39"} +{"current_steps": 1550, "total_steps": 1955, "loss": 1.0, "lr": 1.0015881153175244e-05, "epoch": 3.9641943734015346, "percentage": 79.28, "elapsed_time": "7:48:18", "remaining_time": "2:02:21"} +{"current_steps": 1551, "total_steps": 1955, "loss": 0.9892, "lr": 9.968643459287856e-06, "epoch": 3.9667519181585678, "percentage": 79.34, "elapsed_time": "7:48:36", "remaining_time": "2:02:03"} +{"current_steps": 1552, "total_steps": 1955, "loss": 0.9901, "lr": 9.921501560413555e-06, "epoch": 3.969309462915601, "percentage": 79.39, "elapsed_time": "7:48:54", "remaining_time": "2:01:45"} +{"current_steps": 1553, "total_steps": 1955, "loss": 0.9889, "lr": 9.874455606927134e-06, "epoch": 3.971867007672634, "percentage": 79.44, "elapsed_time": "7:49:12", "remaining_time": "2:01:27"} +{"current_steps": 1554, "total_steps": 1955, "loss": 0.9767, "lr": 9.827505748897308e-06, "epoch": 3.9744245524296673, "percentage": 79.49, "elapsed_time": "7:49:30", "remaining_time": "2:01:09"} +{"current_steps": 1555, "total_steps": 1955, "loss": 0.9987, "lr": 9.780652136086322e-06, "epoch": 3.976982097186701, "percentage": 79.54, "elapsed_time": "7:49:48", "remaining_time": "2:00:51"} +{"current_steps": 1556, "total_steps": 1955, "loss": 1.0099, "lr": 9.733894917949365e-06, "epoch": 3.979539641943734, "percentage": 79.59, "elapsed_time": "7:50:06", "remaining_time": "2:00:32"} +{"current_steps": 1557, "total_steps": 1955, "loss": 0.9791, "lr": 9.687234243634127e-06, "epoch": 3.9820971867007673, "percentage": 79.64, "elapsed_time": "7:50:24", "remaining_time": "2:00:14"} +{"current_steps": 1558, "total_steps": 1955, "loss": 0.9931, "lr": 9.640670261980403e-06, "epoch": 3.9846547314578005, "percentage": 79.69, "elapsed_time": "7:50:42", "remaining_time": "1:59:56"} +{"current_steps": 1559, "total_steps": 1955, "loss": 0.9732, "lr": 9.594203121519508e-06, "epoch": 3.9872122762148337, "percentage": 79.74, "elapsed_time": "7:51:00", "remaining_time": "1:59:38"} +{"current_steps": 1560, "total_steps": 1955, "loss": 0.9874, "lr": 9.547832970473858e-06, "epoch": 3.9897698209718673, "percentage": 79.8, "elapsed_time": "7:51:18", "remaining_time": "1:59:20"} +{"current_steps": 1561, "total_steps": 1955, "loss": 0.9869, "lr": 9.501559956756496e-06, "epoch": 3.9923273657289, "percentage": 79.85, "elapsed_time": "7:51:36", "remaining_time": "1:59:02"} +{"current_steps": 1562, "total_steps": 1955, "loss": 1.0008, "lr": 9.45538422797061e-06, "epoch": 3.9948849104859336, "percentage": 79.9, "elapsed_time": "7:51:54", "remaining_time": "1:58:43"} +{"current_steps": 1563, "total_steps": 1955, "loss": 0.9778, "lr": 9.409305931409066e-06, "epoch": 3.997442455242967, "percentage": 79.95, "elapsed_time": "7:52:12", "remaining_time": "1:58:25"} +{"current_steps": 1564, "total_steps": 1955, "loss": 0.9869, "lr": 9.36332521405392e-06, "epoch": 4.0, "percentage": 80.0, "elapsed_time": "7:52:30", "remaining_time": "1:58:07"} +{"current_steps": 1565, "total_steps": 1955, "loss": 0.9557, "lr": 9.317442222576015e-06, "epoch": 4.002557544757034, "percentage": 80.05, "elapsed_time": "7:53:00", "remaining_time": "1:57:52"} +{"current_steps": 1566, "total_steps": 1955, "loss": 0.9685, "lr": 9.27165710333441e-06, "epoch": 4.005115089514066, "percentage": 80.1, "elapsed_time": "7:53:18", "remaining_time": "1:57:34"} +{"current_steps": 1567, "total_steps": 1955, "loss": 0.9439, "lr": 9.225970002375994e-06, "epoch": 4.0076726342711, "percentage": 80.15, "elapsed_time": "7:53:36", "remaining_time": "1:57:16"} +{"current_steps": 1568, "total_steps": 1955, "loss": 0.9612, "lr": 9.180381065435e-06, "epoch": 4.010230179028133, "percentage": 80.2, "elapsed_time": "7:53:54", "remaining_time": "1:56:57"} +{"current_steps": 1569, "total_steps": 1955, "loss": 0.9726, "lr": 9.13489043793251e-06, "epoch": 4.012787723785166, "percentage": 80.26, "elapsed_time": "7:54:12", "remaining_time": "1:56:39"} +{"current_steps": 1570, "total_steps": 1955, "loss": 0.9744, "lr": 9.089498264976044e-06, "epoch": 4.015345268542199, "percentage": 80.31, "elapsed_time": "7:54:31", "remaining_time": "1:56:21"} +{"current_steps": 1571, "total_steps": 1955, "loss": 0.9713, "lr": 9.044204691359035e-06, "epoch": 4.017902813299233, "percentage": 80.36, "elapsed_time": "7:54:49", "remaining_time": "1:56:03"} +{"current_steps": 1572, "total_steps": 1955, "loss": 0.9579, "lr": 8.99900986156046e-06, "epoch": 4.020460358056266, "percentage": 80.41, "elapsed_time": "7:55:08", "remaining_time": "1:55:45"} +{"current_steps": 1573, "total_steps": 1955, "loss": 0.9623, "lr": 8.953913919744237e-06, "epoch": 4.023017902813299, "percentage": 80.46, "elapsed_time": "7:55:25", "remaining_time": "1:55:27"} +{"current_steps": 1574, "total_steps": 1955, "loss": 0.956, "lr": 8.908917009758902e-06, "epoch": 4.025575447570333, "percentage": 80.51, "elapsed_time": "7:55:44", "remaining_time": "1:55:09"} +{"current_steps": 1575, "total_steps": 1955, "loss": 0.9611, "lr": 8.864019275137096e-06, "epoch": 4.028132992327365, "percentage": 80.56, "elapsed_time": "7:56:02", "remaining_time": "1:54:51"} +{"current_steps": 1576, "total_steps": 1955, "loss": 0.9734, "lr": 8.819220859095096e-06, "epoch": 4.030690537084399, "percentage": 80.61, "elapsed_time": "7:56:20", "remaining_time": "1:54:33"} +{"current_steps": 1577, "total_steps": 1955, "loss": 0.9448, "lr": 8.774521904532349e-06, "epoch": 4.033248081841432, "percentage": 80.66, "elapsed_time": "7:56:38", "remaining_time": "1:54:14"} +{"current_steps": 1578, "total_steps": 1955, "loss": 0.9735, "lr": 8.7299225540311e-06, "epoch": 4.035805626598465, "percentage": 80.72, "elapsed_time": "7:56:56", "remaining_time": "1:53:56"} +{"current_steps": 1579, "total_steps": 1955, "loss": 0.9733, "lr": 8.685422949855785e-06, "epoch": 4.038363171355499, "percentage": 80.77, "elapsed_time": "7:57:14", "remaining_time": "1:53:38"} +{"current_steps": 1580, "total_steps": 1955, "loss": 0.9915, "lr": 8.64102323395272e-06, "epoch": 4.040920716112532, "percentage": 80.82, "elapsed_time": "7:57:32", "remaining_time": "1:53:20"} +{"current_steps": 1581, "total_steps": 1955, "loss": 0.9659, "lr": 8.596723547949603e-06, "epoch": 4.043478260869565, "percentage": 80.87, "elapsed_time": "7:57:50", "remaining_time": "1:53:02"} +{"current_steps": 1582, "total_steps": 1955, "loss": 0.952, "lr": 8.552524033155025e-06, "epoch": 4.046035805626598, "percentage": 80.92, "elapsed_time": "7:58:08", "remaining_time": "1:52:44"} +{"current_steps": 1583, "total_steps": 1955, "loss": 0.9458, "lr": 8.508424830558075e-06, "epoch": 4.048593350383632, "percentage": 80.97, "elapsed_time": "7:58:26", "remaining_time": "1:52:25"} +{"current_steps": 1584, "total_steps": 1955, "loss": 0.9638, "lr": 8.464426080827808e-06, "epoch": 4.051150895140665, "percentage": 81.02, "elapsed_time": "7:58:44", "remaining_time": "1:52:07"} +{"current_steps": 1585, "total_steps": 1955, "loss": 0.9592, "lr": 8.42052792431292e-06, "epoch": 4.053708439897698, "percentage": 81.07, "elapsed_time": "7:59:02", "remaining_time": "1:51:49"} +{"current_steps": 1586, "total_steps": 1955, "loss": 1.0004, "lr": 8.376730501041197e-06, "epoch": 4.056265984654732, "percentage": 81.13, "elapsed_time": "7:59:20", "remaining_time": "1:51:31"} +{"current_steps": 1587, "total_steps": 1955, "loss": 0.9268, "lr": 8.333033950719089e-06, "epoch": 4.0588235294117645, "percentage": 81.18, "elapsed_time": "7:59:38", "remaining_time": "1:51:13"} +{"current_steps": 1588, "total_steps": 1955, "loss": 0.9524, "lr": 8.28943841273132e-06, "epoch": 4.061381074168798, "percentage": 81.23, "elapsed_time": "7:59:56", "remaining_time": "1:50:55"} +{"current_steps": 1589, "total_steps": 1955, "loss": 0.9931, "lr": 8.24594402614038e-06, "epoch": 4.063938618925831, "percentage": 81.28, "elapsed_time": "8:00:15", "remaining_time": "1:50:37"} +{"current_steps": 1590, "total_steps": 1955, "loss": 0.9591, "lr": 8.202550929686061e-06, "epoch": 4.0664961636828645, "percentage": 81.33, "elapsed_time": "8:00:33", "remaining_time": "1:50:18"} +{"current_steps": 1591, "total_steps": 1955, "loss": 0.9966, "lr": 8.15925926178513e-06, "epoch": 4.069053708439898, "percentage": 81.38, "elapsed_time": "8:00:51", "remaining_time": "1:50:00"} +{"current_steps": 1592, "total_steps": 1955, "loss": 0.9862, "lr": 8.116069160530772e-06, "epoch": 4.071611253196931, "percentage": 81.43, "elapsed_time": "8:01:10", "remaining_time": "1:49:42"} +{"current_steps": 1593, "total_steps": 1955, "loss": 0.9959, "lr": 8.072980763692197e-06, "epoch": 4.0741687979539645, "percentage": 81.48, "elapsed_time": "8:01:28", "remaining_time": "1:49:24"} +{"current_steps": 1594, "total_steps": 1955, "loss": 0.9983, "lr": 8.029994208714198e-06, "epoch": 4.076726342710997, "percentage": 81.53, "elapsed_time": "8:01:46", "remaining_time": "1:49:06"} +{"current_steps": 1595, "total_steps": 1955, "loss": 0.9592, "lr": 7.987109632716703e-06, "epoch": 4.079283887468031, "percentage": 81.59, "elapsed_time": "8:02:04", "remaining_time": "1:48:48"} +{"current_steps": 1596, "total_steps": 1955, "loss": 0.9853, "lr": 7.94432717249435e-06, "epoch": 4.081841432225064, "percentage": 81.64, "elapsed_time": "8:02:22", "remaining_time": "1:48:30"} +{"current_steps": 1597, "total_steps": 1955, "loss": 0.9715, "lr": 7.901646964516039e-06, "epoch": 4.084398976982097, "percentage": 81.69, "elapsed_time": "8:02:40", "remaining_time": "1:48:12"} +{"current_steps": 1598, "total_steps": 1955, "loss": 0.9395, "lr": 7.859069144924528e-06, "epoch": 4.086956521739131, "percentage": 81.74, "elapsed_time": "8:02:59", "remaining_time": "1:47:54"} +{"current_steps": 1599, "total_steps": 1955, "loss": 0.9437, "lr": 7.816593849535935e-06, "epoch": 4.089514066496164, "percentage": 81.79, "elapsed_time": "8:03:17", "remaining_time": "1:47:35"} +{"current_steps": 1600, "total_steps": 1955, "loss": 0.9752, "lr": 7.77422121383936e-06, "epoch": 4.092071611253197, "percentage": 81.84, "elapsed_time": "8:03:35", "remaining_time": "1:47:17"} +{"current_steps": 1601, "total_steps": 1955, "loss": 0.9709, "lr": 7.731951372996445e-06, "epoch": 4.09462915601023, "percentage": 81.89, "elapsed_time": "8:03:53", "remaining_time": "1:46:59"} +{"current_steps": 1602, "total_steps": 1955, "loss": 0.957, "lr": 7.689784461840916e-06, "epoch": 4.0971867007672635, "percentage": 81.94, "elapsed_time": "8:04:11", "remaining_time": "1:46:41"} +{"current_steps": 1603, "total_steps": 1955, "loss": 0.9433, "lr": 7.647720614878182e-06, "epoch": 4.099744245524296, "percentage": 81.99, "elapsed_time": "8:04:29", "remaining_time": "1:46:23"} +{"current_steps": 1604, "total_steps": 1955, "loss": 0.9843, "lr": 7.605759966284885e-06, "epoch": 4.10230179028133, "percentage": 82.05, "elapsed_time": "8:04:47", "remaining_time": "1:46:05"} +{"current_steps": 1605, "total_steps": 1955, "loss": 0.9695, "lr": 7.563902649908513e-06, "epoch": 4.1048593350383635, "percentage": 82.1, "elapsed_time": "8:05:05", "remaining_time": "1:45:47"} +{"current_steps": 1606, "total_steps": 1955, "loss": 0.9728, "lr": 7.522148799266888e-06, "epoch": 4.107416879795396, "percentage": 82.15, "elapsed_time": "8:05:23", "remaining_time": "1:45:28"} +{"current_steps": 1607, "total_steps": 1955, "loss": 0.9508, "lr": 7.480498547547825e-06, "epoch": 4.10997442455243, "percentage": 82.2, "elapsed_time": "8:05:41", "remaining_time": "1:45:10"} +{"current_steps": 1608, "total_steps": 1955, "loss": 0.9579, "lr": 7.438952027608688e-06, "epoch": 4.112531969309463, "percentage": 82.25, "elapsed_time": "8:05:59", "remaining_time": "1:44:52"} +{"current_steps": 1609, "total_steps": 1955, "loss": 0.9757, "lr": 7.397509371975929e-06, "epoch": 4.115089514066496, "percentage": 82.3, "elapsed_time": "8:06:18", "remaining_time": "1:44:34"} +{"current_steps": 1610, "total_steps": 1955, "loss": 0.9704, "lr": 7.356170712844703e-06, "epoch": 4.117647058823529, "percentage": 82.35, "elapsed_time": "8:06:36", "remaining_time": "1:44:16"} +{"current_steps": 1611, "total_steps": 1955, "loss": 0.9625, "lr": 7.314936182078427e-06, "epoch": 4.120204603580563, "percentage": 82.4, "elapsed_time": "8:06:54", "remaining_time": "1:43:58"} +{"current_steps": 1612, "total_steps": 1955, "loss": 0.9565, "lr": 7.273805911208374e-06, "epoch": 4.122762148337596, "percentage": 82.46, "elapsed_time": "8:07:12", "remaining_time": "1:43:39"} +{"current_steps": 1613, "total_steps": 1955, "loss": 0.9816, "lr": 7.23278003143324e-06, "epoch": 4.125319693094629, "percentage": 82.51, "elapsed_time": "8:07:30", "remaining_time": "1:43:21"} +{"current_steps": 1614, "total_steps": 1955, "loss": 0.9693, "lr": 7.191858673618713e-06, "epoch": 4.127877237851663, "percentage": 82.56, "elapsed_time": "8:07:48", "remaining_time": "1:43:03"} +{"current_steps": 1615, "total_steps": 1955, "loss": 0.9704, "lr": 7.1510419682971235e-06, "epoch": 4.130434782608695, "percentage": 82.61, "elapsed_time": "8:08:06", "remaining_time": "1:42:45"} +{"current_steps": 1616, "total_steps": 1955, "loss": 0.9686, "lr": 7.110330045666938e-06, "epoch": 4.132992327365729, "percentage": 82.66, "elapsed_time": "8:08:24", "remaining_time": "1:42:27"} +{"current_steps": 1617, "total_steps": 1955, "loss": 0.9643, "lr": 7.069723035592368e-06, "epoch": 4.135549872122763, "percentage": 82.71, "elapsed_time": "8:08:42", "remaining_time": "1:42:09"} +{"current_steps": 1618, "total_steps": 1955, "loss": 0.9638, "lr": 7.02922106760302e-06, "epoch": 4.138107416879795, "percentage": 82.76, "elapsed_time": "8:09:00", "remaining_time": "1:41:50"} +{"current_steps": 1619, "total_steps": 1955, "loss": 0.9392, "lr": 6.988824270893402e-06, "epoch": 4.140664961636829, "percentage": 82.81, "elapsed_time": "8:09:17", "remaining_time": "1:41:32"} +{"current_steps": 1620, "total_steps": 1955, "loss": 0.9921, "lr": 6.948532774322552e-06, "epoch": 4.143222506393862, "percentage": 82.86, "elapsed_time": "8:09:36", "remaining_time": "1:41:14"} +{"current_steps": 1621, "total_steps": 1955, "loss": 0.9759, "lr": 6.908346706413605e-06, "epoch": 4.145780051150895, "percentage": 82.92, "elapsed_time": "8:09:54", "remaining_time": "1:40:56"} +{"current_steps": 1622, "total_steps": 1955, "loss": 0.9725, "lr": 6.868266195353439e-06, "epoch": 4.148337595907928, "percentage": 82.97, "elapsed_time": "8:10:12", "remaining_time": "1:40:38"} +{"current_steps": 1623, "total_steps": 1955, "loss": 0.9881, "lr": 6.828291368992159e-06, "epoch": 4.150895140664962, "percentage": 83.02, "elapsed_time": "8:10:30", "remaining_time": "1:40:20"} +{"current_steps": 1624, "total_steps": 1955, "loss": 0.9523, "lr": 6.7884223548427915e-06, "epoch": 4.153452685421995, "percentage": 83.07, "elapsed_time": "8:10:48", "remaining_time": "1:40:02"} +{"current_steps": 1625, "total_steps": 1955, "loss": 0.9646, "lr": 6.748659280080852e-06, "epoch": 4.156010230179028, "percentage": 83.12, "elapsed_time": "8:11:06", "remaining_time": "1:39:43"} +{"current_steps": 1626, "total_steps": 1955, "loss": 0.9764, "lr": 6.709002271543896e-06, "epoch": 4.158567774936062, "percentage": 83.17, "elapsed_time": "8:11:24", "remaining_time": "1:39:25"} +{"current_steps": 1627, "total_steps": 1955, "loss": 0.9688, "lr": 6.669451455731168e-06, "epoch": 4.161125319693094, "percentage": 83.22, "elapsed_time": "8:11:42", "remaining_time": "1:39:07"} +{"current_steps": 1628, "total_steps": 1955, "loss": 0.9832, "lr": 6.630006958803159e-06, "epoch": 4.163682864450128, "percentage": 83.27, "elapsed_time": "8:12:00", "remaining_time": "1:38:49"} +{"current_steps": 1629, "total_steps": 1955, "loss": 0.9865, "lr": 6.590668906581235e-06, "epoch": 4.166240409207161, "percentage": 83.32, "elapsed_time": "8:12:18", "remaining_time": "1:38:31"} +{"current_steps": 1630, "total_steps": 1955, "loss": 0.9745, "lr": 6.551437424547207e-06, "epoch": 4.168797953964194, "percentage": 83.38, "elapsed_time": "8:12:36", "remaining_time": "1:38:13"} +{"current_steps": 1631, "total_steps": 1955, "loss": 0.9926, "lr": 6.5123126378429455e-06, "epoch": 4.171355498721228, "percentage": 83.43, "elapsed_time": "8:12:54", "remaining_time": "1:37:54"} +{"current_steps": 1632, "total_steps": 1955, "loss": 0.9718, "lr": 6.473294671270011e-06, "epoch": 4.173913043478261, "percentage": 83.48, "elapsed_time": "8:13:12", "remaining_time": "1:37:36"} +{"current_steps": 1633, "total_steps": 1955, "loss": 0.9668, "lr": 6.4343836492891974e-06, "epoch": 4.176470588235294, "percentage": 83.53, "elapsed_time": "8:13:30", "remaining_time": "1:37:18"} +{"current_steps": 1634, "total_steps": 1955, "loss": 0.9631, "lr": 6.395579696020138e-06, "epoch": 4.179028132992327, "percentage": 83.58, "elapsed_time": "8:13:48", "remaining_time": "1:37:00"} +{"current_steps": 1635, "total_steps": 1955, "loss": 0.98, "lr": 6.3568829352409975e-06, "epoch": 4.181585677749361, "percentage": 83.63, "elapsed_time": "8:14:06", "remaining_time": "1:36:42"} +{"current_steps": 1636, "total_steps": 1955, "loss": 1.0123, "lr": 6.3182934903879725e-06, "epoch": 4.1841432225063935, "percentage": 83.68, "elapsed_time": "8:14:24", "remaining_time": "1:36:24"} +{"current_steps": 1637, "total_steps": 1955, "loss": 0.9655, "lr": 6.279811484554943e-06, "epoch": 4.186700767263427, "percentage": 83.73, "elapsed_time": "8:14:42", "remaining_time": "1:36:06"} +{"current_steps": 1638, "total_steps": 1955, "loss": 0.9812, "lr": 6.241437040493088e-06, "epoch": 4.189258312020461, "percentage": 83.79, "elapsed_time": "8:15:00", "remaining_time": "1:35:47"} +{"current_steps": 1639, "total_steps": 1955, "loss": 0.9639, "lr": 6.2031702806104735e-06, "epoch": 4.1918158567774935, "percentage": 83.84, "elapsed_time": "8:15:19", "remaining_time": "1:35:29"} +{"current_steps": 1640, "total_steps": 1955, "loss": 0.9572, "lr": 6.165011326971679e-06, "epoch": 4.194373401534527, "percentage": 83.89, "elapsed_time": "8:15:37", "remaining_time": "1:35:11"} +{"current_steps": 1641, "total_steps": 1955, "loss": 0.9358, "lr": 6.1269603012973795e-06, "epoch": 4.19693094629156, "percentage": 83.94, "elapsed_time": "8:15:55", "remaining_time": "1:34:53"} +{"current_steps": 1642, "total_steps": 1955, "loss": 0.9838, "lr": 6.089017324964017e-06, "epoch": 4.1994884910485935, "percentage": 83.99, "elapsed_time": "8:16:13", "remaining_time": "1:34:35"} +{"current_steps": 1643, "total_steps": 1955, "loss": 0.9413, "lr": 6.051182519003336e-06, "epoch": 4.202046035805626, "percentage": 84.04, "elapsed_time": "8:16:31", "remaining_time": "1:34:17"} +{"current_steps": 1644, "total_steps": 1955, "loss": 0.9833, "lr": 6.013456004102058e-06, "epoch": 4.20460358056266, "percentage": 84.09, "elapsed_time": "8:16:49", "remaining_time": "1:33:59"} +{"current_steps": 1645, "total_steps": 1955, "loss": 0.9699, "lr": 5.97583790060146e-06, "epoch": 4.207161125319693, "percentage": 84.14, "elapsed_time": "8:17:07", "remaining_time": "1:33:40"} +{"current_steps": 1646, "total_steps": 1955, "loss": 0.9504, "lr": 5.938328328497016e-06, "epoch": 4.209718670076726, "percentage": 84.19, "elapsed_time": "8:17:25", "remaining_time": "1:33:22"} +{"current_steps": 1647, "total_steps": 1955, "loss": 0.9521, "lr": 5.900927407437995e-06, "epoch": 4.21227621483376, "percentage": 84.25, "elapsed_time": "8:17:43", "remaining_time": "1:33:04"} +{"current_steps": 1648, "total_steps": 1955, "loss": 0.964, "lr": 5.8636352567270804e-06, "epoch": 4.2148337595907925, "percentage": 84.3, "elapsed_time": "8:18:01", "remaining_time": "1:32:46"} +{"current_steps": 1649, "total_steps": 1955, "loss": 0.9479, "lr": 5.8264519953200375e-06, "epoch": 4.217391304347826, "percentage": 84.35, "elapsed_time": "8:18:19", "remaining_time": "1:32:28"} +{"current_steps": 1650, "total_steps": 1955, "loss": 0.9512, "lr": 5.7893777418252325e-06, "epoch": 4.21994884910486, "percentage": 84.4, "elapsed_time": "8:18:37", "remaining_time": "1:32:10"} +{"current_steps": 1651, "total_steps": 1955, "loss": 0.9714, "lr": 5.75241261450334e-06, "epoch": 4.2225063938618925, "percentage": 84.45, "elapsed_time": "8:18:55", "remaining_time": "1:31:52"} +{"current_steps": 1652, "total_steps": 1955, "loss": 0.9621, "lr": 5.715556731266971e-06, "epoch": 4.225063938618926, "percentage": 84.5, "elapsed_time": "8:19:13", "remaining_time": "1:31:33"} +{"current_steps": 1653, "total_steps": 1955, "loss": 0.9744, "lr": 5.678810209680219e-06, "epoch": 4.227621483375959, "percentage": 84.55, "elapsed_time": "8:19:31", "remaining_time": "1:31:15"} +{"current_steps": 1654, "total_steps": 1955, "loss": 0.9681, "lr": 5.642173166958365e-06, "epoch": 4.2301790281329925, "percentage": 84.6, "elapsed_time": "8:19:49", "remaining_time": "1:30:57"} +{"current_steps": 1655, "total_steps": 1955, "loss": 0.9931, "lr": 5.605645719967445e-06, "epoch": 4.232736572890025, "percentage": 84.65, "elapsed_time": "8:20:08", "remaining_time": "1:30:39"} +{"current_steps": 1656, "total_steps": 1955, "loss": 0.968, "lr": 5.5692279852239286e-06, "epoch": 4.235294117647059, "percentage": 84.71, "elapsed_time": "8:20:26", "remaining_time": "1:30:21"} +{"current_steps": 1657, "total_steps": 1955, "loss": 0.9548, "lr": 5.532920078894299e-06, "epoch": 4.2378516624040925, "percentage": 84.76, "elapsed_time": "8:20:44", "remaining_time": "1:30:03"} +{"current_steps": 1658, "total_steps": 1955, "loss": 0.989, "lr": 5.496722116794715e-06, "epoch": 4.240409207161125, "percentage": 84.81, "elapsed_time": "8:21:02", "remaining_time": "1:29:45"} +{"current_steps": 1659, "total_steps": 1955, "loss": 0.9722, "lr": 5.4606342143906386e-06, "epoch": 4.242966751918159, "percentage": 84.86, "elapsed_time": "8:21:20", "remaining_time": "1:29:26"} +{"current_steps": 1660, "total_steps": 1955, "loss": 0.976, "lr": 5.424656486796465e-06, "epoch": 4.245524296675192, "percentage": 84.91, "elapsed_time": "8:21:38", "remaining_time": "1:29:08"} +{"current_steps": 1661, "total_steps": 1955, "loss": 0.9828, "lr": 5.388789048775103e-06, "epoch": 4.248081841432225, "percentage": 84.96, "elapsed_time": "8:21:56", "remaining_time": "1:28:50"} +{"current_steps": 1662, "total_steps": 1955, "loss": 0.9775, "lr": 5.353032014737718e-06, "epoch": 4.250639386189258, "percentage": 85.01, "elapsed_time": "8:22:15", "remaining_time": "1:28:32"} +{"current_steps": 1663, "total_steps": 1955, "loss": 0.9585, "lr": 5.317385498743273e-06, "epoch": 4.253196930946292, "percentage": 85.06, "elapsed_time": "8:22:33", "remaining_time": "1:28:14"} +{"current_steps": 1664, "total_steps": 1955, "loss": 0.9567, "lr": 5.2818496144982025e-06, "epoch": 4.255754475703325, "percentage": 85.12, "elapsed_time": "8:22:51", "remaining_time": "1:27:56"} +{"current_steps": 1665, "total_steps": 1955, "loss": 0.9615, "lr": 5.246424475356034e-06, "epoch": 4.258312020460358, "percentage": 85.17, "elapsed_time": "8:23:09", "remaining_time": "1:27:38"} +{"current_steps": 1666, "total_steps": 1955, "loss": 0.9626, "lr": 5.211110194317073e-06, "epoch": 4.260869565217392, "percentage": 85.22, "elapsed_time": "8:23:27", "remaining_time": "1:27:20"} +{"current_steps": 1667, "total_steps": 1955, "loss": 0.9741, "lr": 5.17590688402795e-06, "epoch": 4.263427109974424, "percentage": 85.27, "elapsed_time": "8:23:45", "remaining_time": "1:27:01"} +{"current_steps": 1668, "total_steps": 1955, "loss": 0.9226, "lr": 5.140814656781352e-06, "epoch": 4.265984654731458, "percentage": 85.32, "elapsed_time": "8:24:03", "remaining_time": "1:26:43"} +{"current_steps": 1669, "total_steps": 1955, "loss": 0.9648, "lr": 5.105833624515639e-06, "epoch": 4.268542199488491, "percentage": 85.37, "elapsed_time": "8:24:21", "remaining_time": "1:26:25"} +{"current_steps": 1670, "total_steps": 1955, "loss": 0.9779, "lr": 5.070963898814456e-06, "epoch": 4.271099744245524, "percentage": 85.42, "elapsed_time": "8:24:39", "remaining_time": "1:26:07"} +{"current_steps": 1671, "total_steps": 1955, "loss": 0.9539, "lr": 5.036205590906403e-06, "epoch": 4.273657289002558, "percentage": 85.47, "elapsed_time": "8:24:58", "remaining_time": "1:25:49"} +{"current_steps": 1672, "total_steps": 1955, "loss": 0.9636, "lr": 5.0015588116646865e-06, "epoch": 4.276214833759591, "percentage": 85.52, "elapsed_time": "8:25:16", "remaining_time": "1:25:31"} +{"current_steps": 1673, "total_steps": 1955, "loss": 0.9201, "lr": 4.967023671606734e-06, "epoch": 4.278772378516624, "percentage": 85.58, "elapsed_time": "8:25:34", "remaining_time": "1:25:13"} +{"current_steps": 1674, "total_steps": 1955, "loss": 0.9603, "lr": 4.932600280893889e-06, "epoch": 4.281329923273657, "percentage": 85.63, "elapsed_time": "8:25:52", "remaining_time": "1:24:55"} +{"current_steps": 1675, "total_steps": 1955, "loss": 0.9763, "lr": 4.8982887493310085e-06, "epoch": 4.283887468030691, "percentage": 85.68, "elapsed_time": "8:26:11", "remaining_time": "1:24:36"} +{"current_steps": 1676, "total_steps": 1955, "loss": 0.9603, "lr": 4.8640891863661614e-06, "epoch": 4.286445012787723, "percentage": 85.73, "elapsed_time": "8:26:29", "remaining_time": "1:24:18"} +{"current_steps": 1677, "total_steps": 1955, "loss": 0.977, "lr": 4.830001701090248e-06, "epoch": 4.289002557544757, "percentage": 85.78, "elapsed_time": "8:26:47", "remaining_time": "1:24:00"} +{"current_steps": 1678, "total_steps": 1955, "loss": 0.9769, "lr": 4.796026402236624e-06, "epoch": 4.291560102301791, "percentage": 85.83, "elapsed_time": "8:27:05", "remaining_time": "1:23:42"} +{"current_steps": 1679, "total_steps": 1955, "loss": 0.9639, "lr": 4.762163398180852e-06, "epoch": 4.294117647058823, "percentage": 85.88, "elapsed_time": "8:27:23", "remaining_time": "1:23:24"} +{"current_steps": 1680, "total_steps": 1955, "loss": 0.9822, "lr": 4.728412796940251e-06, "epoch": 4.296675191815857, "percentage": 85.93, "elapsed_time": "8:27:41", "remaining_time": "1:23:06"} +{"current_steps": 1681, "total_steps": 1955, "loss": 0.9545, "lr": 4.694774706173602e-06, "epoch": 4.29923273657289, "percentage": 85.98, "elapsed_time": "8:27:59", "remaining_time": "1:22:48"} +{"current_steps": 1682, "total_steps": 1955, "loss": 1.0015, "lr": 4.661249233180787e-06, "epoch": 4.301790281329923, "percentage": 86.04, "elapsed_time": "8:28:17", "remaining_time": "1:22:29"} +{"current_steps": 1683, "total_steps": 1955, "loss": 0.9803, "lr": 4.627836484902495e-06, "epoch": 4.304347826086957, "percentage": 86.09, "elapsed_time": "8:28:35", "remaining_time": "1:22:11"} +{"current_steps": 1684, "total_steps": 1955, "loss": 1.0028, "lr": 4.59453656791979e-06, "epoch": 4.30690537084399, "percentage": 86.14, "elapsed_time": "8:28:53", "remaining_time": "1:21:53"} +{"current_steps": 1685, "total_steps": 1955, "loss": 0.9534, "lr": 4.561349588453849e-06, "epoch": 4.309462915601023, "percentage": 86.19, "elapsed_time": "8:29:11", "remaining_time": "1:21:35"} +{"current_steps": 1686, "total_steps": 1955, "loss": 0.9848, "lr": 4.528275652365608e-06, "epoch": 4.312020460358056, "percentage": 86.24, "elapsed_time": "8:29:29", "remaining_time": "1:21:17"} +{"current_steps": 1687, "total_steps": 1955, "loss": 0.9469, "lr": 4.495314865155393e-06, "epoch": 4.31457800511509, "percentage": 86.29, "elapsed_time": "8:29:48", "remaining_time": "1:20:59"} +{"current_steps": 1688, "total_steps": 1955, "loss": 0.9438, "lr": 4.462467331962614e-06, "epoch": 4.3171355498721224, "percentage": 86.34, "elapsed_time": "8:30:07", "remaining_time": "1:20:41"} +{"current_steps": 1689, "total_steps": 1955, "loss": 0.9493, "lr": 4.4297331575654165e-06, "epoch": 4.319693094629156, "percentage": 86.39, "elapsed_time": "8:30:25", "remaining_time": "1:20:23"} +{"current_steps": 1690, "total_steps": 1955, "loss": 1.01, "lr": 4.397112446380347e-06, "epoch": 4.322250639386189, "percentage": 86.45, "elapsed_time": "8:30:43", "remaining_time": "1:20:05"} +{"current_steps": 1691, "total_steps": 1955, "loss": 0.9614, "lr": 4.3646053024620375e-06, "epoch": 4.324808184143222, "percentage": 86.5, "elapsed_time": "8:31:01", "remaining_time": "1:19:46"} +{"current_steps": 1692, "total_steps": 1955, "loss": 0.9589, "lr": 4.33221182950283e-06, "epoch": 4.327365728900256, "percentage": 86.55, "elapsed_time": "8:31:19", "remaining_time": "1:19:28"} +{"current_steps": 1693, "total_steps": 1955, "loss": 0.9734, "lr": 4.299932130832516e-06, "epoch": 4.329923273657289, "percentage": 86.6, "elapsed_time": "8:31:38", "remaining_time": "1:19:10"} +{"current_steps": 1694, "total_steps": 1955, "loss": 0.9647, "lr": 4.267766309417947e-06, "epoch": 4.332480818414322, "percentage": 86.65, "elapsed_time": "8:31:56", "remaining_time": "1:18:52"} +{"current_steps": 1695, "total_steps": 1955, "loss": 0.9531, "lr": 4.235714467862697e-06, "epoch": 4.335038363171355, "percentage": 86.7, "elapsed_time": "8:32:15", "remaining_time": "1:18:34"} +{"current_steps": 1696, "total_steps": 1955, "loss": 0.9931, "lr": 4.2037767084068105e-06, "epoch": 4.337595907928389, "percentage": 86.75, "elapsed_time": "8:32:33", "remaining_time": "1:18:16"} +{"current_steps": 1697, "total_steps": 1955, "loss": 0.9611, "lr": 4.171953132926403e-06, "epoch": 4.340153452685422, "percentage": 86.8, "elapsed_time": "8:32:51", "remaining_time": "1:17:58"} +{"current_steps": 1698, "total_steps": 1955, "loss": 0.9384, "lr": 4.140243842933353e-06, "epoch": 4.342710997442455, "percentage": 86.85, "elapsed_time": "8:33:09", "remaining_time": "1:17:40"} +{"current_steps": 1699, "total_steps": 1955, "loss": 0.9747, "lr": 4.1086489395750326e-06, "epoch": 4.345268542199489, "percentage": 86.91, "elapsed_time": "8:33:27", "remaining_time": "1:17:22"} +{"current_steps": 1700, "total_steps": 1955, "loss": 0.9708, "lr": 4.077168523633881e-06, "epoch": 4.3478260869565215, "percentage": 86.96, "elapsed_time": "8:33:45", "remaining_time": "1:17:03"} +{"current_steps": 1701, "total_steps": 1955, "loss": 0.9354, "lr": 4.045802695527177e-06, "epoch": 4.350383631713555, "percentage": 87.01, "elapsed_time": "8:34:03", "remaining_time": "1:16:45"} +{"current_steps": 1702, "total_steps": 1955, "loss": 0.9849, "lr": 4.014551555306665e-06, "epoch": 4.352941176470588, "percentage": 87.06, "elapsed_time": "8:34:21", "remaining_time": "1:16:27"} +{"current_steps": 1703, "total_steps": 1955, "loss": 0.9995, "lr": 3.983415202658281e-06, "epoch": 4.3554987212276215, "percentage": 87.11, "elapsed_time": "8:34:39", "remaining_time": "1:16:09"} +{"current_steps": 1704, "total_steps": 1955, "loss": 0.9533, "lr": 3.952393736901785e-06, "epoch": 4.358056265984655, "percentage": 87.16, "elapsed_time": "8:34:57", "remaining_time": "1:15:51"} +{"current_steps": 1705, "total_steps": 1955, "loss": 0.9708, "lr": 3.921487256990473e-06, "epoch": 4.360613810741688, "percentage": 87.21, "elapsed_time": "8:35:15", "remaining_time": "1:15:33"} +{"current_steps": 1706, "total_steps": 1955, "loss": 0.9814, "lr": 3.890695861510856e-06, "epoch": 4.3631713554987215, "percentage": 87.26, "elapsed_time": "8:35:33", "remaining_time": "1:15:14"} +{"current_steps": 1707, "total_steps": 1955, "loss": 0.9717, "lr": 3.8600196486823405e-06, "epoch": 4.365728900255754, "percentage": 87.31, "elapsed_time": "8:35:51", "remaining_time": "1:14:56"} +{"current_steps": 1708, "total_steps": 1955, "loss": 0.9598, "lr": 3.8294587163569194e-06, "epoch": 4.368286445012788, "percentage": 87.37, "elapsed_time": "8:36:09", "remaining_time": "1:14:38"} +{"current_steps": 1709, "total_steps": 1955, "loss": 0.9837, "lr": 3.799013162018881e-06, "epoch": 4.370843989769821, "percentage": 87.42, "elapsed_time": "8:36:27", "remaining_time": "1:14:20"} +{"current_steps": 1710, "total_steps": 1955, "loss": 0.9721, "lr": 3.7686830827844544e-06, "epoch": 4.373401534526854, "percentage": 87.47, "elapsed_time": "8:36:45", "remaining_time": "1:14:02"} +{"current_steps": 1711, "total_steps": 1955, "loss": 0.9946, "lr": 3.7384685754015126e-06, "epoch": 4.375959079283888, "percentage": 87.52, "elapsed_time": "8:37:04", "remaining_time": "1:13:44"} +{"current_steps": 1712, "total_steps": 1955, "loss": 0.9938, "lr": 3.7083697362493066e-06, "epoch": 4.378516624040921, "percentage": 87.57, "elapsed_time": "8:37:22", "remaining_time": "1:13:26"} +{"current_steps": 1713, "total_steps": 1955, "loss": 0.9597, "lr": 3.6783866613381025e-06, "epoch": 4.381074168797954, "percentage": 87.62, "elapsed_time": "8:37:40", "remaining_time": "1:13:07"} +{"current_steps": 1714, "total_steps": 1955, "loss": 0.9816, "lr": 3.6485194463089022e-06, "epoch": 4.383631713554987, "percentage": 87.67, "elapsed_time": "8:37:58", "remaining_time": "1:12:49"} +{"current_steps": 1715, "total_steps": 1955, "loss": 0.9774, "lr": 3.6187681864331235e-06, "epoch": 4.3861892583120206, "percentage": 87.72, "elapsed_time": "8:38:16", "remaining_time": "1:12:31"} +{"current_steps": 1716, "total_steps": 1955, "loss": 0.9663, "lr": 3.5891329766123505e-06, "epoch": 4.388746803069053, "percentage": 87.77, "elapsed_time": "8:38:34", "remaining_time": "1:12:13"} +{"current_steps": 1717, "total_steps": 1955, "loss": 1.0028, "lr": 3.5596139113779215e-06, "epoch": 4.391304347826087, "percentage": 87.83, "elapsed_time": "8:38:52", "remaining_time": "1:11:55"} +{"current_steps": 1718, "total_steps": 1955, "loss": 0.9469, "lr": 3.5302110848907336e-06, "epoch": 4.3938618925831205, "percentage": 87.88, "elapsed_time": "8:39:10", "remaining_time": "1:11:37"} +{"current_steps": 1719, "total_steps": 1955, "loss": 0.9587, "lr": 3.500924590940904e-06, "epoch": 4.396419437340153, "percentage": 87.93, "elapsed_time": "8:39:28", "remaining_time": "1:11:19"} +{"current_steps": 1720, "total_steps": 1955, "loss": 0.9788, "lr": 3.4717545229474524e-06, "epoch": 4.398976982097187, "percentage": 87.98, "elapsed_time": "8:39:46", "remaining_time": "1:11:00"} +{"current_steps": 1721, "total_steps": 1955, "loss": 0.9962, "lr": 3.442700973958024e-06, "epoch": 4.40153452685422, "percentage": 88.03, "elapsed_time": "8:40:04", "remaining_time": "1:10:42"} +{"current_steps": 1722, "total_steps": 1955, "loss": 0.9824, "lr": 3.4137640366485924e-06, "epoch": 4.404092071611253, "percentage": 88.08, "elapsed_time": "8:40:22", "remaining_time": "1:10:24"} +{"current_steps": 1723, "total_steps": 1955, "loss": 0.9865, "lr": 3.384943803323153e-06, "epoch": 4.406649616368286, "percentage": 88.13, "elapsed_time": "8:40:40", "remaining_time": "1:10:06"} +{"current_steps": 1724, "total_steps": 1955, "loss": 0.9551, "lr": 3.356240365913439e-06, "epoch": 4.40920716112532, "percentage": 88.18, "elapsed_time": "8:40:58", "remaining_time": "1:09:48"} +{"current_steps": 1725, "total_steps": 1955, "loss": 0.9706, "lr": 3.3276538159786197e-06, "epoch": 4.411764705882353, "percentage": 88.24, "elapsed_time": "8:41:16", "remaining_time": "1:09:30"} +{"current_steps": 1726, "total_steps": 1955, "loss": 0.9697, "lr": 3.299184244705025e-06, "epoch": 4.414322250639386, "percentage": 88.29, "elapsed_time": "8:41:34", "remaining_time": "1:09:12"} +{"current_steps": 1727, "total_steps": 1955, "loss": 0.9589, "lr": 3.2708317429058334e-06, "epoch": 4.41687979539642, "percentage": 88.34, "elapsed_time": "8:41:52", "remaining_time": "1:08:53"} +{"current_steps": 1728, "total_steps": 1955, "loss": 0.9588, "lr": 3.2425964010207765e-06, "epoch": 4.419437340153452, "percentage": 88.39, "elapsed_time": "8:42:10", "remaining_time": "1:08:35"} +{"current_steps": 1729, "total_steps": 1955, "loss": 0.9641, "lr": 3.214478309115894e-06, "epoch": 4.421994884910486, "percentage": 88.44, "elapsed_time": "8:42:29", "remaining_time": "1:08:17"} +{"current_steps": 1730, "total_steps": 1955, "loss": 1.0013, "lr": 3.1864775568832027e-06, "epoch": 4.42455242966752, "percentage": 88.49, "elapsed_time": "8:42:47", "remaining_time": "1:07:59"} +{"current_steps": 1731, "total_steps": 1955, "loss": 0.9535, "lr": 3.1585942336404262e-06, "epoch": 4.427109974424552, "percentage": 88.54, "elapsed_time": "8:43:05", "remaining_time": "1:07:41"} +{"current_steps": 1732, "total_steps": 1955, "loss": 0.9435, "lr": 3.13082842833071e-06, "epoch": 4.429667519181586, "percentage": 88.59, "elapsed_time": "8:43:23", "remaining_time": "1:07:23"} +{"current_steps": 1733, "total_steps": 1955, "loss": 0.9757, "lr": 3.1031802295223355e-06, "epoch": 4.432225063938619, "percentage": 88.64, "elapsed_time": "8:43:41", "remaining_time": "1:07:05"} +{"current_steps": 1734, "total_steps": 1955, "loss": 0.9593, "lr": 3.075649725408445e-06, "epoch": 4.434782608695652, "percentage": 88.7, "elapsed_time": "8:43:58", "remaining_time": "1:06:46"} +{"current_steps": 1735, "total_steps": 1955, "loss": 0.9811, "lr": 3.048237003806738e-06, "epoch": 4.437340153452685, "percentage": 88.75, "elapsed_time": "8:44:17", "remaining_time": "1:06:28"} +{"current_steps": 1736, "total_steps": 1955, "loss": 1.0008, "lr": 3.0209421521592318e-06, "epoch": 4.439897698209719, "percentage": 88.8, "elapsed_time": "8:44:35", "remaining_time": "1:06:10"} +{"current_steps": 1737, "total_steps": 1955, "loss": 0.9636, "lr": 2.9937652575319444e-06, "epoch": 4.442455242966752, "percentage": 88.85, "elapsed_time": "8:44:53", "remaining_time": "1:05:52"} +{"current_steps": 1738, "total_steps": 1955, "loss": 0.9639, "lr": 2.966706406614628e-06, "epoch": 4.445012787723785, "percentage": 88.9, "elapsed_time": "8:45:11", "remaining_time": "1:05:34"} +{"current_steps": 1739, "total_steps": 1955, "loss": 0.983, "lr": 2.939765685720497e-06, "epoch": 4.447570332480819, "percentage": 88.95, "elapsed_time": "8:45:29", "remaining_time": "1:05:16"} +{"current_steps": 1740, "total_steps": 1955, "loss": 0.9378, "lr": 2.912943180785961e-06, "epoch": 4.450127877237851, "percentage": 89.0, "elapsed_time": "8:45:47", "remaining_time": "1:04:58"} +{"current_steps": 1741, "total_steps": 1955, "loss": 0.9564, "lr": 2.886238977370326e-06, "epoch": 4.452685421994885, "percentage": 89.05, "elapsed_time": "8:46:05", "remaining_time": "1:04:39"} +{"current_steps": 1742, "total_steps": 1955, "loss": 1.0017, "lr": 2.8596531606555377e-06, "epoch": 4.455242966751918, "percentage": 89.1, "elapsed_time": "8:46:23", "remaining_time": "1:04:21"} +{"current_steps": 1743, "total_steps": 1955, "loss": 0.956, "lr": 2.833185815445929e-06, "epoch": 4.457800511508951, "percentage": 89.16, "elapsed_time": "8:46:41", "remaining_time": "1:04:03"} +{"current_steps": 1744, "total_steps": 1955, "loss": 0.9627, "lr": 2.806837026167899e-06, "epoch": 4.460358056265985, "percentage": 89.21, "elapsed_time": "8:46:59", "remaining_time": "1:03:45"} +{"current_steps": 1745, "total_steps": 1955, "loss": 0.9723, "lr": 2.780606876869678e-06, "epoch": 4.462915601023018, "percentage": 89.26, "elapsed_time": "8:47:17", "remaining_time": "1:03:27"} +{"current_steps": 1746, "total_steps": 1955, "loss": 0.9769, "lr": 2.7544954512210842e-06, "epoch": 4.465473145780051, "percentage": 89.31, "elapsed_time": "8:47:35", "remaining_time": "1:03:09"} +{"current_steps": 1747, "total_steps": 1955, "loss": 0.963, "lr": 2.7285028325131892e-06, "epoch": 4.468030690537084, "percentage": 89.36, "elapsed_time": "8:47:53", "remaining_time": "1:02:51"} +{"current_steps": 1748, "total_steps": 1955, "loss": 0.9682, "lr": 2.702629103658114e-06, "epoch": 4.470588235294118, "percentage": 89.41, "elapsed_time": "8:48:11", "remaining_time": "1:02:32"} +{"current_steps": 1749, "total_steps": 1955, "loss": 0.9715, "lr": 2.6768743471887337e-06, "epoch": 4.4731457800511505, "percentage": 89.46, "elapsed_time": "8:48:29", "remaining_time": "1:02:14"} +{"current_steps": 1750, "total_steps": 1955, "loss": 0.9953, "lr": 2.651238645258416e-06, "epoch": 4.475703324808184, "percentage": 89.51, "elapsed_time": "8:48:47", "remaining_time": "1:01:56"} +{"current_steps": 1751, "total_steps": 1955, "loss": 0.9606, "lr": 2.6257220796407756e-06, "epoch": 4.478260869565218, "percentage": 89.57, "elapsed_time": "8:49:05", "remaining_time": "1:01:38"} +{"current_steps": 1752, "total_steps": 1955, "loss": 0.9759, "lr": 2.6003247317293847e-06, "epoch": 4.4808184143222505, "percentage": 89.62, "elapsed_time": "8:49:23", "remaining_time": "1:01:20"} +{"current_steps": 1753, "total_steps": 1955, "loss": 0.9872, "lr": 2.5750466825375585e-06, "epoch": 4.483375959079284, "percentage": 89.67, "elapsed_time": "8:49:41", "remaining_time": "1:01:02"} +{"current_steps": 1754, "total_steps": 1955, "loss": 0.988, "lr": 2.5498880126980518e-06, "epoch": 4.485933503836317, "percentage": 89.72, "elapsed_time": "8:49:59", "remaining_time": "1:00:44"} +{"current_steps": 1755, "total_steps": 1955, "loss": 0.9636, "lr": 2.5248488024628024e-06, "epoch": 4.4884910485933505, "percentage": 89.77, "elapsed_time": "8:50:17", "remaining_time": "1:00:25"} +{"current_steps": 1756, "total_steps": 1955, "loss": 0.9773, "lr": 2.499929131702734e-06, "epoch": 4.491048593350383, "percentage": 89.82, "elapsed_time": "8:50:35", "remaining_time": "1:00:07"} +{"current_steps": 1757, "total_steps": 1955, "loss": 0.9769, "lr": 2.4751290799074213e-06, "epoch": 4.493606138107417, "percentage": 89.87, "elapsed_time": "8:50:53", "remaining_time": "0:59:49"} +{"current_steps": 1758, "total_steps": 1955, "loss": 0.9523, "lr": 2.450448726184891e-06, "epoch": 4.4961636828644505, "percentage": 89.92, "elapsed_time": "8:51:11", "remaining_time": "0:59:31"} +{"current_steps": 1759, "total_steps": 1955, "loss": 0.9389, "lr": 2.425888149261346e-06, "epoch": 4.498721227621483, "percentage": 89.97, "elapsed_time": "8:51:29", "remaining_time": "0:59:13"} +{"current_steps": 1760, "total_steps": 1955, "loss": 0.9895, "lr": 2.401447427480941e-06, "epoch": 4.501278772378517, "percentage": 90.03, "elapsed_time": "8:51:47", "remaining_time": "0:58:55"} +{"current_steps": 1761, "total_steps": 1955, "loss": 0.9817, "lr": 2.3771266388054846e-06, "epoch": 4.5038363171355495, "percentage": 90.08, "elapsed_time": "8:52:05", "remaining_time": "0:58:37"} +{"current_steps": 1762, "total_steps": 1955, "loss": 0.9939, "lr": 2.3529258608142326e-06, "epoch": 4.506393861892583, "percentage": 90.13, "elapsed_time": "8:52:23", "remaining_time": "0:58:18"} +{"current_steps": 1763, "total_steps": 1955, "loss": 0.9793, "lr": 2.3288451707036288e-06, "epoch": 4.508951406649617, "percentage": 90.18, "elapsed_time": "8:52:42", "remaining_time": "0:58:00"} +{"current_steps": 1764, "total_steps": 1955, "loss": 0.9554, "lr": 2.3048846452870597e-06, "epoch": 4.5115089514066495, "percentage": 90.23, "elapsed_time": "8:53:00", "remaining_time": "0:57:42"} +{"current_steps": 1765, "total_steps": 1955, "loss": 0.9705, "lr": 2.281044360994589e-06, "epoch": 4.514066496163683, "percentage": 90.28, "elapsed_time": "8:53:18", "remaining_time": "0:57:24"} +{"current_steps": 1766, "total_steps": 1955, "loss": 0.9569, "lr": 2.257324393872753e-06, "epoch": 4.516624040920716, "percentage": 90.33, "elapsed_time": "8:53:36", "remaining_time": "0:57:06"} +{"current_steps": 1767, "total_steps": 1955, "loss": 0.9875, "lr": 2.233724819584282e-06, "epoch": 4.5191815856777495, "percentage": 90.38, "elapsed_time": "8:53:54", "remaining_time": "0:56:48"} +{"current_steps": 1768, "total_steps": 1955, "loss": 0.9851, "lr": 2.210245713407879e-06, "epoch": 4.521739130434782, "percentage": 90.43, "elapsed_time": "8:54:12", "remaining_time": "0:56:30"} +{"current_steps": 1769, "total_steps": 1955, "loss": 0.9434, "lr": 2.186887150237964e-06, "epoch": 4.524296675191816, "percentage": 90.49, "elapsed_time": "8:54:30", "remaining_time": "0:56:12"} +{"current_steps": 1770, "total_steps": 1955, "loss": 0.9676, "lr": 2.163649204584464e-06, "epoch": 4.526854219948849, "percentage": 90.54, "elapsed_time": "8:54:48", "remaining_time": "0:55:53"} +{"current_steps": 1771, "total_steps": 1955, "loss": 0.9625, "lr": 2.140531950572551e-06, "epoch": 4.529411764705882, "percentage": 90.59, "elapsed_time": "8:55:06", "remaining_time": "0:55:35"} +{"current_steps": 1772, "total_steps": 1955, "loss": 0.9667, "lr": 2.117535461942386e-06, "epoch": 4.531969309462916, "percentage": 90.64, "elapsed_time": "8:55:24", "remaining_time": "0:55:17"} +{"current_steps": 1773, "total_steps": 1955, "loss": 0.9548, "lr": 2.094659812048945e-06, "epoch": 4.534526854219949, "percentage": 90.69, "elapsed_time": "8:55:42", "remaining_time": "0:54:59"} +{"current_steps": 1774, "total_steps": 1955, "loss": 0.9586, "lr": 2.071905073861724e-06, "epoch": 4.537084398976982, "percentage": 90.74, "elapsed_time": "8:56:00", "remaining_time": "0:54:41"} +{"current_steps": 1775, "total_steps": 1955, "loss": 0.9482, "lr": 2.049271319964543e-06, "epoch": 4.539641943734015, "percentage": 90.79, "elapsed_time": "8:56:18", "remaining_time": "0:54:23"} +{"current_steps": 1776, "total_steps": 1955, "loss": 0.9571, "lr": 2.0267586225552892e-06, "epoch": 4.542199488491049, "percentage": 90.84, "elapsed_time": "8:56:36", "remaining_time": "0:54:05"} +{"current_steps": 1777, "total_steps": 1955, "loss": 0.9737, "lr": 2.0043670534457194e-06, "epoch": 4.544757033248082, "percentage": 90.9, "elapsed_time": "8:56:54", "remaining_time": "0:53:46"} +{"current_steps": 1778, "total_steps": 1955, "loss": 0.9848, "lr": 1.982096684061192e-06, "epoch": 4.547314578005115, "percentage": 90.95, "elapsed_time": "8:57:12", "remaining_time": "0:53:28"} +{"current_steps": 1779, "total_steps": 1955, "loss": 0.9695, "lr": 1.959947585440469e-06, "epoch": 4.549872122762149, "percentage": 91.0, "elapsed_time": "8:57:31", "remaining_time": "0:53:10"} +{"current_steps": 1780, "total_steps": 1955, "loss": 0.9621, "lr": 1.9379198282354794e-06, "epoch": 4.552429667519181, "percentage": 91.05, "elapsed_time": "8:57:49", "remaining_time": "0:52:52"} +{"current_steps": 1781, "total_steps": 1955, "loss": 0.9935, "lr": 1.916013482711092e-06, "epoch": 4.554987212276215, "percentage": 91.1, "elapsed_time": "8:58:07", "remaining_time": "0:52:34"} +{"current_steps": 1782, "total_steps": 1955, "loss": 0.9679, "lr": 1.8942286187448865e-06, "epoch": 4.557544757033249, "percentage": 91.15, "elapsed_time": "8:58:25", "remaining_time": "0:52:16"} +{"current_steps": 1783, "total_steps": 1955, "loss": 1.0187, "lr": 1.8725653058269433e-06, "epoch": 4.560102301790281, "percentage": 91.2, "elapsed_time": "8:58:43", "remaining_time": "0:51:58"} +{"current_steps": 1784, "total_steps": 1955, "loss": 0.9438, "lr": 1.8510236130596126e-06, "epoch": 4.562659846547315, "percentage": 91.25, "elapsed_time": "8:59:01", "remaining_time": "0:51:39"} +{"current_steps": 1785, "total_steps": 1955, "loss": 0.9453, "lr": 1.8296036091573022e-06, "epoch": 4.565217391304348, "percentage": 91.3, "elapsed_time": "8:59:19", "remaining_time": "0:51:21"} +{"current_steps": 1786, "total_steps": 1955, "loss": 0.9531, "lr": 1.8083053624462365e-06, "epoch": 4.567774936061381, "percentage": 91.36, "elapsed_time": "8:59:38", "remaining_time": "0:51:03"} +{"current_steps": 1787, "total_steps": 1955, "loss": 0.9635, "lr": 1.7871289408642889e-06, "epoch": 4.570332480818414, "percentage": 91.41, "elapsed_time": "8:59:56", "remaining_time": "0:50:45"} +{"current_steps": 1788, "total_steps": 1955, "loss": 0.9531, "lr": 1.7660744119606876e-06, "epoch": 4.572890025575448, "percentage": 91.46, "elapsed_time": "9:00:14", "remaining_time": "0:50:27"} +{"current_steps": 1789, "total_steps": 1955, "loss": 1.0048, "lr": 1.745141842895861e-06, "epoch": 4.57544757033248, "percentage": 91.51, "elapsed_time": "9:00:32", "remaining_time": "0:50:09"} +{"current_steps": 1790, "total_steps": 1955, "loss": 0.9783, "lr": 1.7243313004412288e-06, "epoch": 4.578005115089514, "percentage": 91.56, "elapsed_time": "9:00:50", "remaining_time": "0:49:51"} +{"current_steps": 1791, "total_steps": 1955, "loss": 0.9669, "lr": 1.7036428509789394e-06, "epoch": 4.580562659846548, "percentage": 91.61, "elapsed_time": "9:01:08", "remaining_time": "0:49:33"} +{"current_steps": 1792, "total_steps": 1955, "loss": 0.9688, "lr": 1.6830765605016842e-06, "epoch": 4.58312020460358, "percentage": 91.66, "elapsed_time": "9:01:26", "remaining_time": "0:49:14"} +{"current_steps": 1793, "total_steps": 1955, "loss": 0.962, "lr": 1.6626324946125061e-06, "epoch": 4.585677749360614, "percentage": 91.71, "elapsed_time": "9:01:44", "remaining_time": "0:48:56"} +{"current_steps": 1794, "total_steps": 1955, "loss": 0.972, "lr": 1.6423107185245557e-06, "epoch": 4.588235294117647, "percentage": 91.76, "elapsed_time": "9:02:02", "remaining_time": "0:48:38"} +{"current_steps": 1795, "total_steps": 1955, "loss": 0.9526, "lr": 1.6221112970609088e-06, "epoch": 4.59079283887468, "percentage": 91.82, "elapsed_time": "9:02:20", "remaining_time": "0:48:20"} +{"current_steps": 1796, "total_steps": 1955, "loss": 0.9635, "lr": 1.60203429465434e-06, "epoch": 4.593350383631714, "percentage": 91.87, "elapsed_time": "9:02:39", "remaining_time": "0:48:02"} +{"current_steps": 1797, "total_steps": 1955, "loss": 0.971, "lr": 1.5820797753471495e-06, "epoch": 4.595907928388747, "percentage": 91.92, "elapsed_time": "9:02:57", "remaining_time": "0:47:44"} +{"current_steps": 1798, "total_steps": 1955, "loss": 0.9482, "lr": 1.562247802790915e-06, "epoch": 4.59846547314578, "percentage": 91.97, "elapsed_time": "9:03:15", "remaining_time": "0:47:26"} +{"current_steps": 1799, "total_steps": 1955, "loss": 0.9568, "lr": 1.542538440246304e-06, "epoch": 4.601023017902813, "percentage": 92.02, "elapsed_time": "9:03:33", "remaining_time": "0:47:08"} +{"current_steps": 1800, "total_steps": 1955, "loss": 0.9892, "lr": 1.5229517505829105e-06, "epoch": 4.603580562659847, "percentage": 92.07, "elapsed_time": "9:03:51", "remaining_time": "0:46:49"} +{"current_steps": 1801, "total_steps": 1955, "loss": 0.9494, "lr": 1.5034877962789973e-06, "epoch": 4.6061381074168795, "percentage": 92.12, "elapsed_time": "9:04:09", "remaining_time": "0:46:31"} +{"current_steps": 1802, "total_steps": 1955, "loss": 0.9392, "lr": 1.4841466394213311e-06, "epoch": 4.608695652173913, "percentage": 92.17, "elapsed_time": "9:04:27", "remaining_time": "0:46:13"} +{"current_steps": 1803, "total_steps": 1955, "loss": 0.9633, "lr": 1.4649283417049653e-06, "epoch": 4.611253196930946, "percentage": 92.23, "elapsed_time": "9:04:46", "remaining_time": "0:45:55"} +{"current_steps": 1804, "total_steps": 1955, "loss": 0.9312, "lr": 1.445832964433076e-06, "epoch": 4.6138107416879794, "percentage": 92.28, "elapsed_time": "9:05:04", "remaining_time": "0:45:37"} +{"current_steps": 1805, "total_steps": 1955, "loss": 0.9746, "lr": 1.4268605685167125e-06, "epoch": 4.616368286445013, "percentage": 92.33, "elapsed_time": "9:05:22", "remaining_time": "0:45:19"} +{"current_steps": 1806, "total_steps": 1955, "loss": 0.9383, "lr": 1.4080112144746516e-06, "epoch": 4.618925831202046, "percentage": 92.38, "elapsed_time": "9:05:40", "remaining_time": "0:45:01"} +{"current_steps": 1807, "total_steps": 1955, "loss": 0.9678, "lr": 1.3892849624331973e-06, "epoch": 4.621483375959079, "percentage": 92.43, "elapsed_time": "9:05:58", "remaining_time": "0:44:43"} +{"current_steps": 1808, "total_steps": 1955, "loss": 1.0104, "lr": 1.3706818721259584e-06, "epoch": 4.624040920716112, "percentage": 92.48, "elapsed_time": "9:06:16", "remaining_time": "0:44:24"} +{"current_steps": 1809, "total_steps": 1955, "loss": 0.9806, "lr": 1.35220200289369e-06, "epoch": 4.626598465473146, "percentage": 92.53, "elapsed_time": "9:06:34", "remaining_time": "0:44:06"} +{"current_steps": 1810, "total_steps": 1955, "loss": 0.9677, "lr": 1.3338454136840872e-06, "epoch": 4.629156010230179, "percentage": 92.58, "elapsed_time": "9:06:52", "remaining_time": "0:43:48"} +{"current_steps": 1811, "total_steps": 1955, "loss": 0.9676, "lr": 1.315612163051605e-06, "epoch": 4.631713554987212, "percentage": 92.63, "elapsed_time": "9:07:10", "remaining_time": "0:43:30"} +{"current_steps": 1812, "total_steps": 1955, "loss": 0.994, "lr": 1.2975023091572657e-06, "epoch": 4.634271099744246, "percentage": 92.69, "elapsed_time": "9:07:28", "remaining_time": "0:43:12"} +{"current_steps": 1813, "total_steps": 1955, "loss": 0.9345, "lr": 1.279515909768483e-06, "epoch": 4.6368286445012785, "percentage": 92.74, "elapsed_time": "9:07:46", "remaining_time": "0:42:54"} +{"current_steps": 1814, "total_steps": 1955, "loss": 1.0019, "lr": 1.2616530222588685e-06, "epoch": 4.639386189258312, "percentage": 92.79, "elapsed_time": "9:08:04", "remaining_time": "0:42:36"} +{"current_steps": 1815, "total_steps": 1955, "loss": 0.9927, "lr": 1.2439137036080573e-06, "epoch": 4.641943734015345, "percentage": 92.84, "elapsed_time": "9:08:22", "remaining_time": "0:42:17"} +{"current_steps": 1816, "total_steps": 1955, "loss": 0.9925, "lr": 1.2262980104014921e-06, "epoch": 4.6445012787723785, "percentage": 92.89, "elapsed_time": "9:08:41", "remaining_time": "0:41:59"} +{"current_steps": 1817, "total_steps": 1955, "loss": 0.9895, "lr": 1.208805998830318e-06, "epoch": 4.647058823529412, "percentage": 92.94, "elapsed_time": "9:08:59", "remaining_time": "0:41:41"} +{"current_steps": 1818, "total_steps": 1955, "loss": 0.9783, "lr": 1.1914377246911158e-06, "epoch": 4.649616368286445, "percentage": 92.99, "elapsed_time": "9:09:17", "remaining_time": "0:41:23"} +{"current_steps": 1819, "total_steps": 1955, "loss": 0.9549, "lr": 1.174193243385786e-06, "epoch": 4.6521739130434785, "percentage": 93.04, "elapsed_time": "9:09:35", "remaining_time": "0:41:05"} +{"current_steps": 1820, "total_steps": 1955, "loss": 0.943, "lr": 1.1570726099213414e-06, "epoch": 4.654731457800511, "percentage": 93.09, "elapsed_time": "9:09:53", "remaining_time": "0:40:47"} +{"current_steps": 1821, "total_steps": 1955, "loss": 0.9443, "lr": 1.1400758789097499e-06, "epoch": 4.657289002557545, "percentage": 93.15, "elapsed_time": "9:10:11", "remaining_time": "0:40:29"} +{"current_steps": 1822, "total_steps": 1955, "loss": 0.9987, "lr": 1.1232031045677405e-06, "epoch": 4.659846547314578, "percentage": 93.2, "elapsed_time": "9:10:29", "remaining_time": "0:40:11"} +{"current_steps": 1823, "total_steps": 1955, "loss": 0.9463, "lr": 1.106454340716643e-06, "epoch": 4.662404092071611, "percentage": 93.25, "elapsed_time": "9:10:47", "remaining_time": "0:39:52"} +{"current_steps": 1824, "total_steps": 1955, "loss": 0.9599, "lr": 1.089829640782223e-06, "epoch": 4.664961636828645, "percentage": 93.3, "elapsed_time": "9:11:05", "remaining_time": "0:39:34"} +{"current_steps": 1825, "total_steps": 1955, "loss": 0.9626, "lr": 1.073329057794501e-06, "epoch": 4.667519181585678, "percentage": 93.35, "elapsed_time": "9:11:23", "remaining_time": "0:39:16"} +{"current_steps": 1826, "total_steps": 1955, "loss": 0.9586, "lr": 1.0569526443875745e-06, "epoch": 4.670076726342711, "percentage": 93.4, "elapsed_time": "9:11:41", "remaining_time": "0:38:58"} +{"current_steps": 1827, "total_steps": 1955, "loss": 0.9603, "lr": 1.0407004527994701e-06, "epoch": 4.672634271099744, "percentage": 93.45, "elapsed_time": "9:11:59", "remaining_time": "0:38:40"} +{"current_steps": 1828, "total_steps": 1955, "loss": 0.9526, "lr": 1.0245725348719681e-06, "epoch": 4.675191815856778, "percentage": 93.5, "elapsed_time": "9:12:17", "remaining_time": "0:38:22"} +{"current_steps": 1829, "total_steps": 1955, "loss": 0.958, "lr": 1.0085689420504274e-06, "epoch": 4.677749360613811, "percentage": 93.55, "elapsed_time": "9:12:35", "remaining_time": "0:38:04"} +{"current_steps": 1830, "total_steps": 1955, "loss": 0.9825, "lr": 9.926897253836442e-07, "epoch": 4.680306905370844, "percentage": 93.61, "elapsed_time": "9:12:54", "remaining_time": "0:37:45"} +{"current_steps": 1831, "total_steps": 1955, "loss": 0.9671, "lr": 9.769349355236702e-07, "epoch": 4.6828644501278776, "percentage": 93.66, "elapsed_time": "9:13:12", "remaining_time": "0:37:27"} +{"current_steps": 1832, "total_steps": 1955, "loss": 0.9883, "lr": 9.613046227256472e-07, "epoch": 4.68542199488491, "percentage": 93.71, "elapsed_time": "9:13:30", "remaining_time": "0:37:09"} +{"current_steps": 1833, "total_steps": 1955, "loss": 0.9578, "lr": 9.457988368476579e-07, "epoch": 4.687979539641944, "percentage": 93.76, "elapsed_time": "9:13:49", "remaining_time": "0:36:51"} +{"current_steps": 1834, "total_steps": 1955, "loss": 1.0142, "lr": 9.304176273505861e-07, "epoch": 4.690537084398977, "percentage": 93.81, "elapsed_time": "9:14:07", "remaining_time": "0:36:33"} +{"current_steps": 1835, "total_steps": 1955, "loss": 0.9775, "lr": 9.151610432979097e-07, "epoch": 4.69309462915601, "percentage": 93.86, "elapsed_time": "9:14:25", "remaining_time": "0:36:15"} +{"current_steps": 1836, "total_steps": 1955, "loss": 0.9895, "lr": 9.000291333555888e-07, "epoch": 4.695652173913043, "percentage": 93.91, "elapsed_time": "9:14:43", "remaining_time": "0:35:57"} +{"current_steps": 1837, "total_steps": 1955, "loss": 0.973, "lr": 8.850219457918929e-07, "epoch": 4.698209718670077, "percentage": 93.96, "elapsed_time": "9:15:01", "remaining_time": "0:35:39"} +{"current_steps": 1838, "total_steps": 1955, "loss": 0.9425, "lr": 8.701395284772407e-07, "epoch": 4.70076726342711, "percentage": 94.02, "elapsed_time": "9:15:19", "remaining_time": "0:35:20"} +{"current_steps": 1839, "total_steps": 1955, "loss": 0.939, "lr": 8.553819288840537e-07, "epoch": 4.703324808184143, "percentage": 94.07, "elapsed_time": "9:15:37", "remaining_time": "0:35:02"} +{"current_steps": 1840, "total_steps": 1955, "loss": 0.9976, "lr": 8.40749194086623e-07, "epoch": 4.705882352941177, "percentage": 94.12, "elapsed_time": "9:15:55", "remaining_time": "0:34:44"} +{"current_steps": 1841, "total_steps": 1955, "loss": 0.9576, "lr": 8.262413707609185e-07, "epoch": 4.708439897698209, "percentage": 94.17, "elapsed_time": "9:16:13", "remaining_time": "0:34:26"} +{"current_steps": 1842, "total_steps": 1955, "loss": 0.9554, "lr": 8.118585051844774e-07, "epoch": 4.710997442455243, "percentage": 94.22, "elapsed_time": "9:16:31", "remaining_time": "0:34:08"} +{"current_steps": 1843, "total_steps": 1955, "loss": 0.9571, "lr": 7.976006432362449e-07, "epoch": 4.713554987212277, "percentage": 94.27, "elapsed_time": "9:16:49", "remaining_time": "0:33:50"} +{"current_steps": 1844, "total_steps": 1955, "loss": 0.9781, "lr": 7.834678303964094e-07, "epoch": 4.716112531969309, "percentage": 94.32, "elapsed_time": "9:17:07", "remaining_time": "0:33:32"} +{"current_steps": 1845, "total_steps": 1955, "loss": 0.9743, "lr": 7.694601117462919e-07, "epoch": 4.718670076726343, "percentage": 94.37, "elapsed_time": "9:17:25", "remaining_time": "0:33:14"} +{"current_steps": 1846, "total_steps": 1955, "loss": 0.9642, "lr": 7.555775319681724e-07, "epoch": 4.721227621483376, "percentage": 94.42, "elapsed_time": "9:17:43", "remaining_time": "0:32:55"} +{"current_steps": 1847, "total_steps": 1955, "loss": 0.9681, "lr": 7.41820135345166e-07, "epoch": 4.723785166240409, "percentage": 94.48, "elapsed_time": "9:18:01", "remaining_time": "0:32:37"} +{"current_steps": 1848, "total_steps": 1955, "loss": 0.9974, "lr": 7.281879657610758e-07, "epoch": 4.726342710997442, "percentage": 94.53, "elapsed_time": "9:18:19", "remaining_time": "0:32:19"} +{"current_steps": 1849, "total_steps": 1955, "loss": 0.9972, "lr": 7.146810667002291e-07, "epoch": 4.728900255754476, "percentage": 94.58, "elapsed_time": "9:18:38", "remaining_time": "0:32:01"} +{"current_steps": 1850, "total_steps": 1955, "loss": 0.965, "lr": 7.012994812473928e-07, "epoch": 4.731457800511509, "percentage": 94.63, "elapsed_time": "9:18:56", "remaining_time": "0:31:43"} +{"current_steps": 1851, "total_steps": 1955, "loss": 0.9666, "lr": 6.880432520875868e-07, "epoch": 4.734015345268542, "percentage": 94.68, "elapsed_time": "9:19:14", "remaining_time": "0:31:25"} +{"current_steps": 1852, "total_steps": 1955, "loss": 0.9584, "lr": 6.749124215059733e-07, "epoch": 4.736572890025576, "percentage": 94.73, "elapsed_time": "9:19:32", "remaining_time": "0:31:07"} +{"current_steps": 1853, "total_steps": 1955, "loss": 0.9999, "lr": 6.619070313877007e-07, "epoch": 4.739130434782608, "percentage": 94.78, "elapsed_time": "9:19:50", "remaining_time": "0:30:49"} +{"current_steps": 1854, "total_steps": 1955, "loss": 0.9938, "lr": 6.490271232177981e-07, "epoch": 4.741687979539642, "percentage": 94.83, "elapsed_time": "9:20:08", "remaining_time": "0:30:30"} +{"current_steps": 1855, "total_steps": 1955, "loss": 0.97, "lr": 6.362727380810185e-07, "epoch": 4.744245524296675, "percentage": 94.88, "elapsed_time": "9:20:26", "remaining_time": "0:30:12"} +{"current_steps": 1856, "total_steps": 1955, "loss": 0.9377, "lr": 6.236439166617204e-07, "epoch": 4.746803069053708, "percentage": 94.94, "elapsed_time": "9:20:44", "remaining_time": "0:29:54"} +{"current_steps": 1857, "total_steps": 1955, "loss": 0.9583, "lr": 6.111406992437374e-07, "epoch": 4.749360613810742, "percentage": 94.99, "elapsed_time": "9:21:02", "remaining_time": "0:29:36"} +{"current_steps": 1858, "total_steps": 1955, "loss": 0.9879, "lr": 5.98763125710251e-07, "epoch": 4.751918158567775, "percentage": 95.04, "elapsed_time": "9:21:20", "remaining_time": "0:29:18"} +{"current_steps": 1859, "total_steps": 1955, "loss": 0.9596, "lr": 5.865112355436431e-07, "epoch": 4.754475703324808, "percentage": 95.09, "elapsed_time": "9:21:38", "remaining_time": "0:29:00"} +{"current_steps": 1860, "total_steps": 1955, "loss": 0.9843, "lr": 5.74385067825407e-07, "epoch": 4.757033248081841, "percentage": 95.14, "elapsed_time": "9:21:56", "remaining_time": "0:28:42"} +{"current_steps": 1861, "total_steps": 1955, "loss": 0.9781, "lr": 5.623846612359884e-07, "epoch": 4.759590792838875, "percentage": 95.19, "elapsed_time": "9:22:14", "remaining_time": "0:28:23"} +{"current_steps": 1862, "total_steps": 1955, "loss": 0.97, "lr": 5.505100540546827e-07, "epoch": 4.762148337595908, "percentage": 95.24, "elapsed_time": "9:22:32", "remaining_time": "0:28:05"} +{"current_steps": 1863, "total_steps": 1955, "loss": 0.9848, "lr": 5.387612841595014e-07, "epoch": 4.764705882352941, "percentage": 95.29, "elapsed_time": "9:22:51", "remaining_time": "0:27:47"} +{"current_steps": 1864, "total_steps": 1955, "loss": 0.9915, "lr": 5.271383890270621e-07, "epoch": 4.767263427109975, "percentage": 95.35, "elapsed_time": "9:23:08", "remaining_time": "0:27:29"} +{"current_steps": 1865, "total_steps": 1955, "loss": 0.958, "lr": 5.156414057324588e-07, "epoch": 4.7698209718670075, "percentage": 95.4, "elapsed_time": "9:23:27", "remaining_time": "0:27:11"} +{"current_steps": 1866, "total_steps": 1955, "loss": 0.9827, "lr": 5.042703709491425e-07, "epoch": 4.772378516624041, "percentage": 95.45, "elapsed_time": "9:23:45", "remaining_time": "0:26:53"} +{"current_steps": 1867, "total_steps": 1955, "loss": 0.9591, "lr": 4.930253209488278e-07, "epoch": 4.774936061381074, "percentage": 95.5, "elapsed_time": "9:24:03", "remaining_time": "0:26:35"} +{"current_steps": 1868, "total_steps": 1955, "loss": 0.9598, "lr": 4.819062916013329e-07, "epoch": 4.7774936061381075, "percentage": 95.55, "elapsed_time": "9:24:21", "remaining_time": "0:26:17"} +{"current_steps": 1869, "total_steps": 1955, "loss": 0.974, "lr": 4.7091331837451736e-07, "epoch": 4.78005115089514, "percentage": 95.6, "elapsed_time": "9:24:39", "remaining_time": "0:25:58"} +{"current_steps": 1870, "total_steps": 1955, "loss": 0.9639, "lr": 4.600464363341228e-07, "epoch": 4.782608695652174, "percentage": 95.65, "elapsed_time": "9:24:57", "remaining_time": "0:25:40"} +{"current_steps": 1871, "total_steps": 1955, "loss": 1.0149, "lr": 4.493056801436879e-07, "epoch": 4.7851662404092075, "percentage": 95.7, "elapsed_time": "9:25:15", "remaining_time": "0:25:22"} +{"current_steps": 1872, "total_steps": 1955, "loss": 0.9489, "lr": 4.386910840644376e-07, "epoch": 4.78772378516624, "percentage": 95.75, "elapsed_time": "9:25:33", "remaining_time": "0:25:04"} +{"current_steps": 1873, "total_steps": 1955, "loss": 0.9543, "lr": 4.2820268195515434e-07, "epoch": 4.790281329923274, "percentage": 95.81, "elapsed_time": "9:25:51", "remaining_time": "0:24:46"} +{"current_steps": 1874, "total_steps": 1955, "loss": 0.9422, "lr": 4.178405072720937e-07, "epoch": 4.792838874680307, "percentage": 95.86, "elapsed_time": "9:26:09", "remaining_time": "0:24:28"} +{"current_steps": 1875, "total_steps": 1955, "loss": 0.9719, "lr": 4.076045930688599e-07, "epoch": 4.79539641943734, "percentage": 95.91, "elapsed_time": "9:26:27", "remaining_time": "0:24:10"} +{"current_steps": 1876, "total_steps": 1955, "loss": 0.9398, "lr": 3.9749497199631724e-07, "epoch": 4.797953964194374, "percentage": 95.96, "elapsed_time": "9:26:45", "remaining_time": "0:23:52"} +{"current_steps": 1877, "total_steps": 1955, "loss": 0.9813, "lr": 3.875116763024611e-07, "epoch": 4.8005115089514065, "percentage": 96.01, "elapsed_time": "9:27:03", "remaining_time": "0:23:33"} +{"current_steps": 1878, "total_steps": 1955, "loss": 0.9674, "lr": 3.7765473783233807e-07, "epoch": 4.80306905370844, "percentage": 96.06, "elapsed_time": "9:27:21", "remaining_time": "0:23:15"} +{"current_steps": 1879, "total_steps": 1955, "loss": 0.952, "lr": 3.6792418802794386e-07, "epoch": 4.805626598465473, "percentage": 96.11, "elapsed_time": "9:27:39", "remaining_time": "0:22:57"} +{"current_steps": 1880, "total_steps": 1955, "loss": 0.9479, "lr": 3.583200579280943e-07, "epoch": 4.8081841432225065, "percentage": 96.16, "elapsed_time": "9:27:57", "remaining_time": "0:22:39"} +{"current_steps": 1881, "total_steps": 1955, "loss": 0.979, "lr": 3.4884237816837696e-07, "epoch": 4.810741687979539, "percentage": 96.21, "elapsed_time": "9:28:15", "remaining_time": "0:22:21"} +{"current_steps": 1882, "total_steps": 1955, "loss": 0.9489, "lr": 3.394911789809996e-07, "epoch": 4.813299232736573, "percentage": 96.27, "elapsed_time": "9:28:33", "remaining_time": "0:22:03"} +{"current_steps": 1883, "total_steps": 1955, "loss": 0.9401, "lr": 3.3026649019471946e-07, "epoch": 4.8158567774936065, "percentage": 96.32, "elapsed_time": "9:28:51", "remaining_time": "0:21:45"} +{"current_steps": 1884, "total_steps": 1955, "loss": 0.9613, "lr": 3.2116834123476326e-07, "epoch": 4.818414322250639, "percentage": 96.37, "elapsed_time": "9:29:09", "remaining_time": "0:21:26"} +{"current_steps": 1885, "total_steps": 1955, "loss": 0.962, "lr": 3.1219676112270723e-07, "epoch": 4.820971867007673, "percentage": 96.42, "elapsed_time": "9:29:28", "remaining_time": "0:21:08"} +{"current_steps": 1886, "total_steps": 1955, "loss": 0.9778, "lr": 3.033517784763884e-07, "epoch": 4.823529411764706, "percentage": 96.47, "elapsed_time": "9:29:45", "remaining_time": "0:20:50"} +{"current_steps": 1887, "total_steps": 1955, "loss": 0.9765, "lr": 2.94633421509829e-07, "epoch": 4.826086956521739, "percentage": 96.52, "elapsed_time": "9:30:04", "remaining_time": "0:20:32"} +{"current_steps": 1888, "total_steps": 1955, "loss": 0.9748, "lr": 2.8604171803312986e-07, "epoch": 4.828644501278772, "percentage": 96.57, "elapsed_time": "9:30:22", "remaining_time": "0:20:14"} +{"current_steps": 1889, "total_steps": 1955, "loss": 0.9611, "lr": 2.775766954523906e-07, "epoch": 4.831202046035806, "percentage": 96.62, "elapsed_time": "9:30:40", "remaining_time": "0:19:56"} +{"current_steps": 1890, "total_steps": 1955, "loss": 0.941, "lr": 2.692383807696164e-07, "epoch": 4.833759590792839, "percentage": 96.68, "elapsed_time": "9:30:58", "remaining_time": "0:19:38"} +{"current_steps": 1891, "total_steps": 1955, "loss": 0.977, "lr": 2.6102680058264664e-07, "epoch": 4.836317135549872, "percentage": 96.73, "elapsed_time": "9:31:16", "remaining_time": "0:19:20"} +{"current_steps": 1892, "total_steps": 1955, "loss": 0.9424, "lr": 2.5294198108504865e-07, "epoch": 4.838874680306906, "percentage": 96.78, "elapsed_time": "9:31:34", "remaining_time": "0:19:01"} +{"current_steps": 1893, "total_steps": 1955, "loss": 0.9217, "lr": 2.449839480660421e-07, "epoch": 4.841432225063938, "percentage": 96.83, "elapsed_time": "9:31:52", "remaining_time": "0:18:43"} +{"current_steps": 1894, "total_steps": 1955, "loss": 0.9712, "lr": 2.371527269104279e-07, "epoch": 4.843989769820972, "percentage": 96.88, "elapsed_time": "9:32:10", "remaining_time": "0:18:25"} +{"current_steps": 1895, "total_steps": 1955, "loss": 0.961, "lr": 2.2944834259849058e-07, "epoch": 4.846547314578006, "percentage": 96.93, "elapsed_time": "9:32:28", "remaining_time": "0:18:07"} +{"current_steps": 1896, "total_steps": 1955, "loss": 0.9934, "lr": 2.2187081970593605e-07, "epoch": 4.849104859335038, "percentage": 96.98, "elapsed_time": "9:32:46", "remaining_time": "0:17:49"} +{"current_steps": 1897, "total_steps": 1955, "loss": 0.9582, "lr": 2.1442018240378948e-07, "epoch": 4.851662404092072, "percentage": 97.03, "elapsed_time": "9:33:04", "remaining_time": "0:17:31"} +{"current_steps": 1898, "total_steps": 1955, "loss": 0.9714, "lr": 2.0709645445834646e-07, "epoch": 4.854219948849105, "percentage": 97.08, "elapsed_time": "9:33:22", "remaining_time": "0:17:13"} +{"current_steps": 1899, "total_steps": 1955, "loss": 1.0043, "lr": 1.998996592310709e-07, "epoch": 4.856777493606138, "percentage": 97.14, "elapsed_time": "9:33:40", "remaining_time": "0:16:55"} +{"current_steps": 1900, "total_steps": 1955, "loss": 0.9557, "lr": 1.928298196785372e-07, "epoch": 4.859335038363171, "percentage": 97.19, "elapsed_time": "9:33:58", "remaining_time": "0:16:36"} +{"current_steps": 1901, "total_steps": 1955, "loss": 0.9563, "lr": 1.8588695835235926e-07, "epoch": 4.861892583120205, "percentage": 97.24, "elapsed_time": "9:34:16", "remaining_time": "0:16:18"} +{"current_steps": 1902, "total_steps": 1955, "loss": 0.9618, "lr": 1.790710973991061e-07, "epoch": 4.864450127877237, "percentage": 97.29, "elapsed_time": "9:34:34", "remaining_time": "0:16:00"} +{"current_steps": 1903, "total_steps": 1955, "loss": 0.9745, "lr": 1.723822585602264e-07, "epoch": 4.867007672634271, "percentage": 97.34, "elapsed_time": "9:34:52", "remaining_time": "0:15:42"} +{"current_steps": 1904, "total_steps": 1955, "loss": 0.9847, "lr": 1.658204631720084e-07, "epoch": 4.869565217391305, "percentage": 97.39, "elapsed_time": "9:35:10", "remaining_time": "0:15:24"} +{"current_steps": 1905, "total_steps": 1955, "loss": 0.9789, "lr": 1.5938573216547792e-07, "epoch": 4.872122762148337, "percentage": 97.44, "elapsed_time": "9:35:28", "remaining_time": "0:15:06"} +{"current_steps": 1906, "total_steps": 1955, "loss": 0.9662, "lr": 1.5307808606635832e-07, "epoch": 4.874680306905371, "percentage": 97.49, "elapsed_time": "9:35:46", "remaining_time": "0:14:48"} +{"current_steps": 1907, "total_steps": 1955, "loss": 0.9715, "lr": 1.4689754499498166e-07, "epoch": 4.877237851662404, "percentage": 97.54, "elapsed_time": "9:36:04", "remaining_time": "0:14:30"} +{"current_steps": 1908, "total_steps": 1955, "loss": 0.9947, "lr": 1.408441286662443e-07, "epoch": 4.879795396419437, "percentage": 97.6, "elapsed_time": "9:36:22", "remaining_time": "0:14:11"} +{"current_steps": 1909, "total_steps": 1955, "loss": 1.0021, "lr": 1.3491785638953592e-07, "epoch": 4.882352941176471, "percentage": 97.65, "elapsed_time": "9:36:40", "remaining_time": "0:13:53"} +{"current_steps": 1910, "total_steps": 1955, "loss": 0.9644, "lr": 1.291187470686728e-07, "epoch": 4.884910485933504, "percentage": 97.7, "elapsed_time": "9:36:58", "remaining_time": "0:13:35"} +{"current_steps": 1911, "total_steps": 1955, "loss": 0.9445, "lr": 1.2344681920184455e-07, "epoch": 4.887468030690537, "percentage": 97.75, "elapsed_time": "9:37:16", "remaining_time": "0:13:17"} +{"current_steps": 1912, "total_steps": 1955, "loss": 0.987, "lr": 1.1790209088155646e-07, "epoch": 4.89002557544757, "percentage": 97.8, "elapsed_time": "9:37:34", "remaining_time": "0:12:59"} +{"current_steps": 1913, "total_steps": 1955, "loss": 0.9693, "lr": 1.1248457979456729e-07, "epoch": 4.892583120204604, "percentage": 97.85, "elapsed_time": "9:37:52", "remaining_time": "0:12:41"} +{"current_steps": 1914, "total_steps": 1955, "loss": 0.9783, "lr": 1.0719430322183144e-07, "epoch": 4.8951406649616365, "percentage": 97.9, "elapsed_time": "9:38:10", "remaining_time": "0:12:23"} +{"current_steps": 1915, "total_steps": 1955, "loss": 0.9698, "lr": 1.0203127803844137e-07, "epoch": 4.89769820971867, "percentage": 97.95, "elapsed_time": "9:38:28", "remaining_time": "0:12:04"} +{"current_steps": 1916, "total_steps": 1955, "loss": 0.9935, "lr": 9.699552071359642e-08, "epoch": 4.900255754475703, "percentage": 98.01, "elapsed_time": "9:38:46", "remaining_time": "0:11:46"} +{"current_steps": 1917, "total_steps": 1955, "loss": 0.9509, "lr": 9.208704731051399e-08, "epoch": 4.9028132992327365, "percentage": 98.06, "elapsed_time": "9:39:05", "remaining_time": "0:11:28"} +{"current_steps": 1918, "total_steps": 1955, "loss": 0.9255, "lr": 8.73058734864074e-08, "epoch": 4.90537084398977, "percentage": 98.11, "elapsed_time": "9:39:23", "remaining_time": "0:11:10"} +{"current_steps": 1919, "total_steps": 1955, "loss": 0.9974, "lr": 8.265201449242367e-08, "epoch": 4.907928388746803, "percentage": 98.16, "elapsed_time": "9:39:41", "remaining_time": "0:10:52"} +{"current_steps": 1920, "total_steps": 1955, "loss": 0.9625, "lr": 7.812548517359908e-08, "epoch": 4.910485933503836, "percentage": 98.21, "elapsed_time": "9:39:59", "remaining_time": "0:10:34"} +{"current_steps": 1921, "total_steps": 1955, "loss": 0.9778, "lr": 7.37262999687971e-08, "epoch": 4.913043478260869, "percentage": 98.26, "elapsed_time": "9:40:17", "remaining_time": "0:10:16"} +{"current_steps": 1922, "total_steps": 1955, "loss": 0.9712, "lr": 6.945447291069052e-08, "epoch": 4.915601023017903, "percentage": 98.31, "elapsed_time": "9:40:35", "remaining_time": "0:09:58"} +{"current_steps": 1923, "total_steps": 1955, "loss": 0.9557, "lr": 6.531001762569044e-08, "epoch": 4.918158567774936, "percentage": 98.36, "elapsed_time": "9:40:53", "remaining_time": "0:09:39"} +{"current_steps": 1924, "total_steps": 1955, "loss": 0.9581, "lr": 6.129294733391966e-08, "epoch": 4.920716112531969, "percentage": 98.41, "elapsed_time": "9:41:11", "remaining_time": "0:09:21"} +{"current_steps": 1925, "total_steps": 1955, "loss": 0.9918, "lr": 5.740327484916819e-08, "epoch": 4.923273657289003, "percentage": 98.47, "elapsed_time": "9:41:29", "remaining_time": "0:09:03"} +{"current_steps": 1926, "total_steps": 1955, "loss": 0.9898, "lr": 5.364101257883114e-08, "epoch": 4.9258312020460355, "percentage": 98.52, "elapsed_time": "9:41:47", "remaining_time": "0:08:45"} +{"current_steps": 1927, "total_steps": 1955, "loss": 0.9926, "lr": 5.000617252389983e-08, "epoch": 4.928388746803069, "percentage": 98.57, "elapsed_time": "9:42:05", "remaining_time": "0:08:27"} +{"current_steps": 1928, "total_steps": 1955, "loss": 0.9555, "lr": 4.649876627890848e-08, "epoch": 4.930946291560103, "percentage": 98.62, "elapsed_time": "9:42:23", "remaining_time": "0:08:09"} +{"current_steps": 1929, "total_steps": 1955, "loss": 0.9761, "lr": 4.311880503189869e-08, "epoch": 4.9335038363171355, "percentage": 98.67, "elapsed_time": "9:42:41", "remaining_time": "0:07:51"} +{"current_steps": 1930, "total_steps": 1955, "loss": 0.9981, "lr": 3.986629956437948e-08, "epoch": 4.936061381074169, "percentage": 98.72, "elapsed_time": "9:42:59", "remaining_time": "0:07:33"} +{"current_steps": 1931, "total_steps": 1955, "loss": 0.9968, "lr": 3.674126025130509e-08, "epoch": 4.938618925831202, "percentage": 98.77, "elapsed_time": "9:43:17", "remaining_time": "0:07:14"} +{"current_steps": 1932, "total_steps": 1955, "loss": 0.978, "lr": 3.374369706102609e-08, "epoch": 4.9411764705882355, "percentage": 98.82, "elapsed_time": "9:43:35", "remaining_time": "0:06:56"} +{"current_steps": 1933, "total_steps": 1955, "loss": 0.9466, "lr": 3.0873619555267244e-08, "epoch": 4.943734015345268, "percentage": 98.87, "elapsed_time": "9:43:53", "remaining_time": "0:06:38"} +{"current_steps": 1934, "total_steps": 1955, "loss": 0.9722, "lr": 2.8131036889096352e-08, "epoch": 4.946291560102302, "percentage": 98.93, "elapsed_time": "9:44:11", "remaining_time": "0:06:20"} +{"current_steps": 1935, "total_steps": 1955, "loss": 0.956, "lr": 2.5515957810897663e-08, "epoch": 4.948849104859335, "percentage": 98.98, "elapsed_time": "9:44:30", "remaining_time": "0:06:02"} +{"current_steps": 1936, "total_steps": 1955, "loss": 0.9641, "lr": 2.302839066233631e-08, "epoch": 4.951406649616368, "percentage": 99.03, "elapsed_time": "9:44:48", "remaining_time": "0:05:44"} +{"current_steps": 1937, "total_steps": 1955, "loss": 0.9671, "lr": 2.066834337833168e-08, "epoch": 4.953964194373402, "percentage": 99.08, "elapsed_time": "9:45:06", "remaining_time": "0:05:26"} +{"current_steps": 1938, "total_steps": 1955, "loss": 0.9409, "lr": 1.843582348704409e-08, "epoch": 4.956521739130435, "percentage": 99.13, "elapsed_time": "9:45:24", "remaining_time": "0:05:08"} +{"current_steps": 1939, "total_steps": 1955, "loss": 0.9782, "lr": 1.6330838109839266e-08, "epoch": 4.959079283887468, "percentage": 99.18, "elapsed_time": "9:45:42", "remaining_time": "0:04:49"} +{"current_steps": 1940, "total_steps": 1955, "loss": 0.9874, "lr": 1.4353393961270556e-08, "epoch": 4.961636828644501, "percentage": 99.23, "elapsed_time": "9:46:00", "remaining_time": "0:04:31"} +{"current_steps": 1941, "total_steps": 1955, "loss": 0.9653, "lr": 1.2503497349047877e-08, "epoch": 4.964194373401535, "percentage": 99.28, "elapsed_time": "9:46:18", "remaining_time": "0:04:13"} +{"current_steps": 1942, "total_steps": 1955, "loss": 0.977, "lr": 1.078115417404213e-08, "epoch": 4.966751918158568, "percentage": 99.34, "elapsed_time": "9:46:36", "remaining_time": "0:03:55"} +{"current_steps": 1943, "total_steps": 1955, "loss": 0.9729, "lr": 9.18636993023192e-09, "epoch": 4.969309462915601, "percentage": 99.39, "elapsed_time": "9:46:54", "remaining_time": "0:03:37"} +{"current_steps": 1944, "total_steps": 1955, "loss": 0.9452, "lr": 7.719149704716878e-09, "epoch": 4.971867007672635, "percentage": 99.44, "elapsed_time": "9:47:12", "remaining_time": "0:03:19"} +{"current_steps": 1945, "total_steps": 1955, "loss": 0.9638, "lr": 6.379498177686572e-09, "epoch": 4.974424552429667, "percentage": 99.49, "elapsed_time": "9:47:30", "remaining_time": "0:03:01"} +{"current_steps": 1946, "total_steps": 1955, "loss": 1.0006, "lr": 5.167419622402747e-09, "epoch": 4.976982097186701, "percentage": 99.54, "elapsed_time": "9:47:48", "remaining_time": "0:02:43"} +{"current_steps": 1947, "total_steps": 1955, "loss": 0.9504, "lr": 4.082917905194883e-09, "epoch": 4.979539641943734, "percentage": 99.59, "elapsed_time": "9:48:06", "remaining_time": "0:02:24"} +{"current_steps": 1948, "total_steps": 1955, "loss": 0.9662, "lr": 3.1259964854379877e-09, "epoch": 4.982097186700767, "percentage": 99.64, "elapsed_time": "9:48:24", "remaining_time": "0:02:06"} +{"current_steps": 1949, "total_steps": 1955, "loss": 0.9928, "lr": 2.2966584155570403e-09, "epoch": 4.9846547314578, "percentage": 99.69, "elapsed_time": "9:48:42", "remaining_time": "0:01:48"} +{"current_steps": 1950, "total_steps": 1955, "loss": 0.9916, "lr": 1.5949063410003462e-09, "epoch": 4.987212276214834, "percentage": 99.74, "elapsed_time": "9:49:00", "remaining_time": "0:01:30"} +{"current_steps": 1951, "total_steps": 1955, "loss": 0.9555, "lr": 1.0207425002395355e-09, "epoch": 4.989769820971867, "percentage": 99.8, "elapsed_time": "9:49:18", "remaining_time": "0:01:12"} +{"current_steps": 1952, "total_steps": 1955, "loss": 0.9803, "lr": 5.74168724760682e-10, "epoch": 4.9923273657289, "percentage": 99.85, "elapsed_time": "9:49:36", "remaining_time": "0:00:54"} +{"current_steps": 1953, "total_steps": 1955, "loss": 0.9432, "lr": 2.551864390598624e-10, "epoch": 4.994884910485934, "percentage": 99.9, "elapsed_time": "9:49:54", "remaining_time": "0:00:36"} +{"current_steps": 1954, "total_steps": 1955, "loss": 0.9505, "lr": 6.379666063871525e-11, "epoch": 4.997442455242966, "percentage": 99.95, "elapsed_time": "9:50:12", "remaining_time": "0:00:18"} +{"current_steps": 1955, "total_steps": 1955, "loss": 0.971, "lr": 0.0, "epoch": 5.0, "percentage": 100.0, "elapsed_time": "9:50:30", "remaining_time": "0:00:00"} +{"current_steps": 1955, "total_steps": 1955, "epoch": 5.0, "percentage": 100.0, "elapsed_time": "9:50:51", "remaining_time": "0:00:00"} diff --git a/trainer_state.json b/trainer_state.json new file mode 100644 index 0000000..4122264 --- /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": 2.956157423011394, + "learning_rate": 4.0816326530612243e-07, + "loss": 1.5213, + "step": 1 + }, + { + "epoch": 0.005115089514066497, + "grad_norm": 2.957156524653666, + "learning_rate": 8.163265306122449e-07, + "loss": 1.4742, + "step": 2 + }, + { + "epoch": 0.0076726342710997444, + "grad_norm": 3.013201374781629, + "learning_rate": 1.2244897959183673e-06, + "loss": 1.4946, + "step": 3 + }, + { + "epoch": 0.010230179028132993, + "grad_norm": 2.9332409751747672, + "learning_rate": 1.6326530612244897e-06, + "loss": 1.4837, + "step": 4 + }, + { + "epoch": 0.01278772378516624, + "grad_norm": 2.8958487377498088, + "learning_rate": 2.0408163265306125e-06, + "loss": 1.4922, + "step": 5 + }, + { + "epoch": 0.015345268542199489, + "grad_norm": 2.7113697659439957, + "learning_rate": 2.4489795918367347e-06, + "loss": 1.4997, + "step": 6 + }, + { + "epoch": 0.017902813299232736, + "grad_norm": 2.55930320591955, + "learning_rate": 2.8571428571428573e-06, + "loss": 1.4709, + "step": 7 + }, + { + "epoch": 0.020460358056265986, + "grad_norm": 2.0258165853063708, + "learning_rate": 3.2653061224489794e-06, + "loss": 1.4673, + "step": 8 + }, + { + "epoch": 0.023017902813299233, + "grad_norm": 1.9427534891501914, + "learning_rate": 3.6734693877551024e-06, + "loss": 1.5002, + "step": 9 + }, + { + "epoch": 0.02557544757033248, + "grad_norm": 1.8741278853652108, + "learning_rate": 4.081632653061225e-06, + "loss": 1.4897, + "step": 10 + }, + { + "epoch": 0.028132992327365727, + "grad_norm": 1.8672035113437018, + "learning_rate": 4.489795918367348e-06, + "loss": 1.4669, + "step": 11 + }, + { + "epoch": 0.030690537084398978, + "grad_norm": 1.8231697254158397, + "learning_rate": 4.897959183673469e-06, + "loss": 1.4282, + "step": 12 + }, + { + "epoch": 0.03324808184143223, + "grad_norm": 1.7349261636368631, + "learning_rate": 5.306122448979593e-06, + "loss": 1.4664, + "step": 13 + }, + { + "epoch": 0.03580562659846547, + "grad_norm": 1.4416073688922821, + "learning_rate": 5.7142857142857145e-06, + "loss": 1.3861, + "step": 14 + }, + { + "epoch": 0.03836317135549872, + "grad_norm": 1.804185644593906, + "learning_rate": 6.122448979591837e-06, + "loss": 1.4201, + "step": 15 + }, + { + "epoch": 0.04092071611253197, + "grad_norm": 1.78311427201953, + "learning_rate": 6.530612244897959e-06, + "loss": 1.3764, + "step": 16 + }, + { + "epoch": 0.043478260869565216, + "grad_norm": 1.5180464959128221, + "learning_rate": 6.938775510204082e-06, + "loss": 1.3926, + "step": 17 + }, + { + "epoch": 0.04603580562659847, + "grad_norm": 1.2280933246833325, + "learning_rate": 7.346938775510205e-06, + "loss": 1.3886, + "step": 18 + }, + { + "epoch": 0.04859335038363171, + "grad_norm": 0.978036927475401, + "learning_rate": 7.755102040816327e-06, + "loss": 1.3666, + "step": 19 + }, + { + "epoch": 0.05115089514066496, + "grad_norm": 1.087810311630016, + "learning_rate": 8.16326530612245e-06, + "loss": 1.3479, + "step": 20 + }, + { + "epoch": 0.05370843989769821, + "grad_norm": 0.890366570770121, + "learning_rate": 8.571428571428571e-06, + "loss": 1.3275, + "step": 21 + }, + { + "epoch": 0.056265984654731455, + "grad_norm": 0.8853934440384114, + "learning_rate": 8.979591836734695e-06, + "loss": 1.3511, + "step": 22 + }, + { + "epoch": 0.058823529411764705, + "grad_norm": 0.7917639991185591, + "learning_rate": 9.387755102040818e-06, + "loss": 1.3045, + "step": 23 + }, + { + "epoch": 0.061381074168797956, + "grad_norm": 0.6651063674631976, + "learning_rate": 9.795918367346939e-06, + "loss": 1.3239, + "step": 24 + }, + { + "epoch": 0.0639386189258312, + "grad_norm": 0.6825326754949458, + "learning_rate": 1.0204081632653063e-05, + "loss": 1.2989, + "step": 25 + }, + { + "epoch": 0.06649616368286446, + "grad_norm": 0.7207844273872734, + "learning_rate": 1.0612244897959186e-05, + "loss": 1.3023, + "step": 26 + }, + { + "epoch": 0.06905370843989769, + "grad_norm": 0.587801356926753, + "learning_rate": 1.1020408163265306e-05, + "loss": 1.2843, + "step": 27 + }, + { + "epoch": 0.07161125319693094, + "grad_norm": 0.49770805079425884, + "learning_rate": 1.1428571428571429e-05, + "loss": 1.3075, + "step": 28 + }, + { + "epoch": 0.0741687979539642, + "grad_norm": 0.545355697285009, + "learning_rate": 1.1836734693877552e-05, + "loss": 1.2835, + "step": 29 + }, + { + "epoch": 0.07672634271099744, + "grad_norm": 0.5693171639034866, + "learning_rate": 1.2244897959183674e-05, + "loss": 1.2971, + "step": 30 + }, + { + "epoch": 0.0792838874680307, + "grad_norm": 0.5161243281317084, + "learning_rate": 1.2653061224489798e-05, + "loss": 1.2911, + "step": 31 + }, + { + "epoch": 0.08184143222506395, + "grad_norm": 0.46725528465384736, + "learning_rate": 1.3061224489795918e-05, + "loss": 1.2638, + "step": 32 + }, + { + "epoch": 0.08439897698209718, + "grad_norm": 0.41798797083329553, + "learning_rate": 1.3469387755102042e-05, + "loss": 1.2964, + "step": 33 + }, + { + "epoch": 0.08695652173913043, + "grad_norm": 0.4623409185375301, + "learning_rate": 1.3877551020408165e-05, + "loss": 1.3138, + "step": 34 + }, + { + "epoch": 0.08951406649616368, + "grad_norm": 0.4374009191444938, + "learning_rate": 1.4285714285714287e-05, + "loss": 1.2555, + "step": 35 + }, + { + "epoch": 0.09207161125319693, + "grad_norm": 0.36271282660839976, + "learning_rate": 1.469387755102041e-05, + "loss": 1.2684, + "step": 36 + }, + { + "epoch": 0.09462915601023018, + "grad_norm": 0.39662907683429816, + "learning_rate": 1.510204081632653e-05, + "loss": 1.2828, + "step": 37 + }, + { + "epoch": 0.09718670076726342, + "grad_norm": 0.3621344734960456, + "learning_rate": 1.5510204081632655e-05, + "loss": 1.2585, + "step": 38 + }, + { + "epoch": 0.09974424552429667, + "grad_norm": 0.38753977253082816, + "learning_rate": 1.5918367346938776e-05, + "loss": 1.2739, + "step": 39 + }, + { + "epoch": 0.10230179028132992, + "grad_norm": 0.32331110740235386, + "learning_rate": 1.63265306122449e-05, + "loss": 1.2148, + "step": 40 + }, + { + "epoch": 0.10485933503836317, + "grad_norm": 0.3235269735931933, + "learning_rate": 1.673469387755102e-05, + "loss": 1.242, + "step": 41 + }, + { + "epoch": 0.10741687979539642, + "grad_norm": 0.3438080240782346, + "learning_rate": 1.7142857142857142e-05, + "loss": 1.2409, + "step": 42 + }, + { + "epoch": 0.10997442455242967, + "grad_norm": 0.2916611186028452, + "learning_rate": 1.7551020408163266e-05, + "loss": 1.2136, + "step": 43 + }, + { + "epoch": 0.11253196930946291, + "grad_norm": 0.3008986538459429, + "learning_rate": 1.795918367346939e-05, + "loss": 1.2245, + "step": 44 + }, + { + "epoch": 0.11508951406649616, + "grad_norm": 0.3060253220280668, + "learning_rate": 1.836734693877551e-05, + "loss": 1.2252, + "step": 45 + }, + { + "epoch": 0.11764705882352941, + "grad_norm": 0.31820392904760986, + "learning_rate": 1.8775510204081636e-05, + "loss": 1.2464, + "step": 46 + }, + { + "epoch": 0.12020460358056266, + "grad_norm": 0.2865033138020395, + "learning_rate": 1.9183673469387756e-05, + "loss": 1.2466, + "step": 47 + }, + { + "epoch": 0.12276214833759591, + "grad_norm": 0.26555310624821354, + "learning_rate": 1.9591836734693877e-05, + "loss": 1.2171, + "step": 48 + }, + { + "epoch": 0.12531969309462915, + "grad_norm": 0.2876989657538828, + "learning_rate": 2e-05, + "loss": 1.2317, + "step": 49 + }, + { + "epoch": 0.1278772378516624, + "grad_norm": 0.247990267920054, + "learning_rate": 2.0408163265306126e-05, + "loss": 1.201, + "step": 50 + }, + { + "epoch": 0.13043478260869565, + "grad_norm": 0.2570347387829217, + "learning_rate": 2.0816326530612247e-05, + "loss": 1.2083, + "step": 51 + }, + { + "epoch": 0.1329923273657289, + "grad_norm": 0.2502935337530908, + "learning_rate": 2.122448979591837e-05, + "loss": 1.189, + "step": 52 + }, + { + "epoch": 0.13554987212276215, + "grad_norm": 0.25819296392357916, + "learning_rate": 2.1632653061224492e-05, + "loss": 1.1904, + "step": 53 + }, + { + "epoch": 0.13810741687979539, + "grad_norm": 0.2557655926265301, + "learning_rate": 2.2040816326530613e-05, + "loss": 1.2268, + "step": 54 + }, + { + "epoch": 0.14066496163682865, + "grad_norm": 0.23483854657893372, + "learning_rate": 2.2448979591836734e-05, + "loss": 1.2391, + "step": 55 + }, + { + "epoch": 0.1432225063938619, + "grad_norm": 0.24717465180571466, + "learning_rate": 2.2857142857142858e-05, + "loss": 1.227, + "step": 56 + }, + { + "epoch": 0.14578005115089515, + "grad_norm": 0.2172921037220085, + "learning_rate": 2.3265306122448982e-05, + "loss": 1.2093, + "step": 57 + }, + { + "epoch": 0.1483375959079284, + "grad_norm": 0.26820055891479894, + "learning_rate": 2.3673469387755103e-05, + "loss": 1.2253, + "step": 58 + }, + { + "epoch": 0.15089514066496162, + "grad_norm": 0.21516549373211177, + "learning_rate": 2.4081632653061227e-05, + "loss": 1.191, + "step": 59 + }, + { + "epoch": 0.1534526854219949, + "grad_norm": 0.27883891563143004, + "learning_rate": 2.448979591836735e-05, + "loss": 1.2151, + "step": 60 + }, + { + "epoch": 0.15601023017902813, + "grad_norm": 0.23221139463765425, + "learning_rate": 2.4897959183673473e-05, + "loss": 1.2032, + "step": 61 + }, + { + "epoch": 0.1585677749360614, + "grad_norm": 0.2902807098346473, + "learning_rate": 2.5306122448979597e-05, + "loss": 1.2352, + "step": 62 + }, + { + "epoch": 0.16112531969309463, + "grad_norm": 0.25945252836481614, + "learning_rate": 2.5714285714285718e-05, + "loss": 1.2182, + "step": 63 + }, + { + "epoch": 0.1636828644501279, + "grad_norm": 0.3154426361540951, + "learning_rate": 2.6122448979591835e-05, + "loss": 1.2128, + "step": 64 + }, + { + "epoch": 0.16624040920716113, + "grad_norm": 0.3917230985512171, + "learning_rate": 2.653061224489796e-05, + "loss": 1.2052, + "step": 65 + }, + { + "epoch": 0.16879795396419436, + "grad_norm": 0.4945933154303559, + "learning_rate": 2.6938775510204084e-05, + "loss": 1.1671, + "step": 66 + }, + { + "epoch": 0.17135549872122763, + "grad_norm": 0.6037374924582795, + "learning_rate": 2.7346938775510205e-05, + "loss": 1.1915, + "step": 67 + }, + { + "epoch": 0.17391304347826086, + "grad_norm": 0.5830383636720183, + "learning_rate": 2.775510204081633e-05, + "loss": 1.1683, + "step": 68 + }, + { + "epoch": 0.17647058823529413, + "grad_norm": 0.4049022994573582, + "learning_rate": 2.816326530612245e-05, + "loss": 1.152, + "step": 69 + }, + { + "epoch": 0.17902813299232737, + "grad_norm": 0.36129854662345895, + "learning_rate": 2.8571428571428574e-05, + "loss": 1.1638, + "step": 70 + }, + { + "epoch": 0.1815856777493606, + "grad_norm": 0.5309473427051697, + "learning_rate": 2.89795918367347e-05, + "loss": 1.1747, + "step": 71 + }, + { + "epoch": 0.18414322250639387, + "grad_norm": 0.6377735187370677, + "learning_rate": 2.938775510204082e-05, + "loss": 1.1894, + "step": 72 + }, + { + "epoch": 0.1867007672634271, + "grad_norm": 0.648301783776856, + "learning_rate": 2.9795918367346944e-05, + "loss": 1.2237, + "step": 73 + }, + { + "epoch": 0.18925831202046037, + "grad_norm": 0.5452770096862445, + "learning_rate": 3.020408163265306e-05, + "loss": 1.1695, + "step": 74 + }, + { + "epoch": 0.1918158567774936, + "grad_norm": 0.47917361255424545, + "learning_rate": 3.061224489795918e-05, + "loss": 1.1671, + "step": 75 + }, + { + "epoch": 0.19437340153452684, + "grad_norm": 0.46650181032832944, + "learning_rate": 3.102040816326531e-05, + "loss": 1.1705, + "step": 76 + }, + { + "epoch": 0.1969309462915601, + "grad_norm": 0.39177334859844115, + "learning_rate": 3.142857142857143e-05, + "loss": 1.1808, + "step": 77 + }, + { + "epoch": 0.19948849104859334, + "grad_norm": 0.499101882627103, + "learning_rate": 3.183673469387755e-05, + "loss": 1.1616, + "step": 78 + }, + { + "epoch": 0.2020460358056266, + "grad_norm": 0.5788889023503736, + "learning_rate": 3.224489795918368e-05, + "loss": 1.1554, + "step": 79 + }, + { + "epoch": 0.20460358056265984, + "grad_norm": 0.5961045731637039, + "learning_rate": 3.26530612244898e-05, + "loss": 1.1725, + "step": 80 + }, + { + "epoch": 0.2071611253196931, + "grad_norm": 0.47119669685447696, + "learning_rate": 3.306122448979592e-05, + "loss": 1.1561, + "step": 81 + }, + { + "epoch": 0.20971867007672634, + "grad_norm": 0.4703018870884221, + "learning_rate": 3.346938775510204e-05, + "loss": 1.153, + "step": 82 + }, + { + "epoch": 0.21227621483375958, + "grad_norm": 0.5556907967995176, + "learning_rate": 3.387755102040817e-05, + "loss": 1.1677, + "step": 83 + }, + { + "epoch": 0.21483375959079284, + "grad_norm": 0.6373677375142429, + "learning_rate": 3.4285714285714284e-05, + "loss": 1.1634, + "step": 84 + }, + { + "epoch": 0.21739130434782608, + "grad_norm": 0.825864360069245, + "learning_rate": 3.469387755102041e-05, + "loss": 1.2139, + "step": 85 + }, + { + "epoch": 0.21994884910485935, + "grad_norm": 1.1333633511553514, + "learning_rate": 3.510204081632653e-05, + "loss": 1.1666, + "step": 86 + }, + { + "epoch": 0.22250639386189258, + "grad_norm": 1.1545216752131862, + "learning_rate": 3.551020408163265e-05, + "loss": 1.1727, + "step": 87 + }, + { + "epoch": 0.22506393861892582, + "grad_norm": 0.842741588438428, + "learning_rate": 3.591836734693878e-05, + "loss": 1.1958, + "step": 88 + }, + { + "epoch": 0.22762148337595908, + "grad_norm": 0.5836488090866901, + "learning_rate": 3.63265306122449e-05, + "loss": 1.1561, + "step": 89 + }, + { + "epoch": 0.23017902813299232, + "grad_norm": 0.6807274898079663, + "learning_rate": 3.673469387755102e-05, + "loss": 1.1866, + "step": 90 + }, + { + "epoch": 0.23273657289002558, + "grad_norm": 0.978642798883616, + "learning_rate": 3.714285714285715e-05, + "loss": 1.1831, + "step": 91 + }, + { + "epoch": 0.23529411764705882, + "grad_norm": 1.1688573758554635, + "learning_rate": 3.755102040816327e-05, + "loss": 1.1865, + "step": 92 + }, + { + "epoch": 0.23785166240409208, + "grad_norm": 0.6824215678902069, + "learning_rate": 3.795918367346939e-05, + "loss": 1.1676, + "step": 93 + }, + { + "epoch": 0.24040920716112532, + "grad_norm": 0.44788443667257866, + "learning_rate": 3.836734693877551e-05, + "loss": 1.1884, + "step": 94 + }, + { + "epoch": 0.24296675191815856, + "grad_norm": 0.5868828512762089, + "learning_rate": 3.8775510204081634e-05, + "loss": 1.1359, + "step": 95 + }, + { + "epoch": 0.24552429667519182, + "grad_norm": 0.9092466565103363, + "learning_rate": 3.9183673469387755e-05, + "loss": 1.1769, + "step": 96 + }, + { + "epoch": 0.24808184143222506, + "grad_norm": 1.0055631954693776, + "learning_rate": 3.959183673469388e-05, + "loss": 1.18, + "step": 97 + }, + { + "epoch": 0.2506393861892583, + "grad_norm": 0.8661368356493842, + "learning_rate": 4e-05, + "loss": 1.1508, + "step": 98 + }, + { + "epoch": 0.2531969309462916, + "grad_norm": 0.6456496755830919, + "learning_rate": 4.0408163265306124e-05, + "loss": 1.1473, + "step": 99 + }, + { + "epoch": 0.2557544757033248, + "grad_norm": 0.8111263397131059, + "learning_rate": 4.081632653061225e-05, + "loss": 1.1413, + "step": 100 + }, + { + "epoch": 0.25831202046035806, + "grad_norm": 1.2262439163561427, + "learning_rate": 4.1224489795918366e-05, + "loss": 1.1647, + "step": 101 + }, + { + "epoch": 0.2608695652173913, + "grad_norm": 0.895523164890546, + "learning_rate": 4.1632653061224494e-05, + "loss": 1.1698, + "step": 102 + }, + { + "epoch": 0.26342710997442453, + "grad_norm": 0.4128663944963409, + "learning_rate": 4.2040816326530615e-05, + "loss": 1.1374, + "step": 103 + }, + { + "epoch": 0.2659846547314578, + "grad_norm": 0.6875687358413721, + "learning_rate": 4.244897959183674e-05, + "loss": 1.1295, + "step": 104 + }, + { + "epoch": 0.26854219948849106, + "grad_norm": 1.2301351848291924, + "learning_rate": 4.2857142857142856e-05, + "loss": 1.1397, + "step": 105 + }, + { + "epoch": 0.2710997442455243, + "grad_norm": 0.7010073231328794, + "learning_rate": 4.3265306122448984e-05, + "loss": 1.1273, + "step": 106 + }, + { + "epoch": 0.27365728900255754, + "grad_norm": 0.5227686699195527, + "learning_rate": 4.3673469387755105e-05, + "loss": 1.1544, + "step": 107 + }, + { + "epoch": 0.27621483375959077, + "grad_norm": 0.5470695476248038, + "learning_rate": 4.4081632653061226e-05, + "loss": 1.134, + "step": 108 + }, + { + "epoch": 0.27877237851662406, + "grad_norm": 0.7660262395726699, + "learning_rate": 4.4489795918367353e-05, + "loss": 1.1582, + "step": 109 + }, + { + "epoch": 0.2813299232736573, + "grad_norm": 1.1805124576184864, + "learning_rate": 4.489795918367347e-05, + "loss": 1.1614, + "step": 110 + }, + { + "epoch": 0.28388746803069054, + "grad_norm": 0.8289200241114665, + "learning_rate": 4.5306122448979595e-05, + "loss": 1.162, + "step": 111 + }, + { + "epoch": 0.2864450127877238, + "grad_norm": 0.6363878509379256, + "learning_rate": 4.5714285714285716e-05, + "loss": 1.1508, + "step": 112 + }, + { + "epoch": 0.289002557544757, + "grad_norm": 0.8223651947759109, + "learning_rate": 4.6122448979591844e-05, + "loss": 1.1246, + "step": 113 + }, + { + "epoch": 0.2915601023017903, + "grad_norm": 0.9658149004434372, + "learning_rate": 4.6530612244897965e-05, + "loss": 1.155, + "step": 114 + }, + { + "epoch": 0.29411764705882354, + "grad_norm": 1.1429259285624525, + "learning_rate": 4.6938775510204086e-05, + "loss": 1.1396, + "step": 115 + }, + { + "epoch": 0.2966751918158568, + "grad_norm": 1.0902172962065055, + "learning_rate": 4.7346938775510206e-05, + "loss": 1.146, + "step": 116 + }, + { + "epoch": 0.29923273657289, + "grad_norm": 0.988988857628424, + "learning_rate": 4.7755102040816334e-05, + "loss": 1.1366, + "step": 117 + }, + { + "epoch": 0.30179028132992325, + "grad_norm": 0.7402542318116936, + "learning_rate": 4.8163265306122455e-05, + "loss": 1.1863, + "step": 118 + }, + { + "epoch": 0.30434782608695654, + "grad_norm": 0.9283436178370745, + "learning_rate": 4.857142857142857e-05, + "loss": 1.1481, + "step": 119 + }, + { + "epoch": 0.3069053708439898, + "grad_norm": 1.1139736386114893, + "learning_rate": 4.89795918367347e-05, + "loss": 1.1574, + "step": 120 + }, + { + "epoch": 0.309462915601023, + "grad_norm": 0.6424105578724688, + "learning_rate": 4.938775510204082e-05, + "loss": 1.1701, + "step": 121 + }, + { + "epoch": 0.31202046035805625, + "grad_norm": 0.5840585035243322, + "learning_rate": 4.9795918367346945e-05, + "loss": 1.1204, + "step": 122 + }, + { + "epoch": 0.3145780051150895, + "grad_norm": 0.8914478021127121, + "learning_rate": 5.0204081632653066e-05, + "loss": 1.1424, + "step": 123 + }, + { + "epoch": 0.3171355498721228, + "grad_norm": 0.8050516376303483, + "learning_rate": 5.0612244897959194e-05, + "loss": 1.1374, + "step": 124 + }, + { + "epoch": 0.319693094629156, + "grad_norm": 0.7768422507226259, + "learning_rate": 5.102040816326531e-05, + "loss": 1.1181, + "step": 125 + }, + { + "epoch": 0.32225063938618925, + "grad_norm": 0.9832467250312055, + "learning_rate": 5.1428571428571436e-05, + "loss": 1.1755, + "step": 126 + }, + { + "epoch": 0.3248081841432225, + "grad_norm": 1.0832409134965832, + "learning_rate": 5.1836734693877557e-05, + "loss": 1.12, + "step": 127 + }, + { + "epoch": 0.3273657289002558, + "grad_norm": 1.05095046390139, + "learning_rate": 5.224489795918367e-05, + "loss": 1.1371, + "step": 128 + }, + { + "epoch": 0.329923273657289, + "grad_norm": 0.8504844773142398, + "learning_rate": 5.26530612244898e-05, + "loss": 1.1289, + "step": 129 + }, + { + "epoch": 0.33248081841432225, + "grad_norm": 1.013882978247601, + "learning_rate": 5.306122448979592e-05, + "loss": 1.1588, + "step": 130 + }, + { + "epoch": 0.3350383631713555, + "grad_norm": 1.4983721735915714, + "learning_rate": 5.346938775510205e-05, + "loss": 1.1252, + "step": 131 + }, + { + "epoch": 0.3375959079283887, + "grad_norm": 0.6395636406832796, + "learning_rate": 5.387755102040817e-05, + "loss": 1.1789, + "step": 132 + }, + { + "epoch": 0.340153452685422, + "grad_norm": 1.1119427951585954, + "learning_rate": 5.4285714285714295e-05, + "loss": 1.1548, + "step": 133 + }, + { + "epoch": 0.34271099744245526, + "grad_norm": 1.50419051921151, + "learning_rate": 5.469387755102041e-05, + "loss": 1.1259, + "step": 134 + }, + { + "epoch": 0.3452685421994885, + "grad_norm": 0.8868212687779602, + "learning_rate": 5.510204081632654e-05, + "loss": 1.1825, + "step": 135 + }, + { + "epoch": 0.34782608695652173, + "grad_norm": 1.1455650604061842, + "learning_rate": 5.551020408163266e-05, + "loss": 1.1453, + "step": 136 + }, + { + "epoch": 0.35038363171355497, + "grad_norm": 1.1928108885641358, + "learning_rate": 5.5918367346938786e-05, + "loss": 1.1554, + "step": 137 + }, + { + "epoch": 0.35294117647058826, + "grad_norm": 1.2946027592370013, + "learning_rate": 5.63265306122449e-05, + "loss": 1.1429, + "step": 138 + }, + { + "epoch": 0.3554987212276215, + "grad_norm": 0.7456369454357881, + "learning_rate": 5.673469387755102e-05, + "loss": 1.116, + "step": 139 + }, + { + "epoch": 0.35805626598465473, + "grad_norm": 0.9145391630298629, + "learning_rate": 5.714285714285715e-05, + "loss": 1.176, + "step": 140 + }, + { + "epoch": 0.36061381074168797, + "grad_norm": 1.0539461991771215, + "learning_rate": 5.755102040816327e-05, + "loss": 1.0897, + "step": 141 + }, + { + "epoch": 0.3631713554987212, + "grad_norm": 1.1574195644796725, + "learning_rate": 5.79591836734694e-05, + "loss": 1.1411, + "step": 142 + }, + { + "epoch": 0.3657289002557545, + "grad_norm": 1.0444683774173515, + "learning_rate": 5.836734693877551e-05, + "loss": 1.1689, + "step": 143 + }, + { + "epoch": 0.36828644501278773, + "grad_norm": 0.8301811817561374, + "learning_rate": 5.877551020408164e-05, + "loss": 1.1345, + "step": 144 + }, + { + "epoch": 0.37084398976982097, + "grad_norm": 0.812249217700949, + "learning_rate": 5.918367346938776e-05, + "loss": 1.1466, + "step": 145 + }, + { + "epoch": 0.3734015345268542, + "grad_norm": 0.7307569711274879, + "learning_rate": 5.959183673469389e-05, + "loss": 1.1303, + "step": 146 + }, + { + "epoch": 0.37595907928388744, + "grad_norm": 0.8056769718133762, + "learning_rate": 6.000000000000001e-05, + "loss": 1.1675, + "step": 147 + }, + { + "epoch": 0.37851662404092073, + "grad_norm": 1.2641845208914293, + "learning_rate": 6.040816326530612e-05, + "loss": 1.13, + "step": 148 + }, + { + "epoch": 0.38107416879795397, + "grad_norm": 1.0107711010817608, + "learning_rate": 6.081632653061225e-05, + "loss": 1.1145, + "step": 149 + }, + { + "epoch": 0.3836317135549872, + "grad_norm": 1.1073005971254093, + "learning_rate": 6.122448979591836e-05, + "loss": 1.1266, + "step": 150 + }, + { + "epoch": 0.38618925831202044, + "grad_norm": 0.8625795523903089, + "learning_rate": 6.163265306122449e-05, + "loss": 1.129, + "step": 151 + }, + { + "epoch": 0.3887468030690537, + "grad_norm": 1.0401585969587819, + "learning_rate": 6.204081632653062e-05, + "loss": 1.1361, + "step": 152 + }, + { + "epoch": 0.391304347826087, + "grad_norm": 1.0394281409109958, + "learning_rate": 6.244897959183675e-05, + "loss": 1.1266, + "step": 153 + }, + { + "epoch": 0.3938618925831202, + "grad_norm": 0.9953073831065966, + "learning_rate": 6.285714285714286e-05, + "loss": 1.1343, + "step": 154 + }, + { + "epoch": 0.39641943734015345, + "grad_norm": 1.2177849301315296, + "learning_rate": 6.326530612244899e-05, + "loss": 1.1463, + "step": 155 + }, + { + "epoch": 0.3989769820971867, + "grad_norm": 0.7596280219343059, + "learning_rate": 6.36734693877551e-05, + "loss": 1.1454, + "step": 156 + }, + { + "epoch": 0.40153452685422, + "grad_norm": 1.0754506244483104, + "learning_rate": 6.408163265306123e-05, + "loss": 1.1616, + "step": 157 + }, + { + "epoch": 0.4040920716112532, + "grad_norm": 1.1630866426285293, + "learning_rate": 6.448979591836736e-05, + "loss": 1.1545, + "step": 158 + }, + { + "epoch": 0.40664961636828645, + "grad_norm": 0.6988890630556159, + "learning_rate": 6.489795918367347e-05, + "loss": 1.1052, + "step": 159 + }, + { + "epoch": 0.4092071611253197, + "grad_norm": 0.7715767301674425, + "learning_rate": 6.53061224489796e-05, + "loss": 1.1264, + "step": 160 + }, + { + "epoch": 0.4117647058823529, + "grad_norm": 0.7984070291716113, + "learning_rate": 6.571428571428571e-05, + "loss": 1.12, + "step": 161 + }, + { + "epoch": 0.4143222506393862, + "grad_norm": 0.8560213949536746, + "learning_rate": 6.612244897959184e-05, + "loss": 1.1171, + "step": 162 + }, + { + "epoch": 0.41687979539641945, + "grad_norm": 0.9369116479027433, + "learning_rate": 6.653061224489797e-05, + "loss": 1.1196, + "step": 163 + }, + { + "epoch": 0.4194373401534527, + "grad_norm": 0.9843108307123993, + "learning_rate": 6.693877551020408e-05, + "loss": 1.1316, + "step": 164 + }, + { + "epoch": 0.4219948849104859, + "grad_norm": 1.0791992478077372, + "learning_rate": 6.734693877551021e-05, + "loss": 1.107, + "step": 165 + }, + { + "epoch": 0.42455242966751916, + "grad_norm": 1.2221610713261708, + "learning_rate": 6.775510204081634e-05, + "loss": 1.1294, + "step": 166 + }, + { + "epoch": 0.42710997442455245, + "grad_norm": 0.9019331212289963, + "learning_rate": 6.816326530612245e-05, + "loss": 1.1245, + "step": 167 + }, + { + "epoch": 0.4296675191815857, + "grad_norm": 0.9269989033015092, + "learning_rate": 6.857142857142857e-05, + "loss": 1.1338, + "step": 168 + }, + { + "epoch": 0.4322250639386189, + "grad_norm": 1.0981263529686658, + "learning_rate": 6.89795918367347e-05, + "loss": 1.1332, + "step": 169 + }, + { + "epoch": 0.43478260869565216, + "grad_norm": 0.9888336583144302, + "learning_rate": 6.938775510204082e-05, + "loss": 1.1127, + "step": 170 + }, + { + "epoch": 0.4373401534526854, + "grad_norm": 1.3074455063450379, + "learning_rate": 6.979591836734695e-05, + "loss": 1.1285, + "step": 171 + }, + { + "epoch": 0.4398976982097187, + "grad_norm": 0.6818203503765217, + "learning_rate": 7.020408163265306e-05, + "loss": 1.1223, + "step": 172 + }, + { + "epoch": 0.4424552429667519, + "grad_norm": 0.8192921172095765, + "learning_rate": 7.061224489795919e-05, + "loss": 1.142, + "step": 173 + }, + { + "epoch": 0.44501278772378516, + "grad_norm": 0.8114896652522754, + "learning_rate": 7.10204081632653e-05, + "loss": 1.151, + "step": 174 + }, + { + "epoch": 0.4475703324808184, + "grad_norm": 0.8520710303534221, + "learning_rate": 7.142857142857143e-05, + "loss": 1.1235, + "step": 175 + }, + { + "epoch": 0.45012787723785164, + "grad_norm": 1.1274861056526013, + "learning_rate": 7.183673469387756e-05, + "loss": 1.1553, + "step": 176 + }, + { + "epoch": 0.45268542199488493, + "grad_norm": 1.0575599991469968, + "learning_rate": 7.224489795918368e-05, + "loss": 1.1281, + "step": 177 + }, + { + "epoch": 0.45524296675191817, + "grad_norm": 0.9955941287548709, + "learning_rate": 7.26530612244898e-05, + "loss": 1.1267, + "step": 178 + }, + { + "epoch": 0.4578005115089514, + "grad_norm": 1.334626282193364, + "learning_rate": 7.306122448979592e-05, + "loss": 1.1147, + "step": 179 + }, + { + "epoch": 0.46035805626598464, + "grad_norm": 1.0475795735421864, + "learning_rate": 7.346938775510205e-05, + "loss": 1.1197, + "step": 180 + }, + { + "epoch": 0.4629156010230179, + "grad_norm": 1.5007093885023364, + "learning_rate": 7.387755102040817e-05, + "loss": 1.1482, + "step": 181 + }, + { + "epoch": 0.46547314578005117, + "grad_norm": 0.7288699412052543, + "learning_rate": 7.42857142857143e-05, + "loss": 1.1206, + "step": 182 + }, + { + "epoch": 0.4680306905370844, + "grad_norm": 1.3525278294620537, + "learning_rate": 7.469387755102041e-05, + "loss": 1.1366, + "step": 183 + }, + { + "epoch": 0.47058823529411764, + "grad_norm": 0.9907423022995694, + "learning_rate": 7.510204081632654e-05, + "loss": 1.1294, + "step": 184 + }, + { + "epoch": 0.4731457800511509, + "grad_norm": 1.4564946558231437, + "learning_rate": 7.551020408163266e-05, + "loss": 1.1395, + "step": 185 + }, + { + "epoch": 0.47570332480818417, + "grad_norm": 0.7257811934036715, + "learning_rate": 7.591836734693878e-05, + "loss": 1.1332, + "step": 186 + }, + { + "epoch": 0.4782608695652174, + "grad_norm": 1.0803795052811809, + "learning_rate": 7.63265306122449e-05, + "loss": 1.1187, + "step": 187 + }, + { + "epoch": 0.48081841432225064, + "grad_norm": 1.0690865063674833, + "learning_rate": 7.673469387755103e-05, + "loss": 1.1428, + "step": 188 + }, + { + "epoch": 0.4833759590792839, + "grad_norm": 1.0845046051039753, + "learning_rate": 7.714285714285715e-05, + "loss": 1.126, + "step": 189 + }, + { + "epoch": 0.4859335038363171, + "grad_norm": 0.9227022064148079, + "learning_rate": 7.755102040816327e-05, + "loss": 1.1262, + "step": 190 + }, + { + "epoch": 0.4884910485933504, + "grad_norm": 1.031390374218521, + "learning_rate": 7.79591836734694e-05, + "loss": 1.0769, + "step": 191 + }, + { + "epoch": 0.49104859335038364, + "grad_norm": 0.9817678976684439, + "learning_rate": 7.836734693877551e-05, + "loss": 1.1177, + "step": 192 + }, + { + "epoch": 0.4936061381074169, + "grad_norm": 1.0891373841172232, + "learning_rate": 7.877551020408164e-05, + "loss": 1.0895, + "step": 193 + }, + { + "epoch": 0.4961636828644501, + "grad_norm": 0.8992782943686848, + "learning_rate": 7.918367346938776e-05, + "loss": 1.0864, + "step": 194 + }, + { + "epoch": 0.49872122762148335, + "grad_norm": 1.0356054173418965, + "learning_rate": 7.959183673469389e-05, + "loss": 1.1628, + "step": 195 + }, + { + "epoch": 0.5012787723785166, + "grad_norm": 1.054829292229048, + "learning_rate": 8e-05, + "loss": 1.1215, + "step": 196 + }, + { + "epoch": 0.5038363171355499, + "grad_norm": 1.1952728526697924, + "learning_rate": 7.999993620333937e-05, + "loss": 1.1072, + "step": 197 + }, + { + "epoch": 0.5063938618925832, + "grad_norm": 1.4069207531654522, + "learning_rate": 7.999974481356096e-05, + "loss": 1.1389, + "step": 198 + }, + { + "epoch": 0.5089514066496164, + "grad_norm": 1.0226573738564007, + "learning_rate": 7.999942583127525e-05, + "loss": 1.1039, + "step": 199 + }, + { + "epoch": 0.5115089514066496, + "grad_norm": 1.237262581205958, + "learning_rate": 7.999897925749977e-05, + "loss": 1.1205, + "step": 200 + }, + { + "epoch": 0.5140664961636828, + "grad_norm": 0.713505858861466, + "learning_rate": 7.9998405093659e-05, + "loss": 1.0991, + "step": 201 + }, + { + "epoch": 0.5166240409207161, + "grad_norm": 0.8851648472236339, + "learning_rate": 7.999770334158445e-05, + "loss": 1.0939, + "step": 202 + }, + { + "epoch": 0.5191815856777494, + "grad_norm": 1.2619572731510282, + "learning_rate": 7.999687400351458e-05, + "loss": 1.1263, + "step": 203 + }, + { + "epoch": 0.5217391304347826, + "grad_norm": 0.650360402034941, + "learning_rate": 7.999591708209482e-05, + "loss": 1.0945, + "step": 204 + }, + { + "epoch": 0.5242966751918159, + "grad_norm": 1.0524079760247882, + "learning_rate": 7.99948325803776e-05, + "loss": 1.1189, + "step": 205 + }, + { + "epoch": 0.5268542199488491, + "grad_norm": 1.1840200840428867, + "learning_rate": 7.999362050182232e-05, + "loss": 1.1079, + "step": 206 + }, + { + "epoch": 0.5294117647058824, + "grad_norm": 0.9689049896840791, + "learning_rate": 7.999228085029529e-05, + "loss": 1.1252, + "step": 207 + }, + { + "epoch": 0.5319693094629157, + "grad_norm": 1.3635698843008814, + "learning_rate": 7.999081363006977e-05, + "loss": 1.1119, + "step": 208 + }, + { + "epoch": 0.5345268542199488, + "grad_norm": 0.7404269657383984, + "learning_rate": 7.998921884582596e-05, + "loss": 1.0998, + "step": 209 + }, + { + "epoch": 0.5370843989769821, + "grad_norm": 0.901502531204049, + "learning_rate": 7.998749650265096e-05, + "loss": 1.0953, + "step": 210 + }, + { + "epoch": 0.5396419437340153, + "grad_norm": 0.8855993913877257, + "learning_rate": 7.998564660603873e-05, + "loss": 1.1146, + "step": 211 + }, + { + "epoch": 0.5421994884910486, + "grad_norm": 0.6583772614283104, + "learning_rate": 7.998366916189016e-05, + "loss": 1.0821, + "step": 212 + }, + { + "epoch": 0.5447570332480819, + "grad_norm": 0.6745326521828335, + "learning_rate": 7.998156417651297e-05, + "loss": 1.1258, + "step": 213 + }, + { + "epoch": 0.5473145780051151, + "grad_norm": 0.9038994334777463, + "learning_rate": 7.997933165662167e-05, + "loss": 1.0766, + "step": 214 + }, + { + "epoch": 0.5498721227621484, + "grad_norm": 0.9031283752629393, + "learning_rate": 7.997697160933767e-05, + "loss": 1.0825, + "step": 215 + }, + { + "epoch": 0.5524296675191815, + "grad_norm": 1.3656136727514145, + "learning_rate": 7.99744840421891e-05, + "loss": 1.1183, + "step": 216 + }, + { + "epoch": 0.5549872122762148, + "grad_norm": 0.9662280283806104, + "learning_rate": 7.997186896311091e-05, + "loss": 1.0898, + "step": 217 + }, + { + "epoch": 0.5575447570332481, + "grad_norm": 1.1227766801563628, + "learning_rate": 7.996912638044474e-05, + "loss": 1.0641, + "step": 218 + }, + { + "epoch": 0.5601023017902813, + "grad_norm": 0.5938083971975682, + "learning_rate": 7.996625630293898e-05, + "loss": 1.087, + "step": 219 + }, + { + "epoch": 0.5626598465473146, + "grad_norm": 0.830004505260378, + "learning_rate": 7.996325873974871e-05, + "loss": 1.1414, + "step": 220 + }, + { + "epoch": 0.5652173913043478, + "grad_norm": 0.688470818535481, + "learning_rate": 7.996013370043562e-05, + "loss": 1.0817, + "step": 221 + }, + { + "epoch": 0.5677749360613811, + "grad_norm": 0.9773236602394363, + "learning_rate": 7.995688119496811e-05, + "loss": 1.0851, + "step": 222 + }, + { + "epoch": 0.5703324808184144, + "grad_norm": 1.584952668299985, + "learning_rate": 7.99535012337211e-05, + "loss": 1.0933, + "step": 223 + }, + { + "epoch": 0.5728900255754475, + "grad_norm": 0.8022915092257511, + "learning_rate": 7.99499938274761e-05, + "loss": 1.1196, + "step": 224 + }, + { + "epoch": 0.5754475703324808, + "grad_norm": 1.3582138669069936, + "learning_rate": 7.994635898742118e-05, + "loss": 1.1199, + "step": 225 + }, + { + "epoch": 0.578005115089514, + "grad_norm": 1.3157395032889772, + "learning_rate": 7.994259672515083e-05, + "loss": 1.0922, + "step": 226 + }, + { + "epoch": 0.5805626598465473, + "grad_norm": 1.0747983662822311, + "learning_rate": 7.993870705266608e-05, + "loss": 1.1012, + "step": 227 + }, + { + "epoch": 0.5831202046035806, + "grad_norm": 0.9000554867539409, + "learning_rate": 7.99346899823743e-05, + "loss": 1.086, + "step": 228 + }, + { + "epoch": 0.5856777493606138, + "grad_norm": 1.1498952956983293, + "learning_rate": 7.993054552708931e-05, + "loss": 1.1069, + "step": 229 + }, + { + "epoch": 0.5882352941176471, + "grad_norm": 0.9821705283797745, + "learning_rate": 7.99262737000312e-05, + "loss": 1.1105, + "step": 230 + }, + { + "epoch": 0.5907928388746803, + "grad_norm": 1.2035932083418026, + "learning_rate": 7.992187451482642e-05, + "loss": 1.1132, + "step": 231 + }, + { + "epoch": 0.5933503836317136, + "grad_norm": 0.9030392431196859, + "learning_rate": 7.991734798550758e-05, + "loss": 1.0944, + "step": 232 + }, + { + "epoch": 0.5959079283887468, + "grad_norm": 0.901349070838302, + "learning_rate": 7.99126941265136e-05, + "loss": 1.1008, + "step": 233 + }, + { + "epoch": 0.59846547314578, + "grad_norm": 0.9137700738163895, + "learning_rate": 7.990791295268949e-05, + "loss": 1.0579, + "step": 234 + }, + { + "epoch": 0.6010230179028133, + "grad_norm": 0.9571473627952674, + "learning_rate": 7.99030044792864e-05, + "loss": 1.0766, + "step": 235 + }, + { + "epoch": 0.6035805626598465, + "grad_norm": 0.9060175971731932, + "learning_rate": 7.989796872196156e-05, + "loss": 1.1069, + "step": 236 + }, + { + "epoch": 0.6061381074168798, + "grad_norm": 0.9882670857555712, + "learning_rate": 7.989280569677817e-05, + "loss": 1.0935, + "step": 237 + }, + { + "epoch": 0.6086956521739131, + "grad_norm": 0.8147636024776135, + "learning_rate": 7.988751542020543e-05, + "loss": 1.0755, + "step": 238 + }, + { + "epoch": 0.6112531969309463, + "grad_norm": 0.9201311976288161, + "learning_rate": 7.988209790911844e-05, + "loss": 1.1183, + "step": 239 + }, + { + "epoch": 0.6138107416879796, + "grad_norm": 1.0312318256040205, + "learning_rate": 7.987655318079816e-05, + "loss": 1.084, + "step": 240 + }, + { + "epoch": 0.6163682864450127, + "grad_norm": 1.0473461685811893, + "learning_rate": 7.987088125293133e-05, + "loss": 1.0765, + "step": 241 + }, + { + "epoch": 0.618925831202046, + "grad_norm": 1.1909721824510273, + "learning_rate": 7.986508214361047e-05, + "loss": 1.1047, + "step": 242 + }, + { + "epoch": 0.6214833759590793, + "grad_norm": 0.6527590513693277, + "learning_rate": 7.985915587133375e-05, + "loss": 1.0975, + "step": 243 + }, + { + "epoch": 0.6240409207161125, + "grad_norm": 0.6644980276462075, + "learning_rate": 7.985310245500502e-05, + "loss": 1.1448, + "step": 244 + }, + { + "epoch": 0.6265984654731458, + "grad_norm": 0.8238984428442946, + "learning_rate": 7.984692191393364e-05, + "loss": 1.1053, + "step": 245 + }, + { + "epoch": 0.629156010230179, + "grad_norm": 0.9188194695868873, + "learning_rate": 7.984061426783452e-05, + "loss": 1.1014, + "step": 246 + }, + { + "epoch": 0.6317135549872123, + "grad_norm": 0.7164005648194159, + "learning_rate": 7.9834179536828e-05, + "loss": 1.0916, + "step": 247 + }, + { + "epoch": 0.6342710997442456, + "grad_norm": 0.6529981781161239, + "learning_rate": 7.982761774143978e-05, + "loss": 1.1379, + "step": 248 + }, + { + "epoch": 0.6368286445012787, + "grad_norm": 0.7893937427043366, + "learning_rate": 7.98209289026009e-05, + "loss": 1.1201, + "step": 249 + }, + { + "epoch": 0.639386189258312, + "grad_norm": 0.6695991863515521, + "learning_rate": 7.981411304164765e-05, + "loss": 1.0988, + "step": 250 + }, + { + "epoch": 0.6419437340153452, + "grad_norm": 0.8732562498954997, + "learning_rate": 7.980717018032147e-05, + "loss": 1.1052, + "step": 251 + }, + { + "epoch": 0.6445012787723785, + "grad_norm": 1.1886830926779333, + "learning_rate": 7.980010034076893e-05, + "loss": 1.1121, + "step": 252 + }, + { + "epoch": 0.6470588235294118, + "grad_norm": 0.7817538943389751, + "learning_rate": 7.979290354554166e-05, + "loss": 1.089, + "step": 253 + }, + { + "epoch": 0.649616368286445, + "grad_norm": 0.7904438046538128, + "learning_rate": 7.978557981759622e-05, + "loss": 1.1395, + "step": 254 + }, + { + "epoch": 0.6521739130434783, + "grad_norm": 0.7725929256674784, + "learning_rate": 7.977812918029407e-05, + "loss": 1.0977, + "step": 255 + }, + { + "epoch": 0.6547314578005116, + "grad_norm": 0.7520483951987087, + "learning_rate": 7.977055165740151e-05, + "loss": 1.0859, + "step": 256 + }, + { + "epoch": 0.6572890025575447, + "grad_norm": 0.6629038851395048, + "learning_rate": 7.976284727308958e-05, + "loss": 1.1208, + "step": 257 + }, + { + "epoch": 0.659846547314578, + "grad_norm": 0.7911526422920417, + "learning_rate": 7.975501605193396e-05, + "loss": 1.0834, + "step": 258 + }, + { + "epoch": 0.6624040920716112, + "grad_norm": 0.9171702660893679, + "learning_rate": 7.974705801891496e-05, + "loss": 1.1104, + "step": 259 + }, + { + "epoch": 0.6649616368286445, + "grad_norm": 1.0462534089765212, + "learning_rate": 7.973897319941736e-05, + "loss": 1.0931, + "step": 260 + }, + { + "epoch": 0.6675191815856778, + "grad_norm": 1.1787829597415223, + "learning_rate": 7.97307616192304e-05, + "loss": 1.0654, + "step": 261 + }, + { + "epoch": 0.670076726342711, + "grad_norm": 0.8204297503121395, + "learning_rate": 7.972242330454762e-05, + "loss": 1.0806, + "step": 262 + }, + { + "epoch": 0.6726342710997443, + "grad_norm": 0.8044857551775063, + "learning_rate": 7.971395828196688e-05, + "loss": 1.1243, + "step": 263 + }, + { + "epoch": 0.6751918158567775, + "grad_norm": 0.7741196371916483, + "learning_rate": 7.970536657849018e-05, + "loss": 1.102, + "step": 264 + }, + { + "epoch": 0.6777493606138107, + "grad_norm": 0.7513622426525168, + "learning_rate": 7.969664822152362e-05, + "loss": 1.1313, + "step": 265 + }, + { + "epoch": 0.680306905370844, + "grad_norm": 0.7139289605653666, + "learning_rate": 7.96878032388773e-05, + "loss": 1.0976, + "step": 266 + }, + { + "epoch": 0.6828644501278772, + "grad_norm": 0.658091891766218, + "learning_rate": 7.967883165876524e-05, + "loss": 1.0522, + "step": 267 + }, + { + "epoch": 0.6854219948849105, + "grad_norm": 0.5913576930217892, + "learning_rate": 7.966973350980528e-05, + "loss": 1.0948, + "step": 268 + }, + { + "epoch": 0.6879795396419437, + "grad_norm": 0.5132302401343739, + "learning_rate": 7.9660508821019e-05, + "loss": 1.0707, + "step": 269 + }, + { + "epoch": 0.690537084398977, + "grad_norm": 0.5780733190718835, + "learning_rate": 7.965115762183163e-05, + "loss": 1.1105, + "step": 270 + }, + { + "epoch": 0.6930946291560103, + "grad_norm": 0.6018356145185099, + "learning_rate": 7.96416799420719e-05, + "loss": 1.1096, + "step": 271 + }, + { + "epoch": 0.6956521739130435, + "grad_norm": 0.6205282943178246, + "learning_rate": 7.963207581197207e-05, + "loss": 1.0721, + "step": 272 + }, + { + "epoch": 0.6982097186700768, + "grad_norm": 0.6061678652911838, + "learning_rate": 7.962234526216767e-05, + "loss": 1.0879, + "step": 273 + }, + { + "epoch": 0.7007672634271099, + "grad_norm": 0.6945934405726988, + "learning_rate": 7.961248832369754e-05, + "loss": 1.1085, + "step": 274 + }, + { + "epoch": 0.7033248081841432, + "grad_norm": 0.5760159589741866, + "learning_rate": 7.96025050280037e-05, + "loss": 1.0864, + "step": 275 + }, + { + "epoch": 0.7058823529411765, + "grad_norm": 0.5565501271336987, + "learning_rate": 7.959239540693115e-05, + "loss": 1.1034, + "step": 276 + }, + { + "epoch": 0.7084398976982097, + "grad_norm": 0.6676422629214439, + "learning_rate": 7.958215949272791e-05, + "loss": 1.1176, + "step": 277 + }, + { + "epoch": 0.710997442455243, + "grad_norm": 0.7153664297383551, + "learning_rate": 7.957179731804485e-05, + "loss": 1.0847, + "step": 278 + }, + { + "epoch": 0.7135549872122762, + "grad_norm": 0.5930553863066046, + "learning_rate": 7.956130891593556e-05, + "loss": 1.114, + "step": 279 + }, + { + "epoch": 0.7161125319693095, + "grad_norm": 0.5888284152918564, + "learning_rate": 7.955069431985632e-05, + "loss": 1.0802, + "step": 280 + }, + { + "epoch": 0.7186700767263428, + "grad_norm": 0.7930110052547374, + "learning_rate": 7.953995356366588e-05, + "loss": 1.0739, + "step": 281 + }, + { + "epoch": 0.7212276214833759, + "grad_norm": 0.6696001227538381, + "learning_rate": 7.952908668162549e-05, + "loss": 1.1003, + "step": 282 + }, + { + "epoch": 0.7237851662404092, + "grad_norm": 0.4757426938641106, + "learning_rate": 7.951809370839867e-05, + "loss": 1.0299, + "step": 283 + }, + { + "epoch": 0.7263427109974424, + "grad_norm": 0.5474713187788883, + "learning_rate": 7.950697467905118e-05, + "loss": 1.0951, + "step": 284 + }, + { + "epoch": 0.7289002557544757, + "grad_norm": 0.5071386457506744, + "learning_rate": 7.949572962905086e-05, + "loss": 1.0772, + "step": 285 + }, + { + "epoch": 0.731457800511509, + "grad_norm": 0.5247168975203097, + "learning_rate": 7.948435859426755e-05, + "loss": 1.1039, + "step": 286 + }, + { + "epoch": 0.7340153452685422, + "grad_norm": 0.6300459744494643, + "learning_rate": 7.947286161097294e-05, + "loss": 1.1019, + "step": 287 + }, + { + "epoch": 0.7365728900255755, + "grad_norm": 0.8793621293918455, + "learning_rate": 7.94612387158405e-05, + "loss": 1.0641, + "step": 288 + }, + { + "epoch": 0.7391304347826086, + "grad_norm": 1.354355970537, + "learning_rate": 7.944948994594532e-05, + "loss": 1.0976, + "step": 289 + }, + { + "epoch": 0.7416879795396419, + "grad_norm": 0.811793129507913, + "learning_rate": 7.943761533876402e-05, + "loss": 1.1066, + "step": 290 + }, + { + "epoch": 0.7442455242966752, + "grad_norm": 0.6985981549179727, + "learning_rate": 7.94256149321746e-05, + "loss": 1.0889, + "step": 291 + }, + { + "epoch": 0.7468030690537084, + "grad_norm": 0.6979422293870231, + "learning_rate": 7.941348876445636e-05, + "loss": 1.0538, + "step": 292 + }, + { + "epoch": 0.7493606138107417, + "grad_norm": 0.5854843946594773, + "learning_rate": 7.940123687428975e-05, + "loss": 1.0817, + "step": 293 + }, + { + "epoch": 0.7519181585677749, + "grad_norm": 0.7081721412507801, + "learning_rate": 7.938885930075628e-05, + "loss": 1.0781, + "step": 294 + }, + { + "epoch": 0.7544757033248082, + "grad_norm": 0.7189613701722457, + "learning_rate": 7.937635608333829e-05, + "loss": 1.0709, + "step": 295 + }, + { + "epoch": 0.7570332480818415, + "grad_norm": 0.9612156787986404, + "learning_rate": 7.936372726191899e-05, + "loss": 1.0792, + "step": 296 + }, + { + "epoch": 0.7595907928388747, + "grad_norm": 1.5095702200934382, + "learning_rate": 7.935097287678222e-05, + "loss": 1.1203, + "step": 297 + }, + { + "epoch": 0.7621483375959079, + "grad_norm": 0.49810286262739045, + "learning_rate": 7.933809296861232e-05, + "loss": 1.0951, + "step": 298 + }, + { + "epoch": 0.7647058823529411, + "grad_norm": 1.094271200719579, + "learning_rate": 7.932508757849403e-05, + "loss": 1.1087, + "step": 299 + }, + { + "epoch": 0.7672634271099744, + "grad_norm": 1.6990794876777922, + "learning_rate": 7.931195674791242e-05, + "loss": 1.0713, + "step": 300 + }, + { + "epoch": 0.7698209718670077, + "grad_norm": 0.6778774182184969, + "learning_rate": 7.929870051875261e-05, + "loss": 1.1025, + "step": 301 + }, + { + "epoch": 0.7723785166240409, + "grad_norm": 1.9523378206319744, + "learning_rate": 7.928531893329977e-05, + "loss": 1.1099, + "step": 302 + }, + { + "epoch": 0.7749360613810742, + "grad_norm": 1.0174747598205622, + "learning_rate": 7.927181203423893e-05, + "loss": 1.0761, + "step": 303 + }, + { + "epoch": 0.7774936061381074, + "grad_norm": 2.097344028961776, + "learning_rate": 7.925817986465483e-05, + "loss": 1.0964, + "step": 304 + }, + { + "epoch": 0.7800511508951407, + "grad_norm": 1.7588955371199144, + "learning_rate": 7.924442246803183e-05, + "loss": 1.1114, + "step": 305 + }, + { + "epoch": 0.782608695652174, + "grad_norm": 1.1038887334502365, + "learning_rate": 7.923053988825372e-05, + "loss": 1.0863, + "step": 306 + }, + { + "epoch": 0.7851662404092071, + "grad_norm": 1.2204166281857975, + "learning_rate": 7.92165321696036e-05, + "loss": 1.0609, + "step": 307 + }, + { + "epoch": 0.7877237851662404, + "grad_norm": 0.6613406186209091, + "learning_rate": 7.920239935676377e-05, + "loss": 1.0832, + "step": 308 + }, + { + "epoch": 0.7902813299232737, + "grad_norm": 0.9099805345560761, + "learning_rate": 7.918814149481553e-05, + "loss": 1.1062, + "step": 309 + }, + { + "epoch": 0.7928388746803069, + "grad_norm": 0.8966794823734242, + "learning_rate": 7.917375862923908e-05, + "loss": 1.0699, + "step": 310 + }, + { + "epoch": 0.7953964194373402, + "grad_norm": 0.6431878954941771, + "learning_rate": 7.915925080591338e-05, + "loss": 1.0999, + "step": 311 + }, + { + "epoch": 0.7979539641943734, + "grad_norm": 0.7980040266876918, + "learning_rate": 7.914461807111595e-05, + "loss": 1.1169, + "step": 312 + }, + { + "epoch": 0.8005115089514067, + "grad_norm": 0.5792659205085796, + "learning_rate": 7.912986047152278e-05, + "loss": 1.0783, + "step": 313 + }, + { + "epoch": 0.80306905370844, + "grad_norm": 0.7059169343501349, + "learning_rate": 7.911497805420811e-05, + "loss": 1.0823, + "step": 314 + }, + { + "epoch": 0.8056265984654731, + "grad_norm": 0.5933585988646929, + "learning_rate": 7.909997086664443e-05, + "loss": 1.0748, + "step": 315 + }, + { + "epoch": 0.8081841432225064, + "grad_norm": 0.6845366174297771, + "learning_rate": 7.908483895670209e-05, + "loss": 1.071, + "step": 316 + }, + { + "epoch": 0.8107416879795396, + "grad_norm": 0.6043958213664847, + "learning_rate": 7.906958237264942e-05, + "loss": 1.0984, + "step": 317 + }, + { + "epoch": 0.8132992327365729, + "grad_norm": 0.8065799320997168, + "learning_rate": 7.905420116315234e-05, + "loss": 1.0908, + "step": 318 + }, + { + "epoch": 0.8158567774936062, + "grad_norm": 0.9613882476117077, + "learning_rate": 7.903869537727437e-05, + "loss": 1.0503, + "step": 319 + }, + { + "epoch": 0.8184143222506394, + "grad_norm": 0.8851238240180742, + "learning_rate": 7.902306506447634e-05, + "loss": 1.0783, + "step": 320 + }, + { + "epoch": 0.8209718670076727, + "grad_norm": 0.9671778018065025, + "learning_rate": 7.900731027461635e-05, + "loss": 1.057, + "step": 321 + }, + { + "epoch": 0.8235294117647058, + "grad_norm": 1.0461389584721013, + "learning_rate": 7.899143105794958e-05, + "loss": 1.087, + "step": 322 + }, + { + "epoch": 0.8260869565217391, + "grad_norm": 0.8134854113246186, + "learning_rate": 7.897542746512804e-05, + "loss": 1.1064, + "step": 323 + }, + { + "epoch": 0.8286445012787724, + "grad_norm": 0.7093110104105061, + "learning_rate": 7.895929954720053e-05, + "loss": 1.1046, + "step": 324 + }, + { + "epoch": 0.8312020460358056, + "grad_norm": 0.4793356093430287, + "learning_rate": 7.894304735561243e-05, + "loss": 1.0652, + "step": 325 + }, + { + "epoch": 0.8337595907928389, + "grad_norm": 0.5002076354784437, + "learning_rate": 7.89266709422055e-05, + "loss": 1.0756, + "step": 326 + }, + { + "epoch": 0.8363171355498721, + "grad_norm": 0.5414464668464288, + "learning_rate": 7.891017035921778e-05, + "loss": 1.1155, + "step": 327 + }, + { + "epoch": 0.8388746803069054, + "grad_norm": 0.6820350461047787, + "learning_rate": 7.889354565928337e-05, + "loss": 1.0879, + "step": 328 + }, + { + "epoch": 0.8414322250639387, + "grad_norm": 0.8316642019332523, + "learning_rate": 7.887679689543227e-05, + "loss": 1.1265, + "step": 329 + }, + { + "epoch": 0.8439897698209718, + "grad_norm": 0.8446501746399396, + "learning_rate": 7.885992412109026e-05, + "loss": 1.0971, + "step": 330 + }, + { + "epoch": 0.8465473145780051, + "grad_norm": 0.8372575100501372, + "learning_rate": 7.884292739007866e-05, + "loss": 1.0698, + "step": 331 + }, + { + "epoch": 0.8491048593350383, + "grad_norm": 0.751312270457101, + "learning_rate": 7.882580675661422e-05, + "loss": 1.0612, + "step": 332 + }, + { + "epoch": 0.8516624040920716, + "grad_norm": 0.8153412859957982, + "learning_rate": 7.88085622753089e-05, + "loss": 1.1092, + "step": 333 + }, + { + "epoch": 0.8542199488491049, + "grad_norm": 0.7887032728971717, + "learning_rate": 7.879119400116969e-05, + "loss": 1.0411, + "step": 334 + }, + { + "epoch": 0.8567774936061381, + "grad_norm": 0.7667290672650495, + "learning_rate": 7.877370198959852e-05, + "loss": 1.0671, + "step": 335 + }, + { + "epoch": 0.8593350383631714, + "grad_norm": 0.6040917626312119, + "learning_rate": 7.875608629639195e-05, + "loss": 1.067, + "step": 336 + }, + { + "epoch": 0.8618925831202046, + "grad_norm": 0.5108422350504948, + "learning_rate": 7.873834697774114e-05, + "loss": 1.0896, + "step": 337 + }, + { + "epoch": 0.8644501278772379, + "grad_norm": 0.6073358430679844, + "learning_rate": 7.872048409023153e-05, + "loss": 1.0945, + "step": 338 + }, + { + "epoch": 0.8670076726342711, + "grad_norm": 0.5811775734604341, + "learning_rate": 7.870249769084274e-05, + "loss": 1.0599, + "step": 339 + }, + { + "epoch": 0.8695652173913043, + "grad_norm": 0.4942443213793738, + "learning_rate": 7.868438783694841e-05, + "loss": 1.0984, + "step": 340 + }, + { + "epoch": 0.8721227621483376, + "grad_norm": 0.44691873435841717, + "learning_rate": 7.866615458631591e-05, + "loss": 1.0551, + "step": 341 + }, + { + "epoch": 0.8746803069053708, + "grad_norm": 0.49127713203197954, + "learning_rate": 7.864779799710631e-05, + "loss": 1.0821, + "step": 342 + }, + { + "epoch": 0.8772378516624041, + "grad_norm": 0.42857623809659445, + "learning_rate": 7.862931812787404e-05, + "loss": 1.0802, + "step": 343 + }, + { + "epoch": 0.8797953964194374, + "grad_norm": 0.5052119789372929, + "learning_rate": 7.86107150375668e-05, + "loss": 1.0876, + "step": 344 + }, + { + "epoch": 0.8823529411764706, + "grad_norm": 0.38100935660159413, + "learning_rate": 7.859198878552536e-05, + "loss": 1.068, + "step": 345 + }, + { + "epoch": 0.8849104859335039, + "grad_norm": 0.44991333074166684, + "learning_rate": 7.85731394314833e-05, + "loss": 1.1331, + "step": 346 + }, + { + "epoch": 0.887468030690537, + "grad_norm": 0.45213307645529244, + "learning_rate": 7.855416703556693e-05, + "loss": 1.0161, + "step": 347 + }, + { + "epoch": 0.8900255754475703, + "grad_norm": 0.4159819987382006, + "learning_rate": 7.853507165829504e-05, + "loss": 1.0811, + "step": 348 + }, + { + "epoch": 0.8925831202046036, + "grad_norm": 0.5486443431245583, + "learning_rate": 7.851585336057868e-05, + "loss": 1.0653, + "step": 349 + }, + { + "epoch": 0.8951406649616368, + "grad_norm": 0.6862683732800593, + "learning_rate": 7.849651220372101e-05, + "loss": 1.0867, + "step": 350 + }, + { + "epoch": 0.8976982097186701, + "grad_norm": 0.6981991056672471, + "learning_rate": 7.847704824941709e-05, + "loss": 1.105, + "step": 351 + }, + { + "epoch": 0.9002557544757033, + "grad_norm": 0.785234109714766, + "learning_rate": 7.84574615597537e-05, + "loss": 1.0596, + "step": 352 + }, + { + "epoch": 0.9028132992327366, + "grad_norm": 0.9770924073440606, + "learning_rate": 7.84377521972091e-05, + "loss": 1.0823, + "step": 353 + }, + { + "epoch": 0.9053708439897699, + "grad_norm": 1.1898436603817308, + "learning_rate": 7.841792022465286e-05, + "loss": 1.0763, + "step": 354 + }, + { + "epoch": 0.907928388746803, + "grad_norm": 0.8386259285895766, + "learning_rate": 7.839796570534566e-05, + "loss": 1.0542, + "step": 355 + }, + { + "epoch": 0.9104859335038363, + "grad_norm": 0.6389869741905022, + "learning_rate": 7.83778887029391e-05, + "loss": 1.0668, + "step": 356 + }, + { + "epoch": 0.9130434782608695, + "grad_norm": 0.5583652456462365, + "learning_rate": 7.835768928147545e-05, + "loss": 1.0759, + "step": 357 + }, + { + "epoch": 0.9156010230179028, + "grad_norm": 0.571315368558641, + "learning_rate": 7.83373675053875e-05, + "loss": 1.0976, + "step": 358 + }, + { + "epoch": 0.9181585677749361, + "grad_norm": 0.6447121021663385, + "learning_rate": 7.831692343949832e-05, + "loss": 1.0792, + "step": 359 + }, + { + "epoch": 0.9207161125319693, + "grad_norm": 0.7085586190843103, + "learning_rate": 7.829635714902108e-05, + "loss": 1.0461, + "step": 360 + }, + { + "epoch": 0.9232736572890026, + "grad_norm": 0.8729805375627833, + "learning_rate": 7.827566869955878e-05, + "loss": 1.0957, + "step": 361 + }, + { + "epoch": 0.9258312020460358, + "grad_norm": 1.1209022936512785, + "learning_rate": 7.825485815710415e-05, + "loss": 1.0246, + "step": 362 + }, + { + "epoch": 0.928388746803069, + "grad_norm": 1.0159476142523682, + "learning_rate": 7.823392558803933e-05, + "loss": 1.0859, + "step": 363 + }, + { + "epoch": 0.9309462915601023, + "grad_norm": 0.8822113574613138, + "learning_rate": 7.821287105913571e-05, + "loss": 1.069, + "step": 364 + }, + { + "epoch": 0.9335038363171355, + "grad_norm": 0.8276272022971064, + "learning_rate": 7.819169463755376e-05, + "loss": 1.0857, + "step": 365 + }, + { + "epoch": 0.9360613810741688, + "grad_norm": 0.814599705940635, + "learning_rate": 7.817039639084271e-05, + "loss": 1.1084, + "step": 366 + }, + { + "epoch": 0.9386189258312021, + "grad_norm": 0.9444127014112077, + "learning_rate": 7.814897638694039e-05, + "loss": 1.0864, + "step": 367 + }, + { + "epoch": 0.9411764705882353, + "grad_norm": 0.9252499734817574, + "learning_rate": 7.812743469417306e-05, + "loss": 1.0797, + "step": 368 + }, + { + "epoch": 0.9437340153452686, + "grad_norm": 0.8304614201741923, + "learning_rate": 7.810577138125512e-05, + "loss": 1.074, + "step": 369 + }, + { + "epoch": 0.9462915601023018, + "grad_norm": 0.7597853417158138, + "learning_rate": 7.808398651728891e-05, + "loss": 1.0704, + "step": 370 + }, + { + "epoch": 0.948849104859335, + "grad_norm": 0.6492633689174822, + "learning_rate": 7.806208017176453e-05, + "loss": 1.0978, + "step": 371 + }, + { + "epoch": 0.9514066496163683, + "grad_norm": 0.7103188739776854, + "learning_rate": 7.804005241455954e-05, + "loss": 1.0799, + "step": 372 + }, + { + "epoch": 0.9539641943734015, + "grad_norm": 0.8080045645046198, + "learning_rate": 7.801790331593882e-05, + "loss": 1.0847, + "step": 373 + }, + { + "epoch": 0.9565217391304348, + "grad_norm": 0.7941792943654038, + "learning_rate": 7.799563294655429e-05, + "loss": 1.0843, + "step": 374 + }, + { + "epoch": 0.959079283887468, + "grad_norm": 0.5866783302080361, + "learning_rate": 7.797324137744472e-05, + "loss": 1.1051, + "step": 375 + }, + { + "epoch": 0.9616368286445013, + "grad_norm": 0.4359647133720881, + "learning_rate": 7.795072868003546e-05, + "loss": 1.0541, + "step": 376 + }, + { + "epoch": 0.9641943734015346, + "grad_norm": 0.5857374136808916, + "learning_rate": 7.792809492613828e-05, + "loss": 1.0867, + "step": 377 + }, + { + "epoch": 0.9667519181585678, + "grad_norm": 0.6385407081997835, + "learning_rate": 7.790534018795106e-05, + "loss": 1.0925, + "step": 378 + }, + { + "epoch": 0.969309462915601, + "grad_norm": 0.6434152574400142, + "learning_rate": 7.788246453805762e-05, + "loss": 1.0609, + "step": 379 + }, + { + "epoch": 0.9718670076726342, + "grad_norm": 0.6915297988962398, + "learning_rate": 7.785946804942746e-05, + "loss": 1.0443, + "step": 380 + }, + { + "epoch": 0.9744245524296675, + "grad_norm": 0.7443780294295859, + "learning_rate": 7.783635079541553e-05, + "loss": 1.0742, + "step": 381 + }, + { + "epoch": 0.9769820971867008, + "grad_norm": 0.7984905238389998, + "learning_rate": 7.781311284976204e-05, + "loss": 1.0784, + "step": 382 + }, + { + "epoch": 0.979539641943734, + "grad_norm": 0.7209084804370026, + "learning_rate": 7.778975428659213e-05, + "loss": 1.0828, + "step": 383 + }, + { + "epoch": 0.9820971867007673, + "grad_norm": 0.5272873910092764, + "learning_rate": 7.776627518041573e-05, + "loss": 1.0359, + "step": 384 + }, + { + "epoch": 0.9846547314578005, + "grad_norm": 0.5081491988480245, + "learning_rate": 7.774267560612725e-05, + "loss": 1.1162, + "step": 385 + }, + { + "epoch": 0.9872122762148338, + "grad_norm": 0.47763924105204525, + "learning_rate": 7.771895563900542e-05, + "loss": 1.0707, + "step": 386 + }, + { + "epoch": 0.989769820971867, + "grad_norm": 0.4293473992712172, + "learning_rate": 7.769511535471294e-05, + "loss": 1.0835, + "step": 387 + }, + { + "epoch": 0.9923273657289002, + "grad_norm": 0.4033861632452256, + "learning_rate": 7.767115482929638e-05, + "loss": 1.0974, + "step": 388 + }, + { + "epoch": 0.9948849104859335, + "grad_norm": 0.5275929751722469, + "learning_rate": 7.764707413918578e-05, + "loss": 1.0806, + "step": 389 + }, + { + "epoch": 0.9974424552429667, + "grad_norm": 0.5525995164607961, + "learning_rate": 7.762287336119453e-05, + "loss": 1.0609, + "step": 390 + }, + { + "epoch": 1.0, + "grad_norm": 0.573425625533648, + "learning_rate": 7.759855257251907e-05, + "loss": 1.0896, + "step": 391 + }, + { + "epoch": 1.0025575447570332, + "grad_norm": 0.6121282952092544, + "learning_rate": 7.757411185073866e-05, + "loss": 1.0565, + "step": 392 + }, + { + "epoch": 1.0051150895140666, + "grad_norm": 0.6487069044059379, + "learning_rate": 7.754955127381512e-05, + "loss": 1.0719, + "step": 393 + }, + { + "epoch": 1.0076726342710998, + "grad_norm": 0.7649726943677306, + "learning_rate": 7.75248709200926e-05, + "loss": 1.044, + "step": 394 + }, + { + "epoch": 1.010230179028133, + "grad_norm": 1.0004112774028269, + "learning_rate": 7.750007086829729e-05, + "loss": 1.0534, + "step": 395 + }, + { + "epoch": 1.0127877237851663, + "grad_norm": 1.3267806534444309, + "learning_rate": 7.747515119753721e-05, + "loss": 1.0731, + "step": 396 + }, + { + "epoch": 1.0153452685421995, + "grad_norm": 0.535485816275368, + "learning_rate": 7.745011198730196e-05, + "loss": 1.0712, + "step": 397 + }, + { + "epoch": 1.0179028132992327, + "grad_norm": 0.6246263933300512, + "learning_rate": 7.742495331746246e-05, + "loss": 1.0622, + "step": 398 + }, + { + "epoch": 1.020460358056266, + "grad_norm": 1.2351151058960912, + "learning_rate": 7.739967526827063e-05, + "loss": 1.0357, + "step": 399 + }, + { + "epoch": 1.0230179028132993, + "grad_norm": 0.9704077261446196, + "learning_rate": 7.737427792035923e-05, + "loss": 1.0885, + "step": 400 + }, + { + "epoch": 1.0255754475703325, + "grad_norm": 0.8543383222043499, + "learning_rate": 7.73487613547416e-05, + "loss": 1.0677, + "step": 401 + }, + { + "epoch": 1.0281329923273657, + "grad_norm": 0.5879183387167717, + "learning_rate": 7.732312565281128e-05, + "loss": 1.0844, + "step": 402 + }, + { + "epoch": 1.030690537084399, + "grad_norm": 0.4537293576835104, + "learning_rate": 7.72973708963419e-05, + "loss": 1.0497, + "step": 403 + }, + { + "epoch": 1.0332480818414322, + "grad_norm": 0.5939375825104586, + "learning_rate": 7.727149716748681e-05, + "loss": 1.0572, + "step": 404 + }, + { + "epoch": 1.0358056265984654, + "grad_norm": 0.6995057421533399, + "learning_rate": 7.724550454877892e-05, + "loss": 1.0549, + "step": 405 + }, + { + "epoch": 1.0383631713554988, + "grad_norm": 0.7494681155888331, + "learning_rate": 7.721939312313033e-05, + "loss": 1.0869, + "step": 406 + }, + { + "epoch": 1.040920716112532, + "grad_norm": 0.7643538285835613, + "learning_rate": 7.71931629738321e-05, + "loss": 1.022, + "step": 407 + }, + { + "epoch": 1.0434782608695652, + "grad_norm": 0.6356534653615626, + "learning_rate": 7.716681418455409e-05, + "loss": 1.0529, + "step": 408 + }, + { + "epoch": 1.0460358056265984, + "grad_norm": 0.5605358498748841, + "learning_rate": 7.714034683934446e-05, + "loss": 1.034, + "step": 409 + }, + { + "epoch": 1.0485933503836318, + "grad_norm": 0.5583980180495551, + "learning_rate": 7.711376102262968e-05, + "loss": 1.0282, + "step": 410 + }, + { + "epoch": 1.051150895140665, + "grad_norm": 0.6738348730192517, + "learning_rate": 7.708705681921405e-05, + "loss": 1.0553, + "step": 411 + }, + { + "epoch": 1.0537084398976981, + "grad_norm": 0.5298166396458623, + "learning_rate": 7.706023431427951e-05, + "loss": 1.0365, + "step": 412 + }, + { + "epoch": 1.0562659846547315, + "grad_norm": 0.4983596119303216, + "learning_rate": 7.703329359338538e-05, + "loss": 1.0437, + "step": 413 + }, + { + "epoch": 1.0588235294117647, + "grad_norm": 0.5960572452353167, + "learning_rate": 7.700623474246806e-05, + "loss": 1.0898, + "step": 414 + }, + { + "epoch": 1.061381074168798, + "grad_norm": 0.5540140213615853, + "learning_rate": 7.697905784784077e-05, + "loss": 1.061, + "step": 415 + }, + { + "epoch": 1.0639386189258313, + "grad_norm": 0.6096623152113402, + "learning_rate": 7.695176299619327e-05, + "loss": 1.0433, + "step": 416 + }, + { + "epoch": 1.0664961636828645, + "grad_norm": 0.5729999229219644, + "learning_rate": 7.692435027459157e-05, + "loss": 1.0268, + "step": 417 + }, + { + "epoch": 1.0690537084398977, + "grad_norm": 0.516351992871556, + "learning_rate": 7.689681977047767e-05, + "loss": 1.0372, + "step": 418 + }, + { + "epoch": 1.0716112531969308, + "grad_norm": 0.5409419169165403, + "learning_rate": 7.68691715716693e-05, + "loss": 1.0924, + "step": 419 + }, + { + "epoch": 1.0741687979539642, + "grad_norm": 0.481066894118381, + "learning_rate": 7.684140576635959e-05, + "loss": 1.0415, + "step": 420 + }, + { + "epoch": 1.0767263427109974, + "grad_norm": 0.4798293644502134, + "learning_rate": 7.68135224431168e-05, + "loss": 1.0556, + "step": 421 + }, + { + "epoch": 1.0792838874680306, + "grad_norm": 0.4778706340999981, + "learning_rate": 7.678552169088411e-05, + "loss": 1.0458, + "step": 422 + }, + { + "epoch": 1.081841432225064, + "grad_norm": 0.47353441131918134, + "learning_rate": 7.675740359897923e-05, + "loss": 1.0314, + "step": 423 + }, + { + "epoch": 1.0843989769820972, + "grad_norm": 0.6103257451747757, + "learning_rate": 7.672916825709417e-05, + "loss": 1.0779, + "step": 424 + }, + { + "epoch": 1.0869565217391304, + "grad_norm": 0.6787803255539844, + "learning_rate": 7.670081575529499e-05, + "loss": 1.0412, + "step": 425 + }, + { + "epoch": 1.0895140664961638, + "grad_norm": 0.5996475603666427, + "learning_rate": 7.667234618402139e-05, + "loss": 1.0312, + "step": 426 + }, + { + "epoch": 1.092071611253197, + "grad_norm": 0.6420838832118385, + "learning_rate": 7.664375963408657e-05, + "loss": 1.0559, + "step": 427 + }, + { + "epoch": 1.0946291560102301, + "grad_norm": 0.6257260302049759, + "learning_rate": 7.661505619667685e-05, + "loss": 1.0367, + "step": 428 + }, + { + "epoch": 1.0971867007672633, + "grad_norm": 0.6078916092596967, + "learning_rate": 7.658623596335141e-05, + "loss": 1.0662, + "step": 429 + }, + { + "epoch": 1.0997442455242967, + "grad_norm": 0.6869715214996625, + "learning_rate": 7.655729902604198e-05, + "loss": 1.0426, + "step": 430 + }, + { + "epoch": 1.10230179028133, + "grad_norm": 0.7202490835032408, + "learning_rate": 7.652824547705255e-05, + "loss": 1.0724, + "step": 431 + }, + { + "epoch": 1.104859335038363, + "grad_norm": 0.6798833175024812, + "learning_rate": 7.64990754090591e-05, + "loss": 1.0899, + "step": 432 + }, + { + "epoch": 1.1074168797953965, + "grad_norm": 0.5492648806120208, + "learning_rate": 7.646978891510927e-05, + "loss": 1.0671, + "step": 433 + }, + { + "epoch": 1.1099744245524297, + "grad_norm": 0.502990998641414, + "learning_rate": 7.64403860886221e-05, + "loss": 1.0558, + "step": 434 + }, + { + "epoch": 1.1125319693094629, + "grad_norm": 0.37487945047582993, + "learning_rate": 7.641086702338766e-05, + "loss": 1.0477, + "step": 435 + }, + { + "epoch": 1.1150895140664963, + "grad_norm": 0.41781785412216255, + "learning_rate": 7.638123181356688e-05, + "loss": 1.046, + "step": 436 + }, + { + "epoch": 1.1176470588235294, + "grad_norm": 0.5447678799969146, + "learning_rate": 7.635148055369111e-05, + "loss": 1.0268, + "step": 437 + }, + { + "epoch": 1.1202046035805626, + "grad_norm": 0.7354119493315371, + "learning_rate": 7.63216133386619e-05, + "loss": 1.0407, + "step": 438 + }, + { + "epoch": 1.1227621483375958, + "grad_norm": 0.8468997724198893, + "learning_rate": 7.62916302637507e-05, + "loss": 1.0613, + "step": 439 + }, + { + "epoch": 1.1253196930946292, + "grad_norm": 0.90648052854961, + "learning_rate": 7.62615314245985e-05, + "loss": 1.0389, + "step": 440 + }, + { + "epoch": 1.1278772378516624, + "grad_norm": 1.090719930551356, + "learning_rate": 7.623131691721556e-05, + "loss": 1.0604, + "step": 441 + }, + { + "epoch": 1.1304347826086956, + "grad_norm": 0.8711758817781659, + "learning_rate": 7.620098683798113e-05, + "loss": 1.0396, + "step": 442 + }, + { + "epoch": 1.132992327365729, + "grad_norm": 0.7941323029712526, + "learning_rate": 7.617054128364308e-05, + "loss": 1.064, + "step": 443 + }, + { + "epoch": 1.1355498721227621, + "grad_norm": 0.7358238997014726, + "learning_rate": 7.613998035131767e-05, + "loss": 1.0302, + "step": 444 + }, + { + "epoch": 1.1381074168797953, + "grad_norm": 0.8267368055242302, + "learning_rate": 7.610930413848916e-05, + "loss": 1.0499, + "step": 445 + }, + { + "epoch": 1.1406649616368287, + "grad_norm": 0.8808186893039806, + "learning_rate": 7.607851274300954e-05, + "loss": 1.0241, + "step": 446 + }, + { + "epoch": 1.143222506393862, + "grad_norm": 0.844367546052809, + "learning_rate": 7.604760626309822e-05, + "loss": 1.0136, + "step": 447 + }, + { + "epoch": 1.145780051150895, + "grad_norm": 0.9008836613292239, + "learning_rate": 7.601658479734172e-05, + "loss": 1.0478, + "step": 448 + }, + { + "epoch": 1.1483375959079285, + "grad_norm": 0.8890437058493534, + "learning_rate": 7.598544844469335e-05, + "loss": 1.0371, + "step": 449 + }, + { + "epoch": 1.1508951406649617, + "grad_norm": 0.7574790514791548, + "learning_rate": 7.595419730447283e-05, + "loss": 1.0871, + "step": 450 + }, + { + "epoch": 1.1534526854219949, + "grad_norm": 0.6622645508305952, + "learning_rate": 7.592283147636613e-05, + "loss": 1.0584, + "step": 451 + }, + { + "epoch": 1.156010230179028, + "grad_norm": 0.5240209699484857, + "learning_rate": 7.589135106042497e-05, + "loss": 1.061, + "step": 452 + }, + { + "epoch": 1.1585677749360614, + "grad_norm": 0.7258085512815178, + "learning_rate": 7.585975615706664e-05, + "loss": 1.0782, + "step": 453 + }, + { + "epoch": 1.1611253196930946, + "grad_norm": 0.8692692426989976, + "learning_rate": 7.582804686707361e-05, + "loss": 1.0487, + "step": 454 + }, + { + "epoch": 1.1636828644501278, + "grad_norm": 0.7256777326774219, + "learning_rate": 7.57962232915932e-05, + "loss": 1.0443, + "step": 455 + }, + { + "epoch": 1.1662404092071612, + "grad_norm": 0.5531298720327349, + "learning_rate": 7.576428553213731e-05, + "loss": 1.0366, + "step": 456 + }, + { + "epoch": 1.1687979539641944, + "grad_norm": 0.6217533009743852, + "learning_rate": 7.573223369058206e-05, + "loss": 1.0295, + "step": 457 + }, + { + "epoch": 1.1713554987212276, + "grad_norm": 0.49232292611041006, + "learning_rate": 7.570006786916749e-05, + "loss": 1.0329, + "step": 458 + }, + { + "epoch": 1.1739130434782608, + "grad_norm": 0.52993398976404, + "learning_rate": 7.566778817049718e-05, + "loss": 1.0402, + "step": 459 + }, + { + "epoch": 1.1764705882352942, + "grad_norm": 0.5585620019915892, + "learning_rate": 7.563539469753798e-05, + "loss": 1.069, + "step": 460 + }, + { + "epoch": 1.1790281329923273, + "grad_norm": 0.6428206102008164, + "learning_rate": 7.560288755361967e-05, + "loss": 1.0908, + "step": 461 + }, + { + "epoch": 1.1815856777493605, + "grad_norm": 0.6280277113340915, + "learning_rate": 7.557026684243459e-05, + "loss": 1.0348, + "step": 462 + }, + { + "epoch": 1.184143222506394, + "grad_norm": 0.43288917306833347, + "learning_rate": 7.55375326680374e-05, + "loss": 1.0241, + "step": 463 + }, + { + "epoch": 1.186700767263427, + "grad_norm": 0.42841127990041067, + "learning_rate": 7.550468513484461e-05, + "loss": 1.0514, + "step": 464 + }, + { + "epoch": 1.1892583120204603, + "grad_norm": 0.4254341549287906, + "learning_rate": 7.54717243476344e-05, + "loss": 1.0606, + "step": 465 + }, + { + "epoch": 1.1918158567774937, + "grad_norm": 0.4171053731115308, + "learning_rate": 7.543865041154616e-05, + "loss": 1.0563, + "step": 466 + }, + { + "epoch": 1.1943734015345269, + "grad_norm": 0.4236528757379926, + "learning_rate": 7.540546343208021e-05, + "loss": 1.0336, + "step": 467 + }, + { + "epoch": 1.19693094629156, + "grad_norm": 0.4994661910864934, + "learning_rate": 7.537216351509752e-05, + "loss": 1.0844, + "step": 468 + }, + { + "epoch": 1.1994884910485935, + "grad_norm": 0.6198060558124957, + "learning_rate": 7.533875076681923e-05, + "loss": 1.0492, + "step": 469 + }, + { + "epoch": 1.2020460358056266, + "grad_norm": 0.583735235841795, + "learning_rate": 7.530522529382641e-05, + "loss": 1.071, + "step": 470 + }, + { + "epoch": 1.2046035805626598, + "grad_norm": 0.5659582359875482, + "learning_rate": 7.527158720305975e-05, + "loss": 1.0429, + "step": 471 + }, + { + "epoch": 1.207161125319693, + "grad_norm": 0.5451963873404054, + "learning_rate": 7.523783660181915e-05, + "loss": 1.0021, + "step": 472 + }, + { + "epoch": 1.2097186700767264, + "grad_norm": 0.4423684081873438, + "learning_rate": 7.520397359776338e-05, + "loss": 1.0574, + "step": 473 + }, + { + "epoch": 1.2122762148337596, + "grad_norm": 0.5198395882214115, + "learning_rate": 7.516999829890976e-05, + "loss": 1.0409, + "step": 474 + }, + { + "epoch": 1.2148337595907928, + "grad_norm": 0.5360908221928974, + "learning_rate": 7.513591081363384e-05, + "loss": 1.0555, + "step": 475 + }, + { + "epoch": 1.2173913043478262, + "grad_norm": 0.40726862330103586, + "learning_rate": 7.5101711250669e-05, + "loss": 1.0253, + "step": 476 + }, + { + "epoch": 1.2199488491048593, + "grad_norm": 0.554526695474263, + "learning_rate": 7.506739971910613e-05, + "loss": 1.0792, + "step": 477 + }, + { + "epoch": 1.2225063938618925, + "grad_norm": 0.6837744946054213, + "learning_rate": 7.503297632839328e-05, + "loss": 1.0618, + "step": 478 + }, + { + "epoch": 1.2250639386189257, + "grad_norm": 0.787840350339849, + "learning_rate": 7.499844118833533e-05, + "loss": 1.0711, + "step": 479 + }, + { + "epoch": 1.227621483375959, + "grad_norm": 0.9184717384704346, + "learning_rate": 7.496379440909361e-05, + "loss": 1.0225, + "step": 480 + }, + { + "epoch": 1.2301790281329923, + "grad_norm": 0.98861238646143, + "learning_rate": 7.492903610118556e-05, + "loss": 1.0322, + "step": 481 + }, + { + "epoch": 1.2327365728900257, + "grad_norm": 0.8602509324185822, + "learning_rate": 7.489416637548436e-05, + "loss": 1.0511, + "step": 482 + }, + { + "epoch": 1.2352941176470589, + "grad_norm": 0.7497024524996155, + "learning_rate": 7.485918534321866e-05, + "loss": 1.0439, + "step": 483 + }, + { + "epoch": 1.237851662404092, + "grad_norm": 0.6658522387073332, + "learning_rate": 7.482409311597206e-05, + "loss": 1.0597, + "step": 484 + }, + { + "epoch": 1.2404092071611252, + "grad_norm": 0.6041271689399434, + "learning_rate": 7.478888980568294e-05, + "loss": 1.0622, + "step": 485 + }, + { + "epoch": 1.2429667519181586, + "grad_norm": 0.5141728116418924, + "learning_rate": 7.475357552464397e-05, + "loss": 1.0559, + "step": 486 + }, + { + "epoch": 1.2455242966751918, + "grad_norm": 0.3924962212572937, + "learning_rate": 7.47181503855018e-05, + "loss": 1.0283, + "step": 487 + }, + { + "epoch": 1.248081841432225, + "grad_norm": 0.4006024520158971, + "learning_rate": 7.468261450125674e-05, + "loss": 1.0463, + "step": 488 + }, + { + "epoch": 1.2506393861892584, + "grad_norm": 0.3732442212525654, + "learning_rate": 7.464696798526229e-05, + "loss": 1.0553, + "step": 489 + }, + { + "epoch": 1.2531969309462916, + "grad_norm": 0.4036756007900454, + "learning_rate": 7.461121095122491e-05, + "loss": 1.0754, + "step": 490 + }, + { + "epoch": 1.2557544757033248, + "grad_norm": 0.4521255607620733, + "learning_rate": 7.457534351320356e-05, + "loss": 1.0475, + "step": 491 + }, + { + "epoch": 1.258312020460358, + "grad_norm": 0.4480587687694347, + "learning_rate": 7.453936578560937e-05, + "loss": 1.0724, + "step": 492 + }, + { + "epoch": 1.2608695652173914, + "grad_norm": 0.4064497930563765, + "learning_rate": 7.450327788320529e-05, + "loss": 1.0411, + "step": 493 + }, + { + "epoch": 1.2634271099744245, + "grad_norm": 0.32223473130563357, + "learning_rate": 7.446707992110571e-05, + "loss": 1.0391, + "step": 494 + }, + { + "epoch": 1.265984654731458, + "grad_norm": 0.33489725476820903, + "learning_rate": 7.443077201477609e-05, + "loss": 1.0631, + "step": 495 + }, + { + "epoch": 1.2685421994884911, + "grad_norm": 0.34146623369427676, + "learning_rate": 7.439435428003256e-05, + "loss": 1.0523, + "step": 496 + }, + { + "epoch": 1.2710997442455243, + "grad_norm": 0.29613834050231674, + "learning_rate": 7.435782683304164e-05, + "loss": 1.0446, + "step": 497 + }, + { + "epoch": 1.2736572890025575, + "grad_norm": 0.3841724281833824, + "learning_rate": 7.43211897903198e-05, + "loss": 1.0306, + "step": 498 + }, + { + "epoch": 1.2762148337595907, + "grad_norm": 0.4433757479687236, + "learning_rate": 7.428444326873303e-05, + "loss": 1.0448, + "step": 499 + }, + { + "epoch": 1.278772378516624, + "grad_norm": 0.36337315524962616, + "learning_rate": 7.424758738549667e-05, + "loss": 1.0201, + "step": 500 + }, + { + "epoch": 1.2813299232736572, + "grad_norm": 0.29686335739489117, + "learning_rate": 7.421062225817478e-05, + "loss": 1.0397, + "step": 501 + }, + { + "epoch": 1.2838874680306906, + "grad_norm": 0.41385410051793525, + "learning_rate": 7.417354800467997e-05, + "loss": 1.0449, + "step": 502 + }, + { + "epoch": 1.2864450127877238, + "grad_norm": 0.41885148772105985, + "learning_rate": 7.413636474327292e-05, + "loss": 1.0642, + "step": 503 + }, + { + "epoch": 1.289002557544757, + "grad_norm": 0.43853476074138853, + "learning_rate": 7.409907259256202e-05, + "loss": 1.0397, + "step": 504 + }, + { + "epoch": 1.2915601023017902, + "grad_norm": 0.6686928411130255, + "learning_rate": 7.4061671671503e-05, + "loss": 1.0542, + "step": 505 + }, + { + "epoch": 1.2941176470588236, + "grad_norm": 0.607563092071724, + "learning_rate": 7.402416209939856e-05, + "loss": 1.0304, + "step": 506 + }, + { + "epoch": 1.2966751918158568, + "grad_norm": 0.580772361466558, + "learning_rate": 7.398654399589795e-05, + "loss": 1.0632, + "step": 507 + }, + { + "epoch": 1.29923273657289, + "grad_norm": 0.7281319746142754, + "learning_rate": 7.394881748099668e-05, + "loss": 1.026, + "step": 508 + }, + { + "epoch": 1.3017902813299234, + "grad_norm": 0.782767959332906, + "learning_rate": 7.391098267503599e-05, + "loss": 1.0584, + "step": 509 + }, + { + "epoch": 1.3043478260869565, + "grad_norm": 0.7334069729083356, + "learning_rate": 7.387303969870262e-05, + "loss": 1.0265, + "step": 510 + }, + { + "epoch": 1.3069053708439897, + "grad_norm": 0.8443472328520687, + "learning_rate": 7.383498867302835e-05, + "loss": 1.0362, + "step": 511 + }, + { + "epoch": 1.309462915601023, + "grad_norm": 0.9845489824701952, + "learning_rate": 7.379682971938953e-05, + "loss": 1.0658, + "step": 512 + }, + { + "epoch": 1.3120204603580563, + "grad_norm": 1.0456842320775128, + "learning_rate": 7.375856295950693e-05, + "loss": 1.0222, + "step": 513 + }, + { + "epoch": 1.3145780051150895, + "grad_norm": 0.872370978809925, + "learning_rate": 7.372018851544507e-05, + "loss": 1.0431, + "step": 514 + }, + { + "epoch": 1.317135549872123, + "grad_norm": 0.5883115072059536, + "learning_rate": 7.368170650961203e-05, + "loss": 1.0349, + "step": 515 + }, + { + "epoch": 1.319693094629156, + "grad_norm": 0.4266657814616918, + "learning_rate": 7.364311706475901e-05, + "loss": 1.0304, + "step": 516 + }, + { + "epoch": 1.3222506393861893, + "grad_norm": 0.5955652733904395, + "learning_rate": 7.360442030397987e-05, + "loss": 1.0267, + "step": 517 + }, + { + "epoch": 1.3248081841432224, + "grad_norm": 0.6659451706135621, + "learning_rate": 7.356561635071082e-05, + "loss": 1.0437, + "step": 518 + }, + { + "epoch": 1.3273657289002558, + "grad_norm": 0.6313803070585998, + "learning_rate": 7.352670532872999e-05, + "loss": 1.0342, + "step": 519 + }, + { + "epoch": 1.329923273657289, + "grad_norm": 0.6906340675596294, + "learning_rate": 7.348768736215706e-05, + "loss": 1.0784, + "step": 520 + }, + { + "epoch": 1.3324808184143222, + "grad_norm": 0.6387943436906747, + "learning_rate": 7.34485625754528e-05, + "loss": 1.0785, + "step": 521 + }, + { + "epoch": 1.3350383631713556, + "grad_norm": 0.6214194954347501, + "learning_rate": 7.340933109341878e-05, + "loss": 1.0554, + "step": 522 + }, + { + "epoch": 1.3375959079283888, + "grad_norm": 0.6783392090338549, + "learning_rate": 7.336999304119685e-05, + "loss": 1.0461, + "step": 523 + }, + { + "epoch": 1.340153452685422, + "grad_norm": 0.7773583234292399, + "learning_rate": 7.333054854426884e-05, + "loss": 1.0714, + "step": 524 + }, + { + "epoch": 1.3427109974424551, + "grad_norm": 0.7619691171883715, + "learning_rate": 7.329099772845611e-05, + "loss": 1.0427, + "step": 525 + }, + { + "epoch": 1.3452685421994885, + "grad_norm": 0.6796465920444608, + "learning_rate": 7.325134071991915e-05, + "loss": 1.0673, + "step": 526 + }, + { + "epoch": 1.3478260869565217, + "grad_norm": 0.5764938509906341, + "learning_rate": 7.321157764515721e-05, + "loss": 1.0172, + "step": 527 + }, + { + "epoch": 1.350383631713555, + "grad_norm": 0.5703898248580829, + "learning_rate": 7.317170863100786e-05, + "loss": 1.0814, + "step": 528 + }, + { + "epoch": 1.3529411764705883, + "grad_norm": 0.543928921118232, + "learning_rate": 7.313173380464658e-05, + "loss": 1.0427, + "step": 529 + }, + { + "epoch": 1.3554987212276215, + "grad_norm": 0.403264231625384, + "learning_rate": 7.309165329358639e-05, + "loss": 1.0141, + "step": 530 + }, + { + "epoch": 1.3580562659846547, + "grad_norm": 0.37834189295305315, + "learning_rate": 7.305146722567746e-05, + "loss": 1.0518, + "step": 531 + }, + { + "epoch": 1.3606138107416879, + "grad_norm": 0.4423087622812266, + "learning_rate": 7.301117572910661e-05, + "loss": 1.0374, + "step": 532 + }, + { + "epoch": 1.3631713554987213, + "grad_norm": 0.6513039474923614, + "learning_rate": 7.297077893239699e-05, + "loss": 1.0751, + "step": 533 + }, + { + "epoch": 1.3657289002557544, + "grad_norm": 0.7901236558090899, + "learning_rate": 7.293027696440763e-05, + "loss": 1.0429, + "step": 534 + }, + { + "epoch": 1.3682864450127878, + "grad_norm": 0.8028952131258908, + "learning_rate": 7.288966995433309e-05, + "loss": 1.0531, + "step": 535 + }, + { + "epoch": 1.370843989769821, + "grad_norm": 0.7531252236472497, + "learning_rate": 7.284895803170288e-05, + "loss": 1.0875, + "step": 536 + }, + { + "epoch": 1.3734015345268542, + "grad_norm": 0.7178997040327029, + "learning_rate": 7.280814132638129e-05, + "loss": 1.0439, + "step": 537 + }, + { + "epoch": 1.3759590792838874, + "grad_norm": 0.5847814083136553, + "learning_rate": 7.276721996856677e-05, + "loss": 1.0554, + "step": 538 + }, + { + "epoch": 1.3785166240409208, + "grad_norm": 0.49255490697808374, + "learning_rate": 7.272619408879164e-05, + "loss": 1.0404, + "step": 539 + }, + { + "epoch": 1.381074168797954, + "grad_norm": 0.45239860846158764, + "learning_rate": 7.268506381792158e-05, + "loss": 1.049, + "step": 540 + }, + { + "epoch": 1.3836317135549872, + "grad_norm": 0.68450243672809, + "learning_rate": 7.26438292871553e-05, + "loss": 1.076, + "step": 541 + }, + { + "epoch": 1.3861892583120206, + "grad_norm": 0.6927411824014928, + "learning_rate": 7.260249062802407e-05, + "loss": 1.0677, + "step": 542 + }, + { + "epoch": 1.3887468030690537, + "grad_norm": 0.6632064448071592, + "learning_rate": 7.256104797239132e-05, + "loss": 1.0227, + "step": 543 + }, + { + "epoch": 1.391304347826087, + "grad_norm": 0.6518144933302445, + "learning_rate": 7.251950145245218e-05, + "loss": 1.044, + "step": 544 + }, + { + "epoch": 1.39386189258312, + "grad_norm": 0.5815850660081673, + "learning_rate": 7.247785120073313e-05, + "loss": 1.0333, + "step": 545 + }, + { + "epoch": 1.3964194373401535, + "grad_norm": 0.47544855954948456, + "learning_rate": 7.24360973500915e-05, + "loss": 1.0128, + "step": 546 + }, + { + "epoch": 1.3989769820971867, + "grad_norm": 0.5253477384563784, + "learning_rate": 7.239424003371511e-05, + "loss": 1.0519, + "step": 547 + }, + { + "epoch": 1.40153452685422, + "grad_norm": 0.54424332711931, + "learning_rate": 7.235227938512182e-05, + "loss": 1.0421, + "step": 548 + }, + { + "epoch": 1.4040920716112533, + "grad_norm": 0.4163026169587667, + "learning_rate": 7.231021553815909e-05, + "loss": 1.0647, + "step": 549 + }, + { + "epoch": 1.4066496163682864, + "grad_norm": 0.43975874335531806, + "learning_rate": 7.226804862700357e-05, + "loss": 1.006, + "step": 550 + }, + { + "epoch": 1.4092071611253196, + "grad_norm": 0.5676277493519121, + "learning_rate": 7.222577878616065e-05, + "loss": 1.018, + "step": 551 + }, + { + "epoch": 1.4117647058823528, + "grad_norm": 0.4945104420529325, + "learning_rate": 7.218340615046407e-05, + "loss": 1.0609, + "step": 552 + }, + { + "epoch": 1.4143222506393862, + "grad_norm": 0.42595338834103746, + "learning_rate": 7.214093085507549e-05, + "loss": 1.0455, + "step": 553 + }, + { + "epoch": 1.4168797953964194, + "grad_norm": 0.3651457329411855, + "learning_rate": 7.209835303548396e-05, + "loss": 1.0558, + "step": 554 + }, + { + "epoch": 1.4194373401534528, + "grad_norm": 0.48372716214113715, + "learning_rate": 7.205567282750567e-05, + "loss": 1.069, + "step": 555 + }, + { + "epoch": 1.421994884910486, + "grad_norm": 0.483272224628568, + "learning_rate": 7.201289036728331e-05, + "loss": 1.0536, + "step": 556 + }, + { + "epoch": 1.4245524296675192, + "grad_norm": 0.4770924027986762, + "learning_rate": 7.197000579128582e-05, + "loss": 1.0403, + "step": 557 + }, + { + "epoch": 1.4271099744245523, + "grad_norm": 0.448305597034261, + "learning_rate": 7.192701923630781e-05, + "loss": 1.0328, + "step": 558 + }, + { + "epoch": 1.4296675191815857, + "grad_norm": 0.495021183649144, + "learning_rate": 7.188393083946923e-05, + "loss": 1.0258, + "step": 559 + }, + { + "epoch": 1.432225063938619, + "grad_norm": 0.4705177505313109, + "learning_rate": 7.184074073821487e-05, + "loss": 1.0174, + "step": 560 + }, + { + "epoch": 1.434782608695652, + "grad_norm": 0.43574757389209073, + "learning_rate": 7.179744907031394e-05, + "loss": 1.0287, + "step": 561 + }, + { + "epoch": 1.4373401534526855, + "grad_norm": 0.3635773785423228, + "learning_rate": 7.175405597385964e-05, + "loss": 1.0412, + "step": 562 + }, + { + "epoch": 1.4398976982097187, + "grad_norm": 0.553156783226794, + "learning_rate": 7.171056158726869e-05, + "loss": 1.0688, + "step": 563 + }, + { + "epoch": 1.4424552429667519, + "grad_norm": 0.6900930611820969, + "learning_rate": 7.166696604928091e-05, + "loss": 1.0668, + "step": 564 + }, + { + "epoch": 1.445012787723785, + "grad_norm": 0.85914870330665, + "learning_rate": 7.162326949895882e-05, + "loss": 1.0752, + "step": 565 + }, + { + "epoch": 1.4475703324808185, + "grad_norm": 1.0719699916305456, + "learning_rate": 7.157947207568708e-05, + "loss": 1.038, + "step": 566 + }, + { + "epoch": 1.4501278772378516, + "grad_norm": 0.8876743742200847, + "learning_rate": 7.153557391917221e-05, + "loss": 1.0186, + "step": 567 + }, + { + "epoch": 1.452685421994885, + "grad_norm": 0.6329728177218358, + "learning_rate": 7.149157516944195e-05, + "loss": 1.0496, + "step": 568 + }, + { + "epoch": 1.4552429667519182, + "grad_norm": 0.4320509800831513, + "learning_rate": 7.144747596684498e-05, + "loss": 1.0401, + "step": 569 + }, + { + "epoch": 1.4578005115089514, + "grad_norm": 0.3695887057019756, + "learning_rate": 7.14032764520504e-05, + "loss": 1.0298, + "step": 570 + }, + { + "epoch": 1.4603580562659846, + "grad_norm": 0.4557157139603631, + "learning_rate": 7.135897676604728e-05, + "loss": 1.0504, + "step": 571 + }, + { + "epoch": 1.4629156010230178, + "grad_norm": 0.5697580796103671, + "learning_rate": 7.131457705014424e-05, + "loss": 1.0553, + "step": 572 + }, + { + "epoch": 1.4654731457800512, + "grad_norm": 0.562535535478939, + "learning_rate": 7.127007744596891e-05, + "loss": 1.0402, + "step": 573 + }, + { + "epoch": 1.4680306905370843, + "grad_norm": 0.5386864181747942, + "learning_rate": 7.122547809546766e-05, + "loss": 1.0266, + "step": 574 + }, + { + "epoch": 1.4705882352941178, + "grad_norm": 0.5604159637885537, + "learning_rate": 7.118077914090492e-05, + "loss": 1.0593, + "step": 575 + }, + { + "epoch": 1.473145780051151, + "grad_norm": 0.5345575941108697, + "learning_rate": 7.113598072486291e-05, + "loss": 1.0628, + "step": 576 + }, + { + "epoch": 1.4757033248081841, + "grad_norm": 0.42108117394775474, + "learning_rate": 7.10910829902411e-05, + "loss": 1.0545, + "step": 577 + }, + { + "epoch": 1.4782608695652173, + "grad_norm": 0.42679316673017587, + "learning_rate": 7.104608608025578e-05, + "loss": 1.0441, + "step": 578 + }, + { + "epoch": 1.4808184143222507, + "grad_norm": 0.4632742996319085, + "learning_rate": 7.100099013843956e-05, + "loss": 1.0233, + "step": 579 + }, + { + "epoch": 1.4833759590792839, + "grad_norm": 0.41059416177562724, + "learning_rate": 7.095579530864097e-05, + "loss": 1.0393, + "step": 580 + }, + { + "epoch": 1.485933503836317, + "grad_norm": 0.49854832145650435, + "learning_rate": 7.091050173502397e-05, + "loss": 1.0578, + "step": 581 + }, + { + "epoch": 1.4884910485933505, + "grad_norm": 0.6121395366510051, + "learning_rate": 7.08651095620675e-05, + "loss": 1.0463, + "step": 582 + }, + { + "epoch": 1.4910485933503836, + "grad_norm": 0.6431404220907431, + "learning_rate": 7.081961893456502e-05, + "loss": 1.0715, + "step": 583 + }, + { + "epoch": 1.4936061381074168, + "grad_norm": 0.8536051526625087, + "learning_rate": 7.077402999762401e-05, + "loss": 1.0774, + "step": 584 + }, + { + "epoch": 1.49616368286445, + "grad_norm": 0.9386047615779075, + "learning_rate": 7.07283428966656e-05, + "loss": 1.0167, + "step": 585 + }, + { + "epoch": 1.4987212276214834, + "grad_norm": 0.8795122294030159, + "learning_rate": 7.068255777742399e-05, + "loss": 1.0325, + "step": 586 + }, + { + "epoch": 1.5012787723785166, + "grad_norm": 0.6822762592618459, + "learning_rate": 7.063667478594609e-05, + "loss": 1.0408, + "step": 587 + }, + { + "epoch": 1.50383631713555, + "grad_norm": 0.5845658862279202, + "learning_rate": 7.059069406859096e-05, + "loss": 1.0671, + "step": 588 + }, + { + "epoch": 1.5063938618925832, + "grad_norm": 0.5244238010393297, + "learning_rate": 7.05446157720294e-05, + "loss": 1.0071, + "step": 589 + }, + { + "epoch": 1.5089514066496164, + "grad_norm": 0.5405468625462244, + "learning_rate": 7.049844004324352e-05, + "loss": 1.0364, + "step": 590 + }, + { + "epoch": 1.5115089514066495, + "grad_norm": 0.5321510176297898, + "learning_rate": 7.045216702952615e-05, + "loss": 1.0099, + "step": 591 + }, + { + "epoch": 1.5140664961636827, + "grad_norm": 0.6304136952687458, + "learning_rate": 7.040579687848051e-05, + "loss": 1.0629, + "step": 592 + }, + { + "epoch": 1.5166240409207161, + "grad_norm": 0.6788574527638351, + "learning_rate": 7.035932973801961e-05, + "loss": 0.9933, + "step": 593 + }, + { + "epoch": 1.5191815856777495, + "grad_norm": 0.5664383321834381, + "learning_rate": 7.031276575636588e-05, + "loss": 1.0581, + "step": 594 + }, + { + "epoch": 1.5217391304347827, + "grad_norm": 0.4764617199761682, + "learning_rate": 7.026610508205065e-05, + "loss": 1.03, + "step": 595 + }, + { + "epoch": 1.5242966751918159, + "grad_norm": 0.4115926551003442, + "learning_rate": 7.021934786391369e-05, + "loss": 1.0404, + "step": 596 + }, + { + "epoch": 1.526854219948849, + "grad_norm": 0.5008700498030078, + "learning_rate": 7.017249425110268e-05, + "loss": 1.0477, + "step": 597 + }, + { + "epoch": 1.5294117647058822, + "grad_norm": 0.5614325700629363, + "learning_rate": 7.012554439307288e-05, + "loss": 1.0496, + "step": 598 + }, + { + "epoch": 1.5319693094629157, + "grad_norm": 0.6424393529952258, + "learning_rate": 7.007849843958646e-05, + "loss": 1.027, + "step": 599 + }, + { + "epoch": 1.5345268542199488, + "grad_norm": 0.7601254190908516, + "learning_rate": 7.003135654071216e-05, + "loss": 1.0771, + "step": 600 + }, + { + "epoch": 1.5370843989769822, + "grad_norm": 0.6683027372767937, + "learning_rate": 6.998411884682476e-05, + "loss": 1.0245, + "step": 601 + }, + { + "epoch": 1.5396419437340154, + "grad_norm": 0.4292563464334402, + "learning_rate": 6.993678550860464e-05, + "loss": 1.0208, + "step": 602 + }, + { + "epoch": 1.5421994884910486, + "grad_norm": 0.37531967359267043, + "learning_rate": 6.988935667703722e-05, + "loss": 1.0341, + "step": 603 + }, + { + "epoch": 1.5447570332480818, + "grad_norm": 0.5020503230133389, + "learning_rate": 6.984183250341257e-05, + "loss": 1.0454, + "step": 604 + }, + { + "epoch": 1.547314578005115, + "grad_norm": 0.46091810824436635, + "learning_rate": 6.979421313932488e-05, + "loss": 1.0315, + "step": 605 + }, + { + "epoch": 1.5498721227621484, + "grad_norm": 0.4736527457196331, + "learning_rate": 6.974649873667194e-05, + "loss": 1.0086, + "step": 606 + }, + { + "epoch": 1.5524296675191815, + "grad_norm": 0.4999676394044973, + "learning_rate": 6.969868944765474e-05, + "loss": 1.0323, + "step": 607 + }, + { + "epoch": 1.554987212276215, + "grad_norm": 0.5697900185173456, + "learning_rate": 6.965078542477693e-05, + "loss": 1.0575, + "step": 608 + }, + { + "epoch": 1.5575447570332481, + "grad_norm": 0.6296338412635226, + "learning_rate": 6.960278682084436e-05, + "loss": 1.0537, + "step": 609 + }, + { + "epoch": 1.5601023017902813, + "grad_norm": 0.564117474441083, + "learning_rate": 6.955469378896456e-05, + "loss": 1.0231, + "step": 610 + }, + { + "epoch": 1.5626598465473145, + "grad_norm": 0.5743418608288569, + "learning_rate": 6.950650648254623e-05, + "loss": 1.0572, + "step": 611 + }, + { + "epoch": 1.5652173913043477, + "grad_norm": 0.48086239097871125, + "learning_rate": 6.945822505529887e-05, + "loss": 1.0466, + "step": 612 + }, + { + "epoch": 1.567774936061381, + "grad_norm": 0.3756345397327383, + "learning_rate": 6.940984966123217e-05, + "loss": 1.0366, + "step": 613 + }, + { + "epoch": 1.5703324808184145, + "grad_norm": 0.4472187990537581, + "learning_rate": 6.936138045465556e-05, + "loss": 1.0415, + "step": 614 + }, + { + "epoch": 1.5728900255754477, + "grad_norm": 0.4852565244536602, + "learning_rate": 6.93128175901777e-05, + "loss": 1.0591, + "step": 615 + }, + { + "epoch": 1.5754475703324808, + "grad_norm": 0.4059570141509962, + "learning_rate": 6.926416122270604e-05, + "loss": 1.0144, + "step": 616 + }, + { + "epoch": 1.578005115089514, + "grad_norm": 0.3681367090287205, + "learning_rate": 6.921541150744624e-05, + "loss": 1.0473, + "step": 617 + }, + { + "epoch": 1.5805626598465472, + "grad_norm": 0.3430601437581608, + "learning_rate": 6.916656859990179e-05, + "loss": 1.0409, + "step": 618 + }, + { + "epoch": 1.5831202046035806, + "grad_norm": 0.3419778416901883, + "learning_rate": 6.911763265587338e-05, + "loss": 1.0274, + "step": 619 + }, + { + "epoch": 1.5856777493606138, + "grad_norm": 0.3796918250923331, + "learning_rate": 6.90686038314585e-05, + "loss": 1.015, + "step": 620 + }, + { + "epoch": 1.5882352941176472, + "grad_norm": 0.49070616775356396, + "learning_rate": 6.901948228305094e-05, + "loss": 1.0446, + "step": 621 + }, + { + "epoch": 1.5907928388746804, + "grad_norm": 0.448562317355006, + "learning_rate": 6.89702681673402e-05, + "loss": 1.0583, + "step": 622 + }, + { + "epoch": 1.5933503836317136, + "grad_norm": 0.41319179373282156, + "learning_rate": 6.892096164131113e-05, + "loss": 1.031, + "step": 623 + }, + { + "epoch": 1.5959079283887467, + "grad_norm": 0.4625806799698519, + "learning_rate": 6.887156286224329e-05, + "loss": 1.0698, + "step": 624 + }, + { + "epoch": 1.59846547314578, + "grad_norm": 0.5803778153984633, + "learning_rate": 6.882207198771056e-05, + "loss": 1.0904, + "step": 625 + }, + { + "epoch": 1.6010230179028133, + "grad_norm": 0.6358263157928301, + "learning_rate": 6.877248917558055e-05, + "loss": 1.0178, + "step": 626 + }, + { + "epoch": 1.6035805626598465, + "grad_norm": 0.6408849160156058, + "learning_rate": 6.872281458401414e-05, + "loss": 1.028, + "step": 627 + }, + { + "epoch": 1.60613810741688, + "grad_norm": 0.6318504394160956, + "learning_rate": 6.867304837146501e-05, + "loss": 1.0669, + "step": 628 + }, + { + "epoch": 1.608695652173913, + "grad_norm": 0.6132529357479547, + "learning_rate": 6.862319069667904e-05, + "loss": 1.0215, + "step": 629 + }, + { + "epoch": 1.6112531969309463, + "grad_norm": 0.5149229199094381, + "learning_rate": 6.857324171869393e-05, + "loss": 1.0268, + "step": 630 + }, + { + "epoch": 1.6138107416879794, + "grad_norm": 0.3860072737272255, + "learning_rate": 6.852320159683853e-05, + "loss": 1.0342, + "step": 631 + }, + { + "epoch": 1.6163682864450126, + "grad_norm": 0.33909485027918096, + "learning_rate": 6.84730704907325e-05, + "loss": 1.0283, + "step": 632 + }, + { + "epoch": 1.618925831202046, + "grad_norm": 0.4191834960800194, + "learning_rate": 6.842284856028571e-05, + "loss": 1.0114, + "step": 633 + }, + { + "epoch": 1.6214833759590794, + "grad_norm": 0.5347405775672891, + "learning_rate": 6.83725359656977e-05, + "loss": 1.0589, + "step": 634 + }, + { + "epoch": 1.6240409207161126, + "grad_norm": 0.6355103271925497, + "learning_rate": 6.832213286745727e-05, + "loss": 1.0119, + "step": 635 + }, + { + "epoch": 1.6265984654731458, + "grad_norm": 0.7345332659340343, + "learning_rate": 6.82716394263419e-05, + "loss": 1.058, + "step": 636 + }, + { + "epoch": 1.629156010230179, + "grad_norm": 0.6965270461851645, + "learning_rate": 6.82210558034172e-05, + "loss": 1.0517, + "step": 637 + }, + { + "epoch": 1.6317135549872122, + "grad_norm": 0.582639080193431, + "learning_rate": 6.817038216003649e-05, + "loss": 1.0361, + "step": 638 + }, + { + "epoch": 1.6342710997442456, + "grad_norm": 0.49231339614396796, + "learning_rate": 6.811961865784026e-05, + "loss": 1.0691, + "step": 639 + }, + { + "epoch": 1.6368286445012787, + "grad_norm": 0.4590551715359045, + "learning_rate": 6.806876545875558e-05, + "loss": 1.004, + "step": 640 + }, + { + "epoch": 1.6393861892583121, + "grad_norm": 0.4934679803392225, + "learning_rate": 6.801782272499567e-05, + "loss": 1.0249, + "step": 641 + }, + { + "epoch": 1.6419437340153453, + "grad_norm": 0.4855827703329849, + "learning_rate": 6.796679061905936e-05, + "loss": 1.0275, + "step": 642 + }, + { + "epoch": 1.6445012787723785, + "grad_norm": 0.5663891915421606, + "learning_rate": 6.791566930373051e-05, + "loss": 1.0431, + "step": 643 + }, + { + "epoch": 1.6470588235294117, + "grad_norm": 0.5449325271192005, + "learning_rate": 6.786445894207763e-05, + "loss": 1.088, + "step": 644 + }, + { + "epoch": 1.6496163682864449, + "grad_norm": 0.4404635331433356, + "learning_rate": 6.781315969745319e-05, + "loss": 0.9978, + "step": 645 + }, + { + "epoch": 1.6521739130434783, + "grad_norm": 0.44006079141319454, + "learning_rate": 6.776177173349322e-05, + "loss": 1.0542, + "step": 646 + }, + { + "epoch": 1.6547314578005117, + "grad_norm": 0.43703882239072955, + "learning_rate": 6.771029521411674e-05, + "loss": 1.0127, + "step": 647 + }, + { + "epoch": 1.6572890025575449, + "grad_norm": 0.5232507591272976, + "learning_rate": 6.765873030352525e-05, + "loss": 1.0307, + "step": 648 + }, + { + "epoch": 1.659846547314578, + "grad_norm": 0.5907572767491934, + "learning_rate": 6.760707716620224e-05, + "loss": 1.0364, + "step": 649 + }, + { + "epoch": 1.6624040920716112, + "grad_norm": 0.5978645336642385, + "learning_rate": 6.755533596691255e-05, + "loss": 1.0197, + "step": 650 + }, + { + "epoch": 1.6649616368286444, + "grad_norm": 0.5721851679271525, + "learning_rate": 6.7503506870702e-05, + "loss": 1.0456, + "step": 651 + }, + { + "epoch": 1.6675191815856778, + "grad_norm": 0.526587743406477, + "learning_rate": 6.74515900428967e-05, + "loss": 1.0698, + "step": 652 + }, + { + "epoch": 1.670076726342711, + "grad_norm": 0.45666448861954223, + "learning_rate": 6.739958564910274e-05, + "loss": 1.0444, + "step": 653 + }, + { + "epoch": 1.6726342710997444, + "grad_norm": 0.33495402182030853, + "learning_rate": 6.734749385520539e-05, + "loss": 1.0494, + "step": 654 + }, + { + "epoch": 1.6751918158567776, + "grad_norm": 0.48053582090285774, + "learning_rate": 6.72953148273688e-05, + "loss": 1.0384, + "step": 655 + }, + { + "epoch": 1.6777493606138107, + "grad_norm": 0.6618699544458244, + "learning_rate": 6.724304873203538e-05, + "loss": 1.0252, + "step": 656 + }, + { + "epoch": 1.680306905370844, + "grad_norm": 0.7825491070996107, + "learning_rate": 6.71906957359252e-05, + "loss": 1.0279, + "step": 657 + }, + { + "epoch": 1.682864450127877, + "grad_norm": 0.7241348855375171, + "learning_rate": 6.71382560060356e-05, + "loss": 1.0492, + "step": 658 + }, + { + "epoch": 1.6854219948849105, + "grad_norm": 0.49266363223214565, + "learning_rate": 6.708572970964054e-05, + "loss": 1.0496, + "step": 659 + }, + { + "epoch": 1.6879795396419437, + "grad_norm": 0.3589459784191605, + "learning_rate": 6.703311701429019e-05, + "loss": 1.0657, + "step": 660 + }, + { + "epoch": 1.690537084398977, + "grad_norm": 0.5098638866479447, + "learning_rate": 6.698041808781021e-05, + "loss": 1.045, + "step": 661 + }, + { + "epoch": 1.6930946291560103, + "grad_norm": 0.5960811936716889, + "learning_rate": 6.69276330983014e-05, + "loss": 1.0372, + "step": 662 + }, + { + "epoch": 1.6956521739130435, + "grad_norm": 0.6441771191728006, + "learning_rate": 6.687476221413907e-05, + "loss": 1.0591, + "step": 663 + }, + { + "epoch": 1.6982097186700766, + "grad_norm": 0.7198375562017726, + "learning_rate": 6.682180560397248e-05, + "loss": 1.045, + "step": 664 + }, + { + "epoch": 1.7007672634271098, + "grad_norm": 0.7219444637646196, + "learning_rate": 6.676876343672441e-05, + "loss": 1.0487, + "step": 665 + }, + { + "epoch": 1.7033248081841432, + "grad_norm": 0.5227925388389713, + "learning_rate": 6.67156358815905e-05, + "loss": 1.0198, + "step": 666 + }, + { + "epoch": 1.7058823529411766, + "grad_norm": 0.465506768448557, + "learning_rate": 6.666242310803879e-05, + "loss": 1.0403, + "step": 667 + }, + { + "epoch": 1.7084398976982098, + "grad_norm": 0.4532826345672689, + "learning_rate": 6.660912528580912e-05, + "loss": 1.0297, + "step": 668 + }, + { + "epoch": 1.710997442455243, + "grad_norm": 0.45327509709330577, + "learning_rate": 6.655574258491267e-05, + "loss": 1.0554, + "step": 669 + }, + { + "epoch": 1.7135549872122762, + "grad_norm": 0.5641203058724333, + "learning_rate": 6.650227517563134e-05, + "loss": 1.0345, + "step": 670 + }, + { + "epoch": 1.7161125319693094, + "grad_norm": 0.5934846994434492, + "learning_rate": 6.644872322851722e-05, + "loss": 1.0512, + "step": 671 + }, + { + "epoch": 1.7186700767263428, + "grad_norm": 0.5243971153118981, + "learning_rate": 6.63950869143921e-05, + "loss": 1.0531, + "step": 672 + }, + { + "epoch": 1.721227621483376, + "grad_norm": 0.47550551440418576, + "learning_rate": 6.634136640434685e-05, + "loss": 1.0305, + "step": 673 + }, + { + "epoch": 1.7237851662404093, + "grad_norm": 0.4645714276383756, + "learning_rate": 6.628756186974093e-05, + "loss": 1.0734, + "step": 674 + }, + { + "epoch": 1.7263427109974425, + "grad_norm": 0.5045165698461913, + "learning_rate": 6.623367348220184e-05, + "loss": 1.044, + "step": 675 + }, + { + "epoch": 1.7289002557544757, + "grad_norm": 0.55969672109987, + "learning_rate": 6.61797014136245e-05, + "loss": 1.01, + "step": 676 + }, + { + "epoch": 1.7314578005115089, + "grad_norm": 0.5692361967437805, + "learning_rate": 6.612564583617085e-05, + "loss": 1.0271, + "step": 677 + }, + { + "epoch": 1.734015345268542, + "grad_norm": 0.4887973521051894, + "learning_rate": 6.607150692226912e-05, + "loss": 1.0289, + "step": 678 + }, + { + "epoch": 1.7365728900255755, + "grad_norm": 0.4760009295478612, + "learning_rate": 6.601728484461342e-05, + "loss": 1.0301, + "step": 679 + }, + { + "epoch": 1.7391304347826086, + "grad_norm": 0.48489041804369454, + "learning_rate": 6.59629797761631e-05, + "loss": 1.0309, + "step": 680 + }, + { + "epoch": 1.741687979539642, + "grad_norm": 0.38491676259975743, + "learning_rate": 6.59085918901423e-05, + "loss": 1.0192, + "step": 681 + }, + { + "epoch": 1.7442455242966752, + "grad_norm": 0.33184552078799817, + "learning_rate": 6.585412136003927e-05, + "loss": 1.0649, + "step": 682 + }, + { + "epoch": 1.7468030690537084, + "grad_norm": 0.29047480084769817, + "learning_rate": 6.579956835960592e-05, + "loss": 1.0573, + "step": 683 + }, + { + "epoch": 1.7493606138107416, + "grad_norm": 0.43218015798550996, + "learning_rate": 6.574493306285719e-05, + "loss": 1.0425, + "step": 684 + }, + { + "epoch": 1.7519181585677748, + "grad_norm": 0.39350320437790215, + "learning_rate": 6.569021564407057e-05, + "loss": 1.0349, + "step": 685 + }, + { + "epoch": 1.7544757033248082, + "grad_norm": 0.28656671953450064, + "learning_rate": 6.56354162777855e-05, + "loss": 1.0648, + "step": 686 + }, + { + "epoch": 1.7570332480818416, + "grad_norm": 0.44037380340676446, + "learning_rate": 6.558053513880278e-05, + "loss": 1.0399, + "step": 687 + }, + { + "epoch": 1.7595907928388748, + "grad_norm": 0.48759264629519933, + "learning_rate": 6.552557240218412e-05, + "loss": 1.0274, + "step": 688 + }, + { + "epoch": 1.762148337595908, + "grad_norm": 0.47090986696853737, + "learning_rate": 6.547052824325142e-05, + "loss": 1.0105, + "step": 689 + }, + { + "epoch": 1.7647058823529411, + "grad_norm": 0.47925643310416566, + "learning_rate": 6.541540283758642e-05, + "loss": 1.0363, + "step": 690 + }, + { + "epoch": 1.7672634271099743, + "grad_norm": 0.3174804641375395, + "learning_rate": 6.53601963610299e-05, + "loss": 1.0469, + "step": 691 + }, + { + "epoch": 1.7698209718670077, + "grad_norm": 0.34413914838661525, + "learning_rate": 6.530490898968135e-05, + "loss": 1.0405, + "step": 692 + }, + { + "epoch": 1.772378516624041, + "grad_norm": 0.44861200909145516, + "learning_rate": 6.524954089989822e-05, + "loss": 1.0554, + "step": 693 + }, + { + "epoch": 1.7749360613810743, + "grad_norm": 0.4663028003033403, + "learning_rate": 6.519409226829549e-05, + "loss": 1.0071, + "step": 694 + }, + { + "epoch": 1.7774936061381075, + "grad_norm": 0.4344266413537487, + "learning_rate": 6.513856327174501e-05, + "loss": 1.0219, + "step": 695 + }, + { + "epoch": 1.7800511508951407, + "grad_norm": 0.44323943000930544, + "learning_rate": 6.508295408737505e-05, + "loss": 1.0553, + "step": 696 + }, + { + "epoch": 1.7826086956521738, + "grad_norm": 0.4656359741585452, + "learning_rate": 6.50272648925696e-05, + "loss": 1.0331, + "step": 697 + }, + { + "epoch": 1.785166240409207, + "grad_norm": 0.4298294147098899, + "learning_rate": 6.49714958649679e-05, + "loss": 1.0474, + "step": 698 + }, + { + "epoch": 1.7877237851662404, + "grad_norm": 0.373839672554522, + "learning_rate": 6.491564718246381e-05, + "loss": 1.0372, + "step": 699 + }, + { + "epoch": 1.7902813299232738, + "grad_norm": 0.3633684264946787, + "learning_rate": 6.485971902320534e-05, + "loss": 1.0545, + "step": 700 + }, + { + "epoch": 1.792838874680307, + "grad_norm": 0.374369128686114, + "learning_rate": 6.480371156559395e-05, + "loss": 1.0207, + "step": 701 + }, + { + "epoch": 1.7953964194373402, + "grad_norm": 0.38960484038509224, + "learning_rate": 6.47476249882841e-05, + "loss": 1.0395, + "step": 702 + }, + { + "epoch": 1.7979539641943734, + "grad_norm": 0.41076940243434146, + "learning_rate": 6.469145947018261e-05, + "loss": 1.0423, + "step": 703 + }, + { + "epoch": 1.8005115089514065, + "grad_norm": 0.4037378076235256, + "learning_rate": 6.463521519044808e-05, + "loss": 1.0228, + "step": 704 + }, + { + "epoch": 1.80306905370844, + "grad_norm": 0.3340754340147289, + "learning_rate": 6.457889232849039e-05, + "loss": 1.0534, + "step": 705 + }, + { + "epoch": 1.8056265984654731, + "grad_norm": 0.2796291247118141, + "learning_rate": 6.452249106397004e-05, + "loss": 1.0043, + "step": 706 + }, + { + "epoch": 1.8081841432225065, + "grad_norm": 0.29928606836625965, + "learning_rate": 6.446601157679768e-05, + "loss": 1.0086, + "step": 707 + }, + { + "epoch": 1.8107416879795397, + "grad_norm": 0.32398878662153513, + "learning_rate": 6.440945404713344e-05, + "loss": 1.0346, + "step": 708 + }, + { + "epoch": 1.813299232736573, + "grad_norm": 0.3448476257962995, + "learning_rate": 6.435281865538638e-05, + "loss": 0.9851, + "step": 709 + }, + { + "epoch": 1.815856777493606, + "grad_norm": 0.3437352152692172, + "learning_rate": 6.429610558221395e-05, + "loss": 1.0297, + "step": 710 + }, + { + "epoch": 1.8184143222506393, + "grad_norm": 0.3285991493351655, + "learning_rate": 6.423931500852139e-05, + "loss": 1.0827, + "step": 711 + }, + { + "epoch": 1.8209718670076727, + "grad_norm": 0.2915476971624437, + "learning_rate": 6.418244711546113e-05, + "loss": 1.0311, + "step": 712 + }, + { + "epoch": 1.8235294117647058, + "grad_norm": 0.3038318588786024, + "learning_rate": 6.412550208443227e-05, + "loss": 1.0402, + "step": 713 + }, + { + "epoch": 1.8260869565217392, + "grad_norm": 0.312782111489241, + "learning_rate": 6.406848009707995e-05, + "loss": 1.0283, + "step": 714 + }, + { + "epoch": 1.8286445012787724, + "grad_norm": 0.2822241296244193, + "learning_rate": 6.401138133529478e-05, + "loss": 1.0125, + "step": 715 + }, + { + "epoch": 1.8312020460358056, + "grad_norm": 0.407257179302736, + "learning_rate": 6.395420598121229e-05, + "loss": 1.0293, + "step": 716 + }, + { + "epoch": 1.8337595907928388, + "grad_norm": 0.4664636629971284, + "learning_rate": 6.389695421721231e-05, + "loss": 0.9997, + "step": 717 + }, + { + "epoch": 1.836317135549872, + "grad_norm": 0.4720160006464846, + "learning_rate": 6.383962622591838e-05, + "loss": 1.0498, + "step": 718 + }, + { + "epoch": 1.8388746803069054, + "grad_norm": 0.45150296457653805, + "learning_rate": 6.378222219019726e-05, + "loss": 1.0491, + "step": 719 + }, + { + "epoch": 1.8414322250639388, + "grad_norm": 0.43237175184442106, + "learning_rate": 6.372474229315823e-05, + "loss": 1.0556, + "step": 720 + }, + { + "epoch": 1.843989769820972, + "grad_norm": 0.3555614837810917, + "learning_rate": 6.366718671815255e-05, + "loss": 1.0566, + "step": 721 + }, + { + "epoch": 1.8465473145780051, + "grad_norm": 0.4043617128578834, + "learning_rate": 6.360955564877289e-05, + "loss": 1.0128, + "step": 722 + }, + { + "epoch": 1.8491048593350383, + "grad_norm": 0.36791381894407194, + "learning_rate": 6.355184926885275e-05, + "loss": 1.0398, + "step": 723 + }, + { + "epoch": 1.8516624040920715, + "grad_norm": 0.41285311729480534, + "learning_rate": 6.349406776246587e-05, + "loss": 1.0303, + "step": 724 + }, + { + "epoch": 1.854219948849105, + "grad_norm": 0.47956295629579015, + "learning_rate": 6.343621131392558e-05, + "loss": 1.0613, + "step": 725 + }, + { + "epoch": 1.856777493606138, + "grad_norm": 0.35715402007757124, + "learning_rate": 6.337828010778427e-05, + "loss": 0.9846, + "step": 726 + }, + { + "epoch": 1.8593350383631715, + "grad_norm": 0.30788741050485563, + "learning_rate": 6.332027432883287e-05, + "loss": 1.01, + "step": 727 + }, + { + "epoch": 1.8618925831202047, + "grad_norm": 0.36516318391298225, + "learning_rate": 6.326219416210009e-05, + "loss": 1.015, + "step": 728 + }, + { + "epoch": 1.8644501278772379, + "grad_norm": 0.4172573664919271, + "learning_rate": 6.320403979285197e-05, + "loss": 1.0721, + "step": 729 + }, + { + "epoch": 1.867007672634271, + "grad_norm": 0.357721021972882, + "learning_rate": 6.314581140659124e-05, + "loss": 1.027, + "step": 730 + }, + { + "epoch": 1.8695652173913042, + "grad_norm": 0.3079770159596423, + "learning_rate": 6.308750918905674e-05, + "loss": 0.9984, + "step": 731 + }, + { + "epoch": 1.8721227621483376, + "grad_norm": 0.3516516243175642, + "learning_rate": 6.302913332622279e-05, + "loss": 1.0354, + "step": 732 + }, + { + "epoch": 1.8746803069053708, + "grad_norm": 0.5749314259929272, + "learning_rate": 6.297068400429866e-05, + "loss": 1.0471, + "step": 733 + }, + { + "epoch": 1.8772378516624042, + "grad_norm": 0.6805580946328815, + "learning_rate": 6.291216140972793e-05, + "loss": 1.0069, + "step": 734 + }, + { + "epoch": 1.8797953964194374, + "grad_norm": 0.7991805134495602, + "learning_rate": 6.28535657291879e-05, + "loss": 1.0364, + "step": 735 + }, + { + "epoch": 1.8823529411764706, + "grad_norm": 0.9755837794916613, + "learning_rate": 6.279489714958902e-05, + "loss": 1.0685, + "step": 736 + }, + { + "epoch": 1.8849104859335037, + "grad_norm": 1.0407798737927882, + "learning_rate": 6.273615585807423e-05, + "loss": 1.0413, + "step": 737 + }, + { + "epoch": 1.887468030690537, + "grad_norm": 0.6895467560527573, + "learning_rate": 6.267734204201848e-05, + "loss": 1.0468, + "step": 738 + }, + { + "epoch": 1.8900255754475703, + "grad_norm": 0.3662552887287085, + "learning_rate": 6.261845588902799e-05, + "loss": 1.0311, + "step": 739 + }, + { + "epoch": 1.8925831202046037, + "grad_norm": 0.5316160201986302, + "learning_rate": 6.255949758693977e-05, + "loss": 1.0032, + "step": 740 + }, + { + "epoch": 1.895140664961637, + "grad_norm": 0.6379429419889909, + "learning_rate": 6.250046732382099e-05, + "loss": 1.0131, + "step": 741 + }, + { + "epoch": 1.89769820971867, + "grad_norm": 0.5804871548161847, + "learning_rate": 6.244136528796827e-05, + "loss": 1.0191, + "step": 742 + }, + { + "epoch": 1.9002557544757033, + "grad_norm": 0.43651704808841196, + "learning_rate": 6.238219166790728e-05, + "loss": 1.0563, + "step": 743 + }, + { + "epoch": 1.9028132992327365, + "grad_norm": 0.4032559565923256, + "learning_rate": 6.232294665239199e-05, + "loss": 1.012, + "step": 744 + }, + { + "epoch": 1.9053708439897699, + "grad_norm": 0.4809968163709181, + "learning_rate": 6.22636304304041e-05, + "loss": 1.0094, + "step": 745 + }, + { + "epoch": 1.907928388746803, + "grad_norm": 0.6033128328270763, + "learning_rate": 6.220424319115243e-05, + "loss": 1.0333, + "step": 746 + }, + { + "epoch": 1.9104859335038364, + "grad_norm": 0.47176355179184654, + "learning_rate": 6.214478512407238e-05, + "loss": 1.025, + "step": 747 + }, + { + "epoch": 1.9130434782608696, + "grad_norm": 0.49617831540264234, + "learning_rate": 6.208525641882527e-05, + "loss": 1.0401, + "step": 748 + }, + { + "epoch": 1.9156010230179028, + "grad_norm": 0.4499870092411898, + "learning_rate": 6.20256572652977e-05, + "loss": 1.0626, + "step": 749 + }, + { + "epoch": 1.918158567774936, + "grad_norm": 0.3753432726335465, + "learning_rate": 6.196598785360104e-05, + "loss": 1.0152, + "step": 750 + }, + { + "epoch": 1.9207161125319692, + "grad_norm": 0.524059742940959, + "learning_rate": 6.190624837407075e-05, + "loss": 1.0711, + "step": 751 + }, + { + "epoch": 1.9232736572890026, + "grad_norm": 0.37911876119146704, + "learning_rate": 6.184643901726578e-05, + "loss": 1.0184, + "step": 752 + }, + { + "epoch": 1.9258312020460358, + "grad_norm": 0.5046505349197199, + "learning_rate": 6.178655997396801e-05, + "loss": 1.0341, + "step": 753 + }, + { + "epoch": 1.9283887468030692, + "grad_norm": 0.5897426305004047, + "learning_rate": 6.172661143518158e-05, + "loss": 1.0288, + "step": 754 + }, + { + "epoch": 1.9309462915601023, + "grad_norm": 0.5277155522871224, + "learning_rate": 6.16665935921323e-05, + "loss": 0.9989, + "step": 755 + }, + { + "epoch": 1.9335038363171355, + "grad_norm": 0.4455680271469787, + "learning_rate": 6.16065066362671e-05, + "loss": 1.0358, + "step": 756 + }, + { + "epoch": 1.9360613810741687, + "grad_norm": 0.37516675425752755, + "learning_rate": 6.154635075925333e-05, + "loss": 1.0328, + "step": 757 + }, + { + "epoch": 1.938618925831202, + "grad_norm": 0.31134972244002584, + "learning_rate": 6.148612615297819e-05, + "loss": 1.0143, + "step": 758 + }, + { + "epoch": 1.9411764705882353, + "grad_norm": 0.30116788123121785, + "learning_rate": 6.14258330095481e-05, + "loss": 1.0603, + "step": 759 + }, + { + "epoch": 1.9437340153452687, + "grad_norm": 0.33789017729539433, + "learning_rate": 6.136547152128817e-05, + "loss": 1.0642, + "step": 760 + }, + { + "epoch": 1.9462915601023019, + "grad_norm": 0.3723406702535513, + "learning_rate": 6.130504188074141e-05, + "loss": 1.0637, + "step": 761 + }, + { + "epoch": 1.948849104859335, + "grad_norm": 0.3074090795104566, + "learning_rate": 6.12445442806683e-05, + "loss": 1.0411, + "step": 762 + }, + { + "epoch": 1.9514066496163682, + "grad_norm": 0.29952175422386457, + "learning_rate": 6.118397891404612e-05, + "loss": 1.0311, + "step": 763 + }, + { + "epoch": 1.9539641943734014, + "grad_norm": 0.34505379977188627, + "learning_rate": 6.112334597406823e-05, + "loss": 1.0455, + "step": 764 + }, + { + "epoch": 1.9565217391304348, + "grad_norm": 0.3773271960134726, + "learning_rate": 6.106264565414363e-05, + "loss": 0.978, + "step": 765 + }, + { + "epoch": 1.959079283887468, + "grad_norm": 0.3577711544939887, + "learning_rate": 6.100187814789617e-05, + "loss": 1.0441, + "step": 766 + }, + { + "epoch": 1.9616368286445014, + "grad_norm": 0.3957345794906749, + "learning_rate": 6.094104364916405e-05, + "loss": 1.0126, + "step": 767 + }, + { + "epoch": 1.9641943734015346, + "grad_norm": 0.34753096665022937, + "learning_rate": 6.088014235199918e-05, + "loss": 1.0374, + "step": 768 + }, + { + "epoch": 1.9667519181585678, + "grad_norm": 0.31284752029157914, + "learning_rate": 6.081917445066651e-05, + "loss": 1.0469, + "step": 769 + }, + { + "epoch": 1.969309462915601, + "grad_norm": 0.42998501605296235, + "learning_rate": 6.07581401396435e-05, + "loss": 1.0298, + "step": 770 + }, + { + "epoch": 1.9718670076726341, + "grad_norm": 0.4320264632598454, + "learning_rate": 6.0697039613619376e-05, + "loss": 1.0209, + "step": 771 + }, + { + "epoch": 1.9744245524296675, + "grad_norm": 0.45573994194726763, + "learning_rate": 6.063587306749463e-05, + "loss": 1.0364, + "step": 772 + }, + { + "epoch": 1.976982097186701, + "grad_norm": 0.4479493314499644, + "learning_rate": 6.057464069638032e-05, + "loss": 1.0454, + "step": 773 + }, + { + "epoch": 1.979539641943734, + "grad_norm": 0.3489694286760604, + "learning_rate": 6.051334269559751e-05, + "loss": 1.0121, + "step": 774 + }, + { + "epoch": 1.9820971867007673, + "grad_norm": 0.2796056957493086, + "learning_rate": 6.045197926067657e-05, + "loss": 1.0229, + "step": 775 + }, + { + "epoch": 1.9846547314578005, + "grad_norm": 0.2899329917764029, + "learning_rate": 6.03905505873566e-05, + "loss": 1.04, + "step": 776 + }, + { + "epoch": 1.9872122762148337, + "grad_norm": 0.31537011362535095, + "learning_rate": 6.032905687158483e-05, + "loss": 1.0201, + "step": 777 + }, + { + "epoch": 1.989769820971867, + "grad_norm": 0.3419075043049475, + "learning_rate": 6.026749830951595e-05, + "loss": 1.0604, + "step": 778 + }, + { + "epoch": 1.9923273657289002, + "grad_norm": 0.3779426657587405, + "learning_rate": 6.0205875097511474e-05, + "loss": 1.042, + "step": 779 + }, + { + "epoch": 1.9948849104859336, + "grad_norm": 0.31387221943228044, + "learning_rate": 6.0144187432139185e-05, + "loss": 1.024, + "step": 780 + }, + { + "epoch": 1.9974424552429668, + "grad_norm": 0.3760802263824882, + "learning_rate": 6.008243551017241e-05, + "loss": 1.0303, + "step": 781 + }, + { + "epoch": 2.0, + "grad_norm": 0.48112235112341956, + "learning_rate": 6.002061952858949e-05, + "loss": 1.0061, + "step": 782 + }, + { + "epoch": 2.002557544757033, + "grad_norm": 0.4843347641579091, + "learning_rate": 5.995873968457307e-05, + "loss": 1.0042, + "step": 783 + }, + { + "epoch": 2.0051150895140664, + "grad_norm": 0.43280419351794497, + "learning_rate": 5.989679617550953e-05, + "loss": 0.9913, + "step": 784 + }, + { + "epoch": 2.0076726342710995, + "grad_norm": 0.33817333447982634, + "learning_rate": 5.983478919898832e-05, + "loss": 0.9942, + "step": 785 + }, + { + "epoch": 2.010230179028133, + "grad_norm": 0.46364451856775074, + "learning_rate": 5.9772718952801335e-05, + "loss": 1.077, + "step": 786 + }, + { + "epoch": 2.0127877237851663, + "grad_norm": 0.6456172502250567, + "learning_rate": 5.9710585634942304e-05, + "loss": 1.0141, + "step": 787 + }, + { + "epoch": 2.0153452685421995, + "grad_norm": 0.7087434988801616, + "learning_rate": 5.9648389443606136e-05, + "loss": 1.024, + "step": 788 + }, + { + "epoch": 2.0179028132992327, + "grad_norm": 0.6206523201919865, + "learning_rate": 5.958613057718828e-05, + "loss": 1.008, + "step": 789 + }, + { + "epoch": 2.020460358056266, + "grad_norm": 0.48723855559372675, + "learning_rate": 5.9523809234284155e-05, + "loss": 0.9975, + "step": 790 + }, + { + "epoch": 2.023017902813299, + "grad_norm": 0.3479935955066568, + "learning_rate": 5.9461425613688416e-05, + "loss": 0.995, + "step": 791 + }, + { + "epoch": 2.0255754475703327, + "grad_norm": 0.40412834234775397, + "learning_rate": 5.9398979914394416e-05, + "loss": 0.9994, + "step": 792 + }, + { + "epoch": 2.028132992327366, + "grad_norm": 0.4719705013531893, + "learning_rate": 5.933647233559349e-05, + "loss": 0.9927, + "step": 793 + }, + { + "epoch": 2.030690537084399, + "grad_norm": 0.39255651120492613, + "learning_rate": 5.9273903076674374e-05, + "loss": 0.9919, + "step": 794 + }, + { + "epoch": 2.0332480818414322, + "grad_norm": 0.3148451039709822, + "learning_rate": 5.921127233722258e-05, + "loss": 0.9973, + "step": 795 + }, + { + "epoch": 2.0358056265984654, + "grad_norm": 0.2961510487834574, + "learning_rate": 5.91485803170197e-05, + "loss": 1.0086, + "step": 796 + }, + { + "epoch": 2.0383631713554986, + "grad_norm": 0.318553289880585, + "learning_rate": 5.908582721604281e-05, + "loss": 1.0301, + "step": 797 + }, + { + "epoch": 2.040920716112532, + "grad_norm": 0.4333673733070479, + "learning_rate": 5.902301323446381e-05, + "loss": 1.0039, + "step": 798 + }, + { + "epoch": 2.0434782608695654, + "grad_norm": 0.5115830801046926, + "learning_rate": 5.896013857264884e-05, + "loss": 1.0043, + "step": 799 + }, + { + "epoch": 2.0460358056265986, + "grad_norm": 0.42276666456237705, + "learning_rate": 5.889720343115755e-05, + "loss": 1.0487, + "step": 800 + }, + { + "epoch": 2.0485933503836318, + "grad_norm": 0.33643099281393857, + "learning_rate": 5.8834208010742534e-05, + "loss": 1.0009, + "step": 801 + }, + { + "epoch": 2.051150895140665, + "grad_norm": 0.3504564236510021, + "learning_rate": 5.8771152512348694e-05, + "loss": 1.0075, + "step": 802 + }, + { + "epoch": 2.053708439897698, + "grad_norm": 0.5251143269792232, + "learning_rate": 5.870803713711251e-05, + "loss": 1.0153, + "step": 803 + }, + { + "epoch": 2.0562659846547313, + "grad_norm": 0.6328580629452392, + "learning_rate": 5.86448620863615e-05, + "loss": 1.0436, + "step": 804 + }, + { + "epoch": 2.0588235294117645, + "grad_norm": 0.6211486830413784, + "learning_rate": 5.8581627561613536e-05, + "loss": 0.9909, + "step": 805 + }, + { + "epoch": 2.061381074168798, + "grad_norm": 0.6716809355299344, + "learning_rate": 5.8518333764576176e-05, + "loss": 1.0042, + "step": 806 + }, + { + "epoch": 2.0639386189258313, + "grad_norm": 0.7599675406888277, + "learning_rate": 5.8454980897146094e-05, + "loss": 1.0454, + "step": 807 + }, + { + "epoch": 2.0664961636828645, + "grad_norm": 0.6684391810626017, + "learning_rate": 5.839156916140832e-05, + "loss": 1.0176, + "step": 808 + }, + { + "epoch": 2.0690537084398977, + "grad_norm": 0.41255587095603724, + "learning_rate": 5.832809875963573e-05, + "loss": 0.9945, + "step": 809 + }, + { + "epoch": 2.071611253196931, + "grad_norm": 0.4429337990160458, + "learning_rate": 5.826456989428832e-05, + "loss": 0.9978, + "step": 810 + }, + { + "epoch": 2.074168797953964, + "grad_norm": 0.5017870623183658, + "learning_rate": 5.820098276801252e-05, + "loss": 1.0212, + "step": 811 + }, + { + "epoch": 2.0767263427109977, + "grad_norm": 0.33363936794703103, + "learning_rate": 5.813733758364069e-05, + "loss": 0.9955, + "step": 812 + }, + { + "epoch": 2.079283887468031, + "grad_norm": 0.3184226812318926, + "learning_rate": 5.8073634544190315e-05, + "loss": 0.9961, + "step": 813 + }, + { + "epoch": 2.081841432225064, + "grad_norm": 0.38065698509998974, + "learning_rate": 5.800987385286346e-05, + "loss": 1.0569, + "step": 814 + }, + { + "epoch": 2.084398976982097, + "grad_norm": 0.34937290071950106, + "learning_rate": 5.79460557130461e-05, + "loss": 0.9854, + "step": 815 + }, + { + "epoch": 2.0869565217391304, + "grad_norm": 0.30190537548927504, + "learning_rate": 5.788218032830742e-05, + "loss": 0.9973, + "step": 816 + }, + { + "epoch": 2.0895140664961636, + "grad_norm": 0.3785973016391384, + "learning_rate": 5.781824790239924e-05, + "loss": 1.0084, + "step": 817 + }, + { + "epoch": 2.0920716112531967, + "grad_norm": 0.3560048267667031, + "learning_rate": 5.7754258639255334e-05, + "loss": 0.9879, + "step": 818 + }, + { + "epoch": 2.0946291560102304, + "grad_norm": 0.27318601205797294, + "learning_rate": 5.769021274299077e-05, + "loss": 0.9916, + "step": 819 + }, + { + "epoch": 2.0971867007672635, + "grad_norm": 0.27970625507687286, + "learning_rate": 5.7626110417901236e-05, + "loss": 0.9926, + "step": 820 + }, + { + "epoch": 2.0997442455242967, + "grad_norm": 0.34794308760067744, + "learning_rate": 5.7561951868462476e-05, + "loss": 1.0049, + "step": 821 + }, + { + "epoch": 2.10230179028133, + "grad_norm": 0.30884358202442075, + "learning_rate": 5.749773729932954e-05, + "loss": 0.99, + "step": 822 + }, + { + "epoch": 2.104859335038363, + "grad_norm": 0.2890559493826192, + "learning_rate": 5.743346691533617e-05, + "loss": 1.0209, + "step": 823 + }, + { + "epoch": 2.1074168797953963, + "grad_norm": 0.3795109731603794, + "learning_rate": 5.73691409214942e-05, + "loss": 1.0133, + "step": 824 + }, + { + "epoch": 2.10997442455243, + "grad_norm": 0.3707355948082582, + "learning_rate": 5.730475952299275e-05, + "loss": 1.0144, + "step": 825 + }, + { + "epoch": 2.112531969309463, + "grad_norm": 0.30752005319876746, + "learning_rate": 5.724032292519777e-05, + "loss": 1.0104, + "step": 826 + }, + { + "epoch": 2.1150895140664963, + "grad_norm": 0.38345043941403895, + "learning_rate": 5.717583133365125e-05, + "loss": 1.0029, + "step": 827 + }, + { + "epoch": 2.1176470588235294, + "grad_norm": 0.39070708528958853, + "learning_rate": 5.7111284954070576e-05, + "loss": 1.0131, + "step": 828 + }, + { + "epoch": 2.1202046035805626, + "grad_norm": 0.45836567684648993, + "learning_rate": 5.704668399234795e-05, + "loss": 1.017, + "step": 829 + }, + { + "epoch": 2.122762148337596, + "grad_norm": 0.5856980096882749, + "learning_rate": 5.6982028654549624e-05, + "loss": 1.0435, + "step": 830 + }, + { + "epoch": 2.125319693094629, + "grad_norm": 0.419392306323121, + "learning_rate": 5.6917319146915354e-05, + "loss": 1.0343, + "step": 831 + }, + { + "epoch": 2.1278772378516626, + "grad_norm": 0.30877657000728803, + "learning_rate": 5.685255567585766e-05, + "loss": 1.0317, + "step": 832 + }, + { + "epoch": 2.130434782608696, + "grad_norm": 0.4491035337879607, + "learning_rate": 5.678773844796118e-05, + "loss": 1.0305, + "step": 833 + }, + { + "epoch": 2.132992327365729, + "grad_norm": 0.5632576746306452, + "learning_rate": 5.6722867669982084e-05, + "loss": 0.9911, + "step": 834 + }, + { + "epoch": 2.135549872122762, + "grad_norm": 0.6141512327447454, + "learning_rate": 5.66579435488473e-05, + "loss": 1.0223, + "step": 835 + }, + { + "epoch": 2.1381074168797953, + "grad_norm": 0.4763196937292799, + "learning_rate": 5.659296629165393e-05, + "loss": 1.0383, + "step": 836 + }, + { + "epoch": 2.1406649616368285, + "grad_norm": 0.30729108731019755, + "learning_rate": 5.652793610566859e-05, + "loss": 0.9768, + "step": 837 + }, + { + "epoch": 2.1432225063938617, + "grad_norm": 0.3227450801791841, + "learning_rate": 5.646285319832671e-05, + "loss": 1.0356, + "step": 838 + }, + { + "epoch": 2.1457800511508953, + "grad_norm": 0.38794788918322937, + "learning_rate": 5.639771777723189e-05, + "loss": 1.0125, + "step": 839 + }, + { + "epoch": 2.1483375959079285, + "grad_norm": 0.4348976014763873, + "learning_rate": 5.633253005015526e-05, + "loss": 1.0369, + "step": 840 + }, + { + "epoch": 2.1508951406649617, + "grad_norm": 0.41931164227914425, + "learning_rate": 5.6267290225034775e-05, + "loss": 1.0142, + "step": 841 + }, + { + "epoch": 2.153452685421995, + "grad_norm": 0.27541715665063, + "learning_rate": 5.620199850997459e-05, + "loss": 1.0262, + "step": 842 + }, + { + "epoch": 2.156010230179028, + "grad_norm": 0.36736598694146594, + "learning_rate": 5.6136655113244386e-05, + "loss": 1.0015, + "step": 843 + }, + { + "epoch": 2.1585677749360612, + "grad_norm": 0.43457711232669183, + "learning_rate": 5.607126024327867e-05, + "loss": 1.0493, + "step": 844 + }, + { + "epoch": 2.1611253196930944, + "grad_norm": 0.3685703029484094, + "learning_rate": 5.600581410867614e-05, + "loss": 1.0086, + "step": 845 + }, + { + "epoch": 2.163682864450128, + "grad_norm": 0.36431748824284954, + "learning_rate": 5.5940316918199084e-05, + "loss": 1.0048, + "step": 846 + }, + { + "epoch": 2.166240409207161, + "grad_norm": 0.35086550909201497, + "learning_rate": 5.587476888077258e-05, + "loss": 0.9661, + "step": 847 + }, + { + "epoch": 2.1687979539641944, + "grad_norm": 0.47340488623900007, + "learning_rate": 5.580917020548391e-05, + "loss": 1.0578, + "step": 848 + }, + { + "epoch": 2.1713554987212276, + "grad_norm": 0.41591848145887617, + "learning_rate": 5.5743521101581934e-05, + "loss": 1.0163, + "step": 849 + }, + { + "epoch": 2.1739130434782608, + "grad_norm": 0.2646123185537355, + "learning_rate": 5.56778217784763e-05, + "loss": 1.0171, + "step": 850 + }, + { + "epoch": 2.176470588235294, + "grad_norm": 0.3511706100428371, + "learning_rate": 5.561207244573688e-05, + "loss": 1.0219, + "step": 851 + }, + { + "epoch": 2.1790281329923276, + "grad_norm": 0.38155483809818286, + "learning_rate": 5.554627331309308e-05, + "loss": 1.0355, + "step": 852 + }, + { + "epoch": 2.1815856777493607, + "grad_norm": 0.44428569276837054, + "learning_rate": 5.548042459043313e-05, + "loss": 0.9744, + "step": 853 + }, + { + "epoch": 2.184143222506394, + "grad_norm": 0.3228503410573803, + "learning_rate": 5.541452648780348e-05, + "loss": 0.9585, + "step": 854 + }, + { + "epoch": 2.186700767263427, + "grad_norm": 0.3380049356254192, + "learning_rate": 5.5348579215408064e-05, + "loss": 1.0107, + "step": 855 + }, + { + "epoch": 2.1892583120204603, + "grad_norm": 0.29475974621755224, + "learning_rate": 5.528258298360767e-05, + "loss": 1.029, + "step": 856 + }, + { + "epoch": 2.1918158567774935, + "grad_norm": 0.3037628643580443, + "learning_rate": 5.521653800291927e-05, + "loss": 1.0448, + "step": 857 + }, + { + "epoch": 2.1943734015345266, + "grad_norm": 0.29688565337946315, + "learning_rate": 5.51504444840153e-05, + "loss": 1.0376, + "step": 858 + }, + { + "epoch": 2.1969309462915603, + "grad_norm": 0.323945767250843, + "learning_rate": 5.508430263772308e-05, + "loss": 1.0336, + "step": 859 + }, + { + "epoch": 2.1994884910485935, + "grad_norm": 0.3284522396502332, + "learning_rate": 5.501811267502403e-05, + "loss": 1.0302, + "step": 860 + }, + { + "epoch": 2.2020460358056266, + "grad_norm": 0.3592349215963694, + "learning_rate": 5.49518748070531e-05, + "loss": 1.023, + "step": 861 + }, + { + "epoch": 2.20460358056266, + "grad_norm": 0.3423858035574432, + "learning_rate": 5.488558924509802e-05, + "loss": 0.9957, + "step": 862 + }, + { + "epoch": 2.207161125319693, + "grad_norm": 0.2626938661984674, + "learning_rate": 5.481925620059866e-05, + "loss": 1.0192, + "step": 863 + }, + { + "epoch": 2.209718670076726, + "grad_norm": 0.31225095162840705, + "learning_rate": 5.4752875885146354e-05, + "loss": 1.0212, + "step": 864 + }, + { + "epoch": 2.21227621483376, + "grad_norm": 0.39268950929217, + "learning_rate": 5.468644851048324e-05, + "loss": 0.9785, + "step": 865 + }, + { + "epoch": 2.214833759590793, + "grad_norm": 0.3479708324986449, + "learning_rate": 5.461997428850154e-05, + "loss": 1.0092, + "step": 866 + }, + { + "epoch": 2.217391304347826, + "grad_norm": 0.30125519235337805, + "learning_rate": 5.4553453431242936e-05, + "loss": 1.0514, + "step": 867 + }, + { + "epoch": 2.2199488491048593, + "grad_norm": 0.3569041941007821, + "learning_rate": 5.448688615089784e-05, + "loss": 0.9922, + "step": 868 + }, + { + "epoch": 2.2225063938618925, + "grad_norm": 0.3578777931289321, + "learning_rate": 5.442027265980478e-05, + "loss": 1.0292, + "step": 869 + }, + { + "epoch": 2.2250639386189257, + "grad_norm": 0.43031174528074956, + "learning_rate": 5.4353613170449636e-05, + "loss": 1.0144, + "step": 870 + }, + { + "epoch": 2.227621483375959, + "grad_norm": 0.3034147546996222, + "learning_rate": 5.42869078954651e-05, + "loss": 0.995, + "step": 871 + }, + { + "epoch": 2.2301790281329925, + "grad_norm": 0.3265430475750126, + "learning_rate": 5.422015704762981e-05, + "loss": 1.016, + "step": 872 + }, + { + "epoch": 2.2327365728900257, + "grad_norm": 0.4632802224731344, + "learning_rate": 5.415336083986786e-05, + "loss": 1.0321, + "step": 873 + }, + { + "epoch": 2.235294117647059, + "grad_norm": 0.42905428916009136, + "learning_rate": 5.408651948524799e-05, + "loss": 1.0206, + "step": 874 + }, + { + "epoch": 2.237851662404092, + "grad_norm": 0.35791980721264094, + "learning_rate": 5.401963319698295e-05, + "loss": 0.9878, + "step": 875 + }, + { + "epoch": 2.2404092071611252, + "grad_norm": 0.3518593213090125, + "learning_rate": 5.395270218842886e-05, + "loss": 1.0073, + "step": 876 + }, + { + "epoch": 2.2429667519181584, + "grad_norm": 0.274369572647955, + "learning_rate": 5.388572667308442e-05, + "loss": 0.9968, + "step": 877 + }, + { + "epoch": 2.2455242966751916, + "grad_norm": 0.4023661915440134, + "learning_rate": 5.381870686459038e-05, + "loss": 1.0038, + "step": 878 + }, + { + "epoch": 2.2480818414322252, + "grad_norm": 0.4463692936387662, + "learning_rate": 5.375164297672872e-05, + "loss": 1.0037, + "step": 879 + }, + { + "epoch": 2.2506393861892584, + "grad_norm": 0.3917913866024726, + "learning_rate": 5.368453522342204e-05, + "loss": 0.9793, + "step": 880 + }, + { + "epoch": 2.2531969309462916, + "grad_norm": 0.2696435570953871, + "learning_rate": 5.361738381873289e-05, + "loss": 0.9997, + "step": 881 + }, + { + "epoch": 2.2557544757033248, + "grad_norm": 0.2954089772852502, + "learning_rate": 5.355018897686302e-05, + "loss": 0.9663, + "step": 882 + }, + { + "epoch": 2.258312020460358, + "grad_norm": 0.30149260615056417, + "learning_rate": 5.348295091215276e-05, + "loss": 1.0061, + "step": 883 + }, + { + "epoch": 2.260869565217391, + "grad_norm": 0.2802190777854028, + "learning_rate": 5.341566983908031e-05, + "loss": 1.0006, + "step": 884 + }, + { + "epoch": 2.2634271099744243, + "grad_norm": 0.2655762208217684, + "learning_rate": 5.3348345972261065e-05, + "loss": 1.0199, + "step": 885 + }, + { + "epoch": 2.265984654731458, + "grad_norm": 0.2351284398309919, + "learning_rate": 5.328097952644691e-05, + "loss": 1.028, + "step": 886 + }, + { + "epoch": 2.268542199488491, + "grad_norm": 0.2557435656189813, + "learning_rate": 5.321357071652557e-05, + "loss": 0.9823, + "step": 887 + }, + { + "epoch": 2.2710997442455243, + "grad_norm": 0.267763753999149, + "learning_rate": 5.3146119757519875e-05, + "loss": 0.9887, + "step": 888 + }, + { + "epoch": 2.2736572890025575, + "grad_norm": 0.27860205774229696, + "learning_rate": 5.3078626864587144e-05, + "loss": 1.0133, + "step": 889 + }, + { + "epoch": 2.2762148337595907, + "grad_norm": 0.2987558728085059, + "learning_rate": 5.3011092253018434e-05, + "loss": 1.0467, + "step": 890 + }, + { + "epoch": 2.2787723785166243, + "grad_norm": 0.269660583834978, + "learning_rate": 5.294351613823787e-05, + "loss": 1.0236, + "step": 891 + }, + { + "epoch": 2.2813299232736575, + "grad_norm": 0.2682105777894826, + "learning_rate": 5.287589873580196e-05, + "loss": 1.0073, + "step": 892 + }, + { + "epoch": 2.2838874680306906, + "grad_norm": 0.26935713463820965, + "learning_rate": 5.280824026139896e-05, + "loss": 1.0137, + "step": 893 + }, + { + "epoch": 2.286445012787724, + "grad_norm": 0.2632358936931599, + "learning_rate": 5.2740540930848094e-05, + "loss": 1.0335, + "step": 894 + }, + { + "epoch": 2.289002557544757, + "grad_norm": 0.26738226375067914, + "learning_rate": 5.267280096009892e-05, + "loss": 1.0228, + "step": 895 + }, + { + "epoch": 2.29156010230179, + "grad_norm": 0.2780005314658838, + "learning_rate": 5.260502056523063e-05, + "loss": 1.0098, + "step": 896 + }, + { + "epoch": 2.2941176470588234, + "grad_norm": 0.2741454954614826, + "learning_rate": 5.253719996245137e-05, + "loss": 1.0088, + "step": 897 + }, + { + "epoch": 2.296675191815857, + "grad_norm": 0.3539765842461498, + "learning_rate": 5.246933936809754e-05, + "loss": 0.9849, + "step": 898 + }, + { + "epoch": 2.29923273657289, + "grad_norm": 0.3350858712179409, + "learning_rate": 5.240143899863312e-05, + "loss": 1.0194, + "step": 899 + }, + { + "epoch": 2.3017902813299234, + "grad_norm": 0.28258375421118737, + "learning_rate": 5.233349907064892e-05, + "loss": 1.0043, + "step": 900 + }, + { + "epoch": 2.3043478260869565, + "grad_norm": 0.2746324008943725, + "learning_rate": 5.2265519800861996e-05, + "loss": 0.9957, + "step": 901 + }, + { + "epoch": 2.3069053708439897, + "grad_norm": 0.284357215500465, + "learning_rate": 5.219750140611485e-05, + "loss": 1.0013, + "step": 902 + }, + { + "epoch": 2.309462915601023, + "grad_norm": 0.3486207256426231, + "learning_rate": 5.212944410337482e-05, + "loss": 1.0235, + "step": 903 + }, + { + "epoch": 2.312020460358056, + "grad_norm": 0.29063234065847066, + "learning_rate": 5.206134810973332e-05, + "loss": 0.9498, + "step": 904 + }, + { + "epoch": 2.3145780051150897, + "grad_norm": 0.2783227068727512, + "learning_rate": 5.19932136424052e-05, + "loss": 0.9856, + "step": 905 + }, + { + "epoch": 2.317135549872123, + "grad_norm": 0.255357028205985, + "learning_rate": 5.192504091872807e-05, + "loss": 0.9942, + "step": 906 + }, + { + "epoch": 2.319693094629156, + "grad_norm": 0.28410051410619885, + "learning_rate": 5.185683015616149e-05, + "loss": 0.9985, + "step": 907 + }, + { + "epoch": 2.3222506393861893, + "grad_norm": 0.27168522625319097, + "learning_rate": 5.178858157228643e-05, + "loss": 1.0386, + "step": 908 + }, + { + "epoch": 2.3248081841432224, + "grad_norm": 0.2974602944219359, + "learning_rate": 5.1720295384804466e-05, + "loss": 0.9756, + "step": 909 + }, + { + "epoch": 2.3273657289002556, + "grad_norm": 0.4194104880758461, + "learning_rate": 5.165197181153713e-05, + "loss": 1.0509, + "step": 910 + }, + { + "epoch": 2.329923273657289, + "grad_norm": 0.34398031740039026, + "learning_rate": 5.158361107042524e-05, + "loss": 1.0279, + "step": 911 + }, + { + "epoch": 2.3324808184143224, + "grad_norm": 0.4011339549083505, + "learning_rate": 5.1515213379528115e-05, + "loss": 0.9884, + "step": 912 + }, + { + "epoch": 2.3350383631713556, + "grad_norm": 0.3280230569283143, + "learning_rate": 5.1446778957022984e-05, + "loss": 1.0194, + "step": 913 + }, + { + "epoch": 2.337595907928389, + "grad_norm": 0.2805049932148967, + "learning_rate": 5.137830802120422e-05, + "loss": 1.0192, + "step": 914 + }, + { + "epoch": 2.340153452685422, + "grad_norm": 0.28082514853803286, + "learning_rate": 5.130980079048269e-05, + "loss": 1.0164, + "step": 915 + }, + { + "epoch": 2.342710997442455, + "grad_norm": 0.3252478527316345, + "learning_rate": 5.124125748338501e-05, + "loss": 1.0374, + "step": 916 + }, + { + "epoch": 2.3452685421994883, + "grad_norm": 0.36012680682896564, + "learning_rate": 5.117267831855289e-05, + "loss": 0.9791, + "step": 917 + }, + { + "epoch": 2.3478260869565215, + "grad_norm": 0.3562744865725541, + "learning_rate": 5.110406351474241e-05, + "loss": 1.0071, + "step": 918 + }, + { + "epoch": 2.350383631713555, + "grad_norm": 0.32552496986950086, + "learning_rate": 5.103541329082334e-05, + "loss": 0.9929, + "step": 919 + }, + { + "epoch": 2.3529411764705883, + "grad_norm": 0.28878679693944026, + "learning_rate": 5.096672786577843e-05, + "loss": 0.9936, + "step": 920 + }, + { + "epoch": 2.3554987212276215, + "grad_norm": 0.33555234908502024, + "learning_rate": 5.089800745870273e-05, + "loss": 0.9976, + "step": 921 + }, + { + "epoch": 2.3580562659846547, + "grad_norm": 0.2852116174307783, + "learning_rate": 5.082925228880284e-05, + "loss": 0.9972, + "step": 922 + }, + { + "epoch": 2.360613810741688, + "grad_norm": 0.2858096756505163, + "learning_rate": 5.076046257539632e-05, + "loss": 0.9994, + "step": 923 + }, + { + "epoch": 2.363171355498721, + "grad_norm": 0.27416120487035023, + "learning_rate": 5.0691638537910806e-05, + "loss": 1.0077, + "step": 924 + }, + { + "epoch": 2.3657289002557547, + "grad_norm": 0.2524911331080638, + "learning_rate": 5.0622780395883526e-05, + "loss": 0.9952, + "step": 925 + }, + { + "epoch": 2.368286445012788, + "grad_norm": 0.43007425455994347, + "learning_rate": 5.055388836896045e-05, + "loss": 0.9719, + "step": 926 + }, + { + "epoch": 2.370843989769821, + "grad_norm": 0.3162275053374832, + "learning_rate": 5.0484962676895624e-05, + "loss": 1.0277, + "step": 927 + }, + { + "epoch": 2.373401534526854, + "grad_norm": 0.3630514647001877, + "learning_rate": 5.041600353955054e-05, + "loss": 1.0331, + "step": 928 + }, + { + "epoch": 2.3759590792838874, + "grad_norm": 0.288338048580676, + "learning_rate": 5.0347011176893284e-05, + "loss": 1.003, + "step": 929 + }, + { + "epoch": 2.3785166240409206, + "grad_norm": 0.3257712128860222, + "learning_rate": 5.0277985808998e-05, + "loss": 1.0268, + "step": 930 + }, + { + "epoch": 2.381074168797954, + "grad_norm": 0.2742545893923996, + "learning_rate": 5.020892765604407e-05, + "loss": 0.9939, + "step": 931 + }, + { + "epoch": 2.3836317135549874, + "grad_norm": 0.34791994805077125, + "learning_rate": 5.013983693831549e-05, + "loss": 1.0188, + "step": 932 + }, + { + "epoch": 2.3861892583120206, + "grad_norm": 0.29742817188293785, + "learning_rate": 5.007071387620009e-05, + "loss": 1.0014, + "step": 933 + }, + { + "epoch": 2.3887468030690537, + "grad_norm": 0.28763416890386856, + "learning_rate": 5.000155869018893e-05, + "loss": 0.9896, + "step": 934 + }, + { + "epoch": 2.391304347826087, + "grad_norm": 0.27895542194014733, + "learning_rate": 4.993237160087549e-05, + "loss": 0.9879, + "step": 935 + }, + { + "epoch": 2.39386189258312, + "grad_norm": 0.2729422261234032, + "learning_rate": 4.986315282895502e-05, + "loss": 1.0062, + "step": 936 + }, + { + "epoch": 2.3964194373401533, + "grad_norm": 0.27170320699542444, + "learning_rate": 4.979390259522386e-05, + "loss": 1.0087, + "step": 937 + }, + { + "epoch": 2.398976982097187, + "grad_norm": 0.30230312142914184, + "learning_rate": 4.972462112057868e-05, + "loss": 1.0173, + "step": 938 + }, + { + "epoch": 2.40153452685422, + "grad_norm": 0.3184224115524826, + "learning_rate": 4.9655308626015835e-05, + "loss": 1.0014, + "step": 939 + }, + { + "epoch": 2.4040920716112533, + "grad_norm": 0.3172883163212683, + "learning_rate": 4.958596533263061e-05, + "loss": 1.0102, + "step": 940 + }, + { + "epoch": 2.4066496163682864, + "grad_norm": 0.303355088910324, + "learning_rate": 4.9516591461616505e-05, + "loss": 0.999, + "step": 941 + }, + { + "epoch": 2.4092071611253196, + "grad_norm": 0.24585768714005402, + "learning_rate": 4.9447187234264615e-05, + "loss": 1.0091, + "step": 942 + }, + { + "epoch": 2.411764705882353, + "grad_norm": 0.3093500884677307, + "learning_rate": 4.937775287196282e-05, + "loss": 1.0122, + "step": 943 + }, + { + "epoch": 2.414322250639386, + "grad_norm": 0.41304662460185293, + "learning_rate": 4.930828859619515e-05, + "loss": 1.0119, + "step": 944 + }, + { + "epoch": 2.4168797953964196, + "grad_norm": 0.4451137669910999, + "learning_rate": 4.923879462854105e-05, + "loss": 0.9659, + "step": 945 + }, + { + "epoch": 2.419437340153453, + "grad_norm": 0.32994372437876596, + "learning_rate": 4.9169271190674656e-05, + "loss": 0.9856, + "step": 946 + }, + { + "epoch": 2.421994884910486, + "grad_norm": 0.22818907249672446, + "learning_rate": 4.909971850436414e-05, + "loss": 1.0361, + "step": 947 + }, + { + "epoch": 2.424552429667519, + "grad_norm": 0.33962850938503897, + "learning_rate": 4.9030136791470976e-05, + "loss": 0.9944, + "step": 948 + }, + { + "epoch": 2.4271099744245523, + "grad_norm": 0.40954174881553046, + "learning_rate": 4.896052627394917e-05, + "loss": 1.0093, + "step": 949 + }, + { + "epoch": 2.4296675191815855, + "grad_norm": 0.4257393909425225, + "learning_rate": 4.889088717384468e-05, + "loss": 0.9912, + "step": 950 + }, + { + "epoch": 2.4322250639386187, + "grad_norm": 0.30412969316946686, + "learning_rate": 4.8821219713294595e-05, + "loss": 1.0071, + "step": 951 + }, + { + "epoch": 2.4347826086956523, + "grad_norm": 0.35293329411505, + "learning_rate": 4.8751524114526486e-05, + "loss": 0.9719, + "step": 952 + }, + { + "epoch": 2.4373401534526855, + "grad_norm": 0.4117484756819428, + "learning_rate": 4.8681800599857687e-05, + "loss": 1.0234, + "step": 953 + }, + { + "epoch": 2.4398976982097187, + "grad_norm": 0.437336308379765, + "learning_rate": 4.861204939169454e-05, + "loss": 1.0258, + "step": 954 + }, + { + "epoch": 2.442455242966752, + "grad_norm": 0.31336295315356344, + "learning_rate": 4.854227071253177e-05, + "loss": 1.0007, + "step": 955 + }, + { + "epoch": 2.445012787723785, + "grad_norm": 0.3033170124494707, + "learning_rate": 4.847246478495172e-05, + "loss": 1.0077, + "step": 956 + }, + { + "epoch": 2.4475703324808182, + "grad_norm": 0.37625594485986924, + "learning_rate": 4.840263183162363e-05, + "loss": 0.9793, + "step": 957 + }, + { + "epoch": 2.4501278772378514, + "grad_norm": 0.4110181159588999, + "learning_rate": 4.833277207530298e-05, + "loss": 0.9742, + "step": 958 + }, + { + "epoch": 2.452685421994885, + "grad_norm": 0.3664571890685922, + "learning_rate": 4.826288573883071e-05, + "loss": 0.9968, + "step": 959 + }, + { + "epoch": 2.455242966751918, + "grad_norm": 0.27080091028479425, + "learning_rate": 4.8192973045132575e-05, + "loss": 0.9728, + "step": 960 + }, + { + "epoch": 2.4578005115089514, + "grad_norm": 0.38098722077365216, + "learning_rate": 4.812303421721837e-05, + "loss": 1.0399, + "step": 961 + }, + { + "epoch": 2.4603580562659846, + "grad_norm": 0.46341954095412086, + "learning_rate": 4.8053069478181315e-05, + "loss": 1.0156, + "step": 962 + }, + { + "epoch": 2.4629156010230178, + "grad_norm": 0.32489657486295764, + "learning_rate": 4.798307905119723e-05, + "loss": 1.023, + "step": 963 + }, + { + "epoch": 2.4654731457800514, + "grad_norm": 0.2790931892033216, + "learning_rate": 4.7913063159523896e-05, + "loss": 1.0285, + "step": 964 + }, + { + "epoch": 2.4680306905370846, + "grad_norm": 0.28666393779100274, + "learning_rate": 4.784302202650031e-05, + "loss": 0.9802, + "step": 965 + }, + { + "epoch": 2.4705882352941178, + "grad_norm": 0.2756414211620851, + "learning_rate": 4.777295587554597e-05, + "loss": 0.9958, + "step": 966 + }, + { + "epoch": 2.473145780051151, + "grad_norm": 0.254767874308196, + "learning_rate": 4.770286493016025e-05, + "loss": 1.0056, + "step": 967 + }, + { + "epoch": 2.475703324808184, + "grad_norm": 0.27565052471697443, + "learning_rate": 4.7632749413921524e-05, + "loss": 1.0109, + "step": 968 + }, + { + "epoch": 2.4782608695652173, + "grad_norm": 2.156916489907195, + "learning_rate": 4.756260955048658e-05, + "loss": 1.0193, + "step": 969 + }, + { + "epoch": 2.4808184143222505, + "grad_norm": 0.34706047060118994, + "learning_rate": 4.74924455635899e-05, + "loss": 0.9803, + "step": 970 + }, + { + "epoch": 2.483375959079284, + "grad_norm": 0.4005154747079565, + "learning_rate": 4.742225767704286e-05, + "loss": 1.0025, + "step": 971 + }, + { + "epoch": 2.4859335038363173, + "grad_norm": 0.3356815450895581, + "learning_rate": 4.7352046114733116e-05, + "loss": 1.0224, + "step": 972 + }, + { + "epoch": 2.4884910485933505, + "grad_norm": 0.34686857253960623, + "learning_rate": 4.728181110062383e-05, + "loss": 1.0031, + "step": 973 + }, + { + "epoch": 2.4910485933503836, + "grad_norm": 0.26287696238858543, + "learning_rate": 4.721155285875294e-05, + "loss": 1.0016, + "step": 974 + }, + { + "epoch": 2.493606138107417, + "grad_norm": 0.34276165243054646, + "learning_rate": 4.714127161323255e-05, + "loss": 0.9816, + "step": 975 + }, + { + "epoch": 2.49616368286445, + "grad_norm": 0.28389546601802473, + "learning_rate": 4.707096758824807e-05, + "loss": 1.0099, + "step": 976 + }, + { + "epoch": 2.498721227621483, + "grad_norm": 0.2515028036738449, + "learning_rate": 4.7000641008057605e-05, + "loss": 0.999, + "step": 977 + }, + { + "epoch": 2.501278772378517, + "grad_norm": 0.2828216002216222, + "learning_rate": 4.6930292096991214e-05, + "loss": 1.0151, + "step": 978 + }, + { + "epoch": 2.50383631713555, + "grad_norm": 0.3728810365049237, + "learning_rate": 4.685992107945016e-05, + "loss": 1.0189, + "step": 979 + }, + { + "epoch": 2.506393861892583, + "grad_norm": 0.31696277982756876, + "learning_rate": 4.6789528179906254e-05, + "loss": 0.9779, + "step": 980 + }, + { + "epoch": 2.5089514066496164, + "grad_norm": 0.2802033478255871, + "learning_rate": 4.67191136229011e-05, + "loss": 0.987, + "step": 981 + }, + { + "epoch": 2.5115089514066495, + "grad_norm": 0.3340397189248965, + "learning_rate": 4.664867763304536e-05, + "loss": 0.9693, + "step": 982 + }, + { + "epoch": 2.5140664961636827, + "grad_norm": 0.31946713078301153, + "learning_rate": 4.657822043501807e-05, + "loss": 0.98, + "step": 983 + }, + { + "epoch": 2.516624040920716, + "grad_norm": 0.2827099960450771, + "learning_rate": 4.6507742253565964e-05, + "loss": 1.0107, + "step": 984 + }, + { + "epoch": 2.5191815856777495, + "grad_norm": 0.27277020153203635, + "learning_rate": 4.643724331350265e-05, + "loss": 1.0085, + "step": 985 + }, + { + "epoch": 2.5217391304347827, + "grad_norm": 0.2950110771549243, + "learning_rate": 4.6366723839707976e-05, + "loss": 0.9817, + "step": 986 + }, + { + "epoch": 2.524296675191816, + "grad_norm": 0.28737673459457397, + "learning_rate": 4.629618405712729e-05, + "loss": 0.9931, + "step": 987 + }, + { + "epoch": 2.526854219948849, + "grad_norm": 0.328976885705382, + "learning_rate": 4.6225624190770725e-05, + "loss": 1.0179, + "step": 988 + }, + { + "epoch": 2.5294117647058822, + "grad_norm": 0.28874722164604466, + "learning_rate": 4.615504446571248e-05, + "loss": 1.0176, + "step": 989 + }, + { + "epoch": 2.531969309462916, + "grad_norm": 0.2514991876721953, + "learning_rate": 4.6084445107090076e-05, + "loss": 1.0257, + "step": 990 + }, + { + "epoch": 2.5345268542199486, + "grad_norm": 0.25531502605497386, + "learning_rate": 4.601382634010367e-05, + "loss": 1.0067, + "step": 991 + }, + { + "epoch": 2.5370843989769822, + "grad_norm": 0.3276255896510962, + "learning_rate": 4.594318839001539e-05, + "loss": 1.0238, + "step": 992 + }, + { + "epoch": 2.5396419437340154, + "grad_norm": 0.26779804167719207, + "learning_rate": 4.587253148214845e-05, + "loss": 1.006, + "step": 993 + }, + { + "epoch": 2.5421994884910486, + "grad_norm": 0.37061278544032084, + "learning_rate": 4.580185584188661e-05, + "loss": 0.9882, + "step": 994 + }, + { + "epoch": 2.544757033248082, + "grad_norm": 0.2609641658380784, + "learning_rate": 4.573116169467336e-05, + "loss": 0.9815, + "step": 995 + }, + { + "epoch": 2.547314578005115, + "grad_norm": 0.38096136166812233, + "learning_rate": 4.5660449266011214e-05, + "loss": 0.9984, + "step": 996 + }, + { + "epoch": 2.5498721227621486, + "grad_norm": 0.33636150127134645, + "learning_rate": 4.558971878146104e-05, + "loss": 1.0026, + "step": 997 + }, + { + "epoch": 2.5524296675191813, + "grad_norm": 0.27604661429078775, + "learning_rate": 4.551897046664124e-05, + "loss": 0.9993, + "step": 998 + }, + { + "epoch": 2.554987212276215, + "grad_norm": 0.31994197051269924, + "learning_rate": 4.5448204547227154e-05, + "loss": 0.9949, + "step": 999 + }, + { + "epoch": 2.557544757033248, + "grad_norm": 0.3511508004578517, + "learning_rate": 4.537742124895024e-05, + "loss": 0.9982, + "step": 1000 + }, + { + "epoch": 2.5601023017902813, + "grad_norm": 0.29090223993859365, + "learning_rate": 4.530662079759738e-05, + "loss": 0.9997, + "step": 1001 + }, + { + "epoch": 2.5626598465473145, + "grad_norm": 0.2462037300168939, + "learning_rate": 4.523580341901023e-05, + "loss": 0.9904, + "step": 1002 + }, + { + "epoch": 2.5652173913043477, + "grad_norm": 0.2746318643511379, + "learning_rate": 4.516496933908438e-05, + "loss": 1.0181, + "step": 1003 + }, + { + "epoch": 2.5677749360613813, + "grad_norm": 0.31680827605801093, + "learning_rate": 4.509411878376873e-05, + "loss": 1.0241, + "step": 1004 + }, + { + "epoch": 2.5703324808184145, + "grad_norm": 0.34253425537117993, + "learning_rate": 4.50232519790647e-05, + "loss": 1.0167, + "step": 1005 + }, + { + "epoch": 2.5728900255754477, + "grad_norm": 0.26338348935218686, + "learning_rate": 4.495236915102559e-05, + "loss": 0.9547, + "step": 1006 + }, + { + "epoch": 2.575447570332481, + "grad_norm": 0.33449339157822405, + "learning_rate": 4.488147052575577e-05, + "loss": 1.0314, + "step": 1007 + }, + { + "epoch": 2.578005115089514, + "grad_norm": 0.42495311822077536, + "learning_rate": 4.481055632941003e-05, + "loss": 1.0195, + "step": 1008 + }, + { + "epoch": 2.580562659846547, + "grad_norm": 0.3543777063618769, + "learning_rate": 4.4739626788192806e-05, + "loss": 1.0179, + "step": 1009 + }, + { + "epoch": 2.5831202046035804, + "grad_norm": 0.2734074675391571, + "learning_rate": 4.4668682128357474e-05, + "loss": 1.0057, + "step": 1010 + }, + { + "epoch": 2.585677749360614, + "grad_norm": 0.3724948524313399, + "learning_rate": 4.45977225762057e-05, + "loss": 1.0057, + "step": 1011 + }, + { + "epoch": 2.588235294117647, + "grad_norm": 0.3795718033122553, + "learning_rate": 4.4526748358086566e-05, + "loss": 1.0112, + "step": 1012 + }, + { + "epoch": 2.5907928388746804, + "grad_norm": 0.30997955818137274, + "learning_rate": 4.445575970039599e-05, + "loss": 0.9937, + "step": 1013 + }, + { + "epoch": 2.5933503836317136, + "grad_norm": 0.27956010657525027, + "learning_rate": 4.438475682957595e-05, + "loss": 1.0112, + "step": 1014 + }, + { + "epoch": 2.5959079283887467, + "grad_norm": 0.3860164174815654, + "learning_rate": 4.431373997211372e-05, + "loss": 0.9941, + "step": 1015 + }, + { + "epoch": 2.59846547314578, + "grad_norm": 0.43208138507444893, + "learning_rate": 4.424270935454124e-05, + "loss": 0.9932, + "step": 1016 + }, + { + "epoch": 2.601023017902813, + "grad_norm": 0.2756081486534577, + "learning_rate": 4.4171665203434336e-05, + "loss": 1.0022, + "step": 1017 + }, + { + "epoch": 2.6035805626598467, + "grad_norm": 0.2986676945368832, + "learning_rate": 4.4100607745411947e-05, + "loss": 1.0067, + "step": 1018 + }, + { + "epoch": 2.60613810741688, + "grad_norm": 0.3273660950180001, + "learning_rate": 4.4029537207135535e-05, + "loss": 0.9788, + "step": 1019 + }, + { + "epoch": 2.608695652173913, + "grad_norm": 0.28650532791024097, + "learning_rate": 4.395845381530823e-05, + "loss": 0.9874, + "step": 1020 + }, + { + "epoch": 2.6112531969309463, + "grad_norm": 0.246812531877351, + "learning_rate": 4.3887357796674174e-05, + "loss": 0.9998, + "step": 1021 + }, + { + "epoch": 2.6138107416879794, + "grad_norm": 0.24906583542460717, + "learning_rate": 4.381624937801784e-05, + "loss": 0.9925, + "step": 1022 + }, + { + "epoch": 2.6163682864450126, + "grad_norm": 0.2738737665378058, + "learning_rate": 4.374512878616315e-05, + "loss": 1.0151, + "step": 1023 + }, + { + "epoch": 2.618925831202046, + "grad_norm": 0.29728516470216987, + "learning_rate": 4.3673996247972966e-05, + "loss": 0.9945, + "step": 1024 + }, + { + "epoch": 2.6214833759590794, + "grad_norm": 0.23273105269614583, + "learning_rate": 4.360285199034819e-05, + "loss": 0.9899, + "step": 1025 + }, + { + "epoch": 2.6240409207161126, + "grad_norm": 0.29094117249252427, + "learning_rate": 4.353169624022713e-05, + "loss": 0.9982, + "step": 1026 + }, + { + "epoch": 2.626598465473146, + "grad_norm": 0.29039727074129185, + "learning_rate": 4.346052922458473e-05, + "loss": 1.005, + "step": 1027 + }, + { + "epoch": 2.629156010230179, + "grad_norm": 0.2914886029801254, + "learning_rate": 4.3389351170431914e-05, + "loss": 1.0143, + "step": 1028 + }, + { + "epoch": 2.631713554987212, + "grad_norm": 0.25123267713852904, + "learning_rate": 4.331816230481478e-05, + "loss": 1.0175, + "step": 1029 + }, + { + "epoch": 2.634271099744246, + "grad_norm": 0.2761513425595422, + "learning_rate": 4.3246962854813904e-05, + "loss": 1.0032, + "step": 1030 + }, + { + "epoch": 2.6368286445012785, + "grad_norm": 0.2952763733952678, + "learning_rate": 4.317575304754367e-05, + "loss": 1.0079, + "step": 1031 + }, + { + "epoch": 2.639386189258312, + "grad_norm": 0.2396174674261751, + "learning_rate": 4.310453311015146e-05, + "loss": 0.9751, + "step": 1032 + }, + { + "epoch": 2.6419437340153453, + "grad_norm": 0.30588512401846724, + "learning_rate": 4.3033303269817e-05, + "loss": 1.0328, + "step": 1033 + }, + { + "epoch": 2.6445012787723785, + "grad_norm": 0.35350331623413067, + "learning_rate": 4.296206375375156e-05, + "loss": 1.011, + "step": 1034 + }, + { + "epoch": 2.6470588235294117, + "grad_norm": 0.28796718761109585, + "learning_rate": 4.289081478919731e-05, + "loss": 1.0149, + "step": 1035 + }, + { + "epoch": 2.649616368286445, + "grad_norm": 0.2383147053710307, + "learning_rate": 4.281955660342656e-05, + "loss": 0.9874, + "step": 1036 + }, + { + "epoch": 2.6521739130434785, + "grad_norm": 0.27157160483463727, + "learning_rate": 4.274828942374103e-05, + "loss": 0.9949, + "step": 1037 + }, + { + "epoch": 2.6547314578005117, + "grad_norm": 0.2534384580666098, + "learning_rate": 4.2677013477471104e-05, + "loss": 0.9997, + "step": 1038 + }, + { + "epoch": 2.657289002557545, + "grad_norm": 0.22567843700077125, + "learning_rate": 4.260572899197517e-05, + "loss": 1.0078, + "step": 1039 + }, + { + "epoch": 2.659846547314578, + "grad_norm": 0.29838728779169416, + "learning_rate": 4.253443619463882e-05, + "loss": 1.0185, + "step": 1040 + }, + { + "epoch": 2.662404092071611, + "grad_norm": 0.25199620198627026, + "learning_rate": 4.246313531287418e-05, + "loss": 0.9965, + "step": 1041 + }, + { + "epoch": 2.6649616368286444, + "grad_norm": 0.25813433899181243, + "learning_rate": 4.2391826574119156e-05, + "loss": 1.0276, + "step": 1042 + }, + { + "epoch": 2.6675191815856776, + "grad_norm": 0.2213842255776451, + "learning_rate": 4.2320510205836714e-05, + "loss": 0.9618, + "step": 1043 + }, + { + "epoch": 2.670076726342711, + "grad_norm": 0.34212252067273924, + "learning_rate": 4.2249186435514184e-05, + "loss": 1.0035, + "step": 1044 + }, + { + "epoch": 2.6726342710997444, + "grad_norm": 0.23979377486467365, + "learning_rate": 4.217785549066245e-05, + "loss": 0.9896, + "step": 1045 + }, + { + "epoch": 2.6751918158567776, + "grad_norm": 0.2138184617430876, + "learning_rate": 4.2106517598815337e-05, + "loss": 0.9743, + "step": 1046 + }, + { + "epoch": 2.6777493606138107, + "grad_norm": 0.2659130978637147, + "learning_rate": 4.203517298752881e-05, + "loss": 1.0077, + "step": 1047 + }, + { + "epoch": 2.680306905370844, + "grad_norm": 0.30697679317607074, + "learning_rate": 4.1963821884380246e-05, + "loss": 1.008, + "step": 1048 + }, + { + "epoch": 2.682864450127877, + "grad_norm": 0.22999291962370588, + "learning_rate": 4.1892464516967776e-05, + "loss": 0.9824, + "step": 1049 + }, + { + "epoch": 2.6854219948849103, + "grad_norm": 0.26998959283360185, + "learning_rate": 4.182110111290948e-05, + "loss": 1.0083, + "step": 1050 + }, + { + "epoch": 2.687979539641944, + "grad_norm": 0.29486316071332647, + "learning_rate": 4.17497318998427e-05, + "loss": 1.0048, + "step": 1051 + }, + { + "epoch": 2.690537084398977, + "grad_norm": 0.2986425273868719, + "learning_rate": 4.1678357105423306e-05, + "loss": 0.9973, + "step": 1052 + }, + { + "epoch": 2.6930946291560103, + "grad_norm": 0.25347000823131294, + "learning_rate": 4.1606976957324976e-05, + "loss": 1.01, + "step": 1053 + }, + { + "epoch": 2.6956521739130435, + "grad_norm": 0.25469040517907776, + "learning_rate": 4.153559168323847e-05, + "loss": 0.9737, + "step": 1054 + }, + { + "epoch": 2.6982097186700766, + "grad_norm": 0.2468042191556332, + "learning_rate": 4.146420151087089e-05, + "loss": 0.9875, + "step": 1055 + }, + { + "epoch": 2.70076726342711, + "grad_norm": 0.3168078654851793, + "learning_rate": 4.1392806667944965e-05, + "loss": 1.0072, + "step": 1056 + }, + { + "epoch": 2.703324808184143, + "grad_norm": 0.2577802930863439, + "learning_rate": 4.132140738219831e-05, + "loss": 1.0074, + "step": 1057 + }, + { + "epoch": 2.7058823529411766, + "grad_norm": 0.22240516058092002, + "learning_rate": 4.125000388138276e-05, + "loss": 0.9964, + "step": 1058 + }, + { + "epoch": 2.70843989769821, + "grad_norm": 0.2899408076786431, + "learning_rate": 4.117859639326352e-05, + "loss": 1.0246, + "step": 1059 + }, + { + "epoch": 2.710997442455243, + "grad_norm": 0.24816779206581488, + "learning_rate": 4.1107185145618574e-05, + "loss": 0.9832, + "step": 1060 + }, + { + "epoch": 2.713554987212276, + "grad_norm": 0.35105713611659695, + "learning_rate": 4.10357703662379e-05, + "loss": 1.0324, + "step": 1061 + }, + { + "epoch": 2.7161125319693094, + "grad_norm": 0.23582025700520193, + "learning_rate": 4.096435228292268e-05, + "loss": 1.0105, + "step": 1062 + }, + { + "epoch": 2.718670076726343, + "grad_norm": 0.224766329004601, + "learning_rate": 4.089293112348469e-05, + "loss": 1.043, + "step": 1063 + }, + { + "epoch": 2.7212276214833757, + "grad_norm": 0.2607269768142398, + "learning_rate": 4.082150711574552e-05, + "loss": 1.007, + "step": 1064 + }, + { + "epoch": 2.7237851662404093, + "grad_norm": 0.22627614919460648, + "learning_rate": 4.0750080487535807e-05, + "loss": 0.9762, + "step": 1065 + }, + { + "epoch": 2.7263427109974425, + "grad_norm": 0.2541558036254898, + "learning_rate": 4.067865146669459e-05, + "loss": 0.9949, + "step": 1066 + }, + { + "epoch": 2.7289002557544757, + "grad_norm": 0.23805701026486054, + "learning_rate": 4.060722028106849e-05, + "loss": 1.0013, + "step": 1067 + }, + { + "epoch": 2.731457800511509, + "grad_norm": 0.26429490418987045, + "learning_rate": 4.053578715851109e-05, + "loss": 1.0213, + "step": 1068 + }, + { + "epoch": 2.734015345268542, + "grad_norm": 0.24156276865731632, + "learning_rate": 4.046435232688211e-05, + "loss": 0.9954, + "step": 1069 + }, + { + "epoch": 2.7365728900255757, + "grad_norm": 0.2882909987018262, + "learning_rate": 4.0392916014046735e-05, + "loss": 1.0414, + "step": 1070 + }, + { + "epoch": 2.7391304347826084, + "grad_norm": 0.20412597587654827, + "learning_rate": 4.0321478447874895e-05, + "loss": 0.9911, + "step": 1071 + }, + { + "epoch": 2.741687979539642, + "grad_norm": 0.24979733236444654, + "learning_rate": 4.0250039856240476e-05, + "loss": 1.0219, + "step": 1072 + }, + { + "epoch": 2.7442455242966752, + "grad_norm": 0.24634392488290774, + "learning_rate": 4.0178600467020664e-05, + "loss": 1.0048, + "step": 1073 + }, + { + "epoch": 2.7468030690537084, + "grad_norm": 0.26158148075151116, + "learning_rate": 4.010716050809517e-05, + "loss": 0.9903, + "step": 1074 + }, + { + "epoch": 2.7493606138107416, + "grad_norm": 0.2179061990457055, + "learning_rate": 4.0035720207345556e-05, + "loss": 1.0435, + "step": 1075 + }, + { + "epoch": 2.7519181585677748, + "grad_norm": 0.2569251284054396, + "learning_rate": 3.9964279792654464e-05, + "loss": 1.0001, + "step": 1076 + }, + { + "epoch": 2.7544757033248084, + "grad_norm": 0.2834854817803271, + "learning_rate": 3.989283949190484e-05, + "loss": 1.0211, + "step": 1077 + }, + { + "epoch": 2.7570332480818416, + "grad_norm": 0.24647968373722906, + "learning_rate": 3.982139953297935e-05, + "loss": 0.9836, + "step": 1078 + }, + { + "epoch": 2.7595907928388748, + "grad_norm": 0.2603116983015655, + "learning_rate": 3.974996014375954e-05, + "loss": 0.9624, + "step": 1079 + }, + { + "epoch": 2.762148337595908, + "grad_norm": 0.26774223660199403, + "learning_rate": 3.967852155212511e-05, + "loss": 1.0087, + "step": 1080 + }, + { + "epoch": 2.764705882352941, + "grad_norm": 0.25111070048702194, + "learning_rate": 3.9607083985953265e-05, + "loss": 0.9716, + "step": 1081 + }, + { + "epoch": 2.7672634271099743, + "grad_norm": 0.27537157062354006, + "learning_rate": 3.95356476731179e-05, + "loss": 1.009, + "step": 1082 + }, + { + "epoch": 2.7698209718670075, + "grad_norm": 0.26627222201056977, + "learning_rate": 3.946421284148892e-05, + "loss": 1.0071, + "step": 1083 + }, + { + "epoch": 2.772378516624041, + "grad_norm": 0.2647256382334767, + "learning_rate": 3.939277971893152e-05, + "loss": 1.0099, + "step": 1084 + }, + { + "epoch": 2.7749360613810743, + "grad_norm": 0.3161240038333297, + "learning_rate": 3.932134853330542e-05, + "loss": 0.9844, + "step": 1085 + }, + { + "epoch": 2.7774936061381075, + "grad_norm": 0.24673702901545966, + "learning_rate": 3.92499195124642e-05, + "loss": 1.0384, + "step": 1086 + }, + { + "epoch": 2.7800511508951407, + "grad_norm": 0.2627376460002893, + "learning_rate": 3.91784928842545e-05, + "loss": 0.9644, + "step": 1087 + }, + { + "epoch": 2.782608695652174, + "grad_norm": 0.25015861128564604, + "learning_rate": 3.9107068876515314e-05, + "loss": 0.9907, + "step": 1088 + }, + { + "epoch": 2.785166240409207, + "grad_norm": 0.25505938008111867, + "learning_rate": 3.9035647717077335e-05, + "loss": 1.0117, + "step": 1089 + }, + { + "epoch": 2.78772378516624, + "grad_norm": 0.2885186518776429, + "learning_rate": 3.896422963376211e-05, + "loss": 0.9782, + "step": 1090 + }, + { + "epoch": 2.790281329923274, + "grad_norm": 0.24772701208848075, + "learning_rate": 3.8892814854381425e-05, + "loss": 0.9543, + "step": 1091 + }, + { + "epoch": 2.792838874680307, + "grad_norm": 0.3043825900169142, + "learning_rate": 3.882140360673648e-05, + "loss": 1.0135, + "step": 1092 + }, + { + "epoch": 2.79539641943734, + "grad_norm": 0.26534262469691305, + "learning_rate": 3.8749996118617256e-05, + "loss": 1.0158, + "step": 1093 + }, + { + "epoch": 2.7979539641943734, + "grad_norm": 0.2439645917934899, + "learning_rate": 3.86785926178017e-05, + "loss": 0.9991, + "step": 1094 + }, + { + "epoch": 2.8005115089514065, + "grad_norm": 0.2963910857686297, + "learning_rate": 3.860719333205505e-05, + "loss": 0.9949, + "step": 1095 + }, + { + "epoch": 2.80306905370844, + "grad_norm": 0.3347355595394114, + "learning_rate": 3.853579848912912e-05, + "loss": 1.043, + "step": 1096 + }, + { + "epoch": 2.805626598465473, + "grad_norm": 0.23698370137396493, + "learning_rate": 3.846440831676153e-05, + "loss": 1.0109, + "step": 1097 + }, + { + "epoch": 2.8081841432225065, + "grad_norm": 0.2553667731091932, + "learning_rate": 3.839302304267504e-05, + "loss": 1.0243, + "step": 1098 + }, + { + "epoch": 2.8107416879795397, + "grad_norm": 0.25553599743585725, + "learning_rate": 3.832164289457671e-05, + "loss": 1.0148, + "step": 1099 + }, + { + "epoch": 2.813299232736573, + "grad_norm": 0.20938583263905827, + "learning_rate": 3.8250268100157304e-05, + "loss": 0.9899, + "step": 1100 + }, + { + "epoch": 2.815856777493606, + "grad_norm": 0.2803140720481866, + "learning_rate": 3.817889888709053e-05, + "loss": 1.0172, + "step": 1101 + }, + { + "epoch": 2.8184143222506393, + "grad_norm": 0.26697443095676276, + "learning_rate": 3.8107535483032224e-05, + "loss": 0.999, + "step": 1102 + }, + { + "epoch": 2.820971867007673, + "grad_norm": 0.24132691739947107, + "learning_rate": 3.803617811561976e-05, + "loss": 1.0137, + "step": 1103 + }, + { + "epoch": 2.8235294117647056, + "grad_norm": 0.28295370026487926, + "learning_rate": 3.796482701247121e-05, + "loss": 0.9906, + "step": 1104 + }, + { + "epoch": 2.8260869565217392, + "grad_norm": 0.2363477732559459, + "learning_rate": 3.789348240118468e-05, + "loss": 0.9807, + "step": 1105 + }, + { + "epoch": 2.8286445012787724, + "grad_norm": 0.2870494002943626, + "learning_rate": 3.782214450933756e-05, + "loss": 1.0106, + "step": 1106 + }, + { + "epoch": 2.8312020460358056, + "grad_norm": 0.24656039873041008, + "learning_rate": 3.775081356448583e-05, + "loss": 1.0069, + "step": 1107 + }, + { + "epoch": 2.833759590792839, + "grad_norm": 0.24824680931599624, + "learning_rate": 3.7679489794163286e-05, + "loss": 1.017, + "step": 1108 + }, + { + "epoch": 2.836317135549872, + "grad_norm": 0.25584210376463995, + "learning_rate": 3.760817342588086e-05, + "loss": 1.0209, + "step": 1109 + }, + { + "epoch": 2.8388746803069056, + "grad_norm": 0.23768463941469484, + "learning_rate": 3.753686468712583e-05, + "loss": 1.002, + "step": 1110 + }, + { + "epoch": 2.8414322250639388, + "grad_norm": 0.2140982655494908, + "learning_rate": 3.746556380536119e-05, + "loss": 1.0018, + "step": 1111 + }, + { + "epoch": 2.843989769820972, + "grad_norm": 0.2322699559270119, + "learning_rate": 3.7394271008024834e-05, + "loss": 1.0102, + "step": 1112 + }, + { + "epoch": 2.846547314578005, + "grad_norm": 0.2541795231409938, + "learning_rate": 3.7322986522528896e-05, + "loss": 0.9998, + "step": 1113 + }, + { + "epoch": 2.8491048593350383, + "grad_norm": 0.2711876988209337, + "learning_rate": 3.7251710576258966e-05, + "loss": 1.0001, + "step": 1114 + }, + { + "epoch": 2.8516624040920715, + "grad_norm": 0.23402242185528463, + "learning_rate": 3.7180443396573445e-05, + "loss": 1.0051, + "step": 1115 + }, + { + "epoch": 2.8542199488491047, + "grad_norm": 0.225618095882578, + "learning_rate": 3.71091852108027e-05, + "loss": 0.997, + "step": 1116 + }, + { + "epoch": 2.8567774936061383, + "grad_norm": 0.2373879922383219, + "learning_rate": 3.7037936246248455e-05, + "loss": 0.9823, + "step": 1117 + }, + { + "epoch": 2.8593350383631715, + "grad_norm": 0.2179619293216389, + "learning_rate": 3.696669673018302e-05, + "loss": 1.014, + "step": 1118 + }, + { + "epoch": 2.8618925831202047, + "grad_norm": 0.23271028414622047, + "learning_rate": 3.689546688984854e-05, + "loss": 0.9867, + "step": 1119 + }, + { + "epoch": 2.864450127877238, + "grad_norm": 0.22959905790536927, + "learning_rate": 3.6824246952456346e-05, + "loss": 1.0419, + "step": 1120 + }, + { + "epoch": 2.867007672634271, + "grad_norm": 0.22427029568274529, + "learning_rate": 3.675303714518611e-05, + "loss": 1.0103, + "step": 1121 + }, + { + "epoch": 2.869565217391304, + "grad_norm": 0.23497147356312345, + "learning_rate": 3.6681837695185235e-05, + "loss": 0.9924, + "step": 1122 + }, + { + "epoch": 2.8721227621483374, + "grad_norm": 0.2313618533257831, + "learning_rate": 3.66106488295681e-05, + "loss": 0.9621, + "step": 1123 + }, + { + "epoch": 2.874680306905371, + "grad_norm": 0.21491888385362423, + "learning_rate": 3.653947077541527e-05, + "loss": 1.0429, + "step": 1124 + }, + { + "epoch": 2.877237851662404, + "grad_norm": 0.2142173562919566, + "learning_rate": 3.646830375977288e-05, + "loss": 1.0197, + "step": 1125 + }, + { + "epoch": 2.8797953964194374, + "grad_norm": 0.22758855581505263, + "learning_rate": 3.639714800965183e-05, + "loss": 1.0112, + "step": 1126 + }, + { + "epoch": 2.8823529411764706, + "grad_norm": 0.2037855866672826, + "learning_rate": 3.632600375202704e-05, + "loss": 0.9783, + "step": 1127 + }, + { + "epoch": 2.8849104859335037, + "grad_norm": 0.245178868595634, + "learning_rate": 3.625487121383686e-05, + "loss": 0.9839, + "step": 1128 + }, + { + "epoch": 2.887468030690537, + "grad_norm": 0.24685844906961088, + "learning_rate": 3.6183750621982176e-05, + "loss": 0.9899, + "step": 1129 + }, + { + "epoch": 2.89002557544757, + "grad_norm": 0.30768839678374543, + "learning_rate": 3.611264220332583e-05, + "loss": 1.0155, + "step": 1130 + }, + { + "epoch": 2.8925831202046037, + "grad_norm": 0.27604427834359535, + "learning_rate": 3.60415461846918e-05, + "loss": 1.0281, + "step": 1131 + }, + { + "epoch": 2.895140664961637, + "grad_norm": 0.2953227328216093, + "learning_rate": 3.5970462792864486e-05, + "loss": 0.9653, + "step": 1132 + }, + { + "epoch": 2.89769820971867, + "grad_norm": 0.2776299135622876, + "learning_rate": 3.589939225458807e-05, + "loss": 0.9844, + "step": 1133 + }, + { + "epoch": 2.9002557544757033, + "grad_norm": 0.25982730487682326, + "learning_rate": 3.582833479656568e-05, + "loss": 0.9857, + "step": 1134 + }, + { + "epoch": 2.9028132992327365, + "grad_norm": 0.272622930597586, + "learning_rate": 3.575729064545876e-05, + "loss": 1.01, + "step": 1135 + }, + { + "epoch": 2.90537084398977, + "grad_norm": 0.28024133642600774, + "learning_rate": 3.568626002788628e-05, + "loss": 0.9812, + "step": 1136 + }, + { + "epoch": 2.907928388746803, + "grad_norm": 0.21927470498778384, + "learning_rate": 3.5615243170424074e-05, + "loss": 1.0438, + "step": 1137 + }, + { + "epoch": 2.9104859335038364, + "grad_norm": 0.2407505627513316, + "learning_rate": 3.5544240299604024e-05, + "loss": 0.9827, + "step": 1138 + }, + { + "epoch": 2.9130434782608696, + "grad_norm": 0.2583021004717082, + "learning_rate": 3.547325164191345e-05, + "loss": 0.9976, + "step": 1139 + }, + { + "epoch": 2.915601023017903, + "grad_norm": 0.23084176420891026, + "learning_rate": 3.5402277423794316e-05, + "loss": 0.9897, + "step": 1140 + }, + { + "epoch": 2.918158567774936, + "grad_norm": 0.258034589423058, + "learning_rate": 3.5331317871642525e-05, + "loss": 1.0057, + "step": 1141 + }, + { + "epoch": 2.920716112531969, + "grad_norm": 0.3037754487191455, + "learning_rate": 3.526037321180722e-05, + "loss": 1.0152, + "step": 1142 + }, + { + "epoch": 2.923273657289003, + "grad_norm": 0.20675604578569703, + "learning_rate": 3.518944367058999e-05, + "loss": 1.0117, + "step": 1143 + }, + { + "epoch": 2.9258312020460355, + "grad_norm": 0.278991208465292, + "learning_rate": 3.511852947424424e-05, + "loss": 0.99, + "step": 1144 + }, + { + "epoch": 2.928388746803069, + "grad_norm": 0.24889238741757388, + "learning_rate": 3.504763084897442e-05, + "loss": 0.9982, + "step": 1145 + }, + { + "epoch": 2.9309462915601023, + "grad_norm": 0.2497728135979749, + "learning_rate": 3.4976748020935305e-05, + "loss": 0.9852, + "step": 1146 + }, + { + "epoch": 2.9335038363171355, + "grad_norm": 0.27461155792883407, + "learning_rate": 3.4905881216231284e-05, + "loss": 1.0029, + "step": 1147 + }, + { + "epoch": 2.9360613810741687, + "grad_norm": 0.22344834081280318, + "learning_rate": 3.4835030660915634e-05, + "loss": 1.0008, + "step": 1148 + }, + { + "epoch": 2.938618925831202, + "grad_norm": 0.25313066405291, + "learning_rate": 3.476419658098978e-05, + "loss": 1.0166, + "step": 1149 + }, + { + "epoch": 2.9411764705882355, + "grad_norm": 0.2368466744031409, + "learning_rate": 3.469337920240263e-05, + "loss": 1.0058, + "step": 1150 + }, + { + "epoch": 2.9437340153452687, + "grad_norm": 0.260205430555691, + "learning_rate": 3.462257875104977e-05, + "loss": 1.0211, + "step": 1151 + }, + { + "epoch": 2.946291560102302, + "grad_norm": 0.24750655038016556, + "learning_rate": 3.455179545277286e-05, + "loss": 0.9922, + "step": 1152 + }, + { + "epoch": 2.948849104859335, + "grad_norm": 0.26061134046269646, + "learning_rate": 3.4481029533358775e-05, + "loss": 1.0137, + "step": 1153 + }, + { + "epoch": 2.9514066496163682, + "grad_norm": 0.23020961645395976, + "learning_rate": 3.441028121853898e-05, + "loss": 1.0183, + "step": 1154 + }, + { + "epoch": 2.9539641943734014, + "grad_norm": 0.22552716653712218, + "learning_rate": 3.43395507339888e-05, + "loss": 1.0343, + "step": 1155 + }, + { + "epoch": 2.9565217391304346, + "grad_norm": 0.22338059564503102, + "learning_rate": 3.426883830532665e-05, + "loss": 1.0078, + "step": 1156 + }, + { + "epoch": 2.959079283887468, + "grad_norm": 0.26129239518427483, + "learning_rate": 3.4198144158113405e-05, + "loss": 0.9989, + "step": 1157 + }, + { + "epoch": 2.9616368286445014, + "grad_norm": 0.2292830652258008, + "learning_rate": 3.412746851785156e-05, + "loss": 0.9938, + "step": 1158 + }, + { + "epoch": 2.9641943734015346, + "grad_norm": 0.24649745683965166, + "learning_rate": 3.405681160998463e-05, + "loss": 0.9934, + "step": 1159 + }, + { + "epoch": 2.9667519181585678, + "grad_norm": 0.23007278451720706, + "learning_rate": 3.3986173659896335e-05, + "loss": 1.015, + "step": 1160 + }, + { + "epoch": 2.969309462915601, + "grad_norm": 0.1758081567513317, + "learning_rate": 3.3915554892909944e-05, + "loss": 0.985, + "step": 1161 + }, + { + "epoch": 2.971867007672634, + "grad_norm": 0.2255832579224465, + "learning_rate": 3.384495553428754e-05, + "loss": 0.9894, + "step": 1162 + }, + { + "epoch": 2.9744245524296673, + "grad_norm": 0.21165101777080628, + "learning_rate": 3.3774375809229275e-05, + "loss": 1.0311, + "step": 1163 + }, + { + "epoch": 2.976982097186701, + "grad_norm": 0.23815079825461055, + "learning_rate": 3.370381594287273e-05, + "loss": 0.985, + "step": 1164 + }, + { + "epoch": 2.979539641943734, + "grad_norm": 0.2196988840192207, + "learning_rate": 3.3633276160292044e-05, + "loss": 0.9935, + "step": 1165 + }, + { + "epoch": 2.9820971867007673, + "grad_norm": 0.20489499569969788, + "learning_rate": 3.3562756686497364e-05, + "loss": 1.0254, + "step": 1166 + }, + { + "epoch": 2.9846547314578005, + "grad_norm": 0.19784059419234884, + "learning_rate": 3.349225774643405e-05, + "loss": 1.0012, + "step": 1167 + }, + { + "epoch": 2.9872122762148337, + "grad_norm": 0.18918392336436632, + "learning_rate": 3.342177956498193e-05, + "loss": 1.0156, + "step": 1168 + }, + { + "epoch": 2.9897698209718673, + "grad_norm": 0.20588207839076775, + "learning_rate": 3.3351322366954656e-05, + "loss": 0.9923, + "step": 1169 + }, + { + "epoch": 2.9923273657289, + "grad_norm": 0.19126183243148004, + "learning_rate": 3.328088637709892e-05, + "loss": 0.9903, + "step": 1170 + }, + { + "epoch": 2.9948849104859336, + "grad_norm": 0.20591684079974318, + "learning_rate": 3.321047182009375e-05, + "loss": 0.9872, + "step": 1171 + }, + { + "epoch": 2.997442455242967, + "grad_norm": 0.20972832124324786, + "learning_rate": 3.314007892054985e-05, + "loss": 1.0294, + "step": 1172 + }, + { + "epoch": 3.0, + "grad_norm": 0.19979271297295348, + "learning_rate": 3.30697079030088e-05, + "loss": 1.0199, + "step": 1173 + }, + { + "epoch": 3.002557544757033, + "grad_norm": 0.20927717303693083, + "learning_rate": 3.29993589919424e-05, + "loss": 1.0065, + "step": 1174 + }, + { + "epoch": 3.0051150895140664, + "grad_norm": 0.1894544897464941, + "learning_rate": 3.292903241175195e-05, + "loss": 1.0053, + "step": 1175 + }, + { + "epoch": 3.0076726342710995, + "grad_norm": 0.19263685473563882, + "learning_rate": 3.285872838676746e-05, + "loss": 0.9849, + "step": 1176 + }, + { + "epoch": 3.010230179028133, + "grad_norm": 0.20499176394344015, + "learning_rate": 3.278844714124707e-05, + "loss": 1.0211, + "step": 1177 + }, + { + "epoch": 3.0127877237851663, + "grad_norm": 0.2267745823697339, + "learning_rate": 3.2718188899376183e-05, + "loss": 1.0096, + "step": 1178 + }, + { + "epoch": 3.0153452685421995, + "grad_norm": 0.19329973297024897, + "learning_rate": 3.264795388526689e-05, + "loss": 0.9608, + "step": 1179 + }, + { + "epoch": 3.0179028132992327, + "grad_norm": 0.2456406558753071, + "learning_rate": 3.257774232295714e-05, + "loss": 0.9707, + "step": 1180 + }, + { + "epoch": 3.020460358056266, + "grad_norm": 0.22184901987844155, + "learning_rate": 3.250755443641012e-05, + "loss": 0.9833, + "step": 1181 + }, + { + "epoch": 3.023017902813299, + "grad_norm": 0.20437847819940258, + "learning_rate": 3.243739044951343e-05, + "loss": 0.9764, + "step": 1182 + }, + { + "epoch": 3.0255754475703327, + "grad_norm": 0.20756135809272994, + "learning_rate": 3.2367250586078496e-05, + "loss": 0.9979, + "step": 1183 + }, + { + "epoch": 3.028132992327366, + "grad_norm": 0.21183456404253598, + "learning_rate": 3.2297135069839766e-05, + "loss": 0.9648, + "step": 1184 + }, + { + "epoch": 3.030690537084399, + "grad_norm": 0.19385697959377585, + "learning_rate": 3.2227044124454024e-05, + "loss": 0.979, + "step": 1185 + }, + { + "epoch": 3.0332480818414322, + "grad_norm": 0.24603982366166482, + "learning_rate": 3.215697797349972e-05, + "loss": 0.9618, + "step": 1186 + }, + { + "epoch": 3.0358056265984654, + "grad_norm": 0.1921193878500632, + "learning_rate": 3.2086936840476124e-05, + "loss": 0.9852, + "step": 1187 + }, + { + "epoch": 3.0383631713554986, + "grad_norm": 0.22131221807833543, + "learning_rate": 3.201692094880278e-05, + "loss": 0.9512, + "step": 1188 + }, + { + "epoch": 3.040920716112532, + "grad_norm": 0.26400713025589573, + "learning_rate": 3.194693052181869e-05, + "loss": 0.9841, + "step": 1189 + }, + { + "epoch": 3.0434782608695654, + "grad_norm": 0.18604109405773514, + "learning_rate": 3.1876965782781626e-05, + "loss": 0.9725, + "step": 1190 + }, + { + "epoch": 3.0460358056265986, + "grad_norm": 0.32633685255896494, + "learning_rate": 3.1807026954867445e-05, + "loss": 0.9523, + "step": 1191 + }, + { + "epoch": 3.0485933503836318, + "grad_norm": 0.21116781754958366, + "learning_rate": 3.173711426116931e-05, + "loss": 0.9898, + "step": 1192 + }, + { + "epoch": 3.051150895140665, + "grad_norm": 0.2464885736886057, + "learning_rate": 3.1667227924697033e-05, + "loss": 0.9975, + "step": 1193 + }, + { + "epoch": 3.053708439897698, + "grad_norm": 0.2174756211201157, + "learning_rate": 3.1597368168376376e-05, + "loss": 0.9695, + "step": 1194 + }, + { + "epoch": 3.0562659846547313, + "grad_norm": 0.21679630916030693, + "learning_rate": 3.1527535215048284e-05, + "loss": 0.9757, + "step": 1195 + }, + { + "epoch": 3.0588235294117645, + "grad_norm": 0.19385880112351023, + "learning_rate": 3.1457729287468225e-05, + "loss": 0.9874, + "step": 1196 + }, + { + "epoch": 3.061381074168798, + "grad_norm": 0.2307908579955019, + "learning_rate": 3.138795060830549e-05, + "loss": 0.9883, + "step": 1197 + }, + { + "epoch": 3.0639386189258313, + "grad_norm": 0.19918037708419395, + "learning_rate": 3.131819940014234e-05, + "loss": 0.9853, + "step": 1198 + }, + { + "epoch": 3.0664961636828645, + "grad_norm": 0.27259883681321456, + "learning_rate": 3.1248475885473534e-05, + "loss": 1.0029, + "step": 1199 + }, + { + "epoch": 3.0690537084398977, + "grad_norm": 0.21850928888775867, + "learning_rate": 3.117878028670542e-05, + "loss": 0.9627, + "step": 1200 + }, + { + "epoch": 3.071611253196931, + "grad_norm": 0.26311180789408495, + "learning_rate": 3.110911282615533e-05, + "loss": 0.9749, + "step": 1201 + }, + { + "epoch": 3.074168797953964, + "grad_norm": 0.2365413768196383, + "learning_rate": 3.103947372605084e-05, + "loss": 0.9971, + "step": 1202 + }, + { + "epoch": 3.0767263427109977, + "grad_norm": 0.20574394116778202, + "learning_rate": 3.0969863208529044e-05, + "loss": 1.0029, + "step": 1203 + }, + { + "epoch": 3.079283887468031, + "grad_norm": 0.2285897523783578, + "learning_rate": 3.090028149563587e-05, + "loss": 1.0025, + "step": 1204 + }, + { + "epoch": 3.081841432225064, + "grad_norm": 0.2062668987784313, + "learning_rate": 3.083072880932535e-05, + "loss": 1.0031, + "step": 1205 + }, + { + "epoch": 3.084398976982097, + "grad_norm": 0.19032654177423064, + "learning_rate": 3.0761205371458953e-05, + "loss": 0.974, + "step": 1206 + }, + { + "epoch": 3.0869565217391304, + "grad_norm": 0.21207164634258652, + "learning_rate": 3.069171140380485e-05, + "loss": 0.9952, + "step": 1207 + }, + { + "epoch": 3.0895140664961636, + "grad_norm": 0.21972821827355038, + "learning_rate": 3.06222471280372e-05, + "loss": 1.0226, + "step": 1208 + }, + { + "epoch": 3.0920716112531967, + "grad_norm": 0.20237512777327865, + "learning_rate": 3.055281276573539e-05, + "loss": 0.9692, + "step": 1209 + }, + { + "epoch": 3.0946291560102304, + "grad_norm": 0.2216867068993571, + "learning_rate": 3.04834085383835e-05, + "loss": 0.9928, + "step": 1210 + }, + { + "epoch": 3.0971867007672635, + "grad_norm": 0.19513855102804212, + "learning_rate": 3.04140346673694e-05, + "loss": 0.9933, + "step": 1211 + }, + { + "epoch": 3.0997442455242967, + "grad_norm": 0.21612553792582045, + "learning_rate": 3.034469137398417e-05, + "loss": 0.9588, + "step": 1212 + }, + { + "epoch": 3.10230179028133, + "grad_norm": 0.197236734236815, + "learning_rate": 3.0275378879421313e-05, + "loss": 1.0255, + "step": 1213 + }, + { + "epoch": 3.104859335038363, + "grad_norm": 0.20305992275112705, + "learning_rate": 3.0206097404776158e-05, + "loss": 0.9745, + "step": 1214 + }, + { + "epoch": 3.1074168797953963, + "grad_norm": 0.20883263282580242, + "learning_rate": 3.0136847171044996e-05, + "loss": 1.0162, + "step": 1215 + }, + { + "epoch": 3.10997442455243, + "grad_norm": 0.1988099822661825, + "learning_rate": 3.0067628399124523e-05, + "loss": 0.9678, + "step": 1216 + }, + { + "epoch": 3.112531969309463, + "grad_norm": 0.19802134385759912, + "learning_rate": 2.999844130981108e-05, + "loss": 1.0046, + "step": 1217 + }, + { + "epoch": 3.1150895140664963, + "grad_norm": 0.21886668903308013, + "learning_rate": 2.9929286123799906e-05, + "loss": 0.9786, + "step": 1218 + }, + { + "epoch": 3.1176470588235294, + "grad_norm": 0.20270734092286255, + "learning_rate": 2.986016306168453e-05, + "loss": 0.9952, + "step": 1219 + }, + { + "epoch": 3.1202046035805626, + "grad_norm": 0.19771293400631856, + "learning_rate": 2.9791072343955946e-05, + "loss": 0.9832, + "step": 1220 + }, + { + "epoch": 3.122762148337596, + "grad_norm": 0.1998240112481755, + "learning_rate": 2.972201419100202e-05, + "loss": 0.9803, + "step": 1221 + }, + { + "epoch": 3.125319693094629, + "grad_norm": 0.21631234902354732, + "learning_rate": 2.9652988823106736e-05, + "loss": 0.9296, + "step": 1222 + }, + { + "epoch": 3.1278772378516626, + "grad_norm": 0.2001811610318111, + "learning_rate": 2.9583996460449474e-05, + "loss": 0.9911, + "step": 1223 + }, + { + "epoch": 3.130434782608696, + "grad_norm": 0.223830854725341, + "learning_rate": 2.9515037323104373e-05, + "loss": 0.9603, + "step": 1224 + }, + { + "epoch": 3.132992327365729, + "grad_norm": 0.18521322384491026, + "learning_rate": 2.9446111631039572e-05, + "loss": 0.9667, + "step": 1225 + }, + { + "epoch": 3.135549872122762, + "grad_norm": 0.21059872337735092, + "learning_rate": 2.9377219604116484e-05, + "loss": 0.9842, + "step": 1226 + }, + { + "epoch": 3.1381074168797953, + "grad_norm": 0.22516840522850048, + "learning_rate": 2.930836146208921e-05, + "loss": 0.9838, + "step": 1227 + }, + { + "epoch": 3.1406649616368285, + "grad_norm": 0.17870913096115748, + "learning_rate": 2.92395374246037e-05, + "loss": 1.0012, + "step": 1228 + }, + { + "epoch": 3.1432225063938617, + "grad_norm": 0.22974849107011203, + "learning_rate": 2.9170747711197154e-05, + "loss": 1.0068, + "step": 1229 + }, + { + "epoch": 3.1457800511508953, + "grad_norm": 0.18080075081074998, + "learning_rate": 2.9101992541297288e-05, + "loss": 0.9905, + "step": 1230 + }, + { + "epoch": 3.1483375959079285, + "grad_norm": 0.22879808464892484, + "learning_rate": 2.9033272134221575e-05, + "loss": 0.999, + "step": 1231 + }, + { + "epoch": 3.1508951406649617, + "grad_norm": 0.22112889389313278, + "learning_rate": 2.8964586709176674e-05, + "loss": 0.9777, + "step": 1232 + }, + { + "epoch": 3.153452685421995, + "grad_norm": 0.19632604879440582, + "learning_rate": 2.88959364852576e-05, + "loss": 1.0168, + "step": 1233 + }, + { + "epoch": 3.156010230179028, + "grad_norm": 0.21685788275652715, + "learning_rate": 2.882732168144712e-05, + "loss": 0.9994, + "step": 1234 + }, + { + "epoch": 3.1585677749360612, + "grad_norm": 0.2179611102756312, + "learning_rate": 2.8758742516614994e-05, + "loss": 0.9893, + "step": 1235 + }, + { + "epoch": 3.1611253196930944, + "grad_norm": 0.19368842885976295, + "learning_rate": 2.8690199209517322e-05, + "loss": 0.9934, + "step": 1236 + }, + { + "epoch": 3.163682864450128, + "grad_norm": 0.19759500447000294, + "learning_rate": 2.8621691978795793e-05, + "loss": 0.9822, + "step": 1237 + }, + { + "epoch": 3.166240409207161, + "grad_norm": 0.19295856916803417, + "learning_rate": 2.8553221042977022e-05, + "loss": 0.9868, + "step": 1238 + }, + { + "epoch": 3.1687979539641944, + "grad_norm": 0.25414221486911515, + "learning_rate": 2.84847866204719e-05, + "loss": 0.9566, + "step": 1239 + }, + { + "epoch": 3.1713554987212276, + "grad_norm": 0.1883908256135399, + "learning_rate": 2.8416388929574768e-05, + "loss": 0.9772, + "step": 1240 + }, + { + "epoch": 3.1739130434782608, + "grad_norm": 0.2326630917084391, + "learning_rate": 2.8348028188462876e-05, + "loss": 0.9663, + "step": 1241 + }, + { + "epoch": 3.176470588235294, + "grad_norm": 0.201683374235404, + "learning_rate": 2.8279704615195554e-05, + "loss": 0.9727, + "step": 1242 + }, + { + "epoch": 3.1790281329923276, + "grad_norm": 0.1971824727000338, + "learning_rate": 2.8211418427713584e-05, + "loss": 0.9812, + "step": 1243 + }, + { + "epoch": 3.1815856777493607, + "grad_norm": 0.19508872782415485, + "learning_rate": 2.8143169843838522e-05, + "loss": 0.9663, + "step": 1244 + }, + { + "epoch": 3.184143222506394, + "grad_norm": 0.18504489219850168, + "learning_rate": 2.807495908127194e-05, + "loss": 0.944, + "step": 1245 + }, + { + "epoch": 3.186700767263427, + "grad_norm": 0.18984378821009806, + "learning_rate": 2.8006786357594795e-05, + "loss": 0.9938, + "step": 1246 + }, + { + "epoch": 3.1892583120204603, + "grad_norm": 0.1875579903849718, + "learning_rate": 2.7938651890266695e-05, + "loss": 1.0008, + "step": 1247 + }, + { + "epoch": 3.1918158567774935, + "grad_norm": 0.18441764733962582, + "learning_rate": 2.7870555896625196e-05, + "loss": 0.9826, + "step": 1248 + }, + { + "epoch": 3.1943734015345266, + "grad_norm": 0.1805205042347647, + "learning_rate": 2.7802498593885155e-05, + "loss": 0.9715, + "step": 1249 + }, + { + "epoch": 3.1969309462915603, + "grad_norm": 0.1793075530333337, + "learning_rate": 2.7734480199138007e-05, + "loss": 0.981, + "step": 1250 + }, + { + "epoch": 3.1994884910485935, + "grad_norm": 0.21281110414647764, + "learning_rate": 2.7666500929351082e-05, + "loss": 0.9591, + "step": 1251 + }, + { + "epoch": 3.2020460358056266, + "grad_norm": 0.1916993153560797, + "learning_rate": 2.7598561001366905e-05, + "loss": 0.9972, + "step": 1252 + }, + { + "epoch": 3.20460358056266, + "grad_norm": 0.19622262403295782, + "learning_rate": 2.7530660631902467e-05, + "loss": 0.9543, + "step": 1253 + }, + { + "epoch": 3.207161125319693, + "grad_norm": 0.17506135270655868, + "learning_rate": 2.7462800037548643e-05, + "loss": 0.9787, + "step": 1254 + }, + { + "epoch": 3.209718670076726, + "grad_norm": 0.19641653510944684, + "learning_rate": 2.7394979434769378e-05, + "loss": 0.9786, + "step": 1255 + }, + { + "epoch": 3.21227621483376, + "grad_norm": 0.191620221833112, + "learning_rate": 2.7327199039901092e-05, + "loss": 0.9703, + "step": 1256 + }, + { + "epoch": 3.214833759590793, + "grad_norm": 0.18678920887787637, + "learning_rate": 2.725945906915191e-05, + "loss": 0.9748, + "step": 1257 + }, + { + "epoch": 3.217391304347826, + "grad_norm": 0.18895520158914417, + "learning_rate": 2.7191759738601052e-05, + "loss": 0.9708, + "step": 1258 + }, + { + "epoch": 3.2199488491048593, + "grad_norm": 0.17852521291354706, + "learning_rate": 2.712410126419805e-05, + "loss": 1.0167, + "step": 1259 + }, + { + "epoch": 3.2225063938618925, + "grad_norm": 0.1975194067747155, + "learning_rate": 2.705648386176215e-05, + "loss": 0.9843, + "step": 1260 + }, + { + "epoch": 3.2250639386189257, + "grad_norm": 0.19210567469178344, + "learning_rate": 2.698890774698158e-05, + "loss": 0.9679, + "step": 1261 + }, + { + "epoch": 3.227621483375959, + "grad_norm": 0.17550111816818414, + "learning_rate": 2.6921373135412852e-05, + "loss": 0.9504, + "step": 1262 + }, + { + "epoch": 3.2301790281329925, + "grad_norm": 0.19205803836193247, + "learning_rate": 2.685388024248013e-05, + "loss": 0.9719, + "step": 1263 + }, + { + "epoch": 3.2327365728900257, + "grad_norm": 0.19464925988695062, + "learning_rate": 2.6786429283474454e-05, + "loss": 0.9774, + "step": 1264 + }, + { + "epoch": 3.235294117647059, + "grad_norm": 0.18695245921746287, + "learning_rate": 2.67190204735531e-05, + "loss": 0.9719, + "step": 1265 + }, + { + "epoch": 3.237851662404092, + "grad_norm": 0.18689844980516787, + "learning_rate": 2.665165402773895e-05, + "loss": 0.9793, + "step": 1266 + }, + { + "epoch": 3.2404092071611252, + "grad_norm": 0.1793689613933117, + "learning_rate": 2.6584330160919693e-05, + "loss": 0.9725, + "step": 1267 + }, + { + "epoch": 3.2429667519181584, + "grad_norm": 0.19239691492067834, + "learning_rate": 2.6517049087847245e-05, + "loss": 0.9351, + "step": 1268 + }, + { + "epoch": 3.2455242966751916, + "grad_norm": 0.18288592509102963, + "learning_rate": 2.6449811023137e-05, + "loss": 0.9962, + "step": 1269 + }, + { + "epoch": 3.2480818414322252, + "grad_norm": 0.231930212219846, + "learning_rate": 2.6382616181267125e-05, + "loss": 0.973, + "step": 1270 + }, + { + "epoch": 3.2506393861892584, + "grad_norm": 0.17786106163226348, + "learning_rate": 2.6315464776577974e-05, + "loss": 0.9895, + "step": 1271 + }, + { + "epoch": 3.2531969309462916, + "grad_norm": 0.2054873075026308, + "learning_rate": 2.6248357023271292e-05, + "loss": 0.9731, + "step": 1272 + }, + { + "epoch": 3.2557544757033248, + "grad_norm": 0.19487399212179374, + "learning_rate": 2.6181293135409628e-05, + "loss": 0.9727, + "step": 1273 + }, + { + "epoch": 3.258312020460358, + "grad_norm": 0.18326903320186239, + "learning_rate": 2.611427332691558e-05, + "loss": 0.9847, + "step": 1274 + }, + { + "epoch": 3.260869565217391, + "grad_norm": 0.18796016490886536, + "learning_rate": 2.6047297811571163e-05, + "loss": 0.9929, + "step": 1275 + }, + { + "epoch": 3.2634271099744243, + "grad_norm": 0.17566753019414313, + "learning_rate": 2.5980366803017063e-05, + "loss": 0.9486, + "step": 1276 + }, + { + "epoch": 3.265984654731458, + "grad_norm": 0.20427285085857036, + "learning_rate": 2.591348051475202e-05, + "loss": 1.0034, + "step": 1277 + }, + { + "epoch": 3.268542199488491, + "grad_norm": 0.1896070148851372, + "learning_rate": 2.5846639160132152e-05, + "loss": 0.9725, + "step": 1278 + }, + { + "epoch": 3.2710997442455243, + "grad_norm": 0.185712139910448, + "learning_rate": 2.5779842952370193e-05, + "loss": 1.0223, + "step": 1279 + }, + { + "epoch": 3.2736572890025575, + "grad_norm": 0.2064687121716206, + "learning_rate": 2.5713092104534926e-05, + "loss": 0.9905, + "step": 1280 + }, + { + "epoch": 3.2762148337595907, + "grad_norm": 0.18607952395957236, + "learning_rate": 2.5646386829550378e-05, + "loss": 0.9996, + "step": 1281 + }, + { + "epoch": 3.2787723785166243, + "grad_norm": 0.20062219284524274, + "learning_rate": 2.557972734019524e-05, + "loss": 0.9859, + "step": 1282 + }, + { + "epoch": 3.2813299232736575, + "grad_norm": 0.2154776394176252, + "learning_rate": 2.5513113849102168e-05, + "loss": 0.9847, + "step": 1283 + }, + { + "epoch": 3.2838874680306906, + "grad_norm": 0.18368848032395593, + "learning_rate": 2.5446546568757074e-05, + "loss": 0.9804, + "step": 1284 + }, + { + "epoch": 3.286445012787724, + "grad_norm": 0.20334568862048236, + "learning_rate": 2.5380025711498463e-05, + "loss": 0.9838, + "step": 1285 + }, + { + "epoch": 3.289002557544757, + "grad_norm": 0.18953720629273552, + "learning_rate": 2.5313551489516773e-05, + "loss": 0.9803, + "step": 1286 + }, + { + "epoch": 3.29156010230179, + "grad_norm": 0.18747324088248227, + "learning_rate": 2.5247124114853656e-05, + "loss": 0.972, + "step": 1287 + }, + { + "epoch": 3.2941176470588234, + "grad_norm": 0.18560605380576226, + "learning_rate": 2.518074379940136e-05, + "loss": 0.9585, + "step": 1288 + }, + { + "epoch": 3.296675191815857, + "grad_norm": 0.18595879737381238, + "learning_rate": 2.5114410754901992e-05, + "loss": 0.9584, + "step": 1289 + }, + { + "epoch": 3.29923273657289, + "grad_norm": 0.18911753423327268, + "learning_rate": 2.5048125192946912e-05, + "loss": 0.9567, + "step": 1290 + }, + { + "epoch": 3.3017902813299234, + "grad_norm": 0.1957348195686964, + "learning_rate": 2.4981887324975988e-05, + "loss": 0.969, + "step": 1291 + }, + { + "epoch": 3.3043478260869565, + "grad_norm": 0.19191922140197118, + "learning_rate": 2.4915697362276935e-05, + "loss": 0.9958, + "step": 1292 + }, + { + "epoch": 3.3069053708439897, + "grad_norm": 0.17974364713669186, + "learning_rate": 2.484955551598471e-05, + "loss": 0.9392, + "step": 1293 + }, + { + "epoch": 3.309462915601023, + "grad_norm": 0.20101316164789793, + "learning_rate": 2.4783461997080745e-05, + "loss": 0.9697, + "step": 1294 + }, + { + "epoch": 3.312020460358056, + "grad_norm": 0.1958479453724846, + "learning_rate": 2.471741701639234e-05, + "loss": 0.9814, + "step": 1295 + }, + { + "epoch": 3.3145780051150897, + "grad_norm": 0.19206314348892373, + "learning_rate": 2.4651420784591942e-05, + "loss": 0.9924, + "step": 1296 + }, + { + "epoch": 3.317135549872123, + "grad_norm": 0.20836822041907688, + "learning_rate": 2.4585473512196537e-05, + "loss": 0.9493, + "step": 1297 + }, + { + "epoch": 3.319693094629156, + "grad_norm": 0.18897201432176333, + "learning_rate": 2.4519575409566888e-05, + "loss": 0.9686, + "step": 1298 + }, + { + "epoch": 3.3222506393861893, + "grad_norm": 0.24239885455652932, + "learning_rate": 2.445372668690694e-05, + "loss": 0.9934, + "step": 1299 + }, + { + "epoch": 3.3248081841432224, + "grad_norm": 0.2116091122884991, + "learning_rate": 2.4387927554263136e-05, + "loss": 1.0145, + "step": 1300 + }, + { + "epoch": 3.3273657289002556, + "grad_norm": 0.22172159721754445, + "learning_rate": 2.432217822152371e-05, + "loss": 0.9585, + "step": 1301 + }, + { + "epoch": 3.329923273657289, + "grad_norm": 0.20040699370284384, + "learning_rate": 2.425647889841808e-05, + "loss": 0.9883, + "step": 1302 + }, + { + "epoch": 3.3324808184143224, + "grad_norm": 0.20978963710409357, + "learning_rate": 2.4190829794516093e-05, + "loss": 0.9681, + "step": 1303 + }, + { + "epoch": 3.3350383631713556, + "grad_norm": 0.2099270440664411, + "learning_rate": 2.412523111922743e-05, + "loss": 0.9839, + "step": 1304 + }, + { + "epoch": 3.337595907928389, + "grad_norm": 0.19154406605739202, + "learning_rate": 2.405968308180093e-05, + "loss": 0.9592, + "step": 1305 + }, + { + "epoch": 3.340153452685422, + "grad_norm": 0.2041595809244005, + "learning_rate": 2.399418589132386e-05, + "loss": 0.9928, + "step": 1306 + }, + { + "epoch": 3.342710997442455, + "grad_norm": 0.18133483534377134, + "learning_rate": 2.392873975672135e-05, + "loss": 0.9465, + "step": 1307 + }, + { + "epoch": 3.3452685421994883, + "grad_norm": 0.1932151711889084, + "learning_rate": 2.386334488675564e-05, + "loss": 0.9885, + "step": 1308 + }, + { + "epoch": 3.3478260869565215, + "grad_norm": 0.16981875008819527, + "learning_rate": 2.3798001490025417e-05, + "loss": 0.9736, + "step": 1309 + }, + { + "epoch": 3.350383631713555, + "grad_norm": 0.19797950201310727, + "learning_rate": 2.3732709774965238e-05, + "loss": 1.0126, + "step": 1310 + }, + { + "epoch": 3.3529411764705883, + "grad_norm": 0.18722532558513932, + "learning_rate": 2.3667469949844744e-05, + "loss": 0.9801, + "step": 1311 + }, + { + "epoch": 3.3554987212276215, + "grad_norm": 0.2168260181647949, + "learning_rate": 2.3602282222768117e-05, + "loss": 0.9735, + "step": 1312 + }, + { + "epoch": 3.3580562659846547, + "grad_norm": 0.19677536362501252, + "learning_rate": 2.353714680167331e-05, + "loss": 0.9759, + "step": 1313 + }, + { + "epoch": 3.360613810741688, + "grad_norm": 0.19959150858815647, + "learning_rate": 2.3472063894331423e-05, + "loss": 0.9853, + "step": 1314 + }, + { + "epoch": 3.363171355498721, + "grad_norm": 0.19210281532956738, + "learning_rate": 2.340703370834608e-05, + "loss": 0.956, + "step": 1315 + }, + { + "epoch": 3.3657289002557547, + "grad_norm": 0.1866860991602217, + "learning_rate": 2.3342056451152713e-05, + "loss": 0.9786, + "step": 1316 + }, + { + "epoch": 3.368286445012788, + "grad_norm": 0.18937893295508668, + "learning_rate": 2.327713233001793e-05, + "loss": 0.9896, + "step": 1317 + }, + { + "epoch": 3.370843989769821, + "grad_norm": 0.1771423252358527, + "learning_rate": 2.3212261552038827e-05, + "loss": 0.9856, + "step": 1318 + }, + { + "epoch": 3.373401534526854, + "grad_norm": 0.2235030616821708, + "learning_rate": 2.314744432414236e-05, + "loss": 0.9846, + "step": 1319 + }, + { + "epoch": 3.3759590792838874, + "grad_norm": 0.1700881379670313, + "learning_rate": 2.3082680853084663e-05, + "loss": 0.96, + "step": 1320 + }, + { + "epoch": 3.3785166240409206, + "grad_norm": 0.2032024302743889, + "learning_rate": 2.301797134545039e-05, + "loss": 0.9602, + "step": 1321 + }, + { + "epoch": 3.381074168797954, + "grad_norm": 0.18105668532410607, + "learning_rate": 2.2953316007652058e-05, + "loss": 0.9701, + "step": 1322 + }, + { + "epoch": 3.3836317135549874, + "grad_norm": 0.19642096592335476, + "learning_rate": 2.2888715045929427e-05, + "loss": 0.9633, + "step": 1323 + }, + { + "epoch": 3.3861892583120206, + "grad_norm": 0.179999924216972, + "learning_rate": 2.2824168666348775e-05, + "loss": 0.9771, + "step": 1324 + }, + { + "epoch": 3.3887468030690537, + "grad_norm": 0.18186351034819495, + "learning_rate": 2.2759677074802237e-05, + "loss": 0.9786, + "step": 1325 + }, + { + "epoch": 3.391304347826087, + "grad_norm": 0.19979550514976036, + "learning_rate": 2.269524047700726e-05, + "loss": 0.981, + "step": 1326 + }, + { + "epoch": 3.39386189258312, + "grad_norm": 0.20152802687590698, + "learning_rate": 2.2630859078505824e-05, + "loss": 0.9674, + "step": 1327 + }, + { + "epoch": 3.3964194373401533, + "grad_norm": 0.2070039830486493, + "learning_rate": 2.256653308466382e-05, + "loss": 0.9556, + "step": 1328 + }, + { + "epoch": 3.398976982097187, + "grad_norm": 0.2135037631517176, + "learning_rate": 2.2502262700670457e-05, + "loss": 1.0158, + "step": 1329 + }, + { + "epoch": 3.40153452685422, + "grad_norm": 0.20419304849123474, + "learning_rate": 2.243804813153754e-05, + "loss": 1.0003, + "step": 1330 + }, + { + "epoch": 3.4040920716112533, + "grad_norm": 0.18452079803418459, + "learning_rate": 2.2373889582098767e-05, + "loss": 0.9725, + "step": 1331 + }, + { + "epoch": 3.4066496163682864, + "grad_norm": 0.21656282916679845, + "learning_rate": 2.2309787257009243e-05, + "loss": 0.9907, + "step": 1332 + }, + { + "epoch": 3.4092071611253196, + "grad_norm": 0.17662188950107724, + "learning_rate": 2.224574136074467e-05, + "loss": 0.968, + "step": 1333 + }, + { + "epoch": 3.411764705882353, + "grad_norm": 0.22073774520396994, + "learning_rate": 2.218175209760077e-05, + "loss": 0.9957, + "step": 1334 + }, + { + "epoch": 3.414322250639386, + "grad_norm": 0.1952236355606058, + "learning_rate": 2.2117819671692596e-05, + "loss": 0.9802, + "step": 1335 + }, + { + "epoch": 3.4168797953964196, + "grad_norm": 0.2123002319054059, + "learning_rate": 2.2053944286953925e-05, + "loss": 0.9802, + "step": 1336 + }, + { + "epoch": 3.419437340153453, + "grad_norm": 0.17112900395184183, + "learning_rate": 2.199012614713655e-05, + "loss": 0.9756, + "step": 1337 + }, + { + "epoch": 3.421994884910486, + "grad_norm": 0.18105726277890447, + "learning_rate": 2.1926365455809684e-05, + "loss": 0.9646, + "step": 1338 + }, + { + "epoch": 3.424552429667519, + "grad_norm": 0.20192420986526344, + "learning_rate": 2.1862662416359316e-05, + "loss": 0.9712, + "step": 1339 + }, + { + "epoch": 3.4271099744245523, + "grad_norm": 0.2191925792086017, + "learning_rate": 2.1799017231987483e-05, + "loss": 1.0229, + "step": 1340 + }, + { + "epoch": 3.4296675191815855, + "grad_norm": 0.20301795120999852, + "learning_rate": 2.1735430105711695e-05, + "loss": 1.012, + "step": 1341 + }, + { + "epoch": 3.4322250639386187, + "grad_norm": 0.23205083833577947, + "learning_rate": 2.1671901240364272e-05, + "loss": 0.9772, + "step": 1342 + }, + { + "epoch": 3.4347826086956523, + "grad_norm": 0.22514218245005138, + "learning_rate": 2.1608430838591685e-05, + "loss": 0.9893, + "step": 1343 + }, + { + "epoch": 3.4373401534526855, + "grad_norm": 0.20385071246928765, + "learning_rate": 2.1545019102853927e-05, + "loss": 0.9717, + "step": 1344 + }, + { + "epoch": 3.4398976982097187, + "grad_norm": 0.22705339818003886, + "learning_rate": 2.148166623542382e-05, + "loss": 0.9616, + "step": 1345 + }, + { + "epoch": 3.442455242966752, + "grad_norm": 0.2084596120525907, + "learning_rate": 2.1418372438386484e-05, + "loss": 1.0011, + "step": 1346 + }, + { + "epoch": 3.445012787723785, + "grad_norm": 0.23886059421463335, + "learning_rate": 2.1355137913638515e-05, + "loss": 0.9772, + "step": 1347 + }, + { + "epoch": 3.4475703324808182, + "grad_norm": 0.18940104107641928, + "learning_rate": 2.1291962862887498e-05, + "loss": 0.9979, + "step": 1348 + }, + { + "epoch": 3.4501278772378514, + "grad_norm": 0.24592659749162, + "learning_rate": 2.1228847487651312e-05, + "loss": 1.0035, + "step": 1349 + }, + { + "epoch": 3.452685421994885, + "grad_norm": 0.16598425333535824, + "learning_rate": 2.116579198925747e-05, + "loss": 0.9917, + "step": 1350 + }, + { + "epoch": 3.455242966751918, + "grad_norm": 0.2205308902985363, + "learning_rate": 2.110279656884246e-05, + "loss": 0.9865, + "step": 1351 + }, + { + "epoch": 3.4578005115089514, + "grad_norm": 0.18936884708047907, + "learning_rate": 2.103986142735118e-05, + "loss": 0.9941, + "step": 1352 + }, + { + "epoch": 3.4603580562659846, + "grad_norm": 0.2113422925950507, + "learning_rate": 2.0976986765536202e-05, + "loss": 0.9584, + "step": 1353 + }, + { + "epoch": 3.4629156010230178, + "grad_norm": 0.1616728561002516, + "learning_rate": 2.0914172783957213e-05, + "loss": 0.9804, + "step": 1354 + }, + { + "epoch": 3.4654731457800514, + "grad_norm": 0.16257254424814638, + "learning_rate": 2.08514196829803e-05, + "loss": 0.9722, + "step": 1355 + }, + { + "epoch": 3.4680306905370846, + "grad_norm": 0.1834674640457499, + "learning_rate": 2.078872766277742e-05, + "loss": 0.9788, + "step": 1356 + }, + { + "epoch": 3.4705882352941178, + "grad_norm": 0.1685618446023564, + "learning_rate": 2.072609692332564e-05, + "loss": 0.9819, + "step": 1357 + }, + { + "epoch": 3.473145780051151, + "grad_norm": 0.17275999233879516, + "learning_rate": 2.0663527664406518e-05, + "loss": 0.9845, + "step": 1358 + }, + { + "epoch": 3.475703324808184, + "grad_norm": 0.17003829121965705, + "learning_rate": 2.0601020085605598e-05, + "loss": 1.0034, + "step": 1359 + }, + { + "epoch": 3.4782608695652173, + "grad_norm": 0.18552163039343017, + "learning_rate": 2.053857438631159e-05, + "loss": 0.992, + "step": 1360 + }, + { + "epoch": 3.4808184143222505, + "grad_norm": 0.1656628875654245, + "learning_rate": 2.0476190765715858e-05, + "loss": 0.9748, + "step": 1361 + }, + { + "epoch": 3.483375959079284, + "grad_norm": 0.18199082040271558, + "learning_rate": 2.041386942281171e-05, + "loss": 0.9809, + "step": 1362 + }, + { + "epoch": 3.4859335038363173, + "grad_norm": 0.16923865119691642, + "learning_rate": 2.0351610556393887e-05, + "loss": 0.9971, + "step": 1363 + }, + { + "epoch": 3.4884910485933505, + "grad_norm": 0.1628005115603011, + "learning_rate": 2.0289414365057716e-05, + "loss": 0.9669, + "step": 1364 + }, + { + "epoch": 3.4910485933503836, + "grad_norm": 0.16855632039381577, + "learning_rate": 2.022728104719867e-05, + "loss": 1.004, + "step": 1365 + }, + { + "epoch": 3.493606138107417, + "grad_norm": 0.16112536512929426, + "learning_rate": 2.0165210801011684e-05, + "loss": 1.0225, + "step": 1366 + }, + { + "epoch": 3.49616368286445, + "grad_norm": 0.17493102634964577, + "learning_rate": 2.010320382449048e-05, + "loss": 1.0037, + "step": 1367 + }, + { + "epoch": 3.498721227621483, + "grad_norm": 0.16471041553434848, + "learning_rate": 2.0041260315426933e-05, + "loss": 0.9444, + "step": 1368 + }, + { + "epoch": 3.501278772378517, + "grad_norm": 0.16903247119557196, + "learning_rate": 1.997938047141052e-05, + "loss": 0.9926, + "step": 1369 + }, + { + "epoch": 3.50383631713555, + "grad_norm": 0.16181490376970553, + "learning_rate": 1.9917564489827595e-05, + "loss": 0.9941, + "step": 1370 + }, + { + "epoch": 3.506393861892583, + "grad_norm": 0.16844937271379926, + "learning_rate": 1.9855812567860832e-05, + "loss": 0.9904, + "step": 1371 + }, + { + "epoch": 3.5089514066496164, + "grad_norm": 0.1955609613101622, + "learning_rate": 1.979412490248852e-05, + "loss": 1.0092, + "step": 1372 + }, + { + "epoch": 3.5115089514066495, + "grad_norm": 0.17584894207107374, + "learning_rate": 1.9732501690484054e-05, + "loss": 0.9686, + "step": 1373 + }, + { + "epoch": 3.5140664961636827, + "grad_norm": 0.1709859307794947, + "learning_rate": 1.9670943128415185e-05, + "loss": 0.9905, + "step": 1374 + }, + { + "epoch": 3.516624040920716, + "grad_norm": 0.18078474312646597, + "learning_rate": 1.9609449412643403e-05, + "loss": 0.9622, + "step": 1375 + }, + { + "epoch": 3.5191815856777495, + "grad_norm": 0.1939597215141259, + "learning_rate": 1.9548020739323443e-05, + "loss": 0.9684, + "step": 1376 + }, + { + "epoch": 3.5217391304347827, + "grad_norm": 0.16884122381730668, + "learning_rate": 1.94866573044025e-05, + "loss": 0.9961, + "step": 1377 + }, + { + "epoch": 3.524296675191816, + "grad_norm": 0.17059684327689034, + "learning_rate": 1.9425359303619685e-05, + "loss": 0.9624, + "step": 1378 + }, + { + "epoch": 3.526854219948849, + "grad_norm": 0.17751695012672875, + "learning_rate": 1.9364126932505383e-05, + "loss": 1.0249, + "step": 1379 + }, + { + "epoch": 3.5294117647058822, + "grad_norm": 0.15312460755115656, + "learning_rate": 1.9302960386380637e-05, + "loss": 0.9665, + "step": 1380 + }, + { + "epoch": 3.531969309462916, + "grad_norm": 0.18097753824526422, + "learning_rate": 1.924185986035652e-05, + "loss": 1.021, + "step": 1381 + }, + { + "epoch": 3.5345268542199486, + "grad_norm": 0.16847855508114354, + "learning_rate": 1.9180825549333483e-05, + "loss": 0.9914, + "step": 1382 + }, + { + "epoch": 3.5370843989769822, + "grad_norm": 0.17393073349900104, + "learning_rate": 1.9119857648000827e-05, + "loss": 0.9762, + "step": 1383 + }, + { + "epoch": 3.5396419437340154, + "grad_norm": 0.17339771912019677, + "learning_rate": 1.9058956350835952e-05, + "loss": 1.0047, + "step": 1384 + }, + { + "epoch": 3.5421994884910486, + "grad_norm": 0.18250020868636335, + "learning_rate": 1.8998121852103843e-05, + "loss": 0.9844, + "step": 1385 + }, + { + "epoch": 3.544757033248082, + "grad_norm": 0.168632483834106, + "learning_rate": 1.8937354345856383e-05, + "loss": 1.0024, + "step": 1386 + }, + { + "epoch": 3.547314578005115, + "grad_norm": 0.16967759514899067, + "learning_rate": 1.8876654025931767e-05, + "loss": 0.9568, + "step": 1387 + }, + { + "epoch": 3.5498721227621486, + "grad_norm": 0.16759243143730015, + "learning_rate": 1.881602108595389e-05, + "loss": 0.9762, + "step": 1388 + }, + { + "epoch": 3.5524296675191813, + "grad_norm": 0.15832481897103784, + "learning_rate": 1.875545571933169e-05, + "loss": 1.0031, + "step": 1389 + }, + { + "epoch": 3.554987212276215, + "grad_norm": 0.1811084645197821, + "learning_rate": 1.8694958119258612e-05, + "loss": 1.0111, + "step": 1390 + }, + { + "epoch": 3.557544757033248, + "grad_norm": 0.1648706646392026, + "learning_rate": 1.863452847871186e-05, + "loss": 0.948, + "step": 1391 + }, + { + "epoch": 3.5601023017902813, + "grad_norm": 0.16239454243103976, + "learning_rate": 1.8574166990451898e-05, + "loss": 0.9727, + "step": 1392 + }, + { + "epoch": 3.5626598465473145, + "grad_norm": 0.17942259610790612, + "learning_rate": 1.8513873847021823e-05, + "loss": 0.9908, + "step": 1393 + }, + { + "epoch": 3.5652173913043477, + "grad_norm": 0.17137935333375814, + "learning_rate": 1.8453649240746675e-05, + "loss": 0.9932, + "step": 1394 + }, + { + "epoch": 3.5677749360613813, + "grad_norm": 0.17038531828743989, + "learning_rate": 1.8393493363732905e-05, + "loss": 0.9989, + "step": 1395 + }, + { + "epoch": 3.5703324808184145, + "grad_norm": 0.16665000610754513, + "learning_rate": 1.8333406407867712e-05, + "loss": 0.9777, + "step": 1396 + }, + { + "epoch": 3.5728900255754477, + "grad_norm": 0.1578795644839414, + "learning_rate": 1.8273388564818443e-05, + "loss": 0.9587, + "step": 1397 + }, + { + "epoch": 3.575447570332481, + "grad_norm": 0.17474995545729136, + "learning_rate": 1.8213440026032013e-05, + "loss": 0.9979, + "step": 1398 + }, + { + "epoch": 3.578005115089514, + "grad_norm": 0.15791533543568376, + "learning_rate": 1.815356098273422e-05, + "loss": 0.9993, + "step": 1399 + }, + { + "epoch": 3.580562659846547, + "grad_norm": 0.1521732452347076, + "learning_rate": 1.8093751625929252e-05, + "loss": 0.9605, + "step": 1400 + }, + { + "epoch": 3.5831202046035804, + "grad_norm": 0.1592417054700969, + "learning_rate": 1.8034012146398972e-05, + "loss": 0.9697, + "step": 1401 + }, + { + "epoch": 3.585677749360614, + "grad_norm": 0.15829811384256975, + "learning_rate": 1.79743427347023e-05, + "loss": 0.9891, + "step": 1402 + }, + { + "epoch": 3.588235294117647, + "grad_norm": 0.16118804739509424, + "learning_rate": 1.791474358117474e-05, + "loss": 0.9841, + "step": 1403 + }, + { + "epoch": 3.5907928388746804, + "grad_norm": 0.16191134259771192, + "learning_rate": 1.785521487592762e-05, + "loss": 0.9899, + "step": 1404 + }, + { + "epoch": 3.5933503836317136, + "grad_norm": 0.1526578683810162, + "learning_rate": 1.7795756808847583e-05, + "loss": 0.9807, + "step": 1405 + }, + { + "epoch": 3.5959079283887467, + "grad_norm": 0.1700342189970984, + "learning_rate": 1.773636956959591e-05, + "loss": 0.9837, + "step": 1406 + }, + { + "epoch": 3.59846547314578, + "grad_norm": 0.17079530240041818, + "learning_rate": 1.7677053347608026e-05, + "loss": 0.9929, + "step": 1407 + }, + { + "epoch": 3.601023017902813, + "grad_norm": 0.16946592118696485, + "learning_rate": 1.7617808332092733e-05, + "loss": 0.9937, + "step": 1408 + }, + { + "epoch": 3.6035805626598467, + "grad_norm": 0.16721155279112196, + "learning_rate": 1.7558634712031733e-05, + "loss": 0.9716, + "step": 1409 + }, + { + "epoch": 3.60613810741688, + "grad_norm": 0.15812399028035554, + "learning_rate": 1.7499532676179027e-05, + "loss": 0.9463, + "step": 1410 + }, + { + "epoch": 3.608695652173913, + "grad_norm": 0.18933737997751232, + "learning_rate": 1.744050241306023e-05, + "loss": 0.9873, + "step": 1411 + }, + { + "epoch": 3.6112531969309463, + "grad_norm": 0.1513273347708644, + "learning_rate": 1.738154411097202e-05, + "loss": 1.0049, + "step": 1412 + }, + { + "epoch": 3.6138107416879794, + "grad_norm": 0.16160857563728534, + "learning_rate": 1.7322657957981543e-05, + "loss": 0.9629, + "step": 1413 + }, + { + "epoch": 3.6163682864450126, + "grad_norm": 0.1575626021393853, + "learning_rate": 1.726384414192578e-05, + "loss": 0.9817, + "step": 1414 + }, + { + "epoch": 3.618925831202046, + "grad_norm": 0.14742706913576417, + "learning_rate": 1.7205102850411e-05, + "loss": 0.9608, + "step": 1415 + }, + { + "epoch": 3.6214833759590794, + "grad_norm": 0.15385433784590577, + "learning_rate": 1.71464342708121e-05, + "loss": 0.9834, + "step": 1416 + }, + { + "epoch": 3.6240409207161126, + "grad_norm": 0.15976764455036044, + "learning_rate": 1.7087838590272067e-05, + "loss": 0.9599, + "step": 1417 + }, + { + "epoch": 3.626598465473146, + "grad_norm": 0.17485245098862065, + "learning_rate": 1.7029315995701355e-05, + "loss": 0.9952, + "step": 1418 + }, + { + "epoch": 3.629156010230179, + "grad_norm": 0.16888749566043196, + "learning_rate": 1.6970866673777217e-05, + "loss": 0.9879, + "step": 1419 + }, + { + "epoch": 3.631713554987212, + "grad_norm": 0.17530855499982936, + "learning_rate": 1.6912490810943273e-05, + "loss": 0.9886, + "step": 1420 + }, + { + "epoch": 3.634271099744246, + "grad_norm": 0.17867965330900965, + "learning_rate": 1.685418859340877e-05, + "loss": 1.004, + "step": 1421 + }, + { + "epoch": 3.6368286445012785, + "grad_norm": 0.16102945276705097, + "learning_rate": 1.6795960207148047e-05, + "loss": 0.9585, + "step": 1422 + }, + { + "epoch": 3.639386189258312, + "grad_norm": 0.1740372821884824, + "learning_rate": 1.673780583789993e-05, + "loss": 0.9864, + "step": 1423 + }, + { + "epoch": 3.6419437340153453, + "grad_norm": 0.1599072680335884, + "learning_rate": 1.6679725671167153e-05, + "loss": 0.9778, + "step": 1424 + }, + { + "epoch": 3.6445012787723785, + "grad_norm": 0.17915224537175092, + "learning_rate": 1.662171989221574e-05, + "loss": 1.0061, + "step": 1425 + }, + { + "epoch": 3.6470588235294117, + "grad_norm": 0.15032289661695453, + "learning_rate": 1.6563788686074437e-05, + "loss": 0.9656, + "step": 1426 + }, + { + "epoch": 3.649616368286445, + "grad_norm": 0.16252679935892114, + "learning_rate": 1.6505932237534135e-05, + "loss": 0.9766, + "step": 1427 + }, + { + "epoch": 3.6521739130434785, + "grad_norm": 0.16372638209261026, + "learning_rate": 1.6448150731147245e-05, + "loss": 0.971, + "step": 1428 + }, + { + "epoch": 3.6547314578005117, + "grad_norm": 0.15286966641481936, + "learning_rate": 1.6390444351227116e-05, + "loss": 0.9615, + "step": 1429 + }, + { + "epoch": 3.657289002557545, + "grad_norm": 0.17422830046285157, + "learning_rate": 1.6332813281847468e-05, + "loss": 0.9918, + "step": 1430 + }, + { + "epoch": 3.659846547314578, + "grad_norm": 0.1536114632035703, + "learning_rate": 1.6275257706841787e-05, + "loss": 0.9853, + "step": 1431 + }, + { + "epoch": 3.662404092071611, + "grad_norm": 0.15498834822619906, + "learning_rate": 1.6217777809802752e-05, + "loss": 0.9889, + "step": 1432 + }, + { + "epoch": 3.6649616368286444, + "grad_norm": 0.1525756152661097, + "learning_rate": 1.6160373774081616e-05, + "loss": 0.9645, + "step": 1433 + }, + { + "epoch": 3.6675191815856776, + "grad_norm": 0.163535134810249, + "learning_rate": 1.610304578278772e-05, + "loss": 0.9704, + "step": 1434 + }, + { + "epoch": 3.670076726342711, + "grad_norm": 0.1439706547451718, + "learning_rate": 1.604579401878773e-05, + "loss": 0.9661, + "step": 1435 + }, + { + "epoch": 3.6726342710997444, + "grad_norm": 0.18467413525193094, + "learning_rate": 1.5988618664705223e-05, + "loss": 0.9958, + "step": 1436 + }, + { + "epoch": 3.6751918158567776, + "grad_norm": 0.14597437711546657, + "learning_rate": 1.593151990292006e-05, + "loss": 0.9723, + "step": 1437 + }, + { + "epoch": 3.6777493606138107, + "grad_norm": 0.19323217406252066, + "learning_rate": 1.587449791556774e-05, + "loss": 1.0298, + "step": 1438 + }, + { + "epoch": 3.680306905370844, + "grad_norm": 0.15130412791127118, + "learning_rate": 1.5817552884538883e-05, + "loss": 0.9652, + "step": 1439 + }, + { + "epoch": 3.682864450127877, + "grad_norm": 0.18744157138750245, + "learning_rate": 1.5760684991478634e-05, + "loss": 0.9658, + "step": 1440 + }, + { + "epoch": 3.6854219948849103, + "grad_norm": 0.19424967162069548, + "learning_rate": 1.5703894417786066e-05, + "loss": 0.9659, + "step": 1441 + }, + { + "epoch": 3.687979539641944, + "grad_norm": 0.16583222812083506, + "learning_rate": 1.564718134461364e-05, + "loss": 0.9634, + "step": 1442 + }, + { + "epoch": 3.690537084398977, + "grad_norm": 0.20374960351527888, + "learning_rate": 1.5590545952866565e-05, + "loss": 0.9473, + "step": 1443 + }, + { + "epoch": 3.6930946291560103, + "grad_norm": 0.16010311366530502, + "learning_rate": 1.553398842320232e-05, + "loss": 0.974, + "step": 1444 + }, + { + "epoch": 3.6956521739130435, + "grad_norm": 0.18480335391304883, + "learning_rate": 1.5477508936029978e-05, + "loss": 0.9945, + "step": 1445 + }, + { + "epoch": 3.6982097186700766, + "grad_norm": 0.1640514417582284, + "learning_rate": 1.542110767150963e-05, + "loss": 0.9586, + "step": 1446 + }, + { + "epoch": 3.70076726342711, + "grad_norm": 0.15448350271744538, + "learning_rate": 1.5364784809551938e-05, + "loss": 0.9853, + "step": 1447 + }, + { + "epoch": 3.703324808184143, + "grad_norm": 0.17847282007082363, + "learning_rate": 1.5308540529817405e-05, + "loss": 1.0062, + "step": 1448 + }, + { + "epoch": 3.7058823529411766, + "grad_norm": 0.15577197474274293, + "learning_rate": 1.5252375011715894e-05, + "loss": 0.9602, + "step": 1449 + }, + { + "epoch": 3.70843989769821, + "grad_norm": 0.18403129017994174, + "learning_rate": 1.5196288434406045e-05, + "loss": 0.9729, + "step": 1450 + }, + { + "epoch": 3.710997442455243, + "grad_norm": 0.16034660354421298, + "learning_rate": 1.514028097679468e-05, + "loss": 0.9534, + "step": 1451 + }, + { + "epoch": 3.713554987212276, + "grad_norm": 0.14891427160648704, + "learning_rate": 1.5084352817536205e-05, + "loss": 0.9829, + "step": 1452 + }, + { + "epoch": 3.7161125319693094, + "grad_norm": 0.1660110312736295, + "learning_rate": 1.5028504135032113e-05, + "loss": 0.9666, + "step": 1453 + }, + { + "epoch": 3.718670076726343, + "grad_norm": 0.14924967707687345, + "learning_rate": 1.4972735107430407e-05, + "loss": 1.0091, + "step": 1454 + }, + { + "epoch": 3.7212276214833757, + "grad_norm": 0.17536986017606493, + "learning_rate": 1.4917045912624954e-05, + "loss": 0.9656, + "step": 1455 + }, + { + "epoch": 3.7237851662404093, + "grad_norm": 0.1545050307942204, + "learning_rate": 1.4861436728254992e-05, + "loss": 0.9626, + "step": 1456 + }, + { + "epoch": 3.7263427109974425, + "grad_norm": 0.16567326268668936, + "learning_rate": 1.480590773170453e-05, + "loss": 1.0058, + "step": 1457 + }, + { + "epoch": 3.7289002557544757, + "grad_norm": 0.159986133602208, + "learning_rate": 1.47504591001018e-05, + "loss": 0.965, + "step": 1458 + }, + { + "epoch": 3.731457800511509, + "grad_norm": 0.1636516639657428, + "learning_rate": 1.469509101031866e-05, + "loss": 0.9992, + "step": 1459 + }, + { + "epoch": 3.734015345268542, + "grad_norm": 0.15877712329888719, + "learning_rate": 1.4639803638970098e-05, + "loss": 0.9818, + "step": 1460 + }, + { + "epoch": 3.7365728900255757, + "grad_norm": 0.15177768229747693, + "learning_rate": 1.4584597162413592e-05, + "loss": 0.9596, + "step": 1461 + }, + { + "epoch": 3.7391304347826084, + "grad_norm": 0.16539728281480964, + "learning_rate": 1.4529471756748579e-05, + "loss": 0.9654, + "step": 1462 + }, + { + "epoch": 3.741687979539642, + "grad_norm": 0.15013106984230754, + "learning_rate": 1.4474427597815898e-05, + "loss": 0.9749, + "step": 1463 + }, + { + "epoch": 3.7442455242966752, + "grad_norm": 0.1607063503736631, + "learning_rate": 1.4419464861197225e-05, + "loss": 0.9936, + "step": 1464 + }, + { + "epoch": 3.7468030690537084, + "grad_norm": 0.15822776373381015, + "learning_rate": 1.4364583722214521e-05, + "loss": 0.9816, + "step": 1465 + }, + { + "epoch": 3.7493606138107416, + "grad_norm": 0.16495768276016773, + "learning_rate": 1.4309784355929429e-05, + "loss": 0.9784, + "step": 1466 + }, + { + "epoch": 3.7519181585677748, + "grad_norm": 0.16464946701490385, + "learning_rate": 1.4255066937142816e-05, + "loss": 0.9801, + "step": 1467 + }, + { + "epoch": 3.7544757033248084, + "grad_norm": 0.16145775604523005, + "learning_rate": 1.4200431640394103e-05, + "loss": 0.9749, + "step": 1468 + }, + { + "epoch": 3.7570332480818416, + "grad_norm": 0.16615520065079115, + "learning_rate": 1.4145878639960735e-05, + "loss": 0.9978, + "step": 1469 + }, + { + "epoch": 3.7595907928388748, + "grad_norm": 0.13971867189011447, + "learning_rate": 1.4091408109857706e-05, + "loss": 0.9557, + "step": 1470 + }, + { + "epoch": 3.762148337595908, + "grad_norm": 0.14959140214161223, + "learning_rate": 1.4037020223836897e-05, + "loss": 0.9881, + "step": 1471 + }, + { + "epoch": 3.764705882352941, + "grad_norm": 0.1682587608263464, + "learning_rate": 1.3982715155386592e-05, + "loss": 0.982, + "step": 1472 + }, + { + "epoch": 3.7672634271099743, + "grad_norm": 0.15458557583904273, + "learning_rate": 1.3928493077730889e-05, + "loss": 0.9652, + "step": 1473 + }, + { + "epoch": 3.7698209718670075, + "grad_norm": 0.1788403999805433, + "learning_rate": 1.3874354163829158e-05, + "loss": 0.9742, + "step": 1474 + }, + { + "epoch": 3.772378516624041, + "grad_norm": 0.15175695136886533, + "learning_rate": 1.3820298586375502e-05, + "loss": 0.9873, + "step": 1475 + }, + { + "epoch": 3.7749360613810743, + "grad_norm": 0.16076876387469338, + "learning_rate": 1.3766326517798168e-05, + "loss": 0.9799, + "step": 1476 + }, + { + "epoch": 3.7774936061381075, + "grad_norm": 0.13980858682129033, + "learning_rate": 1.3712438130259073e-05, + "loss": 0.9489, + "step": 1477 + }, + { + "epoch": 3.7800511508951407, + "grad_norm": 0.15469469304691397, + "learning_rate": 1.3658633595653159e-05, + "loss": 0.9741, + "step": 1478 + }, + { + "epoch": 3.782608695652174, + "grad_norm": 0.16083095141615023, + "learning_rate": 1.3604913085607914e-05, + "loss": 0.959, + "step": 1479 + }, + { + "epoch": 3.785166240409207, + "grad_norm": 0.1527879996311441, + "learning_rate": 1.3551276771482788e-05, + "loss": 0.9692, + "step": 1480 + }, + { + "epoch": 3.78772378516624, + "grad_norm": 0.1501151428939216, + "learning_rate": 1.3497724824368672e-05, + "loss": 0.9696, + "step": 1481 + }, + { + "epoch": 3.790281329923274, + "grad_norm": 0.15367672935484297, + "learning_rate": 1.3444257415087339e-05, + "loss": 0.9879, + "step": 1482 + }, + { + "epoch": 3.792838874680307, + "grad_norm": 0.16397124262551233, + "learning_rate": 1.3390874714190876e-05, + "loss": 1.0394, + "step": 1483 + }, + { + "epoch": 3.79539641943734, + "grad_norm": 0.1590628803184472, + "learning_rate": 1.3337576891961228e-05, + "loss": 0.9854, + "step": 1484 + }, + { + "epoch": 3.7979539641943734, + "grad_norm": 0.18225039937660117, + "learning_rate": 1.3284364118409512e-05, + "loss": 1.0181, + "step": 1485 + }, + { + "epoch": 3.8005115089514065, + "grad_norm": 0.15539862711036953, + "learning_rate": 1.3231236563275593e-05, + "loss": 1.0083, + "step": 1486 + }, + { + "epoch": 3.80306905370844, + "grad_norm": 0.1702708590595948, + "learning_rate": 1.317819439602752e-05, + "loss": 0.993, + "step": 1487 + }, + { + "epoch": 3.805626598465473, + "grad_norm": 0.17494004955408202, + "learning_rate": 1.3125237785860945e-05, + "loss": 0.9749, + "step": 1488 + }, + { + "epoch": 3.8081841432225065, + "grad_norm": 0.1525516290905405, + "learning_rate": 1.3072366901698606e-05, + "loss": 0.9819, + "step": 1489 + }, + { + "epoch": 3.8107416879795397, + "grad_norm": 0.16157170912735652, + "learning_rate": 1.3019581912189794e-05, + "loss": 0.9852, + "step": 1490 + }, + { + "epoch": 3.813299232736573, + "grad_norm": 0.17274917566336145, + "learning_rate": 1.2966882985709824e-05, + "loss": 0.9839, + "step": 1491 + }, + { + "epoch": 3.815856777493606, + "grad_norm": 0.17196613842795713, + "learning_rate": 1.291427029035946e-05, + "loss": 0.9797, + "step": 1492 + }, + { + "epoch": 3.8184143222506393, + "grad_norm": 0.17260762571717933, + "learning_rate": 1.2861743993964409e-05, + "loss": 0.9868, + "step": 1493 + }, + { + "epoch": 3.820971867007673, + "grad_norm": 0.17169441039113015, + "learning_rate": 1.2809304264074807e-05, + "loss": 0.9641, + "step": 1494 + }, + { + "epoch": 3.8235294117647056, + "grad_norm": 0.15100733833702012, + "learning_rate": 1.2756951267964644e-05, + "loss": 1.0115, + "step": 1495 + }, + { + "epoch": 3.8260869565217392, + "grad_norm": 0.16076257621863999, + "learning_rate": 1.2704685172631193e-05, + "loss": 0.9817, + "step": 1496 + }, + { + "epoch": 3.8286445012787724, + "grad_norm": 0.1574213450598427, + "learning_rate": 1.2652506144794612e-05, + "loss": 0.9941, + "step": 1497 + }, + { + "epoch": 3.8312020460358056, + "grad_norm": 0.14534492595767098, + "learning_rate": 1.2600414350897272e-05, + "loss": 0.9637, + "step": 1498 + }, + { + "epoch": 3.833759590792839, + "grad_norm": 0.14698706161648437, + "learning_rate": 1.2548409957103296e-05, + "loss": 0.9718, + "step": 1499 + }, + { + "epoch": 3.836317135549872, + "grad_norm": 0.15605018007996246, + "learning_rate": 1.2496493129298011e-05, + "loss": 0.9997, + "step": 1500 + }, + { + "epoch": 3.8388746803069056, + "grad_norm": 0.1477242511131571, + "learning_rate": 1.2444664033087466e-05, + "loss": 0.9797, + "step": 1501 + }, + { + "epoch": 3.8414322250639388, + "grad_norm": 0.13847001744552098, + "learning_rate": 1.2392922833797783e-05, + "loss": 0.9881, + "step": 1502 + }, + { + "epoch": 3.843989769820972, + "grad_norm": 0.14462911397711195, + "learning_rate": 1.2341269696474747e-05, + "loss": 0.9535, + "step": 1503 + }, + { + "epoch": 3.846547314578005, + "grad_norm": 0.15004821476381025, + "learning_rate": 1.2289704785883272e-05, + "loss": 0.9979, + "step": 1504 + }, + { + "epoch": 3.8491048593350383, + "grad_norm": 0.14681550872608987, + "learning_rate": 1.2238228266506798e-05, + "loss": 0.9988, + "step": 1505 + }, + { + "epoch": 3.8516624040920715, + "grad_norm": 0.13903328529411774, + "learning_rate": 1.2186840302546826e-05, + "loss": 0.9756, + "step": 1506 + }, + { + "epoch": 3.8542199488491047, + "grad_norm": 0.1450706267912665, + "learning_rate": 1.213554105792238e-05, + "loss": 0.9945, + "step": 1507 + }, + { + "epoch": 3.8567774936061383, + "grad_norm": 0.14188979334960153, + "learning_rate": 1.2084330696269491e-05, + "loss": 0.9766, + "step": 1508 + }, + { + "epoch": 3.8593350383631715, + "grad_norm": 0.1504358702702587, + "learning_rate": 1.2033209380940654e-05, + "loss": 1.004, + "step": 1509 + }, + { + "epoch": 3.8618925831202047, + "grad_norm": 0.14446185909795273, + "learning_rate": 1.198217727500432e-05, + "loss": 0.982, + "step": 1510 + }, + { + "epoch": 3.864450127877238, + "grad_norm": 0.14918074072958132, + "learning_rate": 1.1931234541244416e-05, + "loss": 0.9572, + "step": 1511 + }, + { + "epoch": 3.867007672634271, + "grad_norm": 0.1476126427894347, + "learning_rate": 1.1880381342159754e-05, + "loss": 0.9429, + "step": 1512 + }, + { + "epoch": 3.869565217391304, + "grad_norm": 0.14658231627960866, + "learning_rate": 1.1829617839963508e-05, + "loss": 0.9706, + "step": 1513 + }, + { + "epoch": 3.8721227621483374, + "grad_norm": 0.1463434207088499, + "learning_rate": 1.1778944196582814e-05, + "loss": 0.9799, + "step": 1514 + }, + { + "epoch": 3.874680306905371, + "grad_norm": 0.14270360583380973, + "learning_rate": 1.172836057365812e-05, + "loss": 0.9826, + "step": 1515 + }, + { + "epoch": 3.877237851662404, + "grad_norm": 0.15081164075513045, + "learning_rate": 1.1677867132542735e-05, + "loss": 0.993, + "step": 1516 + }, + { + "epoch": 3.8797953964194374, + "grad_norm": 0.14549864989116654, + "learning_rate": 1.162746403430231e-05, + "loss": 1.0152, + "step": 1517 + }, + { + "epoch": 3.8823529411764706, + "grad_norm": 0.1436310825646862, + "learning_rate": 1.157715143971431e-05, + "loss": 0.9881, + "step": 1518 + }, + { + "epoch": 3.8849104859335037, + "grad_norm": 0.14506305658975452, + "learning_rate": 1.1526929509267513e-05, + "loss": 1.0059, + "step": 1519 + }, + { + "epoch": 3.887468030690537, + "grad_norm": 0.13957687924429218, + "learning_rate": 1.1476798403161475e-05, + "loss": 0.9961, + "step": 1520 + }, + { + "epoch": 3.89002557544757, + "grad_norm": 0.15982963311960688, + "learning_rate": 1.1426758281306087e-05, + "loss": 0.9726, + "step": 1521 + }, + { + "epoch": 3.8925831202046037, + "grad_norm": 0.14440211669907038, + "learning_rate": 1.1376809303320964e-05, + "loss": 0.943, + "step": 1522 + }, + { + "epoch": 3.895140664961637, + "grad_norm": 0.15463895891888774, + "learning_rate": 1.1326951628535006e-05, + "loss": 0.9789, + "step": 1523 + }, + { + "epoch": 3.89769820971867, + "grad_norm": 0.14858430492843056, + "learning_rate": 1.1277185415985871e-05, + "loss": 0.9748, + "step": 1524 + }, + { + "epoch": 3.9002557544757033, + "grad_norm": 0.15150163420283375, + "learning_rate": 1.1227510824419468e-05, + "loss": 0.9964, + "step": 1525 + }, + { + "epoch": 3.9028132992327365, + "grad_norm": 0.16315553122177956, + "learning_rate": 1.1177928012289455e-05, + "loss": 0.9832, + "step": 1526 + }, + { + "epoch": 3.90537084398977, + "grad_norm": 0.13639222004794507, + "learning_rate": 1.1128437137756705e-05, + "loss": 0.9744, + "step": 1527 + }, + { + "epoch": 3.907928388746803, + "grad_norm": 0.15405033084464928, + "learning_rate": 1.1079038358688882e-05, + "loss": 0.9882, + "step": 1528 + }, + { + "epoch": 3.9104859335038364, + "grad_norm": 0.1691974950328252, + "learning_rate": 1.1029731832659811e-05, + "loss": 0.9538, + "step": 1529 + }, + { + "epoch": 3.9130434782608696, + "grad_norm": 0.1378352496795574, + "learning_rate": 1.0980517716949075e-05, + "loss": 0.9822, + "step": 1530 + }, + { + "epoch": 3.915601023017903, + "grad_norm": 0.14209307859417014, + "learning_rate": 1.0931396168541509e-05, + "loss": 0.9734, + "step": 1531 + }, + { + "epoch": 3.918158567774936, + "grad_norm": 0.160326538775092, + "learning_rate": 1.088236734412664e-05, + "loss": 0.9877, + "step": 1532 + }, + { + "epoch": 3.920716112531969, + "grad_norm": 0.14428090720467784, + "learning_rate": 1.0833431400098227e-05, + "loss": 0.9846, + "step": 1533 + }, + { + "epoch": 3.923273657289003, + "grad_norm": 0.1465209742252091, + "learning_rate": 1.0784588492553768e-05, + "loss": 0.9964, + "step": 1534 + }, + { + "epoch": 3.9258312020460355, + "grad_norm": 0.16030862232460416, + "learning_rate": 1.0735838777293979e-05, + "loss": 0.9858, + "step": 1535 + }, + { + "epoch": 3.928388746803069, + "grad_norm": 0.13686581434508327, + "learning_rate": 1.0687182409822311e-05, + "loss": 0.9716, + "step": 1536 + }, + { + "epoch": 3.9309462915601023, + "grad_norm": 0.1625327040079931, + "learning_rate": 1.0638619545344447e-05, + "loss": 0.9699, + "step": 1537 + }, + { + "epoch": 3.9335038363171355, + "grad_norm": 0.1525199965864014, + "learning_rate": 1.0590150338767828e-05, + "loss": 0.9877, + "step": 1538 + }, + { + "epoch": 3.9360613810741687, + "grad_norm": 0.13380670697300262, + "learning_rate": 1.054177494470114e-05, + "loss": 0.97, + "step": 1539 + }, + { + "epoch": 3.938618925831202, + "grad_norm": 0.17163164533417571, + "learning_rate": 1.0493493517453777e-05, + "loss": 0.98, + "step": 1540 + }, + { + "epoch": 3.9411764705882355, + "grad_norm": 0.14388648975792426, + "learning_rate": 1.044530621103546e-05, + "loss": 0.9714, + "step": 1541 + }, + { + "epoch": 3.9437340153452687, + "grad_norm": 0.15724845081283767, + "learning_rate": 1.0397213179155647e-05, + "loss": 0.9482, + "step": 1542 + }, + { + "epoch": 3.946291560102302, + "grad_norm": 0.14205243895810854, + "learning_rate": 1.034921457522307e-05, + "loss": 0.9727, + "step": 1543 + }, + { + "epoch": 3.948849104859335, + "grad_norm": 0.14157720210833158, + "learning_rate": 1.0301310552345258e-05, + "loss": 0.9548, + "step": 1544 + }, + { + "epoch": 3.9514066496163682, + "grad_norm": 0.1457113622126631, + "learning_rate": 1.0253501263328075e-05, + "loss": 0.9902, + "step": 1545 + }, + { + "epoch": 3.9539641943734014, + "grad_norm": 0.16390432154458898, + "learning_rate": 1.020578686067514e-05, + "loss": 0.9627, + "step": 1546 + }, + { + "epoch": 3.9565217391304346, + "grad_norm": 0.141405745046886, + "learning_rate": 1.0158167496587428e-05, + "loss": 0.9909, + "step": 1547 + }, + { + "epoch": 3.959079283887468, + "grad_norm": 0.14806685880079012, + "learning_rate": 1.0110643322962779e-05, + "loss": 0.9693, + "step": 1548 + }, + { + "epoch": 3.9616368286445014, + "grad_norm": 0.15773747192555593, + "learning_rate": 1.006321449139537e-05, + "loss": 1.0323, + "step": 1549 + }, + { + "epoch": 3.9641943734015346, + "grad_norm": 0.140456219803408, + "learning_rate": 1.0015881153175244e-05, + "loss": 1.0, + "step": 1550 + }, + { + "epoch": 3.9667519181585678, + "grad_norm": 0.15067874532052797, + "learning_rate": 9.968643459287856e-06, + "loss": 0.9892, + "step": 1551 + }, + { + "epoch": 3.969309462915601, + "grad_norm": 0.14479984414191183, + "learning_rate": 9.921501560413555e-06, + "loss": 0.9901, + "step": 1552 + }, + { + "epoch": 3.971867007672634, + "grad_norm": 0.15329653150211797, + "learning_rate": 9.874455606927134e-06, + "loss": 0.9889, + "step": 1553 + }, + { + "epoch": 3.9744245524296673, + "grad_norm": 0.15721815567916902, + "learning_rate": 9.827505748897308e-06, + "loss": 0.9767, + "step": 1554 + }, + { + "epoch": 3.976982097186701, + "grad_norm": 0.13746284270048956, + "learning_rate": 9.780652136086322e-06, + "loss": 0.9987, + "step": 1555 + }, + { + "epoch": 3.979539641943734, + "grad_norm": 0.1445235042786382, + "learning_rate": 9.733894917949365e-06, + "loss": 1.0099, + "step": 1556 + }, + { + "epoch": 3.9820971867007673, + "grad_norm": 0.15925854158907532, + "learning_rate": 9.687234243634127e-06, + "loss": 0.9791, + "step": 1557 + }, + { + "epoch": 3.9846547314578005, + "grad_norm": 0.1469893002397056, + "learning_rate": 9.640670261980403e-06, + "loss": 0.9931, + "step": 1558 + }, + { + "epoch": 3.9872122762148337, + "grad_norm": 0.14758630633936268, + "learning_rate": 9.594203121519508e-06, + "loss": 0.9732, + "step": 1559 + }, + { + "epoch": 3.9897698209718673, + "grad_norm": 0.14169937812952016, + "learning_rate": 9.547832970473858e-06, + "loss": 0.9874, + "step": 1560 + }, + { + "epoch": 3.9923273657289, + "grad_norm": 0.13845067281699075, + "learning_rate": 9.501559956756496e-06, + "loss": 0.9869, + "step": 1561 + }, + { + "epoch": 3.9948849104859336, + "grad_norm": 0.1347356255658209, + "learning_rate": 9.45538422797061e-06, + "loss": 1.0008, + "step": 1562 + }, + { + "epoch": 3.997442455242967, + "grad_norm": 0.1446889101806559, + "learning_rate": 9.409305931409066e-06, + "loss": 0.9778, + "step": 1563 + }, + { + "epoch": 4.0, + "grad_norm": 0.1456124424171873, + "learning_rate": 9.36332521405392e-06, + "loss": 0.9869, + "step": 1564 + }, + { + "epoch": 4.002557544757034, + "grad_norm": 0.1694434470987983, + "learning_rate": 9.317442222576015e-06, + "loss": 0.9557, + "step": 1565 + }, + { + "epoch": 4.005115089514066, + "grad_norm": 0.14659146019169372, + "learning_rate": 9.27165710333441e-06, + "loss": 0.9685, + "step": 1566 + }, + { + "epoch": 4.0076726342711, + "grad_norm": 0.1469630437189336, + "learning_rate": 9.225970002375994e-06, + "loss": 0.9439, + "step": 1567 + }, + { + "epoch": 4.010230179028133, + "grad_norm": 0.17386186743500315, + "learning_rate": 9.180381065435e-06, + "loss": 0.9612, + "step": 1568 + }, + { + "epoch": 4.012787723785166, + "grad_norm": 0.15084057811060614, + "learning_rate": 9.13489043793251e-06, + "loss": 0.9726, + "step": 1569 + }, + { + "epoch": 4.015345268542199, + "grad_norm": 0.15247338496330756, + "learning_rate": 9.089498264976044e-06, + "loss": 0.9744, + "step": 1570 + }, + { + "epoch": 4.017902813299233, + "grad_norm": 0.1489429891129685, + "learning_rate": 9.044204691359035e-06, + "loss": 0.9713, + "step": 1571 + }, + { + "epoch": 4.020460358056266, + "grad_norm": 0.1480263481725032, + "learning_rate": 8.99900986156046e-06, + "loss": 0.9579, + "step": 1572 + }, + { + "epoch": 4.023017902813299, + "grad_norm": 0.1608202224818316, + "learning_rate": 8.953913919744237e-06, + "loss": 0.9623, + "step": 1573 + }, + { + "epoch": 4.025575447570333, + "grad_norm": 0.14602910658372553, + "learning_rate": 8.908917009758902e-06, + "loss": 0.956, + "step": 1574 + }, + { + "epoch": 4.028132992327365, + "grad_norm": 0.14751895138775156, + "learning_rate": 8.864019275137096e-06, + "loss": 0.9611, + "step": 1575 + }, + { + "epoch": 4.030690537084399, + "grad_norm": 0.13940784210831325, + "learning_rate": 8.819220859095096e-06, + "loss": 0.9734, + "step": 1576 + }, + { + "epoch": 4.033248081841432, + "grad_norm": 0.15771825616041407, + "learning_rate": 8.774521904532349e-06, + "loss": 0.9448, + "step": 1577 + }, + { + "epoch": 4.035805626598465, + "grad_norm": 0.14987870095750866, + "learning_rate": 8.7299225540311e-06, + "loss": 0.9735, + "step": 1578 + }, + { + "epoch": 4.038363171355499, + "grad_norm": 0.13597639144343102, + "learning_rate": 8.685422949855785e-06, + "loss": 0.9733, + "step": 1579 + }, + { + "epoch": 4.040920716112532, + "grad_norm": 0.14348384657538388, + "learning_rate": 8.64102323395272e-06, + "loss": 0.9915, + "step": 1580 + }, + { + "epoch": 4.043478260869565, + "grad_norm": 0.13911803653848545, + "learning_rate": 8.596723547949603e-06, + "loss": 0.9659, + "step": 1581 + }, + { + "epoch": 4.046035805626598, + "grad_norm": 0.1303425224909242, + "learning_rate": 8.552524033155025e-06, + "loss": 0.952, + "step": 1582 + }, + { + "epoch": 4.048593350383632, + "grad_norm": 0.13127570779064995, + "learning_rate": 8.508424830558075e-06, + "loss": 0.9458, + "step": 1583 + }, + { + "epoch": 4.051150895140665, + "grad_norm": 0.13605211935620362, + "learning_rate": 8.464426080827808e-06, + "loss": 0.9638, + "step": 1584 + }, + { + "epoch": 4.053708439897698, + "grad_norm": 0.1353571143519125, + "learning_rate": 8.42052792431292e-06, + "loss": 0.9592, + "step": 1585 + }, + { + "epoch": 4.056265984654732, + "grad_norm": 0.13743810323417613, + "learning_rate": 8.376730501041197e-06, + "loss": 1.0004, + "step": 1586 + }, + { + "epoch": 4.0588235294117645, + "grad_norm": 0.1335028249124241, + "learning_rate": 8.333033950719089e-06, + "loss": 0.9268, + "step": 1587 + }, + { + "epoch": 4.061381074168798, + "grad_norm": 0.14308735627127328, + "learning_rate": 8.28943841273132e-06, + "loss": 0.9524, + "step": 1588 + }, + { + "epoch": 4.063938618925831, + "grad_norm": 0.13927706340176157, + "learning_rate": 8.24594402614038e-06, + "loss": 0.9931, + "step": 1589 + }, + { + "epoch": 4.0664961636828645, + "grad_norm": 0.13527098384462513, + "learning_rate": 8.202550929686061e-06, + "loss": 0.9591, + "step": 1590 + }, + { + "epoch": 4.069053708439898, + "grad_norm": 0.13623666624179986, + "learning_rate": 8.15925926178513e-06, + "loss": 0.9966, + "step": 1591 + }, + { + "epoch": 4.071611253196931, + "grad_norm": 0.13999973460956114, + "learning_rate": 8.116069160530772e-06, + "loss": 0.9862, + "step": 1592 + }, + { + "epoch": 4.0741687979539645, + "grad_norm": 0.13171292461081588, + "learning_rate": 8.072980763692197e-06, + "loss": 0.9959, + "step": 1593 + }, + { + "epoch": 4.076726342710997, + "grad_norm": 0.1367650747219921, + "learning_rate": 8.029994208714198e-06, + "loss": 0.9983, + "step": 1594 + }, + { + "epoch": 4.079283887468031, + "grad_norm": 0.13227052084210594, + "learning_rate": 7.987109632716703e-06, + "loss": 0.9592, + "step": 1595 + }, + { + "epoch": 4.081841432225064, + "grad_norm": 0.14444449340962193, + "learning_rate": 7.94432717249435e-06, + "loss": 0.9853, + "step": 1596 + }, + { + "epoch": 4.084398976982097, + "grad_norm": 0.13614148393534858, + "learning_rate": 7.901646964516039e-06, + "loss": 0.9715, + "step": 1597 + }, + { + "epoch": 4.086956521739131, + "grad_norm": 0.12749269393727558, + "learning_rate": 7.859069144924528e-06, + "loss": 0.9395, + "step": 1598 + }, + { + "epoch": 4.089514066496164, + "grad_norm": 0.13805044128867539, + "learning_rate": 7.816593849535935e-06, + "loss": 0.9437, + "step": 1599 + }, + { + "epoch": 4.092071611253197, + "grad_norm": 0.12754607705203092, + "learning_rate": 7.77422121383936e-06, + "loss": 0.9752, + "step": 1600 + }, + { + "epoch": 4.09462915601023, + "grad_norm": 0.1351624415394795, + "learning_rate": 7.731951372996445e-06, + "loss": 0.9709, + "step": 1601 + }, + { + "epoch": 4.0971867007672635, + "grad_norm": 0.12306767716486365, + "learning_rate": 7.689784461840916e-06, + "loss": 0.957, + "step": 1602 + }, + { + "epoch": 4.099744245524296, + "grad_norm": 0.12976229511348925, + "learning_rate": 7.647720614878182e-06, + "loss": 0.9433, + "step": 1603 + }, + { + "epoch": 4.10230179028133, + "grad_norm": 0.13817478561667218, + "learning_rate": 7.605759966284885e-06, + "loss": 0.9843, + "step": 1604 + }, + { + "epoch": 4.1048593350383635, + "grad_norm": 0.12832872660123104, + "learning_rate": 7.563902649908513e-06, + "loss": 0.9695, + "step": 1605 + }, + { + "epoch": 4.107416879795396, + "grad_norm": 0.1412483260552206, + "learning_rate": 7.522148799266888e-06, + "loss": 0.9728, + "step": 1606 + }, + { + "epoch": 4.10997442455243, + "grad_norm": 0.1365069279515938, + "learning_rate": 7.480498547547825e-06, + "loss": 0.9508, + "step": 1607 + }, + { + "epoch": 4.112531969309463, + "grad_norm": 0.13711455047388862, + "learning_rate": 7.438952027608688e-06, + "loss": 0.9579, + "step": 1608 + }, + { + "epoch": 4.115089514066496, + "grad_norm": 0.12701385871827708, + "learning_rate": 7.397509371975929e-06, + "loss": 0.9757, + "step": 1609 + }, + { + "epoch": 4.117647058823529, + "grad_norm": 0.15072117557519477, + "learning_rate": 7.356170712844703e-06, + "loss": 0.9704, + "step": 1610 + }, + { + "epoch": 4.120204603580563, + "grad_norm": 0.13660469487293564, + "learning_rate": 7.314936182078427e-06, + "loss": 0.9625, + "step": 1611 + }, + { + "epoch": 4.122762148337596, + "grad_norm": 0.1348626532127348, + "learning_rate": 7.273805911208374e-06, + "loss": 0.9565, + "step": 1612 + }, + { + "epoch": 4.125319693094629, + "grad_norm": 0.1558370983441284, + "learning_rate": 7.23278003143324e-06, + "loss": 0.9816, + "step": 1613 + }, + { + "epoch": 4.127877237851663, + "grad_norm": 0.1442560358532163, + "learning_rate": 7.191858673618713e-06, + "loss": 0.9693, + "step": 1614 + }, + { + "epoch": 4.130434782608695, + "grad_norm": 0.12076587398680057, + "learning_rate": 7.1510419682971235e-06, + "loss": 0.9704, + "step": 1615 + }, + { + "epoch": 4.132992327365729, + "grad_norm": 0.13613335771644464, + "learning_rate": 7.110330045666938e-06, + "loss": 0.9686, + "step": 1616 + }, + { + "epoch": 4.135549872122763, + "grad_norm": 0.15273830211192063, + "learning_rate": 7.069723035592368e-06, + "loss": 0.9643, + "step": 1617 + }, + { + "epoch": 4.138107416879795, + "grad_norm": 0.13829841062139175, + "learning_rate": 7.02922106760302e-06, + "loss": 0.9638, + "step": 1618 + }, + { + "epoch": 4.140664961636829, + "grad_norm": 0.1312114107225528, + "learning_rate": 6.988824270893402e-06, + "loss": 0.9392, + "step": 1619 + }, + { + "epoch": 4.143222506393862, + "grad_norm": 0.1411742619531022, + "learning_rate": 6.948532774322552e-06, + "loss": 0.9921, + "step": 1620 + }, + { + "epoch": 4.145780051150895, + "grad_norm": 0.1472736684669091, + "learning_rate": 6.908346706413605e-06, + "loss": 0.9759, + "step": 1621 + }, + { + "epoch": 4.148337595907928, + "grad_norm": 0.13211483942081645, + "learning_rate": 6.868266195353439e-06, + "loss": 0.9725, + "step": 1622 + }, + { + "epoch": 4.150895140664962, + "grad_norm": 0.1328326493254342, + "learning_rate": 6.828291368992159e-06, + "loss": 0.9881, + "step": 1623 + }, + { + "epoch": 4.153452685421995, + "grad_norm": 0.13520210264575472, + "learning_rate": 6.7884223548427915e-06, + "loss": 0.9523, + "step": 1624 + }, + { + "epoch": 4.156010230179028, + "grad_norm": 0.12812477279560075, + "learning_rate": 6.748659280080852e-06, + "loss": 0.9646, + "step": 1625 + }, + { + "epoch": 4.158567774936062, + "grad_norm": 0.1447636884254579, + "learning_rate": 6.709002271543896e-06, + "loss": 0.9764, + "step": 1626 + }, + { + "epoch": 4.161125319693094, + "grad_norm": 0.12397606004342086, + "learning_rate": 6.669451455731168e-06, + "loss": 0.9688, + "step": 1627 + }, + { + "epoch": 4.163682864450128, + "grad_norm": 0.12250284994573661, + "learning_rate": 6.630006958803159e-06, + "loss": 0.9832, + "step": 1628 + }, + { + "epoch": 4.166240409207161, + "grad_norm": 0.12522111574524233, + "learning_rate": 6.590668906581235e-06, + "loss": 0.9865, + "step": 1629 + }, + { + "epoch": 4.168797953964194, + "grad_norm": 0.14426664986318974, + "learning_rate": 6.551437424547207e-06, + "loss": 0.9745, + "step": 1630 + }, + { + "epoch": 4.171355498721228, + "grad_norm": 0.1364036728911068, + "learning_rate": 6.5123126378429455e-06, + "loss": 0.9926, + "step": 1631 + }, + { + "epoch": 4.173913043478261, + "grad_norm": 0.12516777388845976, + "learning_rate": 6.473294671270011e-06, + "loss": 0.9718, + "step": 1632 + }, + { + "epoch": 4.176470588235294, + "grad_norm": 0.1255663623374594, + "learning_rate": 6.4343836492891974e-06, + "loss": 0.9668, + "step": 1633 + }, + { + "epoch": 4.179028132992327, + "grad_norm": 0.12758144062263563, + "learning_rate": 6.395579696020138e-06, + "loss": 0.9631, + "step": 1634 + }, + { + "epoch": 4.181585677749361, + "grad_norm": 0.13530428494260122, + "learning_rate": 6.3568829352409975e-06, + "loss": 0.98, + "step": 1635 + }, + { + "epoch": 4.1841432225063935, + "grad_norm": 0.13016949790316, + "learning_rate": 6.3182934903879725e-06, + "loss": 1.0123, + "step": 1636 + }, + { + "epoch": 4.186700767263427, + "grad_norm": 0.14112114746254337, + "learning_rate": 6.279811484554943e-06, + "loss": 0.9655, + "step": 1637 + }, + { + "epoch": 4.189258312020461, + "grad_norm": 0.1275629640487005, + "learning_rate": 6.241437040493088e-06, + "loss": 0.9812, + "step": 1638 + }, + { + "epoch": 4.1918158567774935, + "grad_norm": 0.14094316110593494, + "learning_rate": 6.2031702806104735e-06, + "loss": 0.9639, + "step": 1639 + }, + { + "epoch": 4.194373401534527, + "grad_norm": 0.13295300129747156, + "learning_rate": 6.165011326971679e-06, + "loss": 0.9572, + "step": 1640 + }, + { + "epoch": 4.19693094629156, + "grad_norm": 0.12981205516840236, + "learning_rate": 6.1269603012973795e-06, + "loss": 0.9358, + "step": 1641 + }, + { + "epoch": 4.1994884910485935, + "grad_norm": 0.12438813099797683, + "learning_rate": 6.089017324964017e-06, + "loss": 0.9838, + "step": 1642 + }, + { + "epoch": 4.202046035805626, + "grad_norm": 0.12621995618989404, + "learning_rate": 6.051182519003336e-06, + "loss": 0.9413, + "step": 1643 + }, + { + "epoch": 4.20460358056266, + "grad_norm": 0.12388214081758457, + "learning_rate": 6.013456004102058e-06, + "loss": 0.9833, + "step": 1644 + }, + { + "epoch": 4.207161125319693, + "grad_norm": 0.13864147929412118, + "learning_rate": 5.97583790060146e-06, + "loss": 0.9699, + "step": 1645 + }, + { + "epoch": 4.209718670076726, + "grad_norm": 0.12152527637941755, + "learning_rate": 5.938328328497016e-06, + "loss": 0.9504, + "step": 1646 + }, + { + "epoch": 4.21227621483376, + "grad_norm": 0.1228917559817376, + "learning_rate": 5.900927407437995e-06, + "loss": 0.9521, + "step": 1647 + }, + { + "epoch": 4.2148337595907925, + "grad_norm": 0.1281261751626848, + "learning_rate": 5.8636352567270804e-06, + "loss": 0.964, + "step": 1648 + }, + { + "epoch": 4.217391304347826, + "grad_norm": 0.13928877396090505, + "learning_rate": 5.8264519953200375e-06, + "loss": 0.9479, + "step": 1649 + }, + { + "epoch": 4.21994884910486, + "grad_norm": 0.12140483592869106, + "learning_rate": 5.7893777418252325e-06, + "loss": 0.9512, + "step": 1650 + }, + { + "epoch": 4.2225063938618925, + "grad_norm": 0.12832906273182243, + "learning_rate": 5.75241261450334e-06, + "loss": 0.9714, + "step": 1651 + }, + { + "epoch": 4.225063938618926, + "grad_norm": 0.12325256644456352, + "learning_rate": 5.715556731266971e-06, + "loss": 0.9621, + "step": 1652 + }, + { + "epoch": 4.227621483375959, + "grad_norm": 0.1231212443972644, + "learning_rate": 5.678810209680219e-06, + "loss": 0.9744, + "step": 1653 + }, + { + "epoch": 4.2301790281329925, + "grad_norm": 0.12843613548840424, + "learning_rate": 5.642173166958365e-06, + "loss": 0.9681, + "step": 1654 + }, + { + "epoch": 4.232736572890025, + "grad_norm": 0.12908136661093464, + "learning_rate": 5.605645719967445e-06, + "loss": 0.9931, + "step": 1655 + }, + { + "epoch": 4.235294117647059, + "grad_norm": 0.1221393745688676, + "learning_rate": 5.5692279852239286e-06, + "loss": 0.968, + "step": 1656 + }, + { + "epoch": 4.2378516624040925, + "grad_norm": 0.1283209209042707, + "learning_rate": 5.532920078894299e-06, + "loss": 0.9548, + "step": 1657 + }, + { + "epoch": 4.240409207161125, + "grad_norm": 0.13088853822976476, + "learning_rate": 5.496722116794715e-06, + "loss": 0.989, + "step": 1658 + }, + { + "epoch": 4.242966751918159, + "grad_norm": 0.12637460822527927, + "learning_rate": 5.4606342143906386e-06, + "loss": 0.9722, + "step": 1659 + }, + { + "epoch": 4.245524296675192, + "grad_norm": 0.12414353807978612, + "learning_rate": 5.424656486796465e-06, + "loss": 0.976, + "step": 1660 + }, + { + "epoch": 4.248081841432225, + "grad_norm": 0.1215866811292329, + "learning_rate": 5.388789048775103e-06, + "loss": 0.9828, + "step": 1661 + }, + { + "epoch": 4.250639386189258, + "grad_norm": 0.12501721399744686, + "learning_rate": 5.353032014737718e-06, + "loss": 0.9775, + "step": 1662 + }, + { + "epoch": 4.253196930946292, + "grad_norm": 0.12240136677562505, + "learning_rate": 5.317385498743273e-06, + "loss": 0.9585, + "step": 1663 + }, + { + "epoch": 4.255754475703325, + "grad_norm": 0.12452692032290599, + "learning_rate": 5.2818496144982025e-06, + "loss": 0.9567, + "step": 1664 + }, + { + "epoch": 4.258312020460358, + "grad_norm": 0.12715363573233318, + "learning_rate": 5.246424475356034e-06, + "loss": 0.9615, + "step": 1665 + }, + { + "epoch": 4.260869565217392, + "grad_norm": 0.12536021121142926, + "learning_rate": 5.211110194317073e-06, + "loss": 0.9626, + "step": 1666 + }, + { + "epoch": 4.263427109974424, + "grad_norm": 0.11961524409682861, + "learning_rate": 5.17590688402795e-06, + "loss": 0.9741, + "step": 1667 + }, + { + "epoch": 4.265984654731458, + "grad_norm": 0.12506582590821555, + "learning_rate": 5.140814656781352e-06, + "loss": 0.9226, + "step": 1668 + }, + { + "epoch": 4.268542199488491, + "grad_norm": 0.14073939782978875, + "learning_rate": 5.105833624515639e-06, + "loss": 0.9648, + "step": 1669 + }, + { + "epoch": 4.271099744245524, + "grad_norm": 0.12278625115353788, + "learning_rate": 5.070963898814456e-06, + "loss": 0.9779, + "step": 1670 + }, + { + "epoch": 4.273657289002558, + "grad_norm": 0.12666237685804943, + "learning_rate": 5.036205590906403e-06, + "loss": 0.9539, + "step": 1671 + }, + { + "epoch": 4.276214833759591, + "grad_norm": 0.13431493290109958, + "learning_rate": 5.0015588116646865e-06, + "loss": 0.9636, + "step": 1672 + }, + { + "epoch": 4.278772378516624, + "grad_norm": 0.12811446773007196, + "learning_rate": 4.967023671606734e-06, + "loss": 0.9201, + "step": 1673 + }, + { + "epoch": 4.281329923273657, + "grad_norm": 0.12424895319204726, + "learning_rate": 4.932600280893889e-06, + "loss": 0.9603, + "step": 1674 + }, + { + "epoch": 4.283887468030691, + "grad_norm": 0.13835712175563045, + "learning_rate": 4.8982887493310085e-06, + "loss": 0.9763, + "step": 1675 + }, + { + "epoch": 4.286445012787723, + "grad_norm": 0.1266202682434657, + "learning_rate": 4.8640891863661614e-06, + "loss": 0.9603, + "step": 1676 + }, + { + "epoch": 4.289002557544757, + "grad_norm": 0.12820404642448693, + "learning_rate": 4.830001701090248e-06, + "loss": 0.977, + "step": 1677 + }, + { + "epoch": 4.291560102301791, + "grad_norm": 0.11736882208459962, + "learning_rate": 4.796026402236624e-06, + "loss": 0.9769, + "step": 1678 + }, + { + "epoch": 4.294117647058823, + "grad_norm": 0.12866380807856084, + "learning_rate": 4.762163398180852e-06, + "loss": 0.9639, + "step": 1679 + }, + { + "epoch": 4.296675191815857, + "grad_norm": 0.13603267144210404, + "learning_rate": 4.728412796940251e-06, + "loss": 0.9822, + "step": 1680 + }, + { + "epoch": 4.29923273657289, + "grad_norm": 0.12397632361985814, + "learning_rate": 4.694774706173602e-06, + "loss": 0.9545, + "step": 1681 + }, + { + "epoch": 4.301790281329923, + "grad_norm": 0.13248904474625434, + "learning_rate": 4.661249233180787e-06, + "loss": 1.0015, + "step": 1682 + }, + { + "epoch": 4.304347826086957, + "grad_norm": 0.13056497746259793, + "learning_rate": 4.627836484902495e-06, + "loss": 0.9803, + "step": 1683 + }, + { + "epoch": 4.30690537084399, + "grad_norm": 0.12577847853997326, + "learning_rate": 4.59453656791979e-06, + "loss": 1.0028, + "step": 1684 + }, + { + "epoch": 4.309462915601023, + "grad_norm": 0.11751400707268564, + "learning_rate": 4.561349588453849e-06, + "loss": 0.9534, + "step": 1685 + }, + { + "epoch": 4.312020460358056, + "grad_norm": 0.1298510057754835, + "learning_rate": 4.528275652365608e-06, + "loss": 0.9848, + "step": 1686 + }, + { + "epoch": 4.31457800511509, + "grad_norm": 0.12778091258561974, + "learning_rate": 4.495314865155393e-06, + "loss": 0.9469, + "step": 1687 + }, + { + "epoch": 4.3171355498721224, + "grad_norm": 0.12331926980523211, + "learning_rate": 4.462467331962614e-06, + "loss": 0.9438, + "step": 1688 + }, + { + "epoch": 4.319693094629156, + "grad_norm": 0.11808697416694176, + "learning_rate": 4.4297331575654165e-06, + "loss": 0.9493, + "step": 1689 + }, + { + "epoch": 4.322250639386189, + "grad_norm": 0.1192941326588608, + "learning_rate": 4.397112446380347e-06, + "loss": 1.01, + "step": 1690 + }, + { + "epoch": 4.324808184143222, + "grad_norm": 0.1167465157397439, + "learning_rate": 4.3646053024620375e-06, + "loss": 0.9614, + "step": 1691 + }, + { + "epoch": 4.327365728900256, + "grad_norm": 0.12977635130536203, + "learning_rate": 4.33221182950283e-06, + "loss": 0.9589, + "step": 1692 + }, + { + "epoch": 4.329923273657289, + "grad_norm": 0.13176862781260584, + "learning_rate": 4.299932130832516e-06, + "loss": 0.9734, + "step": 1693 + }, + { + "epoch": 4.332480818414322, + "grad_norm": 0.13475421457283548, + "learning_rate": 4.267766309417947e-06, + "loss": 0.9647, + "step": 1694 + }, + { + "epoch": 4.335038363171355, + "grad_norm": 0.11898434149307777, + "learning_rate": 4.235714467862697e-06, + "loss": 0.9531, + "step": 1695 + }, + { + "epoch": 4.337595907928389, + "grad_norm": 0.12202881260318174, + "learning_rate": 4.2037767084068105e-06, + "loss": 0.9931, + "step": 1696 + }, + { + "epoch": 4.340153452685422, + "grad_norm": 0.1359310771287758, + "learning_rate": 4.171953132926403e-06, + "loss": 0.9611, + "step": 1697 + }, + { + "epoch": 4.342710997442455, + "grad_norm": 0.12498482699675276, + "learning_rate": 4.140243842933353e-06, + "loss": 0.9384, + "step": 1698 + }, + { + "epoch": 4.345268542199489, + "grad_norm": 0.12177790637791187, + "learning_rate": 4.1086489395750326e-06, + "loss": 0.9747, + "step": 1699 + }, + { + "epoch": 4.3478260869565215, + "grad_norm": 0.12407774991295589, + "learning_rate": 4.077168523633881e-06, + "loss": 0.9708, + "step": 1700 + }, + { + "epoch": 4.350383631713555, + "grad_norm": 0.12887036467594995, + "learning_rate": 4.045802695527177e-06, + "loss": 0.9354, + "step": 1701 + }, + { + "epoch": 4.352941176470588, + "grad_norm": 0.12989274173197723, + "learning_rate": 4.014551555306665e-06, + "loss": 0.9849, + "step": 1702 + }, + { + "epoch": 4.3554987212276215, + "grad_norm": 0.12421979367055937, + "learning_rate": 3.983415202658281e-06, + "loss": 0.9995, + "step": 1703 + }, + { + "epoch": 4.358056265984655, + "grad_norm": 0.1198682959010047, + "learning_rate": 3.952393736901785e-06, + "loss": 0.9533, + "step": 1704 + }, + { + "epoch": 4.360613810741688, + "grad_norm": 0.128521264353093, + "learning_rate": 3.921487256990473e-06, + "loss": 0.9708, + "step": 1705 + }, + { + "epoch": 4.3631713554987215, + "grad_norm": 0.11847073942111394, + "learning_rate": 3.890695861510856e-06, + "loss": 0.9814, + "step": 1706 + }, + { + "epoch": 4.365728900255754, + "grad_norm": 0.12769887613282743, + "learning_rate": 3.8600196486823405e-06, + "loss": 0.9717, + "step": 1707 + }, + { + "epoch": 4.368286445012788, + "grad_norm": 0.12999690542001419, + "learning_rate": 3.8294587163569194e-06, + "loss": 0.9598, + "step": 1708 + }, + { + "epoch": 4.370843989769821, + "grad_norm": 0.11857254485023354, + "learning_rate": 3.799013162018881e-06, + "loss": 0.9837, + "step": 1709 + }, + { + "epoch": 4.373401534526854, + "grad_norm": 0.12046803154933006, + "learning_rate": 3.7686830827844544e-06, + "loss": 0.9721, + "step": 1710 + }, + { + "epoch": 4.375959079283888, + "grad_norm": 0.12149366728466514, + "learning_rate": 3.7384685754015126e-06, + "loss": 0.9946, + "step": 1711 + }, + { + "epoch": 4.378516624040921, + "grad_norm": 0.11870216605190843, + "learning_rate": 3.7083697362493066e-06, + "loss": 0.9938, + "step": 1712 + }, + { + "epoch": 4.381074168797954, + "grad_norm": 0.11844452341376038, + "learning_rate": 3.6783866613381025e-06, + "loss": 0.9597, + "step": 1713 + }, + { + "epoch": 4.383631713554987, + "grad_norm": 0.1243613960126632, + "learning_rate": 3.6485194463089022e-06, + "loss": 0.9816, + "step": 1714 + }, + { + "epoch": 4.3861892583120206, + "grad_norm": 0.11603761858752591, + "learning_rate": 3.6187681864331235e-06, + "loss": 0.9774, + "step": 1715 + }, + { + "epoch": 4.388746803069053, + "grad_norm": 0.11736635034857253, + "learning_rate": 3.5891329766123505e-06, + "loss": 0.9663, + "step": 1716 + }, + { + "epoch": 4.391304347826087, + "grad_norm": 0.11962011769251431, + "learning_rate": 3.5596139113779215e-06, + "loss": 1.0028, + "step": 1717 + }, + { + "epoch": 4.3938618925831205, + "grad_norm": 0.114948006498465, + "learning_rate": 3.5302110848907336e-06, + "loss": 0.9469, + "step": 1718 + }, + { + "epoch": 4.396419437340153, + "grad_norm": 0.10841359436318306, + "learning_rate": 3.500924590940904e-06, + "loss": 0.9587, + "step": 1719 + }, + { + "epoch": 4.398976982097187, + "grad_norm": 0.11627559315730779, + "learning_rate": 3.4717545229474524e-06, + "loss": 0.9788, + "step": 1720 + }, + { + "epoch": 4.40153452685422, + "grad_norm": 0.1152200142947816, + "learning_rate": 3.442700973958024e-06, + "loss": 0.9962, + "step": 1721 + }, + { + "epoch": 4.404092071611253, + "grad_norm": 0.1154591706617095, + "learning_rate": 3.4137640366485924e-06, + "loss": 0.9824, + "step": 1722 + }, + { + "epoch": 4.406649616368286, + "grad_norm": 0.11858603589980796, + "learning_rate": 3.384943803323153e-06, + "loss": 0.9865, + "step": 1723 + }, + { + "epoch": 4.40920716112532, + "grad_norm": 0.12077395065465756, + "learning_rate": 3.356240365913439e-06, + "loss": 0.9551, + "step": 1724 + }, + { + "epoch": 4.411764705882353, + "grad_norm": 0.11389523914601839, + "learning_rate": 3.3276538159786197e-06, + "loss": 0.9706, + "step": 1725 + }, + { + "epoch": 4.414322250639386, + "grad_norm": 0.11263773696411303, + "learning_rate": 3.299184244705025e-06, + "loss": 0.9697, + "step": 1726 + }, + { + "epoch": 4.41687979539642, + "grad_norm": 0.12406578721816504, + "learning_rate": 3.2708317429058334e-06, + "loss": 0.9589, + "step": 1727 + }, + { + "epoch": 4.419437340153452, + "grad_norm": 0.12286088725943455, + "learning_rate": 3.2425964010207765e-06, + "loss": 0.9588, + "step": 1728 + }, + { + "epoch": 4.421994884910486, + "grad_norm": 0.11915716329986083, + "learning_rate": 3.214478309115894e-06, + "loss": 0.9641, + "step": 1729 + }, + { + "epoch": 4.42455242966752, + "grad_norm": 0.12652222267615001, + "learning_rate": 3.1864775568832027e-06, + "loss": 1.0013, + "step": 1730 + }, + { + "epoch": 4.427109974424552, + "grad_norm": 0.11389511903367709, + "learning_rate": 3.1585942336404262e-06, + "loss": 0.9535, + "step": 1731 + }, + { + "epoch": 4.429667519181586, + "grad_norm": 0.12438352231529569, + "learning_rate": 3.13082842833071e-06, + "loss": 0.9435, + "step": 1732 + }, + { + "epoch": 4.432225063938619, + "grad_norm": 0.1175440012829412, + "learning_rate": 3.1031802295223355e-06, + "loss": 0.9757, + "step": 1733 + }, + { + "epoch": 4.434782608695652, + "grad_norm": 0.1258322406199311, + "learning_rate": 3.075649725408445e-06, + "loss": 0.9593, + "step": 1734 + }, + { + "epoch": 4.437340153452685, + "grad_norm": 0.1181435995087277, + "learning_rate": 3.048237003806738e-06, + "loss": 0.9811, + "step": 1735 + }, + { + "epoch": 4.439897698209719, + "grad_norm": 0.12072878038936147, + "learning_rate": 3.0209421521592318e-06, + "loss": 1.0008, + "step": 1736 + }, + { + "epoch": 4.442455242966752, + "grad_norm": 0.1199404755824827, + "learning_rate": 2.9937652575319444e-06, + "loss": 0.9636, + "step": 1737 + }, + { + "epoch": 4.445012787723785, + "grad_norm": 0.11517484111123064, + "learning_rate": 2.966706406614628e-06, + "loss": 0.9639, + "step": 1738 + }, + { + "epoch": 4.447570332480819, + "grad_norm": 0.11599655109609232, + "learning_rate": 2.939765685720497e-06, + "loss": 0.983, + "step": 1739 + }, + { + "epoch": 4.450127877237851, + "grad_norm": 0.11702394553897594, + "learning_rate": 2.912943180785961e-06, + "loss": 0.9378, + "step": 1740 + }, + { + "epoch": 4.452685421994885, + "grad_norm": 0.11662021025404341, + "learning_rate": 2.886238977370326e-06, + "loss": 0.9564, + "step": 1741 + }, + { + "epoch": 4.455242966751918, + "grad_norm": 0.11861552530992366, + "learning_rate": 2.8596531606555377e-06, + "loss": 1.0017, + "step": 1742 + }, + { + "epoch": 4.457800511508951, + "grad_norm": 0.11486045991602647, + "learning_rate": 2.833185815445929e-06, + "loss": 0.956, + "step": 1743 + }, + { + "epoch": 4.460358056265985, + "grad_norm": 0.1164478351871581, + "learning_rate": 2.806837026167899e-06, + "loss": 0.9627, + "step": 1744 + }, + { + "epoch": 4.462915601023018, + "grad_norm": 0.11604852293545323, + "learning_rate": 2.780606876869678e-06, + "loss": 0.9723, + "step": 1745 + }, + { + "epoch": 4.465473145780051, + "grad_norm": 0.11757102549840029, + "learning_rate": 2.7544954512210842e-06, + "loss": 0.9769, + "step": 1746 + }, + { + "epoch": 4.468030690537084, + "grad_norm": 0.12255097610890886, + "learning_rate": 2.7285028325131892e-06, + "loss": 0.963, + "step": 1747 + }, + { + "epoch": 4.470588235294118, + "grad_norm": 0.11399478345208364, + "learning_rate": 2.702629103658114e-06, + "loss": 0.9682, + "step": 1748 + }, + { + "epoch": 4.4731457800511505, + "grad_norm": 0.11142062474858554, + "learning_rate": 2.6768743471887337e-06, + "loss": 0.9715, + "step": 1749 + }, + { + "epoch": 4.475703324808184, + "grad_norm": 0.11410463536786239, + "learning_rate": 2.651238645258416e-06, + "loss": 0.9953, + "step": 1750 + }, + { + "epoch": 4.478260869565218, + "grad_norm": 0.12100539423129517, + "learning_rate": 2.6257220796407756e-06, + "loss": 0.9606, + "step": 1751 + }, + { + "epoch": 4.4808184143222505, + "grad_norm": 0.11758059360916463, + "learning_rate": 2.6003247317293847e-06, + "loss": 0.9759, + "step": 1752 + }, + { + "epoch": 4.483375959079284, + "grad_norm": 0.11144724361162288, + "learning_rate": 2.5750466825375585e-06, + "loss": 0.9872, + "step": 1753 + }, + { + "epoch": 4.485933503836317, + "grad_norm": 0.11347669660230812, + "learning_rate": 2.5498880126980518e-06, + "loss": 0.988, + "step": 1754 + }, + { + "epoch": 4.4884910485933505, + "grad_norm": 0.11874354177271917, + "learning_rate": 2.5248488024628024e-06, + "loss": 0.9636, + "step": 1755 + }, + { + "epoch": 4.491048593350383, + "grad_norm": 0.11925275787467719, + "learning_rate": 2.499929131702734e-06, + "loss": 0.9773, + "step": 1756 + }, + { + "epoch": 4.493606138107417, + "grad_norm": 0.11090730679921106, + "learning_rate": 2.4751290799074213e-06, + "loss": 0.9769, + "step": 1757 + }, + { + "epoch": 4.4961636828644505, + "grad_norm": 0.11558893311677444, + "learning_rate": 2.450448726184891e-06, + "loss": 0.9523, + "step": 1758 + }, + { + "epoch": 4.498721227621483, + "grad_norm": 0.1273459674627714, + "learning_rate": 2.425888149261346e-06, + "loss": 0.9389, + "step": 1759 + }, + { + "epoch": 4.501278772378517, + "grad_norm": 0.11864458292723912, + "learning_rate": 2.401447427480941e-06, + "loss": 0.9895, + "step": 1760 + }, + { + "epoch": 4.5038363171355495, + "grad_norm": 0.10951592315079048, + "learning_rate": 2.3771266388054846e-06, + "loss": 0.9817, + "step": 1761 + }, + { + "epoch": 4.506393861892583, + "grad_norm": 0.11357909677815746, + "learning_rate": 2.3529258608142326e-06, + "loss": 0.9939, + "step": 1762 + }, + { + "epoch": 4.508951406649617, + "grad_norm": 0.11971018322164677, + "learning_rate": 2.3288451707036288e-06, + "loss": 0.9793, + "step": 1763 + }, + { + "epoch": 4.5115089514066495, + "grad_norm": 0.1122150483378089, + "learning_rate": 2.3048846452870597e-06, + "loss": 0.9554, + "step": 1764 + }, + { + "epoch": 4.514066496163683, + "grad_norm": 0.11638676977027419, + "learning_rate": 2.281044360994589e-06, + "loss": 0.9705, + "step": 1765 + }, + { + "epoch": 4.516624040920716, + "grad_norm": 0.11255736814038485, + "learning_rate": 2.257324393872753e-06, + "loss": 0.9569, + "step": 1766 + }, + { + "epoch": 4.5191815856777495, + "grad_norm": 0.1183417374581666, + "learning_rate": 2.233724819584282e-06, + "loss": 0.9875, + "step": 1767 + }, + { + "epoch": 4.521739130434782, + "grad_norm": 0.11525721410180423, + "learning_rate": 2.210245713407879e-06, + "loss": 0.9851, + "step": 1768 + }, + { + "epoch": 4.524296675191816, + "grad_norm": 0.11672302497611096, + "learning_rate": 2.186887150237964e-06, + "loss": 0.9434, + "step": 1769 + }, + { + "epoch": 4.526854219948849, + "grad_norm": 0.11462568134574831, + "learning_rate": 2.163649204584464e-06, + "loss": 0.9676, + "step": 1770 + }, + { + "epoch": 4.529411764705882, + "grad_norm": 0.11637157916960862, + "learning_rate": 2.140531950572551e-06, + "loss": 0.9625, + "step": 1771 + }, + { + "epoch": 4.531969309462916, + "grad_norm": 0.11317862758953197, + "learning_rate": 2.117535461942386e-06, + "loss": 0.9667, + "step": 1772 + }, + { + "epoch": 4.534526854219949, + "grad_norm": 0.1126947562619317, + "learning_rate": 2.094659812048945e-06, + "loss": 0.9548, + "step": 1773 + }, + { + "epoch": 4.537084398976982, + "grad_norm": 0.1135661694913091, + "learning_rate": 2.071905073861724e-06, + "loss": 0.9586, + "step": 1774 + }, + { + "epoch": 4.539641943734015, + "grad_norm": 0.11250229002588859, + "learning_rate": 2.049271319964543e-06, + "loss": 0.9482, + "step": 1775 + }, + { + "epoch": 4.542199488491049, + "grad_norm": 0.11649750185398619, + "learning_rate": 2.0267586225552892e-06, + "loss": 0.9571, + "step": 1776 + }, + { + "epoch": 4.544757033248082, + "grad_norm": 0.12069116537202378, + "learning_rate": 2.0043670534457194e-06, + "loss": 0.9737, + "step": 1777 + }, + { + "epoch": 4.547314578005115, + "grad_norm": 0.11274024393050174, + "learning_rate": 1.982096684061192e-06, + "loss": 0.9848, + "step": 1778 + }, + { + "epoch": 4.549872122762149, + "grad_norm": 0.11791521579577226, + "learning_rate": 1.959947585440469e-06, + "loss": 0.9695, + "step": 1779 + }, + { + "epoch": 4.552429667519181, + "grad_norm": 0.11087965419915281, + "learning_rate": 1.9379198282354794e-06, + "loss": 0.9621, + "step": 1780 + }, + { + "epoch": 4.554987212276215, + "grad_norm": 0.11400022805723166, + "learning_rate": 1.916013482711092e-06, + "loss": 0.9935, + "step": 1781 + }, + { + "epoch": 4.557544757033249, + "grad_norm": 0.11289106885368878, + "learning_rate": 1.8942286187448865e-06, + "loss": 0.9679, + "step": 1782 + }, + { + "epoch": 4.560102301790281, + "grad_norm": 0.11665726379138751, + "learning_rate": 1.8725653058269433e-06, + "loss": 1.0187, + "step": 1783 + }, + { + "epoch": 4.562659846547315, + "grad_norm": 0.1067411918098924, + "learning_rate": 1.8510236130596126e-06, + "loss": 0.9438, + "step": 1784 + }, + { + "epoch": 4.565217391304348, + "grad_norm": 0.11467062095804521, + "learning_rate": 1.8296036091573022e-06, + "loss": 0.9453, + "step": 1785 + }, + { + "epoch": 4.567774936061381, + "grad_norm": 0.1141598096289312, + "learning_rate": 1.8083053624462365e-06, + "loss": 0.9531, + "step": 1786 + }, + { + "epoch": 4.570332480818414, + "grad_norm": 0.11829288762440843, + "learning_rate": 1.7871289408642889e-06, + "loss": 0.9635, + "step": 1787 + }, + { + "epoch": 4.572890025575448, + "grad_norm": 0.11375107040879473, + "learning_rate": 1.7660744119606876e-06, + "loss": 0.9531, + "step": 1788 + }, + { + "epoch": 4.57544757033248, + "grad_norm": 0.11425273488833146, + "learning_rate": 1.745141842895861e-06, + "loss": 1.0048, + "step": 1789 + }, + { + "epoch": 4.578005115089514, + "grad_norm": 0.1136362533257648, + "learning_rate": 1.7243313004412288e-06, + "loss": 0.9783, + "step": 1790 + }, + { + "epoch": 4.580562659846548, + "grad_norm": 0.11456227986474488, + "learning_rate": 1.7036428509789394e-06, + "loss": 0.9669, + "step": 1791 + }, + { + "epoch": 4.58312020460358, + "grad_norm": 0.11073420030884291, + "learning_rate": 1.6830765605016842e-06, + "loss": 0.9688, + "step": 1792 + }, + { + "epoch": 4.585677749360614, + "grad_norm": 0.11285691919453725, + "learning_rate": 1.6626324946125061e-06, + "loss": 0.962, + "step": 1793 + }, + { + "epoch": 4.588235294117647, + "grad_norm": 0.11211314756286417, + "learning_rate": 1.6423107185245557e-06, + "loss": 0.972, + "step": 1794 + }, + { + "epoch": 4.59079283887468, + "grad_norm": 0.11365304032190994, + "learning_rate": 1.6221112970609088e-06, + "loss": 0.9526, + "step": 1795 + }, + { + "epoch": 4.593350383631714, + "grad_norm": 0.10847895079352249, + "learning_rate": 1.60203429465434e-06, + "loss": 0.9635, + "step": 1796 + }, + { + "epoch": 4.595907928388747, + "grad_norm": 0.11376227683469585, + "learning_rate": 1.5820797753471495e-06, + "loss": 0.971, + "step": 1797 + }, + { + "epoch": 4.59846547314578, + "grad_norm": 0.10877615039280635, + "learning_rate": 1.562247802790915e-06, + "loss": 0.9482, + "step": 1798 + }, + { + "epoch": 4.601023017902813, + "grad_norm": 0.1152456202929066, + "learning_rate": 1.542538440246304e-06, + "loss": 0.9568, + "step": 1799 + }, + { + "epoch": 4.603580562659847, + "grad_norm": 0.11554187769428996, + "learning_rate": 1.5229517505829105e-06, + "loss": 0.9892, + "step": 1800 + }, + { + "epoch": 4.6061381074168795, + "grad_norm": 0.11514254996593944, + "learning_rate": 1.5034877962789973e-06, + "loss": 0.9494, + "step": 1801 + }, + { + "epoch": 4.608695652173913, + "grad_norm": 0.11224106372638573, + "learning_rate": 1.4841466394213311e-06, + "loss": 0.9392, + "step": 1802 + }, + { + "epoch": 4.611253196930946, + "grad_norm": 0.11385256614913049, + "learning_rate": 1.4649283417049653e-06, + "loss": 0.9633, + "step": 1803 + }, + { + "epoch": 4.6138107416879794, + "grad_norm": 0.10775692637459648, + "learning_rate": 1.445832964433076e-06, + "loss": 0.9312, + "step": 1804 + }, + { + "epoch": 4.616368286445013, + "grad_norm": 0.10984407586190349, + "learning_rate": 1.4268605685167125e-06, + "loss": 0.9746, + "step": 1805 + }, + { + "epoch": 4.618925831202046, + "grad_norm": 0.11487256938413119, + "learning_rate": 1.4080112144746516e-06, + "loss": 0.9383, + "step": 1806 + }, + { + "epoch": 4.621483375959079, + "grad_norm": 0.11240309733019481, + "learning_rate": 1.3892849624331973e-06, + "loss": 0.9678, + "step": 1807 + }, + { + "epoch": 4.624040920716112, + "grad_norm": 0.1169186360063392, + "learning_rate": 1.3706818721259584e-06, + "loss": 1.0104, + "step": 1808 + }, + { + "epoch": 4.626598465473146, + "grad_norm": 0.11728481679961328, + "learning_rate": 1.35220200289369e-06, + "loss": 0.9806, + "step": 1809 + }, + { + "epoch": 4.629156010230179, + "grad_norm": 0.11281383249587894, + "learning_rate": 1.3338454136840872e-06, + "loss": 0.9677, + "step": 1810 + }, + { + "epoch": 4.631713554987212, + "grad_norm": 0.10499750834126508, + "learning_rate": 1.315612163051605e-06, + "loss": 0.9676, + "step": 1811 + }, + { + "epoch": 4.634271099744246, + "grad_norm": 0.1110910923804026, + "learning_rate": 1.2975023091572657e-06, + "loss": 0.994, + "step": 1812 + }, + { + "epoch": 4.6368286445012785, + "grad_norm": 0.11326212934873885, + "learning_rate": 1.279515909768483e-06, + "loss": 0.9345, + "step": 1813 + }, + { + "epoch": 4.639386189258312, + "grad_norm": 0.11405643543560509, + "learning_rate": 1.2616530222588685e-06, + "loss": 1.0019, + "step": 1814 + }, + { + "epoch": 4.641943734015345, + "grad_norm": 0.1158242012128305, + "learning_rate": 1.2439137036080573e-06, + "loss": 0.9927, + "step": 1815 + }, + { + "epoch": 4.6445012787723785, + "grad_norm": 0.1149985852845067, + "learning_rate": 1.2262980104014921e-06, + "loss": 0.9925, + "step": 1816 + }, + { + "epoch": 4.647058823529412, + "grad_norm": 0.11465565149918291, + "learning_rate": 1.208805998830318e-06, + "loss": 0.9895, + "step": 1817 + }, + { + "epoch": 4.649616368286445, + "grad_norm": 0.11220682285996932, + "learning_rate": 1.1914377246911158e-06, + "loss": 0.9783, + "step": 1818 + }, + { + "epoch": 4.6521739130434785, + "grad_norm": 0.11250586266208777, + "learning_rate": 1.174193243385786e-06, + "loss": 0.9549, + "step": 1819 + }, + { + "epoch": 4.654731457800511, + "grad_norm": 0.10899951533184332, + "learning_rate": 1.1570726099213414e-06, + "loss": 0.943, + "step": 1820 + }, + { + "epoch": 4.657289002557545, + "grad_norm": 0.11251333624607752, + "learning_rate": 1.1400758789097499e-06, + "loss": 0.9443, + "step": 1821 + }, + { + "epoch": 4.659846547314578, + "grad_norm": 0.10854566000476132, + "learning_rate": 1.1232031045677405e-06, + "loss": 0.9987, + "step": 1822 + }, + { + "epoch": 4.662404092071611, + "grad_norm": 0.10722540031517447, + "learning_rate": 1.106454340716643e-06, + "loss": 0.9463, + "step": 1823 + }, + { + "epoch": 4.664961636828645, + "grad_norm": 0.11013365048748615, + "learning_rate": 1.089829640782223e-06, + "loss": 0.9599, + "step": 1824 + }, + { + "epoch": 4.667519181585678, + "grad_norm": 0.10882148143721988, + "learning_rate": 1.073329057794501e-06, + "loss": 0.9626, + "step": 1825 + }, + { + "epoch": 4.670076726342711, + "grad_norm": 0.10362710867323117, + "learning_rate": 1.0569526443875745e-06, + "loss": 0.9586, + "step": 1826 + }, + { + "epoch": 4.672634271099744, + "grad_norm": 0.10858009010596835, + "learning_rate": 1.0407004527994701e-06, + "loss": 0.9603, + "step": 1827 + }, + { + "epoch": 4.675191815856778, + "grad_norm": 0.1044741695809889, + "learning_rate": 1.0245725348719681e-06, + "loss": 0.9526, + "step": 1828 + }, + { + "epoch": 4.677749360613811, + "grad_norm": 0.11369566676381282, + "learning_rate": 1.0085689420504274e-06, + "loss": 0.958, + "step": 1829 + }, + { + "epoch": 4.680306905370844, + "grad_norm": 0.11199398881890137, + "learning_rate": 9.926897253836442e-07, + "loss": 0.9825, + "step": 1830 + }, + { + "epoch": 4.6828644501278776, + "grad_norm": 0.10991238024490946, + "learning_rate": 9.769349355236702e-07, + "loss": 0.9671, + "step": 1831 + }, + { + "epoch": 4.68542199488491, + "grad_norm": 0.1085409176039824, + "learning_rate": 9.613046227256472e-07, + "loss": 0.9883, + "step": 1832 + }, + { + "epoch": 4.687979539641944, + "grad_norm": 0.10448681879944104, + "learning_rate": 9.457988368476579e-07, + "loss": 0.9578, + "step": 1833 + }, + { + "epoch": 4.690537084398977, + "grad_norm": 0.11285313196545035, + "learning_rate": 9.304176273505861e-07, + "loss": 1.0142, + "step": 1834 + }, + { + "epoch": 4.69309462915601, + "grad_norm": 0.11325534614886543, + "learning_rate": 9.151610432979097e-07, + "loss": 0.9775, + "step": 1835 + }, + { + "epoch": 4.695652173913043, + "grad_norm": 0.11030129813920701, + "learning_rate": 9.000291333555888e-07, + "loss": 0.9895, + "step": 1836 + }, + { + "epoch": 4.698209718670077, + "grad_norm": 0.10981087858061067, + "learning_rate": 8.850219457918929e-07, + "loss": 0.973, + "step": 1837 + }, + { + "epoch": 4.70076726342711, + "grad_norm": 0.10905450057218977, + "learning_rate": 8.701395284772407e-07, + "loss": 0.9425, + "step": 1838 + }, + { + "epoch": 4.703324808184143, + "grad_norm": 0.10782067684161685, + "learning_rate": 8.553819288840537e-07, + "loss": 0.939, + "step": 1839 + }, + { + "epoch": 4.705882352941177, + "grad_norm": 0.10938807520129998, + "learning_rate": 8.40749194086623e-07, + "loss": 0.9976, + "step": 1840 + }, + { + "epoch": 4.708439897698209, + "grad_norm": 0.10882193167069305, + "learning_rate": 8.262413707609185e-07, + "loss": 0.9576, + "step": 1841 + }, + { + "epoch": 4.710997442455243, + "grad_norm": 0.11301673326793968, + "learning_rate": 8.118585051844774e-07, + "loss": 0.9554, + "step": 1842 + }, + { + "epoch": 4.713554987212277, + "grad_norm": 0.10441867915032454, + "learning_rate": 7.976006432362449e-07, + "loss": 0.9571, + "step": 1843 + }, + { + "epoch": 4.716112531969309, + "grad_norm": 0.10956276651492997, + "learning_rate": 7.834678303964094e-07, + "loss": 0.9781, + "step": 1844 + }, + { + "epoch": 4.718670076726343, + "grad_norm": 0.10620345574636991, + "learning_rate": 7.694601117462919e-07, + "loss": 0.9743, + "step": 1845 + }, + { + "epoch": 4.721227621483376, + "grad_norm": 0.11385253611249209, + "learning_rate": 7.555775319681724e-07, + "loss": 0.9642, + "step": 1846 + }, + { + "epoch": 4.723785166240409, + "grad_norm": 0.10874528997426589, + "learning_rate": 7.41820135345166e-07, + "loss": 0.9681, + "step": 1847 + }, + { + "epoch": 4.726342710997442, + "grad_norm": 0.10852960725439142, + "learning_rate": 7.281879657610758e-07, + "loss": 0.9974, + "step": 1848 + }, + { + "epoch": 4.728900255754476, + "grad_norm": 0.10745125077889328, + "learning_rate": 7.146810667002291e-07, + "loss": 0.9972, + "step": 1849 + }, + { + "epoch": 4.731457800511509, + "grad_norm": 0.10513964134651423, + "learning_rate": 7.012994812473928e-07, + "loss": 0.965, + "step": 1850 + }, + { + "epoch": 4.734015345268542, + "grad_norm": 0.10468761014491187, + "learning_rate": 6.880432520875868e-07, + "loss": 0.9666, + "step": 1851 + }, + { + "epoch": 4.736572890025576, + "grad_norm": 0.10729241645271993, + "learning_rate": 6.749124215059733e-07, + "loss": 0.9584, + "step": 1852 + }, + { + "epoch": 4.739130434782608, + "grad_norm": 0.11359362236943447, + "learning_rate": 6.619070313877007e-07, + "loss": 0.9999, + "step": 1853 + }, + { + "epoch": 4.741687979539642, + "grad_norm": 0.10962775985169017, + "learning_rate": 6.490271232177981e-07, + "loss": 0.9938, + "step": 1854 + }, + { + "epoch": 4.744245524296675, + "grad_norm": 0.10854938801338014, + "learning_rate": 6.362727380810185e-07, + "loss": 0.97, + "step": 1855 + }, + { + "epoch": 4.746803069053708, + "grad_norm": 0.1068077939245722, + "learning_rate": 6.236439166617204e-07, + "loss": 0.9377, + "step": 1856 + }, + { + "epoch": 4.749360613810742, + "grad_norm": 0.10646672365841833, + "learning_rate": 6.111406992437374e-07, + "loss": 0.9583, + "step": 1857 + }, + { + "epoch": 4.751918158567775, + "grad_norm": 0.10549076785830483, + "learning_rate": 5.98763125710251e-07, + "loss": 0.9879, + "step": 1858 + }, + { + "epoch": 4.754475703324808, + "grad_norm": 0.10927589356320379, + "learning_rate": 5.865112355436431e-07, + "loss": 0.9596, + "step": 1859 + }, + { + "epoch": 4.757033248081841, + "grad_norm": 0.10616809908251783, + "learning_rate": 5.74385067825407e-07, + "loss": 0.9843, + "step": 1860 + }, + { + "epoch": 4.759590792838875, + "grad_norm": 0.1069862495687352, + "learning_rate": 5.623846612359884e-07, + "loss": 0.9781, + "step": 1861 + }, + { + "epoch": 4.762148337595908, + "grad_norm": 0.11193531922331344, + "learning_rate": 5.505100540546827e-07, + "loss": 0.97, + "step": 1862 + }, + { + "epoch": 4.764705882352941, + "grad_norm": 0.10748221296390185, + "learning_rate": 5.387612841595014e-07, + "loss": 0.9848, + "step": 1863 + }, + { + "epoch": 4.767263427109975, + "grad_norm": 0.11368280666790082, + "learning_rate": 5.271383890270621e-07, + "loss": 0.9915, + "step": 1864 + }, + { + "epoch": 4.7698209718670075, + "grad_norm": 0.10555727887171781, + "learning_rate": 5.156414057324588e-07, + "loss": 0.958, + "step": 1865 + }, + { + "epoch": 4.772378516624041, + "grad_norm": 0.11117163332286255, + "learning_rate": 5.042703709491425e-07, + "loss": 0.9827, + "step": 1866 + }, + { + "epoch": 4.774936061381074, + "grad_norm": 0.1117593310611823, + "learning_rate": 4.930253209488278e-07, + "loss": 0.9591, + "step": 1867 + }, + { + "epoch": 4.7774936061381075, + "grad_norm": 0.10350421985765804, + "learning_rate": 4.819062916013329e-07, + "loss": 0.9598, + "step": 1868 + }, + { + "epoch": 4.78005115089514, + "grad_norm": 0.10831994521473463, + "learning_rate": 4.7091331837451736e-07, + "loss": 0.974, + "step": 1869 + }, + { + "epoch": 4.782608695652174, + "grad_norm": 0.10688454108423966, + "learning_rate": 4.600464363341228e-07, + "loss": 0.9639, + "step": 1870 + }, + { + "epoch": 4.7851662404092075, + "grad_norm": 0.10740390940878632, + "learning_rate": 4.493056801436879e-07, + "loss": 1.0149, + "step": 1871 + }, + { + "epoch": 4.78772378516624, + "grad_norm": 0.1076013560128204, + "learning_rate": 4.386910840644376e-07, + "loss": 0.9489, + "step": 1872 + }, + { + "epoch": 4.790281329923274, + "grad_norm": 0.11083779030015688, + "learning_rate": 4.2820268195515434e-07, + "loss": 0.9543, + "step": 1873 + }, + { + "epoch": 4.792838874680307, + "grad_norm": 0.10701125626461934, + "learning_rate": 4.178405072720937e-07, + "loss": 0.9422, + "step": 1874 + }, + { + "epoch": 4.79539641943734, + "grad_norm": 0.10561417083050885, + "learning_rate": 4.076045930688599e-07, + "loss": 0.9719, + "step": 1875 + }, + { + "epoch": 4.797953964194374, + "grad_norm": 0.10517634037309377, + "learning_rate": 3.9749497199631724e-07, + "loss": 0.9398, + "step": 1876 + }, + { + "epoch": 4.8005115089514065, + "grad_norm": 0.10322694629271818, + "learning_rate": 3.875116763024611e-07, + "loss": 0.9813, + "step": 1877 + }, + { + "epoch": 4.80306905370844, + "grad_norm": 0.11224775443862717, + "learning_rate": 3.7765473783233807e-07, + "loss": 0.9674, + "step": 1878 + }, + { + "epoch": 4.805626598465473, + "grad_norm": 0.10356896266412557, + "learning_rate": 3.6792418802794386e-07, + "loss": 0.952, + "step": 1879 + }, + { + "epoch": 4.8081841432225065, + "grad_norm": 0.1064632218381248, + "learning_rate": 3.583200579280943e-07, + "loss": 0.9479, + "step": 1880 + }, + { + "epoch": 4.810741687979539, + "grad_norm": 0.10668965210736751, + "learning_rate": 3.4884237816837696e-07, + "loss": 0.979, + "step": 1881 + }, + { + "epoch": 4.813299232736573, + "grad_norm": 0.11234432110553998, + "learning_rate": 3.394911789809996e-07, + "loss": 0.9489, + "step": 1882 + }, + { + "epoch": 4.8158567774936065, + "grad_norm": 0.10305811442615184, + "learning_rate": 3.3026649019471946e-07, + "loss": 0.9401, + "step": 1883 + }, + { + "epoch": 4.818414322250639, + "grad_norm": 0.10655521657426903, + "learning_rate": 3.2116834123476326e-07, + "loss": 0.9613, + "step": 1884 + }, + { + "epoch": 4.820971867007673, + "grad_norm": 0.10913574692316319, + "learning_rate": 3.1219676112270723e-07, + "loss": 0.962, + "step": 1885 + }, + { + "epoch": 4.823529411764706, + "grad_norm": 0.10237237844134321, + "learning_rate": 3.033517784763884e-07, + "loss": 0.9778, + "step": 1886 + }, + { + "epoch": 4.826086956521739, + "grad_norm": 0.10794182864566411, + "learning_rate": 2.94633421509829e-07, + "loss": 0.9765, + "step": 1887 + }, + { + "epoch": 4.828644501278772, + "grad_norm": 0.10761359006799004, + "learning_rate": 2.8604171803312986e-07, + "loss": 0.9748, + "step": 1888 + }, + { + "epoch": 4.831202046035806, + "grad_norm": 0.10699503894718207, + "learning_rate": 2.775766954523906e-07, + "loss": 0.9611, + "step": 1889 + }, + { + "epoch": 4.833759590792839, + "grad_norm": 0.10534467355372792, + "learning_rate": 2.692383807696164e-07, + "loss": 0.941, + "step": 1890 + }, + { + "epoch": 4.836317135549872, + "grad_norm": 0.10931285876643325, + "learning_rate": 2.6102680058264664e-07, + "loss": 0.977, + "step": 1891 + }, + { + "epoch": 4.838874680306906, + "grad_norm": 0.10276275524204001, + "learning_rate": 2.5294198108504865e-07, + "loss": 0.9424, + "step": 1892 + }, + { + "epoch": 4.841432225063938, + "grad_norm": 0.10844571118942553, + "learning_rate": 2.449839480660421e-07, + "loss": 0.9217, + "step": 1893 + }, + { + "epoch": 4.843989769820972, + "grad_norm": 0.10833347969842211, + "learning_rate": 2.371527269104279e-07, + "loss": 0.9712, + "step": 1894 + }, + { + "epoch": 4.846547314578006, + "grad_norm": 0.11577418267184579, + "learning_rate": 2.2944834259849058e-07, + "loss": 0.961, + "step": 1895 + }, + { + "epoch": 4.849104859335038, + "grad_norm": 0.10873181639986831, + "learning_rate": 2.2187081970593605e-07, + "loss": 0.9934, + "step": 1896 + }, + { + "epoch": 4.851662404092072, + "grad_norm": 0.10735020699640388, + "learning_rate": 2.1442018240378948e-07, + "loss": 0.9582, + "step": 1897 + }, + { + "epoch": 4.854219948849105, + "grad_norm": 0.10907777804447821, + "learning_rate": 2.0709645445834646e-07, + "loss": 0.9714, + "step": 1898 + }, + { + "epoch": 4.856777493606138, + "grad_norm": 0.1096334655026699, + "learning_rate": 1.998996592310709e-07, + "loss": 1.0043, + "step": 1899 + }, + { + "epoch": 4.859335038363171, + "grad_norm": 0.11043084155630652, + "learning_rate": 1.928298196785372e-07, + "loss": 0.9557, + "step": 1900 + }, + { + "epoch": 4.861892583120205, + "grad_norm": 0.10553296566509356, + "learning_rate": 1.8588695835235926e-07, + "loss": 0.9563, + "step": 1901 + }, + { + "epoch": 4.864450127877237, + "grad_norm": 0.10821191093088821, + "learning_rate": 1.790710973991061e-07, + "loss": 0.9618, + "step": 1902 + }, + { + "epoch": 4.867007672634271, + "grad_norm": 0.11019549559652472, + "learning_rate": 1.723822585602264e-07, + "loss": 0.9745, + "step": 1903 + }, + { + "epoch": 4.869565217391305, + "grad_norm": 0.10707545546640006, + "learning_rate": 1.658204631720084e-07, + "loss": 0.9847, + "step": 1904 + }, + { + "epoch": 4.872122762148337, + "grad_norm": 0.10806355656419639, + "learning_rate": 1.5938573216547792e-07, + "loss": 0.9789, + "step": 1905 + }, + { + "epoch": 4.874680306905371, + "grad_norm": 0.10651089976914624, + "learning_rate": 1.5307808606635832e-07, + "loss": 0.9662, + "step": 1906 + }, + { + "epoch": 4.877237851662404, + "grad_norm": 0.10490157621191196, + "learning_rate": 1.4689754499498166e-07, + "loss": 0.9715, + "step": 1907 + }, + { + "epoch": 4.879795396419437, + "grad_norm": 0.10593964096398936, + "learning_rate": 1.408441286662443e-07, + "loss": 0.9947, + "step": 1908 + }, + { + "epoch": 4.882352941176471, + "grad_norm": 0.10970227452154628, + "learning_rate": 1.3491785638953592e-07, + "loss": 1.0021, + "step": 1909 + }, + { + "epoch": 4.884910485933504, + "grad_norm": 0.10530476523192196, + "learning_rate": 1.291187470686728e-07, + "loss": 0.9644, + "step": 1910 + }, + { + "epoch": 4.887468030690537, + "grad_norm": 0.10894549145015038, + "learning_rate": 1.2344681920184455e-07, + "loss": 0.9445, + "step": 1911 + }, + { + "epoch": 4.89002557544757, + "grad_norm": 0.10932447169022794, + "learning_rate": 1.1790209088155646e-07, + "loss": 0.987, + "step": 1912 + }, + { + "epoch": 4.892583120204604, + "grad_norm": 0.10669046030097515, + "learning_rate": 1.1248457979456729e-07, + "loss": 0.9693, + "step": 1913 + }, + { + "epoch": 4.8951406649616365, + "grad_norm": 0.10317584987321615, + "learning_rate": 1.0719430322183144e-07, + "loss": 0.9783, + "step": 1914 + }, + { + "epoch": 4.89769820971867, + "grad_norm": 0.10544431476402813, + "learning_rate": 1.0203127803844137e-07, + "loss": 0.9698, + "step": 1915 + }, + { + "epoch": 4.900255754475703, + "grad_norm": 0.1059139406234035, + "learning_rate": 9.699552071359642e-08, + "loss": 0.9935, + "step": 1916 + }, + { + "epoch": 4.9028132992327365, + "grad_norm": 0.10842738215390242, + "learning_rate": 9.208704731051399e-08, + "loss": 0.9509, + "step": 1917 + }, + { + "epoch": 4.90537084398977, + "grad_norm": 0.10719633245807475, + "learning_rate": 8.73058734864074e-08, + "loss": 0.9255, + "step": 1918 + }, + { + "epoch": 4.907928388746803, + "grad_norm": 0.10801419457325277, + "learning_rate": 8.265201449242367e-08, + "loss": 0.9974, + "step": 1919 + }, + { + "epoch": 4.910485933503836, + "grad_norm": 0.10217294119273099, + "learning_rate": 7.812548517359908e-08, + "loss": 0.9625, + "step": 1920 + }, + { + "epoch": 4.913043478260869, + "grad_norm": 0.10904336094462923, + "learning_rate": 7.37262999687971e-08, + "loss": 0.9778, + "step": 1921 + }, + { + "epoch": 4.915601023017903, + "grad_norm": 0.10284703646879777, + "learning_rate": 6.945447291069052e-08, + "loss": 0.9712, + "step": 1922 + }, + { + "epoch": 4.918158567774936, + "grad_norm": 0.106153765145594, + "learning_rate": 6.531001762569044e-08, + "loss": 0.9557, + "step": 1923 + }, + { + "epoch": 4.920716112531969, + "grad_norm": 0.1074054901476557, + "learning_rate": 6.129294733391966e-08, + "loss": 0.9581, + "step": 1924 + }, + { + "epoch": 4.923273657289003, + "grad_norm": 0.10969417040676402, + "learning_rate": 5.740327484916819e-08, + "loss": 0.9918, + "step": 1925 + }, + { + "epoch": 4.9258312020460355, + "grad_norm": 0.10657300074043048, + "learning_rate": 5.364101257883114e-08, + "loss": 0.9898, + "step": 1926 + }, + { + "epoch": 4.928388746803069, + "grad_norm": 0.10352327904399762, + "learning_rate": 5.000617252389983e-08, + "loss": 0.9926, + "step": 1927 + }, + { + "epoch": 4.930946291560103, + "grad_norm": 0.10995132445135233, + "learning_rate": 4.649876627890848e-08, + "loss": 0.9555, + "step": 1928 + }, + { + "epoch": 4.9335038363171355, + "grad_norm": 0.10681061794840668, + "learning_rate": 4.311880503189869e-08, + "loss": 0.9761, + "step": 1929 + }, + { + "epoch": 4.936061381074169, + "grad_norm": 0.10628340512917848, + "learning_rate": 3.986629956437948e-08, + "loss": 0.9981, + "step": 1930 + }, + { + "epoch": 4.938618925831202, + "grad_norm": 0.10958658587579229, + "learning_rate": 3.674126025130509e-08, + "loss": 0.9968, + "step": 1931 + }, + { + "epoch": 4.9411764705882355, + "grad_norm": 0.1078617025355873, + "learning_rate": 3.374369706102609e-08, + "loss": 0.978, + "step": 1932 + }, + { + "epoch": 4.943734015345268, + "grad_norm": 0.10200017383938678, + "learning_rate": 3.0873619555267244e-08, + "loss": 0.9466, + "step": 1933 + }, + { + "epoch": 4.946291560102302, + "grad_norm": 0.10404264076360889, + "learning_rate": 2.8131036889096352e-08, + "loss": 0.9722, + "step": 1934 + }, + { + "epoch": 4.948849104859335, + "grad_norm": 0.10619021083501325, + "learning_rate": 2.5515957810897663e-08, + "loss": 0.956, + "step": 1935 + }, + { + "epoch": 4.951406649616368, + "grad_norm": 0.10572319413582357, + "learning_rate": 2.302839066233631e-08, + "loss": 0.9641, + "step": 1936 + }, + { + "epoch": 4.953964194373402, + "grad_norm": 0.10543925313697497, + "learning_rate": 2.066834337833168e-08, + "loss": 0.9671, + "step": 1937 + }, + { + "epoch": 4.956521739130435, + "grad_norm": 0.10450226460470871, + "learning_rate": 1.843582348704409e-08, + "loss": 0.9409, + "step": 1938 + }, + { + "epoch": 4.959079283887468, + "grad_norm": 0.10771140839710609, + "learning_rate": 1.6330838109839266e-08, + "loss": 0.9782, + "step": 1939 + }, + { + "epoch": 4.961636828644501, + "grad_norm": 0.24834469255449557, + "learning_rate": 1.4353393961270556e-08, + "loss": 0.9874, + "step": 1940 + }, + { + "epoch": 4.964194373401535, + "grad_norm": 0.10753233482520143, + "learning_rate": 1.2503497349047877e-08, + "loss": 0.9653, + "step": 1941 + }, + { + "epoch": 4.966751918158568, + "grad_norm": 0.11018037804047005, + "learning_rate": 1.078115417404213e-08, + "loss": 0.977, + "step": 1942 + }, + { + "epoch": 4.969309462915601, + "grad_norm": 0.10439962360210749, + "learning_rate": 9.18636993023192e-09, + "loss": 0.9729, + "step": 1943 + }, + { + "epoch": 4.971867007672635, + "grad_norm": 0.10236667855362652, + "learning_rate": 7.719149704716878e-09, + "loss": 0.9452, + "step": 1944 + }, + { + "epoch": 4.974424552429667, + "grad_norm": 0.10489163848161775, + "learning_rate": 6.379498177686572e-09, + "loss": 0.9638, + "step": 1945 + }, + { + "epoch": 4.976982097186701, + "grad_norm": 0.11428348011942069, + "learning_rate": 5.167419622402747e-09, + "loss": 1.0006, + "step": 1946 + }, + { + "epoch": 4.979539641943734, + "grad_norm": 0.1067592882173923, + "learning_rate": 4.082917905194883e-09, + "loss": 0.9504, + "step": 1947 + }, + { + "epoch": 4.982097186700767, + "grad_norm": 0.10497889316311938, + "learning_rate": 3.1259964854379877e-09, + "loss": 0.9662, + "step": 1948 + }, + { + "epoch": 4.9846547314578, + "grad_norm": 0.10799110314474006, + "learning_rate": 2.2966584155570403e-09, + "loss": 0.9928, + "step": 1949 + }, + { + "epoch": 4.987212276214834, + "grad_norm": 0.10574572409471053, + "learning_rate": 1.5949063410003462e-09, + "loss": 0.9916, + "step": 1950 + }, + { + "epoch": 4.989769820971867, + "grad_norm": 0.10267660554290474, + "learning_rate": 1.0207425002395355e-09, + "loss": 0.9555, + "step": 1951 + }, + { + "epoch": 4.9923273657289, + "grad_norm": 0.10468023161937465, + "learning_rate": 5.74168724760682e-10, + "loss": 0.9803, + "step": 1952 + }, + { + "epoch": 4.994884910485934, + "grad_norm": 0.10454422969262668, + "learning_rate": 2.551864390598624e-10, + "loss": 0.9432, + "step": 1953 + }, + { + "epoch": 4.997442455242966, + "grad_norm": 0.10571613867824278, + "learning_rate": 6.379666063871525e-11, + "loss": 0.9505, + "step": 1954 + }, + { + "epoch": 5.0, + "grad_norm": 0.10885932868024188, + "learning_rate": 0.0, + "loss": 0.971, + "step": 1955 + }, + { + "epoch": 5.0, + "step": 1955, + "total_flos": 7122204608430080.0, + "train_loss": 1.0287959775656386, + "train_runtime": 35452.5316, + "train_samples_per_second": 14.103, + "train_steps_per_second": 0.055 + } + ], + "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..1d5fcfe --- /dev/null +++ b/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:74daa30c1246b315d77332da6673207490f6c4e30d218a9262c196b38f2bb054 +size 7224 diff --git a/training_loss.png b/training_loss.png new file mode 100644 index 0000000..c47e9c5 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