Upload continuous evaluation checkpoints, batch 16

.gitattributes

@@ -145,3 +145,7 @@ continuous/qwen3/mc2_g14_378/finetune/tokenizer.json filter=lfs diff=lfs merge=l
 continuous/qwen3/mc2_h14_378/finetune/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 continuous/qwen3/mc2_l14_224/finetune/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 continuous/qwen3/mc2_m16_224/finetune/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+continuous/qwen3/mc2_m16_224_mt5/finetune/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+continuous/qwen3/mc2_m16_384/finetune/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+continuous/qwen3/mc2_s16_224/finetune/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+continuous/qwen3/mc2_s16_224_mt5/finetune/tokenizer.json filter=lfs diff=lfs merge=lfs -text

continuous/qwen3/mc2_m16_224_mt5/finetune/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:df1af92d356350d4d76fc1a908457b0a942e0633fce64204fec054095ff766c2
+size 4161700600

continuous/qwen3/mc2_m16_224_mt5/finetune/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
+  }
+}

continuous/qwen3/mc2_m16_224_mt5/finetune/tokenizer.json

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

continuous/qwen3/mc2_m16_224_mt5/finetune/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 message.content is string 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.content is string %}\n        {%- set content = message.content %}\n    {%- else %}\n        {%- set content = '' %}\n    {%- endif %}\n    {%- if (message.role == \"user\") or (message.role == \"system\" and not loop.first) %}\n        {{- '<|im_start|>' + message.role + '\\n' + content + '<|im_end|>' + '\\n' }}\n    {%- elif message.role == \"assistant\" %}\n        {%- set reasoning_content = '' %}\n        {%- if message.reasoning_content is string %}\n            {%- set reasoning_content = message.reasoning_content %}\n        {%- else %}\n            {%- if '</think>' in content %}\n                {%- set reasoning_content = content.split('</think>')[0].rstrip('\\n').split('<think>')[-1].lstrip('\\n') %}\n                {%- set content = content.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        {{- 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": 8192,
+  "pad_token": "<|endoftext|>",
+  "padding_side": "right",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

continuous/qwen3/mc2_m16_224_mt5/finetune/trainer_state.json

@@ -0,0 +1,1856 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999037443449803,
+  "eval_steps": 500,
+  "global_step": 2597,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.003850226200789296,
+      "grad_norm": 18.24117660522461,
+      "learning_rate": 1.153846153846154e-06,
+      "loss": 2.4511,
+      "step": 10
+    },
+    {
+      "epoch": 0.007700452401578592,
+      "grad_norm": 8.585675239562988,
+      "learning_rate": 2.435897435897436e-06,
+      "loss": 1.9885,
+      "step": 20
+    },
+    {
+      "epoch": 0.01155067860236789,
+      "grad_norm": 4.398549556732178,
+      "learning_rate": 3.7179487179487184e-06,
+      "loss": 1.5038,
+      "step": 30
+    },
+    {
+      "epoch": 0.015400904803157185,
+      "grad_norm": 1.149141550064087,
+      "learning_rate": 5e-06,
+      "loss": 1.1756,
+      "step": 40
+    },
+    {
+      "epoch": 0.019251131003946483,
+      "grad_norm": 0.8464040160179138,
+      "learning_rate": 6.282051282051282e-06,
+      "loss": 1.0645,
+      "step": 50
+    },
+    {
+      "epoch": 0.02310135720473578,
+      "grad_norm": 0.4418105483055115,
+      "learning_rate": 7.564102564102564e-06,
+      "loss": 0.972,
+      "step": 60
+    },
+    {
+      "epoch": 0.026951583405525073,
+      "grad_norm": 0.33305835723876953,
+      "learning_rate": 8.846153846153847e-06,
+      "loss": 0.9205,
+      "step": 70
+    },
+    {
+      "epoch": 0.03080180960631437,
+      "grad_norm": 0.302616685628891,
+      "learning_rate": 9.999996111488723e-06,
+      "loss": 0.9182,
+      "step": 80
+    },
+    {
+      "epoch": 0.034652035807103666,
+      "grad_norm": 0.2908528745174408,
+      "learning_rate": 9.999529497453782e-06,
+      "loss": 0.8844,
+      "step": 90
+    },
+    {
+      "epoch": 0.038502262007892966,
+      "grad_norm": 0.24981921911239624,
+      "learning_rate": 9.998285264324193e-06,
+      "loss": 0.8635,
+      "step": 100
+    },
+    {
+      "epoch": 0.04235248820868226,
+      "grad_norm": 0.2507269084453583,
+      "learning_rate": 9.996263605626052e-06,
+      "loss": 0.8463,
+      "step": 110
+    },
+    {
+      "epoch": 0.04620271440947156,
+      "grad_norm": 0.2601762115955353,
+      "learning_rate": 9.99346483580503e-06,
+      "loss": 0.8558,
+      "step": 120
+    },
+    {
+      "epoch": 0.05005294061026085,
+      "grad_norm": 0.24757453799247742,
+      "learning_rate": 9.989889390177467e-06,
+      "loss": 0.8495,
+      "step": 130
+    },
+    {
+      "epoch": 0.053903166811050146,
+      "grad_norm": 0.29140859842300415,
+      "learning_rate": 9.985537824862647e-06,
+      "loss": 0.8453,
+      "step": 140
+    },
+    {
+      "epoch": 0.057753393011839446,
+      "grad_norm": 0.30317628383636475,
+      "learning_rate": 9.980410816696317e-06,
+      "loss": 0.83,
+      "step": 150
+    },
+    {
+      "epoch": 0.06160361921262874,
+      "grad_norm": 0.2792593836784363,
+      "learning_rate": 9.97450916312541e-06,
+      "loss": 0.8369,
+      "step": 160
+    },
+    {
+      "epoch": 0.06545384541341803,
+      "grad_norm": 0.2370644509792328,
+      "learning_rate": 9.967833782084004e-06,
+      "loss": 0.8369,
+      "step": 170
+    },
+    {
+      "epoch": 0.06930407161420733,
+      "grad_norm": 0.2749076783657074,
+      "learning_rate": 9.960385711850554e-06,
+      "loss": 0.8623,
+      "step": 180
+    },
+    {
+      "epoch": 0.07315429781499663,
+      "grad_norm": 0.261151522397995,
+      "learning_rate": 9.952166110886394e-06,
+      "loss": 0.8536,
+      "step": 190
+    },
+    {
+      "epoch": 0.07700452401578593,
+      "grad_norm": 0.2802771031856537,
+      "learning_rate": 9.943176257655567e-06,
+      "loss": 0.8408,
+      "step": 200
+    },
+    {
+      "epoch": 0.08085475021657522,
+      "grad_norm": 0.2472686618566513,
+      "learning_rate": 9.933417550425952e-06,
+      "loss": 0.8391,
+      "step": 210
+    },
+    {
+      "epoch": 0.08470497641736452,
+      "grad_norm": 0.26069584488868713,
+      "learning_rate": 9.9228915070518e-06,
+      "loss": 0.8235,
+      "step": 220
+    },
+    {
+      "epoch": 0.08855520261815382,
+      "grad_norm": 0.24521784484386444,
+      "learning_rate": 9.911599764737632e-06,
+      "loss": 0.8244,
+      "step": 230
+    },
+    {
+      "epoch": 0.09240542881894312,
+      "grad_norm": 0.2512511610984802,
+      "learning_rate": 9.899544079783606e-06,
+      "loss": 0.819,
+      "step": 240
+    },
+    {
+      "epoch": 0.0962556550197324,
+      "grad_norm": 0.28951480984687805,
+      "learning_rate": 9.886726327312338e-06,
+      "loss": 0.8363,
+      "step": 250
+    },
+    {
+      "epoch": 0.1001058812205217,
+      "grad_norm": 0.24346619844436646,
+      "learning_rate": 9.87314850097724e-06,
+      "loss": 0.8307,
+      "step": 260
+    },
+    {
+      "epoch": 0.103956107421311,
+      "grad_norm": 0.26015767455101013,
+      "learning_rate": 9.858812712652452e-06,
+      "loss": 0.8206,
+      "step": 270
+    },
+    {
+      "epoch": 0.10780633362210029,
+      "grad_norm": 0.24395637214183807,
+      "learning_rate": 9.843721192104341e-06,
+      "loss": 0.8191,
+      "step": 280
+    },
+    {
+      "epoch": 0.11165655982288959,
+      "grad_norm": 0.245630145072937,
+      "learning_rate": 9.827876286644698e-06,
+      "loss": 0.8361,
+      "step": 290
+    },
+    {
+      "epoch": 0.11550678602367889,
+      "grad_norm": 0.2287023663520813,
+      "learning_rate": 9.811280460765643e-06,
+      "loss": 0.7975,
+      "step": 300
+    },
+    {
+      "epoch": 0.11935701222446819,
+      "grad_norm": 0.23688051104545593,
+      "learning_rate": 9.793936295756292e-06,
+      "loss": 0.8102,
+      "step": 310
+    },
+    {
+      "epoch": 0.12320723842525748,
+      "grad_norm": 0.2466711550951004,
+      "learning_rate": 9.775846489301279e-06,
+      "loss": 0.8106,
+      "step": 320
+    },
+    {
+      "epoch": 0.12705746462604678,
+      "grad_norm": 0.2627280652523041,
+      "learning_rate": 9.757013855061157e-06,
+      "loss": 0.8095,
+      "step": 330
+    },
+    {
+      "epoch": 0.13090769082683607,
+      "grad_norm": 0.2565971314907074,
+      "learning_rate": 9.737441322234758e-06,
+      "loss": 0.8215,
+      "step": 340
+    },
+    {
+      "epoch": 0.13475791702762538,
+      "grad_norm": 0.26523566246032715,
+      "learning_rate": 9.717131935103605e-06,
+      "loss": 0.8216,
+      "step": 350
+    },
+    {
+      "epoch": 0.13860814322841467,
+      "grad_norm": 0.2439344823360443,
+      "learning_rate": 9.696088852558398e-06,
+      "loss": 0.8085,
+      "step": 360
+    },
+    {
+      "epoch": 0.14245836942920398,
+      "grad_norm": 0.2628968358039856,
+      "learning_rate": 9.674315347607686e-06,
+      "loss": 0.8347,
+      "step": 370
+    },
+    {
+      "epoch": 0.14630859562999327,
+      "grad_norm": 0.244170680642128,
+      "learning_rate": 9.651814806868796e-06,
+      "loss": 0.7952,
+      "step": 380
+    },
+    {
+      "epoch": 0.15015882183078255,
+      "grad_norm": 0.2361016869544983,
+      "learning_rate": 9.62859073004107e-06,
+      "loss": 0.8,
+      "step": 390
+    },
+    {
+      "epoch": 0.15400904803157187,
+      "grad_norm": 0.24069201946258545,
+      "learning_rate": 9.604646729361547e-06,
+      "loss": 0.8022,
+      "step": 400
+    },
+    {
+      "epoch": 0.15785927423236115,
+      "grad_norm": 0.23068426549434662,
+      "learning_rate": 9.579986529043099e-06,
+      "loss": 0.7909,
+      "step": 410
+    },
+    {
+      "epoch": 0.16170950043315044,
+      "grad_norm": 0.2401607781648636,
+      "learning_rate": 9.554613964695189e-06,
+      "loss": 0.8107,
+      "step": 420
+    },
+    {
+      "epoch": 0.16555972663393975,
+      "grad_norm": 0.24136482179164886,
+      "learning_rate": 9.528532982727281e-06,
+      "loss": 0.8016,
+      "step": 430
+    },
+    {
+      "epoch": 0.16940995283472904,
+      "grad_norm": 0.23997458815574646,
+      "learning_rate": 9.501747639735018e-06,
+      "loss": 0.7977,
+      "step": 440
+    },
+    {
+      "epoch": 0.17326017903551832,
+      "grad_norm": 0.2505742907524109,
+      "learning_rate": 9.474262101869272e-06,
+      "loss": 0.787,
+      "step": 450
+    },
+    {
+      "epoch": 0.17711040523630764,
+      "grad_norm": 0.23870500922203064,
+      "learning_rate": 9.446080644188141e-06,
+      "loss": 0.7937,
+      "step": 460
+    },
+    {
+      "epoch": 0.18096063143709692,
+      "grad_norm": 0.2324884533882141,
+      "learning_rate": 9.417207649992017e-06,
+      "loss": 0.7987,
+      "step": 470
+    },
+    {
+      "epoch": 0.18481085763788624,
+      "grad_norm": 0.24133141338825226,
+      "learning_rate": 9.387647610141816e-06,
+      "loss": 0.7992,
+      "step": 480
+    },
+    {
+      "epoch": 0.18866108383867553,
+      "grad_norm": 0.2273392230272293,
+      "learning_rate": 9.35740512236047e-06,
+      "loss": 0.7853,
+      "step": 490
+    },
+    {
+      "epoch": 0.1925113100394648,
+      "grad_norm": 0.22827138006687164,
+      "learning_rate": 9.326484890517809e-06,
+      "loss": 0.7931,
+      "step": 500
+    },
+    {
+      "epoch": 0.19636153624025413,
+      "grad_norm": 0.23036795854568481,
+      "learning_rate": 9.294891723898926e-06,
+      "loss": 0.7867,
+      "step": 510
+    },
+    {
+      "epoch": 0.2002117624410434,
+      "grad_norm": 0.23646484315395355,
+      "learning_rate": 9.262630536456153e-06,
+      "loss": 0.8167,
+      "step": 520
+    },
+    {
+      "epoch": 0.2040619886418327,
+      "grad_norm": 0.26009538769721985,
+      "learning_rate": 9.229706346044749e-06,
+      "loss": 0.7872,
+      "step": 530
+    },
+    {
+      "epoch": 0.207912214842622,
+      "grad_norm": 0.2590685486793518,
+      "learning_rate": 9.196124273642422e-06,
+      "loss": 0.802,
+      "step": 540
+    },
+    {
+      "epoch": 0.2117624410434113,
+      "grad_norm": 0.2258717119693756,
+      "learning_rate": 9.161889542552842e-06,
+      "loss": 0.8061,
+      "step": 550
+    },
+    {
+      "epoch": 0.21561266724420058,
+      "grad_norm": 0.2175668478012085,
+      "learning_rate": 9.127007477593193e-06,
+      "loss": 0.8074,
+      "step": 560
+    },
+    {
+      "epoch": 0.2194628934449899,
+      "grad_norm": 0.2743118405342102,
+      "learning_rate": 9.091483504265974e-06,
+      "loss": 0.8143,
+      "step": 570
+    },
+    {
+      "epoch": 0.22331311964577918,
+      "grad_norm": 0.24809886515140533,
+      "learning_rate": 9.055323147915116e-06,
+      "loss": 0.7972,
+      "step": 580
+    },
+    {
+      "epoch": 0.2271633458465685,
+      "grad_norm": 0.2565881013870239,
+      "learning_rate": 9.018532032866585e-06,
+      "loss": 0.8059,
+      "step": 590
+    },
+    {
+      "epoch": 0.23101357204735778,
+      "grad_norm": 0.2181779146194458,
+      "learning_rate": 8.981115881553579e-06,
+      "loss": 0.798,
+      "step": 600
+    },
+    {
+      "epoch": 0.23486379824814707,
+      "grad_norm": 0.25907787680625916,
+      "learning_rate": 8.943080513626476e-06,
+      "loss": 0.8182,
+      "step": 610
+    },
+    {
+      "epoch": 0.23871402444893638,
+      "grad_norm": 0.23706406354904175,
+      "learning_rate": 8.904431845047641e-06,
+      "loss": 0.8019,
+      "step": 620
+    },
+    {
+      "epoch": 0.24256425064972567,
+      "grad_norm": 0.24183323979377747,
+      "learning_rate": 8.86517588717128e-06,
+      "loss": 0.8179,
+      "step": 630
+    },
+    {
+      "epoch": 0.24641447685051496,
+      "grad_norm": 0.23398582637310028,
+      "learning_rate": 8.82531874580844e-06,
+      "loss": 0.8095,
+      "step": 640
+    },
+    {
+      "epoch": 0.25026470305130427,
+      "grad_norm": 0.21613161265850067,
+      "learning_rate": 8.784866620277314e-06,
+      "loss": 0.7667,
+      "step": 650
+    },
+    {
+      "epoch": 0.25411492925209356,
+      "grad_norm": 0.24961239099502563,
+      "learning_rate": 8.743825802439018e-06,
+      "loss": 0.7992,
+      "step": 660
+    },
+    {
+      "epoch": 0.25796515545288284,
+      "grad_norm": 0.2629765272140503,
+      "learning_rate": 8.702202675718956e-06,
+      "loss": 0.8082,
+      "step": 670
+    },
+    {
+      "epoch": 0.26181538165367213,
+      "grad_norm": 0.23281703889369965,
+      "learning_rate": 8.660003714113959e-06,
+      "loss": 0.7756,
+      "step": 680
+    },
+    {
+      "epoch": 0.26566560785446147,
+      "grad_norm": 0.2330506294965744,
+      "learning_rate": 8.617235481185323e-06,
+      "loss": 0.7997,
+      "step": 690
+    },
+    {
+      "epoch": 0.26951583405525076,
+      "grad_norm": 0.24595999717712402,
+      "learning_rate": 8.573904629037931e-06,
+      "loss": 0.8061,
+      "step": 700
+    },
+    {
+      "epoch": 0.27336606025604004,
+      "grad_norm": 0.23494243621826172,
+      "learning_rate": 8.530017897285585e-06,
+      "loss": 0.7879,
+      "step": 710
+    },
+    {
+      "epoch": 0.27721628645682933,
+      "grad_norm": 0.21570800244808197,
+      "learning_rate": 8.485582112002746e-06,
+      "loss": 0.7919,
+      "step": 720
+    },
+    {
+      "epoch": 0.2810665126576186,
+      "grad_norm": 0.2443312555551529,
+      "learning_rate": 8.440604184662811e-06,
+      "loss": 0.7962,
+      "step": 730
+    },
+    {
+      "epoch": 0.28491673885840796,
+      "grad_norm": 0.25895556807518005,
+      "learning_rate": 8.395091111063118e-06,
+      "loss": 0.7828,
+      "step": 740
+    },
+    {
+      "epoch": 0.28876696505919724,
+      "grad_norm": 0.2196115106344223,
+      "learning_rate": 8.349049970236822e-06,
+      "loss": 0.7848,
+      "step": 750
+    },
+    {
+      "epoch": 0.29261719125998653,
+      "grad_norm": 0.23485928773880005,
+      "learning_rate": 8.302487923351842e-06,
+      "loss": 0.7771,
+      "step": 760
+    },
+    {
+      "epoch": 0.2964674174607758,
+      "grad_norm": 0.23182353377342224,
+      "learning_rate": 8.255412212597021e-06,
+      "loss": 0.7944,
+      "step": 770
+    },
+    {
+      "epoch": 0.3003176436615651,
+      "grad_norm": 0.2356346696615219,
+      "learning_rate": 8.207830160055688e-06,
+      "loss": 0.7917,
+      "step": 780
+    },
+    {
+      "epoch": 0.3041678698623544,
+      "grad_norm": 0.24302947521209717,
+      "learning_rate": 8.159749166566786e-06,
+      "loss": 0.7754,
+      "step": 790
+    },
+    {
+      "epoch": 0.30801809606314373,
+      "grad_norm": 0.2251223921775818,
+      "learning_rate": 8.11117671057377e-06,
+      "loss": 0.803,
+      "step": 800
+    },
+    {
+      "epoch": 0.311868322263933,
+      "grad_norm": 0.24077272415161133,
+      "learning_rate": 8.062120346961404e-06,
+      "loss": 0.7951,
+      "step": 810
+    },
+    {
+      "epoch": 0.3157185484647223,
+      "grad_norm": 0.2308858186006546,
+      "learning_rate": 8.012587705880705e-06,
+      "loss": 0.7876,
+      "step": 820
+    },
+    {
+      "epoch": 0.3195687746655116,
+      "grad_norm": 0.2361987829208374,
+      "learning_rate": 7.962586491562146e-06,
+      "loss": 0.7736,
+      "step": 830
+    },
+    {
+      "epoch": 0.3234190008663009,
+      "grad_norm": 0.2420838177204132,
+      "learning_rate": 7.912124481117357e-06,
+      "loss": 0.7802,
+      "step": 840
+    },
+    {
+      "epoch": 0.3272692270670902,
+      "grad_norm": 0.2423592507839203,
+      "learning_rate": 7.861209523329494e-06,
+      "loss": 0.7927,
+      "step": 850
+    },
+    {
+      "epoch": 0.3311194532678795,
+      "grad_norm": 0.23329907655715942,
+      "learning_rate": 7.809849537432432e-06,
+      "loss": 0.7736,
+      "step": 860
+    },
+    {
+      "epoch": 0.3349696794686688,
+      "grad_norm": 0.22841912508010864,
+      "learning_rate": 7.758052511879035e-06,
+      "loss": 0.7855,
+      "step": 870
+    },
+    {
+      "epoch": 0.3388199056694581,
+      "grad_norm": 0.2264324575662613,
+      "learning_rate": 7.705826503098642e-06,
+      "loss": 0.7803,
+      "step": 880
+    },
+    {
+      "epoch": 0.34267013187024736,
+      "grad_norm": 0.2276768535375595,
+      "learning_rate": 7.653179634243979e-06,
+      "loss": 0.7809,
+      "step": 890
+    },
+    {
+      "epoch": 0.34652035807103665,
+      "grad_norm": 0.21411298215389252,
+      "learning_rate": 7.600120093927696e-06,
+      "loss": 0.784,
+      "step": 900
+    },
+    {
+      "epoch": 0.350370584271826,
+      "grad_norm": 0.23348477482795715,
+      "learning_rate": 7.546656134948729e-06,
+      "loss": 0.7737,
+      "step": 910
+    },
+    {
+      "epoch": 0.3542208104726153,
+      "grad_norm": 0.2350476235151291,
+      "learning_rate": 7.492796073008661e-06,
+      "loss": 0.7749,
+      "step": 920
+    },
+    {
+      "epoch": 0.35807103667340456,
+      "grad_norm": 0.24385926127433777,
+      "learning_rate": 7.438548285418326e-06,
+      "loss": 0.7601,
+      "step": 930
+    },
+    {
+      "epoch": 0.36192126287419385,
+      "grad_norm": 0.24922111630439758,
+      "learning_rate": 7.3839212097948e-06,
+      "loss": 0.7947,
+      "step": 940
+    },
+    {
+      "epoch": 0.36577148907498314,
+      "grad_norm": 0.23613817989826202,
+      "learning_rate": 7.328923342749039e-06,
+      "loss": 0.809,
+      "step": 950
+    },
+    {
+      "epoch": 0.3696217152757725,
+      "grad_norm": 0.25151118636131287,
+      "learning_rate": 7.273563238564319e-06,
+      "loss": 0.7935,
+      "step": 960
+    },
+    {
+      "epoch": 0.37347194147656176,
+      "grad_norm": 0.24006135761737823,
+      "learning_rate": 7.217849507865724e-06,
+      "loss": 0.773,
+      "step": 970
+    },
+    {
+      "epoch": 0.37732216767735105,
+      "grad_norm": 0.22567878663539886,
+      "learning_rate": 7.161790816280858e-06,
+      "loss": 0.7803,
+      "step": 980
+    },
+    {
+      "epoch": 0.38117239387814034,
+      "grad_norm": 0.24229654669761658,
+      "learning_rate": 7.1053958830920125e-06,
+      "loss": 0.7629,
+      "step": 990
+    },
+    {
+      "epoch": 0.3850226200789296,
+      "grad_norm": 0.23457102477550507,
+      "learning_rate": 7.0486734798799715e-06,
+      "loss": 0.7825,
+      "step": 1000
+    },
+    {
+      "epoch": 0.3888728462797189,
+      "grad_norm": 0.23638352751731873,
+      "learning_rate": 6.991632429159708e-06,
+      "loss": 0.8054,
+      "step": 1010
+    },
+    {
+      "epoch": 0.39272307248050825,
+      "grad_norm": 0.2624192237854004,
+      "learning_rate": 6.934281603008132e-06,
+      "loss": 0.7959,
+      "step": 1020
+    },
+    {
+      "epoch": 0.39657329868129754,
+      "grad_norm": 0.23721525073051453,
+      "learning_rate": 6.876629921684143e-06,
+      "loss": 0.7929,
+      "step": 1030
+    },
+    {
+      "epoch": 0.4004235248820868,
+      "grad_norm": 0.2391992211341858,
+      "learning_rate": 6.818686352241199e-06,
+      "loss": 0.7837,
+      "step": 1040
+    },
+    {
+      "epoch": 0.4042737510828761,
+      "grad_norm": 0.21158599853515625,
+      "learning_rate": 6.760459907132573e-06,
+      "loss": 0.7788,
+      "step": 1050
+    },
+    {
+      "epoch": 0.4081239772836654,
+      "grad_norm": 0.21910063922405243,
+      "learning_rate": 6.701959642809589e-06,
+      "loss": 0.7845,
+      "step": 1060
+    },
+    {
+      "epoch": 0.41197420348445474,
+      "grad_norm": 0.23000028729438782,
+      "learning_rate": 6.643194658312987e-06,
+      "loss": 0.7772,
+      "step": 1070
+    },
+    {
+      "epoch": 0.415824429685244,
+      "grad_norm": 0.22396276891231537,
+      "learning_rate": 6.584174093857676e-06,
+      "loss": 0.8013,
+      "step": 1080
+    },
+    {
+      "epoch": 0.4196746558860333,
+      "grad_norm": 0.26117366552352905,
+      "learning_rate": 6.5249071294110735e-06,
+      "loss": 0.7833,
+      "step": 1090
+    },
+    {
+      "epoch": 0.4235248820868226,
+      "grad_norm": 0.24859198927879333,
+      "learning_rate": 6.465402983265274e-06,
+      "loss": 0.7972,
+      "step": 1100
+    },
+    {
+      "epoch": 0.4273751082876119,
+      "grad_norm": 0.2363201379776001,
+      "learning_rate": 6.405670910603243e-06,
+      "loss": 0.7922,
+      "step": 1110
+    },
+    {
+      "epoch": 0.43122533448840117,
+      "grad_norm": 0.2567698061466217,
+      "learning_rate": 6.345720202059291e-06,
+      "loss": 0.7879,
+      "step": 1120
+    },
+    {
+      "epoch": 0.4350755606891905,
+      "grad_norm": 0.23966087400913239,
+      "learning_rate": 6.285560182274008e-06,
+      "loss": 0.7667,
+      "step": 1130
+    },
+    {
+      "epoch": 0.4389257868899798,
+      "grad_norm": 0.23583926260471344,
+      "learning_rate": 6.225200208443936e-06,
+      "loss": 0.7896,
+      "step": 1140
+    },
+    {
+      "epoch": 0.4427760130907691,
+      "grad_norm": 0.22360938787460327,
+      "learning_rate": 6.16464966886616e-06,
+      "loss": 0.7786,
+      "step": 1150
+    },
+    {
+      "epoch": 0.44662623929155837,
+      "grad_norm": 0.27130186557769775,
+      "learning_rate": 6.103917981478065e-06,
+      "loss": 0.8025,
+      "step": 1160
+    },
+    {
+      "epoch": 0.45047646549234766,
+      "grad_norm": 0.21909376978874207,
+      "learning_rate": 6.043014592392489e-06,
+      "loss": 0.8131,
+      "step": 1170
+    },
+    {
+      "epoch": 0.454326691693137,
+      "grad_norm": 0.22135120630264282,
+      "learning_rate": 5.981948974428485e-06,
+      "loss": 0.7887,
+      "step": 1180
+    },
+    {
+      "epoch": 0.4581769178939263,
+      "grad_norm": 0.24061603844165802,
+      "learning_rate": 5.920730625637934e-06,
+      "loss": 0.7653,
+      "step": 1190
+    },
+    {
+      "epoch": 0.46202714409471557,
+      "grad_norm": 0.247635617852211,
+      "learning_rate": 5.8593690678282355e-06,
+      "loss": 0.7782,
+      "step": 1200
+    },
+    {
+      "epoch": 0.46587737029550486,
+      "grad_norm": 0.2462078183889389,
+      "learning_rate": 5.797873845081301e-06,
+      "loss": 0.7826,
+      "step": 1210
+    },
+    {
+      "epoch": 0.46972759649629414,
+      "grad_norm": 0.2337314486503601,
+      "learning_rate": 5.736254522269075e-06,
+      "loss": 0.7986,
+      "step": 1220
+    },
+    {
+      "epoch": 0.47357782269708343,
+      "grad_norm": 0.24164052307605743,
+      "learning_rate": 5.6745206835658386e-06,
+      "loss": 0.8009,
+      "step": 1230
+    },
+    {
+      "epoch": 0.47742804889787277,
+      "grad_norm": 0.22943809628486633,
+      "learning_rate": 5.612681930957495e-06,
+      "loss": 0.7723,
+      "step": 1240
+    },
+    {
+      "epoch": 0.48127827509866206,
+      "grad_norm": 0.22526513040065765,
+      "learning_rate": 5.550747882748095e-06,
+      "loss": 0.7697,
+      "step": 1250
+    },
+    {
+      "epoch": 0.48512850129945134,
+      "grad_norm": 0.2629677355289459,
+      "learning_rate": 5.4887281720638195e-06,
+      "loss": 0.777,
+      "step": 1260
+    },
+    {
+      "epoch": 0.48897872750024063,
+      "grad_norm": 0.2622842490673065,
+      "learning_rate": 5.42663244535466e-06,
+      "loss": 0.7608,
+      "step": 1270
+    },
+    {
+      "epoch": 0.4928289537010299,
+      "grad_norm": 0.25055283308029175,
+      "learning_rate": 5.3644703608940205e-06,
+      "loss": 0.7693,
+      "step": 1280
+    },
+    {
+      "epoch": 0.49667917990181926,
+      "grad_norm": 0.23238544166088104,
+      "learning_rate": 5.302251587276491e-06,
+      "loss": 0.7809,
+      "step": 1290
+    },
+    {
+      "epoch": 0.5005294061026085,
+      "grad_norm": 0.2264910638332367,
+      "learning_rate": 5.2399858019140005e-06,
+      "loss": 0.7805,
+      "step": 1300
+    },
+    {
+      "epoch": 0.5043796323033978,
+      "grad_norm": 0.22763748466968536,
+      "learning_rate": 5.177682689530615e-06,
+      "loss": 0.7961,
+      "step": 1310
+    },
+    {
+      "epoch": 0.5082298585041871,
+      "grad_norm": 0.24264036118984222,
+      "learning_rate": 5.11535194065619e-06,
+      "loss": 0.763,
+      "step": 1320
+    },
+    {
+      "epoch": 0.5120800847049765,
+      "grad_norm": 0.25384509563446045,
+      "learning_rate": 5.053003250119113e-06,
+      "loss": 0.7724,
+      "step": 1330
+    },
+    {
+      "epoch": 0.5159303109057657,
+      "grad_norm": 0.23815804719924927,
+      "learning_rate": 4.9906463155383925e-06,
+      "loss": 0.7823,
+      "step": 1340
+    },
+    {
+      "epoch": 0.519780537106555,
+      "grad_norm": 0.23131294548511505,
+      "learning_rate": 4.9282908358153046e-06,
+      "loss": 0.7746,
+      "step": 1350
+    },
+    {
+      "epoch": 0.5236307633073443,
+      "grad_norm": 0.2700917720794678,
+      "learning_rate": 4.8659465096248345e-06,
+      "loss": 0.7907,
+      "step": 1360
+    },
+    {
+      "epoch": 0.5274809895081336,
+      "grad_norm": 0.25883063673973083,
+      "learning_rate": 4.803623033907169e-06,
+      "loss": 0.789,
+      "step": 1370
+    },
+    {
+      "epoch": 0.5313312157089229,
+      "grad_norm": 0.22812847793102264,
+      "learning_rate": 4.741330102359441e-06,
+      "loss": 0.7951,
+      "step": 1380
+    },
+    {
+      "epoch": 0.5351814419097122,
+      "grad_norm": 0.25260505080223083,
+      "learning_rate": 4.679077403927996e-06,
+      "loss": 0.7745,
+      "step": 1390
+    },
+    {
+      "epoch": 0.5390316681105015,
+      "grad_norm": 0.22949516773223877,
+      "learning_rate": 4.616874621301373e-06,
+      "loss": 0.7747,
+      "step": 1400
+    },
+    {
+      "epoch": 0.5428818943112907,
+      "grad_norm": 0.24113507568836212,
+      "learning_rate": 4.554731429404293e-06,
+      "loss": 0.7898,
+      "step": 1410
+    },
+    {
+      "epoch": 0.5467321205120801,
+      "grad_norm": 0.2551576495170593,
+      "learning_rate": 4.492657493892822e-06,
+      "loss": 0.7727,
+      "step": 1420
+    },
+    {
+      "epoch": 0.5505823467128694,
+      "grad_norm": 0.26276305317878723,
+      "learning_rate": 4.430662469650999e-06,
+      "loss": 0.7883,
+      "step": 1430
+    },
+    {
+      "epoch": 0.5544325729136587,
+      "grad_norm": 0.257389098405838,
+      "learning_rate": 4.368755999289118e-06,
+      "loss": 0.7924,
+      "step": 1440
+    },
+    {
+      "epoch": 0.558282799114448,
+      "grad_norm": 0.26279523968696594,
+      "learning_rate": 4.306947711643949e-06,
+      "loss": 0.7767,
+      "step": 1450
+    },
+    {
+      "epoch": 0.5621330253152372,
+      "grad_norm": 0.2398219257593155,
+      "learning_rate": 4.245247220281068e-06,
+      "loss": 0.7795,
+      "step": 1460
+    },
+    {
+      "epoch": 0.5659832515160266,
+      "grad_norm": 0.23530074954032898,
+      "learning_rate": 4.183664121999587e-06,
+      "loss": 0.7771,
+      "step": 1470
+    },
+    {
+      "epoch": 0.5698334777168159,
+      "grad_norm": 0.2413652092218399,
+      "learning_rate": 4.122207995339474e-06,
+      "loss": 0.7741,
+      "step": 1480
+    },
+    {
+      "epoch": 0.5736837039176051,
+      "grad_norm": 0.23564939200878143,
+      "learning_rate": 4.060888399091734e-06,
+      "loss": 0.7782,
+      "step": 1490
+    },
+    {
+      "epoch": 0.5775339301183945,
+      "grad_norm": 0.23303566873073578,
+      "learning_rate": 3.999714870811639e-06,
+      "loss": 0.7882,
+      "step": 1500
+    },
+    {
+      "epoch": 0.5813841563191837,
+      "grad_norm": 0.23346631228923798,
+      "learning_rate": 3.938696925335275e-06,
+      "loss": 0.7807,
+      "step": 1510
+    },
+    {
+      "epoch": 0.5852343825199731,
+      "grad_norm": 0.23135046660900116,
+      "learning_rate": 3.87784405329962e-06,
+      "loss": 0.7885,
+      "step": 1520
+    },
+    {
+      "epoch": 0.5890846087207623,
+      "grad_norm": 0.23338374495506287,
+      "learning_rate": 3.817165719666382e-06,
+      "loss": 0.7897,
+      "step": 1530
+    },
+    {
+      "epoch": 0.5929348349215516,
+      "grad_norm": 0.2296774536371231,
+      "learning_rate": 3.756671362249841e-06,
+      "loss": 0.794,
+      "step": 1540
+    },
+    {
+      "epoch": 0.596785061122341,
+      "grad_norm": 0.23290300369262695,
+      "learning_rate": 3.6963703902488907e-06,
+      "loss": 0.7808,
+      "step": 1550
+    },
+    {
+      "epoch": 0.6006352873231302,
+      "grad_norm": 0.23095272481441498,
+      "learning_rate": 3.6362721827835655e-06,
+      "loss": 0.7976,
+      "step": 1560
+    },
+    {
+      "epoch": 0.6044855135239195,
+      "grad_norm": 0.24652154743671417,
+      "learning_rate": 3.576386087436212e-06,
+      "loss": 0.7935,
+      "step": 1570
+    },
+    {
+      "epoch": 0.6083357397247088,
+      "grad_norm": 0.2430710792541504,
+      "learning_rate": 3.5167214187975913e-06,
+      "loss": 0.7827,
+      "step": 1580
+    },
+    {
+      "epoch": 0.6121859659254981,
+      "grad_norm": 0.2058751881122589,
+      "learning_rate": 3.457287457018086e-06,
+      "loss": 0.7854,
+      "step": 1590
+    },
+    {
+      "epoch": 0.6160361921262875,
+      "grad_norm": 0.22652670741081238,
+      "learning_rate": 3.3980934463643003e-06,
+      "loss": 0.7733,
+      "step": 1600
+    },
+    {
+      "epoch": 0.6198864183270767,
+      "grad_norm": 0.21496686339378357,
+      "learning_rate": 3.3391485937812053e-06,
+      "loss": 0.7562,
+      "step": 1610
+    },
+    {
+      "epoch": 0.623736644527866,
+      "grad_norm": 0.23435050249099731,
+      "learning_rate": 3.280462067460115e-06,
+      "loss": 0.7542,
+      "step": 1620
+    },
+    {
+      "epoch": 0.6275868707286553,
+      "grad_norm": 0.23746740818023682,
+      "learning_rate": 3.222042995412669e-06,
+      "loss": 0.7855,
+      "step": 1630
+    },
+    {
+      "epoch": 0.6314370969294446,
+      "grad_norm": 0.21494735777378082,
+      "learning_rate": 3.163900464051089e-06,
+      "loss": 0.7752,
+      "step": 1640
+    },
+    {
+      "epoch": 0.635287323130234,
+      "grad_norm": 0.23886409401893616,
+      "learning_rate": 3.1060435167748826e-06,
+      "loss": 0.785,
+      "step": 1650
+    },
+    {
+      "epoch": 0.6391375493310232,
+      "grad_norm": 0.2335311472415924,
+      "learning_rate": 3.048481152564253e-06,
+      "loss": 0.7655,
+      "step": 1660
+    },
+    {
+      "epoch": 0.6429877755318125,
+      "grad_norm": 0.2433008849620819,
+      "learning_rate": 2.9912223245804017e-06,
+      "loss": 0.7886,
+      "step": 1670
+    },
+    {
+      "epoch": 0.6468380017326018,
+      "grad_norm": 0.24375061690807343,
+      "learning_rate": 2.934275938772981e-06,
+      "loss": 0.7826,
+      "step": 1680
+    },
+    {
+      "epoch": 0.6506882279333911,
+      "grad_norm": 0.2640165686607361,
+      "learning_rate": 2.8776508524948676e-06,
+      "loss": 0.7736,
+      "step": 1690
+    },
+    {
+      "epoch": 0.6545384541341804,
+      "grad_norm": 0.21457648277282715,
+      "learning_rate": 2.821355873124503e-06,
+      "loss": 0.7708,
+      "step": 1700
+    },
+    {
+      "epoch": 0.6583886803349697,
+      "grad_norm": 0.24571336805820465,
+      "learning_rate": 2.7653997566960144e-06,
+      "loss": 0.7868,
+      "step": 1710
+    },
+    {
+      "epoch": 0.662238906535759,
+      "grad_norm": 0.24042902886867523,
+      "learning_rate": 2.709791206537309e-06,
+      "loss": 0.7755,
+      "step": 1720
+    },
+    {
+      "epoch": 0.6660891327365482,
+      "grad_norm": 0.2489984929561615,
+      "learning_rate": 2.6545388719163747e-06,
+      "loss": 0.7807,
+      "step": 1730
+    },
+    {
+      "epoch": 0.6699393589373376,
+      "grad_norm": 0.22991426289081573,
+      "learning_rate": 2.599651346695979e-06,
+      "loss": 0.7796,
+      "step": 1740
+    },
+    {
+      "epoch": 0.6737895851381268,
+      "grad_norm": 0.23901109397411346,
+      "learning_rate": 2.545137167997007e-06,
+      "loss": 0.7755,
+      "step": 1750
+    },
+    {
+      "epoch": 0.6776398113389162,
+      "grad_norm": 0.2259925752878189,
+      "learning_rate": 2.4910048148705972e-06,
+      "loss": 0.7632,
+      "step": 1760
+    },
+    {
+      "epoch": 0.6814900375397055,
+      "grad_norm": 0.2568780183792114,
+      "learning_rate": 2.4372627069793327e-06,
+      "loss": 0.8078,
+      "step": 1770
+    },
+    {
+      "epoch": 0.6853402637404947,
+      "grad_norm": 0.2356659471988678,
+      "learning_rate": 2.3839192032876486e-06,
+      "loss": 0.7932,
+      "step": 1780
+    },
+    {
+      "epoch": 0.6891904899412841,
+      "grad_norm": 0.22632069885730743,
+      "learning_rate": 2.3309826007617047e-06,
+      "loss": 0.804,
+      "step": 1790
+    },
+    {
+      "epoch": 0.6930407161420733,
+      "grad_norm": 0.2328559309244156,
+      "learning_rate": 2.278461133078881e-06,
+      "loss": 0.7679,
+      "step": 1800
+    },
+    {
+      "epoch": 0.6968909423428626,
+      "grad_norm": 0.23814328014850616,
+      "learning_rate": 2.2263629693471258e-06,
+      "loss": 0.7592,
+      "step": 1810
+    },
+    {
+      "epoch": 0.700741168543652,
+      "grad_norm": 0.2266077995300293,
+      "learning_rate": 2.1746962128343414e-06,
+      "loss": 0.7826,
+      "step": 1820
+    },
+    {
+      "epoch": 0.7045913947444412,
+      "grad_norm": 0.229709193110466,
+      "learning_rate": 2.1234688997080183e-06,
+      "loss": 0.7724,
+      "step": 1830
+    },
+    {
+      "epoch": 0.7084416209452306,
+      "grad_norm": 0.24625201523303986,
+      "learning_rate": 2.0726889977853053e-06,
+      "loss": 0.7769,
+      "step": 1840
+    },
+    {
+      "epoch": 0.7122918471460198,
+      "grad_norm": 0.2473742961883545,
+      "learning_rate": 2.022364405293703e-06,
+      "loss": 0.7899,
+      "step": 1850
+    },
+    {
+      "epoch": 0.7161420733468091,
+      "grad_norm": 0.23134638369083405,
+      "learning_rate": 1.9725029496425814e-06,
+      "loss": 0.7815,
+      "step": 1860
+    },
+    {
+      "epoch": 0.7199922995475985,
+      "grad_norm": 0.24431107938289642,
+      "learning_rate": 1.923112386205727e-06,
+      "loss": 0.7789,
+      "step": 1870
+    },
+    {
+      "epoch": 0.7238425257483877,
+      "grad_norm": 0.22074857354164124,
+      "learning_rate": 1.874200397115079e-06,
+      "loss": 0.7861,
+      "step": 1880
+    },
+    {
+      "epoch": 0.727692751949177,
+      "grad_norm": 0.2445300966501236,
+      "learning_rate": 1.825774590065857e-06,
+      "loss": 0.7707,
+      "step": 1890
+    },
+    {
+      "epoch": 0.7315429781499663,
+      "grad_norm": 0.24017222225666046,
+      "learning_rate": 1.7778424971332813e-06,
+      "loss": 0.7775,
+      "step": 1900
+    },
+    {
+      "epoch": 0.7353932043507556,
+      "grad_norm": 0.23203350603580475,
+      "learning_rate": 1.7304115736010407e-06,
+      "loss": 0.7753,
+      "step": 1910
+    },
+    {
+      "epoch": 0.739243430551545,
+      "grad_norm": 0.2249097228050232,
+      "learning_rate": 1.6834891968017108e-06,
+      "loss": 0.7696,
+      "step": 1920
+    },
+    {
+      "epoch": 0.7430936567523342,
+      "grad_norm": 0.24349232017993927,
+      "learning_rate": 1.6370826649692884e-06,
+      "loss": 0.7863,
+      "step": 1930
+    },
+    {
+      "epoch": 0.7469438829531235,
+      "grad_norm": 0.24131162464618683,
+      "learning_rate": 1.5911991961040463e-06,
+      "loss": 0.7948,
+      "step": 1940
+    },
+    {
+      "epoch": 0.7507941091539128,
+      "grad_norm": 0.2385329008102417,
+      "learning_rate": 1.545845926849851e-06,
+      "loss": 0.7827,
+      "step": 1950
+    },
+    {
+      "epoch": 0.7546443353547021,
+      "grad_norm": 0.24677498638629913,
+      "learning_rate": 1.5010299113841397e-06,
+      "loss": 0.7986,
+      "step": 1960
+    },
+    {
+      "epoch": 0.7584945615554913,
+      "grad_norm": 0.23505358397960663,
+      "learning_rate": 1.4567581203207238e-06,
+      "loss": 0.7897,
+      "step": 1970
+    },
+    {
+      "epoch": 0.7623447877562807,
+      "grad_norm": 0.2415883094072342,
+      "learning_rate": 1.413037439625593e-06,
+      "loss": 0.786,
+      "step": 1980
+    },
+    {
+      "epoch": 0.76619501395707,
+      "grad_norm": 0.24950726330280304,
+      "learning_rate": 1.3698746695458847e-06,
+      "loss": 0.7821,
+      "step": 1990
+    },
+    {
+      "epoch": 0.7700452401578592,
+      "grad_norm": 0.2284851372241974,
+      "learning_rate": 1.3272765235521745e-06,
+      "loss": 0.7728,
+      "step": 2000
+    },
+    {
+      "epoch": 0.7738954663586486,
+      "grad_norm": 0.23085665702819824,
+      "learning_rate": 1.2852496272942877e-06,
+      "loss": 0.7806,
+      "step": 2010
+    },
+    {
+      "epoch": 0.7777456925594378,
+      "grad_norm": 0.22835133969783783,
+      "learning_rate": 1.2438005175707396e-06,
+      "loss": 0.769,
+      "step": 2020
+    },
+    {
+      "epoch": 0.7815959187602272,
+      "grad_norm": 0.2362927943468094,
+      "learning_rate": 1.2029356413120297e-06,
+      "loss": 0.7533,
+      "step": 2030
+    },
+    {
+      "epoch": 0.7854461449610165,
+      "grad_norm": 0.23217064142227173,
+      "learning_rate": 1.1626613545778786e-06,
+      "loss": 0.7715,
+      "step": 2040
+    },
+    {
+      "epoch": 0.7892963711618057,
+      "grad_norm": 0.22042995691299438,
+      "learning_rate": 1.1229839215686294e-06,
+      "loss": 0.7762,
+      "step": 2050
+    },
+    {
+      "epoch": 0.7931465973625951,
+      "grad_norm": 0.22093623876571655,
+      "learning_rate": 1.0839095136509188e-06,
+      "loss": 0.7599,
+      "step": 2060
+    },
+    {
+      "epoch": 0.7969968235633843,
+      "grad_norm": 0.24193589389324188,
+      "learning_rate": 1.045444208397791e-06,
+      "loss": 0.7818,
+      "step": 2070
+    },
+    {
+      "epoch": 0.8008470497641736,
+      "grad_norm": 0.23195093870162964,
+      "learning_rate": 1.007593988643402e-06,
+      "loss": 0.7637,
+      "step": 2080
+    },
+    {
+      "epoch": 0.804697275964963,
+      "grad_norm": 0.23101268708705902,
+      "learning_rate": 9.703647415524613e-07,
+      "loss": 0.7852,
+      "step": 2090
+    },
+    {
+      "epoch": 0.8085475021657522,
+      "grad_norm": 0.23141689598560333,
+      "learning_rate": 9.337622577045485e-07,
+      "loss": 0.7738,
+      "step": 2100
+    },
+    {
+      "epoch": 0.8123977283665416,
+      "grad_norm": 0.2301773577928543,
+      "learning_rate": 8.977922301934588e-07,
+      "loss": 0.7841,
+      "step": 2110
+    },
+    {
+      "epoch": 0.8162479545673308,
+      "grad_norm": 0.23733402788639069,
+      "learning_rate": 8.624602537417009e-07,
+      "loss": 0.7856,
+      "step": 2120
+    },
+    {
+      "epoch": 0.8200981807681201,
+      "grad_norm": 0.24629302322864532,
+      "learning_rate": 8.277718238303123e-07,
+      "loss": 0.7742,
+      "step": 2130
+    },
+    {
+      "epoch": 0.8239484069689095,
+      "grad_norm": 0.23207615315914154,
+      "learning_rate": 7.937323358440935e-07,
+      "loss": 0.7761,
+      "step": 2140
+    },
+    {
+      "epoch": 0.8277986331696987,
+      "grad_norm": 0.2413422018289566,
+      "learning_rate": 7.603470842324223e-07,
+      "loss": 0.7687,
+      "step": 2150
+    },
+    {
+      "epoch": 0.831648859370488,
+      "grad_norm": 0.22652852535247803,
+      "learning_rate": 7.276212616857575e-07,
+      "loss": 0.7852,
+      "step": 2160
+    },
+    {
+      "epoch": 0.8354990855712773,
+      "grad_norm": 0.2251061201095581,
+      "learning_rate": 6.955599583279837e-07,
+      "loss": 0.784,
+      "step": 2170
+    },
+    {
+      "epoch": 0.8393493117720666,
+      "grad_norm": 0.24336068332195282,
+      "learning_rate": 6.641681609246981e-07,
+      "loss": 0.7767,
+      "step": 2180
+    },
+    {
+      "epoch": 0.8431995379728559,
+      "grad_norm": 0.22582781314849854,
+      "learning_rate": 6.334507521075711e-07,
+      "loss": 0.7831,
+      "step": 2190
+    },
+    {
+      "epoch": 0.8470497641736452,
+      "grad_norm": 0.2513900399208069,
+      "learning_rate": 6.034125096149179e-07,
+      "loss": 0.7777,
+      "step": 2200
+    },
+    {
+      "epoch": 0.8508999903744345,
+      "grad_norm": 0.23998549580574036,
+      "learning_rate": 5.740581055485722e-07,
+      "loss": 0.7501,
+      "step": 2210
+    },
+    {
+      "epoch": 0.8547502165752238,
+      "grad_norm": 0.22569261491298676,
+      "learning_rate": 5.453921056471961e-07,
+      "loss": 0.7839,
+      "step": 2220
+    },
+    {
+      "epoch": 0.8586004427760131,
+      "grad_norm": 0.23608243465423584,
+      "learning_rate": 5.174189685761288e-07,
+      "loss": 0.7777,
+      "step": 2230
+    },
+    {
+      "epoch": 0.8624506689768023,
+      "grad_norm": 0.22512704133987427,
+      "learning_rate": 4.901430452338979e-07,
+      "loss": 0.7894,
+      "step": 2240
+    },
+    {
+      "epoch": 0.8663008951775917,
+      "grad_norm": 0.23011514544487,
+      "learning_rate": 4.6356857807548286e-07,
+      "loss": 0.7718,
+      "step": 2250
+    },
+    {
+      "epoch": 0.870151121378381,
+      "grad_norm": 0.2433038055896759,
+      "learning_rate": 4.376997004524541e-07,
+      "loss": 0.7614,
+      "step": 2260
+    },
+    {
+      "epoch": 0.8740013475791703,
+      "grad_norm": 0.23164549469947815,
+      "learning_rate": 4.12540435970073e-07,
+      "loss": 0.7708,
+      "step": 2270
+    },
+    {
+      "epoch": 0.8778515737799596,
+      "grad_norm": 0.22599858045578003,
+      "learning_rate": 3.880946978614703e-07,
+      "loss": 0.7831,
+      "step": 2280
+    },
+    {
+      "epoch": 0.8817017999807488,
+      "grad_norm": 0.22510623931884766,
+      "learning_rate": 3.643662883789878e-07,
+      "loss": 0.7869,
+      "step": 2290
+    },
+    {
+      "epoch": 0.8855520261815382,
+      "grad_norm": 0.22136008739471436,
+      "learning_rate": 3.41358898202781e-07,
+      "loss": 0.7639,
+      "step": 2300
+    },
+    {
+      "epoch": 0.8894022523823275,
+      "grad_norm": 0.21951663494110107,
+      "learning_rate": 3.190761058667741e-07,
+      "loss": 0.774,
+      "step": 2310
+    },
+    {
+      "epoch": 0.8932524785831167,
+      "grad_norm": 0.2309655249118805,
+      "learning_rate": 2.9752137720206463e-07,
+      "loss": 0.7799,
+      "step": 2320
+    },
+    {
+      "epoch": 0.8971027047839061,
+      "grad_norm": 0.23155641555786133,
+      "learning_rate": 2.7669806479785067e-07,
+      "loss": 0.7852,
+      "step": 2330
+    },
+    {
+      "epoch": 0.9009529309846953,
+      "grad_norm": 0.24085073173046112,
+      "learning_rate": 2.566094074799763e-07,
+      "loss": 0.7796,
+      "step": 2340
+    },
+    {
+      "epoch": 0.9048031571854847,
+      "grad_norm": 0.2275749295949936,
+      "learning_rate": 2.3725852980716557e-07,
+      "loss": 0.7616,
+      "step": 2350
+    },
+    {
+      "epoch": 0.908653383386274,
+      "grad_norm": 0.22886724770069122,
+      "learning_rate": 2.1864844158503983e-07,
+      "loss": 0.7745,
+      "step": 2360
+    },
+    {
+      "epoch": 0.9125036095870632,
+      "grad_norm": 0.2575795352458954,
+      "learning_rate": 2.007820373979702e-07,
+      "loss": 0.7673,
+      "step": 2370
+    },
+    {
+      "epoch": 0.9163538357878526,
+      "grad_norm": 0.23114149272441864,
+      "learning_rate": 1.8366209615886532e-07,
+      "loss": 0.7752,
+      "step": 2380
+    },
+    {
+      "epoch": 0.9202040619886418,
+      "grad_norm": 0.22906091809272766,
+      "learning_rate": 1.6729128067693613e-07,
+      "loss": 0.7874,
+      "step": 2390
+    },
+    {
+      "epoch": 0.9240542881894311,
+      "grad_norm": 0.2446931153535843,
+      "learning_rate": 1.5167213724353426e-07,
+      "loss": 0.7878,
+      "step": 2400
+    },
+    {
+      "epoch": 0.9279045143902205,
+      "grad_norm": 0.22670775651931763,
+      "learning_rate": 1.3680709523610058e-07,
+      "loss": 0.7705,
+      "step": 2410
+    },
+    {
+      "epoch": 0.9317547405910097,
+      "grad_norm": 0.22094565629959106,
+      "learning_rate": 1.2269846674030418e-07,
+      "loss": 0.7914,
+      "step": 2420
+    },
+    {
+      "epoch": 0.935604966791799,
+      "grad_norm": 0.23575392365455627,
+      "learning_rate": 1.0934844619042595e-07,
+      "loss": 0.7899,
+      "step": 2430
+    },
+    {
+      "epoch": 0.9394551929925883,
+      "grad_norm": 0.24262778460979462,
+      "learning_rate": 9.675911002803961e-08,
+      "loss": 0.7767,
+      "step": 2440
+    },
+    {
+      "epoch": 0.9433054191933776,
+      "grad_norm": 0.24935580790042877,
+      "learning_rate": 8.493241637904381e-08,
+      "loss": 0.7739,
+      "step": 2450
+    },
+    {
+      "epoch": 0.9471556453941669,
+      "grad_norm": 0.23954641819000244,
+      "learning_rate": 7.387020474909689e-08,
+      "loss": 0.7949,
+      "step": 2460
+    },
+    {
+      "epoch": 0.9510058715949562,
+      "grad_norm": 0.2483944445848465,
+      "learning_rate": 6.357419573750856e-08,
+      "loss": 0.788,
+      "step": 2470
+    },
+    {
+      "epoch": 0.9548560977957455,
+      "grad_norm": 0.23027725517749786,
+      "learning_rate": 5.4045990769613365e-08,
+      "loss": 0.7714,
+      "step": 2480
+    },
+    {
+      "epoch": 0.9587063239965348,
+      "grad_norm": 0.2882312536239624,
+      "learning_rate": 4.5287071847694434e-08,
+      "loss": 0.7754,
+      "step": 2490
+    },
+    {
+      "epoch": 0.9625565501973241,
+      "grad_norm": 0.22553925216197968,
+      "learning_rate": 3.7298801320467817e-08,
+      "loss": 0.7716,
+      "step": 2500
+    },
+    {
+      "epoch": 0.9664067763981133,
+      "grad_norm": 0.2308730036020279,
+      "learning_rate": 3.0082421671192576e-08,
+      "loss": 0.7648,
+      "step": 2510
+    },
+    {
+      "epoch": 0.9702570025989027,
+      "grad_norm": 0.23178935050964355,
+      "learning_rate": 2.3639055324410355e-08,
+      "loss": 0.7879,
+      "step": 2520
+    },
+    {
+      "epoch": 0.974107228799692,
+      "grad_norm": 0.23110386729240417,
+      "learning_rate": 1.7969704471369477e-08,
+      "loss": 0.7702,
+      "step": 2530
+    },
+    {
+      "epoch": 0.9779574550004813,
+      "grad_norm": 0.22606335580348969,
+      "learning_rate": 1.3075250914142413e-08,
+      "loss": 0.7741,
+      "step": 2540
+    },
+    {
+      "epoch": 0.9818076812012706,
+      "grad_norm": 0.22583748400211334,
+      "learning_rate": 8.956455928476049e-09,
+      "loss": 0.7824,
+      "step": 2550
+    },
+    {
+      "epoch": 0.9856579074020598,
+      "grad_norm": 0.23683346807956696,
+      "learning_rate": 5.6139601453786275e-09,
+      "loss": 0.7734,
+      "step": 2560
+    },
+    {
+      "epoch": 0.9895081336028492,
+      "grad_norm": 0.2346523255109787,
+      "learning_rate": 3.0482834514805602e-09,
+      "loss": 0.7633,
+      "step": 2570
+    },
+    {
+      "epoch": 0.9933583598036385,
+      "grad_norm": 0.20977047085762024,
+      "learning_rate": 1.2598249081713366e-09,
+      "loss": 0.7581,
+      "step": 2580
+    },
+    {
+      "epoch": 0.9972085860044277,
+      "grad_norm": 0.26096582412719727,
+      "learning_rate": 2.488626895302826e-10,
+      "loss": 0.7892,
+      "step": 2590
+    },
+    {
+      "epoch": 0.9999037443449803,
+      "step": 2597,
+      "total_flos": 3.652195211100029e+18,
+      "train_loss": 0.8070575117743561,
+      "train_runtime": 10325.4226,
+      "train_samples_per_second": 64.395,
+      "train_steps_per_second": 0.252
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 2597,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.652195211100029e+18,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}

continuous/qwen3/mc2_m16_224_mt5/finetune/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:90749a38f59248eadd6cf9c6b3d921ebd8c86c55d93d7243e511dce883fe1b6b
+size 6136

continuous/qwen3/mc2_m16_224_mt5/pretrain/config.json

@@ -0,0 +1,60 @@
+{
+  "_attn_implementation_autoset": true,
+  "add_faster_video": false,
+  "add_time_instruction": false,
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "eos_token_id": 151645,
+  "faster_token_stride": 10,
+  "force_sample": false,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 2048,
+  "image_aspect_ratio": "square",
+  "image_crop_resolution": null,
+  "image_grid_pinpoints": null,
+  "image_split_resolution": null,
+  "initializer_range": 0.02,
+  "intermediate_size": 6144,
+  "max_position_embeddings": 40960,
+  "max_window_layers": 28,
+  "mm_hidden_size": 512,
+  "mm_newline_position": "grid",
+  "mm_patch_merge_type": "flat",
+  "mm_projector_lr": null,
+  "mm_projector_type": "mlp2x_gelu",
+  "mm_resampler_type": null,
+  "mm_spatial_pool_mode": "bilinear",
+  "mm_spatial_pool_stride": null,
+  "mm_tunable_parts": "mm_mlp_adapter",
+  "mm_use_im_patch_token": false,
+  "mm_use_im_start_end": false,
+  "mm_vision_select_feature": "patch",
+  "mm_vision_select_layer": -2,
+  "mm_vision_tower": "open_clip_hub:ViT-M-16",
+  "mm_vision_tower_lr": null,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pos_skipping_range": 4096,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000,
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "tokenizer_model_max_length": 8192,
+  "tokenizer_padding_side": "right",
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.51.3",
+  "use_cache": true,
+  "use_mm_proj": true,
+  "use_pos_skipping": false,
+  "use_sliding_window": false,
+  "vision_tower_pretrained": "/cache/ckpt/download/tokenizer/continuous/mc2/mc2_m16_224_mt5.pt",
+  "vocab_size": 151936
+}

continuous/qwen3/mc2_m16_224_mt5/pretrain/mm_projector.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b4de9a60c6bd92811c9e99d7cfe14d919f80f84b657856c097977f8747d087d4
+size 20990008

continuous/qwen3/mc2_m16_224_mt5/pretrain/trainer_state.json

@@ -0,0 +1,421 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 545,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01834862385321101,
+      "grad_norm": 10.316337585449219,
+      "learning_rate": 0.00032142857142857147,
+      "loss": 9.0662,
+      "step": 10
+    },
+    {
+      "epoch": 0.03669724770642202,
+      "grad_norm": 2.140162229537964,
+      "learning_rate": 0.0006785714285714287,
+      "loss": 6.0012,
+      "step": 20
+    },
+    {
+      "epoch": 0.05504587155963303,
+      "grad_norm": 0.5349360108375549,
+      "learning_rate": 0.0009999907688176173,
+      "loss": 5.1116,
+      "step": 30
+    },
+    {
+      "epoch": 0.07339449541284404,
+      "grad_norm": 0.2544049322605133,
+      "learning_rate": 0.0009988834393115766,
+      "loss": 4.7906,
+      "step": 40
+    },
+    {
+      "epoch": 0.09174311926605505,
+      "grad_norm": 0.24782566726207733,
+      "learning_rate": 0.0009959345572940772,
+      "loss": 4.6802,
+      "step": 50
+    },
+    {
+      "epoch": 0.11009174311926606,
+      "grad_norm": 0.2708393633365631,
+      "learning_rate": 0.0009911550081155983,
+      "loss": 4.6151,
+      "step": 60
+    },
+    {
+      "epoch": 0.12844036697247707,
+      "grad_norm": 0.23898950219154358,
+      "learning_rate": 0.0009845624347566433,
+      "loss": 4.5487,
+      "step": 70
+    },
+    {
+      "epoch": 0.14678899082568808,
+      "grad_norm": 0.2382814735174179,
+      "learning_rate": 0.0009761811727013547,
+      "loss": 4.5149,
+      "step": 80
+    },
+    {
+      "epoch": 0.1651376146788991,
+      "grad_norm": 0.20825235545635223,
+      "learning_rate": 0.0009660421601067666,
+      "loss": 4.4724,
+      "step": 90
+    },
+    {
+      "epoch": 0.1834862385321101,
+      "grad_norm": 0.1908266693353653,
+      "learning_rate": 0.0009541828235992926,
+      "loss": 4.4479,
+      "step": 100
+    },
+    {
+      "epoch": 0.2018348623853211,
+      "grad_norm": 0.19392043352127075,
+      "learning_rate": 0.0009406469401200151,
+      "loss": 4.4134,
+      "step": 110
+    },
+    {
+      "epoch": 0.22018348623853212,
+      "grad_norm": 0.23125256597995758,
+      "learning_rate": 0.0009254844753287492,
+      "loss": 4.3995,
+      "step": 120
+    },
+    {
+      "epoch": 0.23853211009174313,
+      "grad_norm": 0.1960584819316864,
+      "learning_rate": 0.0009087513991633924,
+      "loss": 4.3708,
+      "step": 130
+    },
+    {
+      "epoch": 0.25688073394495414,
+      "grad_norm": 0.2294335663318634,
+      "learning_rate": 0.0008905094792353917,
+      "loss": 4.3494,
+      "step": 140
+    },
+    {
+      "epoch": 0.27522935779816515,
+      "grad_norm": 0.20732808113098145,
+      "learning_rate": 0.0008708260528239789,
+      "loss": 4.3455,
+      "step": 150
+    },
+    {
+      "epoch": 0.29357798165137616,
+      "grad_norm": 0.2399912327528,
+      "learning_rate": 0.0008497737783108237,
+      "loss": 4.3272,
+      "step": 160
+    },
+    {
+      "epoch": 0.3119266055045872,
+      "grad_norm": 0.22740620374679565,
+      "learning_rate": 0.0008274303669726426,
+      "loss": 4.3125,
+      "step": 170
+    },
+    {
+      "epoch": 0.3302752293577982,
+      "grad_norm": 0.22209100425243378,
+      "learning_rate": 0.0008038782961218137,
+      "loss": 4.2984,
+      "step": 180
+    },
+    {
+      "epoch": 0.3486238532110092,
+      "grad_norm": 0.28716766834259033,
+      "learning_rate": 0.000779204504653894,
+      "loss": 4.2902,
+      "step": 190
+    },
+    {
+      "epoch": 0.3669724770642202,
+      "grad_norm": 0.1883043795824051,
+      "learning_rate": 0.0007535000721258791,
+      "loss": 4.2684,
+      "step": 200
+    },
+    {
+      "epoch": 0.3853211009174312,
+      "grad_norm": 0.2107715606689453,
+      "learning_rate": 0.00072685988254984,
+      "loss": 4.2605,
+      "step": 210
+    },
+    {
+      "epoch": 0.4036697247706422,
+      "grad_norm": 0.23885343968868256,
+      "learning_rate": 0.0006993822741429906,
+      "loss": 4.2736,
+      "step": 220
+    },
+    {
+      "epoch": 0.42201834862385323,
+      "grad_norm": 0.20772460103034973,
+      "learning_rate": 0.0006711686763270818,
+      "loss": 4.2674,
+      "step": 230
+    },
+    {
+      "epoch": 0.44036697247706424,
+      "grad_norm": 0.19894476234912872,
+      "learning_rate": 0.0006423232353170831,
+      "loss": 4.2512,
+      "step": 240
+    },
+    {
+      "epoch": 0.45871559633027525,
+      "grad_norm": 0.20970451831817627,
+      "learning_rate": 0.0006129524296812335,
+      "loss": 4.2309,
+      "step": 250
+    },
+    {
+      "epoch": 0.47706422018348627,
+      "grad_norm": 0.19239820539951324,
+      "learning_rate": 0.0005831646772915651,
+      "loss": 4.2384,
+      "step": 260
+    },
+    {
+      "epoch": 0.4954128440366973,
+      "grad_norm": 0.31041330099105835,
+      "learning_rate": 0.0005530699351157809,
+      "loss": 4.2125,
+      "step": 270
+    },
+    {
+      "epoch": 0.5137614678899083,
+      "grad_norm": 0.22544287145137787,
+      "learning_rate": 0.0005227792933277943,
+      "loss": 4.1992,
+      "step": 280
+    },
+    {
+      "epoch": 0.5321100917431193,
+      "grad_norm": 0.22341442108154297,
+      "learning_rate": 0.0004924045652352092,
+      "loss": 4.174,
+      "step": 290
+    },
+    {
+      "epoch": 0.5504587155963303,
+      "grad_norm": 0.22654026746749878,
+      "learning_rate": 0.0004620578745374633,
+      "loss": 4.1616,
+      "step": 300
+    },
+    {
+      "epoch": 0.5688073394495413,
+      "grad_norm": 0.2821761667728424,
+      "learning_rate": 0.0004318512414382059,
+      "loss": 4.1678,
+      "step": 310
+    },
+    {
+      "epoch": 0.5871559633027523,
+      "grad_norm": 0.21506866812705994,
+      "learning_rate": 0.0004018961691397148,
+      "loss": 4.1502,
+      "step": 320
+    },
+    {
+      "epoch": 0.6055045871559633,
+      "grad_norm": 0.2050533890724182,
+      "learning_rate": 0.0003723032322457458,
+      "loss": 4.1294,
+      "step": 330
+    },
+    {
+      "epoch": 0.6238532110091743,
+      "grad_norm": 0.21028739213943481,
+      "learning_rate": 0.0003431816685921625,
+      "loss": 4.122,
+      "step": 340
+    },
+    {
+      "epoch": 0.6422018348623854,
+      "grad_norm": 0.2093348205089569,
+      "learning_rate": 0.0003146389760120416,
+      "loss": 4.1211,
+      "step": 350
+    },
+    {
+      "epoch": 0.6605504587155964,
+      "grad_norm": 0.2153186947107315,
+      "learning_rate": 0.00028678051552373485,
+      "loss": 4.1219,
+      "step": 360
+    },
+    {
+      "epoch": 0.6788990825688074,
+      "grad_norm": 0.20171812176704407,
+      "learning_rate": 0.0002597091224066581,
+      "loss": 4.1027,
+      "step": 370
+    },
+    {
+      "epoch": 0.6972477064220184,
+      "grad_norm": 0.22647275030612946,
+      "learning_rate": 0.0002335247266004629,
+      "loss": 4.1161,
+      "step": 380
+    },
+    {
+      "epoch": 0.7155963302752294,
+      "grad_norm": 0.21645371615886688,
+      "learning_rate": 0.000208323983828828,
+      "loss": 4.0978,
+      "step": 390
+    },
+    {
+      "epoch": 0.7339449541284404,
+      "grad_norm": 0.20616964995861053,
+      "learning_rate": 0.0001841999188095224,
+      "loss": 4.0934,
+      "step": 400
+    },
+    {
+      "epoch": 0.7522935779816514,
+      "grad_norm": 0.20119550824165344,
+      "learning_rate": 0.00016124158186777676,
+      "loss": 4.0951,
+      "step": 410
+    },
+    {
+      "epoch": 0.7706422018348624,
+      "grad_norm": 0.2039160430431366,
+      "learning_rate": 0.0001395337202205211,
+      "loss": 4.0874,
+      "step": 420
+    },
+    {
+      "epoch": 0.7889908256880734,
+      "grad_norm": 0.20110921561717987,
+      "learning_rate": 0.00011915646514489292,
+      "loss": 4.0872,
+      "step": 430
+    },
+    {
+      "epoch": 0.8073394495412844,
+      "grad_norm": 0.2009619176387787,
+      "learning_rate": 0.00010018503618578818,
+      "loss": 4.0936,
+      "step": 440
+    },
+    {
+      "epoch": 0.8256880733944955,
+      "grad_norm": 0.18924878537654877,
+      "learning_rate": 8.268946349432583e-05,
+      "loss": 4.0852,
+      "step": 450
+    },
+    {
+      "epoch": 0.8440366972477065,
+      "grad_norm": 0.20672276616096497,
+      "learning_rate": 6.673432932217117e-05,
+      "loss": 4.0777,
+      "step": 460
+    },
+    {
+      "epoch": 0.8623853211009175,
+      "grad_norm": 0.1975499987602234,
+      "learning_rate": 5.2378529625954684e-05,
+      "loss": 4.0648,
+      "step": 470
+    },
+    {
+      "epoch": 0.8807339449541285,
+      "grad_norm": 0.2015974074602127,
+      "learning_rate": 3.9675056661785556e-05,
+      "loss": 4.0716,
+      "step": 480
+    },
+    {
+      "epoch": 0.8990825688073395,
+      "grad_norm": 0.18186846375465393,
+      "learning_rate": 2.8670803372379538e-05,
+      "loss": 4.0743,
+      "step": 490
+    },
+    {
+      "epoch": 0.9174311926605505,
+      "grad_norm": 0.1904570758342743,
+      "learning_rate": 1.9406390288876586e-05,
+      "loss": 4.0857,
+      "step": 500
+    },
+    {
+      "epoch": 0.9357798165137615,
+      "grad_norm": 0.18695923686027527,
+      "learning_rate": 1.191601558631461e-05,
+      "loss": 4.0755,
+      "step": 510
+    },
+    {
+      "epoch": 0.9541284403669725,
+      "grad_norm": 0.1921919882297516,
+      "learning_rate": 6.2273288462559305e-06,
+      "loss": 4.0716,
+      "step": 520
+    },
+    {
+      "epoch": 0.9724770642201835,
+      "grad_norm": 0.1934342235326767,
+      "learning_rate": 2.361328992552314e-06,
+      "loss": 4.0737,
+      "step": 530
+    },
+    {
+      "epoch": 0.9908256880733946,
+      "grad_norm": 0.1876281499862671,
+      "learning_rate": 3.322867770044202e-07,
+      "loss": 4.0718,
+      "step": 540
+    },
+    {
+      "epoch": 1.0,
+      "step": 545,
+      "total_flos": 2.998547778843443e+17,
+      "train_loss": 4.372127798937877,
+      "train_runtime": 5668.2519,
+      "train_samples_per_second": 98.466,
+      "train_steps_per_second": 0.096
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 545,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": false,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.998547778843443e+17,
+  "train_batch_size": 128,
+  "trial_name": null,
+  "trial_params": null
+}

continuous/qwen3/mc2_m16_384/finetune/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
+}

continuous/qwen3/mc2_m16_384/finetune/config.json

@@ -0,0 +1,59 @@
+{
+  "add_faster_video": false,
+  "add_time_instruction": false,
+  "architectures": [
+    "LlavaQwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "eos_token_id": 151645,
+  "faster_token_stride": 10,
+  "force_sample": false,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 2048,
+  "image_aspect_ratio": "square",
+  "image_crop_resolution": null,
+  "image_grid_pinpoints": null,
+  "image_split_resolution": null,
+  "initializer_range": 0.02,
+  "intermediate_size": 6144,
+  "max_position_embeddings": 40960,
+  "max_window_layers": 28,
+  "mm_hidden_size": 512,
+  "mm_newline_position": "grid",
+  "mm_patch_merge_type": "flat",
+  "mm_projector_lr": 0.0001,
+  "mm_projector_type": "mlp2x_gelu",
+  "mm_resampler_type": null,
+  "mm_spatial_pool_mode": "bilinear",
+  "mm_spatial_pool_stride": null,
+  "mm_tunable_parts": "mm_mlp_adapter,mm_language_model",
+  "mm_use_im_patch_token": false,
+  "mm_use_im_start_end": false,
+  "mm_vision_select_feature": "patch",
+  "mm_vision_select_layer": -2,
+  "mm_vision_tower": "open_clip_hub:ViT-M-16",
+  "mm_vision_tower_lr": null,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pos_skipping_range": 4096,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000,
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "tokenizer_model_max_length": 8192,
+  "tokenizer_padding_side": "right",
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.51.3",
+  "use_cache": true,
+  "use_mm_proj": true,
+  "use_pos_skipping": false,
+  "use_sliding_window": false,
+  "vision_tower_pretrained": "/cache/ckpt/download/tokenizer/continuous/mc2/mc2_m16_384.pt",
+  "vocab_size": 151936
+}

continuous/qwen3/mc2_m16_384/finetune/generation_config.json

@@ -0,0 +1,13 @@
+{
+  "bos_token_id": 151643,
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "pad_token_id": 151643,
+  "temperature": 0.6,
+  "top_k": 20,
+  "top_p": 0.95,
+  "transformers_version": "4.51.3"
+}

continuous/qwen3/mc2_m16_384/finetune/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8f6f31b137b425699a7a8b59ed4a71a1924bb8bc9a607904d5e1dd83a8990c53
+size 4162089720

continuous/qwen3/mc2_m16_384/finetune/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
+  }
+}

continuous/qwen3/mc2_m16_384/finetune/tokenizer.json

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

continuous/qwen3/mc2_m16_384/finetune/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 message.content is string 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.content is string %}\n        {%- set content = message.content %}\n    {%- else %}\n        {%- set content = '' %}\n    {%- endif %}\n    {%- if (message.role == \"user\") or (message.role == \"system\" and not loop.first) %}\n        {{- '<|im_start|>' + message.role + '\\n' + content + '<|im_end|>' + '\\n' }}\n    {%- elif message.role == \"assistant\" %}\n        {%- set reasoning_content = '' %}\n        {%- if message.reasoning_content is string %}\n            {%- set reasoning_content = message.reasoning_content %}\n        {%- else %}\n            {%- if '</think>' in content %}\n                {%- set reasoning_content = content.split('</think>')[0].rstrip('\\n').split('<think>')[-1].lstrip('\\n') %}\n                {%- set content = content.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        {{- 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": 8192,
+  "pad_token": "<|endoftext|>",
+  "padding_side": "right",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

continuous/qwen3/mc2_m16_384/finetune/trainer_state.json

@@ -0,0 +1,1856 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999037443449803,
+  "eval_steps": 500,
+  "global_step": 2597,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.003850226200789296,
+      "grad_norm": 10.426959991455078,
+      "learning_rate": 1.153846153846154e-06,
+      "loss": 2.4864,
+      "step": 10
+    },
+    {
+      "epoch": 0.007700452401578592,
+      "grad_norm": 3.5048065185546875,
+      "learning_rate": 2.435897435897436e-06,
+      "loss": 1.9349,
+      "step": 20
+    },
+    {
+      "epoch": 0.01155067860236789,
+      "grad_norm": 2.362255334854126,
+      "learning_rate": 3.7179487179487184e-06,
+      "loss": 1.486,
+      "step": 30
+    },
+    {
+      "epoch": 0.015400904803157185,
+      "grad_norm": 0.5864666700363159,
+      "learning_rate": 5e-06,
+      "loss": 1.1338,
+      "step": 40
+    },
+    {
+      "epoch": 0.019251131003946483,
+      "grad_norm": 0.47236353158950806,
+      "learning_rate": 6.282051282051282e-06,
+      "loss": 1.0315,
+      "step": 50
+    },
+    {
+      "epoch": 0.02310135720473578,
+      "grad_norm": 0.2322143167257309,
+      "learning_rate": 7.564102564102564e-06,
+      "loss": 0.9523,
+      "step": 60
+    },
+    {
+      "epoch": 0.026951583405525073,
+      "grad_norm": 0.19395074248313904,
+      "learning_rate": 8.846153846153847e-06,
+      "loss": 0.9109,
+      "step": 70
+    },
+    {
+      "epoch": 0.03080180960631437,
+      "grad_norm": 0.1664077341556549,
+      "learning_rate": 9.999996111488723e-06,
+      "loss": 0.9101,
+      "step": 80
+    },
+    {
+      "epoch": 0.034652035807103666,
+      "grad_norm": 0.17666219174861908,
+      "learning_rate": 9.999529497453782e-06,
+      "loss": 0.8853,
+      "step": 90
+    },
+    {
+      "epoch": 0.038502262007892966,
+      "grad_norm": 0.1467515081167221,
+      "learning_rate": 9.998285264324193e-06,
+      "loss": 0.8595,
+      "step": 100
+    },
+    {
+      "epoch": 0.04235248820868226,
+      "grad_norm": 0.1578163504600525,
+      "learning_rate": 9.996263605626052e-06,
+      "loss": 0.8375,
+      "step": 110
+    },
+    {
+      "epoch": 0.04620271440947156,
+      "grad_norm": 0.14854232966899872,
+      "learning_rate": 9.99346483580503e-06,
+      "loss": 0.8458,
+      "step": 120
+    },
+    {
+      "epoch": 0.05005294061026085,
+      "grad_norm": 0.1380491554737091,
+      "learning_rate": 9.989889390177467e-06,
+      "loss": 0.8338,
+      "step": 130
+    },
+    {
+      "epoch": 0.053903166811050146,
+      "grad_norm": 0.16309493780136108,
+      "learning_rate": 9.985537824862647e-06,
+      "loss": 0.8336,
+      "step": 140
+    },
+    {
+      "epoch": 0.057753393011839446,
+      "grad_norm": 0.1726323813199997,
+      "learning_rate": 9.980410816696317e-06,
+      "loss": 0.8278,
+      "step": 150
+    },
+    {
+      "epoch": 0.06160361921262874,
+      "grad_norm": 0.15688160061836243,
+      "learning_rate": 9.97450916312541e-06,
+      "loss": 0.8248,
+      "step": 160
+    },
+    {
+      "epoch": 0.06545384541341803,
+      "grad_norm": 0.1321345567703247,
+      "learning_rate": 9.967833782084004e-06,
+      "loss": 0.8225,
+      "step": 170
+    },
+    {
+      "epoch": 0.06930407161420733,
+      "grad_norm": 0.14647983014583588,
+      "learning_rate": 9.960385711850554e-06,
+      "loss": 0.8444,
+      "step": 180
+    },
+    {
+      "epoch": 0.07315429781499663,
+      "grad_norm": 0.14560836553573608,
+      "learning_rate": 9.952166110886394e-06,
+      "loss": 0.8469,
+      "step": 190
+    },
+    {
+      "epoch": 0.07700452401578593,
+      "grad_norm": 0.15247638523578644,
+      "learning_rate": 9.943176257655567e-06,
+      "loss": 0.8212,
+      "step": 200
+    },
+    {
+      "epoch": 0.08085475021657522,
+      "grad_norm": 0.13554003834724426,
+      "learning_rate": 9.933417550425952e-06,
+      "loss": 0.8325,
+      "step": 210
+    },
+    {
+      "epoch": 0.08470497641736452,
+      "grad_norm": 0.14426331222057343,
+      "learning_rate": 9.9228915070518e-06,
+      "loss": 0.8067,
+      "step": 220
+    },
+    {
+      "epoch": 0.08855520261815382,
+      "grad_norm": 0.14931245148181915,
+      "learning_rate": 9.911599764737632e-06,
+      "loss": 0.8123,
+      "step": 230
+    },
+    {
+      "epoch": 0.09240542881894312,
+      "grad_norm": 0.14041395485401154,
+      "learning_rate": 9.899544079783606e-06,
+      "loss": 0.7987,
+      "step": 240
+    },
+    {
+      "epoch": 0.0962556550197324,
+      "grad_norm": 0.15251077711582184,
+      "learning_rate": 9.886726327312338e-06,
+      "loss": 0.8194,
+      "step": 250
+    },
+    {
+      "epoch": 0.1001058812205217,
+      "grad_norm": 0.13117128610610962,
+      "learning_rate": 9.87314850097724e-06,
+      "loss": 0.8197,
+      "step": 260
+    },
+    {
+      "epoch": 0.103956107421311,
+      "grad_norm": 0.13382817804813385,
+      "learning_rate": 9.858812712652452e-06,
+      "loss": 0.8074,
+      "step": 270
+    },
+    {
+      "epoch": 0.10780633362210029,
+      "grad_norm": 0.1412690430879593,
+      "learning_rate": 9.843721192104341e-06,
+      "loss": 0.8081,
+      "step": 280
+    },
+    {
+      "epoch": 0.11165655982288959,
+      "grad_norm": 0.13202448189258575,
+      "learning_rate": 9.827876286644698e-06,
+      "loss": 0.8219,
+      "step": 290
+    },
+    {
+      "epoch": 0.11550678602367889,
+      "grad_norm": 0.12715892493724823,
+      "learning_rate": 9.811280460765643e-06,
+      "loss": 0.7839,
+      "step": 300
+    },
+    {
+      "epoch": 0.11935701222446819,
+      "grad_norm": 0.1344890594482422,
+      "learning_rate": 9.793936295756292e-06,
+      "loss": 0.8005,
+      "step": 310
+    },
+    {
+      "epoch": 0.12320723842525748,
+      "grad_norm": 0.1415622979402542,
+      "learning_rate": 9.775846489301279e-06,
+      "loss": 0.7967,
+      "step": 320
+    },
+    {
+      "epoch": 0.12705746462604678,
+      "grad_norm": 0.13349561393260956,
+      "learning_rate": 9.757013855061157e-06,
+      "loss": 0.7989,
+      "step": 330
+    },
+    {
+      "epoch": 0.13090769082683607,
+      "grad_norm": 0.1327776163816452,
+      "learning_rate": 9.737441322234758e-06,
+      "loss": 0.8067,
+      "step": 340
+    },
+    {
+      "epoch": 0.13475791702762538,
+      "grad_norm": 0.14556820690631866,
+      "learning_rate": 9.717131935103605e-06,
+      "loss": 0.8143,
+      "step": 350
+    },
+    {
+      "epoch": 0.13860814322841467,
+      "grad_norm": 0.13287591934204102,
+      "learning_rate": 9.696088852558398e-06,
+      "loss": 0.7984,
+      "step": 360
+    },
+    {
+      "epoch": 0.14245836942920398,
+      "grad_norm": 0.13083535432815552,
+      "learning_rate": 9.674315347607686e-06,
+      "loss": 0.8233,
+      "step": 370
+    },
+    {
+      "epoch": 0.14630859562999327,
+      "grad_norm": 0.12564091384410858,
+      "learning_rate": 9.651814806868796e-06,
+      "loss": 0.7921,
+      "step": 380
+    },
+    {
+      "epoch": 0.15015882183078255,
+      "grad_norm": 0.14943233132362366,
+      "learning_rate": 9.62859073004107e-06,
+      "loss": 0.7944,
+      "step": 390
+    },
+    {
+      "epoch": 0.15400904803157187,
+      "grad_norm": 0.13010723888874054,
+      "learning_rate": 9.604646729361547e-06,
+      "loss": 0.7959,
+      "step": 400
+    },
+    {
+      "epoch": 0.15785927423236115,
+      "grad_norm": 0.1298939436674118,
+      "learning_rate": 9.579986529043099e-06,
+      "loss": 0.7749,
+      "step": 410
+    },
+    {
+      "epoch": 0.16170950043315044,
+      "grad_norm": 0.13276731967926025,
+      "learning_rate": 9.554613964695189e-06,
+      "loss": 0.7968,
+      "step": 420
+    },
+    {
+      "epoch": 0.16555972663393975,
+      "grad_norm": 0.13559892773628235,
+      "learning_rate": 9.528532982727281e-06,
+      "loss": 0.7895,
+      "step": 430
+    },
+    {
+      "epoch": 0.16940995283472904,
+      "grad_norm": 0.14169320464134216,
+      "learning_rate": 9.501747639735018e-06,
+      "loss": 0.7851,
+      "step": 440
+    },
+    {
+      "epoch": 0.17326017903551832,
+      "grad_norm": 0.130951389670372,
+      "learning_rate": 9.474262101869272e-06,
+      "loss": 0.7713,
+      "step": 450
+    },
+    {
+      "epoch": 0.17711040523630764,
+      "grad_norm": 0.13912425935268402,
+      "learning_rate": 9.446080644188141e-06,
+      "loss": 0.7885,
+      "step": 460
+    },
+    {
+      "epoch": 0.18096063143709692,
+      "grad_norm": 0.14031025767326355,
+      "learning_rate": 9.417207649992017e-06,
+      "loss": 0.7912,
+      "step": 470
+    },
+    {
+      "epoch": 0.18481085763788624,
+      "grad_norm": 0.13050790131092072,
+      "learning_rate": 9.387647610141816e-06,
+      "loss": 0.7831,
+      "step": 480
+    },
+    {
+      "epoch": 0.18866108383867553,
+      "grad_norm": 0.12025883793830872,
+      "learning_rate": 9.35740512236047e-06,
+      "loss": 0.7753,
+      "step": 490
+    },
+    {
+      "epoch": 0.1925113100394648,
+      "grad_norm": 0.12513162195682526,
+      "learning_rate": 9.326484890517809e-06,
+      "loss": 0.7816,
+      "step": 500
+    },
+    {
+      "epoch": 0.19636153624025413,
+      "grad_norm": 0.13071326911449432,
+      "learning_rate": 9.294891723898926e-06,
+      "loss": 0.7658,
+      "step": 510
+    },
+    {
+      "epoch": 0.2002117624410434,
+      "grad_norm": 0.13292165100574493,
+      "learning_rate": 9.262630536456153e-06,
+      "loss": 0.8049,
+      "step": 520
+    },
+    {
+      "epoch": 0.2040619886418327,
+      "grad_norm": 0.13903024792671204,
+      "learning_rate": 9.229706346044749e-06,
+      "loss": 0.7722,
+      "step": 530
+    },
+    {
+      "epoch": 0.207912214842622,
+      "grad_norm": 0.14837901294231415,
+      "learning_rate": 9.196124273642422e-06,
+      "loss": 0.7931,
+      "step": 540
+    },
+    {
+      "epoch": 0.2117624410434113,
+      "grad_norm": 0.12491396814584732,
+      "learning_rate": 9.161889542552842e-06,
+      "loss": 0.8008,
+      "step": 550
+    },
+    {
+      "epoch": 0.21561266724420058,
+      "grad_norm": 0.12683063745498657,
+      "learning_rate": 9.127007477593193e-06,
+      "loss": 0.7965,
+      "step": 560
+    },
+    {
+      "epoch": 0.2194628934449899,
+      "grad_norm": 0.14064358174800873,
+      "learning_rate": 9.091483504265974e-06,
+      "loss": 0.8043,
+      "step": 570
+    },
+    {
+      "epoch": 0.22331311964577918,
+      "grad_norm": 0.13511013984680176,
+      "learning_rate": 9.055323147915116e-06,
+      "loss": 0.782,
+      "step": 580
+    },
+    {
+      "epoch": 0.2271633458465685,
+      "grad_norm": 0.13587257266044617,
+      "learning_rate": 9.018532032866585e-06,
+      "loss": 0.7984,
+      "step": 590
+    },
+    {
+      "epoch": 0.23101357204735778,
+      "grad_norm": 0.1325037181377411,
+      "learning_rate": 8.981115881553579e-06,
+      "loss": 0.7922,
+      "step": 600
+    },
+    {
+      "epoch": 0.23486379824814707,
+      "grad_norm": 0.14388161897659302,
+      "learning_rate": 8.943080513626476e-06,
+      "loss": 0.8066,
+      "step": 610
+    },
+    {
+      "epoch": 0.23871402444893638,
+      "grad_norm": 0.1333983987569809,
+      "learning_rate": 8.904431845047641e-06,
+      "loss": 0.7961,
+      "step": 620
+    },
+    {
+      "epoch": 0.24256425064972567,
+      "grad_norm": 0.14152297377586365,
+      "learning_rate": 8.86517588717128e-06,
+      "loss": 0.8003,
+      "step": 630
+    },
+    {
+      "epoch": 0.24641447685051496,
+      "grad_norm": 0.12305159121751785,
+      "learning_rate": 8.82531874580844e-06,
+      "loss": 0.7994,
+      "step": 640
+    },
+    {
+      "epoch": 0.25026470305130427,
+      "grad_norm": 0.122881218791008,
+      "learning_rate": 8.784866620277314e-06,
+      "loss": 0.7558,
+      "step": 650
+    },
+    {
+      "epoch": 0.25411492925209356,
+      "grad_norm": 0.14221997559070587,
+      "learning_rate": 8.743825802439018e-06,
+      "loss": 0.7757,
+      "step": 660
+    },
+    {
+      "epoch": 0.25796515545288284,
+      "grad_norm": 0.15257683396339417,
+      "learning_rate": 8.702202675718956e-06,
+      "loss": 0.7979,
+      "step": 670
+    },
+    {
+      "epoch": 0.26181538165367213,
+      "grad_norm": 0.12376541644334793,
+      "learning_rate": 8.660003714113959e-06,
+      "loss": 0.7711,
+      "step": 680
+    },
+    {
+      "epoch": 0.26566560785446147,
+      "grad_norm": 0.12757772207260132,
+      "learning_rate": 8.617235481185323e-06,
+      "loss": 0.7823,
+      "step": 690
+    },
+    {
+      "epoch": 0.26951583405525076,
+      "grad_norm": 0.1348702609539032,
+      "learning_rate": 8.573904629037931e-06,
+      "loss": 0.791,
+      "step": 700
+    },
+    {
+      "epoch": 0.27336606025604004,
+      "grad_norm": 0.12504084408283234,
+      "learning_rate": 8.530017897285585e-06,
+      "loss": 0.7753,
+      "step": 710
+    },
+    {
+      "epoch": 0.27721628645682933,
+      "grad_norm": 0.11880157887935638,
+      "learning_rate": 8.485582112002746e-06,
+      "loss": 0.7807,
+      "step": 720
+    },
+    {
+      "epoch": 0.2810665126576186,
+      "grad_norm": 0.13829629123210907,
+      "learning_rate": 8.440604184662811e-06,
+      "loss": 0.7873,
+      "step": 730
+    },
+    {
+      "epoch": 0.28491673885840796,
+      "grad_norm": 0.1341078281402588,
+      "learning_rate": 8.395091111063118e-06,
+      "loss": 0.7703,
+      "step": 740
+    },
+    {
+      "epoch": 0.28876696505919724,
+      "grad_norm": 0.12302078306674957,
+      "learning_rate": 8.349049970236822e-06,
+      "loss": 0.7735,
+      "step": 750
+    },
+    {
+      "epoch": 0.29261719125998653,
+      "grad_norm": 0.12622518837451935,
+      "learning_rate": 8.302487923351842e-06,
+      "loss": 0.7635,
+      "step": 760
+    },
+    {
+      "epoch": 0.2964674174607758,
+      "grad_norm": 0.13887855410575867,
+      "learning_rate": 8.255412212597021e-06,
+      "loss": 0.7804,
+      "step": 770
+    },
+    {
+      "epoch": 0.3003176436615651,
+      "grad_norm": 0.13842009007930756,
+      "learning_rate": 8.207830160055688e-06,
+      "loss": 0.7777,
+      "step": 780
+    },
+    {
+      "epoch": 0.3041678698623544,
+      "grad_norm": 0.1605684459209442,
+      "learning_rate": 8.159749166566786e-06,
+      "loss": 0.7636,
+      "step": 790
+    },
+    {
+      "epoch": 0.30801809606314373,
+      "grad_norm": 0.13192898035049438,
+      "learning_rate": 8.11117671057377e-06,
+      "loss": 0.7905,
+      "step": 800
+    },
+    {
+      "epoch": 0.311868322263933,
+      "grad_norm": 0.14384149014949799,
+      "learning_rate": 8.062120346961404e-06,
+      "loss": 0.7755,
+      "step": 810
+    },
+    {
+      "epoch": 0.3157185484647223,
+      "grad_norm": 0.12282013893127441,
+      "learning_rate": 8.012587705880705e-06,
+      "loss": 0.7806,
+      "step": 820
+    },
+    {
+      "epoch": 0.3195687746655116,
+      "grad_norm": 0.13046863675117493,
+      "learning_rate": 7.962586491562146e-06,
+      "loss": 0.7656,
+      "step": 830
+    },
+    {
+      "epoch": 0.3234190008663009,
+      "grad_norm": 0.13952922821044922,
+      "learning_rate": 7.912124481117357e-06,
+      "loss": 0.7714,
+      "step": 840
+    },
+    {
+      "epoch": 0.3272692270670902,
+      "grad_norm": 0.12606029212474823,
+      "learning_rate": 7.861209523329494e-06,
+      "loss": 0.7749,
+      "step": 850
+    },
+    {
+      "epoch": 0.3311194532678795,
+      "grad_norm": 0.12955816090106964,
+      "learning_rate": 7.809849537432432e-06,
+      "loss": 0.7639,
+      "step": 860
+    },
+    {
+      "epoch": 0.3349696794686688,
+      "grad_norm": 0.15477503836154938,
+      "learning_rate": 7.758052511879035e-06,
+      "loss": 0.7632,
+      "step": 870
+    },
+    {
+      "epoch": 0.3388199056694581,
+      "grad_norm": 0.12045598030090332,
+      "learning_rate": 7.705826503098642e-06,
+      "loss": 0.7706,
+      "step": 880
+    },
+    {
+      "epoch": 0.34267013187024736,
+      "grad_norm": 0.12975232303142548,
+      "learning_rate": 7.653179634243979e-06,
+      "loss": 0.7627,
+      "step": 890
+    },
+    {
+      "epoch": 0.34652035807103665,
+      "grad_norm": 0.12093327194452286,
+      "learning_rate": 7.600120093927696e-06,
+      "loss": 0.7618,
+      "step": 900
+    },
+    {
+      "epoch": 0.350370584271826,
+      "grad_norm": 0.13550744950771332,
+      "learning_rate": 7.546656134948729e-06,
+      "loss": 0.7609,
+      "step": 910
+    },
+    {
+      "epoch": 0.3542208104726153,
+      "grad_norm": 0.13092990219593048,
+      "learning_rate": 7.492796073008661e-06,
+      "loss": 0.7548,
+      "step": 920
+    },
+    {
+      "epoch": 0.35807103667340456,
+      "grad_norm": 0.1370266079902649,
+      "learning_rate": 7.438548285418326e-06,
+      "loss": 0.7474,
+      "step": 930
+    },
+    {
+      "epoch": 0.36192126287419385,
+      "grad_norm": 0.14137309789657593,
+      "learning_rate": 7.3839212097948e-06,
+      "loss": 0.786,
+      "step": 940
+    },
+    {
+      "epoch": 0.36577148907498314,
+      "grad_norm": 0.12099441885948181,
+      "learning_rate": 7.328923342749039e-06,
+      "loss": 0.7965,
+      "step": 950
+    },
+    {
+      "epoch": 0.3696217152757725,
+      "grad_norm": 0.1384241133928299,
+      "learning_rate": 7.273563238564319e-06,
+      "loss": 0.773,
+      "step": 960
+    },
+    {
+      "epoch": 0.37347194147656176,
+      "grad_norm": 0.12875156104564667,
+      "learning_rate": 7.217849507865724e-06,
+      "loss": 0.7533,
+      "step": 970
+    },
+    {
+      "epoch": 0.37732216767735105,
+      "grad_norm": 0.12601423263549805,
+      "learning_rate": 7.161790816280858e-06,
+      "loss": 0.7648,
+      "step": 980
+    },
+    {
+      "epoch": 0.38117239387814034,
+      "grad_norm": 0.14266720414161682,
+      "learning_rate": 7.1053958830920125e-06,
+      "loss": 0.7458,
+      "step": 990
+    },
+    {
+      "epoch": 0.3850226200789296,
+      "grad_norm": 0.13349124789237976,
+      "learning_rate": 7.0486734798799715e-06,
+      "loss": 0.7752,
+      "step": 1000
+    },
+    {
+      "epoch": 0.3888728462797189,
+      "grad_norm": 0.13659334182739258,
+      "learning_rate": 6.991632429159708e-06,
+      "loss": 0.7942,
+      "step": 1010
+    },
+    {
+      "epoch": 0.39272307248050825,
+      "grad_norm": 0.13839192688465118,
+      "learning_rate": 6.934281603008132e-06,
+      "loss": 0.7704,
+      "step": 1020
+    },
+    {
+      "epoch": 0.39657329868129754,
+      "grad_norm": 0.13028165698051453,
+      "learning_rate": 6.876629921684143e-06,
+      "loss": 0.781,
+      "step": 1030
+    },
+    {
+      "epoch": 0.4004235248820868,
+      "grad_norm": 0.13299152255058289,
+      "learning_rate": 6.818686352241199e-06,
+      "loss": 0.7802,
+      "step": 1040
+    },
+    {
+      "epoch": 0.4042737510828761,
+      "grad_norm": 0.121186263859272,
+      "learning_rate": 6.760459907132573e-06,
+      "loss": 0.7667,
+      "step": 1050
+    },
+    {
+      "epoch": 0.4081239772836654,
+      "grad_norm": 0.13482633233070374,
+      "learning_rate": 6.701959642809589e-06,
+      "loss": 0.772,
+      "step": 1060
+    },
+    {
+      "epoch": 0.41197420348445474,
+      "grad_norm": 0.12591473758220673,
+      "learning_rate": 6.643194658312987e-06,
+      "loss": 0.7562,
+      "step": 1070
+    },
+    {
+      "epoch": 0.415824429685244,
+      "grad_norm": 0.11694116145372391,
+      "learning_rate": 6.584174093857676e-06,
+      "loss": 0.7877,
+      "step": 1080
+    },
+    {
+      "epoch": 0.4196746558860333,
+      "grad_norm": 0.1456041932106018,
+      "learning_rate": 6.5249071294110735e-06,
+      "loss": 0.7644,
+      "step": 1090
+    },
+    {
+      "epoch": 0.4235248820868226,
+      "grad_norm": 0.14559213817119598,
+      "learning_rate": 6.465402983265274e-06,
+      "loss": 0.7828,
+      "step": 1100
+    },
+    {
+      "epoch": 0.4273751082876119,
+      "grad_norm": 0.13032405078411102,
+      "learning_rate": 6.405670910603243e-06,
+      "loss": 0.7789,
+      "step": 1110
+    },
+    {
+      "epoch": 0.43122533448840117,
+      "grad_norm": 0.14029555022716522,
+      "learning_rate": 6.345720202059291e-06,
+      "loss": 0.7704,
+      "step": 1120
+    },
+    {
+      "epoch": 0.4350755606891905,
+      "grad_norm": 0.13209089636802673,
+      "learning_rate": 6.285560182274008e-06,
+      "loss": 0.7592,
+      "step": 1130
+    },
+    {
+      "epoch": 0.4389257868899798,
+      "grad_norm": 0.1249668076634407,
+      "learning_rate": 6.225200208443936e-06,
+      "loss": 0.7659,
+      "step": 1140
+    },
+    {
+      "epoch": 0.4427760130907691,
+      "grad_norm": 0.12155070900917053,
+      "learning_rate": 6.16464966886616e-06,
+      "loss": 0.7612,
+      "step": 1150
+    },
+    {
+      "epoch": 0.44662623929155837,
+      "grad_norm": 0.14608614146709442,
+      "learning_rate": 6.103917981478065e-06,
+      "loss": 0.7898,
+      "step": 1160
+    },
+    {
+      "epoch": 0.45047646549234766,
+      "grad_norm": 0.12289571762084961,
+      "learning_rate": 6.043014592392489e-06,
+      "loss": 0.7946,
+      "step": 1170
+    },
+    {
+      "epoch": 0.454326691693137,
+      "grad_norm": 0.11624269187450409,
+      "learning_rate": 5.981948974428485e-06,
+      "loss": 0.7695,
+      "step": 1180
+    },
+    {
+      "epoch": 0.4581769178939263,
+      "grad_norm": 0.12709057331085205,
+      "learning_rate": 5.920730625637934e-06,
+      "loss": 0.7469,
+      "step": 1190
+    },
+    {
+      "epoch": 0.46202714409471557,
+      "grad_norm": 0.13609179854393005,
+      "learning_rate": 5.8593690678282355e-06,
+      "loss": 0.7578,
+      "step": 1200
+    },
+    {
+      "epoch": 0.46587737029550486,
+      "grad_norm": 0.13217753171920776,
+      "learning_rate": 5.797873845081301e-06,
+      "loss": 0.7633,
+      "step": 1210
+    },
+    {
+      "epoch": 0.46972759649629414,
+      "grad_norm": 0.12627682089805603,
+      "learning_rate": 5.736254522269075e-06,
+      "loss": 0.7778,
+      "step": 1220
+    },
+    {
+      "epoch": 0.47357782269708343,
+      "grad_norm": 0.12332972884178162,
+      "learning_rate": 5.6745206835658386e-06,
+      "loss": 0.7801,
+      "step": 1230
+    },
+    {
+      "epoch": 0.47742804889787277,
+      "grad_norm": 0.12149509787559509,
+      "learning_rate": 5.612681930957495e-06,
+      "loss": 0.7599,
+      "step": 1240
+    },
+    {
+      "epoch": 0.48127827509866206,
+      "grad_norm": 0.12278099358081818,
+      "learning_rate": 5.550747882748095e-06,
+      "loss": 0.755,
+      "step": 1250
+    },
+    {
+      "epoch": 0.48512850129945134,
+      "grad_norm": 0.12988902628421783,
+      "learning_rate": 5.4887281720638195e-06,
+      "loss": 0.7579,
+      "step": 1260
+    },
+    {
+      "epoch": 0.48897872750024063,
+      "grad_norm": 0.15063133835792542,
+      "learning_rate": 5.42663244535466e-06,
+      "loss": 0.7447,
+      "step": 1270
+    },
+    {
+      "epoch": 0.4928289537010299,
+      "grad_norm": 0.13177147507667542,
+      "learning_rate": 5.3644703608940205e-06,
+      "loss": 0.7515,
+      "step": 1280
+    },
+    {
+      "epoch": 0.49667917990181926,
+      "grad_norm": 0.12496476620435715,
+      "learning_rate": 5.302251587276491e-06,
+      "loss": 0.7743,
+      "step": 1290
+    },
+    {
+      "epoch": 0.5005294061026085,
+      "grad_norm": 0.13080374896526337,
+      "learning_rate": 5.2399858019140005e-06,
+      "loss": 0.7676,
+      "step": 1300
+    },
+    {
+      "epoch": 0.5043796323033978,
+      "grad_norm": 0.1356072872877121,
+      "learning_rate": 5.177682689530615e-06,
+      "loss": 0.7827,
+      "step": 1310
+    },
+    {
+      "epoch": 0.5082298585041871,
+      "grad_norm": 0.14104081690311432,
+      "learning_rate": 5.11535194065619e-06,
+      "loss": 0.7563,
+      "step": 1320
+    },
+    {
+      "epoch": 0.5120800847049765,
+      "grad_norm": 0.14419083297252655,
+      "learning_rate": 5.053003250119113e-06,
+      "loss": 0.7579,
+      "step": 1330
+    },
+    {
+      "epoch": 0.5159303109057657,
+      "grad_norm": 0.13484668731689453,
+      "learning_rate": 4.9906463155383925e-06,
+      "loss": 0.7677,
+      "step": 1340
+    },
+    {
+      "epoch": 0.519780537106555,
+      "grad_norm": 0.13017813861370087,
+      "learning_rate": 4.9282908358153046e-06,
+      "loss": 0.7592,
+      "step": 1350
+    },
+    {
+      "epoch": 0.5236307633073443,
+      "grad_norm": 0.14236843585968018,
+      "learning_rate": 4.8659465096248345e-06,
+      "loss": 0.7677,
+      "step": 1360
+    },
+    {
+      "epoch": 0.5274809895081336,
+      "grad_norm": 0.12917667627334595,
+      "learning_rate": 4.803623033907169e-06,
+      "loss": 0.7737,
+      "step": 1370
+    },
+    {
+      "epoch": 0.5313312157089229,
+      "grad_norm": 0.12443782389163971,
+      "learning_rate": 4.741330102359441e-06,
+      "loss": 0.7794,
+      "step": 1380
+    },
+    {
+      "epoch": 0.5351814419097122,
+      "grad_norm": 0.13911321759223938,
+      "learning_rate": 4.679077403927996e-06,
+      "loss": 0.7534,
+      "step": 1390
+    },
+    {
+      "epoch": 0.5390316681105015,
+      "grad_norm": 0.13194623589515686,
+      "learning_rate": 4.616874621301373e-06,
+      "loss": 0.76,
+      "step": 1400
+    },
+    {
+      "epoch": 0.5428818943112907,
+      "grad_norm": 0.13341842591762543,
+      "learning_rate": 4.554731429404293e-06,
+      "loss": 0.7775,
+      "step": 1410
+    },
+    {
+      "epoch": 0.5467321205120801,
+      "grad_norm": 0.15359541773796082,
+      "learning_rate": 4.492657493892822e-06,
+      "loss": 0.7581,
+      "step": 1420
+    },
+    {
+      "epoch": 0.5505823467128694,
+      "grad_norm": 0.14718052744865417,
+      "learning_rate": 4.430662469650999e-06,
+      "loss": 0.7707,
+      "step": 1430
+    },
+    {
+      "epoch": 0.5544325729136587,
+      "grad_norm": 0.13260136544704437,
+      "learning_rate": 4.368755999289118e-06,
+      "loss": 0.774,
+      "step": 1440
+    },
+    {
+      "epoch": 0.558282799114448,
+      "grad_norm": 0.13983632624149323,
+      "learning_rate": 4.306947711643949e-06,
+      "loss": 0.7573,
+      "step": 1450
+    },
+    {
+      "epoch": 0.5621330253152372,
+      "grad_norm": 0.13076947629451752,
+      "learning_rate": 4.245247220281068e-06,
+      "loss": 0.7626,
+      "step": 1460
+    },
+    {
+      "epoch": 0.5659832515160266,
+      "grad_norm": 0.12532980740070343,
+      "learning_rate": 4.183664121999587e-06,
+      "loss": 0.7575,
+      "step": 1470
+    },
+    {
+      "epoch": 0.5698334777168159,
+      "grad_norm": 0.13432535529136658,
+      "learning_rate": 4.122207995339474e-06,
+      "loss": 0.7581,
+      "step": 1480
+    },
+    {
+      "epoch": 0.5736837039176051,
+      "grad_norm": 0.13435795903205872,
+      "learning_rate": 4.060888399091734e-06,
+      "loss": 0.7634,
+      "step": 1490
+    },
+    {
+      "epoch": 0.5775339301183945,
+      "grad_norm": 0.13242164254188538,
+      "learning_rate": 3.999714870811639e-06,
+      "loss": 0.7736,
+      "step": 1500
+    },
+    {
+      "epoch": 0.5813841563191837,
+      "grad_norm": 0.12284350395202637,
+      "learning_rate": 3.938696925335275e-06,
+      "loss": 0.7626,
+      "step": 1510
+    },
+    {
+      "epoch": 0.5852343825199731,
+      "grad_norm": 0.14273084700107574,
+      "learning_rate": 3.87784405329962e-06,
+      "loss": 0.7707,
+      "step": 1520
+    },
+    {
+      "epoch": 0.5890846087207623,
+      "grad_norm": 0.1340179592370987,
+      "learning_rate": 3.817165719666382e-06,
+      "loss": 0.7686,
+      "step": 1530
+    },
+    {
+      "epoch": 0.5929348349215516,
+      "grad_norm": 0.12685668468475342,
+      "learning_rate": 3.756671362249841e-06,
+      "loss": 0.7776,
+      "step": 1540
+    },
+    {
+      "epoch": 0.596785061122341,
+      "grad_norm": 0.13374850153923035,
+      "learning_rate": 3.6963703902488907e-06,
+      "loss": 0.7694,
+      "step": 1550
+    },
+    {
+      "epoch": 0.6006352873231302,
+      "grad_norm": 0.13426467776298523,
+      "learning_rate": 3.6362721827835655e-06,
+      "loss": 0.7807,
+      "step": 1560
+    },
+    {
+      "epoch": 0.6044855135239195,
+      "grad_norm": 0.1366957724094391,
+      "learning_rate": 3.576386087436212e-06,
+      "loss": 0.7706,
+      "step": 1570
+    },
+    {
+      "epoch": 0.6083357397247088,
+      "grad_norm": 0.13442905247211456,
+      "learning_rate": 3.5167214187975913e-06,
+      "loss": 0.7722,
+      "step": 1580
+    },
+    {
+      "epoch": 0.6121859659254981,
+      "grad_norm": 0.11772517114877701,
+      "learning_rate": 3.457287457018086e-06,
+      "loss": 0.7703,
+      "step": 1590
+    },
+    {
+      "epoch": 0.6160361921262875,
+      "grad_norm": 0.12089650332927704,
+      "learning_rate": 3.3980934463643003e-06,
+      "loss": 0.7534,
+      "step": 1600
+    },
+    {
+      "epoch": 0.6198864183270767,
+      "grad_norm": 0.11867595463991165,
+      "learning_rate": 3.3391485937812053e-06,
+      "loss": 0.7358,
+      "step": 1610
+    },
+    {
+      "epoch": 0.623736644527866,
+      "grad_norm": 0.12567250430583954,
+      "learning_rate": 3.280462067460115e-06,
+      "loss": 0.7369,
+      "step": 1620
+    },
+    {
+      "epoch": 0.6275868707286553,
+      "grad_norm": 0.12301046401262283,
+      "learning_rate": 3.222042995412669e-06,
+      "loss": 0.7759,
+      "step": 1630
+    },
+    {
+      "epoch": 0.6314370969294446,
+      "grad_norm": 0.11937104165554047,
+      "learning_rate": 3.163900464051089e-06,
+      "loss": 0.7586,
+      "step": 1640
+    },
+    {
+      "epoch": 0.635287323130234,
+      "grad_norm": 0.13481391966342926,
+      "learning_rate": 3.1060435167748826e-06,
+      "loss": 0.7685,
+      "step": 1650
+    },
+    {
+      "epoch": 0.6391375493310232,
+      "grad_norm": 0.12800006568431854,
+      "learning_rate": 3.048481152564253e-06,
+      "loss": 0.7533,
+      "step": 1660
+    },
+    {
+      "epoch": 0.6429877755318125,
+      "grad_norm": 0.13377708196640015,
+      "learning_rate": 2.9912223245804017e-06,
+      "loss": 0.778,
+      "step": 1670
+    },
+    {
+      "epoch": 0.6468380017326018,
+      "grad_norm": 0.13154065608978271,
+      "learning_rate": 2.934275938772981e-06,
+      "loss": 0.7592,
+      "step": 1680
+    },
+    {
+      "epoch": 0.6506882279333911,
+      "grad_norm": 0.13539618253707886,
+      "learning_rate": 2.8776508524948676e-06,
+      "loss": 0.7498,
+      "step": 1690
+    },
+    {
+      "epoch": 0.6545384541341804,
+      "grad_norm": 0.1216798722743988,
+      "learning_rate": 2.821355873124503e-06,
+      "loss": 0.7471,
+      "step": 1700
+    },
+    {
+      "epoch": 0.6583886803349697,
+      "grad_norm": 0.12857292592525482,
+      "learning_rate": 2.7653997566960144e-06,
+      "loss": 0.7685,
+      "step": 1710
+    },
+    {
+      "epoch": 0.662238906535759,
+      "grad_norm": 0.13037486374378204,
+      "learning_rate": 2.709791206537309e-06,
+      "loss": 0.7564,
+      "step": 1720
+    },
+    {
+      "epoch": 0.6660891327365482,
+      "grad_norm": 0.12307510524988174,
+      "learning_rate": 2.6545388719163747e-06,
+      "loss": 0.7588,
+      "step": 1730
+    },
+    {
+      "epoch": 0.6699393589373376,
+      "grad_norm": 0.12428302317857742,
+      "learning_rate": 2.599651346695979e-06,
+      "loss": 0.7663,
+      "step": 1740
+    },
+    {
+      "epoch": 0.6737895851381268,
+      "grad_norm": 0.13857893645763397,
+      "learning_rate": 2.545137167997007e-06,
+      "loss": 0.7595,
+      "step": 1750
+    },
+    {
+      "epoch": 0.6776398113389162,
+      "grad_norm": 0.11809112876653671,
+      "learning_rate": 2.4910048148705972e-06,
+      "loss": 0.7464,
+      "step": 1760
+    },
+    {
+      "epoch": 0.6814900375397055,
+      "grad_norm": 0.13366936147212982,
+      "learning_rate": 2.4372627069793327e-06,
+      "loss": 0.7884,
+      "step": 1770
+    },
+    {
+      "epoch": 0.6853402637404947,
+      "grad_norm": 0.13095703721046448,
+      "learning_rate": 2.3839192032876486e-06,
+      "loss": 0.7712,
+      "step": 1780
+    },
+    {
+      "epoch": 0.6891904899412841,
+      "grad_norm": 0.121241495013237,
+      "learning_rate": 2.3309826007617047e-06,
+      "loss": 0.7825,
+      "step": 1790
+    },
+    {
+      "epoch": 0.6930407161420733,
+      "grad_norm": 0.1243000403046608,
+      "learning_rate": 2.278461133078881e-06,
+      "loss": 0.7494,
+      "step": 1800
+    },
+    {
+      "epoch": 0.6968909423428626,
+      "grad_norm": 0.12692585587501526,
+      "learning_rate": 2.2263629693471258e-06,
+      "loss": 0.7437,
+      "step": 1810
+    },
+    {
+      "epoch": 0.700741168543652,
+      "grad_norm": 0.12234779447317123,
+      "learning_rate": 2.1746962128343414e-06,
+      "loss": 0.7683,
+      "step": 1820
+    },
+    {
+      "epoch": 0.7045913947444412,
+      "grad_norm": 0.12647537887096405,
+      "learning_rate": 2.1234688997080183e-06,
+      "loss": 0.7648,
+      "step": 1830
+    },
+    {
+      "epoch": 0.7084416209452306,
+      "grad_norm": 0.1402587890625,
+      "learning_rate": 2.0726889977853053e-06,
+      "loss": 0.7624,
+      "step": 1840
+    },
+    {
+      "epoch": 0.7122918471460198,
+      "grad_norm": 0.131031796336174,
+      "learning_rate": 2.022364405293703e-06,
+      "loss": 0.7723,
+      "step": 1850
+    },
+    {
+      "epoch": 0.7161420733468091,
+      "grad_norm": 0.12695756554603577,
+      "learning_rate": 1.9725029496425814e-06,
+      "loss": 0.7687,
+      "step": 1860
+    },
+    {
+      "epoch": 0.7199922995475985,
+      "grad_norm": 0.13097985088825226,
+      "learning_rate": 1.923112386205727e-06,
+      "loss": 0.7635,
+      "step": 1870
+    },
+    {
+      "epoch": 0.7238425257483877,
+      "grad_norm": 0.13546185195446014,
+      "learning_rate": 1.874200397115079e-06,
+      "loss": 0.7697,
+      "step": 1880
+    },
+    {
+      "epoch": 0.727692751949177,
+      "grad_norm": 0.13352108001708984,
+      "learning_rate": 1.825774590065857e-06,
+      "loss": 0.7567,
+      "step": 1890
+    },
+    {
+      "epoch": 0.7315429781499663,
+      "grad_norm": 0.1355399787425995,
+      "learning_rate": 1.7778424971332813e-06,
+      "loss": 0.7652,
+      "step": 1900
+    },
+    {
+      "epoch": 0.7353932043507556,
+      "grad_norm": 0.12489397823810577,
+      "learning_rate": 1.7304115736010407e-06,
+      "loss": 0.763,
+      "step": 1910
+    },
+    {
+      "epoch": 0.739243430551545,
+      "grad_norm": 0.1240018755197525,
+      "learning_rate": 1.6834891968017108e-06,
+      "loss": 0.7554,
+      "step": 1920
+    },
+    {
+      "epoch": 0.7430936567523342,
+      "grad_norm": 0.1305236965417862,
+      "learning_rate": 1.6370826649692884e-06,
+      "loss": 0.7699,
+      "step": 1930
+    },
+    {
+      "epoch": 0.7469438829531235,
+      "grad_norm": 0.12964355945587158,
+      "learning_rate": 1.5911991961040463e-06,
+      "loss": 0.7803,
+      "step": 1940
+    },
+    {
+      "epoch": 0.7507941091539128,
+      "grad_norm": 0.1294676959514618,
+      "learning_rate": 1.545845926849851e-06,
+      "loss": 0.7671,
+      "step": 1950
+    },
+    {
+      "epoch": 0.7546443353547021,
+      "grad_norm": 0.13649147748947144,
+      "learning_rate": 1.5010299113841397e-06,
+      "loss": 0.7829,
+      "step": 1960
+    },
+    {
+      "epoch": 0.7584945615554913,
+      "grad_norm": 0.12891341745853424,
+      "learning_rate": 1.4567581203207238e-06,
+      "loss": 0.7792,
+      "step": 1970
+    },
+    {
+      "epoch": 0.7623447877562807,
+      "grad_norm": 0.13368885219097137,
+      "learning_rate": 1.413037439625593e-06,
+      "loss": 0.762,
+      "step": 1980
+    },
+    {
+      "epoch": 0.76619501395707,
+      "grad_norm": 0.1444050371646881,
+      "learning_rate": 1.3698746695458847e-06,
+      "loss": 0.767,
+      "step": 1990
+    },
+    {
+      "epoch": 0.7700452401578592,
+      "grad_norm": 0.12430238723754883,
+      "learning_rate": 1.3272765235521745e-06,
+      "loss": 0.7659,
+      "step": 2000
+    },
+    {
+      "epoch": 0.7738954663586486,
+      "grad_norm": 0.1277063488960266,
+      "learning_rate": 1.2852496272942877e-06,
+      "loss": 0.7637,
+      "step": 2010
+    },
+    {
+      "epoch": 0.7777456925594378,
+      "grad_norm": 0.11850254237651825,
+      "learning_rate": 1.2438005175707396e-06,
+      "loss": 0.7527,
+      "step": 2020
+    },
+    {
+      "epoch": 0.7815959187602272,
+      "grad_norm": 0.12748679518699646,
+      "learning_rate": 1.2029356413120297e-06,
+      "loss": 0.7365,
+      "step": 2030
+    },
+    {
+      "epoch": 0.7854461449610165,
+      "grad_norm": 0.1298525184392929,
+      "learning_rate": 1.1626613545778786e-06,
+      "loss": 0.757,
+      "step": 2040
+    },
+    {
+      "epoch": 0.7892963711618057,
+      "grad_norm": 0.11869119852781296,
+      "learning_rate": 1.1229839215686294e-06,
+      "loss": 0.7626,
+      "step": 2050
+    },
+    {
+      "epoch": 0.7931465973625951,
+      "grad_norm": 0.1212410032749176,
+      "learning_rate": 1.0839095136509188e-06,
+      "loss": 0.7448,
+      "step": 2060
+    },
+    {
+      "epoch": 0.7969968235633843,
+      "grad_norm": 0.1335805207490921,
+      "learning_rate": 1.045444208397791e-06,
+      "loss": 0.7564,
+      "step": 2070
+    },
+    {
+      "epoch": 0.8008470497641736,
+      "grad_norm": 0.12781178951263428,
+      "learning_rate": 1.007593988643402e-06,
+      "loss": 0.7428,
+      "step": 2080
+    },
+    {
+      "epoch": 0.804697275964963,
+      "grad_norm": 0.12488113343715668,
+      "learning_rate": 9.703647415524613e-07,
+      "loss": 0.7658,
+      "step": 2090
+    },
+    {
+      "epoch": 0.8085475021657522,
+      "grad_norm": 0.12729336321353912,
+      "learning_rate": 9.337622577045485e-07,
+      "loss": 0.7614,
+      "step": 2100
+    },
+    {
+      "epoch": 0.8123977283665416,
+      "grad_norm": 0.12341848760843277,
+      "learning_rate": 8.977922301934588e-07,
+      "loss": 0.7732,
+      "step": 2110
+    },
+    {
+      "epoch": 0.8162479545673308,
+      "grad_norm": 0.1305152326822281,
+      "learning_rate": 8.624602537417009e-07,
+      "loss": 0.7696,
+      "step": 2120
+    },
+    {
+      "epoch": 0.8200981807681201,
+      "grad_norm": 0.13209085166454315,
+      "learning_rate": 8.277718238303123e-07,
+      "loss": 0.7554,
+      "step": 2130
+    },
+    {
+      "epoch": 0.8239484069689095,
+      "grad_norm": 0.12885673344135284,
+      "learning_rate": 7.937323358440935e-07,
+      "loss": 0.7659,
+      "step": 2140
+    },
+    {
+      "epoch": 0.8277986331696987,
+      "grad_norm": 0.13965924084186554,
+      "learning_rate": 7.603470842324223e-07,
+      "loss": 0.7509,
+      "step": 2150
+    },
+    {
+      "epoch": 0.831648859370488,
+      "grad_norm": 0.12412252277135849,
+      "learning_rate": 7.276212616857575e-07,
+      "loss": 0.7681,
+      "step": 2160
+    },
+    {
+      "epoch": 0.8354990855712773,
+      "grad_norm": 0.1225537583231926,
+      "learning_rate": 6.955599583279837e-07,
+      "loss": 0.7686,
+      "step": 2170
+    },
+    {
+      "epoch": 0.8393493117720666,
+      "grad_norm": 0.12548023462295532,
+      "learning_rate": 6.641681609246981e-07,
+      "loss": 0.7584,
+      "step": 2180
+    },
+    {
+      "epoch": 0.8431995379728559,
+      "grad_norm": 0.12710238993167877,
+      "learning_rate": 6.334507521075711e-07,
+      "loss": 0.7686,
+      "step": 2190
+    },
+    {
+      "epoch": 0.8470497641736452,
+      "grad_norm": 0.13170848786830902,
+      "learning_rate": 6.034125096149179e-07,
+      "loss": 0.753,
+      "step": 2200
+    },
+    {
+      "epoch": 0.8508999903744345,
+      "grad_norm": 0.1452515423297882,
+      "learning_rate": 5.740581055485722e-07,
+      "loss": 0.7431,
+      "step": 2210
+    },
+    {
+      "epoch": 0.8547502165752238,
+      "grad_norm": 0.11501020193099976,
+      "learning_rate": 5.453921056471961e-07,
+      "loss": 0.7647,
+      "step": 2220
+    },
+    {
+      "epoch": 0.8586004427760131,
+      "grad_norm": 0.1348080188035965,
+      "learning_rate": 5.174189685761288e-07,
+      "loss": 0.7559,
+      "step": 2230
+    },
+    {
+      "epoch": 0.8624506689768023,
+      "grad_norm": 0.12723205983638763,
+      "learning_rate": 4.901430452338979e-07,
+      "loss": 0.7754,
+      "step": 2240
+    },
+    {
+      "epoch": 0.8663008951775917,
+      "grad_norm": 0.1316486895084381,
+      "learning_rate": 4.6356857807548286e-07,
+      "loss": 0.7575,
+      "step": 2250
+    },
+    {
+      "epoch": 0.870151121378381,
+      "grad_norm": 0.1338786780834198,
+      "learning_rate": 4.376997004524541e-07,
+      "loss": 0.7465,
+      "step": 2260
+    },
+    {
+      "epoch": 0.8740013475791703,
+      "grad_norm": 0.12255477160215378,
+      "learning_rate": 4.12540435970073e-07,
+      "loss": 0.7484,
+      "step": 2270
+    },
+    {
+      "epoch": 0.8778515737799596,
+      "grad_norm": 0.12812486290931702,
+      "learning_rate": 3.880946978614703e-07,
+      "loss": 0.7676,
+      "step": 2280
+    },
+    {
+      "epoch": 0.8817017999807488,
+      "grad_norm": 0.12441135197877884,
+      "learning_rate": 3.643662883789878e-07,
+      "loss": 0.769,
+      "step": 2290
+    },
+    {
+      "epoch": 0.8855520261815382,
+      "grad_norm": 0.13148459792137146,
+      "learning_rate": 3.41358898202781e-07,
+      "loss": 0.7517,
+      "step": 2300
+    },
+    {
+      "epoch": 0.8894022523823275,
+      "grad_norm": 0.1197814792394638,
+      "learning_rate": 3.190761058667741e-07,
+      "loss": 0.7517,
+      "step": 2310
+    },
+    {
+      "epoch": 0.8932524785831167,
+      "grad_norm": 0.11836831271648407,
+      "learning_rate": 2.9752137720206463e-07,
+      "loss": 0.762,
+      "step": 2320
+    },
+    {
+      "epoch": 0.8971027047839061,
+      "grad_norm": 0.13113580644130707,
+      "learning_rate": 2.7669806479785067e-07,
+      "loss": 0.7727,
+      "step": 2330
+    },
+    {
+      "epoch": 0.9009529309846953,
+      "grad_norm": 0.1284720003604889,
+      "learning_rate": 2.566094074799763e-07,
+      "loss": 0.7639,
+      "step": 2340
+    },
+    {
+      "epoch": 0.9048031571854847,
+      "grad_norm": 0.12138517200946808,
+      "learning_rate": 2.3725852980716557e-07,
+      "loss": 0.7458,
+      "step": 2350
+    },
+    {
+      "epoch": 0.908653383386274,
+      "grad_norm": 0.12775886058807373,
+      "learning_rate": 2.1864844158503983e-07,
+      "loss": 0.7633,
+      "step": 2360
+    },
+    {
+      "epoch": 0.9125036095870632,
+      "grad_norm": 0.137359157204628,
+      "learning_rate": 2.007820373979702e-07,
+      "loss": 0.752,
+      "step": 2370
+    },
+    {
+      "epoch": 0.9163538357878526,
+      "grad_norm": 0.13013143837451935,
+      "learning_rate": 1.8366209615886532e-07,
+      "loss": 0.7534,
+      "step": 2380
+    },
+    {
+      "epoch": 0.9202040619886418,
+      "grad_norm": 0.12357207387685776,
+      "learning_rate": 1.6729128067693613e-07,
+      "loss": 0.7751,
+      "step": 2390
+    },
+    {
+      "epoch": 0.9240542881894311,
+      "grad_norm": 0.1325681060552597,
+      "learning_rate": 1.5167213724353426e-07,
+      "loss": 0.7646,
+      "step": 2400
+    },
+    {
+      "epoch": 0.9279045143902205,
+      "grad_norm": 0.12471011281013489,
+      "learning_rate": 1.3680709523610058e-07,
+      "loss": 0.7536,
+      "step": 2410
+    },
+    {
+      "epoch": 0.9317547405910097,
+      "grad_norm": 0.12029357999563217,
+      "learning_rate": 1.2269846674030418e-07,
+      "loss": 0.7781,
+      "step": 2420
+    },
+    {
+      "epoch": 0.935604966791799,
+      "grad_norm": 0.1299075484275818,
+      "learning_rate": 1.0934844619042595e-07,
+      "loss": 0.7738,
+      "step": 2430
+    },
+    {
+      "epoch": 0.9394551929925883,
+      "grad_norm": 0.13423195481300354,
+      "learning_rate": 9.675911002803961e-08,
+      "loss": 0.7602,
+      "step": 2440
+    },
+    {
+      "epoch": 0.9433054191933776,
+      "grad_norm": 0.13226471841335297,
+      "learning_rate": 8.493241637904381e-08,
+      "loss": 0.7569,
+      "step": 2450
+    },
+    {
+      "epoch": 0.9471556453941669,
+      "grad_norm": 0.1321547031402588,
+      "learning_rate": 7.387020474909689e-08,
+      "loss": 0.7777,
+      "step": 2460
+    },
+    {
+      "epoch": 0.9510058715949562,
+      "grad_norm": 0.1283348798751831,
+      "learning_rate": 6.357419573750856e-08,
+      "loss": 0.7614,
+      "step": 2470
+    },
+    {
+      "epoch": 0.9548560977957455,
+      "grad_norm": 0.12811590731143951,
+      "learning_rate": 5.4045990769613365e-08,
+      "loss": 0.7584,
+      "step": 2480
+    },
+    {
+      "epoch": 0.9587063239965348,
+      "grad_norm": 0.14459218084812164,
+      "learning_rate": 4.5287071847694434e-08,
+      "loss": 0.7585,
+      "step": 2490
+    },
+    {
+      "epoch": 0.9625565501973241,
+      "grad_norm": 0.13033607602119446,
+      "learning_rate": 3.7298801320467817e-08,
+      "loss": 0.7558,
+      "step": 2500
+    },
+    {
+      "epoch": 0.9664067763981133,
+      "grad_norm": 0.13140419125556946,
+      "learning_rate": 3.0082421671192576e-08,
+      "loss": 0.7506,
+      "step": 2510
+    },
+    {
+      "epoch": 0.9702570025989027,
+      "grad_norm": 0.1235627830028534,
+      "learning_rate": 2.3639055324410355e-08,
+      "loss": 0.7666,
+      "step": 2520
+    },
+    {
+      "epoch": 0.974107228799692,
+      "grad_norm": 0.12780387699604034,
+      "learning_rate": 1.7969704471369477e-08,
+      "loss": 0.7539,
+      "step": 2530
+    },
+    {
+      "epoch": 0.9779574550004813,
+      "grad_norm": 0.12176188826560974,
+      "learning_rate": 1.3075250914142413e-08,
+      "loss": 0.759,
+      "step": 2540
+    },
+    {
+      "epoch": 0.9818076812012706,
+      "grad_norm": 0.117548868060112,
+      "learning_rate": 8.956455928476049e-09,
+      "loss": 0.7682,
+      "step": 2550
+    },
+    {
+      "epoch": 0.9856579074020598,
+      "grad_norm": 0.11594812572002411,
+      "learning_rate": 5.6139601453786275e-09,
+      "loss": 0.7484,
+      "step": 2560
+    },
+    {
+      "epoch": 0.9895081336028492,
+      "grad_norm": 0.13381458818912506,
+      "learning_rate": 3.0482834514805602e-09,
+      "loss": 0.7517,
+      "step": 2570
+    },
+    {
+      "epoch": 0.9933583598036385,
+      "grad_norm": 0.11651048064231873,
+      "learning_rate": 1.2598249081713366e-09,
+      "loss": 0.7388,
+      "step": 2580
+    },
+    {
+      "epoch": 0.9972085860044277,
+      "grad_norm": 0.12992671132087708,
+      "learning_rate": 2.488626895302826e-10,
+      "loss": 0.7691,
+      "step": 2590
+    },
+    {
+      "epoch": 0.9999037443449803,
+      "step": 2597,
+      "total_flos": 2.8967692650571366e+18,
+      "train_loss": 0.7923428122704425,
+      "train_runtime": 13091.1989,
+      "train_samples_per_second": 50.79,
+      "train_steps_per_second": 0.198
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 2597,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.8967692650571366e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}

continuous/qwen3/mc2_m16_384/finetune/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:699ec8339a1a35161f67ef6664252bf6edcdb5c8ed719eeb73ca381807ef3085
+size 6136

continuous/qwen3/mc2_m16_384/pretrain/config.json

@@ -0,0 +1,60 @@
+{
+  "_attn_implementation_autoset": true,
+  "add_faster_video": false,
+  "add_time_instruction": false,
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "eos_token_id": 151645,
+  "faster_token_stride": 10,
+  "force_sample": false,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 2048,
+  "image_aspect_ratio": "square",
+  "image_crop_resolution": null,
+  "image_grid_pinpoints": null,
+  "image_split_resolution": null,
+  "initializer_range": 0.02,
+  "intermediate_size": 6144,
+  "max_position_embeddings": 40960,
+  "max_window_layers": 28,
+  "mm_hidden_size": 512,
+  "mm_newline_position": "grid",
+  "mm_patch_merge_type": "flat",
+  "mm_projector_lr": null,
+  "mm_projector_type": "mlp2x_gelu",
+  "mm_resampler_type": null,
+  "mm_spatial_pool_mode": "bilinear",
+  "mm_spatial_pool_stride": null,
+  "mm_tunable_parts": "mm_mlp_adapter",
+  "mm_use_im_patch_token": false,
+  "mm_use_im_start_end": false,
+  "mm_vision_select_feature": "patch",
+  "mm_vision_select_layer": -2,
+  "mm_vision_tower": "open_clip_hub:ViT-M-16",
+  "mm_vision_tower_lr": null,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pos_skipping_range": 4096,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000,
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "tokenizer_model_max_length": 8192,
+  "tokenizer_padding_side": "right",
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.51.3",
+  "use_cache": true,
+  "use_mm_proj": true,
+  "use_pos_skipping": false,
+  "use_sliding_window": false,
+  "vision_tower_pretrained": "/cache/ckpt/download/tokenizer/continuous/mc2/mc2_m16_384.pt",
+  "vocab_size": 151936
+}

continuous/qwen3/mc2_m16_384/pretrain/mm_projector.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:02a712e345c832e1e9828c4bced6db6f582cefc73035b4b912da77a368ff35f9
+size 20990008

continuous/qwen3/mc2_m16_384/pretrain/trainer_state.json

@@ -0,0 +1,421 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 545,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01834862385321101,
+      "grad_norm": 8.000633239746094,
+      "learning_rate": 0.00032142857142857147,
+      "loss": 8.2443,
+      "step": 10
+    },
+    {
+      "epoch": 0.03669724770642202,
+      "grad_norm": 0.16249574720859528,
+      "learning_rate": 0.0006785714285714287,
+      "loss": 5.2747,
+      "step": 20
+    },
+    {
+      "epoch": 0.05504587155963303,
+      "grad_norm": 0.0708988755941391,
+      "learning_rate": 0.0009999907688176173,
+      "loss": 4.7575,
+      "step": 30
+    },
+    {
+      "epoch": 0.07339449541284404,
+      "grad_norm": 0.04504955932497978,
+      "learning_rate": 0.0009988834393115766,
+      "loss": 4.6742,
+      "step": 40
+    },
+    {
+      "epoch": 0.09174311926605505,
+      "grad_norm": 0.036903876811265945,
+      "learning_rate": 0.0009959345572940772,
+      "loss": 4.6328,
+      "step": 50
+    },
+    {
+      "epoch": 0.11009174311926606,
+      "grad_norm": 0.03924962133169174,
+      "learning_rate": 0.0009911550081155983,
+      "loss": 4.6057,
+      "step": 60
+    },
+    {
+      "epoch": 0.12844036697247707,
+      "grad_norm": 0.05502660199999809,
+      "learning_rate": 0.0009845624347566433,
+      "loss": 4.5629,
+      "step": 70
+    },
+    {
+      "epoch": 0.14678899082568808,
+      "grad_norm": 0.05512475594878197,
+      "learning_rate": 0.0009761811727013547,
+      "loss": 4.5328,
+      "step": 80
+    },
+    {
+      "epoch": 0.1651376146788991,
+      "grad_norm": 0.06463111191987991,
+      "learning_rate": 0.0009660421601067666,
+      "loss": 4.4869,
+      "step": 90
+    },
+    {
+      "epoch": 0.1834862385321101,
+      "grad_norm": 0.057016074657440186,
+      "learning_rate": 0.0009541828235992926,
+      "loss": 4.4554,
+      "step": 100
+    },
+    {
+      "epoch": 0.2018348623853211,
+      "grad_norm": 0.0610203854739666,
+      "learning_rate": 0.0009406469401200151,
+      "loss": 4.4139,
+      "step": 110
+    },
+    {
+      "epoch": 0.22018348623853212,
+      "grad_norm": 0.05788898468017578,
+      "learning_rate": 0.0009254844753287492,
+      "loss": 4.3929,
+      "step": 120
+    },
+    {
+      "epoch": 0.23853211009174313,
+      "grad_norm": 0.05752696841955185,
+      "learning_rate": 0.0009087513991633924,
+      "loss": 4.3535,
+      "step": 130
+    },
+    {
+      "epoch": 0.25688073394495414,
+      "grad_norm": 0.06006615608930588,
+      "learning_rate": 0.0008905094792353917,
+      "loss": 4.3278,
+      "step": 140
+    },
+    {
+      "epoch": 0.27522935779816515,
+      "grad_norm": 0.050621695816516876,
+      "learning_rate": 0.0008708260528239789,
+      "loss": 4.3186,
+      "step": 150
+    },
+    {
+      "epoch": 0.29357798165137616,
+      "grad_norm": 0.06644929945468903,
+      "learning_rate": 0.0008497737783108237,
+      "loss": 4.2961,
+      "step": 160
+    },
+    {
+      "epoch": 0.3119266055045872,
+      "grad_norm": 0.06210792809724808,
+      "learning_rate": 0.0008274303669726426,
+      "loss": 4.2765,
+      "step": 170
+    },
+    {
+      "epoch": 0.3302752293577982,
+      "grad_norm": 0.06392346322536469,
+      "learning_rate": 0.0008038782961218137,
+      "loss": 4.2602,
+      "step": 180
+    },
+    {
+      "epoch": 0.3486238532110092,
+      "grad_norm": 0.06315244734287262,
+      "learning_rate": 0.000779204504653894,
+      "loss": 4.2498,
+      "step": 190
+    },
+    {
+      "epoch": 0.3669724770642202,
+      "grad_norm": 0.06137508898973465,
+      "learning_rate": 0.0007535000721258791,
+      "loss": 4.2258,
+      "step": 200
+    },
+    {
+      "epoch": 0.3853211009174312,
+      "grad_norm": 0.06133707985281944,
+      "learning_rate": 0.00072685988254984,
+      "loss": 4.2123,
+      "step": 210
+    },
+    {
+      "epoch": 0.4036697247706422,
+      "grad_norm": 0.07866296917200089,
+      "learning_rate": 0.0006993822741429906,
+      "loss": 4.2106,
+      "step": 220
+    },
+    {
+      "epoch": 0.42201834862385323,
+      "grad_norm": 0.06616409868001938,
+      "learning_rate": 0.0006711686763270818,
+      "loss": 4.1816,
+      "step": 230
+    },
+    {
+      "epoch": 0.44036697247706424,
+      "grad_norm": 0.0659882128238678,
+      "learning_rate": 0.0006423232353170831,
+      "loss": 4.1463,
+      "step": 240
+    },
+    {
+      "epoch": 0.45871559633027525,
+      "grad_norm": 0.06331344693899155,
+      "learning_rate": 0.0006129524296812335,
+      "loss": 4.1176,
+      "step": 250
+    },
+    {
+      "epoch": 0.47706422018348627,
+      "grad_norm": 0.05836929753422737,
+      "learning_rate": 0.0005831646772915651,
+      "loss": 4.1119,
+      "step": 260
+    },
+    {
+      "epoch": 0.4954128440366973,
+      "grad_norm": 0.06842830777168274,
+      "learning_rate": 0.0005530699351157809,
+      "loss": 4.0808,
+      "step": 270
+    },
+    {
+      "epoch": 0.5137614678899083,
+      "grad_norm": 0.05747227370738983,
+      "learning_rate": 0.0005227792933277943,
+      "loss": 4.0712,
+      "step": 280
+    },
+    {
+      "epoch": 0.5321100917431193,
+      "grad_norm": 0.05645981803536415,
+      "learning_rate": 0.0004924045652352092,
+      "loss": 4.0603,
+      "step": 290
+    },
+    {
+      "epoch": 0.5504587155963303,
+      "grad_norm": 0.058747123926877975,
+      "learning_rate": 0.0004620578745374633,
+      "loss": 4.0533,
+      "step": 300
+    },
+    {
+      "epoch": 0.5688073394495413,
+      "grad_norm": 0.05816194787621498,
+      "learning_rate": 0.0004318512414382059,
+      "loss": 4.0627,
+      "step": 310
+    },
+    {
+      "epoch": 0.5871559633027523,
+      "grad_norm": 0.05628599599003792,
+      "learning_rate": 0.0004018961691397148,
+      "loss": 4.0501,
+      "step": 320
+    },
+    {
+      "epoch": 0.6055045871559633,
+      "grad_norm": 0.06160425767302513,
+      "learning_rate": 0.0003723032322457458,
+      "loss": 4.036,
+      "step": 330
+    },
+    {
+      "epoch": 0.6238532110091743,
+      "grad_norm": 0.05798739194869995,
+      "learning_rate": 0.0003431816685921625,
+      "loss": 4.0346,
+      "step": 340
+    },
+    {
+      "epoch": 0.6422018348623854,
+      "grad_norm": 0.05170373618602753,
+      "learning_rate": 0.0003146389760120416,
+      "loss": 4.0356,
+      "step": 350
+    },
+    {
+      "epoch": 0.6605504587155964,
+      "grad_norm": 0.0745982676744461,
+      "learning_rate": 0.00028678051552373485,
+      "loss": 4.0418,
+      "step": 360
+    },
+    {
+      "epoch": 0.6788990825688074,
+      "grad_norm": 0.053786348551511765,
+      "learning_rate": 0.0002597091224066581,
+      "loss": 4.0255,
+      "step": 370
+    },
+    {
+      "epoch": 0.6972477064220184,
+      "grad_norm": 0.05319962278008461,
+      "learning_rate": 0.0002335247266004629,
+      "loss": 4.0386,
+      "step": 380
+    },
+    {
+      "epoch": 0.7155963302752294,
+      "grad_norm": 0.05973520874977112,
+      "learning_rate": 0.000208323983828828,
+      "loss": 4.0223,
+      "step": 390
+    },
+    {
+      "epoch": 0.7339449541284404,
+      "grad_norm": 0.06226144731044769,
+      "learning_rate": 0.0001841999188095224,
+      "loss": 4.0198,
+      "step": 400
+    },
+    {
+      "epoch": 0.7522935779816514,
+      "grad_norm": 0.05117224529385567,
+      "learning_rate": 0.00016124158186777676,
+      "loss": 4.0246,
+      "step": 410
+    },
+    {
+      "epoch": 0.7706422018348624,
+      "grad_norm": 0.05750180035829544,
+      "learning_rate": 0.0001395337202205211,
+      "loss": 4.0188,
+      "step": 420
+    },
+    {
+      "epoch": 0.7889908256880734,
+      "grad_norm": 0.05148438736796379,
+      "learning_rate": 0.00011915646514489292,
+      "loss": 4.016,
+      "step": 430
+    },
+    {
+      "epoch": 0.8073394495412844,
+      "grad_norm": 0.0604918897151947,
+      "learning_rate": 0.00010018503618578818,
+      "loss": 4.0253,
+      "step": 440
+    },
+    {
+      "epoch": 0.8256880733944955,
+      "grad_norm": 0.05597606301307678,
+      "learning_rate": 8.268946349432583e-05,
+      "loss": 4.0162,
+      "step": 450
+    },
+    {
+      "epoch": 0.8440366972477065,
+      "grad_norm": 0.050633255392313004,
+      "learning_rate": 6.673432932217117e-05,
+      "loss": 4.0099,
+      "step": 460
+    },
+    {
+      "epoch": 0.8623853211009175,
+      "grad_norm": 0.04821436107158661,
+      "learning_rate": 5.2378529625954684e-05,
+      "loss": 3.9982,
+      "step": 470
+    },
+    {
+      "epoch": 0.8807339449541285,
+      "grad_norm": 0.05487411469221115,
+      "learning_rate": 3.9675056661785556e-05,
+      "loss": 4.0049,
+      "step": 480
+    },
+    {
+      "epoch": 0.8990825688073395,
+      "grad_norm": 0.048592761158943176,
+      "learning_rate": 2.8670803372379538e-05,
+      "loss": 4.0079,
+      "step": 490
+    },
+    {
+      "epoch": 0.9174311926605505,
+      "grad_norm": 0.04980834946036339,
+      "learning_rate": 1.9406390288876586e-05,
+      "loss": 4.0191,
+      "step": 500
+    },
+    {
+      "epoch": 0.9357798165137615,
+      "grad_norm": 0.04803913086652756,
+      "learning_rate": 1.191601558631461e-05,
+      "loss": 4.0082,
+      "step": 510
+    },
+    {
+      "epoch": 0.9541284403669725,
+      "grad_norm": 0.04756157100200653,
+      "learning_rate": 6.2273288462559305e-06,
+      "loss": 4.0034,
+      "step": 520
+    },
+    {
+      "epoch": 0.9724770642201835,
+      "grad_norm": 0.04829639941453934,
+      "learning_rate": 2.361328992552314e-06,
+      "loss": 4.0086,
+      "step": 530
+    },
+    {
+      "epoch": 0.9908256880733946,
+      "grad_norm": 0.04997587203979492,
+      "learning_rate": 3.322867770044202e-07,
+      "loss": 4.0047,
+      "step": 540
+    },
+    {
+      "epoch": 1.0,
+      "step": 545,
+      "total_flos": 2.7815808298962125e+17,
+      "train_loss": 4.2777741773412865,
+      "train_runtime": 5701.8772,
+      "train_samples_per_second": 97.885,
+      "train_steps_per_second": 0.096
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 545,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": false,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.7815808298962125e+17,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}

continuous/qwen3/mc2_s16_224/finetune/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
+}

continuous/qwen3/mc2_s16_224/finetune/config.json

@@ -0,0 +1,59 @@
+{
+  "add_faster_video": false,
+  "add_time_instruction": false,
+  "architectures": [
+    "LlavaQwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "eos_token_id": 151645,
+  "faster_token_stride": 10,
+  "force_sample": false,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 2048,
+  "image_aspect_ratio": "square",
+  "image_crop_resolution": null,
+  "image_grid_pinpoints": null,
+  "image_split_resolution": null,
+  "initializer_range": 0.02,
+  "intermediate_size": 6144,
+  "max_position_embeddings": 40960,
+  "max_window_layers": 28,
+  "mm_hidden_size": 384,
+  "mm_newline_position": "grid",
+  "mm_patch_merge_type": "flat",
+  "mm_projector_lr": 0.0001,
+  "mm_projector_type": "mlp2x_gelu",
+  "mm_resampler_type": null,
+  "mm_spatial_pool_mode": "bilinear",
+  "mm_spatial_pool_stride": null,
+  "mm_tunable_parts": "mm_mlp_adapter,mm_language_model",
+  "mm_use_im_patch_token": false,
+  "mm_use_im_start_end": false,
+  "mm_vision_select_feature": "patch",
+  "mm_vision_select_layer": -2,
+  "mm_vision_tower": "open_clip_hub:ViT-S-16",
+  "mm_vision_tower_lr": null,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pos_skipping_range": 4096,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000,
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "tokenizer_model_max_length": 8192,
+  "tokenizer_padding_side": "right",
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.51.3",
+  "use_cache": true,
+  "use_mm_proj": true,
+  "use_pos_skipping": false,
+  "use_sliding_window": false,
+  "vision_tower_pretrained": "/cache/ckpt/download/tokenizer/continuous/mc2/mc2_s16_224.pt",
+  "vocab_size": 151936
+}

continuous/qwen3/mc2_s16_224/finetune/generation_config.json

@@ -0,0 +1,13 @@
+{
+  "bos_token_id": 151643,
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "pad_token_id": 151643,
+  "temperature": 0.6,
+  "top_k": 20,
+  "top_p": 0.95,
+  "transformers_version": "4.51.3"
+}

continuous/qwen3/mc2_s16_224/finetune/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d08ec83b7c49761d9e8f411eeabad967de1efad5f89770123c1c9f18827573ae
+size 4127104248

continuous/qwen3/mc2_s16_224/finetune/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
+  }
+}

continuous/qwen3/mc2_s16_224/finetune/tokenizer.json

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

continuous/qwen3/mc2_s16_224/finetune/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 message.content is string 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.content is string %}\n        {%- set content = message.content %}\n    {%- else %}\n        {%- set content = '' %}\n    {%- endif %}\n    {%- if (message.role == \"user\") or (message.role == \"system\" and not loop.first) %}\n        {{- '<|im_start|>' + message.role + '\\n' + content + '<|im_end|>' + '\\n' }}\n    {%- elif message.role == \"assistant\" %}\n        {%- set reasoning_content = '' %}\n        {%- if message.reasoning_content is string %}\n            {%- set reasoning_content = message.reasoning_content %}\n        {%- else %}\n            {%- if '</think>' in content %}\n                {%- set reasoning_content = content.split('</think>')[0].rstrip('\\n').split('<think>')[-1].lstrip('\\n') %}\n                {%- set content = content.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        {{- 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": 8192,
+  "pad_token": "<|endoftext|>",
+  "padding_side": "right",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

continuous/qwen3/mc2_s16_224/finetune/trainer_state.json

@@ -0,0 +1,1856 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999037443449803,
+  "eval_steps": 500,
+  "global_step": 2597,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.003850226200789296,
+      "grad_norm": 24.930252075195312,
+      "learning_rate": 1.153846153846154e-06,
+      "loss": 2.5272,
+      "step": 10
+    },
+    {
+      "epoch": 0.007700452401578592,
+      "grad_norm": 8.364778518676758,
+      "learning_rate": 2.435897435897436e-06,
+      "loss": 2.0071,
+      "step": 20
+    },
+    {
+      "epoch": 0.01155067860236789,
+      "grad_norm": 4.828099250793457,
+      "learning_rate": 3.7179487179487184e-06,
+      "loss": 1.5223,
+      "step": 30
+    },
+    {
+      "epoch": 0.015400904803157185,
+      "grad_norm": 1.5815353393554688,
+      "learning_rate": 5e-06,
+      "loss": 1.173,
+      "step": 40
+    },
+    {
+      "epoch": 0.019251131003946483,
+      "grad_norm": 0.8073214888572693,
+      "learning_rate": 6.282051282051282e-06,
+      "loss": 1.0487,
+      "step": 50
+    },
+    {
+      "epoch": 0.02310135720473578,
+      "grad_norm": 0.431977778673172,
+      "learning_rate": 7.564102564102564e-06,
+      "loss": 0.9602,
+      "step": 60
+    },
+    {
+      "epoch": 0.026951583405525073,
+      "grad_norm": 0.3276802599430084,
+      "learning_rate": 8.846153846153847e-06,
+      "loss": 0.9109,
+      "step": 70
+    },
+    {
+      "epoch": 0.03080180960631437,
+      "grad_norm": 0.3161572217941284,
+      "learning_rate": 9.999996111488723e-06,
+      "loss": 0.9083,
+      "step": 80
+    },
+    {
+      "epoch": 0.034652035807103666,
+      "grad_norm": 0.28097808361053467,
+      "learning_rate": 9.999529497453782e-06,
+      "loss": 0.8762,
+      "step": 90
+    },
+    {
+      "epoch": 0.038502262007892966,
+      "grad_norm": 0.26152050495147705,
+      "learning_rate": 9.998285264324193e-06,
+      "loss": 0.8564,
+      "step": 100
+    },
+    {
+      "epoch": 0.04235248820868226,
+      "grad_norm": 0.265025794506073,
+      "learning_rate": 9.996263605626052e-06,
+      "loss": 0.8395,
+      "step": 110
+    },
+    {
+      "epoch": 0.04620271440947156,
+      "grad_norm": 0.26058533787727356,
+      "learning_rate": 9.99346483580503e-06,
+      "loss": 0.8484,
+      "step": 120
+    },
+    {
+      "epoch": 0.05005294061026085,
+      "grad_norm": 0.2646158039569855,
+      "learning_rate": 9.989889390177467e-06,
+      "loss": 0.844,
+      "step": 130
+    },
+    {
+      "epoch": 0.053903166811050146,
+      "grad_norm": 0.2930638790130615,
+      "learning_rate": 9.985537824862647e-06,
+      "loss": 0.8389,
+      "step": 140
+    },
+    {
+      "epoch": 0.057753393011839446,
+      "grad_norm": 0.30196458101272583,
+      "learning_rate": 9.980410816696317e-06,
+      "loss": 0.8243,
+      "step": 150
+    },
+    {
+      "epoch": 0.06160361921262874,
+      "grad_norm": 0.287474125623703,
+      "learning_rate": 9.97450916312541e-06,
+      "loss": 0.8317,
+      "step": 160
+    },
+    {
+      "epoch": 0.06545384541341803,
+      "grad_norm": 0.2336154282093048,
+      "learning_rate": 9.967833782084004e-06,
+      "loss": 0.8324,
+      "step": 170
+    },
+    {
+      "epoch": 0.06930407161420733,
+      "grad_norm": 0.27090609073638916,
+      "learning_rate": 9.960385711850554e-06,
+      "loss": 0.8577,
+      "step": 180
+    },
+    {
+      "epoch": 0.07315429781499663,
+      "grad_norm": 0.2528645694255829,
+      "learning_rate": 9.952166110886394e-06,
+      "loss": 0.8484,
+      "step": 190
+    },
+    {
+      "epoch": 0.07700452401578593,
+      "grad_norm": 0.29072919487953186,
+      "learning_rate": 9.943176257655567e-06,
+      "loss": 0.8363,
+      "step": 200
+    },
+    {
+      "epoch": 0.08085475021657522,
+      "grad_norm": 0.25932714343070984,
+      "learning_rate": 9.933417550425952e-06,
+      "loss": 0.8348,
+      "step": 210
+    },
+    {
+      "epoch": 0.08470497641736452,
+      "grad_norm": 0.2577572762966156,
+      "learning_rate": 9.9228915070518e-06,
+      "loss": 0.8187,
+      "step": 220
+    },
+    {
+      "epoch": 0.08855520261815382,
+      "grad_norm": 0.2519455552101135,
+      "learning_rate": 9.911599764737632e-06,
+      "loss": 0.82,
+      "step": 230
+    },
+    {
+      "epoch": 0.09240542881894312,
+      "grad_norm": 0.25836172699928284,
+      "learning_rate": 9.899544079783606e-06,
+      "loss": 0.8145,
+      "step": 240
+    },
+    {
+      "epoch": 0.0962556550197324,
+      "grad_norm": 0.29591765999794006,
+      "learning_rate": 9.886726327312338e-06,
+      "loss": 0.8327,
+      "step": 250
+    },
+    {
+      "epoch": 0.1001058812205217,
+      "grad_norm": 0.24245598912239075,
+      "learning_rate": 9.87314850097724e-06,
+      "loss": 0.8272,
+      "step": 260
+    },
+    {
+      "epoch": 0.103956107421311,
+      "grad_norm": 0.2511278986930847,
+      "learning_rate": 9.858812712652452e-06,
+      "loss": 0.817,
+      "step": 270
+    },
+    {
+      "epoch": 0.10780633362210029,
+      "grad_norm": 0.24759146571159363,
+      "learning_rate": 9.843721192104341e-06,
+      "loss": 0.8156,
+      "step": 280
+    },
+    {
+      "epoch": 0.11165655982288959,
+      "grad_norm": 0.24512644112110138,
+      "learning_rate": 9.827876286644698e-06,
+      "loss": 0.8339,
+      "step": 290
+    },
+    {
+      "epoch": 0.11550678602367889,
+      "grad_norm": 0.22913429141044617,
+      "learning_rate": 9.811280460765643e-06,
+      "loss": 0.7955,
+      "step": 300
+    },
+    {
+      "epoch": 0.11935701222446819,
+      "grad_norm": 0.23896601796150208,
+      "learning_rate": 9.793936295756292e-06,
+      "loss": 0.8081,
+      "step": 310
+    },
+    {
+      "epoch": 0.12320723842525748,
+      "grad_norm": 0.24847586452960968,
+      "learning_rate": 9.775846489301279e-06,
+      "loss": 0.8089,
+      "step": 320
+    },
+    {
+      "epoch": 0.12705746462604678,
+      "grad_norm": 0.2601083219051361,
+      "learning_rate": 9.757013855061157e-06,
+      "loss": 0.8082,
+      "step": 330
+    },
+    {
+      "epoch": 0.13090769082683607,
+      "grad_norm": 0.2637288272380829,
+      "learning_rate": 9.737441322234758e-06,
+      "loss": 0.8198,
+      "step": 340
+    },
+    {
+      "epoch": 0.13475791702762538,
+      "grad_norm": 0.27966251969337463,
+      "learning_rate": 9.717131935103605e-06,
+      "loss": 0.8204,
+      "step": 350
+    },
+    {
+      "epoch": 0.13860814322841467,
+      "grad_norm": 0.2612977921962738,
+      "learning_rate": 9.696088852558398e-06,
+      "loss": 0.8077,
+      "step": 360
+    },
+    {
+      "epoch": 0.14245836942920398,
+      "grad_norm": 0.2601540982723236,
+      "learning_rate": 9.674315347607686e-06,
+      "loss": 0.8338,
+      "step": 370
+    },
+    {
+      "epoch": 0.14630859562999327,
+      "grad_norm": 0.24349334836006165,
+      "learning_rate": 9.651814806868796e-06,
+      "loss": 0.7942,
+      "step": 380
+    },
+    {
+      "epoch": 0.15015882183078255,
+      "grad_norm": 0.2400379627943039,
+      "learning_rate": 9.62859073004107e-06,
+      "loss": 0.7985,
+      "step": 390
+    },
+    {
+      "epoch": 0.15400904803157187,
+      "grad_norm": 0.2475101351737976,
+      "learning_rate": 9.604646729361547e-06,
+      "loss": 0.8007,
+      "step": 400
+    },
+    {
+      "epoch": 0.15785927423236115,
+      "grad_norm": 0.24037882685661316,
+      "learning_rate": 9.579986529043099e-06,
+      "loss": 0.7896,
+      "step": 410
+    },
+    {
+      "epoch": 0.16170950043315044,
+      "grad_norm": 0.245247021317482,
+      "learning_rate": 9.554613964695189e-06,
+      "loss": 0.8076,
+      "step": 420
+    },
+    {
+      "epoch": 0.16555972663393975,
+      "grad_norm": 0.24911640584468842,
+      "learning_rate": 9.528532982727281e-06,
+      "loss": 0.7998,
+      "step": 430
+    },
+    {
+      "epoch": 0.16940995283472904,
+      "grad_norm": 0.25109395384788513,
+      "learning_rate": 9.501747639735018e-06,
+      "loss": 0.7949,
+      "step": 440
+    },
+    {
+      "epoch": 0.17326017903551832,
+      "grad_norm": 0.25223881006240845,
+      "learning_rate": 9.474262101869272e-06,
+      "loss": 0.7848,
+      "step": 450
+    },
+    {
+      "epoch": 0.17711040523630764,
+      "grad_norm": 0.24371950328350067,
+      "learning_rate": 9.446080644188141e-06,
+      "loss": 0.7924,
+      "step": 460
+    },
+    {
+      "epoch": 0.18096063143709692,
+      "grad_norm": 0.24263577163219452,
+      "learning_rate": 9.417207649992017e-06,
+      "loss": 0.797,
+      "step": 470
+    },
+    {
+      "epoch": 0.18481085763788624,
+      "grad_norm": 0.2421039193868637,
+      "learning_rate": 9.387647610141816e-06,
+      "loss": 0.7961,
+      "step": 480
+    },
+    {
+      "epoch": 0.18866108383867553,
+      "grad_norm": 0.2329459935426712,
+      "learning_rate": 9.35740512236047e-06,
+      "loss": 0.7833,
+      "step": 490
+    },
+    {
+      "epoch": 0.1925113100394648,
+      "grad_norm": 0.23070144653320312,
+      "learning_rate": 9.326484890517809e-06,
+      "loss": 0.7906,
+      "step": 500
+    },
+    {
+      "epoch": 0.19636153624025413,
+      "grad_norm": 0.23558636009693146,
+      "learning_rate": 9.294891723898926e-06,
+      "loss": 0.7834,
+      "step": 510
+    },
+    {
+      "epoch": 0.2002117624410434,
+      "grad_norm": 0.2453930526971817,
+      "learning_rate": 9.262630536456153e-06,
+      "loss": 0.8142,
+      "step": 520
+    },
+    {
+      "epoch": 0.2040619886418327,
+      "grad_norm": 0.25539153814315796,
+      "learning_rate": 9.229706346044749e-06,
+      "loss": 0.7854,
+      "step": 530
+    },
+    {
+      "epoch": 0.207912214842622,
+      "grad_norm": 0.2589377760887146,
+      "learning_rate": 9.196124273642422e-06,
+      "loss": 0.8001,
+      "step": 540
+    },
+    {
+      "epoch": 0.2117624410434113,
+      "grad_norm": 0.23099522292613983,
+      "learning_rate": 9.161889542552842e-06,
+      "loss": 0.804,
+      "step": 550
+    },
+    {
+      "epoch": 0.21561266724420058,
+      "grad_norm": 0.22399269044399261,
+      "learning_rate": 9.127007477593193e-06,
+      "loss": 0.8055,
+      "step": 560
+    },
+    {
+      "epoch": 0.2194628934449899,
+      "grad_norm": 0.26366788148880005,
+      "learning_rate": 9.091483504265974e-06,
+      "loss": 0.8129,
+      "step": 570
+    },
+    {
+      "epoch": 0.22331311964577918,
+      "grad_norm": 0.25229495763778687,
+      "learning_rate": 9.055323147915116e-06,
+      "loss": 0.7953,
+      "step": 580
+    },
+    {
+      "epoch": 0.2271633458465685,
+      "grad_norm": 0.2591622471809387,
+      "learning_rate": 9.018532032866585e-06,
+      "loss": 0.8036,
+      "step": 590
+    },
+    {
+      "epoch": 0.23101357204735778,
+      "grad_norm": 0.23078176379203796,
+      "learning_rate": 8.981115881553579e-06,
+      "loss": 0.7953,
+      "step": 600
+    },
+    {
+      "epoch": 0.23486379824814707,
+      "grad_norm": 0.25762850046157837,
+      "learning_rate": 8.943080513626476e-06,
+      "loss": 0.8162,
+      "step": 610
+    },
+    {
+      "epoch": 0.23871402444893638,
+      "grad_norm": 0.23695436120033264,
+      "learning_rate": 8.904431845047641e-06,
+      "loss": 0.8003,
+      "step": 620
+    },
+    {
+      "epoch": 0.24256425064972567,
+      "grad_norm": 0.25825950503349304,
+      "learning_rate": 8.86517588717128e-06,
+      "loss": 0.8165,
+      "step": 630
+    },
+    {
+      "epoch": 0.24641447685051496,
+      "grad_norm": 0.24561099708080292,
+      "learning_rate": 8.82531874580844e-06,
+      "loss": 0.808,
+      "step": 640
+    },
+    {
+      "epoch": 0.25026470305130427,
+      "grad_norm": 0.22692140936851501,
+      "learning_rate": 8.784866620277314e-06,
+      "loss": 0.7645,
+      "step": 650
+    },
+    {
+      "epoch": 0.25411492925209356,
+      "grad_norm": 0.2530440092086792,
+      "learning_rate": 8.743825802439018e-06,
+      "loss": 0.7985,
+      "step": 660
+    },
+    {
+      "epoch": 0.25796515545288284,
+      "grad_norm": 0.25808992981910706,
+      "learning_rate": 8.702202675718956e-06,
+      "loss": 0.807,
+      "step": 670
+    },
+    {
+      "epoch": 0.26181538165367213,
+      "grad_norm": 0.23839479684829712,
+      "learning_rate": 8.660003714113959e-06,
+      "loss": 0.7744,
+      "step": 680
+    },
+    {
+      "epoch": 0.26566560785446147,
+      "grad_norm": 0.24054235219955444,
+      "learning_rate": 8.617235481185323e-06,
+      "loss": 0.7979,
+      "step": 690
+    },
+    {
+      "epoch": 0.26951583405525076,
+      "grad_norm": 0.25057265162467957,
+      "learning_rate": 8.573904629037931e-06,
+      "loss": 0.8046,
+      "step": 700
+    },
+    {
+      "epoch": 0.27336606025604004,
+      "grad_norm": 0.24375540018081665,
+      "learning_rate": 8.530017897285585e-06,
+      "loss": 0.7866,
+      "step": 710
+    },
+    {
+      "epoch": 0.27721628645682933,
+      "grad_norm": 0.21812061965465546,
+      "learning_rate": 8.485582112002746e-06,
+      "loss": 0.7906,
+      "step": 720
+    },
+    {
+      "epoch": 0.2810665126576186,
+      "grad_norm": 0.23558159172534943,
+      "learning_rate": 8.440604184662811e-06,
+      "loss": 0.7945,
+      "step": 730
+    },
+    {
+      "epoch": 0.28491673885840796,
+      "grad_norm": 0.24016810953617096,
+      "learning_rate": 8.395091111063118e-06,
+      "loss": 0.7818,
+      "step": 740
+    },
+    {
+      "epoch": 0.28876696505919724,
+      "grad_norm": 0.2288990616798401,
+      "learning_rate": 8.349049970236822e-06,
+      "loss": 0.7831,
+      "step": 750
+    },
+    {
+      "epoch": 0.29261719125998653,
+      "grad_norm": 0.24293552339076996,
+      "learning_rate": 8.302487923351842e-06,
+      "loss": 0.7755,
+      "step": 760
+    },
+    {
+      "epoch": 0.2964674174607758,
+      "grad_norm": 0.242442324757576,
+      "learning_rate": 8.255412212597021e-06,
+      "loss": 0.7931,
+      "step": 770
+    },
+    {
+      "epoch": 0.3003176436615651,
+      "grad_norm": 0.23354770243167877,
+      "learning_rate": 8.207830160055688e-06,
+      "loss": 0.7913,
+      "step": 780
+    },
+    {
+      "epoch": 0.3041678698623544,
+      "grad_norm": 0.24598103761672974,
+      "learning_rate": 8.159749166566786e-06,
+      "loss": 0.7734,
+      "step": 790
+    },
+    {
+      "epoch": 0.30801809606314373,
+      "grad_norm": 0.23429104685783386,
+      "learning_rate": 8.11117671057377e-06,
+      "loss": 0.8023,
+      "step": 800
+    },
+    {
+      "epoch": 0.311868322263933,
+      "grad_norm": 0.2554655075073242,
+      "learning_rate": 8.062120346961404e-06,
+      "loss": 0.7937,
+      "step": 810
+    },
+    {
+      "epoch": 0.3157185484647223,
+      "grad_norm": 0.2309216409921646,
+      "learning_rate": 8.012587705880705e-06,
+      "loss": 0.787,
+      "step": 820
+    },
+    {
+      "epoch": 0.3195687746655116,
+      "grad_norm": 0.23415620625019073,
+      "learning_rate": 7.962586491562146e-06,
+      "loss": 0.7724,
+      "step": 830
+    },
+    {
+      "epoch": 0.3234190008663009,
+      "grad_norm": 0.24677950143814087,
+      "learning_rate": 7.912124481117357e-06,
+      "loss": 0.7786,
+      "step": 840
+    },
+    {
+      "epoch": 0.3272692270670902,
+      "grad_norm": 0.2464367002248764,
+      "learning_rate": 7.861209523329494e-06,
+      "loss": 0.7918,
+      "step": 850
+    },
+    {
+      "epoch": 0.3311194532678795,
+      "grad_norm": 0.2307090014219284,
+      "learning_rate": 7.809849537432432e-06,
+      "loss": 0.7724,
+      "step": 860
+    },
+    {
+      "epoch": 0.3349696794686688,
+      "grad_norm": 0.23632977902889252,
+      "learning_rate": 7.758052511879035e-06,
+      "loss": 0.7835,
+      "step": 870
+    },
+    {
+      "epoch": 0.3388199056694581,
+      "grad_norm": 0.2247433364391327,
+      "learning_rate": 7.705826503098642e-06,
+      "loss": 0.7794,
+      "step": 880
+    },
+    {
+      "epoch": 0.34267013187024736,
+      "grad_norm": 0.23233595490455627,
+      "learning_rate": 7.653179634243979e-06,
+      "loss": 0.779,
+      "step": 890
+    },
+    {
+      "epoch": 0.34652035807103665,
+      "grad_norm": 0.21591071784496307,
+      "learning_rate": 7.600120093927696e-06,
+      "loss": 0.7836,
+      "step": 900
+    },
+    {
+      "epoch": 0.350370584271826,
+      "grad_norm": 0.23742516338825226,
+      "learning_rate": 7.546656134948729e-06,
+      "loss": 0.7725,
+      "step": 910
+    },
+    {
+      "epoch": 0.3542208104726153,
+      "grad_norm": 0.23720425367355347,
+      "learning_rate": 7.492796073008661e-06,
+      "loss": 0.7734,
+      "step": 920
+    },
+    {
+      "epoch": 0.35807103667340456,
+      "grad_norm": 0.25946494936943054,
+      "learning_rate": 7.438548285418326e-06,
+      "loss": 0.759,
+      "step": 930
+    },
+    {
+      "epoch": 0.36192126287419385,
+      "grad_norm": 0.26092779636383057,
+      "learning_rate": 7.3839212097948e-06,
+      "loss": 0.7926,
+      "step": 940
+    },
+    {
+      "epoch": 0.36577148907498314,
+      "grad_norm": 0.24079164862632751,
+      "learning_rate": 7.328923342749039e-06,
+      "loss": 0.8087,
+      "step": 950
+    },
+    {
+      "epoch": 0.3696217152757725,
+      "grad_norm": 0.2565212845802307,
+      "learning_rate": 7.273563238564319e-06,
+      "loss": 0.7925,
+      "step": 960
+    },
+    {
+      "epoch": 0.37347194147656176,
+      "grad_norm": 0.24005284905433655,
+      "learning_rate": 7.217849507865724e-06,
+      "loss": 0.7715,
+      "step": 970
+    },
+    {
+      "epoch": 0.37732216767735105,
+      "grad_norm": 0.24328698217868805,
+      "learning_rate": 7.161790816280858e-06,
+      "loss": 0.7788,
+      "step": 980
+    },
+    {
+      "epoch": 0.38117239387814034,
+      "grad_norm": 0.25015416741371155,
+      "learning_rate": 7.1053958830920125e-06,
+      "loss": 0.762,
+      "step": 990
+    },
+    {
+      "epoch": 0.3850226200789296,
+      "grad_norm": 0.24694310128688812,
+      "learning_rate": 7.0486734798799715e-06,
+      "loss": 0.7818,
+      "step": 1000
+    },
+    {
+      "epoch": 0.3888728462797189,
+      "grad_norm": 0.24083629250526428,
+      "learning_rate": 6.991632429159708e-06,
+      "loss": 0.8043,
+      "step": 1010
+    },
+    {
+      "epoch": 0.39272307248050825,
+      "grad_norm": 0.26479294896125793,
+      "learning_rate": 6.934281603008132e-06,
+      "loss": 0.7951,
+      "step": 1020
+    },
+    {
+      "epoch": 0.39657329868129754,
+      "grad_norm": 0.24434252083301544,
+      "learning_rate": 6.876629921684143e-06,
+      "loss": 0.7922,
+      "step": 1030
+    },
+    {
+      "epoch": 0.4004235248820868,
+      "grad_norm": 0.24188467860221863,
+      "learning_rate": 6.818686352241199e-06,
+      "loss": 0.7826,
+      "step": 1040
+    },
+    {
+      "epoch": 0.4042737510828761,
+      "grad_norm": 0.22415053844451904,
+      "learning_rate": 6.760459907132573e-06,
+      "loss": 0.7784,
+      "step": 1050
+    },
+    {
+      "epoch": 0.4081239772836654,
+      "grad_norm": 0.23733168840408325,
+      "learning_rate": 6.701959642809589e-06,
+      "loss": 0.784,
+      "step": 1060
+    },
+    {
+      "epoch": 0.41197420348445474,
+      "grad_norm": 0.24528081715106964,
+      "learning_rate": 6.643194658312987e-06,
+      "loss": 0.7769,
+      "step": 1070
+    },
+    {
+      "epoch": 0.415824429685244,
+      "grad_norm": 0.2333805114030838,
+      "learning_rate": 6.584174093857676e-06,
+      "loss": 0.8001,
+      "step": 1080
+    },
+    {
+      "epoch": 0.4196746558860333,
+      "grad_norm": 0.27196750044822693,
+      "learning_rate": 6.5249071294110735e-06,
+      "loss": 0.7826,
+      "step": 1090
+    },
+    {
+      "epoch": 0.4235248820868226,
+      "grad_norm": 0.24602298438549042,
+      "learning_rate": 6.465402983265274e-06,
+      "loss": 0.7961,
+      "step": 1100
+    },
+    {
+      "epoch": 0.4273751082876119,
+      "grad_norm": 0.24273204803466797,
+      "learning_rate": 6.405670910603243e-06,
+      "loss": 0.792,
+      "step": 1110
+    },
+    {
+      "epoch": 0.43122533448840117,
+      "grad_norm": 0.2729332149028778,
+      "learning_rate": 6.345720202059291e-06,
+      "loss": 0.7877,
+      "step": 1120
+    },
+    {
+      "epoch": 0.4350755606891905,
+      "grad_norm": 0.24860818684101105,
+      "learning_rate": 6.285560182274008e-06,
+      "loss": 0.7654,
+      "step": 1130
+    },
+    {
+      "epoch": 0.4389257868899798,
+      "grad_norm": 0.23598256707191467,
+      "learning_rate": 6.225200208443936e-06,
+      "loss": 0.7886,
+      "step": 1140
+    },
+    {
+      "epoch": 0.4427760130907691,
+      "grad_norm": 0.2242962270975113,
+      "learning_rate": 6.16464966886616e-06,
+      "loss": 0.7774,
+      "step": 1150
+    },
+    {
+      "epoch": 0.44662623929155837,
+      "grad_norm": 0.27243927121162415,
+      "learning_rate": 6.103917981478065e-06,
+      "loss": 0.8021,
+      "step": 1160
+    },
+    {
+      "epoch": 0.45047646549234766,
+      "grad_norm": 0.234565868973732,
+      "learning_rate": 6.043014592392489e-06,
+      "loss": 0.8121,
+      "step": 1170
+    },
+    {
+      "epoch": 0.454326691693137,
+      "grad_norm": 0.22716280817985535,
+      "learning_rate": 5.981948974428485e-06,
+      "loss": 0.7876,
+      "step": 1180
+    },
+    {
+      "epoch": 0.4581769178939263,
+      "grad_norm": 0.2523665428161621,
+      "learning_rate": 5.920730625637934e-06,
+      "loss": 0.7652,
+      "step": 1190
+    },
+    {
+      "epoch": 0.46202714409471557,
+      "grad_norm": 0.2453547567129135,
+      "learning_rate": 5.8593690678282355e-06,
+      "loss": 0.7782,
+      "step": 1200
+    },
+    {
+      "epoch": 0.46587737029550486,
+      "grad_norm": 0.2525818645954132,
+      "learning_rate": 5.797873845081301e-06,
+      "loss": 0.7824,
+      "step": 1210
+    },
+    {
+      "epoch": 0.46972759649629414,
+      "grad_norm": 0.2391364574432373,
+      "learning_rate": 5.736254522269075e-06,
+      "loss": 0.7981,
+      "step": 1220
+    },
+    {
+      "epoch": 0.47357782269708343,
+      "grad_norm": 0.24407285451889038,
+      "learning_rate": 5.6745206835658386e-06,
+      "loss": 0.8002,
+      "step": 1230
+    },
+    {
+      "epoch": 0.47742804889787277,
+      "grad_norm": 0.23029199242591858,
+      "learning_rate": 5.612681930957495e-06,
+      "loss": 0.7718,
+      "step": 1240
+    },
+    {
+      "epoch": 0.48127827509866206,
+      "grad_norm": 0.23474030196666718,
+      "learning_rate": 5.550747882748095e-06,
+      "loss": 0.7689,
+      "step": 1250
+    },
+    {
+      "epoch": 0.48512850129945134,
+      "grad_norm": 0.2509559094905853,
+      "learning_rate": 5.4887281720638195e-06,
+      "loss": 0.7766,
+      "step": 1260
+    },
+    {
+      "epoch": 0.48897872750024063,
+      "grad_norm": 0.26771098375320435,
+      "learning_rate": 5.42663244535466e-06,
+      "loss": 0.7601,
+      "step": 1270
+    },
+    {
+      "epoch": 0.4928289537010299,
+      "grad_norm": 0.2531026601791382,
+      "learning_rate": 5.3644703608940205e-06,
+      "loss": 0.769,
+      "step": 1280
+    },
+    {
+      "epoch": 0.49667917990181926,
+      "grad_norm": 0.24358202517032623,
+      "learning_rate": 5.302251587276491e-06,
+      "loss": 0.7803,
+      "step": 1290
+    },
+    {
+      "epoch": 0.5005294061026085,
+      "grad_norm": 0.2326783388853073,
+      "learning_rate": 5.2399858019140005e-06,
+      "loss": 0.7809,
+      "step": 1300
+    },
+    {
+      "epoch": 0.5043796323033978,
+      "grad_norm": 0.2279393970966339,
+      "learning_rate": 5.177682689530615e-06,
+      "loss": 0.7961,
+      "step": 1310
+    },
+    {
+      "epoch": 0.5082298585041871,
+      "grad_norm": 0.2483544498682022,
+      "learning_rate": 5.11535194065619e-06,
+      "loss": 0.7621,
+      "step": 1320
+    },
+    {
+      "epoch": 0.5120800847049765,
+      "grad_norm": 0.2608610987663269,
+      "learning_rate": 5.053003250119113e-06,
+      "loss": 0.7725,
+      "step": 1330
+    },
+    {
+      "epoch": 0.5159303109057657,
+      "grad_norm": 0.23802487552165985,
+      "learning_rate": 4.9906463155383925e-06,
+      "loss": 0.7805,
+      "step": 1340
+    },
+    {
+      "epoch": 0.519780537106555,
+      "grad_norm": 0.23979498445987701,
+      "learning_rate": 4.9282908358153046e-06,
+      "loss": 0.7748,
+      "step": 1350
+    },
+    {
+      "epoch": 0.5236307633073443,
+      "grad_norm": 0.26058897376060486,
+      "learning_rate": 4.8659465096248345e-06,
+      "loss": 0.79,
+      "step": 1360
+    },
+    {
+      "epoch": 0.5274809895081336,
+      "grad_norm": 0.2527788281440735,
+      "learning_rate": 4.803623033907169e-06,
+      "loss": 0.7887,
+      "step": 1370
+    },
+    {
+      "epoch": 0.5313312157089229,
+      "grad_norm": 0.23414915800094604,
+      "learning_rate": 4.741330102359441e-06,
+      "loss": 0.7945,
+      "step": 1380
+    },
+    {
+      "epoch": 0.5351814419097122,
+      "grad_norm": 0.2545657157897949,
+      "learning_rate": 4.679077403927996e-06,
+      "loss": 0.7757,
+      "step": 1390
+    },
+    {
+      "epoch": 0.5390316681105015,
+      "grad_norm": 0.23327654600143433,
+      "learning_rate": 4.616874621301373e-06,
+      "loss": 0.7745,
+      "step": 1400
+    },
+    {
+      "epoch": 0.5428818943112907,
+      "grad_norm": 0.24724364280700684,
+      "learning_rate": 4.554731429404293e-06,
+      "loss": 0.789,
+      "step": 1410
+    },
+    {
+      "epoch": 0.5467321205120801,
+      "grad_norm": 0.26878786087036133,
+      "learning_rate": 4.492657493892822e-06,
+      "loss": 0.7724,
+      "step": 1420
+    },
+    {
+      "epoch": 0.5505823467128694,
+      "grad_norm": 0.2753806412220001,
+      "learning_rate": 4.430662469650999e-06,
+      "loss": 0.7882,
+      "step": 1430
+    },
+    {
+      "epoch": 0.5544325729136587,
+      "grad_norm": 0.25860077142715454,
+      "learning_rate": 4.368755999289118e-06,
+      "loss": 0.7919,
+      "step": 1440
+    },
+    {
+      "epoch": 0.558282799114448,
+      "grad_norm": 0.26412490010261536,
+      "learning_rate": 4.306947711643949e-06,
+      "loss": 0.7759,
+      "step": 1450
+    },
+    {
+      "epoch": 0.5621330253152372,
+      "grad_norm": 0.25285953283309937,
+      "learning_rate": 4.245247220281068e-06,
+      "loss": 0.7799,
+      "step": 1460
+    },
+    {
+      "epoch": 0.5659832515160266,
+      "grad_norm": 0.2405516356229782,
+      "learning_rate": 4.183664121999587e-06,
+      "loss": 0.7764,
+      "step": 1470
+    },
+    {
+      "epoch": 0.5698334777168159,
+      "grad_norm": 0.2483542114496231,
+      "learning_rate": 4.122207995339474e-06,
+      "loss": 0.7746,
+      "step": 1480
+    },
+    {
+      "epoch": 0.5736837039176051,
+      "grad_norm": 0.23140136897563934,
+      "learning_rate": 4.060888399091734e-06,
+      "loss": 0.7783,
+      "step": 1490
+    },
+    {
+      "epoch": 0.5775339301183945,
+      "grad_norm": 0.24071085453033447,
+      "learning_rate": 3.999714870811639e-06,
+      "loss": 0.7879,
+      "step": 1500
+    },
+    {
+      "epoch": 0.5813841563191837,
+      "grad_norm": 0.23427248001098633,
+      "learning_rate": 3.938696925335275e-06,
+      "loss": 0.7804,
+      "step": 1510
+    },
+    {
+      "epoch": 0.5852343825199731,
+      "grad_norm": 0.24083785712718964,
+      "learning_rate": 3.87784405329962e-06,
+      "loss": 0.7882,
+      "step": 1520
+    },
+    {
+      "epoch": 0.5890846087207623,
+      "grad_norm": 0.2424783706665039,
+      "learning_rate": 3.817165719666382e-06,
+      "loss": 0.7902,
+      "step": 1530
+    },
+    {
+      "epoch": 0.5929348349215516,
+      "grad_norm": 0.23882929980754852,
+      "learning_rate": 3.756671362249841e-06,
+      "loss": 0.7939,
+      "step": 1540
+    },
+    {
+      "epoch": 0.596785061122341,
+      "grad_norm": 0.24228492379188538,
+      "learning_rate": 3.6963703902488907e-06,
+      "loss": 0.7802,
+      "step": 1550
+    },
+    {
+      "epoch": 0.6006352873231302,
+      "grad_norm": 0.24791893362998962,
+      "learning_rate": 3.6362721827835655e-06,
+      "loss": 0.7964,
+      "step": 1560
+    },
+    {
+      "epoch": 0.6044855135239195,
+      "grad_norm": 0.25479963421821594,
+      "learning_rate": 3.576386087436212e-06,
+      "loss": 0.7923,
+      "step": 1570
+    },
+    {
+      "epoch": 0.6083357397247088,
+      "grad_norm": 0.256637305021286,
+      "learning_rate": 3.5167214187975913e-06,
+      "loss": 0.7823,
+      "step": 1580
+    },
+    {
+      "epoch": 0.6121859659254981,
+      "grad_norm": 0.212455615401268,
+      "learning_rate": 3.457287457018086e-06,
+      "loss": 0.7859,
+      "step": 1590
+    },
+    {
+      "epoch": 0.6160361921262875,
+      "grad_norm": 0.23600450158119202,
+      "learning_rate": 3.3980934463643003e-06,
+      "loss": 0.7726,
+      "step": 1600
+    },
+    {
+      "epoch": 0.6198864183270767,
+      "grad_norm": 0.22204211354255676,
+      "learning_rate": 3.3391485937812053e-06,
+      "loss": 0.7559,
+      "step": 1610
+    },
+    {
+      "epoch": 0.623736644527866,
+      "grad_norm": 0.2446575164794922,
+      "learning_rate": 3.280462067460115e-06,
+      "loss": 0.7541,
+      "step": 1620
+    },
+    {
+      "epoch": 0.6275868707286553,
+      "grad_norm": 0.23379679024219513,
+      "learning_rate": 3.222042995412669e-06,
+      "loss": 0.7858,
+      "step": 1630
+    },
+    {
+      "epoch": 0.6314370969294446,
+      "grad_norm": 0.2182314246892929,
+      "learning_rate": 3.163900464051089e-06,
+      "loss": 0.7753,
+      "step": 1640
+    },
+    {
+      "epoch": 0.635287323130234,
+      "grad_norm": 0.23608042299747467,
+      "learning_rate": 3.1060435167748826e-06,
+      "loss": 0.7857,
+      "step": 1650
+    },
+    {
+      "epoch": 0.6391375493310232,
+      "grad_norm": 0.23846349120140076,
+      "learning_rate": 3.048481152564253e-06,
+      "loss": 0.7654,
+      "step": 1660
+    },
+    {
+      "epoch": 0.6429877755318125,
+      "grad_norm": 0.2520001530647278,
+      "learning_rate": 2.9912223245804017e-06,
+      "loss": 0.7885,
+      "step": 1670
+    },
+    {
+      "epoch": 0.6468380017326018,
+      "grad_norm": 0.24394695460796356,
+      "learning_rate": 2.934275938772981e-06,
+      "loss": 0.7823,
+      "step": 1680
+    },
+    {
+      "epoch": 0.6506882279333911,
+      "grad_norm": 0.27288705110549927,
+      "learning_rate": 2.8776508524948676e-06,
+      "loss": 0.7734,
+      "step": 1690
+    },
+    {
+      "epoch": 0.6545384541341804,
+      "grad_norm": 0.21876105666160583,
+      "learning_rate": 2.821355873124503e-06,
+      "loss": 0.7707,
+      "step": 1700
+    },
+    {
+      "epoch": 0.6583886803349697,
+      "grad_norm": 0.24807816743850708,
+      "learning_rate": 2.7653997566960144e-06,
+      "loss": 0.7865,
+      "step": 1710
+    },
+    {
+      "epoch": 0.662238906535759,
+      "grad_norm": 0.25018468499183655,
+      "learning_rate": 2.709791206537309e-06,
+      "loss": 0.7749,
+      "step": 1720
+    },
+    {
+      "epoch": 0.6660891327365482,
+      "grad_norm": 0.24278303980827332,
+      "learning_rate": 2.6545388719163747e-06,
+      "loss": 0.7813,
+      "step": 1730
+    },
+    {
+      "epoch": 0.6699393589373376,
+      "grad_norm": 0.24660739302635193,
+      "learning_rate": 2.599651346695979e-06,
+      "loss": 0.7798,
+      "step": 1740
+    },
+    {
+      "epoch": 0.6737895851381268,
+      "grad_norm": 0.24126556515693665,
+      "learning_rate": 2.545137167997007e-06,
+      "loss": 0.776,
+      "step": 1750
+    },
+    {
+      "epoch": 0.6776398113389162,
+      "grad_norm": 0.22667951881885529,
+      "learning_rate": 2.4910048148705972e-06,
+      "loss": 0.7634,
+      "step": 1760
+    },
+    {
+      "epoch": 0.6814900375397055,
+      "grad_norm": 0.26011061668395996,
+      "learning_rate": 2.4372627069793327e-06,
+      "loss": 0.8075,
+      "step": 1770
+    },
+    {
+      "epoch": 0.6853402637404947,
+      "grad_norm": 0.236318901181221,
+      "learning_rate": 2.3839192032876486e-06,
+      "loss": 0.7928,
+      "step": 1780
+    },
+    {
+      "epoch": 0.6891904899412841,
+      "grad_norm": 0.23253974318504333,
+      "learning_rate": 2.3309826007617047e-06,
+      "loss": 0.8041,
+      "step": 1790
+    },
+    {
+      "epoch": 0.6930407161420733,
+      "grad_norm": 0.23733313381671906,
+      "learning_rate": 2.278461133078881e-06,
+      "loss": 0.7685,
+      "step": 1800
+    },
+    {
+      "epoch": 0.6968909423428626,
+      "grad_norm": 0.24212266504764557,
+      "learning_rate": 2.2263629693471258e-06,
+      "loss": 0.7593,
+      "step": 1810
+    },
+    {
+      "epoch": 0.700741168543652,
+      "grad_norm": 0.2339828908443451,
+      "learning_rate": 2.1746962128343414e-06,
+      "loss": 0.7829,
+      "step": 1820
+    },
+    {
+      "epoch": 0.7045913947444412,
+      "grad_norm": 0.23947788774967194,
+      "learning_rate": 2.1234688997080183e-06,
+      "loss": 0.7723,
+      "step": 1830
+    },
+    {
+      "epoch": 0.7084416209452306,
+      "grad_norm": 0.2602895200252533,
+      "learning_rate": 2.0726889977853053e-06,
+      "loss": 0.7774,
+      "step": 1840
+    },
+    {
+      "epoch": 0.7122918471460198,
+      "grad_norm": 0.25034359097480774,
+      "learning_rate": 2.022364405293703e-06,
+      "loss": 0.7908,
+      "step": 1850
+    },
+    {
+      "epoch": 0.7161420733468091,
+      "grad_norm": 0.23105847835540771,
+      "learning_rate": 1.9725029496425814e-06,
+      "loss": 0.782,
+      "step": 1860
+    },
+    {
+      "epoch": 0.7199922995475985,
+      "grad_norm": 0.2451004683971405,
+      "learning_rate": 1.923112386205727e-06,
+      "loss": 0.779,
+      "step": 1870
+    },
+    {
+      "epoch": 0.7238425257483877,
+      "grad_norm": 0.226190447807312,
+      "learning_rate": 1.874200397115079e-06,
+      "loss": 0.7859,
+      "step": 1880
+    },
+    {
+      "epoch": 0.727692751949177,
+      "grad_norm": 0.2438235729932785,
+      "learning_rate": 1.825774590065857e-06,
+      "loss": 0.7708,
+      "step": 1890
+    },
+    {
+      "epoch": 0.7315429781499663,
+      "grad_norm": 0.24786034226417542,
+      "learning_rate": 1.7778424971332813e-06,
+      "loss": 0.7776,
+      "step": 1900
+    },
+    {
+      "epoch": 0.7353932043507556,
+      "grad_norm": 0.23320272564888,
+      "learning_rate": 1.7304115736010407e-06,
+      "loss": 0.7748,
+      "step": 1910
+    },
+    {
+      "epoch": 0.739243430551545,
+      "grad_norm": 0.23742364346981049,
+      "learning_rate": 1.6834891968017108e-06,
+      "loss": 0.7695,
+      "step": 1920
+    },
+    {
+      "epoch": 0.7430936567523342,
+      "grad_norm": 0.24859724938869476,
+      "learning_rate": 1.6370826649692884e-06,
+      "loss": 0.7871,
+      "step": 1930
+    },
+    {
+      "epoch": 0.7469438829531235,
+      "grad_norm": 0.24485638737678528,
+      "learning_rate": 1.5911991961040463e-06,
+      "loss": 0.7946,
+      "step": 1940
+    },
+    {
+      "epoch": 0.7507941091539128,
+      "grad_norm": 0.25188320875167847,
+      "learning_rate": 1.545845926849851e-06,
+      "loss": 0.7828,
+      "step": 1950
+    },
+    {
+      "epoch": 0.7546443353547021,
+      "grad_norm": 0.25637540221214294,
+      "learning_rate": 1.5010299113841397e-06,
+      "loss": 0.7992,
+      "step": 1960
+    },
+    {
+      "epoch": 0.7584945615554913,
+      "grad_norm": 0.24335750937461853,
+      "learning_rate": 1.4567581203207238e-06,
+      "loss": 0.7898,
+      "step": 1970
+    },
+    {
+      "epoch": 0.7623447877562807,
+      "grad_norm": 0.2540871500968933,
+      "learning_rate": 1.413037439625593e-06,
+      "loss": 0.7858,
+      "step": 1980
+    },
+    {
+      "epoch": 0.76619501395707,
+      "grad_norm": 0.2636030316352844,
+      "learning_rate": 1.3698746695458847e-06,
+      "loss": 0.7824,
+      "step": 1990
+    },
+    {
+      "epoch": 0.7700452401578592,
+      "grad_norm": 0.23577453196048737,
+      "learning_rate": 1.3272765235521745e-06,
+      "loss": 0.7741,
+      "step": 2000
+    },
+    {
+      "epoch": 0.7738954663586486,
+      "grad_norm": 0.23341624438762665,
+      "learning_rate": 1.2852496272942877e-06,
+      "loss": 0.7816,
+      "step": 2010
+    },
+    {
+      "epoch": 0.7777456925594378,
+      "grad_norm": 0.22716988623142242,
+      "learning_rate": 1.2438005175707396e-06,
+      "loss": 0.7691,
+      "step": 2020
+    },
+    {
+      "epoch": 0.7815959187602272,
+      "grad_norm": 0.23518064618110657,
+      "learning_rate": 1.2029356413120297e-06,
+      "loss": 0.7539,
+      "step": 2030
+    },
+    {
+      "epoch": 0.7854461449610165,
+      "grad_norm": 0.23897691071033478,
+      "learning_rate": 1.1626613545778786e-06,
+      "loss": 0.7716,
+      "step": 2040
+    },
+    {
+      "epoch": 0.7892963711618057,
+      "grad_norm": 0.2286107987165451,
+      "learning_rate": 1.1229839215686294e-06,
+      "loss": 0.7765,
+      "step": 2050
+    },
+    {
+      "epoch": 0.7931465973625951,
+      "grad_norm": 0.22036826610565186,
+      "learning_rate": 1.0839095136509188e-06,
+      "loss": 0.7598,
+      "step": 2060
+    },
+    {
+      "epoch": 0.7969968235633843,
+      "grad_norm": 0.2591765224933624,
+      "learning_rate": 1.045444208397791e-06,
+      "loss": 0.7826,
+      "step": 2070
+    },
+    {
+      "epoch": 0.8008470497641736,
+      "grad_norm": 0.24606357514858246,
+      "learning_rate": 1.007593988643402e-06,
+      "loss": 0.764,
+      "step": 2080
+    },
+    {
+      "epoch": 0.804697275964963,
+      "grad_norm": 0.24214646220207214,
+      "learning_rate": 9.703647415524613e-07,
+      "loss": 0.7857,
+      "step": 2090
+    },
+    {
+      "epoch": 0.8085475021657522,
+      "grad_norm": 0.23447002470493317,
+      "learning_rate": 9.337622577045485e-07,
+      "loss": 0.7739,
+      "step": 2100
+    },
+    {
+      "epoch": 0.8123977283665416,
+      "grad_norm": 0.23982714116573334,
+      "learning_rate": 8.977922301934588e-07,
+      "loss": 0.7842,
+      "step": 2110
+    },
+    {
+      "epoch": 0.8162479545673308,
+      "grad_norm": 0.23694707453250885,
+      "learning_rate": 8.624602537417009e-07,
+      "loss": 0.7855,
+      "step": 2120
+    },
+    {
+      "epoch": 0.8200981807681201,
+      "grad_norm": 0.24654211103916168,
+      "learning_rate": 8.277718238303123e-07,
+      "loss": 0.7746,
+      "step": 2130
+    },
+    {
+      "epoch": 0.8239484069689095,
+      "grad_norm": 0.2400912195444107,
+      "learning_rate": 7.937323358440935e-07,
+      "loss": 0.7771,
+      "step": 2140
+    },
+    {
+      "epoch": 0.8277986331696987,
+      "grad_norm": 0.25079700350761414,
+      "learning_rate": 7.603470842324223e-07,
+      "loss": 0.7691,
+      "step": 2150
+    },
+    {
+      "epoch": 0.831648859370488,
+      "grad_norm": 0.2329913079738617,
+      "learning_rate": 7.276212616857575e-07,
+      "loss": 0.7854,
+      "step": 2160
+    },
+    {
+      "epoch": 0.8354990855712773,
+      "grad_norm": 0.23093287646770477,
+      "learning_rate": 6.955599583279837e-07,
+      "loss": 0.7849,
+      "step": 2170
+    },
+    {
+      "epoch": 0.8393493117720666,
+      "grad_norm": 0.24958983063697815,
+      "learning_rate": 6.641681609246981e-07,
+      "loss": 0.7765,
+      "step": 2180
+    },
+    {
+      "epoch": 0.8431995379728559,
+      "grad_norm": 0.2327955812215805,
+      "learning_rate": 6.334507521075711e-07,
+      "loss": 0.7835,
+      "step": 2190
+    },
+    {
+      "epoch": 0.8470497641736452,
+      "grad_norm": 0.2577342987060547,
+      "learning_rate": 6.034125096149179e-07,
+      "loss": 0.778,
+      "step": 2200
+    },
+    {
+      "epoch": 0.8508999903744345,
+      "grad_norm": 0.23735512793064117,
+      "learning_rate": 5.740581055485722e-07,
+      "loss": 0.7506,
+      "step": 2210
+    },
+    {
+      "epoch": 0.8547502165752238,
+      "grad_norm": 0.22282099723815918,
+      "learning_rate": 5.453921056471961e-07,
+      "loss": 0.7839,
+      "step": 2220
+    },
+    {
+      "epoch": 0.8586004427760131,
+      "grad_norm": 0.242647185921669,
+      "learning_rate": 5.174189685761288e-07,
+      "loss": 0.7778,
+      "step": 2230
+    },
+    {
+      "epoch": 0.8624506689768023,
+      "grad_norm": 0.22908435761928558,
+      "learning_rate": 4.901430452338979e-07,
+      "loss": 0.7896,
+      "step": 2240
+    },
+    {
+      "epoch": 0.8663008951775917,
+      "grad_norm": 0.23686227202415466,
+      "learning_rate": 4.6356857807548286e-07,
+      "loss": 0.7714,
+      "step": 2250
+    },
+    {
+      "epoch": 0.870151121378381,
+      "grad_norm": 0.24194329977035522,
+      "learning_rate": 4.376997004524541e-07,
+      "loss": 0.7616,
+      "step": 2260
+    },
+    {
+      "epoch": 0.8740013475791703,
+      "grad_norm": 0.24324660003185272,
+      "learning_rate": 4.12540435970073e-07,
+      "loss": 0.7714,
+      "step": 2270
+    },
+    {
+      "epoch": 0.8778515737799596,
+      "grad_norm": 0.22939905524253845,
+      "learning_rate": 3.880946978614703e-07,
+      "loss": 0.7834,
+      "step": 2280
+    },
+    {
+      "epoch": 0.8817017999807488,
+      "grad_norm": 0.23358911275863647,
+      "learning_rate": 3.643662883789878e-07,
+      "loss": 0.7865,
+      "step": 2290
+    },
+    {
+      "epoch": 0.8855520261815382,
+      "grad_norm": 0.21970824897289276,
+      "learning_rate": 3.41358898202781e-07,
+      "loss": 0.7646,
+      "step": 2300
+    },
+    {
+      "epoch": 0.8894022523823275,
+      "grad_norm": 0.2278887927532196,
+      "learning_rate": 3.190761058667741e-07,
+      "loss": 0.775,
+      "step": 2310
+    },
+    {
+      "epoch": 0.8932524785831167,
+      "grad_norm": 0.2295532524585724,
+      "learning_rate": 2.9752137720206463e-07,
+      "loss": 0.7803,
+      "step": 2320
+    },
+    {
+      "epoch": 0.8971027047839061,
+      "grad_norm": 0.24007730185985565,
+      "learning_rate": 2.7669806479785067e-07,
+      "loss": 0.7858,
+      "step": 2330
+    },
+    {
+      "epoch": 0.9009529309846953,
+      "grad_norm": 0.24610590934753418,
+      "learning_rate": 2.566094074799763e-07,
+      "loss": 0.7807,
+      "step": 2340
+    },
+    {
+      "epoch": 0.9048031571854847,
+      "grad_norm": 0.2347695231437683,
+      "learning_rate": 2.3725852980716557e-07,
+      "loss": 0.7616,
+      "step": 2350
+    },
+    {
+      "epoch": 0.908653383386274,
+      "grad_norm": 0.2322073131799698,
+      "learning_rate": 2.1864844158503983e-07,
+      "loss": 0.7747,
+      "step": 2360
+    },
+    {
+      "epoch": 0.9125036095870632,
+      "grad_norm": 0.2580222487449646,
+      "learning_rate": 2.007820373979702e-07,
+      "loss": 0.768,
+      "step": 2370
+    },
+    {
+      "epoch": 0.9163538357878526,
+      "grad_norm": 0.23572196066379547,
+      "learning_rate": 1.8366209615886532e-07,
+      "loss": 0.7748,
+      "step": 2380
+    },
+    {
+      "epoch": 0.9202040619886418,
+      "grad_norm": 0.23647673428058624,
+      "learning_rate": 1.6729128067693613e-07,
+      "loss": 0.7882,
+      "step": 2390
+    },
+    {
+      "epoch": 0.9240542881894311,
+      "grad_norm": 0.25139668583869934,
+      "learning_rate": 1.5167213724353426e-07,
+      "loss": 0.7878,
+      "step": 2400
+    },
+    {
+      "epoch": 0.9279045143902205,
+      "grad_norm": 0.23504218459129333,
+      "learning_rate": 1.3680709523610058e-07,
+      "loss": 0.7709,
+      "step": 2410
+    },
+    {
+      "epoch": 0.9317547405910097,
+      "grad_norm": 0.23704379796981812,
+      "learning_rate": 1.2269846674030418e-07,
+      "loss": 0.7917,
+      "step": 2420
+    },
+    {
+      "epoch": 0.935604966791799,
+      "grad_norm": 0.23923741281032562,
+      "learning_rate": 1.0934844619042595e-07,
+      "loss": 0.79,
+      "step": 2430
+    },
+    {
+      "epoch": 0.9394551929925883,
+      "grad_norm": 0.24860647320747375,
+      "learning_rate": 9.675911002803961e-08,
+      "loss": 0.7776,
+      "step": 2440
+    },
+    {
+      "epoch": 0.9433054191933776,
+      "grad_norm": 0.2501905858516693,
+      "learning_rate": 8.493241637904381e-08,
+      "loss": 0.7738,
+      "step": 2450
+    },
+    {
+      "epoch": 0.9471556453941669,
+      "grad_norm": 0.24374371767044067,
+      "learning_rate": 7.387020474909689e-08,
+      "loss": 0.7954,
+      "step": 2460
+    },
+    {
+      "epoch": 0.9510058715949562,
+      "grad_norm": 0.24756571650505066,
+      "learning_rate": 6.357419573750856e-08,
+      "loss": 0.7884,
+      "step": 2470
+    },
+    {
+      "epoch": 0.9548560977957455,
+      "grad_norm": 0.23955398797988892,
+      "learning_rate": 5.4045990769613365e-08,
+      "loss": 0.7729,
+      "step": 2480
+    },
+    {
+      "epoch": 0.9587063239965348,
+      "grad_norm": 0.2842100262641907,
+      "learning_rate": 4.5287071847694434e-08,
+      "loss": 0.7763,
+      "step": 2490
+    },
+    {
+      "epoch": 0.9625565501973241,
+      "grad_norm": 0.23990611732006073,
+      "learning_rate": 3.7298801320467817e-08,
+      "loss": 0.7714,
+      "step": 2500
+    },
+    {
+      "epoch": 0.9664067763981133,
+      "grad_norm": 0.2365291565656662,
+      "learning_rate": 3.0082421671192576e-08,
+      "loss": 0.7647,
+      "step": 2510
+    },
+    {
+      "epoch": 0.9702570025989027,
+      "grad_norm": 0.2390899956226349,
+      "learning_rate": 2.3639055324410355e-08,
+      "loss": 0.7876,
+      "step": 2520
+    },
+    {
+      "epoch": 0.974107228799692,
+      "grad_norm": 0.23528046905994415,
+      "learning_rate": 1.7969704471369477e-08,
+      "loss": 0.7705,
+      "step": 2530
+    },
+    {
+      "epoch": 0.9779574550004813,
+      "grad_norm": 0.22818730771541595,
+      "learning_rate": 1.3075250914142413e-08,
+      "loss": 0.7736,
+      "step": 2540
+    },
+    {
+      "epoch": 0.9818076812012706,
+      "grad_norm": 0.22822588682174683,
+      "learning_rate": 8.956455928476049e-09,
+      "loss": 0.7826,
+      "step": 2550
+    },
+    {
+      "epoch": 0.9856579074020598,
+      "grad_norm": 0.24411576986312866,
+      "learning_rate": 5.6139601453786275e-09,
+      "loss": 0.7747,
+      "step": 2560
+    },
+    {
+      "epoch": 0.9895081336028492,
+      "grad_norm": 0.2396223545074463,
+      "learning_rate": 3.0482834514805602e-09,
+      "loss": 0.7636,
+      "step": 2570
+    },
+    {
+      "epoch": 0.9933583598036385,
+      "grad_norm": 0.21465757489204407,
+      "learning_rate": 1.2598249081713366e-09,
+      "loss": 0.7591,
+      "step": 2580
+    },
+    {
+      "epoch": 0.9972085860044277,
+      "grad_norm": 0.24594566226005554,
+      "learning_rate": 2.488626895302826e-10,
+      "loss": 0.7903,
+      "step": 2590
+    },
+    {
+      "epoch": 0.9999037443449803,
+      "step": 2597,
+      "total_flos": 3.609381421241598e+18,
+      "train_loss": 0.8064731716145723,
+      "train_runtime": 10319.6676,
+      "train_samples_per_second": 64.431,
+      "train_steps_per_second": 0.252
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 2597,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.609381421241598e+18,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}

continuous/qwen3/mc2_s16_224/finetune/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1faa3a9073a633f173e8f8a09e27d711a506d1a8250c83448213c05c831be8f3
+size 6136

continuous/qwen3/mc2_s16_224/pretrain/config.json

@@ -0,0 +1,60 @@
+{
+  "_attn_implementation_autoset": true,
+  "add_faster_video": false,
+  "add_time_instruction": false,
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "eos_token_id": 151645,
+  "faster_token_stride": 10,
+  "force_sample": false,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 2048,
+  "image_aspect_ratio": "square",
+  "image_crop_resolution": null,
+  "image_grid_pinpoints": null,
+  "image_split_resolution": null,
+  "initializer_range": 0.02,
+  "intermediate_size": 6144,
+  "max_position_embeddings": 40960,
+  "max_window_layers": 28,
+  "mm_hidden_size": 384,
+  "mm_newline_position": "grid",
+  "mm_patch_merge_type": "flat",
+  "mm_projector_lr": null,
+  "mm_projector_type": "mlp2x_gelu",
+  "mm_resampler_type": null,
+  "mm_spatial_pool_mode": "bilinear",
+  "mm_spatial_pool_stride": null,
+  "mm_tunable_parts": "mm_mlp_adapter",
+  "mm_use_im_patch_token": false,
+  "mm_use_im_start_end": false,
+  "mm_vision_select_feature": "patch",
+  "mm_vision_select_layer": -2,
+  "mm_vision_tower": "open_clip_hub:ViT-S-16",
+  "mm_vision_tower_lr": null,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pos_skipping_range": 4096,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000,
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "tokenizer_model_max_length": 8192,
+  "tokenizer_padding_side": "right",
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.51.3",
+  "use_cache": true,
+  "use_mm_proj": true,
+  "use_pos_skipping": false,
+  "use_sliding_window": false,
+  "vision_tower_pretrained": "/cache/ckpt/download/tokenizer/continuous/mc2/mc2_s16_224.pt",
+  "vocab_size": 151936
+}

continuous/qwen3/mc2_s16_224/pretrain/mm_projector.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:040890e0e76db3726aca1a94c45d64efe2142fd3fccc5cc499b80fb1e9ad46af
+size 19941432

continuous/qwen3/mc2_s16_224/pretrain/trainer_state.json

@@ -0,0 +1,421 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 545,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01834862385321101,
+      "grad_norm": 30.35206413269043,
+      "learning_rate": 0.00032142857142857147,
+      "loss": 10.4032,
+      "step": 10
+    },
+    {
+      "epoch": 0.03669724770642202,
+      "grad_norm": 1.3426218032836914,
+      "learning_rate": 0.0006785714285714287,
+      "loss": 5.8879,
+      "step": 20
+    },
+    {
+      "epoch": 0.05504587155963303,
+      "grad_norm": 0.21277467906475067,
+      "learning_rate": 0.0009999907688176173,
+      "loss": 4.8139,
+      "step": 30
+    },
+    {
+      "epoch": 0.07339449541284404,
+      "grad_norm": 0.24309980869293213,
+      "learning_rate": 0.0009988834393115766,
+      "loss": 4.6806,
+      "step": 40
+    },
+    {
+      "epoch": 0.09174311926605505,
+      "grad_norm": 0.15426211059093475,
+      "learning_rate": 0.0009959345572940772,
+      "loss": 4.6246,
+      "step": 50
+    },
+    {
+      "epoch": 0.11009174311926606,
+      "grad_norm": 0.13272696733474731,
+      "learning_rate": 0.0009911550081155983,
+      "loss": 4.5807,
+      "step": 60
+    },
+    {
+      "epoch": 0.12844036697247707,
+      "grad_norm": 0.21833860874176025,
+      "learning_rate": 0.0009845624347566433,
+      "loss": 4.5267,
+      "step": 70
+    },
+    {
+      "epoch": 0.14678899082568808,
+      "grad_norm": 0.13713011145591736,
+      "learning_rate": 0.0009761811727013547,
+      "loss": 4.4914,
+      "step": 80
+    },
+    {
+      "epoch": 0.1651376146788991,
+      "grad_norm": 0.1842188984155655,
+      "learning_rate": 0.0009660421601067666,
+      "loss": 4.4403,
+      "step": 90
+    },
+    {
+      "epoch": 0.1834862385321101,
+      "grad_norm": 0.2132931500673294,
+      "learning_rate": 0.0009541828235992926,
+      "loss": 4.409,
+      "step": 100
+    },
+    {
+      "epoch": 0.2018348623853211,
+      "grad_norm": 0.22319301962852478,
+      "learning_rate": 0.0009406469401200151,
+      "loss": 4.3694,
+      "step": 110
+    },
+    {
+      "epoch": 0.22018348623853212,
+      "grad_norm": 0.20336176455020905,
+      "learning_rate": 0.0009254844753287492,
+      "loss": 4.3532,
+      "step": 120
+    },
+    {
+      "epoch": 0.23853211009174313,
+      "grad_norm": 0.1778409481048584,
+      "learning_rate": 0.0009087513991633924,
+      "loss": 4.3081,
+      "step": 130
+    },
+    {
+      "epoch": 0.25688073394495414,
+      "grad_norm": 0.2476678341627121,
+      "learning_rate": 0.0008905094792353917,
+      "loss": 4.2785,
+      "step": 140
+    },
+    {
+      "epoch": 0.27522935779816515,
+      "grad_norm": 0.2011418491601944,
+      "learning_rate": 0.0008708260528239789,
+      "loss": 4.2513,
+      "step": 150
+    },
+    {
+      "epoch": 0.29357798165137616,
+      "grad_norm": 0.21509084105491638,
+      "learning_rate": 0.0008497737783108237,
+      "loss": 4.2157,
+      "step": 160
+    },
+    {
+      "epoch": 0.3119266055045872,
+      "grad_norm": 0.18161150813102722,
+      "learning_rate": 0.0008274303669726426,
+      "loss": 4.1878,
+      "step": 170
+    },
+    {
+      "epoch": 0.3302752293577982,
+      "grad_norm": 0.21620866656303406,
+      "learning_rate": 0.0008038782961218137,
+      "loss": 4.175,
+      "step": 180
+    },
+    {
+      "epoch": 0.3486238532110092,
+      "grad_norm": 0.20195560157299042,
+      "learning_rate": 0.000779204504653894,
+      "loss": 4.164,
+      "step": 190
+    },
+    {
+      "epoch": 0.3669724770642202,
+      "grad_norm": 0.1793253868818283,
+      "learning_rate": 0.0007535000721258791,
+      "loss": 4.1417,
+      "step": 200
+    },
+    {
+      "epoch": 0.3853211009174312,
+      "grad_norm": 0.19129818677902222,
+      "learning_rate": 0.00072685988254984,
+      "loss": 4.1289,
+      "step": 210
+    },
+    {
+      "epoch": 0.4036697247706422,
+      "grad_norm": 0.2865113615989685,
+      "learning_rate": 0.0006993822741429906,
+      "loss": 4.1401,
+      "step": 220
+    },
+    {
+      "epoch": 0.42201834862385323,
+      "grad_norm": 0.1894138753414154,
+      "learning_rate": 0.0006711686763270818,
+      "loss": 4.1338,
+      "step": 230
+    },
+    {
+      "epoch": 0.44036697247706424,
+      "grad_norm": 0.15442484617233276,
+      "learning_rate": 0.0006423232353170831,
+      "loss": 4.1171,
+      "step": 240
+    },
+    {
+      "epoch": 0.45871559633027525,
+      "grad_norm": 0.1790081262588501,
+      "learning_rate": 0.0006129524296812335,
+      "loss": 4.1055,
+      "step": 250
+    },
+    {
+      "epoch": 0.47706422018348627,
+      "grad_norm": 0.17230738699436188,
+      "learning_rate": 0.0005831646772915651,
+      "loss": 4.1179,
+      "step": 260
+    },
+    {
+      "epoch": 0.4954128440366973,
+      "grad_norm": 0.210103377699852,
+      "learning_rate": 0.0005530699351157809,
+      "loss": 4.0973,
+      "step": 270
+    },
+    {
+      "epoch": 0.5137614678899083,
+      "grad_norm": 0.20966532826423645,
+      "learning_rate": 0.0005227792933277943,
+      "loss": 4.0976,
+      "step": 280
+    },
+    {
+      "epoch": 0.5321100917431193,
+      "grad_norm": 0.18907999992370605,
+      "learning_rate": 0.0004924045652352092,
+      "loss": 4.0877,
+      "step": 290
+    },
+    {
+      "epoch": 0.5504587155963303,
+      "grad_norm": 0.172409325838089,
+      "learning_rate": 0.0004620578745374633,
+      "loss": 4.0862,
+      "step": 300
+    },
+    {
+      "epoch": 0.5688073394495413,
+      "grad_norm": 0.20657303929328918,
+      "learning_rate": 0.0004318512414382059,
+      "loss": 4.0966,
+      "step": 310
+    },
+    {
+      "epoch": 0.5871559633027523,
+      "grad_norm": 0.16348327696323395,
+      "learning_rate": 0.0004018961691397148,
+      "loss": 4.0882,
+      "step": 320
+    },
+    {
+      "epoch": 0.6055045871559633,
+      "grad_norm": 0.1543281376361847,
+      "learning_rate": 0.0003723032322457458,
+      "loss": 4.0738,
+      "step": 330
+    },
+    {
+      "epoch": 0.6238532110091743,
+      "grad_norm": 0.2603055536746979,
+      "learning_rate": 0.0003431816685921625,
+      "loss": 4.0736,
+      "step": 340
+    },
+    {
+      "epoch": 0.6422018348623854,
+      "grad_norm": 0.1702510267496109,
+      "learning_rate": 0.0003146389760120416,
+      "loss": 4.0776,
+      "step": 350
+    },
+    {
+      "epoch": 0.6605504587155964,
+      "grad_norm": 0.2150663584470749,
+      "learning_rate": 0.00028678051552373485,
+      "loss": 4.0834,
+      "step": 360
+    },
+    {
+      "epoch": 0.6788990825688074,
+      "grad_norm": 0.16770724952220917,
+      "learning_rate": 0.0002597091224066581,
+      "loss": 4.072,
+      "step": 370
+    },
+    {
+      "epoch": 0.6972477064220184,
+      "grad_norm": 0.1589813530445099,
+      "learning_rate": 0.0002335247266004629,
+      "loss": 4.0838,
+      "step": 380
+    },
+    {
+      "epoch": 0.7155963302752294,
+      "grad_norm": 0.1580507457256317,
+      "learning_rate": 0.000208323983828828,
+      "loss": 4.066,
+      "step": 390
+    },
+    {
+      "epoch": 0.7339449541284404,
+      "grad_norm": 0.16531822085380554,
+      "learning_rate": 0.0001841999188095224,
+      "loss": 4.0642,
+      "step": 400
+    },
+    {
+      "epoch": 0.7522935779816514,
+      "grad_norm": 0.14605429768562317,
+      "learning_rate": 0.00016124158186777676,
+      "loss": 4.0695,
+      "step": 410
+    },
+    {
+      "epoch": 0.7706422018348624,
+      "grad_norm": 0.1449277698993683,
+      "learning_rate": 0.0001395337202205211,
+      "loss": 4.0643,
+      "step": 420
+    },
+    {
+      "epoch": 0.7889908256880734,
+      "grad_norm": 0.14998669922351837,
+      "learning_rate": 0.00011915646514489292,
+      "loss": 4.0651,
+      "step": 430
+    },
+    {
+      "epoch": 0.8073394495412844,
+      "grad_norm": 0.17550690472126007,
+      "learning_rate": 0.00010018503618578818,
+      "loss": 4.0732,
+      "step": 440
+    },
+    {
+      "epoch": 0.8256880733944955,
+      "grad_norm": 0.14613936841487885,
+      "learning_rate": 8.268946349432583e-05,
+      "loss": 4.0624,
+      "step": 450
+    },
+    {
+      "epoch": 0.8440366972477065,
+      "grad_norm": 0.15143834054470062,
+      "learning_rate": 6.673432932217117e-05,
+      "loss": 4.0553,
+      "step": 460
+    },
+    {
+      "epoch": 0.8623853211009175,
+      "grad_norm": 0.15412414073944092,
+      "learning_rate": 5.2378529625954684e-05,
+      "loss": 4.0448,
+      "step": 470
+    },
+    {
+      "epoch": 0.8807339449541285,
+      "grad_norm": 0.1447436660528183,
+      "learning_rate": 3.9675056661785556e-05,
+      "loss": 4.0532,
+      "step": 480
+    },
+    {
+      "epoch": 0.8990825688073395,
+      "grad_norm": 0.14502595365047455,
+      "learning_rate": 2.8670803372379538e-05,
+      "loss": 4.0567,
+      "step": 490
+    },
+    {
+      "epoch": 0.9174311926605505,
+      "grad_norm": 0.14784513413906097,
+      "learning_rate": 1.9406390288876586e-05,
+      "loss": 4.0685,
+      "step": 500
+    },
+    {
+      "epoch": 0.9357798165137615,
+      "grad_norm": 0.1444227248430252,
+      "learning_rate": 1.191601558631461e-05,
+      "loss": 4.0575,
+      "step": 510
+    },
+    {
+      "epoch": 0.9541284403669725,
+      "grad_norm": 0.14080111682415009,
+      "learning_rate": 6.2273288462559305e-06,
+      "loss": 4.0522,
+      "step": 520
+    },
+    {
+      "epoch": 0.9724770642201835,
+      "grad_norm": 0.14046412706375122,
+      "learning_rate": 2.361328992552314e-06,
+      "loss": 4.0533,
+      "step": 530
+    },
+    {
+      "epoch": 0.9908256880733946,
+      "grad_norm": 0.14061950147151947,
+      "learning_rate": 3.322867770044202e-07,
+      "loss": 4.0533,
+      "step": 540
+    },
+    {
+      "epoch": 1.0,
+      "step": 545,
+      "total_flos": 2.963396558533427e+17,
+      "train_loss": 4.331399077669196,
+      "train_runtime": 5637.084,
+      "train_samples_per_second": 99.01,
+      "train_steps_per_second": 0.097
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 545,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": false,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.963396558533427e+17,
+  "train_batch_size": 128,
+  "trial_name": null,
+  "trial_params": null
+}

continuous/qwen3/mc2_s16_224_mt5/finetune/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
+}

continuous/qwen3/mc2_s16_224_mt5/finetune/config.json

@@ -0,0 +1,59 @@
+{
+  "add_faster_video": false,
+  "add_time_instruction": false,
+  "architectures": [
+    "LlavaQwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "eos_token_id": 151645,
+  "faster_token_stride": 10,
+  "force_sample": false,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 2048,
+  "image_aspect_ratio": "square",
+  "image_crop_resolution": null,
+  "image_grid_pinpoints": null,
+  "image_split_resolution": null,
+  "initializer_range": 0.02,
+  "intermediate_size": 6144,
+  "max_position_embeddings": 40960,
+  "max_window_layers": 28,
+  "mm_hidden_size": 384,
+  "mm_newline_position": "grid",
+  "mm_patch_merge_type": "flat",
+  "mm_projector_lr": 0.0001,
+  "mm_projector_type": "mlp2x_gelu",
+  "mm_resampler_type": null,
+  "mm_spatial_pool_mode": "bilinear",
+  "mm_spatial_pool_stride": null,
+  "mm_tunable_parts": "mm_mlp_adapter,mm_language_model",
+  "mm_use_im_patch_token": false,
+  "mm_use_im_start_end": false,
+  "mm_vision_select_feature": "patch",
+  "mm_vision_select_layer": -2,
+  "mm_vision_tower": "open_clip_hub:ViT-S-16",
+  "mm_vision_tower_lr": null,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pos_skipping_range": 4096,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000,
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "tokenizer_model_max_length": 8192,
+  "tokenizer_padding_side": "right",
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.51.3",
+  "use_cache": true,
+  "use_mm_proj": true,
+  "use_pos_skipping": false,
+  "use_sliding_window": false,
+  "vision_tower_pretrained": "/cache/ckpt/download/tokenizer/continuous/mc2/mc2_s16_224_mt5.pt",
+  "vocab_size": 151936
+}

continuous/qwen3/mc2_s16_224_mt5/finetune/generation_config.json

@@ -0,0 +1,13 @@
+{
+  "bos_token_id": 151643,
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "pad_token_id": 151643,
+  "temperature": 0.6,
+  "top_k": 20,
+  "top_p": 0.95,
+  "transformers_version": "4.51.3"
+}

continuous/qwen3/mc2_s16_224_mt5/finetune/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:514dbf224136dea463ed1c4e9da8b4a7bf05e3e0593de0d83fa6030835bacb69
+size 4127104248

continuous/qwen3/mc2_s16_224_mt5/finetune/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
+  }
+}

continuous/qwen3/mc2_s16_224_mt5/finetune/tokenizer.json

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

continuous/qwen3/mc2_s16_224_mt5/finetune/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 message.content is string 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.content is string %}\n        {%- set content = message.content %}\n    {%- else %}\n        {%- set content = '' %}\n    {%- endif %}\n    {%- if (message.role == \"user\") or (message.role == \"system\" and not loop.first) %}\n        {{- '<|im_start|>' + message.role + '\\n' + content + '<|im_end|>' + '\\n' }}\n    {%- elif message.role == \"assistant\" %}\n        {%- set reasoning_content = '' %}\n        {%- if message.reasoning_content is string %}\n            {%- set reasoning_content = message.reasoning_content %}\n        {%- else %}\n            {%- if '</think>' in content %}\n                {%- set reasoning_content = content.split('</think>')[0].rstrip('\\n').split('<think>')[-1].lstrip('\\n') %}\n                {%- set content = content.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        {{- 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": 8192,
+  "pad_token": "<|endoftext|>",
+  "padding_side": "right",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

continuous/qwen3/mc2_s16_224_mt5/finetune/trainer_state.json

@@ -0,0 +1,1856 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999037443449803,
+  "eval_steps": 500,
+  "global_step": 2597,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.003850226200789296,
+      "grad_norm": 22.892704010009766,
+      "learning_rate": 1.153846153846154e-06,
+      "loss": 2.5131,
+      "step": 10
+    },
+    {
+      "epoch": 0.007700452401578592,
+      "grad_norm": 6.93652868270874,
+      "learning_rate": 2.435897435897436e-06,
+      "loss": 1.9472,
+      "step": 20
+    },
+    {
+      "epoch": 0.01155067860236789,
+      "grad_norm": 4.588390350341797,
+      "learning_rate": 3.7179487179487184e-06,
+      "loss": 1.5137,
+      "step": 30
+    },
+    {
+      "epoch": 0.015400904803157185,
+      "grad_norm": 1.2024660110473633,
+      "learning_rate": 5e-06,
+      "loss": 1.1729,
+      "step": 40
+    },
+    {
+      "epoch": 0.019251131003946483,
+      "grad_norm": 0.8227046132087708,
+      "learning_rate": 6.282051282051282e-06,
+      "loss": 1.0582,
+      "step": 50
+    },
+    {
+      "epoch": 0.02310135720473578,
+      "grad_norm": 0.40982484817504883,
+      "learning_rate": 7.564102564102564e-06,
+      "loss": 0.9685,
+      "step": 60
+    },
+    {
+      "epoch": 0.026951583405525073,
+      "grad_norm": 0.3134242296218872,
+      "learning_rate": 8.846153846153847e-06,
+      "loss": 0.9205,
+      "step": 70
+    },
+    {
+      "epoch": 0.03080180960631437,
+      "grad_norm": 0.29037871956825256,
+      "learning_rate": 9.999996111488723e-06,
+      "loss": 0.9166,
+      "step": 80
+    },
+    {
+      "epoch": 0.034652035807103666,
+      "grad_norm": 0.2813657820224762,
+      "learning_rate": 9.999529497453782e-06,
+      "loss": 0.8839,
+      "step": 90
+    },
+    {
+      "epoch": 0.038502262007892966,
+      "grad_norm": 0.25447848439216614,
+      "learning_rate": 9.998285264324193e-06,
+      "loss": 0.8649,
+      "step": 100
+    },
+    {
+      "epoch": 0.04235248820868226,
+      "grad_norm": 0.25465282797813416,
+      "learning_rate": 9.996263605626052e-06,
+      "loss": 0.8472,
+      "step": 110
+    },
+    {
+      "epoch": 0.04620271440947156,
+      "grad_norm": 0.256450891494751,
+      "learning_rate": 9.99346483580503e-06,
+      "loss": 0.8567,
+      "step": 120
+    },
+    {
+      "epoch": 0.05005294061026085,
+      "grad_norm": 0.25389862060546875,
+      "learning_rate": 9.989889390177467e-06,
+      "loss": 0.8511,
+      "step": 130
+    },
+    {
+      "epoch": 0.053903166811050146,
+      "grad_norm": 0.28586211800575256,
+      "learning_rate": 9.985537824862647e-06,
+      "loss": 0.8464,
+      "step": 140
+    },
+    {
+      "epoch": 0.057753393011839446,
+      "grad_norm": 0.29856494069099426,
+      "learning_rate": 9.980410816696317e-06,
+      "loss": 0.8316,
+      "step": 150
+    },
+    {
+      "epoch": 0.06160361921262874,
+      "grad_norm": 0.26229092478752136,
+      "learning_rate": 9.97450916312541e-06,
+      "loss": 0.8376,
+      "step": 160
+    },
+    {
+      "epoch": 0.06545384541341803,
+      "grad_norm": 0.22727970778942108,
+      "learning_rate": 9.967833782084004e-06,
+      "loss": 0.8382,
+      "step": 170
+    },
+    {
+      "epoch": 0.06930407161420733,
+      "grad_norm": 0.267502099275589,
+      "learning_rate": 9.960385711850554e-06,
+      "loss": 0.8648,
+      "step": 180
+    },
+    {
+      "epoch": 0.07315429781499663,
+      "grad_norm": 0.257737934589386,
+      "learning_rate": 9.952166110886394e-06,
+      "loss": 0.8552,
+      "step": 190
+    },
+    {
+      "epoch": 0.07700452401578593,
+      "grad_norm": 0.28025615215301514,
+      "learning_rate": 9.943176257655567e-06,
+      "loss": 0.8426,
+      "step": 200
+    },
+    {
+      "epoch": 0.08085475021657522,
+      "grad_norm": 0.2587888240814209,
+      "learning_rate": 9.933417550425952e-06,
+      "loss": 0.8397,
+      "step": 210
+    },
+    {
+      "epoch": 0.08470497641736452,
+      "grad_norm": 0.25643810629844666,
+      "learning_rate": 9.9228915070518e-06,
+      "loss": 0.8253,
+      "step": 220
+    },
+    {
+      "epoch": 0.08855520261815382,
+      "grad_norm": 0.24446739256381989,
+      "learning_rate": 9.911599764737632e-06,
+      "loss": 0.8259,
+      "step": 230
+    },
+    {
+      "epoch": 0.09240542881894312,
+      "grad_norm": 0.2536662817001343,
+      "learning_rate": 9.899544079783606e-06,
+      "loss": 0.8201,
+      "step": 240
+    },
+    {
+      "epoch": 0.0962556550197324,
+      "grad_norm": 0.2899295389652252,
+      "learning_rate": 9.886726327312338e-06,
+      "loss": 0.8385,
+      "step": 250
+    },
+    {
+      "epoch": 0.1001058812205217,
+      "grad_norm": 0.23787197470664978,
+      "learning_rate": 9.87314850097724e-06,
+      "loss": 0.8332,
+      "step": 260
+    },
+    {
+      "epoch": 0.103956107421311,
+      "grad_norm": 0.2487773895263672,
+      "learning_rate": 9.858812712652452e-06,
+      "loss": 0.8221,
+      "step": 270
+    },
+    {
+      "epoch": 0.10780633362210029,
+      "grad_norm": 0.23249875009059906,
+      "learning_rate": 9.843721192104341e-06,
+      "loss": 0.8214,
+      "step": 280
+    },
+    {
+      "epoch": 0.11165655982288959,
+      "grad_norm": 0.239415243268013,
+      "learning_rate": 9.827876286644698e-06,
+      "loss": 0.8386,
+      "step": 290
+    },
+    {
+      "epoch": 0.11550678602367889,
+      "grad_norm": 0.22825098037719727,
+      "learning_rate": 9.811280460765643e-06,
+      "loss": 0.8007,
+      "step": 300
+    },
+    {
+      "epoch": 0.11935701222446819,
+      "grad_norm": 0.23969928920269012,
+      "learning_rate": 9.793936295756292e-06,
+      "loss": 0.8132,
+      "step": 310
+    },
+    {
+      "epoch": 0.12320723842525748,
+      "grad_norm": 0.25044894218444824,
+      "learning_rate": 9.775846489301279e-06,
+      "loss": 0.8143,
+      "step": 320
+    },
+    {
+      "epoch": 0.12705746462604678,
+      "grad_norm": 0.2465069442987442,
+      "learning_rate": 9.757013855061157e-06,
+      "loss": 0.8137,
+      "step": 330
+    },
+    {
+      "epoch": 0.13090769082683607,
+      "grad_norm": 0.2557399868965149,
+      "learning_rate": 9.737441322234758e-06,
+      "loss": 0.8244,
+      "step": 340
+    },
+    {
+      "epoch": 0.13475791702762538,
+      "grad_norm": 0.2511087954044342,
+      "learning_rate": 9.717131935103605e-06,
+      "loss": 0.8249,
+      "step": 350
+    },
+    {
+      "epoch": 0.13860814322841467,
+      "grad_norm": 0.238852858543396,
+      "learning_rate": 9.696088852558398e-06,
+      "loss": 0.8119,
+      "step": 360
+    },
+    {
+      "epoch": 0.14245836942920398,
+      "grad_norm": 0.26223963499069214,
+      "learning_rate": 9.674315347607686e-06,
+      "loss": 0.8379,
+      "step": 370
+    },
+    {
+      "epoch": 0.14630859562999327,
+      "grad_norm": 0.2426629364490509,
+      "learning_rate": 9.651814806868796e-06,
+      "loss": 0.7983,
+      "step": 380
+    },
+    {
+      "epoch": 0.15015882183078255,
+      "grad_norm": 0.23303556442260742,
+      "learning_rate": 9.62859073004107e-06,
+      "loss": 0.8037,
+      "step": 390
+    },
+    {
+      "epoch": 0.15400904803157187,
+      "grad_norm": 0.24447718262672424,
+      "learning_rate": 9.604646729361547e-06,
+      "loss": 0.8054,
+      "step": 400
+    },
+    {
+      "epoch": 0.15785927423236115,
+      "grad_norm": 0.22934141755104065,
+      "learning_rate": 9.579986529043099e-06,
+      "loss": 0.795,
+      "step": 410
+    },
+    {
+      "epoch": 0.16170950043315044,
+      "grad_norm": 0.23649121820926666,
+      "learning_rate": 9.554613964695189e-06,
+      "loss": 0.8132,
+      "step": 420
+    },
+    {
+      "epoch": 0.16555972663393975,
+      "grad_norm": 0.23390144109725952,
+      "learning_rate": 9.528532982727281e-06,
+      "loss": 0.8039,
+      "step": 430
+    },
+    {
+      "epoch": 0.16940995283472904,
+      "grad_norm": 0.23824650049209595,
+      "learning_rate": 9.501747639735018e-06,
+      "loss": 0.7996,
+      "step": 440
+    },
+    {
+      "epoch": 0.17326017903551832,
+      "grad_norm": 0.251240074634552,
+      "learning_rate": 9.474262101869272e-06,
+      "loss": 0.7904,
+      "step": 450
+    },
+    {
+      "epoch": 0.17711040523630764,
+      "grad_norm": 0.24060536921024323,
+      "learning_rate": 9.446080644188141e-06,
+      "loss": 0.7971,
+      "step": 460
+    },
+    {
+      "epoch": 0.18096063143709692,
+      "grad_norm": 0.24634629487991333,
+      "learning_rate": 9.417207649992017e-06,
+      "loss": 0.802,
+      "step": 470
+    },
+    {
+      "epoch": 0.18481085763788624,
+      "grad_norm": 0.24451257288455963,
+      "learning_rate": 9.387647610141816e-06,
+      "loss": 0.8017,
+      "step": 480
+    },
+    {
+      "epoch": 0.18866108383867553,
+      "grad_norm": 0.2326030135154724,
+      "learning_rate": 9.35740512236047e-06,
+      "loss": 0.7879,
+      "step": 490
+    },
+    {
+      "epoch": 0.1925113100394648,
+      "grad_norm": 0.22699305415153503,
+      "learning_rate": 9.326484890517809e-06,
+      "loss": 0.7955,
+      "step": 500
+    },
+    {
+      "epoch": 0.19636153624025413,
+      "grad_norm": 0.23437446355819702,
+      "learning_rate": 9.294891723898926e-06,
+      "loss": 0.7895,
+      "step": 510
+    },
+    {
+      "epoch": 0.2002117624410434,
+      "grad_norm": 0.24088536202907562,
+      "learning_rate": 9.262630536456153e-06,
+      "loss": 0.8198,
+      "step": 520
+    },
+    {
+      "epoch": 0.2040619886418327,
+      "grad_norm": 0.25662100315093994,
+      "learning_rate": 9.229706346044749e-06,
+      "loss": 0.7907,
+      "step": 530
+    },
+    {
+      "epoch": 0.207912214842622,
+      "grad_norm": 0.259194552898407,
+      "learning_rate": 9.196124273642422e-06,
+      "loss": 0.8045,
+      "step": 540
+    },
+    {
+      "epoch": 0.2117624410434113,
+      "grad_norm": 0.2270226776599884,
+      "learning_rate": 9.161889542552842e-06,
+      "loss": 0.8084,
+      "step": 550
+    },
+    {
+      "epoch": 0.21561266724420058,
+      "grad_norm": 0.223994180560112,
+      "learning_rate": 9.127007477593193e-06,
+      "loss": 0.8102,
+      "step": 560
+    },
+    {
+      "epoch": 0.2194628934449899,
+      "grad_norm": 0.26642847061157227,
+      "learning_rate": 9.091483504265974e-06,
+      "loss": 0.8179,
+      "step": 570
+    },
+    {
+      "epoch": 0.22331311964577918,
+      "grad_norm": 0.2565174400806427,
+      "learning_rate": 9.055323147915116e-06,
+      "loss": 0.8004,
+      "step": 580
+    },
+    {
+      "epoch": 0.2271633458465685,
+      "grad_norm": 0.24466301500797272,
+      "learning_rate": 9.018532032866585e-06,
+      "loss": 0.8088,
+      "step": 590
+    },
+    {
+      "epoch": 0.23101357204735778,
+      "grad_norm": 0.22675062716007233,
+      "learning_rate": 8.981115881553579e-06,
+      "loss": 0.8006,
+      "step": 600
+    },
+    {
+      "epoch": 0.23486379824814707,
+      "grad_norm": 0.2694375514984131,
+      "learning_rate": 8.943080513626476e-06,
+      "loss": 0.8208,
+      "step": 610
+    },
+    {
+      "epoch": 0.23871402444893638,
+      "grad_norm": 0.241071417927742,
+      "learning_rate": 8.904431845047641e-06,
+      "loss": 0.8045,
+      "step": 620
+    },
+    {
+      "epoch": 0.24256425064972567,
+      "grad_norm": 0.2514018416404724,
+      "learning_rate": 8.86517588717128e-06,
+      "loss": 0.8215,
+      "step": 630
+    },
+    {
+      "epoch": 0.24641447685051496,
+      "grad_norm": 0.23445287346839905,
+      "learning_rate": 8.82531874580844e-06,
+      "loss": 0.8128,
+      "step": 640
+    },
+    {
+      "epoch": 0.25026470305130427,
+      "grad_norm": 0.21542055904865265,
+      "learning_rate": 8.784866620277314e-06,
+      "loss": 0.769,
+      "step": 650
+    },
+    {
+      "epoch": 0.25411492925209356,
+      "grad_norm": 0.25378331542015076,
+      "learning_rate": 8.743825802439018e-06,
+      "loss": 0.8033,
+      "step": 660
+    },
+    {
+      "epoch": 0.25796515545288284,
+      "grad_norm": 0.2578502893447876,
+      "learning_rate": 8.702202675718956e-06,
+      "loss": 0.811,
+      "step": 670
+    },
+    {
+      "epoch": 0.26181538165367213,
+      "grad_norm": 0.2330646812915802,
+      "learning_rate": 8.660003714113959e-06,
+      "loss": 0.7788,
+      "step": 680
+    },
+    {
+      "epoch": 0.26566560785446147,
+      "grad_norm": 0.230241060256958,
+      "learning_rate": 8.617235481185323e-06,
+      "loss": 0.8024,
+      "step": 690
+    },
+    {
+      "epoch": 0.26951583405525076,
+      "grad_norm": 0.24212613701820374,
+      "learning_rate": 8.573904629037931e-06,
+      "loss": 0.809,
+      "step": 700
+    },
+    {
+      "epoch": 0.27336606025604004,
+      "grad_norm": 0.23603206872940063,
+      "learning_rate": 8.530017897285585e-06,
+      "loss": 0.7906,
+      "step": 710
+    },
+    {
+      "epoch": 0.27721628645682933,
+      "grad_norm": 0.22476592659950256,
+      "learning_rate": 8.485582112002746e-06,
+      "loss": 0.7943,
+      "step": 720
+    },
+    {
+      "epoch": 0.2810665126576186,
+      "grad_norm": 0.23908984661102295,
+      "learning_rate": 8.440604184662811e-06,
+      "loss": 0.7995,
+      "step": 730
+    },
+    {
+      "epoch": 0.28491673885840796,
+      "grad_norm": 0.24330322444438934,
+      "learning_rate": 8.395091111063118e-06,
+      "loss": 0.7865,
+      "step": 740
+    },
+    {
+      "epoch": 0.28876696505919724,
+      "grad_norm": 0.2262437492609024,
+      "learning_rate": 8.349049970236822e-06,
+      "loss": 0.7875,
+      "step": 750
+    },
+    {
+      "epoch": 0.29261719125998653,
+      "grad_norm": 0.23922546207904816,
+      "learning_rate": 8.302487923351842e-06,
+      "loss": 0.7803,
+      "step": 760
+    },
+    {
+      "epoch": 0.2964674174607758,
+      "grad_norm": 0.23725269734859467,
+      "learning_rate": 8.255412212597021e-06,
+      "loss": 0.798,
+      "step": 770
+    },
+    {
+      "epoch": 0.3003176436615651,
+      "grad_norm": 0.23205368220806122,
+      "learning_rate": 8.207830160055688e-06,
+      "loss": 0.7952,
+      "step": 780
+    },
+    {
+      "epoch": 0.3041678698623544,
+      "grad_norm": 0.24076993763446808,
+      "learning_rate": 8.159749166566786e-06,
+      "loss": 0.7782,
+      "step": 790
+    },
+    {
+      "epoch": 0.30801809606314373,
+      "grad_norm": 0.22522635757923126,
+      "learning_rate": 8.11117671057377e-06,
+      "loss": 0.8061,
+      "step": 800
+    },
+    {
+      "epoch": 0.311868322263933,
+      "grad_norm": 0.24596919119358063,
+      "learning_rate": 8.062120346961404e-06,
+      "loss": 0.7978,
+      "step": 810
+    },
+    {
+      "epoch": 0.3157185484647223,
+      "grad_norm": 0.23313851654529572,
+      "learning_rate": 8.012587705880705e-06,
+      "loss": 0.7911,
+      "step": 820
+    },
+    {
+      "epoch": 0.3195687746655116,
+      "grad_norm": 0.24417811632156372,
+      "learning_rate": 7.962586491562146e-06,
+      "loss": 0.7762,
+      "step": 830
+    },
+    {
+      "epoch": 0.3234190008663009,
+      "grad_norm": 0.2386322319507599,
+      "learning_rate": 7.912124481117357e-06,
+      "loss": 0.7841,
+      "step": 840
+    },
+    {
+      "epoch": 0.3272692270670902,
+      "grad_norm": 0.24011024832725525,
+      "learning_rate": 7.861209523329494e-06,
+      "loss": 0.7954,
+      "step": 850
+    },
+    {
+      "epoch": 0.3311194532678795,
+      "grad_norm": 0.23627501726150513,
+      "learning_rate": 7.809849537432432e-06,
+      "loss": 0.7765,
+      "step": 860
+    },
+    {
+      "epoch": 0.3349696794686688,
+      "grad_norm": 0.2356085181236267,
+      "learning_rate": 7.758052511879035e-06,
+      "loss": 0.7882,
+      "step": 870
+    },
+    {
+      "epoch": 0.3388199056694581,
+      "grad_norm": 0.22468313574790955,
+      "learning_rate": 7.705826503098642e-06,
+      "loss": 0.7833,
+      "step": 880
+    },
+    {
+      "epoch": 0.34267013187024736,
+      "grad_norm": 0.2310420721769333,
+      "learning_rate": 7.653179634243979e-06,
+      "loss": 0.7836,
+      "step": 890
+    },
+    {
+      "epoch": 0.34652035807103665,
+      "grad_norm": 0.21967269480228424,
+      "learning_rate": 7.600120093927696e-06,
+      "loss": 0.788,
+      "step": 900
+    },
+    {
+      "epoch": 0.350370584271826,
+      "grad_norm": 0.23642711341381073,
+      "learning_rate": 7.546656134948729e-06,
+      "loss": 0.777,
+      "step": 910
+    },
+    {
+      "epoch": 0.3542208104726153,
+      "grad_norm": 0.2328341156244278,
+      "learning_rate": 7.492796073008661e-06,
+      "loss": 0.7771,
+      "step": 920
+    },
+    {
+      "epoch": 0.35807103667340456,
+      "grad_norm": 0.24391691386699677,
+      "learning_rate": 7.438548285418326e-06,
+      "loss": 0.763,
+      "step": 930
+    },
+    {
+      "epoch": 0.36192126287419385,
+      "grad_norm": 0.25144580006599426,
+      "learning_rate": 7.3839212097948e-06,
+      "loss": 0.7976,
+      "step": 940
+    },
+    {
+      "epoch": 0.36577148907498314,
+      "grad_norm": 0.2433897703886032,
+      "learning_rate": 7.328923342749039e-06,
+      "loss": 0.8126,
+      "step": 950
+    },
+    {
+      "epoch": 0.3696217152757725,
+      "grad_norm": 0.25676780939102173,
+      "learning_rate": 7.273563238564319e-06,
+      "loss": 0.7963,
+      "step": 960
+    },
+    {
+      "epoch": 0.37347194147656176,
+      "grad_norm": 0.24101212620735168,
+      "learning_rate": 7.217849507865724e-06,
+      "loss": 0.7759,
+      "step": 970
+    },
+    {
+      "epoch": 0.37732216767735105,
+      "grad_norm": 0.2482462078332901,
+      "learning_rate": 7.161790816280858e-06,
+      "loss": 0.7833,
+      "step": 980
+    },
+    {
+      "epoch": 0.38117239387814034,
+      "grad_norm": 0.2459997832775116,
+      "learning_rate": 7.1053958830920125e-06,
+      "loss": 0.7663,
+      "step": 990
+    },
+    {
+      "epoch": 0.3850226200789296,
+      "grad_norm": 0.24794042110443115,
+      "learning_rate": 7.0486734798799715e-06,
+      "loss": 0.7862,
+      "step": 1000
+    },
+    {
+      "epoch": 0.3888728462797189,
+      "grad_norm": 0.23651771247386932,
+      "learning_rate": 6.991632429159708e-06,
+      "loss": 0.8081,
+      "step": 1010
+    },
+    {
+      "epoch": 0.39272307248050825,
+      "grad_norm": 0.25460511445999146,
+      "learning_rate": 6.934281603008132e-06,
+      "loss": 0.7994,
+      "step": 1020
+    },
+    {
+      "epoch": 0.39657329868129754,
+      "grad_norm": 0.23222504556179047,
+      "learning_rate": 6.876629921684143e-06,
+      "loss": 0.7956,
+      "step": 1030
+    },
+    {
+      "epoch": 0.4004235248820868,
+      "grad_norm": 0.24026376008987427,
+      "learning_rate": 6.818686352241199e-06,
+      "loss": 0.7866,
+      "step": 1040
+    },
+    {
+      "epoch": 0.4042737510828761,
+      "grad_norm": 0.21768783032894135,
+      "learning_rate": 6.760459907132573e-06,
+      "loss": 0.7821,
+      "step": 1050
+    },
+    {
+      "epoch": 0.4081239772836654,
+      "grad_norm": 0.2209068238735199,
+      "learning_rate": 6.701959642809589e-06,
+      "loss": 0.7886,
+      "step": 1060
+    },
+    {
+      "epoch": 0.41197420348445474,
+      "grad_norm": 0.23500356078147888,
+      "learning_rate": 6.643194658312987e-06,
+      "loss": 0.7813,
+      "step": 1070
+    },
+    {
+      "epoch": 0.415824429685244,
+      "grad_norm": 0.24396458268165588,
+      "learning_rate": 6.584174093857676e-06,
+      "loss": 0.8046,
+      "step": 1080
+    },
+    {
+      "epoch": 0.4196746558860333,
+      "grad_norm": 0.26096853613853455,
+      "learning_rate": 6.5249071294110735e-06,
+      "loss": 0.7862,
+      "step": 1090
+    },
+    {
+      "epoch": 0.4235248820868226,
+      "grad_norm": 0.2532426714897156,
+      "learning_rate": 6.465402983265274e-06,
+      "loss": 0.7992,
+      "step": 1100
+    },
+    {
+      "epoch": 0.4273751082876119,
+      "grad_norm": 0.23486118018627167,
+      "learning_rate": 6.405670910603243e-06,
+      "loss": 0.7958,
+      "step": 1110
+    },
+    {
+      "epoch": 0.43122533448840117,
+      "grad_norm": 0.2639244794845581,
+      "learning_rate": 6.345720202059291e-06,
+      "loss": 0.7918,
+      "step": 1120
+    },
+    {
+      "epoch": 0.4350755606891905,
+      "grad_norm": 0.24045570194721222,
+      "learning_rate": 6.285560182274008e-06,
+      "loss": 0.7695,
+      "step": 1130
+    },
+    {
+      "epoch": 0.4389257868899798,
+      "grad_norm": 0.23822703957557678,
+      "learning_rate": 6.225200208443936e-06,
+      "loss": 0.7927,
+      "step": 1140
+    },
+    {
+      "epoch": 0.4427760130907691,
+      "grad_norm": 0.21907489001750946,
+      "learning_rate": 6.16464966886616e-06,
+      "loss": 0.7826,
+      "step": 1150
+    },
+    {
+      "epoch": 0.44662623929155837,
+      "grad_norm": 0.26173973083496094,
+      "learning_rate": 6.103917981478065e-06,
+      "loss": 0.8053,
+      "step": 1160
+    },
+    {
+      "epoch": 0.45047646549234766,
+      "grad_norm": 0.23562347888946533,
+      "learning_rate": 6.043014592392489e-06,
+      "loss": 0.8164,
+      "step": 1170
+    },
+    {
+      "epoch": 0.454326691693137,
+      "grad_norm": 0.22702163457870483,
+      "learning_rate": 5.981948974428485e-06,
+      "loss": 0.7927,
+      "step": 1180
+    },
+    {
+      "epoch": 0.4581769178939263,
+      "grad_norm": 0.2484121173620224,
+      "learning_rate": 5.920730625637934e-06,
+      "loss": 0.7692,
+      "step": 1190
+    },
+    {
+      "epoch": 0.46202714409471557,
+      "grad_norm": 0.24692852795124054,
+      "learning_rate": 5.8593690678282355e-06,
+      "loss": 0.7822,
+      "step": 1200
+    },
+    {
+      "epoch": 0.46587737029550486,
+      "grad_norm": 0.24580352008342743,
+      "learning_rate": 5.797873845081301e-06,
+      "loss": 0.786,
+      "step": 1210
+    },
+    {
+      "epoch": 0.46972759649629414,
+      "grad_norm": 0.23985104262828827,
+      "learning_rate": 5.736254522269075e-06,
+      "loss": 0.8027,
+      "step": 1220
+    },
+    {
+      "epoch": 0.47357782269708343,
+      "grad_norm": 0.23553666472434998,
+      "learning_rate": 5.6745206835658386e-06,
+      "loss": 0.8039,
+      "step": 1230
+    },
+    {
+      "epoch": 0.47742804889787277,
+      "grad_norm": 0.22955124080181122,
+      "learning_rate": 5.612681930957495e-06,
+      "loss": 0.7753,
+      "step": 1240
+    },
+    {
+      "epoch": 0.48127827509866206,
+      "grad_norm": 0.21946637332439423,
+      "learning_rate": 5.550747882748095e-06,
+      "loss": 0.7725,
+      "step": 1250
+    },
+    {
+      "epoch": 0.48512850129945134,
+      "grad_norm": 0.25327205657958984,
+      "learning_rate": 5.4887281720638195e-06,
+      "loss": 0.7797,
+      "step": 1260
+    },
+    {
+      "epoch": 0.48897872750024063,
+      "grad_norm": 0.27041447162628174,
+      "learning_rate": 5.42663244535466e-06,
+      "loss": 0.7641,
+      "step": 1270
+    },
+    {
+      "epoch": 0.4928289537010299,
+      "grad_norm": 0.25960516929626465,
+      "learning_rate": 5.3644703608940205e-06,
+      "loss": 0.7719,
+      "step": 1280
+    },
+    {
+      "epoch": 0.49667917990181926,
+      "grad_norm": 0.23603346943855286,
+      "learning_rate": 5.302251587276491e-06,
+      "loss": 0.784,
+      "step": 1290
+    },
+    {
+      "epoch": 0.5005294061026085,
+      "grad_norm": 0.22157606482505798,
+      "learning_rate": 5.2399858019140005e-06,
+      "loss": 0.7844,
+      "step": 1300
+    },
+    {
+      "epoch": 0.5043796323033978,
+      "grad_norm": 0.2334025353193283,
+      "learning_rate": 5.177682689530615e-06,
+      "loss": 0.7993,
+      "step": 1310
+    },
+    {
+      "epoch": 0.5082298585041871,
+      "grad_norm": 0.24116690456867218,
+      "learning_rate": 5.11535194065619e-06,
+      "loss": 0.7665,
+      "step": 1320
+    },
+    {
+      "epoch": 0.5120800847049765,
+      "grad_norm": 0.2622392177581787,
+      "learning_rate": 5.053003250119113e-06,
+      "loss": 0.7767,
+      "step": 1330
+    },
+    {
+      "epoch": 0.5159303109057657,
+      "grad_norm": 0.23988305032253265,
+      "learning_rate": 4.9906463155383925e-06,
+      "loss": 0.784,
+      "step": 1340
+    },
+    {
+      "epoch": 0.519780537106555,
+      "grad_norm": 0.23128847777843475,
+      "learning_rate": 4.9282908358153046e-06,
+      "loss": 0.7781,
+      "step": 1350
+    },
+    {
+      "epoch": 0.5236307633073443,
+      "grad_norm": 0.2620639503002167,
+      "learning_rate": 4.8659465096248345e-06,
+      "loss": 0.7937,
+      "step": 1360
+    },
+    {
+      "epoch": 0.5274809895081336,
+      "grad_norm": 0.25802871584892273,
+      "learning_rate": 4.803623033907169e-06,
+      "loss": 0.7925,
+      "step": 1370
+    },
+    {
+      "epoch": 0.5313312157089229,
+      "grad_norm": 0.23871995508670807,
+      "learning_rate": 4.741330102359441e-06,
+      "loss": 0.7983,
+      "step": 1380
+    },
+    {
+      "epoch": 0.5351814419097122,
+      "grad_norm": 0.25129371881484985,
+      "learning_rate": 4.679077403927996e-06,
+      "loss": 0.7788,
+      "step": 1390
+    },
+    {
+      "epoch": 0.5390316681105015,
+      "grad_norm": 0.23405329883098602,
+      "learning_rate": 4.616874621301373e-06,
+      "loss": 0.7781,
+      "step": 1400
+    },
+    {
+      "epoch": 0.5428818943112907,
+      "grad_norm": 0.2454117238521576,
+      "learning_rate": 4.554731429404293e-06,
+      "loss": 0.793,
+      "step": 1410
+    },
+    {
+      "epoch": 0.5467321205120801,
+      "grad_norm": 0.25545695424079895,
+      "learning_rate": 4.492657493892822e-06,
+      "loss": 0.7768,
+      "step": 1420
+    },
+    {
+      "epoch": 0.5505823467128694,
+      "grad_norm": 0.25972822308540344,
+      "learning_rate": 4.430662469650999e-06,
+      "loss": 0.7917,
+      "step": 1430
+    },
+    {
+      "epoch": 0.5544325729136587,
+      "grad_norm": 0.2612539529800415,
+      "learning_rate": 4.368755999289118e-06,
+      "loss": 0.7952,
+      "step": 1440
+    },
+    {
+      "epoch": 0.558282799114448,
+      "grad_norm": 0.2546044588088989,
+      "learning_rate": 4.306947711643949e-06,
+      "loss": 0.7799,
+      "step": 1450
+    },
+    {
+      "epoch": 0.5621330253152372,
+      "grad_norm": 0.24463561177253723,
+      "learning_rate": 4.245247220281068e-06,
+      "loss": 0.7827,
+      "step": 1460
+    },
+    {
+      "epoch": 0.5659832515160266,
+      "grad_norm": 0.23627039790153503,
+      "learning_rate": 4.183664121999587e-06,
+      "loss": 0.7813,
+      "step": 1470
+    },
+    {
+      "epoch": 0.5698334777168159,
+      "grad_norm": 0.24719104170799255,
+      "learning_rate": 4.122207995339474e-06,
+      "loss": 0.7769,
+      "step": 1480
+    },
+    {
+      "epoch": 0.5736837039176051,
+      "grad_norm": 0.2294992059469223,
+      "learning_rate": 4.060888399091734e-06,
+      "loss": 0.7823,
+      "step": 1490
+    },
+    {
+      "epoch": 0.5775339301183945,
+      "grad_norm": 0.2362632006406784,
+      "learning_rate": 3.999714870811639e-06,
+      "loss": 0.791,
+      "step": 1500
+    },
+    {
+      "epoch": 0.5813841563191837,
+      "grad_norm": 0.23332621157169342,
+      "learning_rate": 3.938696925335275e-06,
+      "loss": 0.7837,
+      "step": 1510
+    },
+    {
+      "epoch": 0.5852343825199731,
+      "grad_norm": 0.23063881695270538,
+      "learning_rate": 3.87784405329962e-06,
+      "loss": 0.7918,
+      "step": 1520
+    },
+    {
+      "epoch": 0.5890846087207623,
+      "grad_norm": 0.2321629673242569,
+      "learning_rate": 3.817165719666382e-06,
+      "loss": 0.794,
+      "step": 1530
+    },
+    {
+      "epoch": 0.5929348349215516,
+      "grad_norm": 0.2358732670545578,
+      "learning_rate": 3.756671362249841e-06,
+      "loss": 0.7972,
+      "step": 1540
+    },
+    {
+      "epoch": 0.596785061122341,
+      "grad_norm": 0.23626816272735596,
+      "learning_rate": 3.6963703902488907e-06,
+      "loss": 0.7835,
+      "step": 1550
+    },
+    {
+      "epoch": 0.6006352873231302,
+      "grad_norm": 0.23058828711509705,
+      "learning_rate": 3.6362721827835655e-06,
+      "loss": 0.8012,
+      "step": 1560
+    },
+    {
+      "epoch": 0.6044855135239195,
+      "grad_norm": 0.24320633709430695,
+      "learning_rate": 3.576386087436212e-06,
+      "loss": 0.7959,
+      "step": 1570
+    },
+    {
+      "epoch": 0.6083357397247088,
+      "grad_norm": 0.2469819188117981,
+      "learning_rate": 3.5167214187975913e-06,
+      "loss": 0.7857,
+      "step": 1580
+    },
+    {
+      "epoch": 0.6121859659254981,
+      "grad_norm": 0.21447835862636566,
+      "learning_rate": 3.457287457018086e-06,
+      "loss": 0.789,
+      "step": 1590
+    },
+    {
+      "epoch": 0.6160361921262875,
+      "grad_norm": 0.2365657091140747,
+      "learning_rate": 3.3980934463643003e-06,
+      "loss": 0.7756,
+      "step": 1600
+    },
+    {
+      "epoch": 0.6198864183270767,
+      "grad_norm": 0.2202494591474533,
+      "learning_rate": 3.3391485937812053e-06,
+      "loss": 0.7602,
+      "step": 1610
+    },
+    {
+      "epoch": 0.623736644527866,
+      "grad_norm": 0.2382076382637024,
+      "learning_rate": 3.280462067460115e-06,
+      "loss": 0.7579,
+      "step": 1620
+    },
+    {
+      "epoch": 0.6275868707286553,
+      "grad_norm": 0.2392725944519043,
+      "learning_rate": 3.222042995412669e-06,
+      "loss": 0.7891,
+      "step": 1630
+    },
+    {
+      "epoch": 0.6314370969294446,
+      "grad_norm": 0.22062397003173828,
+      "learning_rate": 3.163900464051089e-06,
+      "loss": 0.778,
+      "step": 1640
+    },
+    {
+      "epoch": 0.635287323130234,
+      "grad_norm": 0.2426428496837616,
+      "learning_rate": 3.1060435167748826e-06,
+      "loss": 0.7885,
+      "step": 1650
+    },
+    {
+      "epoch": 0.6391375493310232,
+      "grad_norm": 0.22918730974197388,
+      "learning_rate": 3.048481152564253e-06,
+      "loss": 0.7687,
+      "step": 1660
+    },
+    {
+      "epoch": 0.6429877755318125,
+      "grad_norm": 0.24407172203063965,
+      "learning_rate": 2.9912223245804017e-06,
+      "loss": 0.7921,
+      "step": 1670
+    },
+    {
+      "epoch": 0.6468380017326018,
+      "grad_norm": 0.24335666000843048,
+      "learning_rate": 2.934275938772981e-06,
+      "loss": 0.7867,
+      "step": 1680
+    },
+    {
+      "epoch": 0.6506882279333911,
+      "grad_norm": 0.2539544105529785,
+      "learning_rate": 2.8776508524948676e-06,
+      "loss": 0.7766,
+      "step": 1690
+    },
+    {
+      "epoch": 0.6545384541341804,
+      "grad_norm": 0.216528058052063,
+      "learning_rate": 2.821355873124503e-06,
+      "loss": 0.7747,
+      "step": 1700
+    },
+    {
+      "epoch": 0.6583886803349697,
+      "grad_norm": 0.2588459849357605,
+      "learning_rate": 2.7653997566960144e-06,
+      "loss": 0.7902,
+      "step": 1710
+    },
+    {
+      "epoch": 0.662238906535759,
+      "grad_norm": 0.24680231511592865,
+      "learning_rate": 2.709791206537309e-06,
+      "loss": 0.779,
+      "step": 1720
+    },
+    {
+      "epoch": 0.6660891327365482,
+      "grad_norm": 0.24186529219150543,
+      "learning_rate": 2.6545388719163747e-06,
+      "loss": 0.7845,
+      "step": 1730
+    },
+    {
+      "epoch": 0.6699393589373376,
+      "grad_norm": 0.23537187278270721,
+      "learning_rate": 2.599651346695979e-06,
+      "loss": 0.7835,
+      "step": 1740
+    },
+    {
+      "epoch": 0.6737895851381268,
+      "grad_norm": 0.24043671786785126,
+      "learning_rate": 2.545137167997007e-06,
+      "loss": 0.779,
+      "step": 1750
+    },
+    {
+      "epoch": 0.6776398113389162,
+      "grad_norm": 0.2242034524679184,
+      "learning_rate": 2.4910048148705972e-06,
+      "loss": 0.7669,
+      "step": 1760
+    },
+    {
+      "epoch": 0.6814900375397055,
+      "grad_norm": 0.25486162304878235,
+      "learning_rate": 2.4372627069793327e-06,
+      "loss": 0.8115,
+      "step": 1770
+    },
+    {
+      "epoch": 0.6853402637404947,
+      "grad_norm": 0.24263730645179749,
+      "learning_rate": 2.3839192032876486e-06,
+      "loss": 0.7961,
+      "step": 1780
+    },
+    {
+      "epoch": 0.6891904899412841,
+      "grad_norm": 0.23182813823223114,
+      "learning_rate": 2.3309826007617047e-06,
+      "loss": 0.8079,
+      "step": 1790
+    },
+    {
+      "epoch": 0.6930407161420733,
+      "grad_norm": 0.2287960648536682,
+      "learning_rate": 2.278461133078881e-06,
+      "loss": 0.7714,
+      "step": 1800
+    },
+    {
+      "epoch": 0.6968909423428626,
+      "grad_norm": 0.24396851658821106,
+      "learning_rate": 2.2263629693471258e-06,
+      "loss": 0.7624,
+      "step": 1810
+    },
+    {
+      "epoch": 0.700741168543652,
+      "grad_norm": 0.22870798408985138,
+      "learning_rate": 2.1746962128343414e-06,
+      "loss": 0.7868,
+      "step": 1820
+    },
+    {
+      "epoch": 0.7045913947444412,
+      "grad_norm": 0.22622276842594147,
+      "learning_rate": 2.1234688997080183e-06,
+      "loss": 0.7754,
+      "step": 1830
+    },
+    {
+      "epoch": 0.7084416209452306,
+      "grad_norm": 0.24980680644512177,
+      "learning_rate": 2.0726889977853053e-06,
+      "loss": 0.7802,
+      "step": 1840
+    },
+    {
+      "epoch": 0.7122918471460198,
+      "grad_norm": 0.24849389493465424,
+      "learning_rate": 2.022364405293703e-06,
+      "loss": 0.7936,
+      "step": 1850
+    },
+    {
+      "epoch": 0.7161420733468091,
+      "grad_norm": 0.23007017374038696,
+      "learning_rate": 1.9725029496425814e-06,
+      "loss": 0.7855,
+      "step": 1860
+    },
+    {
+      "epoch": 0.7199922995475985,
+      "grad_norm": 0.24519391357898712,
+      "learning_rate": 1.923112386205727e-06,
+      "loss": 0.7819,
+      "step": 1870
+    },
+    {
+      "epoch": 0.7238425257483877,
+      "grad_norm": 0.22870104014873505,
+      "learning_rate": 1.874200397115079e-06,
+      "loss": 0.7894,
+      "step": 1880
+    },
+    {
+      "epoch": 0.727692751949177,
+      "grad_norm": 0.2449340522289276,
+      "learning_rate": 1.825774590065857e-06,
+      "loss": 0.7738,
+      "step": 1890
+    },
+    {
+      "epoch": 0.7315429781499663,
+      "grad_norm": 0.2439175844192505,
+      "learning_rate": 1.7778424971332813e-06,
+      "loss": 0.7816,
+      "step": 1900
+    },
+    {
+      "epoch": 0.7353932043507556,
+      "grad_norm": 0.2305448204278946,
+      "learning_rate": 1.7304115736010407e-06,
+      "loss": 0.7783,
+      "step": 1910
+    },
+    {
+      "epoch": 0.739243430551545,
+      "grad_norm": 0.23463775217533112,
+      "learning_rate": 1.6834891968017108e-06,
+      "loss": 0.7723,
+      "step": 1920
+    },
+    {
+      "epoch": 0.7430936567523342,
+      "grad_norm": 0.24560582637786865,
+      "learning_rate": 1.6370826649692884e-06,
+      "loss": 0.7909,
+      "step": 1930
+    },
+    {
+      "epoch": 0.7469438829531235,
+      "grad_norm": 0.24097521603107452,
+      "learning_rate": 1.5911991961040463e-06,
+      "loss": 0.7977,
+      "step": 1940
+    },
+    {
+      "epoch": 0.7507941091539128,
+      "grad_norm": 0.24416764080524445,
+      "learning_rate": 1.545845926849851e-06,
+      "loss": 0.7858,
+      "step": 1950
+    },
+    {
+      "epoch": 0.7546443353547021,
+      "grad_norm": 0.2517002522945404,
+      "learning_rate": 1.5010299113841397e-06,
+      "loss": 0.8028,
+      "step": 1960
+    },
+    {
+      "epoch": 0.7584945615554913,
+      "grad_norm": 0.24171540141105652,
+      "learning_rate": 1.4567581203207238e-06,
+      "loss": 0.7928,
+      "step": 1970
+    },
+    {
+      "epoch": 0.7623447877562807,
+      "grad_norm": 0.25159895420074463,
+      "learning_rate": 1.413037439625593e-06,
+      "loss": 0.7899,
+      "step": 1980
+    },
+    {
+      "epoch": 0.76619501395707,
+      "grad_norm": 0.2537609338760376,
+      "learning_rate": 1.3698746695458847e-06,
+      "loss": 0.786,
+      "step": 1990
+    },
+    {
+      "epoch": 0.7700452401578592,
+      "grad_norm": 0.23461776971817017,
+      "learning_rate": 1.3272765235521745e-06,
+      "loss": 0.7771,
+      "step": 2000
+    },
+    {
+      "epoch": 0.7738954663586486,
+      "grad_norm": 0.2353181689977646,
+      "learning_rate": 1.2852496272942877e-06,
+      "loss": 0.7842,
+      "step": 2010
+    },
+    {
+      "epoch": 0.7777456925594378,
+      "grad_norm": 0.22375857830047607,
+      "learning_rate": 1.2438005175707396e-06,
+      "loss": 0.7729,
+      "step": 2020
+    },
+    {
+      "epoch": 0.7815959187602272,
+      "grad_norm": 0.23166850209236145,
+      "learning_rate": 1.2029356413120297e-06,
+      "loss": 0.7572,
+      "step": 2030
+    },
+    {
+      "epoch": 0.7854461449610165,
+      "grad_norm": 0.23196128010749817,
+      "learning_rate": 1.1626613545778786e-06,
+      "loss": 0.7747,
+      "step": 2040
+    },
+    {
+      "epoch": 0.7892963711618057,
+      "grad_norm": 0.22527064383029938,
+      "learning_rate": 1.1229839215686294e-06,
+      "loss": 0.7798,
+      "step": 2050
+    },
+    {
+      "epoch": 0.7931465973625951,
+      "grad_norm": 0.22074390947818756,
+      "learning_rate": 1.0839095136509188e-06,
+      "loss": 0.7633,
+      "step": 2060
+    },
+    {
+      "epoch": 0.7969968235633843,
+      "grad_norm": 0.25366634130477905,
+      "learning_rate": 1.045444208397791e-06,
+      "loss": 0.7858,
+      "step": 2070
+    },
+    {
+      "epoch": 0.8008470497641736,
+      "grad_norm": 0.23344238102436066,
+      "learning_rate": 1.007593988643402e-06,
+      "loss": 0.7671,
+      "step": 2080
+    },
+    {
+      "epoch": 0.804697275964963,
+      "grad_norm": 0.23032239079475403,
+      "learning_rate": 9.703647415524613e-07,
+      "loss": 0.7881,
+      "step": 2090
+    },
+    {
+      "epoch": 0.8085475021657522,
+      "grad_norm": 0.2407863885164261,
+      "learning_rate": 9.337622577045485e-07,
+      "loss": 0.7775,
+      "step": 2100
+    },
+    {
+      "epoch": 0.8123977283665416,
+      "grad_norm": 0.2328445315361023,
+      "learning_rate": 8.977922301934588e-07,
+      "loss": 0.7865,
+      "step": 2110
+    },
+    {
+      "epoch": 0.8162479545673308,
+      "grad_norm": 0.23883043229579926,
+      "learning_rate": 8.624602537417009e-07,
+      "loss": 0.7891,
+      "step": 2120
+    },
+    {
+      "epoch": 0.8200981807681201,
+      "grad_norm": 0.25502902269363403,
+      "learning_rate": 8.277718238303123e-07,
+      "loss": 0.7769,
+      "step": 2130
+    },
+    {
+      "epoch": 0.8239484069689095,
+      "grad_norm": 0.23556572198867798,
+      "learning_rate": 7.937323358440935e-07,
+      "loss": 0.7802,
+      "step": 2140
+    },
+    {
+      "epoch": 0.8277986331696987,
+      "grad_norm": 0.24671900272369385,
+      "learning_rate": 7.603470842324223e-07,
+      "loss": 0.7724,
+      "step": 2150
+    },
+    {
+      "epoch": 0.831648859370488,
+      "grad_norm": 0.2238111048936844,
+      "learning_rate": 7.276212616857575e-07,
+      "loss": 0.789,
+      "step": 2160
+    },
+    {
+      "epoch": 0.8354990855712773,
+      "grad_norm": 0.2295861393213272,
+      "learning_rate": 6.955599583279837e-07,
+      "loss": 0.7878,
+      "step": 2170
+    },
+    {
+      "epoch": 0.8393493117720666,
+      "grad_norm": 0.2400270700454712,
+      "learning_rate": 6.641681609246981e-07,
+      "loss": 0.7805,
+      "step": 2180
+    },
+    {
+      "epoch": 0.8431995379728559,
+      "grad_norm": 0.2362034171819687,
+      "learning_rate": 6.334507521075711e-07,
+      "loss": 0.7874,
+      "step": 2190
+    },
+    {
+      "epoch": 0.8470497641736452,
+      "grad_norm": 0.2524145245552063,
+      "learning_rate": 6.034125096149179e-07,
+      "loss": 0.781,
+      "step": 2200
+    },
+    {
+      "epoch": 0.8508999903744345,
+      "grad_norm": 0.23924002051353455,
+      "learning_rate": 5.740581055485722e-07,
+      "loss": 0.7542,
+      "step": 2210
+    },
+    {
+      "epoch": 0.8547502165752238,
+      "grad_norm": 0.2233283519744873,
+      "learning_rate": 5.453921056471961e-07,
+      "loss": 0.7875,
+      "step": 2220
+    },
+    {
+      "epoch": 0.8586004427760131,
+      "grad_norm": 0.2467336505651474,
+      "learning_rate": 5.174189685761288e-07,
+      "loss": 0.7811,
+      "step": 2230
+    },
+    {
+      "epoch": 0.8624506689768023,
+      "grad_norm": 0.22662894427776337,
+      "learning_rate": 4.901430452338979e-07,
+      "loss": 0.7932,
+      "step": 2240
+    },
+    {
+      "epoch": 0.8663008951775917,
+      "grad_norm": 0.23071379959583282,
+      "learning_rate": 4.6356857807548286e-07,
+      "loss": 0.7755,
+      "step": 2250
+    },
+    {
+      "epoch": 0.870151121378381,
+      "grad_norm": 0.23938743770122528,
+      "learning_rate": 4.376997004524541e-07,
+      "loss": 0.7649,
+      "step": 2260
+    },
+    {
+      "epoch": 0.8740013475791703,
+      "grad_norm": 0.23746173083782196,
+      "learning_rate": 4.12540435970073e-07,
+      "loss": 0.7745,
+      "step": 2270
+    },
+    {
+      "epoch": 0.8778515737799596,
+      "grad_norm": 0.23484936356544495,
+      "learning_rate": 3.880946978614703e-07,
+      "loss": 0.787,
+      "step": 2280
+    },
+    {
+      "epoch": 0.8817017999807488,
+      "grad_norm": 0.23110522329807281,
+      "learning_rate": 3.643662883789878e-07,
+      "loss": 0.7904,
+      "step": 2290
+    },
+    {
+      "epoch": 0.8855520261815382,
+      "grad_norm": 0.21828646957874298,
+      "learning_rate": 3.41358898202781e-07,
+      "loss": 0.767,
+      "step": 2300
+    },
+    {
+      "epoch": 0.8894022523823275,
+      "grad_norm": 0.22388425469398499,
+      "learning_rate": 3.190761058667741e-07,
+      "loss": 0.7779,
+      "step": 2310
+    },
+    {
+      "epoch": 0.8932524785831167,
+      "grad_norm": 0.23523356020450592,
+      "learning_rate": 2.9752137720206463e-07,
+      "loss": 0.7838,
+      "step": 2320
+    },
+    {
+      "epoch": 0.8971027047839061,
+      "grad_norm": 0.23452426493167877,
+      "learning_rate": 2.7669806479785067e-07,
+      "loss": 0.789,
+      "step": 2330
+    },
+    {
+      "epoch": 0.9009529309846953,
+      "grad_norm": 0.24388420581817627,
+      "learning_rate": 2.566094074799763e-07,
+      "loss": 0.7836,
+      "step": 2340
+    },
+    {
+      "epoch": 0.9048031571854847,
+      "grad_norm": 0.23441962897777557,
+      "learning_rate": 2.3725852980716557e-07,
+      "loss": 0.7651,
+      "step": 2350
+    },
+    {
+      "epoch": 0.908653383386274,
+      "grad_norm": 0.22986198961734772,
+      "learning_rate": 2.1864844158503983e-07,
+      "loss": 0.7788,
+      "step": 2360
+    },
+    {
+      "epoch": 0.9125036095870632,
+      "grad_norm": 0.2651938498020172,
+      "learning_rate": 2.007820373979702e-07,
+      "loss": 0.7706,
+      "step": 2370
+    },
+    {
+      "epoch": 0.9163538357878526,
+      "grad_norm": 0.2372904270887375,
+      "learning_rate": 1.8366209615886532e-07,
+      "loss": 0.778,
+      "step": 2380
+    },
+    {
+      "epoch": 0.9202040619886418,
+      "grad_norm": 0.2266344577074051,
+      "learning_rate": 1.6729128067693613e-07,
+      "loss": 0.7905,
+      "step": 2390
+    },
+    {
+      "epoch": 0.9240542881894311,
+      "grad_norm": 0.24602758884429932,
+      "learning_rate": 1.5167213724353426e-07,
+      "loss": 0.7906,
+      "step": 2400
+    },
+    {
+      "epoch": 0.9279045143902205,
+      "grad_norm": 0.22600361704826355,
+      "learning_rate": 1.3680709523610058e-07,
+      "loss": 0.7749,
+      "step": 2410
+    },
+    {
+      "epoch": 0.9317547405910097,
+      "grad_norm": 0.2215646356344223,
+      "learning_rate": 1.2269846674030418e-07,
+      "loss": 0.7955,
+      "step": 2420
+    },
+    {
+      "epoch": 0.935604966791799,
+      "grad_norm": 0.23332564532756805,
+      "learning_rate": 1.0934844619042595e-07,
+      "loss": 0.7925,
+      "step": 2430
+    },
+    {
+      "epoch": 0.9394551929925883,
+      "grad_norm": 0.24401888251304626,
+      "learning_rate": 9.675911002803961e-08,
+      "loss": 0.7802,
+      "step": 2440
+    },
+    {
+      "epoch": 0.9433054191933776,
+      "grad_norm": 0.2565440237522125,
+      "learning_rate": 8.493241637904381e-08,
+      "loss": 0.7776,
+      "step": 2450
+    },
+    {
+      "epoch": 0.9471556453941669,
+      "grad_norm": 0.24122463166713715,
+      "learning_rate": 7.387020474909689e-08,
+      "loss": 0.7986,
+      "step": 2460
+    },
+    {
+      "epoch": 0.9510058715949562,
+      "grad_norm": 0.255474716424942,
+      "learning_rate": 6.357419573750856e-08,
+      "loss": 0.7916,
+      "step": 2470
+    },
+    {
+      "epoch": 0.9548560977957455,
+      "grad_norm": 0.2360447198152542,
+      "learning_rate": 5.4045990769613365e-08,
+      "loss": 0.7754,
+      "step": 2480
+    },
+    {
+      "epoch": 0.9587063239965348,
+      "grad_norm": 0.2830134630203247,
+      "learning_rate": 4.5287071847694434e-08,
+      "loss": 0.7799,
+      "step": 2490
+    },
+    {
+      "epoch": 0.9625565501973241,
+      "grad_norm": 0.23016749322414398,
+      "learning_rate": 3.7298801320467817e-08,
+      "loss": 0.7745,
+      "step": 2500
+    },
+    {
+      "epoch": 0.9664067763981133,
+      "grad_norm": 0.23500953614711761,
+      "learning_rate": 3.0082421671192576e-08,
+      "loss": 0.7675,
+      "step": 2510
+    },
+    {
+      "epoch": 0.9702570025989027,
+      "grad_norm": 0.2370045781135559,
+      "learning_rate": 2.3639055324410355e-08,
+      "loss": 0.7912,
+      "step": 2520
+    },
+    {
+      "epoch": 0.974107228799692,
+      "grad_norm": 0.2312004566192627,
+      "learning_rate": 1.7969704471369477e-08,
+      "loss": 0.7738,
+      "step": 2530
+    },
+    {
+      "epoch": 0.9779574550004813,
+      "grad_norm": 0.22838528454303741,
+      "learning_rate": 1.3075250914142413e-08,
+      "loss": 0.7776,
+      "step": 2540
+    },
+    {
+      "epoch": 0.9818076812012706,
+      "grad_norm": 0.2291276603937149,
+      "learning_rate": 8.956455928476049e-09,
+      "loss": 0.7865,
+      "step": 2550
+    },
+    {
+      "epoch": 0.9856579074020598,
+      "grad_norm": 0.23963087797164917,
+      "learning_rate": 5.6139601453786275e-09,
+      "loss": 0.7778,
+      "step": 2560
+    },
+    {
+      "epoch": 0.9895081336028492,
+      "grad_norm": 0.22988466918468475,
+      "learning_rate": 3.0482834514805602e-09,
+      "loss": 0.7666,
+      "step": 2570
+    },
+    {
+      "epoch": 0.9933583598036385,
+      "grad_norm": 0.2149686962366104,
+      "learning_rate": 1.2598249081713366e-09,
+      "loss": 0.7621,
+      "step": 2580
+    },
+    {
+      "epoch": 0.9972085860044277,
+      "grad_norm": 0.2561975419521332,
+      "learning_rate": 2.488626895302826e-10,
+      "loss": 0.7923,
+      "step": 2590
+    },
+    {
+      "epoch": 0.9999037443449803,
+      "step": 2597,
+      "total_flos": 3.609381421241598e+18,
+      "train_loss": 0.810214508851308,
+      "train_runtime": 10346.9478,
+      "train_samples_per_second": 64.261,
+      "train_steps_per_second": 0.251
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 2597,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.609381421241598e+18,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}

continuous/qwen3/mc2_s16_224_mt5/finetune/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:00c7a90d54c322ba721c840166baf37959e2542e96a08c794b56a0722dee3f0c
+size 6136