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": false,
+  "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:aaf7a1dcc0f5ad0b0e4e4574aced916ec29c5d8e34e206958aef0af5138a65fb
+size 4997184968

model-00002-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f1dda93c713cf46335efd5b7d1c306f4263636d3b9b7b1469a03ee42e521423b
+size 4997741608

model-00003-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:18dd07a0a01d1a5f8c04f32285d0a6f4af1e8ee6cd1d580f0fcaab453b4ab991
+size 4997742208

model-00004-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e69a412d175a4109ca639df5b2b78f70b63f5f644893751c6c53009471827ade
+size 4997743184

model-00005-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fb8a58cddc4c832ec9f918ce0fcda42209b3700b37c5ad9d9e21705c9438e55c
+size 4997743184

model-00006-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:baa9bd912e44f361be00dd79cac43178e2021c47fc2bb0cd742f4f0f9fb2b096
+size 4997743184

model-00007-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5a9b493fa256a7b3eec1a5704d9192a0fa698299bf5b1d71904cb9efbf9d03a3
+size 4997743184

model-00008-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b3e96e1611d6a8464d2be97638c83ad89f14fe90e99a4c42da63c0b4f635d016
+size 4997743184

model-00009-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e2a55b149f7fdf51c21053ae4c305c20834d10140371675f903543c3bb287f31
+size 4997743184

model-00010-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d1c2a601ab0817130903452feb19969a7fd48e471583baf1ba6241111d245600
+size 4997743184

model-00011-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:33c7dd376af7abae6ff8b28a4333b1f1cfd67be65f691a49e33988afeaa11e27
+size 4997743184

model-00012-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e64e570e097754c1152c888e47ed0185309883469701f2307e20158fed6f9350
+size 4997743184

model-00013-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6063609e14ae23fe5d24f82aca980cc87086171f5bd1ec9ae2cea61c0dc50b5b
+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": 63.15584646766748,
+      "learning_rate": 0.0,
+      "loss": 1.3331,
+      "step": 1
+    },
+    {
+      "epoch": 0.03508771929824561,
+      "grad_norm": 65.37999240061016,
+      "learning_rate": 6.666666666666667e-07,
+      "loss": 1.3643,
+      "step": 2
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 58.76204353998122,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 1.5736,
+      "step": 3
+    },
+    {
+      "epoch": 0.07017543859649122,
+      "grad_norm": 94.22910235879955,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.6305,
+      "step": 4
+    },
+    {
+      "epoch": 0.08771929824561403,
+      "grad_norm": 63.240336626135914,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 1.2903,
+      "step": 5
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 62.34285462175898,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.4876,
+      "step": 6
+    },
+    {
+      "epoch": 0.12280701754385964,
+      "grad_norm": 38.87129959087156,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.3969,
+      "step": 7
+    },
+    {
+      "epoch": 0.14035087719298245,
+      "grad_norm": 10.342991864065548,
+      "learning_rate": 4.666666666666667e-06,
+      "loss": 1.3244,
+      "step": 8
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 5.942719289647008,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 1.2312,
+      "step": 9
+    },
+    {
+      "epoch": 0.17543859649122806,
+      "grad_norm": 5.474776769364296,
+      "learning_rate": 6e-06,
+      "loss": 1.2549,
+      "step": 10
+    },
+    {
+      "epoch": 0.19298245614035087,
+      "grad_norm": 4.282499929517234,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.2712,
+      "step": 11
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 2.957156365671227,
+      "learning_rate": 7.333333333333333e-06,
+      "loss": 1.0496,
+      "step": 12
+    },
+    {
+      "epoch": 0.22807017543859648,
+      "grad_norm": 2.1951967154579703,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.0609,
+      "step": 13
+    },
+    {
+      "epoch": 0.24561403508771928,
+      "grad_norm": 1.8998626223389208,
+      "learning_rate": 8.666666666666668e-06,
+      "loss": 1.0354,
+      "step": 14
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 1.5287799247220097,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 1.0815,
+      "step": 15
+    },
+    {
+      "epoch": 0.2807017543859649,
+      "grad_norm": 1.3903798573645059,
+      "learning_rate": 1e-05,
+      "loss": 1.0065,
+      "step": 16
+    },
+    {
+      "epoch": 0.2982456140350877,
+      "grad_norm": 1.6539944664702928,
+      "learning_rate": 9.999661540018812e-06,
+      "loss": 1.1115,
+      "step": 17
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 1.227494438263455,
+      "learning_rate": 9.99864620589731e-06,
+      "loss": 0.9712,
+      "step": 18
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 1.0886829004299536,
+      "learning_rate": 9.99695413509548e-06,
+      "loss": 0.9896,
+      "step": 19
+    },
+    {
+      "epoch": 0.3508771929824561,
+      "grad_norm": 0.9462016469544118,
+      "learning_rate": 9.994585556692624e-06,
+      "loss": 1.0194,
+      "step": 20
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 0.8343449744183404,
+      "learning_rate": 9.991540791356342e-06,
+      "loss": 0.9435,
+      "step": 21
+    },
+    {
+      "epoch": 0.38596491228070173,
+      "grad_norm": 0.7938863164587222,
+      "learning_rate": 9.987820251299121e-06,
+      "loss": 0.8057,
+      "step": 22
+    },
+    {
+      "epoch": 0.40350877192982454,
+      "grad_norm": 0.8633971461176382,
+      "learning_rate": 9.98342444022253e-06,
+      "loss": 0.9424,
+      "step": 23
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 0.8092584043672173,
+      "learning_rate": 9.978353953249023e-06,
+      "loss": 0.8359,
+      "step": 24
+    },
+    {
+      "epoch": 0.43859649122807015,
+      "grad_norm": 0.7342380217574445,
+      "learning_rate": 9.972609476841368e-06,
+      "loss": 0.9682,
+      "step": 25
+    },
+    {
+      "epoch": 0.45614035087719296,
+      "grad_norm": 0.7978382483785499,
+      "learning_rate": 9.966191788709716e-06,
+      "loss": 0.9308,
+      "step": 26
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 0.7523338860744757,
+      "learning_rate": 9.959101757706308e-06,
+      "loss": 0.841,
+      "step": 27
+    },
+    {
+      "epoch": 0.49122807017543857,
+      "grad_norm": 0.7100125850436978,
+      "learning_rate": 9.951340343707852e-06,
+      "loss": 0.9126,
+      "step": 28
+    },
+    {
+      "epoch": 0.5087719298245614,
+      "grad_norm": 0.6432035176703957,
+      "learning_rate": 9.942908597485558e-06,
+      "loss": 0.93,
+      "step": 29
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 0.5575650398795282,
+      "learning_rate": 9.933807660562898e-06,
+      "loss": 0.8113,
+      "step": 30
+    },
+    {
+      "epoch": 0.543859649122807,
+      "grad_norm": 0.5538633586514855,
+      "learning_rate": 9.924038765061042e-06,
+      "loss": 0.8689,
+      "step": 31
+    },
+    {
+      "epoch": 0.5614035087719298,
+      "grad_norm": 0.5540704105571124,
+      "learning_rate": 9.913603233532067e-06,
+      "loss": 0.7981,
+      "step": 32
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 0.6864556027313015,
+      "learning_rate": 9.902502478779897e-06,
+      "loss": 0.8981,
+      "step": 33
+    },
+    {
+      "epoch": 0.5964912280701754,
+      "grad_norm": 0.6024572093985017,
+      "learning_rate": 9.890738003669029e-06,
+      "loss": 0.9963,
+      "step": 34
+    },
+    {
+      "epoch": 0.6140350877192983,
+      "grad_norm": 0.5535510234500352,
+      "learning_rate": 9.878311400921072e-06,
+      "loss": 0.8349,
+      "step": 35
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 0.5300668292678892,
+      "learning_rate": 9.86522435289912e-06,
+      "loss": 0.8523,
+      "step": 36
+    },
+    {
+      "epoch": 0.6491228070175439,
+      "grad_norm": 0.5886120290128514,
+      "learning_rate": 9.851478631379982e-06,
+      "loss": 0.889,
+      "step": 37
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 0.5688139181593664,
+      "learning_rate": 9.83707609731432e-06,
+      "loss": 0.8988,
+      "step": 38
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.6119281790052227,
+      "learning_rate": 9.822018700574696e-06,
+      "loss": 0.9201,
+      "step": 39
+    },
+    {
+      "epoch": 0.7017543859649122,
+      "grad_norm": 0.5162998849300403,
+      "learning_rate": 9.806308479691595e-06,
+      "loss": 0.8834,
+      "step": 40
+    },
+    {
+      "epoch": 0.7192982456140351,
+      "grad_norm": 0.5471391274601726,
+      "learning_rate": 9.789947561577445e-06,
+      "loss": 0.8902,
+      "step": 41
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.5058167109969809,
+      "learning_rate": 9.77293816123866e-06,
+      "loss": 0.8224,
+      "step": 42
+    },
+    {
+      "epoch": 0.7543859649122807,
+      "grad_norm": 0.5302134290223017,
+      "learning_rate": 9.755282581475769e-06,
+      "loss": 0.855,
+      "step": 43
+    },
+    {
+      "epoch": 0.7719298245614035,
+      "grad_norm": 0.4881324513673794,
+      "learning_rate": 9.736983212571646e-06,
+      "loss": 0.7701,
+      "step": 44
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.4957819394220034,
+      "learning_rate": 9.718042531967918e-06,
+      "loss": 0.8879,
+      "step": 45
+    },
+    {
+      "epoch": 0.8070175438596491,
+      "grad_norm": 0.6010796184130253,
+      "learning_rate": 9.698463103929542e-06,
+      "loss": 0.8622,
+      "step": 46
+    },
+    {
+      "epoch": 0.8245614035087719,
+      "grad_norm": 0.5546469825759003,
+      "learning_rate": 9.678247579197658e-06,
+      "loss": 0.8712,
+      "step": 47
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.47834181259316355,
+      "learning_rate": 9.657398694630713e-06,
+      "loss": 0.8512,
+      "step": 48
+    },
+    {
+      "epoch": 0.8596491228070176,
+      "grad_norm": 0.5216399443120093,
+      "learning_rate": 9.635919272833938e-06,
+      "loss": 0.9462,
+      "step": 49
+    },
+    {
+      "epoch": 0.8771929824561403,
+      "grad_norm": 0.48944474987036246,
+      "learning_rate": 9.613812221777212e-06,
+      "loss": 0.6809,
+      "step": 50
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 0.5067793225025444,
+      "learning_rate": 9.591080534401371e-06,
+      "loss": 0.9127,
+      "step": 51
+    },
+    {
+      "epoch": 0.9122807017543859,
+      "grad_norm": 0.5700994240773224,
+      "learning_rate": 9.567727288213005e-06,
+      "loss": 0.8831,
+      "step": 52
+    },
+    {
+      "epoch": 0.9298245614035088,
+      "grad_norm": 0.4587934065575008,
+      "learning_rate": 9.543755644867823e-06,
+      "loss": 0.833,
+      "step": 53
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.550141469856127,
+      "learning_rate": 9.519168849742603e-06,
+      "loss": 0.7839,
+      "step": 54
+    },
+    {
+      "epoch": 0.9649122807017544,
+      "grad_norm": 0.54146367658382,
+      "learning_rate": 9.493970231495836e-06,
+      "loss": 0.8254,
+      "step": 55
+    },
+    {
+      "epoch": 0.9824561403508771,
+      "grad_norm": 0.46888157273309244,
+      "learning_rate": 9.468163201617063e-06,
+      "loss": 0.8368,
+      "step": 56
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5244954404043076,
+      "learning_rate": 9.441751253965022e-06,
+      "loss": 0.7685,
+      "step": 57
+    },
+    {
+      "epoch": 1.0175438596491229,
+      "grad_norm": 0.5715240469472487,
+      "learning_rate": 9.414737964294636e-06,
+      "loss": 0.8686,
+      "step": 58
+    },
+    {
+      "epoch": 1.0350877192982457,
+      "grad_norm": 0.5839176316188016,
+      "learning_rate": 9.38712698977291e-06,
+      "loss": 0.753,
+      "step": 59
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.5408800036441663,
+      "learning_rate": 9.358922068483813e-06,
+      "loss": 0.8449,
+      "step": 60
+    },
+    {
+      "epoch": 1.0701754385964912,
+      "grad_norm": 0.5223729459379649,
+      "learning_rate": 9.330127018922195e-06,
+      "loss": 0.7153,
+      "step": 61
+    },
+    {
+      "epoch": 1.087719298245614,
+      "grad_norm": 0.5319407489367689,
+      "learning_rate": 9.30074573947683e-06,
+      "loss": 0.7918,
+      "step": 62
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 0.7512029629847802,
+      "learning_rate": 9.27078220790263e-06,
+      "loss": 0.8333,
+      "step": 63
+    },
+    {
+      "epoch": 1.1228070175438596,
+      "grad_norm": 0.6304754836687595,
+      "learning_rate": 9.24024048078213e-06,
+      "loss": 0.8217,
+      "step": 64
+    },
+    {
+      "epoch": 1.1403508771929824,
+      "grad_norm": 0.5144643188045851,
+      "learning_rate": 9.209124692976287e-06,
+      "loss": 0.7389,
+      "step": 65
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.6878682898184404,
+      "learning_rate": 9.177439057064684e-06,
+      "loss": 0.7412,
+      "step": 66
+    },
+    {
+      "epoch": 1.1754385964912282,
+      "grad_norm": 0.5638913441884815,
+      "learning_rate": 9.145187862775208e-06,
+      "loss": 0.7467,
+      "step": 67
+    },
+    {
+      "epoch": 1.1929824561403508,
+      "grad_norm": 0.5103894720788118,
+      "learning_rate": 9.112375476403313e-06,
+      "loss": 0.7824,
+      "step": 68
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.8340978298682767,
+      "learning_rate": 9.079006340220862e-06,
+      "loss": 0.8395,
+      "step": 69
+    },
+    {
+      "epoch": 1.2280701754385965,
+      "grad_norm": 0.5332830565878104,
+      "learning_rate": 9.045084971874738e-06,
+      "loss": 0.7808,
+      "step": 70
+    },
+    {
+      "epoch": 1.2456140350877192,
+      "grad_norm": 0.4929268516063678,
+      "learning_rate": 9.01061596377522e-06,
+      "loss": 0.8323,
+      "step": 71
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 1.1376889717363314,
+      "learning_rate": 8.97560398247424e-06,
+      "loss": 0.8044,
+      "step": 72
+    },
+    {
+      "epoch": 1.280701754385965,
+      "grad_norm": 0.5300338359407787,
+      "learning_rate": 8.94005376803361e-06,
+      "loss": 0.7985,
+      "step": 73
+    },
+    {
+      "epoch": 1.2982456140350878,
+      "grad_norm": 0.4845039519748431,
+      "learning_rate": 8.903970133383297e-06,
+      "loss": 0.7169,
+      "step": 74
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.6016486595486619,
+      "learning_rate": 8.867357963669821e-06,
+      "loss": 0.8491,
+      "step": 75
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 0.47683104254439873,
+      "learning_rate": 8.83022221559489e-06,
+      "loss": 0.7325,
+      "step": 76
+    },
+    {
+      "epoch": 1.3508771929824561,
+      "grad_norm": 0.5013506543532272,
+      "learning_rate": 8.792567916744346e-06,
+      "loss": 0.7095,
+      "step": 77
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.6777934168520869,
+      "learning_rate": 8.754400164907496e-06,
+      "loss": 0.8317,
+      "step": 78
+    },
+    {
+      "epoch": 1.3859649122807016,
+      "grad_norm": 0.5891901349953744,
+      "learning_rate": 8.715724127386971e-06,
+      "loss": 0.8131,
+      "step": 79
+    },
+    {
+      "epoch": 1.4035087719298245,
+      "grad_norm": 0.4873488027466944,
+      "learning_rate": 8.676545040299145e-06,
+      "loss": 0.7663,
+      "step": 80
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 0.49134574827754873,
+      "learning_rate": 8.636868207865244e-06,
+      "loss": 0.7946,
+      "step": 81
+    },
+    {
+      "epoch": 1.4385964912280702,
+      "grad_norm": 0.5435760861018574,
+      "learning_rate": 8.596699001693257e-06,
+      "loss": 0.7656,
+      "step": 82
+    },
+    {
+      "epoch": 1.456140350877193,
+      "grad_norm": 0.5136619978494325,
+      "learning_rate": 8.556042860050686e-06,
+      "loss": 0.7075,
+      "step": 83
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.4804522550868662,
+      "learning_rate": 8.51490528712831e-06,
+      "loss": 0.6317,
+      "step": 84
+    },
+    {
+      "epoch": 1.4912280701754386,
+      "grad_norm": 0.6008641247045057,
+      "learning_rate": 8.473291852294986e-06,
+      "loss": 0.7932,
+      "step": 85
+    },
+    {
+      "epoch": 1.5087719298245614,
+      "grad_norm": 0.5279922446205254,
+      "learning_rate": 8.43120818934367e-06,
+      "loss": 0.7702,
+      "step": 86
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.5150719148635752,
+      "learning_rate": 8.388659995728662e-06,
+      "loss": 0.9113,
+      "step": 87
+    },
+    {
+      "epoch": 1.543859649122807,
+      "grad_norm": 0.5789890170019726,
+      "learning_rate": 8.345653031794292e-06,
+      "loss": 0.7873,
+      "step": 88
+    },
+    {
+      "epoch": 1.5614035087719298,
+      "grad_norm": 0.46610784951869666,
+      "learning_rate": 8.302193119995038e-06,
+      "loss": 0.8267,
+      "step": 89
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.4750765112665647,
+      "learning_rate": 8.258286144107277e-06,
+      "loss": 0.7293,
+      "step": 90
+    },
+    {
+      "epoch": 1.5964912280701755,
+      "grad_norm": 0.532780121125905,
+      "learning_rate": 8.213938048432697e-06,
+      "loss": 0.7705,
+      "step": 91
+    },
+    {
+      "epoch": 1.6140350877192984,
+      "grad_norm": 0.47147285753477935,
+      "learning_rate": 8.16915483699355e-06,
+      "loss": 0.7719,
+      "step": 92
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.43093450285496204,
+      "learning_rate": 8.123942572719801e-06,
+      "loss": 0.7191,
+      "step": 93
+    },
+    {
+      "epoch": 1.6491228070175439,
+      "grad_norm": 0.4202810357942653,
+      "learning_rate": 8.078307376628292e-06,
+      "loss": 0.7182,
+      "step": 94
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 0.47158236528019215,
+      "learning_rate": 8.032255426994069e-06,
+      "loss": 0.8182,
+      "step": 95
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.47926786500731233,
+      "learning_rate": 7.985792958513932e-06,
+      "loss": 0.7254,
+      "step": 96
+    },
+    {
+      "epoch": 1.7017543859649122,
+      "grad_norm": 0.5563809037423963,
+      "learning_rate": 7.938926261462366e-06,
+      "loss": 0.9024,
+      "step": 97
+    },
+    {
+      "epoch": 1.719298245614035,
+      "grad_norm": 0.44641435414441344,
+      "learning_rate": 7.891661680839932e-06,
+      "loss": 0.7725,
+      "step": 98
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.5040938263823853,
+      "learning_rate": 7.84400561551426e-06,
+      "loss": 0.8062,
+      "step": 99
+    },
+    {
+      "epoch": 1.7543859649122808,
+      "grad_norm": 0.4856518637743601,
+      "learning_rate": 7.795964517353734e-06,
+      "loss": 0.7534,
+      "step": 100
+    },
+    {
+      "epoch": 1.7719298245614035,
+      "grad_norm": 0.5769975852409818,
+      "learning_rate": 7.747544890354031e-06,
+      "loss": 0.8282,
+      "step": 101
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.4188630617545806,
+      "learning_rate": 7.698753289757565e-06,
+      "loss": 0.8134,
+      "step": 102
+    },
+    {
+      "epoch": 1.807017543859649,
+      "grad_norm": 0.5711445074593904,
+      "learning_rate": 7.649596321166024e-06,
+      "loss": 0.853,
+      "step": 103
+    },
+    {
+      "epoch": 1.8245614035087718,
+      "grad_norm": 0.5352181452255025,
+      "learning_rate": 7.600080639646077e-06,
+      "loss": 0.6861,
+      "step": 104
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.45506688804740625,
+      "learning_rate": 7.550212948828377e-06,
+      "loss": 0.6558,
+      "step": 105
+    },
+    {
+      "epoch": 1.8596491228070176,
+      "grad_norm": 0.8043046467422099,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.8641,
+      "step": 106
+    },
+    {
+      "epoch": 1.8771929824561404,
+      "grad_norm": 0.5537560542043134,
+      "learning_rate": 7.449448591190436e-06,
+      "loss": 0.7842,
+      "step": 107
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.5781806163988695,
+      "learning_rate": 7.398565566251232e-06,
+      "loss": 0.8085,
+      "step": 108
+    },
+    {
+      "epoch": 1.912280701754386,
+      "grad_norm": 0.47955611331508013,
+      "learning_rate": 7.347357813929455e-06,
+      "loss": 0.7131,
+      "step": 109
+    },
+    {
+      "epoch": 1.9298245614035088,
+      "grad_norm": 0.5804297378962103,
+      "learning_rate": 7.295832266935059e-06,
+      "loss": 0.8253,
+      "step": 110
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.47267047576613697,
+      "learning_rate": 7.243995901002312e-06,
+      "loss": 0.8292,
+      "step": 111
+    },
+    {
+      "epoch": 1.9649122807017543,
+      "grad_norm": 0.5372078338450347,
+      "learning_rate": 7.191855733945388e-06,
+      "loss": 0.7958,
+      "step": 112
+    },
+    {
+      "epoch": 1.9824561403508771,
+      "grad_norm": 0.5243454602483745,
+      "learning_rate": 7.1394188247082715e-06,
+      "loss": 0.8748,
+      "step": 113
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.6795293196463823,
+      "learning_rate": 7.08669227240909e-06,
+      "loss": 0.7771,
+      "step": 114
+    },
+    {
+      "epoch": 2.017543859649123,
+      "grad_norm": 0.7664553169146359,
+      "learning_rate": 7.033683215379002e-06,
+      "loss": 0.6468,
+      "step": 115
+    },
+    {
+      "epoch": 2.0350877192982457,
+      "grad_norm": 0.44409475504057067,
+      "learning_rate": 6.980398830195785e-06,
+      "loss": 0.6895,
+      "step": 116
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.539799520580517,
+      "learning_rate": 6.9268463307122425e-06,
+      "loss": 0.8799,
+      "step": 117
+    },
+    {
+      "epoch": 2.0701754385964914,
+      "grad_norm": 1.0598183919066515,
+      "learning_rate": 6.873032967079562e-06,
+      "loss": 0.742,
+      "step": 118
+    },
+    {
+      "epoch": 2.087719298245614,
+      "grad_norm": 0.7108752118882978,
+      "learning_rate": 6.818966024765758e-06,
+      "loss": 0.7121,
+      "step": 119
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.5844290762112625,
+      "learning_rate": 6.7646528235693445e-06,
+      "loss": 0.6895,
+      "step": 120
+    },
+    {
+      "epoch": 2.1228070175438596,
+      "grad_norm": 0.6184978048452946,
+      "learning_rate": 6.710100716628345e-06,
+      "loss": 0.6837,
+      "step": 121
+    },
+    {
+      "epoch": 2.1403508771929824,
+      "grad_norm": 0.7683043042585884,
+      "learning_rate": 6.655317089424791e-06,
+      "loss": 0.6333,
+      "step": 122
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.6415739416293404,
+      "learning_rate": 6.600309358784858e-06,
+      "loss": 0.7935,
+      "step": 123
+    },
+    {
+      "epoch": 2.175438596491228,
+      "grad_norm": 0.5940773413366346,
+      "learning_rate": 6.545084971874738e-06,
+      "loss": 0.6889,
+      "step": 124
+    },
+    {
+      "epoch": 2.192982456140351,
+      "grad_norm": 0.6615441876651051,
+      "learning_rate": 6.48965140519241e-06,
+      "loss": 0.7954,
+      "step": 125
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 1.2467698683548056,
+      "learning_rate": 6.434016163555452e-06,
+      "loss": 0.697,
+      "step": 126
+    },
+    {
+      "epoch": 2.2280701754385963,
+      "grad_norm": 0.6176668182244752,
+      "learning_rate": 6.378186779084996e-06,
+      "loss": 0.6529,
+      "step": 127
+    },
+    {
+      "epoch": 2.245614035087719,
+      "grad_norm": 0.6210320340351719,
+      "learning_rate": 6.322170810186013e-06,
+      "loss": 0.659,
+      "step": 128
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.5359174259526259,
+      "learning_rate": 6.26597584052401e-06,
+      "loss": 0.7157,
+      "step": 129
+    },
+    {
+      "epoch": 2.280701754385965,
+      "grad_norm": 0.4417287658476391,
+      "learning_rate": 6.209609477998339e-06,
+      "loss": 0.7255,
+      "step": 130
+    },
+    {
+      "epoch": 2.2982456140350878,
+      "grad_norm": 0.7555828952688477,
+      "learning_rate": 6.153079353712201e-06,
+      "loss": 0.7036,
+      "step": 131
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.5226529983835565,
+      "learning_rate": 6.0963931209395165e-06,
+      "loss": 0.6612,
+      "step": 132
+    },
+    {
+      "epoch": 2.3333333333333335,
+      "grad_norm": 0.6603488308666825,
+      "learning_rate": 6.039558454088796e-06,
+      "loss": 0.7656,
+      "step": 133
+    },
+    {
+      "epoch": 2.3508771929824563,
+      "grad_norm": 0.557865400608557,
+      "learning_rate": 5.982583047664151e-06,
+      "loss": 0.6708,
+      "step": 134
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.45812899843964155,
+      "learning_rate": 5.925474615223573e-06,
+      "loss": 0.7062,
+      "step": 135
+    },
+    {
+      "epoch": 2.3859649122807016,
+      "grad_norm": 0.6174931154694858,
+      "learning_rate": 5.8682408883346535e-06,
+      "loss": 0.6836,
+      "step": 136
+    },
+    {
+      "epoch": 2.4035087719298245,
+      "grad_norm": 0.6633009664798852,
+      "learning_rate": 5.810889615527839e-06,
+      "loss": 0.813,
+      "step": 137
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.7017215850356698,
+      "learning_rate": 5.753428561247416e-06,
+      "loss": 0.7838,
+      "step": 138
+    },
+    {
+      "epoch": 2.43859649122807,
+      "grad_norm": 0.5218400680548609,
+      "learning_rate": 5.695865504800328e-06,
+      "loss": 0.7441,
+      "step": 139
+    },
+    {
+      "epoch": 2.456140350877193,
+      "grad_norm": 0.5532971919778912,
+      "learning_rate": 5.638208239302975e-06,
+      "loss": 0.6975,
+      "step": 140
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.5300461531388235,
+      "learning_rate": 5.5804645706261515e-06,
+      "loss": 0.7416,
+      "step": 141
+    },
+    {
+      "epoch": 2.4912280701754383,
+      "grad_norm": 0.4889112138846317,
+      "learning_rate": 5.522642316338268e-06,
+      "loss": 0.7461,
+      "step": 142
+    },
+    {
+      "epoch": 2.5087719298245617,
+      "grad_norm": 0.9300278760417229,
+      "learning_rate": 5.464749304646963e-06,
+      "loss": 0.5929,
+      "step": 143
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.5675375162238812,
+      "learning_rate": 5.406793373339292e-06,
+      "loss": 0.6783,
+      "step": 144
+    },
+    {
+      "epoch": 2.543859649122807,
+      "grad_norm": 0.48860421723000824,
+      "learning_rate": 5.348782368720627e-06,
+      "loss": 0.6913,
+      "step": 145
+    },
+    {
+      "epoch": 2.56140350877193,
+      "grad_norm": 0.8042262529119318,
+      "learning_rate": 5.290724144552379e-06,
+      "loss": 0.7545,
+      "step": 146
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.6159428981497649,
+      "learning_rate": 5.232626560988735e-06,
+      "loss": 0.6977,
+      "step": 147
+    },
+    {
+      "epoch": 2.5964912280701755,
+      "grad_norm": 0.48056693904137376,
+      "learning_rate": 5.174497483512506e-06,
+      "loss": 0.7129,
+      "step": 148
+    },
+    {
+      "epoch": 2.6140350877192984,
+      "grad_norm": 0.5105363726314144,
+      "learning_rate": 5.116344781870282e-06,
+      "loss": 0.6914,
+      "step": 149
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.7599960328923713,
+      "learning_rate": 5.0581763290069865e-06,
+      "loss": 0.7797,
+      "step": 150
+    },
+    {
+      "epoch": 2.6491228070175437,
+      "grad_norm": 0.5403843680422834,
+      "learning_rate": 5e-06,
+      "loss": 0.7456,
+      "step": 151
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 0.5283139298044032,
+      "learning_rate": 4.941823670993016e-06,
+      "loss": 0.7563,
+      "step": 152
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.4757438822907835,
+      "learning_rate": 4.883655218129719e-06,
+      "loss": 0.7414,
+      "step": 153
+    },
+    {
+      "epoch": 2.7017543859649122,
+      "grad_norm": 0.4649046708706869,
+      "learning_rate": 4.825502516487497e-06,
+      "loss": 0.7404,
+      "step": 154
+    },
+    {
+      "epoch": 2.719298245614035,
+      "grad_norm": 0.5592085409232698,
+      "learning_rate": 4.767373439011267e-06,
+      "loss": 0.7587,
+      "step": 155
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.49429843648544647,
+      "learning_rate": 4.7092758554476215e-06,
+      "loss": 0.6999,
+      "step": 156
+    },
+    {
+      "epoch": 2.754385964912281,
+      "grad_norm": 0.4666906553981525,
+      "learning_rate": 4.651217631279374e-06,
+      "loss": 0.7131,
+      "step": 157
+    },
+    {
+      "epoch": 2.7719298245614032,
+      "grad_norm": 0.4764920863502334,
+      "learning_rate": 4.59320662666071e-06,
+      "loss": 0.6217,
+      "step": 158
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.47665196182907754,
+      "learning_rate": 4.53525069535304e-06,
+      "loss": 0.682,
+      "step": 159
+    },
+    {
+      "epoch": 2.807017543859649,
+      "grad_norm": 0.5051602390399056,
+      "learning_rate": 4.477357683661734e-06,
+      "loss": 0.5861,
+      "step": 160
+    },
+    {
+      "epoch": 2.824561403508772,
+      "grad_norm": 0.5221299500292211,
+      "learning_rate": 4.4195354293738484e-06,
+      "loss": 0.7017,
+      "step": 161
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.5350502642098378,
+      "learning_rate": 4.361791760697027e-06,
+      "loss": 0.7304,
+      "step": 162
+    },
+    {
+      "epoch": 2.8596491228070176,
+      "grad_norm": 0.4892026194771734,
+      "learning_rate": 4.304134495199675e-06,
+      "loss": 0.7408,
+      "step": 163
+    },
+    {
+      "epoch": 2.8771929824561404,
+      "grad_norm": 0.44814147207516225,
+      "learning_rate": 4.246571438752585e-06,
+      "loss": 0.7064,
+      "step": 164
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.5711782493148895,
+      "learning_rate": 4.189110384472164e-06,
+      "loss": 0.7778,
+      "step": 165
+    },
+    {
+      "epoch": 2.912280701754386,
+      "grad_norm": 0.4161252035605757,
+      "learning_rate": 4.131759111665349e-06,
+      "loss": 0.7146,
+      "step": 166
+    },
+    {
+      "epoch": 2.9298245614035086,
+      "grad_norm": 0.5159557005074171,
+      "learning_rate": 4.074525384776428e-06,
+      "loss": 0.6429,
+      "step": 167
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.4831174686326969,
+      "learning_rate": 4.017416952335849e-06,
+      "loss": 0.6891,
+      "step": 168
+    },
+    {
+      "epoch": 2.9649122807017543,
+      "grad_norm": 0.4326754209969522,
+      "learning_rate": 3.960441545911205e-06,
+      "loss": 0.708,
+      "step": 169
+    },
+    {
+      "epoch": 2.982456140350877,
+      "grad_norm": 0.4561948514895945,
+      "learning_rate": 3.903606879060483e-06,
+      "loss": 0.7233,
+      "step": 170
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.5487062441627052,
+      "learning_rate": 3.8469206462878e-06,
+      "loss": 0.5455,
+      "step": 171
+    },
+    {
+      "epoch": 3.017543859649123,
+      "grad_norm": 0.7490225933534763,
+      "learning_rate": 3.790390522001662e-06,
+      "loss": 0.5827,
+      "step": 172
+    },
+    {
+      "epoch": 3.0350877192982457,
+      "grad_norm": 0.44497647466561857,
+      "learning_rate": 3.7340241594759917e-06,
+      "loss": 0.6996,
+      "step": 173
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.5857538987132415,
+      "learning_rate": 3.6778291898139907e-06,
+      "loss": 0.7281,
+      "step": 174
+    },
+    {
+      "epoch": 3.0701754385964914,
+      "grad_norm": 0.6112073048306884,
+      "learning_rate": 3.6218132209150047e-06,
+      "loss": 0.7177,
+      "step": 175
+    },
+    {
+      "epoch": 3.087719298245614,
+      "grad_norm": 0.5814034947816205,
+      "learning_rate": 3.5659838364445505e-06,
+      "loss": 0.6406,
+      "step": 176
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.7563856057732478,
+      "learning_rate": 3.51034859480759e-06,
+      "loss": 0.7529,
+      "step": 177
+    },
+    {
+      "epoch": 3.1228070175438596,
+      "grad_norm": 0.6639137347154231,
+      "learning_rate": 3.4549150281252635e-06,
+      "loss": 0.7206,
+      "step": 178
+    },
+    {
+      "epoch": 3.1403508771929824,
+      "grad_norm": 0.4826176774807268,
+      "learning_rate": 3.399690641215142e-06,
+      "loss": 0.6114,
+      "step": 179
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.5420392340089536,
+      "learning_rate": 3.3446829105752103e-06,
+      "loss": 0.6233,
+      "step": 180
+    },
+    {
+      "epoch": 3.175438596491228,
+      "grad_norm": 0.4568079326628623,
+      "learning_rate": 3.289899283371657e-06,
+      "loss": 0.6587,
+      "step": 181
+    },
+    {
+      "epoch": 3.192982456140351,
+      "grad_norm": 0.7625415796991232,
+      "learning_rate": 3.2353471764306567e-06,
+      "loss": 0.7964,
+      "step": 182
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.5844341400388982,
+      "learning_rate": 3.1810339752342446e-06,
+      "loss": 0.7036,
+      "step": 183
+    },
+    {
+      "epoch": 3.2280701754385963,
+      "grad_norm": 0.521230563734182,
+      "learning_rate": 3.12696703292044e-06,
+      "loss": 0.6464,
+      "step": 184
+    },
+    {
+      "epoch": 3.245614035087719,
+      "grad_norm": 0.4238718196830614,
+      "learning_rate": 3.0731536692877596e-06,
+      "loss": 0.6318,
+      "step": 185
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.6299494548697394,
+      "learning_rate": 3.019601169804216e-06,
+      "loss": 0.5587,
+      "step": 186
+    },
+    {
+      "epoch": 3.280701754385965,
+      "grad_norm": 0.6353313899198718,
+      "learning_rate": 2.966316784621e-06,
+      "loss": 0.6487,
+      "step": 187
+    },
+    {
+      "epoch": 3.2982456140350878,
+      "grad_norm": 0.5028615630609446,
+      "learning_rate": 2.9133077275909112e-06,
+      "loss": 0.7141,
+      "step": 188
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.8022511847139254,
+      "learning_rate": 2.86058117529173e-06,
+      "loss": 0.6222,
+      "step": 189
+    },
+    {
+      "epoch": 3.3333333333333335,
+      "grad_norm": 0.4940462249079852,
+      "learning_rate": 2.8081442660546126e-06,
+      "loss": 0.6395,
+      "step": 190
+    },
+    {
+      "epoch": 3.3508771929824563,
+      "grad_norm": 0.5103743275328774,
+      "learning_rate": 2.7560040989976894e-06,
+      "loss": 0.6651,
+      "step": 191
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.42462365504941096,
+      "learning_rate": 2.7041677330649408e-06,
+      "loss": 0.5694,
+      "step": 192
+    },
+    {
+      "epoch": 3.3859649122807016,
+      "grad_norm": 0.46073929876061015,
+      "learning_rate": 2.6526421860705474e-06,
+      "loss": 0.5869,
+      "step": 193
+    },
+    {
+      "epoch": 3.4035087719298245,
+      "grad_norm": 0.5768914012401148,
+      "learning_rate": 2.601434433748771e-06,
+      "loss": 0.6139,
+      "step": 194
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.6901511256700266,
+      "learning_rate": 2.550551408809566e-06,
+      "loss": 0.5507,
+      "step": 195
+    },
+    {
+      "epoch": 3.43859649122807,
+      "grad_norm": 0.536661518552196,
+      "learning_rate": 2.5000000000000015e-06,
+      "loss": 0.6678,
+      "step": 196
+    },
+    {
+      "epoch": 3.456140350877193,
+      "grad_norm": 0.6253252899607207,
+      "learning_rate": 2.4497870511716237e-06,
+      "loss": 0.4905,
+      "step": 197
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.5220620720879305,
+      "learning_rate": 2.3999193603539234e-06,
+      "loss": 0.6132,
+      "step": 198
+    },
+    {
+      "epoch": 3.4912280701754383,
+      "grad_norm": 0.48419112630764066,
+      "learning_rate": 2.3504036788339763e-06,
+      "loss": 0.6503,
+      "step": 199
+    },
+    {
+      "epoch": 3.5087719298245617,
+      "grad_norm": 0.4196170231527922,
+      "learning_rate": 2.3012467102424373e-06,
+      "loss": 0.604,
+      "step": 200
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.47707756956491976,
+      "learning_rate": 2.2524551096459703e-06,
+      "loss": 0.6639,
+      "step": 201
+    },
+    {
+      "epoch": 3.543859649122807,
+      "grad_norm": 0.45698520194225517,
+      "learning_rate": 2.204035482646267e-06,
+      "loss": 0.5353,
+      "step": 202
+    },
+    {
+      "epoch": 3.56140350877193,
+      "grad_norm": 0.4659983194824815,
+      "learning_rate": 2.155994384485742e-06,
+      "loss": 0.7457,
+      "step": 203
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.5503579954840856,
+      "learning_rate": 2.1083383191600676e-06,
+      "loss": 0.6049,
+      "step": 204
+    },
+    {
+      "epoch": 3.5964912280701755,
+      "grad_norm": 0.5530544671573033,
+      "learning_rate": 2.061073738537635e-06,
+      "loss": 0.6299,
+      "step": 205
+    },
+    {
+      "epoch": 3.6140350877192984,
+      "grad_norm": 0.44375239245186526,
+      "learning_rate": 2.0142070414860704e-06,
+      "loss": 0.5419,
+      "step": 206
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.5739419491104609,
+      "learning_rate": 1.9677445730059348e-06,
+      "loss": 0.5619,
+      "step": 207
+    },
+    {
+      "epoch": 3.6491228070175437,
+      "grad_norm": 0.5078515782999371,
+      "learning_rate": 1.9216926233717087e-06,
+      "loss": 0.5915,
+      "step": 208
+    },
+    {
+      "epoch": 3.6666666666666665,
+      "grad_norm": 0.5298505895421687,
+      "learning_rate": 1.8760574272802002e-06,
+      "loss": 0.6672,
+      "step": 209
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.4863055522030361,
+      "learning_rate": 1.8308451630064484e-06,
+      "loss": 0.5836,
+      "step": 210
+    },
+    {
+      "epoch": 3.7017543859649122,
+      "grad_norm": 0.4820057101613254,
+      "learning_rate": 1.7860619515673034e-06,
+      "loss": 0.6679,
+      "step": 211
+    },
+    {
+      "epoch": 3.719298245614035,
+      "grad_norm": 0.4974206274184026,
+      "learning_rate": 1.7417138558927244e-06,
+      "loss": 0.6436,
+      "step": 212
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.512515881410681,
+      "learning_rate": 1.6978068800049624e-06,
+      "loss": 0.7007,
+      "step": 213
+    },
+    {
+      "epoch": 3.754385964912281,
+      "grad_norm": 0.4957119738463247,
+      "learning_rate": 1.6543469682057105e-06,
+      "loss": 0.7088,
+      "step": 214
+    },
+    {
+      "epoch": 3.7719298245614032,
+      "grad_norm": 0.44653983709178674,
+      "learning_rate": 1.611340004271339e-06,
+      "loss": 0.743,
+      "step": 215
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.5008066790687598,
+      "learning_rate": 1.5687918106563326e-06,
+      "loss": 0.6507,
+      "step": 216
+    },
+    {
+      "epoch": 3.807017543859649,
+      "grad_norm": 0.4892182274778227,
+      "learning_rate": 1.5267081477050132e-06,
+      "loss": 0.6839,
+      "step": 217
+    },
+    {
+      "epoch": 3.824561403508772,
+      "grad_norm": 0.48865392490573634,
+      "learning_rate": 1.4850947128716914e-06,
+      "loss": 0.7049,
+      "step": 218
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.48547105963711196,
+      "learning_rate": 1.4439571399493146e-06,
+      "loss": 0.6675,
+      "step": 219
+    },
+    {
+      "epoch": 3.8596491228070176,
+      "grad_norm": 0.5262312865201039,
+      "learning_rate": 1.4033009983067454e-06,
+      "loss": 0.5713,
+      "step": 220
+    },
+    {
+      "epoch": 3.8771929824561404,
+      "grad_norm": 0.4773548880334042,
+      "learning_rate": 1.3631317921347564e-06,
+      "loss": 0.6657,
+      "step": 221
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.5029287487968093,
+      "learning_rate": 1.3234549597008572e-06,
+      "loss": 0.6461,
+      "step": 222
+    },
+    {
+      "epoch": 3.912280701754386,
+      "grad_norm": 0.41704059643249763,
+      "learning_rate": 1.2842758726130283e-06,
+      "loss": 0.5772,
+      "step": 223
+    },
+    {
+      "epoch": 3.9298245614035086,
+      "grad_norm": 0.4861201391088637,
+      "learning_rate": 1.2455998350925042e-06,
+      "loss": 0.5911,
+      "step": 224
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.5696096357478914,
+      "learning_rate": 1.2074320832556558e-06,
+      "loss": 0.7097,
+      "step": 225
+    },
+    {
+      "epoch": 3.9649122807017543,
+      "grad_norm": 0.47797887333704275,
+      "learning_rate": 1.1697777844051105e-06,
+      "loss": 0.6667,
+      "step": 226
+    },
+    {
+      "epoch": 3.982456140350877,
+      "grad_norm": 0.43111294734231914,
+      "learning_rate": 1.132642036330181e-06,
+      "loss": 0.689,
+      "step": 227
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.5655747134876244,
+      "learning_rate": 1.096029866616704e-06,
+      "loss": 0.5972,
+      "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 sha256:0e6270a7737d26aa3d033366f072b4a2aadf6318801b293e0f33e4e583b74079
+size 8081