Upload folder using huggingface_hub

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,61 @@
+---
+library_name: transformers
+license: other
+base_model: Qwen/Qwen3-30B-A3B-Base
+tags:
+- llama-factory
+- full
+- generated_from_trainer
+model-index:
+- name: base_gsm_bs32_lr1e-5_zero3
+  results: []
+---
+
+<!-- This model card has been generated automatically according to the information the Trainer had access to. You
+should probably proofread and complete it, then remove this comment. -->
+
+# base_gsm_bs32_lr1e-5_zero3
+
+This model is a fine-tuned version of [Qwen/Qwen3-30B-A3B-Base](https://huggingface.co/Qwen/Qwen3-30B-A3B-Base) on the gsm 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: 1e-05
+- train_batch_size: 8
+- eval_batch_size: 8
+- seed: 42
+- distributed_type: multi-GPU
+- num_devices: 8
+- gradient_accumulation_steps: 4
+- total_train_batch_size: 256
+- total_eval_batch_size: 64
+- 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: 3.0
+
+### Training results
+
+
+
+### Framework versions
+
+- Transformers 4.51.3
+- Pytorch 2.6.0+cu124
+- Datasets 3.5.0
+- Tokenizers 0.21.1

added_tokens.json

@@ -0,0 +1,28 @@
+{
+  "</think>": 151668,
+  "</tool_call>": 151658,
+  "</tool_response>": 151666,
+  "<think>": 151667,
+  "<tool_call>": 151657,
+  "<tool_response>": 151665,
+  "<|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
+}

all_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 2.927536231884058,
+    "total_flos": 10226403115008.0,
+    "train_loss": 0.12416269268621416,
+    "train_runtime": 5225.659,
+    "train_samples_per_second": 5.047,
+    "train_steps_per_second": 0.02
+}

config.json

@@ -0,0 +1,38 @@
+{
+  "architectures": [
+    "Qwen3MoeForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "decoder_sparse_step": 1,
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 2048,
+  "initializer_range": 0.02,
+  "intermediate_size": 6144,
+  "max_position_embeddings": 32768,
+  "max_window_layers": 48,
+  "mlp_only_layers": [],
+  "model_type": "qwen3_moe",
+  "moe_intermediate_size": 768,
+  "norm_topk_prob": true,
+  "num_attention_heads": 32,
+  "num_experts": 128,
+  "num_experts_per_tok": 8,
+  "num_hidden_layers": 48,
+  "num_key_value_heads": 4,
+  "output_router_logits": false,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000.0,
+  "router_aux_loss_coef": 0.001,
+  "sliding_window": null,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.51.3",
+  "use_cache": false,
+  "use_sliding_window": false,
+  "vocab_size": 151936
+}

generation_config.json

@@ -0,0 +1,6 @@
+{
+  "bos_token_id": 151643,
+  "eos_token_id": 151643,
+  "max_new_tokens": 2048,
+  "transformers_version": "4.51.3"
+}

model-00001-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3ee53e065505f8d759bd943f84df041af27196c3bcd6c201a18186409e7f1c33
+size 4997184968

model-00002-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c3aeaa1b487c55a16712310f7175db73c9fcf2192771deddbd41e1f0e131bc3a
+size 4997741608

model-00003-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7b6e840081cfa63a27706c5e0b8a99255604fd6bf646d71c0e6ddd15c06e7e8c
+size 4997742208

model-00004-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4d94b0bd027f00c6757aa4a59af9621bca5c46fd3a0055d60e3e39657bf5c5b1
+size 4997743184

model-00005-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3233dc56093d12944cc8c2df15a52e1b0d754d56d4af626bf2fbc528eb75cec5
+size 4997743184

model-00006-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e6dde003e80ccb1a53d9b3a414ebda0f86c240c96edfdf489a23d016d3b2d16e
+size 4997743184

model-00007-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e4475914f1f5e47d051bbca6b379c0102b9357b27d5b06953c86371067c1ea24
+size 4997743184

model-00008-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ddf0383022e177b143f12ad208c55c4b39d7cb39b1bb120ebd39d54f0ccfd66b
+size 4997743184

model-00009-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6c41fc9fd8c076d60bbd7f3c9bb8e610bdb19222dc6064b4f14c9bc91a9698ae
+size 4997743184

model-00010-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:33b4e7d6f59e3fb88196b7a02de0ed60af0efb8ff8b64d5926d72db57dc26750
+size 4997743184

model-00011-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4159754a77a76caa3e3b99733214b00c5a9ee60eb7e5f9f016d1b43a1ae11b75
+size 4997743184

model-00012-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3ccf622173db595f0bf0196492c38c9c8591d01b68794e8e3b809a0803aa9021
+size 4997743184

model-00013-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2d0567ecb5ae3a5946ab1363977668dc65b32a45a883fd98d7b324333b0b2aec
+size 1094220288

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": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aeb13307a71acd8fe81861d94ad54ab689df773318809eed3cbe794b4492dae4
+size 11422654

tokenizer_config.json

@@ -0,0 +1,241 @@
+{
+  "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": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "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
+    },
+    "151665": {
+      "content": "<tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151666": {
+      "content": "</tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151667": {
+      "content": "<think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151668": {
+      "content": "</think>",
+      "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\\n' }}\n    {%- endif %}\n    {{- \"# Tools\\n\\nYou may call one or more functions to assist with the user query.\\n\\nYou are provided with function signatures within <tools></tools> XML tags:\\n<tools>\" }}\n    {%- for tool in tools %}\n        {{- \"\\n\" }}\n        {{- tool | tojson }}\n    {%- endfor %}\n    {{- \"\\n</tools>\\n\\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\\n<tool_call>\\n{\\\"name\\\": <function-name>, \\\"arguments\\\": <args-json-object>}\\n</tool_call><|im_end|>\\n\" }}\n{%- else %}\n    {%- if messages[0].role == 'system' %}\n        {{- '<|im_start|>system\\n' + messages[0].content + '<|im_end|>\\n' }}\n    {%- endif %}\n{%- endif %}\n{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}\n{%- for message in messages[::-1] %}\n    {%- set index = (messages|length - 1) - loop.index0 %}\n    {%- if ns.multi_step_tool and message.role == \"user\" and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}\n        {%- set ns.multi_step_tool = false %}\n        {%- set ns.last_query_index = index %}\n    {%- endif %}\n{%- endfor %}\n{%- for message in messages %}\n    {%- if (message.role == \"user\") or (message.role == \"system\" and not loop.first) %}\n        {{- '<|im_start|>' + message.role + '\\n' + message.content + '<|im_end|>' + '\\n' }}\n    {%- elif message.role == \"assistant\" %}\n        {%- set content = message.content %}\n        {%- set reasoning_content = '' %}\n        {%- if message.reasoning_content is defined and message.reasoning_content is not none %}\n            {%- set reasoning_content = message.reasoning_content %}\n        {%- else %}\n            {%- if '</think>' in message.content %}\n                {%- set content = message.content.split('</think>')[-1].lstrip('\\n') %}\n                {%- set reasoning_content = message.content.split('</think>')[0].rstrip('\\n').split('<think>')[-1].lstrip('\\n') %}\n            {%- endif %}\n        {%- endif %}\n        {%- if loop.index0 > ns.last_query_index %}\n            {%- if loop.last or (not loop.last and reasoning_content) %}\n                {{- '<|im_start|>' + message.role + '\\n<think>\\n' + reasoning_content.strip('\\n') + '\\n</think>\\n\\n' + content.lstrip('\\n') }}\n            {%- else %}\n                {{- '<|im_start|>' + message.role + '\\n' + content }}\n            {%- endif %}\n        {%- else %}\n            {{- '<|im_start|>' + message.role + '\\n' + content }}\n        {%- endif %}\n        {%- if message.tool_calls %}\n            {%- for tool_call in message.tool_calls %}\n                {%- if (loop.first and content) or (not loop.first) %}\n                    {{- '\\n' }}\n                {%- endif %}\n                {%- if tool_call.function %}\n                    {%- set tool_call = tool_call.function %}\n                {%- endif %}\n                {{- '<tool_call>\\n{\"name\": \"' }}\n                {{- tool_call.name }}\n                {{- '\", \"arguments\": ' }}\n                {%- if tool_call.arguments is string %}\n                    {{- tool_call.arguments }}\n                {%- else %}\n                    {{- tool_call.arguments | tojson }}\n                {%- endif %}\n                {{- '}\\n</tool_call>' }}\n            {%- endfor %}\n        {%- endif %}\n        {{- '<|im_end|>\\n' }}\n    {%- elif message.role == \"tool\" %}\n        {%- if loop.first or (messages[loop.index0 - 1].role != \"tool\") %}\n            {{- '<|im_start|>user' }}\n        {%- endif %}\n        {{- '\\n<tool_response>\\n' }}\n        {{- message.content }}\n        {{- '\\n</tool_response>' }}\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    {%- if enable_thinking is defined and enable_thinking is false %}\n        {{- '<think>\\n\\n</think>\\n\\n' }}\n    {%- endif %}\n{%- endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "padding_side": "right",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

train_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 2.927536231884058,
+    "total_flos": 10226403115008.0,
+    "train_loss": 0.12416269268621416,
+    "train_runtime": 5225.659,
+    "train_samples_per_second": 5.047,
+    "train_steps_per_second": 0.02
+}

trainer_log.jsonl

@@ -0,0 +1,103 @@
+{"current_steps": 1, "total_steps": 102, "loss": 0.6357, "lr": 0.0, "epoch": 0.028985507246376812, "percentage": 0.98, "elapsed_time": "0:00:58", "remaining_time": "1:38:52"}
+{"current_steps": 2, "total_steps": 102, "loss": 0.6666, "lr": 9.090909090909091e-07, "epoch": 0.057971014492753624, "percentage": 1.96, "elapsed_time": "0:01:47", "remaining_time": "1:29:13"}
+{"current_steps": 3, "total_steps": 102, "loss": 0.6451, "lr": 1.8181818181818183e-06, "epoch": 0.08695652173913043, "percentage": 2.94, "elapsed_time": "0:02:35", "remaining_time": "1:25:21"}
+{"current_steps": 4, "total_steps": 102, "loss": 0.6048, "lr": 2.7272727272727272e-06, "epoch": 0.11594202898550725, "percentage": 3.92, "elapsed_time": "0:03:23", "remaining_time": "1:23:00"}
+{"current_steps": 5, "total_steps": 102, "loss": 0.5937, "lr": 3.6363636363636366e-06, "epoch": 0.14492753623188406, "percentage": 4.9, "elapsed_time": "0:04:11", "remaining_time": "1:21:22"}
+{"current_steps": 6, "total_steps": 102, "loss": 0.4943, "lr": 4.5454545454545455e-06, "epoch": 0.17391304347826086, "percentage": 5.88, "elapsed_time": "0:04:59", "remaining_time": "1:19:53"}
+{"current_steps": 7, "total_steps": 102, "loss": 0.4247, "lr": 5.4545454545454545e-06, "epoch": 0.2028985507246377, "percentage": 6.86, "elapsed_time": "0:05:48", "remaining_time": "1:18:50"}
+{"current_steps": 8, "total_steps": 102, "loss": 0.302, "lr": 6.363636363636364e-06, "epoch": 0.2318840579710145, "percentage": 7.84, "elapsed_time": "0:06:37", "remaining_time": "1:17:46"}
+{"current_steps": 9, "total_steps": 102, "loss": 0.2751, "lr": 7.272727272727273e-06, "epoch": 0.2608695652173913, "percentage": 8.82, "elapsed_time": "0:07:25", "remaining_time": "1:16:39"}
+{"current_steps": 10, "total_steps": 102, "loss": 0.2311, "lr": 8.181818181818183e-06, "epoch": 0.2898550724637681, "percentage": 9.8, "elapsed_time": "0:08:12", "remaining_time": "1:15:35"}
+{"current_steps": 11, "total_steps": 102, "loss": 0.2133, "lr": 9.090909090909091e-06, "epoch": 0.3188405797101449, "percentage": 10.78, "elapsed_time": "0:09:01", "remaining_time": "1:14:37"}
+{"current_steps": 12, "total_steps": 102, "loss": 0.2021, "lr": 1e-05, "epoch": 0.34782608695652173, "percentage": 11.76, "elapsed_time": "0:09:49", "remaining_time": "1:13:39"}
+{"current_steps": 13, "total_steps": 102, "loss": 0.2086, "lr": 9.997020702755353e-06, "epoch": 0.37681159420289856, "percentage": 12.75, "elapsed_time": "0:10:37", "remaining_time": "1:12:41"}
+{"current_steps": 14, "total_steps": 102, "loss": 0.1844, "lr": 9.98808636150624e-06, "epoch": 0.4057971014492754, "percentage": 13.73, "elapsed_time": "0:11:25", "remaining_time": "1:11:46"}
+{"current_steps": 15, "total_steps": 102, "loss": 0.1561, "lr": 9.973207623475964e-06, "epoch": 0.43478260869565216, "percentage": 14.71, "elapsed_time": "0:12:12", "remaining_time": "1:10:51"}
+{"current_steps": 16, "total_steps": 102, "loss": 0.1561, "lr": 9.952402219937817e-06, "epoch": 0.463768115942029, "percentage": 15.69, "elapsed_time": "0:13:01", "remaining_time": "1:10:01"}
+{"current_steps": 17, "total_steps": 102, "loss": 0.1359, "lr": 9.925694945084369e-06, "epoch": 0.4927536231884058, "percentage": 16.67, "elapsed_time": "0:13:50", "remaining_time": "1:09:12"}
+{"current_steps": 18, "total_steps": 102, "loss": 0.1422, "lr": 9.893117626479778e-06, "epoch": 0.5217391304347826, "percentage": 17.65, "elapsed_time": "0:14:38", "remaining_time": "1:08:19"}
+{"current_steps": 19, "total_steps": 102, "loss": 0.1448, "lr": 9.854709087130261e-06, "epoch": 0.5507246376811594, "percentage": 18.63, "elapsed_time": "0:15:26", "remaining_time": "1:07:26"}
+{"current_steps": 20, "total_steps": 102, "loss": 0.1276, "lr": 9.810515099218004e-06, "epoch": 0.5797101449275363, "percentage": 19.61, "elapsed_time": "0:16:14", "remaining_time": "1:06:34"}
+{"current_steps": 21, "total_steps": 102, "loss": 0.1177, "lr": 9.76058832955357e-06, "epoch": 0.6086956521739131, "percentage": 20.59, "elapsed_time": "0:17:02", "remaining_time": "1:05:43"}
+{"current_steps": 22, "total_steps": 102, "loss": 0.1086, "lr": 9.704988276811883e-06, "epoch": 0.6376811594202898, "percentage": 21.57, "elapsed_time": "0:17:50", "remaining_time": "1:04:52"}
+{"current_steps": 23, "total_steps": 102, "loss": 0.1291, "lr": 9.643781200626512e-06, "epoch": 0.6666666666666666, "percentage": 22.55, "elapsed_time": "0:18:38", "remaining_time": "1:04:01"}
+{"current_steps": 24, "total_steps": 102, "loss": 0.1183, "lr": 9.577040042626832e-06, "epoch": 0.6956521739130435, "percentage": 23.53, "elapsed_time": "0:19:27", "remaining_time": "1:03:12"}
+{"current_steps": 25, "total_steps": 102, "loss": 0.1213, "lr": 9.504844339512096e-06, "epoch": 0.7246376811594203, "percentage": 24.51, "elapsed_time": "0:20:15", "remaining_time": "1:02:24"}
+{"current_steps": 26, "total_steps": 102, "loss": 0.1177, "lr": 9.427280128266049e-06, "epoch": 0.7536231884057971, "percentage": 25.49, "elapsed_time": "0:21:04", "remaining_time": "1:01:35"}
+{"current_steps": 27, "total_steps": 102, "loss": 0.1152, "lr": 9.344439843625034e-06, "epoch": 0.782608695652174, "percentage": 26.47, "elapsed_time": "0:21:52", "remaining_time": "1:00:46"}
+{"current_steps": 28, "total_steps": 102, "loss": 0.1246, "lr": 9.256422207921757e-06, "epoch": 0.8115942028985508, "percentage": 27.45, "elapsed_time": "0:22:40", "remaining_time": "0:59:56"}
+{"current_steps": 29, "total_steps": 102, "loss": 0.1184, "lr": 9.163332113436031e-06, "epoch": 0.8405797101449275, "percentage": 28.43, "elapsed_time": "0:23:28", "remaining_time": "0:59:06"}
+{"current_steps": 30, "total_steps": 102, "loss": 0.1096, "lr": 9.065280497392663e-06, "epoch": 0.8695652173913043, "percentage": 29.41, "elapsed_time": "0:24:17", "remaining_time": "0:58:16"}
+{"current_steps": 31, "total_steps": 102, "loss": 0.1195, "lr": 8.962384209755453e-06, "epoch": 0.8985507246376812, "percentage": 30.39, "elapsed_time": "0:25:04", "remaining_time": "0:57:26"}
+{"current_steps": 32, "total_steps": 102, "loss": 0.1105, "lr": 8.854765873974898e-06, "epoch": 0.927536231884058, "percentage": 31.37, "elapsed_time": "0:25:52", "remaining_time": "0:56:37"}
+{"current_steps": 33, "total_steps": 102, "loss": 0.1175, "lr": 8.742553740855507e-06, "epoch": 0.9565217391304348, "percentage": 32.35, "elapsed_time": "0:26:41", "remaining_time": "0:55:48"}
+{"current_steps": 34, "total_steps": 102, "loss": 0.1058, "lr": 8.625881535716883e-06, "epoch": 0.9855072463768116, "percentage": 33.33, "elapsed_time": "0:27:30", "remaining_time": "0:55:00"}
+{"current_steps": 35, "total_steps": 102, "loss": 0.1001, "lr": 8.504888299030748e-06, "epoch": 1.0, "percentage": 34.31, "elapsed_time": "0:27:53", "remaining_time": "0:53:24"}
+{"current_steps": 36, "total_steps": 102, "loss": 0.0727, "lr": 8.379718220723772e-06, "epoch": 1.0289855072463767, "percentage": 35.29, "elapsed_time": "0:30:31", "remaining_time": "0:55:58"}
+{"current_steps": 37, "total_steps": 102, "loss": 0.0871, "lr": 8.250520468343722e-06, "epoch": 1.0579710144927537, "percentage": 36.27, "elapsed_time": "0:31:19", "remaining_time": "0:55:02"}
+{"current_steps": 38, "total_steps": 102, "loss": 0.0752, "lr": 8.117449009293668e-06, "epoch": 1.0869565217391304, "percentage": 37.25, "elapsed_time": "0:32:07", "remaining_time": "0:54:06"}
+{"current_steps": 39, "total_steps": 102, "loss": 0.0862, "lr": 7.980662427346127e-06, "epoch": 1.1159420289855073, "percentage": 38.24, "elapsed_time": "0:32:55", "remaining_time": "0:53:11"}
+{"current_steps": 40, "total_steps": 102, "loss": 0.0928, "lr": 7.84032373365578e-06, "epoch": 1.144927536231884, "percentage": 39.22, "elapsed_time": "0:33:44", "remaining_time": "0:52:17"}
+{"current_steps": 41, "total_steps": 102, "loss": 0.0757, "lr": 7.696600172495997e-06, "epoch": 1.1739130434782608, "percentage": 40.2, "elapsed_time": "0:34:33", "remaining_time": "0:51:24"}
+{"current_steps": 42, "total_steps": 102, "loss": 0.0621, "lr": 7.5496630219506805e-06, "epoch": 1.2028985507246377, "percentage": 41.18, "elapsed_time": "0:35:23", "remaining_time": "0:50:33"}
+{"current_steps": 43, "total_steps": 102, "loss": 0.0855, "lr": 7.399687389798933e-06, "epoch": 1.2318840579710144, "percentage": 42.16, "elapsed_time": "0:36:11", "remaining_time": "0:49:38"}
+{"current_steps": 44, "total_steps": 102, "loss": 0.084, "lr": 7.246852004835807e-06, "epoch": 1.2608695652173914, "percentage": 43.14, "elapsed_time": "0:36:59", "remaining_time": "0:48:45"}
+{"current_steps": 45, "total_steps": 102, "loss": 0.0767, "lr": 7.091339003877826e-06, "epoch": 1.289855072463768, "percentage": 44.12, "elapsed_time": "0:37:47", "remaining_time": "0:47:52"}
+{"current_steps": 46, "total_steps": 102, "loss": 0.0806, "lr": 6.933333714707094e-06, "epoch": 1.318840579710145, "percentage": 45.1, "elapsed_time": "0:38:35", "remaining_time": "0:46:58"}
+{"current_steps": 47, "total_steps": 102, "loss": 0.0686, "lr": 6.773024435212678e-06, "epoch": 1.3478260869565217, "percentage": 46.08, "elapsed_time": "0:39:23", "remaining_time": "0:46:06"}
+{"current_steps": 48, "total_steps": 102, "loss": 0.0913, "lr": 6.6106022089924535e-06, "epoch": 1.3768115942028984, "percentage": 47.06, "elapsed_time": "0:40:11", "remaining_time": "0:45:13"}
+{"current_steps": 49, "total_steps": 102, "loss": 0.0746, "lr": 6.4462605976828395e-06, "epoch": 1.4057971014492754, "percentage": 48.04, "elapsed_time": "0:40:59", "remaining_time": "0:44:20"}
+{"current_steps": 50, "total_steps": 102, "loss": 0.0976, "lr": 6.280195450287736e-06, "epoch": 1.434782608695652, "percentage": 49.02, "elapsed_time": "0:41:48", "remaining_time": "0:43:28"}
+{"current_steps": 51, "total_steps": 102, "loss": 0.0833, "lr": 6.112604669781572e-06, "epoch": 1.463768115942029, "percentage": 50.0, "elapsed_time": "0:42:36", "remaining_time": "0:42:36"}
+{"current_steps": 52, "total_steps": 102, "loss": 0.0784, "lr": 5.943687977264584e-06, "epoch": 1.4927536231884058, "percentage": 50.98, "elapsed_time": "0:43:25", "remaining_time": "0:41:45"}
+{"current_steps": 53, "total_steps": 102, "loss": 0.0838, "lr": 5.773646673951406e-06, "epoch": 1.5217391304347827, "percentage": 51.96, "elapsed_time": "0:44:13", "remaining_time": "0:40:53"}
+{"current_steps": 54, "total_steps": 102, "loss": 0.0828, "lr": 5.6026834012766155e-06, "epoch": 1.5507246376811594, "percentage": 52.94, "elapsed_time": "0:45:01", "remaining_time": "0:40:01"}
+{"current_steps": 55, "total_steps": 102, "loss": 0.0797, "lr": 5.4310018994030974e-06, "epoch": 1.5797101449275361, "percentage": 53.92, "elapsed_time": "0:45:49", "remaining_time": "0:39:09"}
+{"current_steps": 56, "total_steps": 102, "loss": 0.0772, "lr": 5.258806764421048e-06, "epoch": 1.608695652173913, "percentage": 54.9, "elapsed_time": "0:46:37", "remaining_time": "0:38:18"}
+{"current_steps": 57, "total_steps": 102, "loss": 0.0766, "lr": 5.0863032045269435e-06, "epoch": 1.6376811594202898, "percentage": 55.88, "elapsed_time": "0:47:25", "remaining_time": "0:37:26"}
+{"current_steps": 58, "total_steps": 102, "loss": 0.0818, "lr": 4.913696795473058e-06, "epoch": 1.6666666666666665, "percentage": 56.86, "elapsed_time": "0:48:13", "remaining_time": "0:36:35"}
+{"current_steps": 59, "total_steps": 102, "loss": 0.0927, "lr": 4.741193235578953e-06, "epoch": 1.6956521739130435, "percentage": 57.84, "elapsed_time": "0:49:02", "remaining_time": "0:35:44"}
+{"current_steps": 60, "total_steps": 102, "loss": 0.0665, "lr": 4.568998100596903e-06, "epoch": 1.7246376811594204, "percentage": 58.82, "elapsed_time": "0:49:52", "remaining_time": "0:34:54"}
+{"current_steps": 61, "total_steps": 102, "loss": 0.0839, "lr": 4.397316598723385e-06, "epoch": 1.7536231884057971, "percentage": 59.8, "elapsed_time": "0:50:40", "remaining_time": "0:34:03"}
+{"current_steps": 62, "total_steps": 102, "loss": 0.0773, "lr": 4.226353326048594e-06, "epoch": 1.7826086956521738, "percentage": 60.78, "elapsed_time": "0:51:28", "remaining_time": "0:33:12"}
+{"current_steps": 63, "total_steps": 102, "loss": 0.0809, "lr": 4.056312022735417e-06, "epoch": 1.8115942028985508, "percentage": 61.76, "elapsed_time": "0:52:16", "remaining_time": "0:32:21"}
+{"current_steps": 64, "total_steps": 102, "loss": 0.0782, "lr": 3.887395330218429e-06, "epoch": 1.8405797101449275, "percentage": 62.75, "elapsed_time": "0:53:04", "remaining_time": "0:31:30"}
+{"current_steps": 65, "total_steps": 102, "loss": 0.0774, "lr": 3.7198045497122647e-06, "epoch": 1.8695652173913042, "percentage": 63.73, "elapsed_time": "0:53:52", "remaining_time": "0:30:40"}
+{"current_steps": 66, "total_steps": 102, "loss": 0.0767, "lr": 3.553739402317162e-06, "epoch": 1.8985507246376812, "percentage": 64.71, "elapsed_time": "0:54:40", "remaining_time": "0:29:49"}
+{"current_steps": 67, "total_steps": 102, "loss": 0.0984, "lr": 3.389397791007548e-06, "epoch": 1.927536231884058, "percentage": 65.69, "elapsed_time": "0:55:29", "remaining_time": "0:28:59"}
+{"current_steps": 68, "total_steps": 102, "loss": 0.074, "lr": 3.226975564787322e-06, "epoch": 1.9565217391304348, "percentage": 66.67, "elapsed_time": "0:56:18", "remaining_time": "0:28:09"}
+{"current_steps": 69, "total_steps": 102, "loss": 0.0785, "lr": 3.0666662852929063e-06, "epoch": 1.9855072463768115, "percentage": 67.65, "elapsed_time": "0:57:05", "remaining_time": "0:27:18"}
+{"current_steps": 70, "total_steps": 102, "loss": 0.0659, "lr": 2.9086609961221758e-06, "epoch": 2.0, "percentage": 68.63, "elapsed_time": "0:57:31", "remaining_time": "0:26:17"}
+{"current_steps": 71, "total_steps": 102, "loss": 0.0501, "lr": 2.7531479951641928e-06, "epoch": 2.028985507246377, "percentage": 69.61, "elapsed_time": "1:00:11", "remaining_time": "0:26:16"}
+{"current_steps": 72, "total_steps": 102, "loss": 0.0487, "lr": 2.6003126102010696e-06, "epoch": 2.0579710144927534, "percentage": 70.59, "elapsed_time": "1:00:59", "remaining_time": "0:25:24"}
+{"current_steps": 73, "total_steps": 102, "loss": 0.051, "lr": 2.450336978049322e-06, "epoch": 2.0869565217391304, "percentage": 71.57, "elapsed_time": "1:01:47", "remaining_time": "0:24:32"}
+{"current_steps": 74, "total_steps": 102, "loss": 0.0564, "lr": 2.3033998275040047e-06, "epoch": 2.1159420289855073, "percentage": 72.55, "elapsed_time": "1:02:37", "remaining_time": "0:23:41"}
+{"current_steps": 75, "total_steps": 102, "loss": 0.0497, "lr": 2.159676266344222e-06, "epoch": 2.1449275362318843, "percentage": 73.53, "elapsed_time": "1:03:25", "remaining_time": "0:22:49"}
+{"current_steps": 76, "total_steps": 102, "loss": 0.0608, "lr": 2.0193375726538737e-06, "epoch": 2.1739130434782608, "percentage": 74.51, "elapsed_time": "1:04:13", "remaining_time": "0:21:58"}
+{"current_steps": 77, "total_steps": 102, "loss": 0.0466, "lr": 1.8825509907063328e-06, "epoch": 2.2028985507246377, "percentage": 75.49, "elapsed_time": "1:05:03", "remaining_time": "0:21:07"}
+{"current_steps": 78, "total_steps": 102, "loss": 0.0519, "lr": 1.7494795316562791e-06, "epoch": 2.2318840579710146, "percentage": 76.47, "elapsed_time": "1:05:52", "remaining_time": "0:20:16"}
+{"current_steps": 79, "total_steps": 102, "loss": 0.0475, "lr": 1.6202817792762283e-06, "epoch": 2.260869565217391, "percentage": 77.45, "elapsed_time": "1:06:40", "remaining_time": "0:19:24"}
+{"current_steps": 80, "total_steps": 102, "loss": 0.0439, "lr": 1.4951117009692528e-06, "epoch": 2.289855072463768, "percentage": 78.43, "elapsed_time": "1:07:28", "remaining_time": "0:18:33"}
+{"current_steps": 81, "total_steps": 102, "loss": 0.0447, "lr": 1.374118464283119e-06, "epoch": 2.318840579710145, "percentage": 79.41, "elapsed_time": "1:08:16", "remaining_time": "0:17:42"}
+{"current_steps": 82, "total_steps": 102, "loss": 0.0398, "lr": 1.257446259144494e-06, "epoch": 2.3478260869565215, "percentage": 80.39, "elapsed_time": "1:09:05", "remaining_time": "0:16:51"}
+{"current_steps": 83, "total_steps": 102, "loss": 0.0522, "lr": 1.145234126025102e-06, "epoch": 2.3768115942028984, "percentage": 81.37, "elapsed_time": "1:09:53", "remaining_time": "0:16:00"}
+{"current_steps": 84, "total_steps": 102, "loss": 0.0542, "lr": 1.037615790244549e-06, "epoch": 2.4057971014492754, "percentage": 82.35, "elapsed_time": "1:10:42", "remaining_time": "0:15:09"}
+{"current_steps": 85, "total_steps": 102, "loss": 0.053, "lr": 9.347195026073369e-07, "epoch": 2.4347826086956523, "percentage": 83.33, "elapsed_time": "1:11:30", "remaining_time": "0:14:18"}
+{"current_steps": 86, "total_steps": 102, "loss": 0.0533, "lr": 8.366678865639688e-07, "epoch": 2.463768115942029, "percentage": 84.31, "elapsed_time": "1:12:19", "remaining_time": "0:13:27"}
+{"current_steps": 87, "total_steps": 102, "loss": 0.0451, "lr": 7.435777920782444e-07, "epoch": 2.4927536231884058, "percentage": 85.29, "elapsed_time": "1:13:08", "remaining_time": "0:12:36"}
+{"current_steps": 88, "total_steps": 102, "loss": 0.0638, "lr": 6.555601563749675e-07, "epoch": 2.5217391304347827, "percentage": 86.27, "elapsed_time": "1:13:56", "remaining_time": "0:11:45"}
+{"current_steps": 89, "total_steps": 102, "loss": 0.0507, "lr": 5.727198717339511e-07, "epoch": 2.550724637681159, "percentage": 87.25, "elapsed_time": "1:14:44", "remaining_time": "0:10:55"}
+{"current_steps": 90, "total_steps": 102, "loss": 0.0498, "lr": 4.951556604879049e-07, "epoch": 2.579710144927536, "percentage": 88.24, "elapsed_time": "1:15:33", "remaining_time": "0:10:04"}
+{"current_steps": 91, "total_steps": 102, "loss": 0.0487, "lr": 4.2295995737316854e-07, "epoch": 2.608695652173913, "percentage": 89.22, "elapsed_time": "1:16:21", "remaining_time": "0:09:13"}
+{"current_steps": 92, "total_steps": 102, "loss": 0.0409, "lr": 3.5621879937348836e-07, "epoch": 2.63768115942029, "percentage": 90.2, "elapsed_time": "1:17:10", "remaining_time": "0:08:23"}
+{"current_steps": 93, "total_steps": 102, "loss": 0.0464, "lr": 2.9501172318811834e-07, "epoch": 2.6666666666666665, "percentage": 91.18, "elapsed_time": "1:17:58", "remaining_time": "0:07:32"}
+{"current_steps": 94, "total_steps": 102, "loss": 0.0559, "lr": 2.394116704464294e-07, "epoch": 2.6956521739130435, "percentage": 92.16, "elapsed_time": "1:18:46", "remaining_time": "0:06:42"}
+{"current_steps": 95, "total_steps": 102, "loss": 0.0419, "lr": 1.8948490078199767e-07, "epoch": 2.7246376811594204, "percentage": 93.14, "elapsed_time": "1:19:36", "remaining_time": "0:05:51"}
+{"current_steps": 96, "total_steps": 102, "loss": 0.0467, "lr": 1.4529091286973994e-07, "epoch": 2.753623188405797, "percentage": 94.12, "elapsed_time": "1:20:24", "remaining_time": "0:05:01"}
+{"current_steps": 97, "total_steps": 102, "loss": 0.05, "lr": 1.0688237352022346e-07, "epoch": 2.782608695652174, "percentage": 95.1, "elapsed_time": "1:21:12", "remaining_time": "0:04:11"}
+{"current_steps": 98, "total_steps": 102, "loss": 0.0373, "lr": 7.430505491563101e-08, "epoch": 2.8115942028985508, "percentage": 96.08, "elapsed_time": "1:22:00", "remaining_time": "0:03:20"}
+{"current_steps": 99, "total_steps": 102, "loss": 0.0496, "lr": 4.759778006218407e-08, "epoch": 2.8405797101449277, "percentage": 97.06, "elapsed_time": "1:22:48", "remaining_time": "0:02:30"}
+{"current_steps": 100, "total_steps": 102, "loss": 0.0511, "lr": 2.6792376524036878e-08, "epoch": 2.869565217391304, "percentage": 98.04, "elapsed_time": "1:23:36", "remaining_time": "0:01:40"}
+{"current_steps": 101, "total_steps": 102, "loss": 0.0498, "lr": 1.1913638493762369e-08, "epoch": 2.898550724637681, "percentage": 99.02, "elapsed_time": "1:24:25", "remaining_time": "0:00:50"}
+{"current_steps": 102, "total_steps": 102, "loss": 0.0503, "lr": 2.9792972446479605e-09, "epoch": 2.927536231884058, "percentage": 100.0, "elapsed_time": "1:25:13", "remaining_time": "0:00:00"}
+{"current_steps": 102, "total_steps": 102, "epoch": 2.927536231884058, "percentage": 100.0, "elapsed_time": "1:27:03", "remaining_time": "0:00:00"}

trainer_state.json

@@ -0,0 +1,757 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.927536231884058,
+  "eval_steps": 500,
+  "global_step": 102,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.028985507246376812,
+      "grad_norm": 11.274454167048727,
+      "learning_rate": 0.0,
+      "loss": 0.6357,
+      "step": 1
+    },
+    {
+      "epoch": 0.057971014492753624,
+      "grad_norm": 10.143348640958312,
+      "learning_rate": 9.090909090909091e-07,
+      "loss": 0.6666,
+      "step": 2
+    },
+    {
+      "epoch": 0.08695652173913043,
+      "grad_norm": 9.086409087474683,
+      "learning_rate": 1.8181818181818183e-06,
+      "loss": 0.6451,
+      "step": 3
+    },
+    {
+      "epoch": 0.11594202898550725,
+      "grad_norm": 8.70666602748876,
+      "learning_rate": 2.7272727272727272e-06,
+      "loss": 0.6048,
+      "step": 4
+    },
+    {
+      "epoch": 0.14492753623188406,
+      "grad_norm": 8.976482480167713,
+      "learning_rate": 3.6363636363636366e-06,
+      "loss": 0.5937,
+      "step": 5
+    },
+    {
+      "epoch": 0.17391304347826086,
+      "grad_norm": 6.543196052759934,
+      "learning_rate": 4.5454545454545455e-06,
+      "loss": 0.4943,
+      "step": 6
+    },
+    {
+      "epoch": 0.2028985507246377,
+      "grad_norm": 7.511256155710179,
+      "learning_rate": 5.4545454545454545e-06,
+      "loss": 0.4247,
+      "step": 7
+    },
+    {
+      "epoch": 0.2318840579710145,
+      "grad_norm": 3.274069972289533,
+      "learning_rate": 6.363636363636364e-06,
+      "loss": 0.302,
+      "step": 8
+    },
+    {
+      "epoch": 0.2608695652173913,
+      "grad_norm": 5.507527578193957,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 0.2751,
+      "step": 9
+    },
+    {
+      "epoch": 0.2898550724637681,
+      "grad_norm": 3.651481150079746,
+      "learning_rate": 8.181818181818183e-06,
+      "loss": 0.2311,
+      "step": 10
+    },
+    {
+      "epoch": 0.3188405797101449,
+      "grad_norm": 2.9570009811876314,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 0.2133,
+      "step": 11
+    },
+    {
+      "epoch": 0.34782608695652173,
+      "grad_norm": 2.5680951348234355,
+      "learning_rate": 1e-05,
+      "loss": 0.2021,
+      "step": 12
+    },
+    {
+      "epoch": 0.37681159420289856,
+      "grad_norm": 2.4726534615132394,
+      "learning_rate": 9.997020702755353e-06,
+      "loss": 0.2086,
+      "step": 13
+    },
+    {
+      "epoch": 0.4057971014492754,
+      "grad_norm": 2.0168121750509207,
+      "learning_rate": 9.98808636150624e-06,
+      "loss": 0.1844,
+      "step": 14
+    },
+    {
+      "epoch": 0.43478260869565216,
+      "grad_norm": 1.988964439347895,
+      "learning_rate": 9.973207623475964e-06,
+      "loss": 0.1561,
+      "step": 15
+    },
+    {
+      "epoch": 0.463768115942029,
+      "grad_norm": 1.703404936865681,
+      "learning_rate": 9.952402219937817e-06,
+      "loss": 0.1561,
+      "step": 16
+    },
+    {
+      "epoch": 0.4927536231884058,
+      "grad_norm": 1.4224019007386586,
+      "learning_rate": 9.925694945084369e-06,
+      "loss": 0.1359,
+      "step": 17
+    },
+    {
+      "epoch": 0.5217391304347826,
+      "grad_norm": 1.390079768168245,
+      "learning_rate": 9.893117626479778e-06,
+      "loss": 0.1422,
+      "step": 18
+    },
+    {
+      "epoch": 0.5507246376811594,
+      "grad_norm": 1.363268530198362,
+      "learning_rate": 9.854709087130261e-06,
+      "loss": 0.1448,
+      "step": 19
+    },
+    {
+      "epoch": 0.5797101449275363,
+      "grad_norm": 1.2167507253689909,
+      "learning_rate": 9.810515099218004e-06,
+      "loss": 0.1276,
+      "step": 20
+    },
+    {
+      "epoch": 0.6086956521739131,
+      "grad_norm": 1.1574568817316992,
+      "learning_rate": 9.76058832955357e-06,
+      "loss": 0.1177,
+      "step": 21
+    },
+    {
+      "epoch": 0.6376811594202898,
+      "grad_norm": 1.2101868179533353,
+      "learning_rate": 9.704988276811883e-06,
+      "loss": 0.1086,
+      "step": 22
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 1.1144422829365617,
+      "learning_rate": 9.643781200626512e-06,
+      "loss": 0.1291,
+      "step": 23
+    },
+    {
+      "epoch": 0.6956521739130435,
+      "grad_norm": 1.2450794816372637,
+      "learning_rate": 9.577040042626832e-06,
+      "loss": 0.1183,
+      "step": 24
+    },
+    {
+      "epoch": 0.7246376811594203,
+      "grad_norm": 1.2167862630409976,
+      "learning_rate": 9.504844339512096e-06,
+      "loss": 0.1213,
+      "step": 25
+    },
+    {
+      "epoch": 0.7536231884057971,
+      "grad_norm": 1.248557822116599,
+      "learning_rate": 9.427280128266049e-06,
+      "loss": 0.1177,
+      "step": 26
+    },
+    {
+      "epoch": 0.782608695652174,
+      "grad_norm": 1.113733087204506,
+      "learning_rate": 9.344439843625034e-06,
+      "loss": 0.1152,
+      "step": 27
+    },
+    {
+      "epoch": 0.8115942028985508,
+      "grad_norm": 1.1038118432089006,
+      "learning_rate": 9.256422207921757e-06,
+      "loss": 0.1246,
+      "step": 28
+    },
+    {
+      "epoch": 0.8405797101449275,
+      "grad_norm": 1.100623510584588,
+      "learning_rate": 9.163332113436031e-06,
+      "loss": 0.1184,
+      "step": 29
+    },
+    {
+      "epoch": 0.8695652173913043,
+      "grad_norm": 1.0179727735610475,
+      "learning_rate": 9.065280497392663e-06,
+      "loss": 0.1096,
+      "step": 30
+    },
+    {
+      "epoch": 0.8985507246376812,
+      "grad_norm": 1.2701457238654967,
+      "learning_rate": 8.962384209755453e-06,
+      "loss": 0.1195,
+      "step": 31
+    },
+    {
+      "epoch": 0.927536231884058,
+      "grad_norm": 1.1294896292998675,
+      "learning_rate": 8.854765873974898e-06,
+      "loss": 0.1105,
+      "step": 32
+    },
+    {
+      "epoch": 0.9565217391304348,
+      "grad_norm": 1.1171814840270677,
+      "learning_rate": 8.742553740855507e-06,
+      "loss": 0.1175,
+      "step": 33
+    },
+    {
+      "epoch": 0.9855072463768116,
+      "grad_norm": 0.9886273247393406,
+      "learning_rate": 8.625881535716883e-06,
+      "loss": 0.1058,
+      "step": 34
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.9886273247393406,
+      "learning_rate": 8.504888299030748e-06,
+      "loss": 0.1001,
+      "step": 35
+    },
+    {
+      "epoch": 1.0289855072463767,
+      "grad_norm": 1.5562818130049305,
+      "learning_rate": 8.379718220723772e-06,
+      "loss": 0.0727,
+      "step": 36
+    },
+    {
+      "epoch": 1.0579710144927537,
+      "grad_norm": 0.8147596643901716,
+      "learning_rate": 8.250520468343722e-06,
+      "loss": 0.0871,
+      "step": 37
+    },
+    {
+      "epoch": 1.0869565217391304,
+      "grad_norm": 0.785533808227011,
+      "learning_rate": 8.117449009293668e-06,
+      "loss": 0.0752,
+      "step": 38
+    },
+    {
+      "epoch": 1.1159420289855073,
+      "grad_norm": 1.067197574454516,
+      "learning_rate": 7.980662427346127e-06,
+      "loss": 0.0862,
+      "step": 39
+    },
+    {
+      "epoch": 1.144927536231884,
+      "grad_norm": 1.0067934230612672,
+      "learning_rate": 7.84032373365578e-06,
+      "loss": 0.0928,
+      "step": 40
+    },
+    {
+      "epoch": 1.1739130434782608,
+      "grad_norm": 1.027473082943366,
+      "learning_rate": 7.696600172495997e-06,
+      "loss": 0.0757,
+      "step": 41
+    },
+    {
+      "epoch": 1.2028985507246377,
+      "grad_norm": 0.9703955436663084,
+      "learning_rate": 7.5496630219506805e-06,
+      "loss": 0.0621,
+      "step": 42
+    },
+    {
+      "epoch": 1.2318840579710144,
+      "grad_norm": 0.9885265576714246,
+      "learning_rate": 7.399687389798933e-06,
+      "loss": 0.0855,
+      "step": 43
+    },
+    {
+      "epoch": 1.2608695652173914,
+      "grad_norm": 0.9758981949816004,
+      "learning_rate": 7.246852004835807e-06,
+      "loss": 0.084,
+      "step": 44
+    },
+    {
+      "epoch": 1.289855072463768,
+      "grad_norm": 0.9024594434416954,
+      "learning_rate": 7.091339003877826e-06,
+      "loss": 0.0767,
+      "step": 45
+    },
+    {
+      "epoch": 1.318840579710145,
+      "grad_norm": 0.830224350365886,
+      "learning_rate": 6.933333714707094e-06,
+      "loss": 0.0806,
+      "step": 46
+    },
+    {
+      "epoch": 1.3478260869565217,
+      "grad_norm": 0.8475268092069895,
+      "learning_rate": 6.773024435212678e-06,
+      "loss": 0.0686,
+      "step": 47
+    },
+    {
+      "epoch": 1.3768115942028984,
+      "grad_norm": 0.8152669559501473,
+      "learning_rate": 6.6106022089924535e-06,
+      "loss": 0.0913,
+      "step": 48
+    },
+    {
+      "epoch": 1.4057971014492754,
+      "grad_norm": 0.8455140548788239,
+      "learning_rate": 6.4462605976828395e-06,
+      "loss": 0.0746,
+      "step": 49
+    },
+    {
+      "epoch": 1.434782608695652,
+      "grad_norm": 0.817492348372533,
+      "learning_rate": 6.280195450287736e-06,
+      "loss": 0.0976,
+      "step": 50
+    },
+    {
+      "epoch": 1.463768115942029,
+      "grad_norm": 0.880952546066598,
+      "learning_rate": 6.112604669781572e-06,
+      "loss": 0.0833,
+      "step": 51
+    },
+    {
+      "epoch": 1.4927536231884058,
+      "grad_norm": 0.8489601481147181,
+      "learning_rate": 5.943687977264584e-06,
+      "loss": 0.0784,
+      "step": 52
+    },
+    {
+      "epoch": 1.5217391304347827,
+      "grad_norm": 0.8142122839258903,
+      "learning_rate": 5.773646673951406e-06,
+      "loss": 0.0838,
+      "step": 53
+    },
+    {
+      "epoch": 1.5507246376811594,
+      "grad_norm": 0.8922929202950094,
+      "learning_rate": 5.6026834012766155e-06,
+      "loss": 0.0828,
+      "step": 54
+    },
+    {
+      "epoch": 1.5797101449275361,
+      "grad_norm": 0.7122624377371234,
+      "learning_rate": 5.4310018994030974e-06,
+      "loss": 0.0797,
+      "step": 55
+    },
+    {
+      "epoch": 1.608695652173913,
+      "grad_norm": 0.9217981197726534,
+      "learning_rate": 5.258806764421048e-06,
+      "loss": 0.0772,
+      "step": 56
+    },
+    {
+      "epoch": 1.6376811594202898,
+      "grad_norm": 0.9257391753751778,
+      "learning_rate": 5.0863032045269435e-06,
+      "loss": 0.0766,
+      "step": 57
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 0.8307688279849595,
+      "learning_rate": 4.913696795473058e-06,
+      "loss": 0.0818,
+      "step": 58
+    },
+    {
+      "epoch": 1.6956521739130435,
+      "grad_norm": 0.9173954819802914,
+      "learning_rate": 4.741193235578953e-06,
+      "loss": 0.0927,
+      "step": 59
+    },
+    {
+      "epoch": 1.7246376811594204,
+      "grad_norm": 0.8708594385417067,
+      "learning_rate": 4.568998100596903e-06,
+      "loss": 0.0665,
+      "step": 60
+    },
+    {
+      "epoch": 1.7536231884057971,
+      "grad_norm": 0.8126974465329198,
+      "learning_rate": 4.397316598723385e-06,
+      "loss": 0.0839,
+      "step": 61
+    },
+    {
+      "epoch": 1.7826086956521738,
+      "grad_norm": 0.853158086441921,
+      "learning_rate": 4.226353326048594e-06,
+      "loss": 0.0773,
+      "step": 62
+    },
+    {
+      "epoch": 1.8115942028985508,
+      "grad_norm": 0.8525307891771787,
+      "learning_rate": 4.056312022735417e-06,
+      "loss": 0.0809,
+      "step": 63
+    },
+    {
+      "epoch": 1.8405797101449275,
+      "grad_norm": 1.0467111292284437,
+      "learning_rate": 3.887395330218429e-06,
+      "loss": 0.0782,
+      "step": 64
+    },
+    {
+      "epoch": 1.8695652173913042,
+      "grad_norm": 0.9286296034389645,
+      "learning_rate": 3.7198045497122647e-06,
+      "loss": 0.0774,
+      "step": 65
+    },
+    {
+      "epoch": 1.8985507246376812,
+      "grad_norm": 0.7656482475139829,
+      "learning_rate": 3.553739402317162e-06,
+      "loss": 0.0767,
+      "step": 66
+    },
+    {
+      "epoch": 1.927536231884058,
+      "grad_norm": 0.7766347569440668,
+      "learning_rate": 3.389397791007548e-06,
+      "loss": 0.0984,
+      "step": 67
+    },
+    {
+      "epoch": 1.9565217391304348,
+      "grad_norm": 0.7336465690307817,
+      "learning_rate": 3.226975564787322e-06,
+      "loss": 0.074,
+      "step": 68
+    },
+    {
+      "epoch": 1.9855072463768115,
+      "grad_norm": 0.795203124951574,
+      "learning_rate": 3.0666662852929063e-06,
+      "loss": 0.0785,
+      "step": 69
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.2688010574239046,
+      "learning_rate": 2.9086609961221758e-06,
+      "loss": 0.0659,
+      "step": 70
+    },
+    {
+      "epoch": 2.028985507246377,
+      "grad_norm": 0.5159106800993555,
+      "learning_rate": 2.7531479951641928e-06,
+      "loss": 0.0501,
+      "step": 71
+    },
+    {
+      "epoch": 2.0579710144927534,
+      "grad_norm": 0.49731065359536447,
+      "learning_rate": 2.6003126102010696e-06,
+      "loss": 0.0487,
+      "step": 72
+    },
+    {
+      "epoch": 2.0869565217391304,
+      "grad_norm": 0.4862325168980843,
+      "learning_rate": 2.450336978049322e-06,
+      "loss": 0.051,
+      "step": 73
+    },
+    {
+      "epoch": 2.1159420289855073,
+      "grad_norm": 0.4949710047982997,
+      "learning_rate": 2.3033998275040047e-06,
+      "loss": 0.0564,
+      "step": 74
+    },
+    {
+      "epoch": 2.1449275362318843,
+      "grad_norm": 0.5420095199622179,
+      "learning_rate": 2.159676266344222e-06,
+      "loss": 0.0497,
+      "step": 75
+    },
+    {
+      "epoch": 2.1739130434782608,
+      "grad_norm": 0.5223890245623899,
+      "learning_rate": 2.0193375726538737e-06,
+      "loss": 0.0608,
+      "step": 76
+    },
+    {
+      "epoch": 2.2028985507246377,
+      "grad_norm": 0.5072028301413968,
+      "learning_rate": 1.8825509907063328e-06,
+      "loss": 0.0466,
+      "step": 77
+    },
+    {
+      "epoch": 2.2318840579710146,
+      "grad_norm": 0.6058827129392587,
+      "learning_rate": 1.7494795316562791e-06,
+      "loss": 0.0519,
+      "step": 78
+    },
+    {
+      "epoch": 2.260869565217391,
+      "grad_norm": 0.5967337337925352,
+      "learning_rate": 1.6202817792762283e-06,
+      "loss": 0.0475,
+      "step": 79
+    },
+    {
+      "epoch": 2.289855072463768,
+      "grad_norm": 0.45709533088838783,
+      "learning_rate": 1.4951117009692528e-06,
+      "loss": 0.0439,
+      "step": 80
+    },
+    {
+      "epoch": 2.318840579710145,
+      "grad_norm": 0.5908455438244339,
+      "learning_rate": 1.374118464283119e-06,
+      "loss": 0.0447,
+      "step": 81
+    },
+    {
+      "epoch": 2.3478260869565215,
+      "grad_norm": 0.5958774937866624,
+      "learning_rate": 1.257446259144494e-06,
+      "loss": 0.0398,
+      "step": 82
+    },
+    {
+      "epoch": 2.3768115942028984,
+      "grad_norm": 0.5185817292507452,
+      "learning_rate": 1.145234126025102e-06,
+      "loss": 0.0522,
+      "step": 83
+    },
+    {
+      "epoch": 2.4057971014492754,
+      "grad_norm": 0.517132336779929,
+      "learning_rate": 1.037615790244549e-06,
+      "loss": 0.0542,
+      "step": 84
+    },
+    {
+      "epoch": 2.4347826086956523,
+      "grad_norm": 0.5753126450523465,
+      "learning_rate": 9.347195026073369e-07,
+      "loss": 0.053,
+      "step": 85
+    },
+    {
+      "epoch": 2.463768115942029,
+      "grad_norm": 0.5537929012260736,
+      "learning_rate": 8.366678865639688e-07,
+      "loss": 0.0533,
+      "step": 86
+    },
+    {
+      "epoch": 2.4927536231884058,
+      "grad_norm": 0.5648244812023223,
+      "learning_rate": 7.435777920782444e-07,
+      "loss": 0.0451,
+      "step": 87
+    },
+    {
+      "epoch": 2.5217391304347827,
+      "grad_norm": 0.6470131986288629,
+      "learning_rate": 6.555601563749675e-07,
+      "loss": 0.0638,
+      "step": 88
+    },
+    {
+      "epoch": 2.550724637681159,
+      "grad_norm": 0.682212123629125,
+      "learning_rate": 5.727198717339511e-07,
+      "loss": 0.0507,
+      "step": 89
+    },
+    {
+      "epoch": 2.579710144927536,
+      "grad_norm": 0.5044269543104078,
+      "learning_rate": 4.951556604879049e-07,
+      "loss": 0.0498,
+      "step": 90
+    },
+    {
+      "epoch": 2.608695652173913,
+      "grad_norm": 0.5906232388239259,
+      "learning_rate": 4.2295995737316854e-07,
+      "loss": 0.0487,
+      "step": 91
+    },
+    {
+      "epoch": 2.63768115942029,
+      "grad_norm": 0.6323799418493858,
+      "learning_rate": 3.5621879937348836e-07,
+      "loss": 0.0409,
+      "step": 92
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 0.5896524185850267,
+      "learning_rate": 2.9501172318811834e-07,
+      "loss": 0.0464,
+      "step": 93
+    },
+    {
+      "epoch": 2.6956521739130435,
+      "grad_norm": 0.5444149761155485,
+      "learning_rate": 2.394116704464294e-07,
+      "loss": 0.0559,
+      "step": 94
+    },
+    {
+      "epoch": 2.7246376811594204,
+      "grad_norm": 0.5667834464277357,
+      "learning_rate": 1.8948490078199767e-07,
+      "loss": 0.0419,
+      "step": 95
+    },
+    {
+      "epoch": 2.753623188405797,
+      "grad_norm": 0.607364619958279,
+      "learning_rate": 1.4529091286973994e-07,
+      "loss": 0.0467,
+      "step": 96
+    },
+    {
+      "epoch": 2.782608695652174,
+      "grad_norm": 0.5868613570667187,
+      "learning_rate": 1.0688237352022346e-07,
+      "loss": 0.05,
+      "step": 97
+    },
+    {
+      "epoch": 2.8115942028985508,
+      "grad_norm": 0.558060235895656,
+      "learning_rate": 7.430505491563101e-08,
+      "loss": 0.0373,
+      "step": 98
+    },
+    {
+      "epoch": 2.8405797101449277,
+      "grad_norm": 0.6239726216026107,
+      "learning_rate": 4.759778006218407e-08,
+      "loss": 0.0496,
+      "step": 99
+    },
+    {
+      "epoch": 2.869565217391304,
+      "grad_norm": 0.5091602751990856,
+      "learning_rate": 2.6792376524036878e-08,
+      "loss": 0.0511,
+      "step": 100
+    },
+    {
+      "epoch": 2.898550724637681,
+      "grad_norm": 0.5073066164113447,
+      "learning_rate": 1.1913638493762369e-08,
+      "loss": 0.0498,
+      "step": 101
+    },
+    {
+      "epoch": 2.927536231884058,
+      "grad_norm": 0.6581905704459963,
+      "learning_rate": 2.9792972446479605e-09,
+      "loss": 0.0503,
+      "step": 102
+    },
+    {
+      "epoch": 2.927536231884058,
+      "step": 102,
+      "total_flos": 10226403115008.0,
+      "train_loss": 0.12416269268621416,
+      "train_runtime": 5225.659,
+      "train_samples_per_second": 5.047,
+      "train_steps_per_second": 0.02
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 102,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 10226403115008.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:534bc8e474028dbf459153f9e2e4d91c0f9abbba10a638ebb072d0c0a02971de
+size 7608