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

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
+}

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 sha256:4d7b9e8a6ec433b477074112989f6a8fe28b49b91fece7dc0b01087b206f38dd
+size 4997184968

model-00002-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1f23316762273e3316b404260c058087810e5d4fcda9cfeedbf988af06a3430a
+size 4997741608

model-00003-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:81d55f1b8d478069d29ebeb8d36960b041895a67c031a52fd68f44e779286a05
+size 4997742208

model-00004-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8f3390b1672d14f72fd9c16f725e89ff6520f2c0f3cfeeb44749d8cd51a7160f
+size 4997743184

model-00005-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b1ed13f60bbc108f66e926c221cbd50e523d50bcc0d0df646a0275da67e94217
+size 4997743184

model-00006-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:55bbee5b8d8b2ba3d667b85f0097971da52b79aac5d00e610e19f6aeeec58808
+size 4997743184

model-00007-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e36838bcb6b6598ccc070b0cfd27cdb3ed43b0387afe5d0d2b27690bf6f48d4d
+size 4997743184

model-00008-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6de8700fe0a88306ef00a2645a056bdd0fceb4d6e54bd6a0767c36297ef34a5c
+size 4997743184

model-00009-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c68661fc96b95c249ced28b8d5af7c53ecd8e773fe49d60efd4db5614df4d4c1
+size 4997743184

model-00010-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a838d205fbf3683f39e6c47c88b4eb4d4a8a8cbeeb7d70562e7682679acf4c18
+size 4997743184

model-00011-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bfa8b8e1fc33ba078132f6cda31d09a9040863ccc094957d3487bffba55ac9a8
+size 4997743184

model-00012-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5171772f8882e4c9da20a6879cdbde20969322fa3db9e617c95d031b4ef377d9
+size 4997743184

model-00013-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4d8b5b609d5f159b4a740bbdfa8cc3156adc3b08a9f584370bddf6517f366d01
+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
+}

trainer_state.json

@@ -0,0 +1,1630 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "eval_steps": 500,
+  "global_step": 228,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.017543859649122806,
+      "grad_norm": 2.1737393862889918,
+      "learning_rate": 0.0,
+      "loss": 0.8483,
+      "step": 1
+    },
+    {
+      "epoch": 0.03508771929824561,
+      "grad_norm": 2.065849016353641,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.8763,
+      "step": 2
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 2.2405802416848637,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.0189,
+      "step": 3
+    },
+    {
+      "epoch": 0.07017543859649122,
+      "grad_norm": 1.8818697785146044,
+      "learning_rate": 1e-05,
+      "loss": 1.0442,
+      "step": 4
+    },
+    {
+      "epoch": 0.08771929824561403,
+      "grad_norm": 1.7909780788976988,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.8302,
+      "step": 5
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 1.326496650918511,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.9662,
+      "step": 6
+    },
+    {
+      "epoch": 0.12280701754385964,
+      "grad_norm": 2.2923825086858054,
+      "learning_rate": 2e-05,
+      "loss": 0.9657,
+      "step": 7
+    },
+    {
+      "epoch": 0.14035087719298245,
+      "grad_norm": 1.7709874116952984,
+      "learning_rate": 2.3333333333333336e-05,
+      "loss": 0.9053,
+      "step": 8
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 1.917852629249726,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.929,
+      "step": 9
+    },
+    {
+      "epoch": 0.17543859649122806,
+      "grad_norm": 1.73186050223783,
+      "learning_rate": 3e-05,
+      "loss": 0.9575,
+      "step": 10
+    },
+    {
+      "epoch": 0.19298245614035087,
+      "grad_norm": 1.7935808552502084,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.9943,
+      "step": 11
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 1.054785055716573,
+      "learning_rate": 3.6666666666666666e-05,
+      "loss": 0.8707,
+      "step": 12
+    },
+    {
+      "epoch": 0.22807017543859648,
+      "grad_norm": 1.1678253946017734,
+      "learning_rate": 4e-05,
+      "loss": 0.9013,
+      "step": 13
+    },
+    {
+      "epoch": 0.24561403508771928,
+      "grad_norm": 1.2354003623651981,
+      "learning_rate": 4.3333333333333334e-05,
+      "loss": 0.8911,
+      "step": 14
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 1.1568095206536237,
+      "learning_rate": 4.666666666666667e-05,
+      "loss": 0.9503,
+      "step": 15
+    },
+    {
+      "epoch": 0.2807017543859649,
+      "grad_norm": 0.8495331099528948,
+      "learning_rate": 5e-05,
+      "loss": 0.9027,
+      "step": 16
+    },
+    {
+      "epoch": 0.2982456140350877,
+      "grad_norm": 1.2538751518161324,
+      "learning_rate": 4.999830770009406e-05,
+      "loss": 1.0006,
+      "step": 17
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 1.0265477578740414,
+      "learning_rate": 4.9993231029486544e-05,
+      "loss": 0.8824,
+      "step": 18
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 0.7707819215922666,
+      "learning_rate": 4.99847706754774e-05,
+      "loss": 0.9055,
+      "step": 19
+    },
+    {
+      "epoch": 0.3508771929824561,
+      "grad_norm": 0.8509611053188778,
+      "learning_rate": 4.997292778346312e-05,
+      "loss": 0.9463,
+      "step": 20
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 0.9277926520245647,
+      "learning_rate": 4.995770395678171e-05,
+      "loss": 0.8836,
+      "step": 21
+    },
+    {
+      "epoch": 0.38596491228070173,
+      "grad_norm": 1.1149774662360015,
+      "learning_rate": 4.993910125649561e-05,
+      "loss": 0.7536,
+      "step": 22
+    },
+    {
+      "epoch": 0.40350877192982454,
+      "grad_norm": 0.7943650957585513,
+      "learning_rate": 4.9917122201112656e-05,
+      "loss": 0.8899,
+      "step": 23
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 0.8441796882708572,
+      "learning_rate": 4.989176976624511e-05,
+      "loss": 0.7896,
+      "step": 24
+    },
+    {
+      "epoch": 0.43859649122807015,
+      "grad_norm": 0.635410105590828,
+      "learning_rate": 4.9863047384206835e-05,
+      "loss": 0.9227,
+      "step": 25
+    },
+    {
+      "epoch": 0.45614035087719296,
+      "grad_norm": 0.7423716066814047,
+      "learning_rate": 4.983095894354858e-05,
+      "loss": 0.8919,
+      "step": 26
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 0.7104290541961332,
+      "learning_rate": 4.979550878853154e-05,
+      "loss": 0.8047,
+      "step": 27
+    },
+    {
+      "epoch": 0.49122807017543857,
+      "grad_norm": 0.6031808840111699,
+      "learning_rate": 4.975670171853926e-05,
+      "loss": 0.8762,
+      "step": 28
+    },
+    {
+      "epoch": 0.5087719298245614,
+      "grad_norm": 0.5779272293658767,
+      "learning_rate": 4.971454298742779e-05,
+      "loss": 0.8985,
+      "step": 29
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 0.6445795617091271,
+      "learning_rate": 4.966903830281449e-05,
+      "loss": 0.7863,
+      "step": 30
+    },
+    {
+      "epoch": 0.543859649122807,
+      "grad_norm": 0.7209168276499288,
+      "learning_rate": 4.962019382530521e-05,
+      "loss": 0.8408,
+      "step": 31
+    },
+    {
+      "epoch": 0.5614035087719298,
+      "grad_norm": 0.5698219842212006,
+      "learning_rate": 4.9568016167660334e-05,
+      "loss": 0.7785,
+      "step": 32
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 0.6370166993760419,
+      "learning_rate": 4.951251239389948e-05,
+      "loss": 0.8726,
+      "step": 33
+    },
+    {
+      "epoch": 0.5964912280701754,
+      "grad_norm": 0.7100597543856979,
+      "learning_rate": 4.9453690018345144e-05,
+      "loss": 0.9768,
+      "step": 34
+    },
+    {
+      "epoch": 0.6140350877192983,
+      "grad_norm": 0.5651052866267301,
+      "learning_rate": 4.939155700460536e-05,
+      "loss": 0.8185,
+      "step": 35
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 0.731828562220152,
+      "learning_rate": 4.9326121764495596e-05,
+      "loss": 0.8378,
+      "step": 36
+    },
+    {
+      "epoch": 0.6491228070175439,
+      "grad_norm": 0.6726714827790597,
+      "learning_rate": 4.925739315689991e-05,
+      "loss": 0.8762,
+      "step": 37
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 0.5848273690388421,
+      "learning_rate": 4.9185380486571595e-05,
+      "loss": 0.8833,
+      "step": 38
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.644260996985412,
+      "learning_rate": 4.9110093502873476e-05,
+      "loss": 0.9069,
+      "step": 39
+    },
+    {
+      "epoch": 0.7017543859649122,
+      "grad_norm": 0.5091485420643463,
+      "learning_rate": 4.9031542398457974e-05,
+      "loss": 0.8708,
+      "step": 40
+    },
+    {
+      "epoch": 0.7192982456140351,
+      "grad_norm": 0.5623929735083179,
+      "learning_rate": 4.894973780788722e-05,
+      "loss": 0.8808,
+      "step": 41
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.5195014046396914,
+      "learning_rate": 4.88646908061933e-05,
+      "loss": 0.8106,
+      "step": 42
+    },
+    {
+      "epoch": 0.7543859649122807,
+      "grad_norm": 0.5774984023385406,
+      "learning_rate": 4.877641290737884e-05,
+      "loss": 0.8488,
+      "step": 43
+    },
+    {
+      "epoch": 0.7719298245614035,
+      "grad_norm": 0.5193829880107578,
+      "learning_rate": 4.868491606285823e-05,
+      "loss": 0.7655,
+      "step": 44
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.5365181328093115,
+      "learning_rate": 4.859021265983959e-05,
+      "loss": 0.8806,
+      "step": 45
+    },
+    {
+      "epoch": 0.8070175438596491,
+      "grad_norm": 0.5789874199030948,
+      "learning_rate": 4.849231551964771e-05,
+      "loss": 0.8563,
+      "step": 46
+    },
+    {
+      "epoch": 0.8245614035087719,
+      "grad_norm": 0.6212353702348333,
+      "learning_rate": 4.839123789598829e-05,
+      "loss": 0.8667,
+      "step": 47
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.5701391975310476,
+      "learning_rate": 4.828699347315356e-05,
+      "loss": 0.8426,
+      "step": 48
+    },
+    {
+      "epoch": 0.8596491228070176,
+      "grad_norm": 0.6511736836282973,
+      "learning_rate": 4.817959636416969e-05,
+      "loss": 0.9415,
+      "step": 49
+    },
+    {
+      "epoch": 0.8771929824561403,
+      "grad_norm": 0.4803935151868872,
+      "learning_rate": 4.806906110888606e-05,
+      "loss": 0.6745,
+      "step": 50
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 0.7255949431050441,
+      "learning_rate": 4.7955402672006854e-05,
+      "loss": 0.9098,
+      "step": 51
+    },
+    {
+      "epoch": 0.9122807017543859,
+      "grad_norm": 0.5377201244233609,
+      "learning_rate": 4.783863644106502e-05,
+      "loss": 0.8827,
+      "step": 52
+    },
+    {
+      "epoch": 0.9298245614035088,
+      "grad_norm": 0.5638038206248809,
+      "learning_rate": 4.771877822433911e-05,
+      "loss": 0.8328,
+      "step": 53
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.49220472544815935,
+      "learning_rate": 4.759584424871302e-05,
+      "loss": 0.7827,
+      "step": 54
+    },
+    {
+      "epoch": 0.9649122807017544,
+      "grad_norm": 0.5426889947396931,
+      "learning_rate": 4.7469851157479177e-05,
+      "loss": 0.8224,
+      "step": 55
+    },
+    {
+      "epoch": 0.9824561403508771,
+      "grad_norm": 0.5174411953230827,
+      "learning_rate": 4.734081600808531e-05,
+      "loss": 0.833,
+      "step": 56
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7444376196945202,
+      "learning_rate": 4.7208756269825104e-05,
+      "loss": 0.6926,
+      "step": 57
+    },
+    {
+      "epoch": 1.0175438596491229,
+      "grad_norm": 0.788387022701859,
+      "learning_rate": 4.707368982147318e-05,
+      "loss": 0.7366,
+      "step": 58
+    },
+    {
+      "epoch": 1.0350877192982457,
+      "grad_norm": 0.6610866627123646,
+      "learning_rate": 4.693563494886455e-05,
+      "loss": 0.6288,
+      "step": 59
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.9705336832406968,
+      "learning_rate": 4.679461034241906e-05,
+      "loss": 0.7268,
+      "step": 60
+    },
+    {
+      "epoch": 1.0701754385964912,
+      "grad_norm": 0.9034123796440131,
+      "learning_rate": 4.665063509461097e-05,
+      "loss": 0.6198,
+      "step": 61
+    },
+    {
+      "epoch": 1.087719298245614,
+      "grad_norm": 0.8791095306428166,
+      "learning_rate": 4.650372869738414e-05,
+      "loss": 0.6547,
+      "step": 62
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 1.2043672957317344,
+      "learning_rate": 4.6353911039513145e-05,
+      "loss": 0.7271,
+      "step": 63
+    },
+    {
+      "epoch": 1.1228070175438596,
+      "grad_norm": 0.8784522060594879,
+      "learning_rate": 4.620120240391065e-05,
+      "loss": 0.673,
+      "step": 64
+    },
+    {
+      "epoch": 1.1403508771929824,
+      "grad_norm": 0.7462938442417025,
+      "learning_rate": 4.604562346488144e-05,
+      "loss": 0.6199,
+      "step": 65
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.7282681308735043,
+      "learning_rate": 4.588719528532342e-05,
+      "loss": 0.6381,
+      "step": 66
+    },
+    {
+      "epoch": 1.1754385964912282,
+      "grad_norm": 0.6803141685436093,
+      "learning_rate": 4.572593931387604e-05,
+      "loss": 0.623,
+      "step": 67
+    },
+    {
+      "epoch": 1.1929824561403508,
+      "grad_norm": 0.7325213999855061,
+      "learning_rate": 4.556187738201656e-05,
+      "loss": 0.662,
+      "step": 68
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.7691432804985348,
+      "learning_rate": 4.539503170110431e-05,
+      "loss": 0.6993,
+      "step": 69
+    },
+    {
+      "epoch": 1.2280701754385965,
+      "grad_norm": 0.561067659730212,
+      "learning_rate": 4.522542485937369e-05,
+      "loss": 0.6557,
+      "step": 70
+    },
+    {
+      "epoch": 1.2456140350877192,
+      "grad_norm": 0.7753717205544434,
+      "learning_rate": 4.50530798188761e-05,
+      "loss": 0.722,
+      "step": 71
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 1.512136203917636,
+      "learning_rate": 4.48780199123712e-05,
+      "loss": 0.6832,
+      "step": 72
+    },
+    {
+      "epoch": 1.280701754385965,
+      "grad_norm": 0.6133486927867576,
+      "learning_rate": 4.4700268840168045e-05,
+      "loss": 0.6692,
+      "step": 73
+    },
+    {
+      "epoch": 1.2982456140350878,
+      "grad_norm": 0.5693268023489857,
+      "learning_rate": 4.4519850666916484e-05,
+      "loss": 0.6019,
+      "step": 74
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.6141770853666199,
+      "learning_rate": 4.43367898183491e-05,
+      "loss": 0.7295,
+      "step": 75
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 0.48983184623143194,
+      "learning_rate": 4.415111107797445e-05,
+      "loss": 0.6095,
+      "step": 76
+    },
+    {
+      "epoch": 1.3508771929824561,
+      "grad_norm": 0.5514436368786412,
+      "learning_rate": 4.396283958372173e-05,
+      "loss": 0.601,
+      "step": 77
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.7257113094006158,
+      "learning_rate": 4.377200082453749e-05,
+      "loss": 0.7115,
+      "step": 78
+    },
+    {
+      "epoch": 1.3859649122807016,
+      "grad_norm": 0.604893232531512,
+      "learning_rate": 4.357862063693486e-05,
+      "loss": 0.6896,
+      "step": 79
+    },
+    {
+      "epoch": 1.4035087719298245,
+      "grad_norm": 0.6150678289713601,
+      "learning_rate": 4.3382725201495723e-05,
+      "loss": 0.6629,
+      "step": 80
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 0.5662279941837114,
+      "learning_rate": 4.318434103932622e-05,
+      "loss": 0.6675,
+      "step": 81
+    },
+    {
+      "epoch": 1.4385964912280702,
+      "grad_norm": 0.5613748309599569,
+      "learning_rate": 4.2983495008466276e-05,
+      "loss": 0.6526,
+      "step": 82
+    },
+    {
+      "epoch": 1.456140350877193,
+      "grad_norm": 0.5554128045167495,
+      "learning_rate": 4.278021430025343e-05,
+      "loss": 0.5805,
+      "step": 83
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.4640230873372006,
+      "learning_rate": 4.257452643564155e-05,
+      "loss": 0.5021,
+      "step": 84
+    },
+    {
+      "epoch": 1.4912280701754386,
+      "grad_norm": 0.7044680212736634,
+      "learning_rate": 4.2366459261474933e-05,
+      "loss": 0.6677,
+      "step": 85
+    },
+    {
+      "epoch": 1.5087719298245614,
+      "grad_norm": 0.5247365062928431,
+      "learning_rate": 4.215604094671835e-05,
+      "loss": 0.6426,
+      "step": 86
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.5533183681382117,
+      "learning_rate": 4.194329997864331e-05,
+      "loss": 0.765,
+      "step": 87
+    },
+    {
+      "epoch": 1.543859649122807,
+      "grad_norm": 0.6034264380245536,
+      "learning_rate": 4.172826515897146e-05,
+      "loss": 0.664,
+      "step": 88
+    },
+    {
+      "epoch": 1.5614035087719298,
+      "grad_norm": 0.5230196503836612,
+      "learning_rate": 4.1510965599975196e-05,
+      "loss": 0.7011,
+      "step": 89
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.5828358559586317,
+      "learning_rate": 4.129143072053638e-05,
+      "loss": 0.6038,
+      "step": 90
+    },
+    {
+      "epoch": 1.5964912280701755,
+      "grad_norm": 0.547733265549105,
+      "learning_rate": 4.1069690242163484e-05,
+      "loss": 0.6353,
+      "step": 91
+    },
+    {
+      "epoch": 1.6140350877192984,
+      "grad_norm": 0.5250083745984832,
+      "learning_rate": 4.0845774184967754e-05,
+      "loss": 0.6515,
+      "step": 92
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.5063751430984663,
+      "learning_rate": 4.0619712863599e-05,
+      "loss": 0.5934,
+      "step": 93
+    },
+    {
+      "epoch": 1.6491228070175439,
+      "grad_norm": 0.4615079321893893,
+      "learning_rate": 4.039153688314145e-05,
+      "loss": 0.601,
+      "step": 94
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 0.5360613197615536,
+      "learning_rate": 4.0161277134970345e-05,
+      "loss": 0.6986,
+      "step": 95
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.5236096239190716,
+      "learning_rate": 3.9928964792569655e-05,
+      "loss": 0.6124,
+      "step": 96
+    },
+    {
+      "epoch": 1.7017543859649122,
+      "grad_norm": 0.5550920780280213,
+      "learning_rate": 3.969463130731183e-05,
+      "loss": 0.7707,
+      "step": 97
+    },
+    {
+      "epoch": 1.719298245614035,
+      "grad_norm": 0.6272595343106637,
+      "learning_rate": 3.945830840419966e-05,
+      "loss": 0.6626,
+      "step": 98
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.5123303866602584,
+      "learning_rate": 3.9220028077571295e-05,
+      "loss": 0.6797,
+      "step": 99
+    },
+    {
+      "epoch": 1.7543859649122808,
+      "grad_norm": 0.5176125046966141,
+      "learning_rate": 3.897982258676867e-05,
+      "loss": 0.6152,
+      "step": 100
+    },
+    {
+      "epoch": 1.7719298245614035,
+      "grad_norm": 0.5415145180664312,
+      "learning_rate": 3.873772445177015e-05,
+      "loss": 0.7027,
+      "step": 101
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.47863117651487286,
+      "learning_rate": 3.8493766448787825e-05,
+      "loss": 0.6934,
+      "step": 102
+    },
+    {
+      "epoch": 1.807017543859649,
+      "grad_norm": 0.6373144922403705,
+      "learning_rate": 3.824798160583012e-05,
+      "loss": 0.6943,
+      "step": 103
+    },
+    {
+      "epoch": 1.8245614035087718,
+      "grad_norm": 0.5302740878348818,
+      "learning_rate": 3.8000403198230387e-05,
+      "loss": 0.5755,
+      "step": 104
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.4610854847881297,
+      "learning_rate": 3.775106474414188e-05,
+      "loss": 0.5399,
+      "step": 105
+    },
+    {
+      "epoch": 1.8596491228070176,
+      "grad_norm": 0.7494981458587492,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.7576,
+      "step": 106
+    },
+    {
+      "epoch": 1.8771929824561404,
+      "grad_norm": 0.7550924000614921,
+      "learning_rate": 3.7247242955952175e-05,
+      "loss": 0.6694,
+      "step": 107
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.5441407432942561,
+      "learning_rate": 3.699282783125616e-05,
+      "loss": 0.6648,
+      "step": 108
+    },
+    {
+      "epoch": 1.912280701754386,
+      "grad_norm": 0.5303134950522279,
+      "learning_rate": 3.673678906964727e-05,
+      "loss": 0.5801,
+      "step": 109
+    },
+    {
+      "epoch": 1.9298245614035088,
+      "grad_norm": 0.5865681284021387,
+      "learning_rate": 3.6479161334675296e-05,
+      "loss": 0.6989,
+      "step": 110
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.6194750092091883,
+      "learning_rate": 3.621997950501156e-05,
+      "loss": 0.7117,
+      "step": 111
+    },
+    {
+      "epoch": 1.9649122807017543,
+      "grad_norm": 0.6062058641940399,
+      "learning_rate": 3.5959278669726935e-05,
+      "loss": 0.6757,
+      "step": 112
+    },
+    {
+      "epoch": 1.9824561403508771,
+      "grad_norm": 0.5802932099330305,
+      "learning_rate": 3.569709412354136e-05,
+      "loss": 0.7369,
+      "step": 113
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.7731983699378179,
+      "learning_rate": 3.543346136204545e-05,
+      "loss": 0.5665,
+      "step": 114
+    },
+    {
+      "epoch": 2.017543859649123,
+      "grad_norm": 0.9639546114758945,
+      "learning_rate": 3.516841607689501e-05,
+      "loss": 0.4182,
+      "step": 115
+    },
+    {
+      "epoch": 2.0350877192982457,
+      "grad_norm": 1.0914799505172577,
+      "learning_rate": 3.490199415097892e-05,
+      "loss": 0.4514,
+      "step": 116
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 1.2237851418448777,
+      "learning_rate": 3.463423165356121e-05,
+      "loss": 0.604,
+      "step": 117
+    },
+    {
+      "epoch": 2.0701754385964914,
+      "grad_norm": 0.7423286146663943,
+      "learning_rate": 3.436516483539781e-05,
+      "loss": 0.461,
+      "step": 118
+    },
+    {
+      "epoch": 2.087719298245614,
+      "grad_norm": 0.9155391813155248,
+      "learning_rate": 3.409483012382879e-05,
+      "loss": 0.4713,
+      "step": 119
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.8003519891181342,
+      "learning_rate": 3.382326411784672e-05,
+      "loss": 0.4457,
+      "step": 120
+    },
+    {
+      "epoch": 2.1228070175438596,
+      "grad_norm": 0.5467492402868155,
+      "learning_rate": 3.355050358314172e-05,
+      "loss": 0.4325,
+      "step": 121
+    },
+    {
+      "epoch": 2.1403508771929824,
+      "grad_norm": 0.7059153118181073,
+      "learning_rate": 3.327658544712395e-05,
+      "loss": 0.3819,
+      "step": 122
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.686389123168142,
+      "learning_rate": 3.3001546793924285e-05,
+      "loss": 0.5352,
+      "step": 123
+    },
+    {
+      "epoch": 2.175438596491228,
+      "grad_norm": 0.5786551548347438,
+      "learning_rate": 3.272542485937369e-05,
+      "loss": 0.4219,
+      "step": 124
+    },
+    {
+      "epoch": 2.192982456140351,
+      "grad_norm": 0.5926666178690436,
+      "learning_rate": 3.244825702596205e-05,
+      "loss": 0.5009,
+      "step": 125
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.6141029942521959,
+      "learning_rate": 3.217008081777726e-05,
+      "loss": 0.4298,
+      "step": 126
+    },
+    {
+      "epoch": 2.2280701754385963,
+      "grad_norm": 0.6947870611067657,
+      "learning_rate": 3.1890933895424976e-05,
+      "loss": 0.413,
+      "step": 127
+    },
+    {
+      "epoch": 2.245614035087719,
+      "grad_norm": 0.569362234800568,
+      "learning_rate": 3.161085405093006e-05,
+      "loss": 0.4346,
+      "step": 128
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.5903701306972321,
+      "learning_rate": 3.132987920262005e-05,
+      "loss": 0.4709,
+      "step": 129
+    },
+    {
+      "epoch": 2.280701754385965,
+      "grad_norm": 0.53245169979005,
+      "learning_rate": 3.104804738999169e-05,
+      "loss": 0.4534,
+      "step": 130
+    },
+    {
+      "epoch": 2.2982456140350878,
+      "grad_norm": 0.5937571058925619,
+      "learning_rate": 3.076539676856101e-05,
+      "loss": 0.4618,
+      "step": 131
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.5716558504183931,
+      "learning_rate": 3.048196560469758e-05,
+      "loss": 0.4041,
+      "step": 132
+    },
+    {
+      "epoch": 2.3333333333333335,
+      "grad_norm": 0.634625392073859,
+      "learning_rate": 3.0197792270443982e-05,
+      "loss": 0.4921,
+      "step": 133
+    },
+    {
+      "epoch": 2.3508771929824563,
+      "grad_norm": 0.6559610523673894,
+      "learning_rate": 2.9912915238320754e-05,
+      "loss": 0.4342,
+      "step": 134
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.5397048873931066,
+      "learning_rate": 2.9627373076117863e-05,
+      "loss": 0.4486,
+      "step": 135
+    },
+    {
+      "epoch": 2.3859649122807016,
+      "grad_norm": 0.5797600619266271,
+      "learning_rate": 2.9341204441673266e-05,
+      "loss": 0.4289,
+      "step": 136
+    },
+    {
+      "epoch": 2.4035087719298245,
+      "grad_norm": 0.9033150356178492,
+      "learning_rate": 2.905444807763919e-05,
+      "loss": 0.5367,
+      "step": 137
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.556800021535163,
+      "learning_rate": 2.876714280623708e-05,
+      "loss": 0.5174,
+      "step": 138
+    },
+    {
+      "epoch": 2.43859649122807,
+      "grad_norm": 0.616682217046264,
+      "learning_rate": 2.8479327524001636e-05,
+      "loss": 0.4688,
+      "step": 139
+    },
+    {
+      "epoch": 2.456140350877193,
+      "grad_norm": 0.5552569152326735,
+      "learning_rate": 2.8191041196514873e-05,
+      "loss": 0.4491,
+      "step": 140
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.6477053194551796,
+      "learning_rate": 2.7902322853130757e-05,
+      "loss": 0.4827,
+      "step": 141
+    },
+    {
+      "epoch": 2.4912280701754383,
+      "grad_norm": 0.5491240861447861,
+      "learning_rate": 2.761321158169134e-05,
+      "loss": 0.4692,
+      "step": 142
+    },
+    {
+      "epoch": 2.5087719298245617,
+      "grad_norm": 0.602312657799136,
+      "learning_rate": 2.732374652323481e-05,
+      "loss": 0.3241,
+      "step": 143
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.6974645276221735,
+      "learning_rate": 2.7033966866696457e-05,
+      "loss": 0.4047,
+      "step": 144
+    },
+    {
+      "epoch": 2.543859649122807,
+      "grad_norm": 0.5527674831438305,
+      "learning_rate": 2.674391184360313e-05,
+      "loss": 0.4347,
+      "step": 145
+    },
+    {
+      "epoch": 2.56140350877193,
+      "grad_norm": 0.6452937810917556,
+      "learning_rate": 2.6453620722761896e-05,
+      "loss": 0.5426,
+      "step": 146
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.5175735957709079,
+      "learning_rate": 2.6163132804943675e-05,
+      "loss": 0.4494,
+      "step": 147
+    },
+    {
+      "epoch": 2.5964912280701755,
+      "grad_norm": 0.619694612710559,
+      "learning_rate": 2.587248741756253e-05,
+      "loss": 0.4411,
+      "step": 148
+    },
+    {
+      "epoch": 2.6140350877192984,
+      "grad_norm": 0.6178101742888467,
+      "learning_rate": 2.5581723909351406e-05,
+      "loss": 0.4064,
+      "step": 149
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.6954237921716457,
+      "learning_rate": 2.5290881645034932e-05,
+      "loss": 0.4798,
+      "step": 150
+    },
+    {
+      "epoch": 2.6491228070175437,
+      "grad_norm": 0.570691242690676,
+      "learning_rate": 2.5e-05,
+      "loss": 0.4998,
+      "step": 151
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 0.6082517842768821,
+      "learning_rate": 2.470911835496508e-05,
+      "loss": 0.4812,
+      "step": 152
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.6250425262111831,
+      "learning_rate": 2.4418276090648596e-05,
+      "loss": 0.4813,
+      "step": 153
+    },
+    {
+      "epoch": 2.7017543859649122,
+      "grad_norm": 0.627043458132832,
+      "learning_rate": 2.4127512582437485e-05,
+      "loss": 0.4756,
+      "step": 154
+    },
+    {
+      "epoch": 2.719298245614035,
+      "grad_norm": 0.49199769861698356,
+      "learning_rate": 2.3836867195056335e-05,
+      "loss": 0.486,
+      "step": 155
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.4990139529400001,
+      "learning_rate": 2.3546379277238107e-05,
+      "loss": 0.4435,
+      "step": 156
+    },
+    {
+      "epoch": 2.754385964912281,
+      "grad_norm": 0.4749085128267926,
+      "learning_rate": 2.3256088156396868e-05,
+      "loss": 0.4423,
+      "step": 157
+    },
+    {
+      "epoch": 2.7719298245614032,
+      "grad_norm": 0.4307154819359441,
+      "learning_rate": 2.2966033133303545e-05,
+      "loss": 0.3969,
+      "step": 158
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.489813416980427,
+      "learning_rate": 2.2676253476765196e-05,
+      "loss": 0.4193,
+      "step": 159
+    },
+    {
+      "epoch": 2.807017543859649,
+      "grad_norm": 0.52180420921395,
+      "learning_rate": 2.238678841830867e-05,
+      "loss": 0.3252,
+      "step": 160
+    },
+    {
+      "epoch": 2.824561403508772,
+      "grad_norm": 0.6188954986933813,
+      "learning_rate": 2.2097677146869242e-05,
+      "loss": 0.4225,
+      "step": 161
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.6410957980691102,
+      "learning_rate": 2.1808958803485136e-05,
+      "loss": 0.4468,
+      "step": 162
+    },
+    {
+      "epoch": 2.8596491228070176,
+      "grad_norm": 0.5645208549946816,
+      "learning_rate": 2.1520672475998373e-05,
+      "loss": 0.4591,
+      "step": 163
+    },
+    {
+      "epoch": 2.8771929824561404,
+      "grad_norm": 0.49641673967611727,
+      "learning_rate": 2.1232857193762924e-05,
+      "loss": 0.4674,
+      "step": 164
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.6055492811016956,
+      "learning_rate": 2.0945551922360818e-05,
+      "loss": 0.4929,
+      "step": 165
+    },
+    {
+      "epoch": 2.912280701754386,
+      "grad_norm": 0.42221476564217447,
+      "learning_rate": 2.0658795558326743e-05,
+      "loss": 0.4702,
+      "step": 166
+    },
+    {
+      "epoch": 2.9298245614035086,
+      "grad_norm": 0.6260672302555065,
+      "learning_rate": 2.037262692388214e-05,
+      "loss": 0.4138,
+      "step": 167
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.4813998932824152,
+      "learning_rate": 2.0087084761679245e-05,
+      "loss": 0.4126,
+      "step": 168
+    },
+    {
+      "epoch": 2.9649122807017543,
+      "grad_norm": 0.4863424952466412,
+      "learning_rate": 1.980220772955602e-05,
+      "loss": 0.4428,
+      "step": 169
+    },
+    {
+      "epoch": 2.982456140350877,
+      "grad_norm": 0.4914395244827182,
+      "learning_rate": 1.9518034395302414e-05,
+      "loss": 0.4524,
+      "step": 170
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.43936171740815994,
+      "learning_rate": 1.9234603231438995e-05,
+      "loss": 0.2268,
+      "step": 171
+    },
+    {
+      "epoch": 3.017543859649123,
+      "grad_norm": 1.1355956358305708,
+      "learning_rate": 1.895195261000831e-05,
+      "loss": 0.2215,
+      "step": 172
+    },
+    {
+      "epoch": 3.0350877192982457,
+      "grad_norm": 0.8879169891556423,
+      "learning_rate": 1.8670120797379958e-05,
+      "loss": 0.2855,
+      "step": 173
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.665813131495675,
+      "learning_rate": 1.838914594906995e-05,
+      "loss": 0.2994,
+      "step": 174
+    },
+    {
+      "epoch": 3.0701754385964914,
+      "grad_norm": 1.881390073909953,
+      "learning_rate": 1.8109066104575023e-05,
+      "loss": 0.3107,
+      "step": 175
+    },
+    {
+      "epoch": 3.087719298245614,
+      "grad_norm": 1.6532943065317012,
+      "learning_rate": 1.7829919182222752e-05,
+      "loss": 0.2783,
+      "step": 176
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.9133784251968402,
+      "learning_rate": 1.755174297403795e-05,
+      "loss": 0.2966,
+      "step": 177
+    },
+    {
+      "epoch": 3.1228070175438596,
+      "grad_norm": 0.5705573928382189,
+      "learning_rate": 1.7274575140626318e-05,
+      "loss": 0.2718,
+      "step": 178
+    },
+    {
+      "epoch": 3.1403508771929824,
+      "grad_norm": 0.6881219091003131,
+      "learning_rate": 1.699845320607571e-05,
+      "loss": 0.2347,
+      "step": 179
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.7303903413309477,
+      "learning_rate": 1.672341455287605e-05,
+      "loss": 0.2591,
+      "step": 180
+    },
+    {
+      "epoch": 3.175438596491228,
+      "grad_norm": 0.7801781557780155,
+      "learning_rate": 1.6449496416858284e-05,
+      "loss": 0.2476,
+      "step": 181
+    },
+    {
+      "epoch": 3.192982456140351,
+      "grad_norm": 0.7419803348350285,
+      "learning_rate": 1.617673588215328e-05,
+      "loss": 0.3191,
+      "step": 182
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.5556644625179776,
+      "learning_rate": 1.5905169876171223e-05,
+      "loss": 0.2725,
+      "step": 183
+    },
+    {
+      "epoch": 3.2280701754385963,
+      "grad_norm": 0.6205553909876562,
+      "learning_rate": 1.56348351646022e-05,
+      "loss": 0.2242,
+      "step": 184
+    },
+    {
+      "epoch": 3.245614035087719,
+      "grad_norm": 0.7346333350296814,
+      "learning_rate": 1.5365768346438797e-05,
+      "loss": 0.2494,
+      "step": 185
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.7115758357646363,
+      "learning_rate": 1.509800584902108e-05,
+      "loss": 0.1999,
+      "step": 186
+    },
+    {
+      "epoch": 3.280701754385965,
+      "grad_norm": 0.6512394196525719,
+      "learning_rate": 1.4831583923104999e-05,
+      "loss": 0.2414,
+      "step": 187
+    },
+    {
+      "epoch": 3.2982456140350878,
+      "grad_norm": 0.5541981090984397,
+      "learning_rate": 1.4566538637954554e-05,
+      "loss": 0.2786,
+      "step": 188
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.5528407471656007,
+      "learning_rate": 1.430290587645865e-05,
+      "loss": 0.2326,
+      "step": 189
+    },
+    {
+      "epoch": 3.3333333333333335,
+      "grad_norm": 0.6075883585844755,
+      "learning_rate": 1.4040721330273062e-05,
+      "loss": 0.2318,
+      "step": 190
+    },
+    {
+      "epoch": 3.3508771929824563,
+      "grad_norm": 0.6185476753367053,
+      "learning_rate": 1.3780020494988446e-05,
+      "loss": 0.2443,
+      "step": 191
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.4377019300771013,
+      "learning_rate": 1.3520838665324703e-05,
+      "loss": 0.2066,
+      "step": 192
+    },
+    {
+      "epoch": 3.3859649122807016,
+      "grad_norm": 0.4914459309520691,
+      "learning_rate": 1.3263210930352737e-05,
+      "loss": 0.2418,
+      "step": 193
+    },
+    {
+      "epoch": 3.4035087719298245,
+      "grad_norm": 0.5415582033178936,
+      "learning_rate": 1.3007172168743854e-05,
+      "loss": 0.2318,
+      "step": 194
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.45406450767877093,
+      "learning_rate": 1.2752757044047827e-05,
+      "loss": 0.195,
+      "step": 195
+    },
+    {
+      "epoch": 3.43859649122807,
+      "grad_norm": 0.5488803395501969,
+      "learning_rate": 1.2500000000000006e-05,
+      "loss": 0.251,
+      "step": 196
+    },
+    {
+      "epoch": 3.456140350877193,
+      "grad_norm": 0.522803773721285,
+      "learning_rate": 1.2248935255858117e-05,
+      "loss": 0.1593,
+      "step": 197
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.47485404771046913,
+      "learning_rate": 1.1999596801769616e-05,
+      "loss": 0.1976,
+      "step": 198
+    },
+    {
+      "epoch": 3.4912280701754383,
+      "grad_norm": 0.45733928647155575,
+      "learning_rate": 1.175201839416988e-05,
+      "loss": 0.2484,
+      "step": 199
+    },
+    {
+      "epoch": 3.5087719298245617,
+      "grad_norm": 0.5003022712086583,
+      "learning_rate": 1.1506233551212186e-05,
+      "loss": 0.2256,
+      "step": 200
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.46801680996440476,
+      "learning_rate": 1.126227554822985e-05,
+      "loss": 0.2361,
+      "step": 201
+    },
+    {
+      "epoch": 3.543859649122807,
+      "grad_norm": 0.41528464228686823,
+      "learning_rate": 1.1020177413231334e-05,
+      "loss": 0.1803,
+      "step": 202
+    },
+    {
+      "epoch": 3.56140350877193,
+      "grad_norm": 0.5094107474878486,
+      "learning_rate": 1.0779971922428711e-05,
+      "loss": 0.2889,
+      "step": 203
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.9262271323270204,
+      "learning_rate": 1.0541691595800337e-05,
+      "loss": 0.2008,
+      "step": 204
+    },
+    {
+      "epoch": 3.5964912280701755,
+      "grad_norm": 0.48893637894975905,
+      "learning_rate": 1.0305368692688174e-05,
+      "loss": 0.2175,
+      "step": 205
+    },
+    {
+      "epoch": 3.6140350877192984,
+      "grad_norm": 0.44430495764638256,
+      "learning_rate": 1.0071035207430352e-05,
+      "loss": 0.195,
+      "step": 206
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.3973316120978888,
+      "learning_rate": 9.838722865029673e-06,
+      "loss": 0.1843,
+      "step": 207
+    },
+    {
+      "epoch": 3.6491228070175437,
+      "grad_norm": 0.4732358148087521,
+      "learning_rate": 9.608463116858542e-06,
+      "loss": 0.1883,
+      "step": 208
+    },
+    {
+      "epoch": 3.6666666666666665,
+      "grad_norm": 0.527945690801211,
+      "learning_rate": 9.380287136401e-06,
+      "loss": 0.2278,
+      "step": 209
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.478450642755059,
+      "learning_rate": 9.154225815032242e-06,
+      "loss": 0.2136,
+      "step": 210
+    },
+    {
+      "epoch": 3.7017543859649122,
+      "grad_norm": 0.569419085702566,
+      "learning_rate": 8.930309757836517e-06,
+      "loss": 0.2364,
+      "step": 211
+    },
+    {
+      "epoch": 3.719298245614035,
+      "grad_norm": 0.4498015088386311,
+      "learning_rate": 8.708569279463622e-06,
+      "loss": 0.2383,
+      "step": 212
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.5301991480165194,
+      "learning_rate": 8.489034400024812e-06,
+      "loss": 0.2815,
+      "step": 213
+    },
+    {
+      "epoch": 3.754385964912281,
+      "grad_norm": 0.5082746052836551,
+      "learning_rate": 8.271734841028553e-06,
+      "loss": 0.2875,
+      "step": 214
+    },
+    {
+      "epoch": 3.7719298245614032,
+      "grad_norm": 0.5385567049286154,
+      "learning_rate": 8.056700021356694e-06,
+      "loss": 0.3047,
+      "step": 215
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.4477012157390016,
+      "learning_rate": 7.843959053281663e-06,
+      "loss": 0.2192,
+      "step": 216
+    },
+    {
+      "epoch": 3.807017543859649,
+      "grad_norm": 0.47201031934382787,
+      "learning_rate": 7.633540738525066e-06,
+      "loss": 0.2535,
+      "step": 217
+    },
+    {
+      "epoch": 3.824561403508772,
+      "grad_norm": 0.49653745081511047,
+      "learning_rate": 7.4254735643584564e-06,
+      "loss": 0.2446,
+      "step": 218
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.4839157057987588,
+      "learning_rate": 7.219785699746573e-06,
+      "loss": 0.2269,
+      "step": 219
+    },
+    {
+      "epoch": 3.8596491228070176,
+      "grad_norm": 0.5128804127093707,
+      "learning_rate": 7.016504991533726e-06,
+      "loss": 0.1725,
+      "step": 220
+    },
+    {
+      "epoch": 3.8771929824561404,
+      "grad_norm": 0.4252860730860363,
+      "learning_rate": 6.815658960673782e-06,
+      "loss": 0.2413,
+      "step": 221
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.4768997514329751,
+      "learning_rate": 6.617274798504286e-06,
+      "loss": 0.2081,
+      "step": 222
+    },
+    {
+      "epoch": 3.912280701754386,
+      "grad_norm": 0.40163031023486234,
+      "learning_rate": 6.421379363065142e-06,
+      "loss": 0.2014,
+      "step": 223
+    },
+    {
+      "epoch": 3.9298245614035086,
+      "grad_norm": 0.49380964047273734,
+      "learning_rate": 6.22799917546252e-06,
+      "loss": 0.2047,
+      "step": 224
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.5120667197627459,
+      "learning_rate": 6.037160416278278e-06,
+      "loss": 0.2462,
+      "step": 225
+    },
+    {
+      "epoch": 3.9649122807017543,
+      "grad_norm": 0.501745766842808,
+      "learning_rate": 5.848888922025553e-06,
+      "loss": 0.2638,
+      "step": 226
+    },
+    {
+      "epoch": 3.982456140350877,
+      "grad_norm": 0.4382166272747192,
+      "learning_rate": 5.663210181650905e-06,
+      "loss": 0.2614,
+      "step": 227
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.408700828752829,
+      "learning_rate": 5.48014933308352e-06,
+      "loss": 0.1385,
+      "step": 228
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 285,
+  "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": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 52405071052800.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2f7fd226f2e0ad24fee5a35130a1e108a0e743561f941c221ba049dbdc6ca8dc
+size 8081