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:f4d2325b0404ae55b8f0c1375b344fd75e8ab47d807fc553a3595ea011e3800e
+size 4997184968

model-00002-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d4009703c415c5a6c27cdcea2eb2e9d957b3cbdc0089ac7c2794df53850b865e
+size 4997741608

model-00003-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:13296a42b8db68d7686a18b97d93f02bbbd3fb9a6c49fe4d0cab5f5e73712a12
+size 4997742208

model-00004-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a3c3f6f12239e3097a4867e77a2855ecd423549c698abefc3c4053443898d9d4
+size 4997743184

model-00005-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0d572542f848d5f374cfca898f133940fd5028ba81570e28d6cdc35c8de825f9
+size 4997743184

model-00006-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b5cf916b1d2c2efbf6c488e8bc3b03dcbf990219bd4b2985a0c41bc103009f28
+size 4997743184

model-00007-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:89d3211e74b95ad952bd496ba538d0e92f51e301f6ce8de60cfbee3135c8fc73
+size 4997743184

model-00008-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:989c7ed0e003f1af0df4a03ccea1b327752618808bb2d173f89165ead59093d3
+size 4997743184

model-00009-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8e2ef22a062e1f1fb3c12a2d9dbbc5c29bf70ddfb80b60af75bc935e10f98e20
+size 4997743184

model-00010-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:023af18188aa6de03697f9edb774bf93da31dac135abb1a59fed18c7dabd1bfc
+size 4997743184

model-00011-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:78c9ff833fcd7e8460995a2c3d44edb64e16f5484def59bab3cbbdac880c775d
+size 4997743184

model-00012-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f463fca977f5a2dddfad8b775e38459518c349cc699c189f7241d70c405d738d
+size 4997743184

model-00013-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d309a299bc31df722c86247ef0e8d7daac26b820f00eb56b4b93e1bf4697dc64
+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": 2.1660402021693237,
+      "learning_rate": 0.0,
+      "loss": 0.8483,
+      "step": 1
+    },
+    {
+      "epoch": 0.03508771929824561,
+      "grad_norm": 2.066760278075337,
+      "learning_rate": 6.666666666666667e-07,
+      "loss": 0.8763,
+      "step": 2
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 2.336128090338058,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 1.0207,
+      "step": 3
+    },
+    {
+      "epoch": 0.07017543859649122,
+      "grad_norm": 2.3414067030265864,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.0621,
+      "step": 4
+    },
+    {
+      "epoch": 0.08771929824561403,
+      "grad_norm": 1.8795815283512405,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 0.8563,
+      "step": 5
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 2.0304843539502357,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.008,
+      "step": 6
+    },
+    {
+      "epoch": 0.12280701754385964,
+      "grad_norm": 1.7342900365375629,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.9974,
+      "step": 7
+    },
+    {
+      "epoch": 0.14035087719298245,
+      "grad_norm": 1.7130893901160444,
+      "learning_rate": 4.666666666666667e-06,
+      "loss": 0.9462,
+      "step": 8
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 1.4339421646705366,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 0.9611,
+      "step": 9
+    },
+    {
+      "epoch": 0.17543859649122806,
+      "grad_norm": 1.250245872035883,
+      "learning_rate": 6e-06,
+      "loss": 0.9861,
+      "step": 10
+    },
+    {
+      "epoch": 0.19298245614035087,
+      "grad_norm": 1.3695896453736731,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.0293,
+      "step": 11
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 1.3836439558544265,
+      "learning_rate": 7.333333333333333e-06,
+      "loss": 0.8978,
+      "step": 12
+    },
+    {
+      "epoch": 0.22807017543859648,
+      "grad_norm": 1.4917947699421266,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.9339,
+      "step": 13
+    },
+    {
+      "epoch": 0.24561403508771928,
+      "grad_norm": 1.4339515433488395,
+      "learning_rate": 8.666666666666668e-06,
+      "loss": 0.9121,
+      "step": 14
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 1.2541506853697504,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 0.9702,
+      "step": 15
+    },
+    {
+      "epoch": 0.2807017543859649,
+      "grad_norm": 1.3275538971055934,
+      "learning_rate": 1e-05,
+      "loss": 0.9215,
+      "step": 16
+    },
+    {
+      "epoch": 0.2982456140350877,
+      "grad_norm": 1.3917244346005202,
+      "learning_rate": 9.999661540018812e-06,
+      "loss": 1.0186,
+      "step": 17
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 1.361545760509589,
+      "learning_rate": 9.99864620589731e-06,
+      "loss": 0.902,
+      "step": 18
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 1.1217959407592388,
+      "learning_rate": 9.99695413509548e-06,
+      "loss": 0.9221,
+      "step": 19
+    },
+    {
+      "epoch": 0.3508771929824561,
+      "grad_norm": 1.014157243869696,
+      "learning_rate": 9.994585556692624e-06,
+      "loss": 0.9581,
+      "step": 20
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 0.8686227582358618,
+      "learning_rate": 9.991540791356342e-06,
+      "loss": 0.8953,
+      "step": 21
+    },
+    {
+      "epoch": 0.38596491228070173,
+      "grad_norm": 0.7872806059961304,
+      "learning_rate": 9.987820251299121e-06,
+      "loss": 0.7629,
+      "step": 22
+    },
+    {
+      "epoch": 0.40350877192982454,
+      "grad_norm": 0.9119427318951265,
+      "learning_rate": 9.98342444022253e-06,
+      "loss": 0.8971,
+      "step": 23
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 0.7480898242075966,
+      "learning_rate": 9.978353953249023e-06,
+      "loss": 0.7949,
+      "step": 24
+    },
+    {
+      "epoch": 0.43859649122807015,
+      "grad_norm": 0.8256867027063682,
+      "learning_rate": 9.972609476841368e-06,
+      "loss": 0.9236,
+      "step": 25
+    },
+    {
+      "epoch": 0.45614035087719296,
+      "grad_norm": 0.9297755915177187,
+      "learning_rate": 9.966191788709716e-06,
+      "loss": 0.8878,
+      "step": 26
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 0.8391196889638465,
+      "learning_rate": 9.959101757706308e-06,
+      "loss": 0.8082,
+      "step": 27
+    },
+    {
+      "epoch": 0.49122807017543857,
+      "grad_norm": 0.7189700520044675,
+      "learning_rate": 9.951340343707852e-06,
+      "loss": 0.8765,
+      "step": 28
+    },
+    {
+      "epoch": 0.5087719298245614,
+      "grad_norm": 0.5846628679801547,
+      "learning_rate": 9.942908597485558e-06,
+      "loss": 0.8978,
+      "step": 29
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 0.6168160394185087,
+      "learning_rate": 9.933807660562898e-06,
+      "loss": 0.7833,
+      "step": 30
+    },
+    {
+      "epoch": 0.543859649122807,
+      "grad_norm": 0.5362022880933746,
+      "learning_rate": 9.924038765061042e-06,
+      "loss": 0.8397,
+      "step": 31
+    },
+    {
+      "epoch": 0.5614035087719298,
+      "grad_norm": 0.555268220129423,
+      "learning_rate": 9.913603233532067e-06,
+      "loss": 0.7705,
+      "step": 32
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 3.4423211486243988,
+      "learning_rate": 9.902502478779897e-06,
+      "loss": 0.8688,
+      "step": 33
+    },
+    {
+      "epoch": 0.5964912280701754,
+      "grad_norm": 0.7279458596154308,
+      "learning_rate": 9.890738003669029e-06,
+      "loss": 0.9688,
+      "step": 34
+    },
+    {
+      "epoch": 0.6140350877192983,
+      "grad_norm": 0.6799464505722792,
+      "learning_rate": 9.878311400921072e-06,
+      "loss": 0.8129,
+      "step": 35
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 0.6444622306171982,
+      "learning_rate": 9.86522435289912e-06,
+      "loss": 0.8282,
+      "step": 36
+    },
+    {
+      "epoch": 0.6491228070175439,
+      "grad_norm": 0.5798791871759987,
+      "learning_rate": 9.851478631379982e-06,
+      "loss": 0.8631,
+      "step": 37
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 0.5543709338798263,
+      "learning_rate": 9.83707609731432e-06,
+      "loss": 0.8756,
+      "step": 38
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.5889746910989742,
+      "learning_rate": 9.822018700574696e-06,
+      "loss": 0.8984,
+      "step": 39
+    },
+    {
+      "epoch": 0.7017543859649122,
+      "grad_norm": 0.4753624542604962,
+      "learning_rate": 9.806308479691595e-06,
+      "loss": 0.8641,
+      "step": 40
+    },
+    {
+      "epoch": 0.7192982456140351,
+      "grad_norm": 0.5751109704231023,
+      "learning_rate": 9.789947561577445e-06,
+      "loss": 0.8687,
+      "step": 41
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.5583826594544032,
+      "learning_rate": 9.77293816123866e-06,
+      "loss": 0.8037,
+      "step": 42
+    },
+    {
+      "epoch": 0.7543859649122807,
+      "grad_norm": 0.5397303962731949,
+      "learning_rate": 9.755282581475769e-06,
+      "loss": 0.8364,
+      "step": 43
+    },
+    {
+      "epoch": 0.7719298245614035,
+      "grad_norm": 0.526733335023532,
+      "learning_rate": 9.736983212571646e-06,
+      "loss": 0.7532,
+      "step": 44
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.505071250741911,
+      "learning_rate": 9.718042531967918e-06,
+      "loss": 0.8691,
+      "step": 45
+    },
+    {
+      "epoch": 0.8070175438596491,
+      "grad_norm": 0.543790864733672,
+      "learning_rate": 9.698463103929542e-06,
+      "loss": 0.8461,
+      "step": 46
+    },
+    {
+      "epoch": 0.8245614035087719,
+      "grad_norm": 1.2888799576021894,
+      "learning_rate": 9.678247579197658e-06,
+      "loss": 0.8505,
+      "step": 47
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.43256000451405635,
+      "learning_rate": 9.657398694630713e-06,
+      "loss": 0.8345,
+      "step": 48
+    },
+    {
+      "epoch": 0.8596491228070176,
+      "grad_norm": 0.5019354212033829,
+      "learning_rate": 9.635919272833938e-06,
+      "loss": 0.9263,
+      "step": 49
+    },
+    {
+      "epoch": 0.8771929824561403,
+      "grad_norm": 0.4681295329066555,
+      "learning_rate": 9.613812221777212e-06,
+      "loss": 0.6669,
+      "step": 50
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 0.4877429316552433,
+      "learning_rate": 9.591080534401371e-06,
+      "loss": 0.8957,
+      "step": 51
+    },
+    {
+      "epoch": 0.9122807017543859,
+      "grad_norm": 0.4930148726421091,
+      "learning_rate": 9.567727288213005e-06,
+      "loss": 0.8642,
+      "step": 52
+    },
+    {
+      "epoch": 0.9298245614035088,
+      "grad_norm": 0.45304651127445467,
+      "learning_rate": 9.543755644867823e-06,
+      "loss": 0.816,
+      "step": 53
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.49190524120268453,
+      "learning_rate": 9.519168849742603e-06,
+      "loss": 0.7711,
+      "step": 54
+    },
+    {
+      "epoch": 0.9649122807017544,
+      "grad_norm": 0.5036422964531686,
+      "learning_rate": 9.493970231495836e-06,
+      "loss": 0.8103,
+      "step": 55
+    },
+    {
+      "epoch": 0.9824561403508771,
+      "grad_norm": 0.47457915853612875,
+      "learning_rate": 9.468163201617063e-06,
+      "loss": 0.8217,
+      "step": 56
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.542410510587192,
+      "learning_rate": 9.441751253965022e-06,
+      "loss": 0.7332,
+      "step": 57
+    },
+    {
+      "epoch": 1.0175438596491229,
+      "grad_norm": 0.5669063634388459,
+      "learning_rate": 9.414737964294636e-06,
+      "loss": 0.8413,
+      "step": 58
+    },
+    {
+      "epoch": 1.0350877192982457,
+      "grad_norm": 0.5139063790936388,
+      "learning_rate": 9.38712698977291e-06,
+      "loss": 0.73,
+      "step": 59
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.5257839792559575,
+      "learning_rate": 9.358922068483813e-06,
+      "loss": 0.8163,
+      "step": 60
+    },
+    {
+      "epoch": 1.0701754385964912,
+      "grad_norm": 0.4995493978469428,
+      "learning_rate": 9.330127018922195e-06,
+      "loss": 0.6913,
+      "step": 61
+    },
+    {
+      "epoch": 1.087719298245614,
+      "grad_norm": 0.4997592964667348,
+      "learning_rate": 9.30074573947683e-06,
+      "loss": 0.7771,
+      "step": 62
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 0.609895138625011,
+      "learning_rate": 9.27078220790263e-06,
+      "loss": 0.8055,
+      "step": 63
+    },
+    {
+      "epoch": 1.1228070175438596,
+      "grad_norm": 0.5725193602176493,
+      "learning_rate": 9.24024048078213e-06,
+      "loss": 0.7941,
+      "step": 64
+    },
+    {
+      "epoch": 1.1403508771929824,
+      "grad_norm": 0.4577922503628831,
+      "learning_rate": 9.209124692976287e-06,
+      "loss": 0.7205,
+      "step": 65
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.5180354968954766,
+      "learning_rate": 9.177439057064684e-06,
+      "loss": 0.7195,
+      "step": 66
+    },
+    {
+      "epoch": 1.1754385964912282,
+      "grad_norm": 0.5546103097757371,
+      "learning_rate": 9.145187862775208e-06,
+      "loss": 0.7167,
+      "step": 67
+    },
+    {
+      "epoch": 1.1929824561403508,
+      "grad_norm": 0.5110846501385273,
+      "learning_rate": 9.112375476403313e-06,
+      "loss": 0.7667,
+      "step": 68
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.5600079369794507,
+      "learning_rate": 9.079006340220862e-06,
+      "loss": 0.8244,
+      "step": 69
+    },
+    {
+      "epoch": 1.2280701754385965,
+      "grad_norm": 0.485898327690189,
+      "learning_rate": 9.045084971874738e-06,
+      "loss": 0.7602,
+      "step": 70
+    },
+    {
+      "epoch": 1.2456140350877192,
+      "grad_norm": 0.4730650001066968,
+      "learning_rate": 9.01061596377522e-06,
+      "loss": 0.8109,
+      "step": 71
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.5740085400544082,
+      "learning_rate": 8.97560398247424e-06,
+      "loss": 0.7747,
+      "step": 72
+    },
+    {
+      "epoch": 1.280701754385965,
+      "grad_norm": 0.5081779390222444,
+      "learning_rate": 8.94005376803361e-06,
+      "loss": 0.776,
+      "step": 73
+    },
+    {
+      "epoch": 1.2982456140350878,
+      "grad_norm": 0.45617579227052085,
+      "learning_rate": 8.903970133383297e-06,
+      "loss": 0.6957,
+      "step": 74
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.4748587212385723,
+      "learning_rate": 8.867357963669821e-06,
+      "loss": 0.8237,
+      "step": 75
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 0.4264863459857496,
+      "learning_rate": 8.83022221559489e-06,
+      "loss": 0.7191,
+      "step": 76
+    },
+    {
+      "epoch": 1.3508771929824561,
+      "grad_norm": 0.4828407300056768,
+      "learning_rate": 8.792567916744346e-06,
+      "loss": 0.6896,
+      "step": 77
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.5823481428367913,
+      "learning_rate": 8.754400164907496e-06,
+      "loss": 0.8076,
+      "step": 78
+    },
+    {
+      "epoch": 1.3859649122807016,
+      "grad_norm": 0.543985499862031,
+      "learning_rate": 8.715724127386971e-06,
+      "loss": 0.7921,
+      "step": 79
+    },
+    {
+      "epoch": 1.4035087719298245,
+      "grad_norm": 0.470464475738681,
+      "learning_rate": 8.676545040299145e-06,
+      "loss": 0.7418,
+      "step": 80
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 0.5833913891874376,
+      "learning_rate": 8.636868207865244e-06,
+      "loss": 0.7657,
+      "step": 81
+    },
+    {
+      "epoch": 1.4385964912280702,
+      "grad_norm": 0.5007872118568844,
+      "learning_rate": 8.596699001693257e-06,
+      "loss": 0.7516,
+      "step": 82
+    },
+    {
+      "epoch": 1.456140350877193,
+      "grad_norm": 0.4812479534876582,
+      "learning_rate": 8.556042860050686e-06,
+      "loss": 0.6863,
+      "step": 83
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.4557048077866418,
+      "learning_rate": 8.51490528712831e-06,
+      "loss": 0.6181,
+      "step": 84
+    },
+    {
+      "epoch": 1.4912280701754386,
+      "grad_norm": 0.4737386987709376,
+      "learning_rate": 8.473291852294986e-06,
+      "loss": 0.7704,
+      "step": 85
+    },
+    {
+      "epoch": 1.5087719298245614,
+      "grad_norm": 0.5232119821777624,
+      "learning_rate": 8.43120818934367e-06,
+      "loss": 0.7651,
+      "step": 86
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.46057752266184454,
+      "learning_rate": 8.388659995728662e-06,
+      "loss": 0.884,
+      "step": 87
+    },
+    {
+      "epoch": 1.543859649122807,
+      "grad_norm": 0.44290438508250374,
+      "learning_rate": 8.345653031794292e-06,
+      "loss": 0.7692,
+      "step": 88
+    },
+    {
+      "epoch": 1.5614035087719298,
+      "grad_norm": 0.4338999688228569,
+      "learning_rate": 8.302193119995038e-06,
+      "loss": 0.8082,
+      "step": 89
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.45036462410127814,
+      "learning_rate": 8.258286144107277e-06,
+      "loss": 0.7116,
+      "step": 90
+    },
+    {
+      "epoch": 1.5964912280701755,
+      "grad_norm": 0.5010305972777187,
+      "learning_rate": 8.213938048432697e-06,
+      "loss": 0.7455,
+      "step": 91
+    },
+    {
+      "epoch": 1.6140350877192984,
+      "grad_norm": 0.4490390837532202,
+      "learning_rate": 8.16915483699355e-06,
+      "loss": 0.7559,
+      "step": 92
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.4012083319569404,
+      "learning_rate": 8.123942572719801e-06,
+      "loss": 0.6984,
+      "step": 93
+    },
+    {
+      "epoch": 1.6491228070175439,
+      "grad_norm": 0.40187726090980297,
+      "learning_rate": 8.078307376628292e-06,
+      "loss": 0.7012,
+      "step": 94
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 0.4955120113441123,
+      "learning_rate": 8.032255426994069e-06,
+      "loss": 0.8023,
+      "step": 95
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.4382099371460066,
+      "learning_rate": 7.985792958513932e-06,
+      "loss": 0.7079,
+      "step": 96
+    },
+    {
+      "epoch": 1.7017543859649122,
+      "grad_norm": 0.5002258411852988,
+      "learning_rate": 7.938926261462366e-06,
+      "loss": 0.8722,
+      "step": 97
+    },
+    {
+      "epoch": 1.719298245614035,
+      "grad_norm": 0.45700036961609564,
+      "learning_rate": 7.891661680839932e-06,
+      "loss": 0.7493,
+      "step": 98
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.5874621412536426,
+      "learning_rate": 7.84400561551426e-06,
+      "loss": 0.7848,
+      "step": 99
+    },
+    {
+      "epoch": 1.7543859649122808,
+      "grad_norm": 0.4317640646106818,
+      "learning_rate": 7.795964517353734e-06,
+      "loss": 0.7308,
+      "step": 100
+    },
+    {
+      "epoch": 1.7719298245614035,
+      "grad_norm": 0.5408126651945797,
+      "learning_rate": 7.747544890354031e-06,
+      "loss": 0.7973,
+      "step": 101
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.43108178788348095,
+      "learning_rate": 7.698753289757565e-06,
+      "loss": 0.7964,
+      "step": 102
+    },
+    {
+      "epoch": 1.807017543859649,
+      "grad_norm": 0.5342248380167852,
+      "learning_rate": 7.649596321166024e-06,
+      "loss": 0.8131,
+      "step": 103
+    },
+    {
+      "epoch": 1.8245614035087718,
+      "grad_norm": 0.4432424592735311,
+      "learning_rate": 7.600080639646077e-06,
+      "loss": 0.6702,
+      "step": 104
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.4516614598334642,
+      "learning_rate": 7.550212948828377e-06,
+      "loss": 0.6417,
+      "step": 105
+    },
+    {
+      "epoch": 1.8596491228070176,
+      "grad_norm": 0.6130179819111611,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.8331,
+      "step": 106
+    },
+    {
+      "epoch": 1.8771929824561404,
+      "grad_norm": 0.5257558821927422,
+      "learning_rate": 7.449448591190436e-06,
+      "loss": 0.763,
+      "step": 107
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.46986051203352247,
+      "learning_rate": 7.398565566251232e-06,
+      "loss": 0.7845,
+      "step": 108
+    },
+    {
+      "epoch": 1.912280701754386,
+      "grad_norm": 0.44686310461762097,
+      "learning_rate": 7.347357813929455e-06,
+      "loss": 0.6991,
+      "step": 109
+    },
+    {
+      "epoch": 1.9298245614035088,
+      "grad_norm": 0.5170290400152182,
+      "learning_rate": 7.295832266935059e-06,
+      "loss": 0.8038,
+      "step": 110
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.5261204167045516,
+      "learning_rate": 7.243995901002312e-06,
+      "loss": 0.804,
+      "step": 111
+    },
+    {
+      "epoch": 1.9649122807017543,
+      "grad_norm": 0.5092010260972257,
+      "learning_rate": 7.191855733945388e-06,
+      "loss": 0.7742,
+      "step": 112
+    },
+    {
+      "epoch": 1.9824561403508771,
+      "grad_norm": 0.4980357283687526,
+      "learning_rate": 7.1394188247082715e-06,
+      "loss": 0.848,
+      "step": 113
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.593649456241614,
+      "learning_rate": 7.08669227240909e-06,
+      "loss": 0.753,
+      "step": 114
+    },
+    {
+      "epoch": 2.017543859649123,
+      "grad_norm": 0.6680821344003489,
+      "learning_rate": 7.033683215379002e-06,
+      "loss": 0.6322,
+      "step": 115
+    },
+    {
+      "epoch": 2.0350877192982457,
+      "grad_norm": 0.47737821757077353,
+      "learning_rate": 6.980398830195785e-06,
+      "loss": 0.6719,
+      "step": 116
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.5898593946674421,
+      "learning_rate": 6.9268463307122425e-06,
+      "loss": 0.848,
+      "step": 117
+    },
+    {
+      "epoch": 2.0701754385964914,
+      "grad_norm": 0.9800544780036837,
+      "learning_rate": 6.873032967079562e-06,
+      "loss": 0.711,
+      "step": 118
+    },
+    {
+      "epoch": 2.087719298245614,
+      "grad_norm": 0.5696743185574907,
+      "learning_rate": 6.818966024765758e-06,
+      "loss": 0.6902,
+      "step": 119
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.5255113989360072,
+      "learning_rate": 6.7646528235693445e-06,
+      "loss": 0.6721,
+      "step": 120
+    },
+    {
+      "epoch": 2.1228070175438596,
+      "grad_norm": 0.6310001045851209,
+      "learning_rate": 6.710100716628345e-06,
+      "loss": 0.6626,
+      "step": 121
+    },
+    {
+      "epoch": 2.1403508771929824,
+      "grad_norm": 0.7341630672380693,
+      "learning_rate": 6.655317089424791e-06,
+      "loss": 0.6225,
+      "step": 122
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.4957450021119023,
+      "learning_rate": 6.600309358784858e-06,
+      "loss": 0.7674,
+      "step": 123
+    },
+    {
+      "epoch": 2.175438596491228,
+      "grad_norm": 0.5643940708429578,
+      "learning_rate": 6.545084971874738e-06,
+      "loss": 0.6719,
+      "step": 124
+    },
+    {
+      "epoch": 2.192982456140351,
+      "grad_norm": 0.790170917856681,
+      "learning_rate": 6.48965140519241e-06,
+      "loss": 0.7664,
+      "step": 125
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.5698899323282437,
+      "learning_rate": 6.434016163555452e-06,
+      "loss": 0.6754,
+      "step": 126
+    },
+    {
+      "epoch": 2.2280701754385963,
+      "grad_norm": 0.590125688694081,
+      "learning_rate": 6.378186779084996e-06,
+      "loss": 0.6214,
+      "step": 127
+    },
+    {
+      "epoch": 2.245614035087719,
+      "grad_norm": 0.5135768379522944,
+      "learning_rate": 6.322170810186013e-06,
+      "loss": 0.6463,
+      "step": 128
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.5752270296523769,
+      "learning_rate": 6.26597584052401e-06,
+      "loss": 0.6924,
+      "step": 129
+    },
+    {
+      "epoch": 2.280701754385965,
+      "grad_norm": 0.6836486114054455,
+      "learning_rate": 6.209609477998339e-06,
+      "loss": 0.7045,
+      "step": 130
+    },
+    {
+      "epoch": 2.2982456140350878,
+      "grad_norm": 0.5593380267846941,
+      "learning_rate": 6.153079353712201e-06,
+      "loss": 0.6836,
+      "step": 131
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.4906882073271772,
+      "learning_rate": 6.0963931209395165e-06,
+      "loss": 0.646,
+      "step": 132
+    },
+    {
+      "epoch": 2.3333333333333335,
+      "grad_norm": 0.7826423134186171,
+      "learning_rate": 6.039558454088796e-06,
+      "loss": 0.7445,
+      "step": 133
+    },
+    {
+      "epoch": 2.3508771929824563,
+      "grad_norm": 0.6785590925380028,
+      "learning_rate": 5.982583047664151e-06,
+      "loss": 0.6597,
+      "step": 134
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.465375039314735,
+      "learning_rate": 5.925474615223573e-06,
+      "loss": 0.6717,
+      "step": 135
+    },
+    {
+      "epoch": 2.3859649122807016,
+      "grad_norm": 0.7177144944335283,
+      "learning_rate": 5.8682408883346535e-06,
+      "loss": 0.665,
+      "step": 136
+    },
+    {
+      "epoch": 2.4035087719298245,
+      "grad_norm": 0.5735799303530826,
+      "learning_rate": 5.810889615527839e-06,
+      "loss": 0.7826,
+      "step": 137
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.5286502802421806,
+      "learning_rate": 5.753428561247416e-06,
+      "loss": 0.7587,
+      "step": 138
+    },
+    {
+      "epoch": 2.43859649122807,
+      "grad_norm": 0.5403249749485503,
+      "learning_rate": 5.695865504800328e-06,
+      "loss": 0.7162,
+      "step": 139
+    },
+    {
+      "epoch": 2.456140350877193,
+      "grad_norm": 0.4334762037650122,
+      "learning_rate": 5.638208239302975e-06,
+      "loss": 0.6737,
+      "step": 140
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.5935027147735242,
+      "learning_rate": 5.5804645706261515e-06,
+      "loss": 0.7183,
+      "step": 141
+    },
+    {
+      "epoch": 2.4912280701754383,
+      "grad_norm": 0.5051267174145002,
+      "learning_rate": 5.522642316338268e-06,
+      "loss": 0.721,
+      "step": 142
+    },
+    {
+      "epoch": 2.5087719298245617,
+      "grad_norm": 0.4950014803911829,
+      "learning_rate": 5.464749304646963e-06,
+      "loss": 0.5692,
+      "step": 143
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.4713012868997546,
+      "learning_rate": 5.406793373339292e-06,
+      "loss": 0.6579,
+      "step": 144
+    },
+    {
+      "epoch": 2.543859649122807,
+      "grad_norm": 0.4295753154632374,
+      "learning_rate": 5.348782368720627e-06,
+      "loss": 0.6764,
+      "step": 145
+    },
+    {
+      "epoch": 2.56140350877193,
+      "grad_norm": 0.6088407511386773,
+      "learning_rate": 5.290724144552379e-06,
+      "loss": 0.7354,
+      "step": 146
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.44984667778112336,
+      "learning_rate": 5.232626560988735e-06,
+      "loss": 0.6749,
+      "step": 147
+    },
+    {
+      "epoch": 2.5964912280701755,
+      "grad_norm": 0.44568603289791675,
+      "learning_rate": 5.174497483512506e-06,
+      "loss": 0.693,
+      "step": 148
+    },
+    {
+      "epoch": 2.6140350877192984,
+      "grad_norm": 0.5086878738536396,
+      "learning_rate": 5.116344781870282e-06,
+      "loss": 0.6651,
+      "step": 149
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 1.5360743816680642,
+      "learning_rate": 5.0581763290069865e-06,
+      "loss": 0.7561,
+      "step": 150
+    },
+    {
+      "epoch": 2.6491228070175437,
+      "grad_norm": 0.44075260614863965,
+      "learning_rate": 5e-06,
+      "loss": 0.7387,
+      "step": 151
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 0.5018289937382203,
+      "learning_rate": 4.941823670993016e-06,
+      "loss": 0.7315,
+      "step": 152
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.6753971097001877,
+      "learning_rate": 4.883655218129719e-06,
+      "loss": 0.7251,
+      "step": 153
+    },
+    {
+      "epoch": 2.7017543859649122,
+      "grad_norm": 0.512372179333182,
+      "learning_rate": 4.825502516487497e-06,
+      "loss": 0.7221,
+      "step": 154
+    },
+    {
+      "epoch": 2.719298245614035,
+      "grad_norm": 0.46972995173073046,
+      "learning_rate": 4.767373439011267e-06,
+      "loss": 0.7343,
+      "step": 155
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.4190343839351779,
+      "learning_rate": 4.7092758554476215e-06,
+      "loss": 0.6753,
+      "step": 156
+    },
+    {
+      "epoch": 2.754385964912281,
+      "grad_norm": 0.4513624502323654,
+      "learning_rate": 4.651217631279374e-06,
+      "loss": 0.6913,
+      "step": 157
+    },
+    {
+      "epoch": 2.7719298245614032,
+      "grad_norm": 0.46487454000935413,
+      "learning_rate": 4.59320662666071e-06,
+      "loss": 0.6029,
+      "step": 158
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.43915967996747707,
+      "learning_rate": 4.53525069535304e-06,
+      "loss": 0.6683,
+      "step": 159
+    },
+    {
+      "epoch": 2.807017543859649,
+      "grad_norm": 0.4771514997767183,
+      "learning_rate": 4.477357683661734e-06,
+      "loss": 0.5754,
+      "step": 160
+    },
+    {
+      "epoch": 2.824561403508772,
+      "grad_norm": 0.49051416963609645,
+      "learning_rate": 4.4195354293738484e-06,
+      "loss": 0.6733,
+      "step": 161
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.5533541947029572,
+      "learning_rate": 4.361791760697027e-06,
+      "loss": 0.7034,
+      "step": 162
+    },
+    {
+      "epoch": 2.8596491228070176,
+      "grad_norm": 0.49994436627541317,
+      "learning_rate": 4.304134495199675e-06,
+      "loss": 0.7157,
+      "step": 163
+    },
+    {
+      "epoch": 2.8771929824561404,
+      "grad_norm": 0.4397854458939821,
+      "learning_rate": 4.246571438752585e-06,
+      "loss": 0.6887,
+      "step": 164
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.47809507434717013,
+      "learning_rate": 4.189110384472164e-06,
+      "loss": 0.7504,
+      "step": 165
+    },
+    {
+      "epoch": 2.912280701754386,
+      "grad_norm": 0.39910241714687517,
+      "learning_rate": 4.131759111665349e-06,
+      "loss": 0.6851,
+      "step": 166
+    },
+    {
+      "epoch": 2.9298245614035086,
+      "grad_norm": 0.5335107368208035,
+      "learning_rate": 4.074525384776428e-06,
+      "loss": 0.6261,
+      "step": 167
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.46564637369296086,
+      "learning_rate": 4.017416952335849e-06,
+      "loss": 0.6681,
+      "step": 168
+    },
+    {
+      "epoch": 2.9649122807017543,
+      "grad_norm": 0.4377732908902593,
+      "learning_rate": 3.960441545911205e-06,
+      "loss": 0.6928,
+      "step": 169
+    },
+    {
+      "epoch": 2.982456140350877,
+      "grad_norm": 0.46202011890866834,
+      "learning_rate": 3.903606879060483e-06,
+      "loss": 0.7066,
+      "step": 170
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.5150974337868895,
+      "learning_rate": 3.8469206462878e-06,
+      "loss": 0.5344,
+      "step": 171
+    },
+    {
+      "epoch": 3.017543859649123,
+      "grad_norm": 0.8387697125510133,
+      "learning_rate": 3.790390522001662e-06,
+      "loss": 0.5573,
+      "step": 172
+    },
+    {
+      "epoch": 3.0350877192982457,
+      "grad_norm": 0.48498662697593053,
+      "learning_rate": 3.7340241594759917e-06,
+      "loss": 0.6791,
+      "step": 173
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.7156827789386713,
+      "learning_rate": 3.6778291898139907e-06,
+      "loss": 0.708,
+      "step": 174
+    },
+    {
+      "epoch": 3.0701754385964914,
+      "grad_norm": 0.6772167563154711,
+      "learning_rate": 3.6218132209150047e-06,
+      "loss": 0.6928,
+      "step": 175
+    },
+    {
+      "epoch": 3.087719298245614,
+      "grad_norm": 0.7815727834542585,
+      "learning_rate": 3.5659838364445505e-06,
+      "loss": 0.6183,
+      "step": 176
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.6069404616797929,
+      "learning_rate": 3.51034859480759e-06,
+      "loss": 0.7278,
+      "step": 177
+    },
+    {
+      "epoch": 3.1228070175438596,
+      "grad_norm": 0.5545541428468601,
+      "learning_rate": 3.4549150281252635e-06,
+      "loss": 0.6928,
+      "step": 178
+    },
+    {
+      "epoch": 3.1403508771929824,
+      "grad_norm": 0.5520744872532563,
+      "learning_rate": 3.399690641215142e-06,
+      "loss": 0.596,
+      "step": 179
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.7628842324504479,
+      "learning_rate": 3.3446829105752103e-06,
+      "loss": 0.5981,
+      "step": 180
+    },
+    {
+      "epoch": 3.175438596491228,
+      "grad_norm": 0.5211775941808514,
+      "learning_rate": 3.289899283371657e-06,
+      "loss": 0.6393,
+      "step": 181
+    },
+    {
+      "epoch": 3.192982456140351,
+      "grad_norm": 0.48995495142896,
+      "learning_rate": 3.2353471764306567e-06,
+      "loss": 0.7737,
+      "step": 182
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.5638908248411786,
+      "learning_rate": 3.1810339752342446e-06,
+      "loss": 0.685,
+      "step": 183
+    },
+    {
+      "epoch": 3.2280701754385963,
+      "grad_norm": 0.7903413235219549,
+      "learning_rate": 3.12696703292044e-06,
+      "loss": 0.6232,
+      "step": 184
+    },
+    {
+      "epoch": 3.245614035087719,
+      "grad_norm": 0.4770124134849873,
+      "learning_rate": 3.0731536692877596e-06,
+      "loss": 0.6105,
+      "step": 185
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.4522190380030322,
+      "learning_rate": 3.019601169804216e-06,
+      "loss": 0.5449,
+      "step": 186
+    },
+    {
+      "epoch": 3.280701754385965,
+      "grad_norm": 0.47554535118489655,
+      "learning_rate": 2.966316784621e-06,
+      "loss": 0.6389,
+      "step": 187
+    },
+    {
+      "epoch": 3.2982456140350878,
+      "grad_norm": 0.48723802157353846,
+      "learning_rate": 2.9133077275909112e-06,
+      "loss": 0.6922,
+      "step": 188
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.6087930082371377,
+      "learning_rate": 2.86058117529173e-06,
+      "loss": 0.607,
+      "step": 189
+    },
+    {
+      "epoch": 3.3333333333333335,
+      "grad_norm": 0.5216776908363867,
+      "learning_rate": 2.8081442660546126e-06,
+      "loss": 0.6155,
+      "step": 190
+    },
+    {
+      "epoch": 3.3508771929824563,
+      "grad_norm": 0.5200850074463185,
+      "learning_rate": 2.7560040989976894e-06,
+      "loss": 0.6461,
+      "step": 191
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.4610651901621547,
+      "learning_rate": 2.7041677330649408e-06,
+      "loss": 0.5462,
+      "step": 192
+    },
+    {
+      "epoch": 3.3859649122807016,
+      "grad_norm": 0.40157642158758705,
+      "learning_rate": 2.6526421860705474e-06,
+      "loss": 0.5644,
+      "step": 193
+    },
+    {
+      "epoch": 3.4035087719298245,
+      "grad_norm": 0.4762540380342894,
+      "learning_rate": 2.601434433748771e-06,
+      "loss": 0.5981,
+      "step": 194
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.8962420963971451,
+      "learning_rate": 2.550551408809566e-06,
+      "loss": 0.5239,
+      "step": 195
+    },
+    {
+      "epoch": 3.43859649122807,
+      "grad_norm": 0.4699770415730479,
+      "learning_rate": 2.5000000000000015e-06,
+      "loss": 0.6502,
+      "step": 196
+    },
+    {
+      "epoch": 3.456140350877193,
+      "grad_norm": 0.6775729804455639,
+      "learning_rate": 2.4497870511716237e-06,
+      "loss": 0.4795,
+      "step": 197
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.4664299837506905,
+      "learning_rate": 2.3999193603539234e-06,
+      "loss": 0.5949,
+      "step": 198
+    },
+    {
+      "epoch": 3.4912280701754383,
+      "grad_norm": 0.4683189077817546,
+      "learning_rate": 2.3504036788339763e-06,
+      "loss": 0.6331,
+      "step": 199
+    },
+    {
+      "epoch": 3.5087719298245617,
+      "grad_norm": 0.455012947898177,
+      "learning_rate": 2.3012467102424373e-06,
+      "loss": 0.5819,
+      "step": 200
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.5411207640806618,
+      "learning_rate": 2.2524551096459703e-06,
+      "loss": 0.6417,
+      "step": 201
+    },
+    {
+      "epoch": 3.543859649122807,
+      "grad_norm": 0.5748637247425,
+      "learning_rate": 2.204035482646267e-06,
+      "loss": 0.5131,
+      "step": 202
+    },
+    {
+      "epoch": 3.56140350877193,
+      "grad_norm": 0.44508751154711507,
+      "learning_rate": 2.155994384485742e-06,
+      "loss": 0.7128,
+      "step": 203
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.5471196085215465,
+      "learning_rate": 2.1083383191600676e-06,
+      "loss": 0.5798,
+      "step": 204
+    },
+    {
+      "epoch": 3.5964912280701755,
+      "grad_norm": 0.5709780522197079,
+      "learning_rate": 2.061073738537635e-06,
+      "loss": 0.5938,
+      "step": 205
+    },
+    {
+      "epoch": 3.6140350877192984,
+      "grad_norm": 0.5169192987172408,
+      "learning_rate": 2.0142070414860704e-06,
+      "loss": 0.5251,
+      "step": 206
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.731366983720658,
+      "learning_rate": 1.9677445730059348e-06,
+      "loss": 0.5401,
+      "step": 207
+    },
+    {
+      "epoch": 3.6491228070175437,
+      "grad_norm": 0.6268094134763738,
+      "learning_rate": 1.9216926233717087e-06,
+      "loss": 0.5729,
+      "step": 208
+    },
+    {
+      "epoch": 3.6666666666666665,
+      "grad_norm": 0.5062693136480574,
+      "learning_rate": 1.8760574272802002e-06,
+      "loss": 0.6337,
+      "step": 209
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.4842291341957393,
+      "learning_rate": 1.8308451630064484e-06,
+      "loss": 0.5668,
+      "step": 210
+    },
+    {
+      "epoch": 3.7017543859649122,
+      "grad_norm": 0.5328377038178864,
+      "learning_rate": 1.7860619515673034e-06,
+      "loss": 0.6513,
+      "step": 211
+    },
+    {
+      "epoch": 3.719298245614035,
+      "grad_norm": 0.5355496345065546,
+      "learning_rate": 1.7417138558927244e-06,
+      "loss": 0.6226,
+      "step": 212
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.45704011601378447,
+      "learning_rate": 1.6978068800049624e-06,
+      "loss": 0.6777,
+      "step": 213
+    },
+    {
+      "epoch": 3.754385964912281,
+      "grad_norm": 0.44210078265241176,
+      "learning_rate": 1.6543469682057105e-06,
+      "loss": 0.6827,
+      "step": 214
+    },
+    {
+      "epoch": 3.7719298245614032,
+      "grad_norm": 0.45433914058704983,
+      "learning_rate": 1.611340004271339e-06,
+      "loss": 0.7184,
+      "step": 215
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.4909566653829169,
+      "learning_rate": 1.5687918106563326e-06,
+      "loss": 0.6198,
+      "step": 216
+    },
+    {
+      "epoch": 3.807017543859649,
+      "grad_norm": 0.49694357245411236,
+      "learning_rate": 1.5267081477050132e-06,
+      "loss": 0.6638,
+      "step": 217
+    },
+    {
+      "epoch": 3.824561403508772,
+      "grad_norm": 0.4902393217229584,
+      "learning_rate": 1.4850947128716914e-06,
+      "loss": 0.6818,
+      "step": 218
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.49626411004393584,
+      "learning_rate": 1.4439571399493146e-06,
+      "loss": 0.6449,
+      "step": 219
+    },
+    {
+      "epoch": 3.8596491228070176,
+      "grad_norm": 0.45655426220241163,
+      "learning_rate": 1.4033009983067454e-06,
+      "loss": 0.5492,
+      "step": 220
+    },
+    {
+      "epoch": 3.8771929824561404,
+      "grad_norm": 0.4399760432036042,
+      "learning_rate": 1.3631317921347564e-06,
+      "loss": 0.6375,
+      "step": 221
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.5439997575999478,
+      "learning_rate": 1.3234549597008572e-06,
+      "loss": 0.6291,
+      "step": 222
+    },
+    {
+      "epoch": 3.912280701754386,
+      "grad_norm": 0.41449228900459967,
+      "learning_rate": 1.2842758726130283e-06,
+      "loss": 0.5616,
+      "step": 223
+    },
+    {
+      "epoch": 3.9298245614035086,
+      "grad_norm": 0.5927173476590777,
+      "learning_rate": 1.2455998350925042e-06,
+      "loss": 0.5756,
+      "step": 224
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.5309053645487309,
+      "learning_rate": 1.2074320832556558e-06,
+      "loss": 0.6774,
+      "step": 225
+    },
+    {
+      "epoch": 3.9649122807017543,
+      "grad_norm": 0.6803116781756926,
+      "learning_rate": 1.1697777844051105e-06,
+      "loss": 0.65,
+      "step": 226
+    },
+    {
+      "epoch": 3.982456140350877,
+      "grad_norm": 0.5497151811664531,
+      "learning_rate": 1.132642036330181e-06,
+      "loss": 0.673,
+      "step": 227
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.521937596996345,
+      "learning_rate": 1.096029866616704e-06,
+      "loss": 0.5678,
+      "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 sha256:297d9b63058dba7a150b243742f94a93914859f46dfa5e504fc6f028978c0739
+size 8145