Upload folder using huggingface_hub

.gitattributes

@@ -33,3 +33,8 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+checkpoint-1074/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-1611/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-2148/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-537/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

chat_template.jinja

@@ -0,0 +1,120 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {%- if messages[0].content is string %}
+            {{- messages[0].content }}
+        {%- else %}
+            {%- for content in messages[0].content %}
+                {%- if 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '\n\n' }}
+    {%- endif %}
+    {{- "# 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>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\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" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' }}
+        {%- if messages[0].content is string %}
+            {{- messages[0].content }}
+        {%- else %}
+            {%- for content in messages[0].content %}
+                {%- if 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- for message in messages %}
+    {%- if message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- else %}
+            {%- for content in message.content %}
+                {%- if content.type == 'image' or 'image' in content or 'image_url' in content %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                    {%- if add_vision_id %}Picture {{ image_count.value }}: {% endif -%}
+                    <|vision_start|><|image_pad|><|vision_end|>
+                {%- elif content.type == 'video' or 'video' in content %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                    {%- if add_vision_id %}Video {{ video_count.value }}: {% endif -%}
+                    <|vision_start|><|video_pad|><|vision_end|>
+                {%- elif 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role + '\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- else %}
+            {%- for content_item in message.content %}
+                {%- if 'text' in content_item %}
+                    {{- content_item.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and message.content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- else %}
+            {%- for content in message.content %}
+                {%- if content.type == 'image' or 'image' in content or 'image_url' in content %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                    {%- if add_vision_id %}Picture {{ image_count.value }}: {% endif -%}
+                    <|vision_start|><|image_pad|><|vision_end|>
+                {%- elif content.type == 'video' or 'video' in content %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                    {%- if add_vision_id %}Video {{ video_count.value }}: {% endif -%}
+                    <|vision_start|><|video_pad|><|vision_end|>
+                {%- elif 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}

checkpoint-1074/chat_template.jinja

@@ -0,0 +1,120 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {%- if messages[0].content is string %}
+            {{- messages[0].content }}
+        {%- else %}
+            {%- for content in messages[0].content %}
+                {%- if 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '\n\n' }}
+    {%- endif %}
+    {{- "# 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>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\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" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' }}
+        {%- if messages[0].content is string %}
+            {{- messages[0].content }}
+        {%- else %}
+            {%- for content in messages[0].content %}
+                {%- if 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- for message in messages %}
+    {%- if message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- else %}
+            {%- for content in message.content %}
+                {%- if content.type == 'image' or 'image' in content or 'image_url' in content %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                    {%- if add_vision_id %}Picture {{ image_count.value }}: {% endif -%}
+                    <|vision_start|><|image_pad|><|vision_end|>
+                {%- elif content.type == 'video' or 'video' in content %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                    {%- if add_vision_id %}Video {{ video_count.value }}: {% endif -%}
+                    <|vision_start|><|video_pad|><|vision_end|>
+                {%- elif 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role + '\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- else %}
+            {%- for content_item in message.content %}
+                {%- if 'text' in content_item %}
+                    {{- content_item.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and message.content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- else %}
+            {%- for content in message.content %}
+                {%- if content.type == 'image' or 'image' in content or 'image_url' in content %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                    {%- if add_vision_id %}Picture {{ image_count.value }}: {% endif -%}
+                    <|vision_start|><|image_pad|><|vision_end|>
+                {%- elif content.type == 'video' or 'video' in content %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                    {%- if add_vision_id %}Video {{ video_count.value }}: {% endif -%}
+                    <|vision_start|><|video_pad|><|vision_end|>
+                {%- elif 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}

checkpoint-1074/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0a822468e63ccc5e3a89ea813c674821eca56c240c0ad411c5748ad6b794bac2
+size 4885870044

checkpoint-1074/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:da0749c3bb3346b04b9a602a711d76441b26fd2eed502530ad4e47ec2220a1b3
+size 8527467778

checkpoint-1074/processor_config.json

@@ -0,0 +1,63 @@
+{
+  "image_processor": {
+    "data_format": "channels_first",
+    "do_convert_rgb": true,
+    "do_normalize": true,
+    "do_rescale": true,
+    "do_resize": true,
+    "image_mean": [
+      0.5,
+      0.5,
+      0.5
+    ],
+    "image_processor_type": "Qwen2VLImageProcessorFast",
+    "image_std": [
+      0.5,
+      0.5,
+      0.5
+    ],
+    "merge_size": 2,
+    "patch_size": 16,
+    "resample": 3,
+    "rescale_factor": 0.00392156862745098,
+    "size": {
+      "longest_edge": 16777216,
+      "shortest_edge": 65536
+    },
+    "temporal_patch_size": 2
+  },
+  "processor_class": "Qwen3VLProcessor",
+  "video_processor": {
+    "data_format": "channels_first",
+    "default_to_square": true,
+    "do_convert_rgb": true,
+    "do_normalize": true,
+    "do_rescale": true,
+    "do_resize": true,
+    "do_sample_frames": true,
+    "fps": 2,
+    "image_mean": [
+      0.5,
+      0.5,
+      0.5
+    ],
+    "image_std": [
+      0.5,
+      0.5,
+      0.5
+    ],
+    "max_frames": 768,
+    "merge_size": 2,
+    "min_frames": 4,
+    "patch_size": 16,
+    "resample": 3,
+    "rescale_factor": 0.00392156862745098,
+    "return_metadata": false,
+    "size": {
+      "longest_edge": 25165824,
+      "shortest_edge": 4096
+    },
+    "temporal_patch_size": 2,
+    "video_processor_type": "Qwen3VLVideoProcessor"
+  }
+}

checkpoint-1074/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:659b1cdee2219458dd84ce6a632a595465680b8080e5c44bd600ff97eca8d752
+size 15429

checkpoint-1074/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:86accf27064cdd503053e90476a6bd10de333d4ff0594535ad55ea13a473c91d
+size 15429

checkpoint-1074/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:18ca8d714ef40be035404c1957b5a4dee84e1f43980408393f8aa710552ee6f6
+size 15429

checkpoint-1074/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2cfdebe99e40accc9c9d8f09c63136a14abda997d9b501969ec8e16e9d183179
+size 15429

checkpoint-1074/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:03087e7be8d47f2b2f6c10e1f57235a3d5d197cc743ae7ab37aeb711f41ebee8
+size 1465

checkpoint-1074/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8579e1ca7cc5d82a9e0202eed555529996f4ffe7f563c2979a0290cf3db452d3
+size 11422818

checkpoint-1074/tokenizer_config.json

@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_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|>"
+  ],
+  "is_local": false,
+  "model_max_length": 262144,
+  "pad_token": "<|endoftext|>",
+  "processor_class": "Qwen3VLProcessor",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

checkpoint-1074/trainer_state.json

@@ -0,0 +1,1658 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 50,
+  "global_step": 1074,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.009324009324009324,
+      "grad_norm": 269.7618103027344,
+      "learning_rate": 1.8604651162790698e-07,
+      "loss": 13.883856201171875,
+      "step": 5
+    },
+    {
+      "epoch": 0.018648018648018648,
+      "grad_norm": 256.5934143066406,
+      "learning_rate": 4.186046511627907e-07,
+      "loss": 13.883564758300782,
+      "step": 10
+    },
+    {
+      "epoch": 0.027972027972027972,
+      "grad_norm": 247.4810028076172,
+      "learning_rate": 6.511627906976745e-07,
+      "loss": 13.845947265625,
+      "step": 15
+    },
+    {
+      "epoch": 0.037296037296037296,
+      "grad_norm": 243.70272827148438,
+      "learning_rate": 8.837209302325582e-07,
+      "loss": 13.705046081542969,
+      "step": 20
+    },
+    {
+      "epoch": 0.046620046620046623,
+      "grad_norm": 238.2322235107422,
+      "learning_rate": 1.116279069767442e-06,
+      "loss": 13.556683349609376,
+      "step": 25
+    },
+    {
+      "epoch": 0.055944055944055944,
+      "grad_norm": 241.71214294433594,
+      "learning_rate": 1.3488372093023258e-06,
+      "loss": 13.285650634765625,
+      "step": 30
+    },
+    {
+      "epoch": 0.06526806526806526,
+      "grad_norm": 221.19960021972656,
+      "learning_rate": 1.5813953488372093e-06,
+      "loss": 13.05435791015625,
+      "step": 35
+    },
+    {
+      "epoch": 0.07459207459207459,
+      "grad_norm": 205.08326721191406,
+      "learning_rate": 1.8139534883720933e-06,
+      "loss": 12.770874786376954,
+      "step": 40
+    },
+    {
+      "epoch": 0.08391608391608392,
+      "grad_norm": 170.5838623046875,
+      "learning_rate": 2.0465116279069768e-06,
+      "loss": 12.280790710449219,
+      "step": 45
+    },
+    {
+      "epoch": 0.09324009324009325,
+      "grad_norm": 96.90670776367188,
+      "learning_rate": 2.2790697674418607e-06,
+      "loss": 11.730733489990234,
+      "step": 50
+    },
+    {
+      "epoch": 0.09324009324009325,
+      "eval_com_accuracy": 0.2,
+      "eval_rew_accuracy": 0.276,
+      "step": 50
+    },
+    {
+      "epoch": 0.10256410256410256,
+      "grad_norm": 61.14415740966797,
+      "learning_rate": 2.5116279069767446e-06,
+      "loss": 11.25308609008789,
+      "step": 55
+    },
+    {
+      "epoch": 0.11188811188811189,
+      "grad_norm": 56.027095794677734,
+      "learning_rate": 2.744186046511628e-06,
+      "loss": 11.197045135498048,
+      "step": 60
+    },
+    {
+      "epoch": 0.12121212121212122,
+      "grad_norm": 46.54409408569336,
+      "learning_rate": 2.9767441860465116e-06,
+      "loss": 10.874276733398437,
+      "step": 65
+    },
+    {
+      "epoch": 0.13053613053613053,
+      "grad_norm": 40.62468719482422,
+      "learning_rate": 3.2093023255813956e-06,
+      "loss": 10.69852294921875,
+      "step": 70
+    },
+    {
+      "epoch": 0.13986013986013987,
+      "grad_norm": 37.44975662231445,
+      "learning_rate": 3.4418604651162795e-06,
+      "loss": 10.560739135742187,
+      "step": 75
+    },
+    {
+      "epoch": 0.14918414918414918,
+      "grad_norm": 32.19042205810547,
+      "learning_rate": 3.674418604651163e-06,
+      "loss": 10.408053588867187,
+      "step": 80
+    },
+    {
+      "epoch": 0.1585081585081585,
+      "grad_norm": 27.208942413330078,
+      "learning_rate": 3.906976744186047e-06,
+      "loss": 10.108213806152344,
+      "step": 85
+    },
+    {
+      "epoch": 0.16783216783216784,
+      "grad_norm": 24.76529312133789,
+      "learning_rate": 4.1395348837209304e-06,
+      "loss": 9.929393005371093,
+      "step": 90
+    },
+    {
+      "epoch": 0.17715617715617715,
+      "grad_norm": 23.917495727539062,
+      "learning_rate": 4.372093023255815e-06,
+      "loss": 9.677659606933593,
+      "step": 95
+    },
+    {
+      "epoch": 0.1864801864801865,
+      "grad_norm": 22.98892593383789,
+      "learning_rate": 4.604651162790698e-06,
+      "loss": 9.481817626953125,
+      "step": 100
+    },
+    {
+      "epoch": 0.1864801864801865,
+      "eval_com_accuracy": 0.176,
+      "eval_rew_accuracy": 0.264,
+      "step": 100
+    },
+    {
+      "epoch": 0.1958041958041958,
+      "grad_norm": 25.177200317382812,
+      "learning_rate": 4.837209302325582e-06,
+      "loss": 9.42150650024414,
+      "step": 105
+    },
+    {
+      "epoch": 0.20512820512820512,
+      "grad_norm": 18.95603370666504,
+      "learning_rate": 5.069767441860466e-06,
+      "loss": 9.267645263671875,
+      "step": 110
+    },
+    {
+      "epoch": 0.21445221445221446,
+      "grad_norm": 22.155452728271484,
+      "learning_rate": 5.302325581395349e-06,
+      "loss": 9.06854248046875,
+      "step": 115
+    },
+    {
+      "epoch": 0.22377622377622378,
+      "grad_norm": 20.26909065246582,
+      "learning_rate": 5.534883720930233e-06,
+      "loss": 9.045303344726562,
+      "step": 120
+    },
+    {
+      "epoch": 0.2331002331002331,
+      "grad_norm": 26.49457359313965,
+      "learning_rate": 5.7674418604651175e-06,
+      "loss": 8.91327362060547,
+      "step": 125
+    },
+    {
+      "epoch": 0.24242424242424243,
+      "grad_norm": 19.886178970336914,
+      "learning_rate": 6e-06,
+      "loss": 8.950289154052735,
+      "step": 130
+    },
+    {
+      "epoch": 0.2517482517482518,
+      "grad_norm": 18.22088050842285,
+      "learning_rate": 6.2325581395348845e-06,
+      "loss": 8.710862731933593,
+      "step": 135
+    },
+    {
+      "epoch": 0.26107226107226106,
+      "grad_norm": 21.09984016418457,
+      "learning_rate": 6.465116279069767e-06,
+      "loss": 8.762718200683594,
+      "step": 140
+    },
+    {
+      "epoch": 0.2703962703962704,
+      "grad_norm": 17.96813201904297,
+      "learning_rate": 6.6976744186046515e-06,
+      "loss": 8.533232116699219,
+      "step": 145
+    },
+    {
+      "epoch": 0.27972027972027974,
+      "grad_norm": 20.52909278869629,
+      "learning_rate": 6.930232558139536e-06,
+      "loss": 8.279258728027344,
+      "step": 150
+    },
+    {
+      "epoch": 0.27972027972027974,
+      "eval_com_accuracy": 0.152,
+      "eval_rew_accuracy": 0.252,
+      "step": 150
+    },
+    {
+      "epoch": 0.289044289044289,
+      "grad_norm": 19.475040435791016,
+      "learning_rate": 7.1627906976744185e-06,
+      "loss": 8.369435882568359,
+      "step": 155
+    },
+    {
+      "epoch": 0.29836829836829837,
+      "grad_norm": 16.741973876953125,
+      "learning_rate": 7.395348837209303e-06,
+      "loss": 8.280664825439453,
+      "step": 160
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 19.79235076904297,
+      "learning_rate": 7.627906976744187e-06,
+      "loss": 8.200827026367188,
+      "step": 165
+    },
+    {
+      "epoch": 0.317016317016317,
+      "grad_norm": 32.991554260253906,
+      "learning_rate": 7.86046511627907e-06,
+      "loss": 7.973581695556641,
+      "step": 170
+    },
+    {
+      "epoch": 0.32634032634032634,
+      "grad_norm": 17.949445724487305,
+      "learning_rate": 8.093023255813955e-06,
+      "loss": 8.141377258300782,
+      "step": 175
+    },
+    {
+      "epoch": 0.3356643356643357,
+      "grad_norm": 16.19559097290039,
+      "learning_rate": 8.325581395348837e-06,
+      "loss": 7.837094116210937,
+      "step": 180
+    },
+    {
+      "epoch": 0.34498834498834496,
+      "grad_norm": 23.260622024536133,
+      "learning_rate": 8.558139534883722e-06,
+      "loss": 7.763467407226562,
+      "step": 185
+    },
+    {
+      "epoch": 0.3543123543123543,
+      "grad_norm": 24.858335494995117,
+      "learning_rate": 8.790697674418606e-06,
+      "loss": 7.886874389648438,
+      "step": 190
+    },
+    {
+      "epoch": 0.36363636363636365,
+      "grad_norm": 18.59360122680664,
+      "learning_rate": 9.023255813953489e-06,
+      "loss": 7.7624969482421875,
+      "step": 195
+    },
+    {
+      "epoch": 0.372960372960373,
+      "grad_norm": 18.056495666503906,
+      "learning_rate": 9.255813953488373e-06,
+      "loss": 7.4621437072753904,
+      "step": 200
+    },
+    {
+      "epoch": 0.372960372960373,
+      "eval_com_accuracy": 0.144,
+      "eval_rew_accuracy": 0.332,
+      "step": 200
+    },
+    {
+      "epoch": 0.3822843822843823,
+      "grad_norm": 16.496261596679688,
+      "learning_rate": 9.488372093023258e-06,
+      "loss": 7.526261901855468,
+      "step": 205
+    },
+    {
+      "epoch": 0.3916083916083916,
+      "grad_norm": 16.77888298034668,
+      "learning_rate": 9.72093023255814e-06,
+      "loss": 7.624967193603515,
+      "step": 210
+    },
+    {
+      "epoch": 0.40093240093240096,
+      "grad_norm": 17.096778869628906,
+      "learning_rate": 9.953488372093025e-06,
+      "loss": 7.337849426269531,
+      "step": 215
+    },
+    {
+      "epoch": 0.41025641025641024,
+      "grad_norm": 18.3544979095459,
+      "learning_rate": 9.999894343918674e-06,
+      "loss": 7.354197692871094,
+      "step": 220
+    },
+    {
+      "epoch": 0.4195804195804196,
+      "grad_norm": 15.287137985229492,
+      "learning_rate": 9.999465123741172e-06,
+      "loss": 7.085916137695312,
+      "step": 225
+    },
+    {
+      "epoch": 0.4289044289044289,
+      "grad_norm": 17.03432273864746,
+      "learning_rate": 9.998705764284132e-06,
+      "loss": 7.059596252441406,
+      "step": 230
+    },
+    {
+      "epoch": 0.4382284382284382,
+      "grad_norm": 15.470928192138672,
+      "learning_rate": 9.99761631569179e-06,
+      "loss": 7.118296813964844,
+      "step": 235
+    },
+    {
+      "epoch": 0.44755244755244755,
+      "grad_norm": 15.732117652893066,
+      "learning_rate": 9.9961968499058e-06,
+      "loss": 6.981436157226563,
+      "step": 240
+    },
+    {
+      "epoch": 0.4568764568764569,
+      "grad_norm": 20.92630958557129,
+      "learning_rate": 9.994447460660473e-06,
+      "loss": 6.96953125,
+      "step": 245
+    },
+    {
+      "epoch": 0.4662004662004662,
+      "grad_norm": 14.848095893859863,
+      "learning_rate": 9.992368263476585e-06,
+      "loss": 6.831211090087891,
+      "step": 250
+    },
+    {
+      "epoch": 0.4662004662004662,
+      "eval_com_accuracy": 0.132,
+      "eval_rew_accuracy": 0.312,
+      "step": 250
+    },
+    {
+      "epoch": 0.4755244755244755,
+      "grad_norm": 18.59755516052246,
+      "learning_rate": 9.989959395653756e-06,
+      "loss": 6.873024749755859,
+      "step": 255
+    },
+    {
+      "epoch": 0.48484848484848486,
+      "grad_norm": 15.705364227294922,
+      "learning_rate": 9.98722101626138e-06,
+      "loss": 6.633563232421875,
+      "step": 260
+    },
+    {
+      "epoch": 0.49417249417249415,
+      "grad_norm": 39.2294807434082,
+      "learning_rate": 9.984153306128124e-06,
+      "loss": 6.63948974609375,
+      "step": 265
+    },
+    {
+      "epoch": 0.5034965034965035,
+      "grad_norm": 13.209358215332031,
+      "learning_rate": 9.980756467829977e-06,
+      "loss": 6.688297271728516,
+      "step": 270
+    },
+    {
+      "epoch": 0.5128205128205128,
+      "grad_norm": 14.645393371582031,
+      "learning_rate": 9.977030725676887e-06,
+      "loss": 6.4885505676269535,
+      "step": 275
+    },
+    {
+      "epoch": 0.5221445221445221,
+      "grad_norm": 17.008819580078125,
+      "learning_rate": 9.972976325697938e-06,
+      "loss": 6.78935546875,
+      "step": 280
+    },
+    {
+      "epoch": 0.5314685314685315,
+      "grad_norm": 16.029518127441406,
+      "learning_rate": 9.96859353562511e-06,
+      "loss": 6.681609344482422,
+      "step": 285
+    },
+    {
+      "epoch": 0.5407925407925408,
+      "grad_norm": 13.878348350524902,
+      "learning_rate": 9.963882644875594e-06,
+      "loss": 6.528868103027344,
+      "step": 290
+    },
+    {
+      "epoch": 0.5501165501165501,
+      "grad_norm": 13.996464729309082,
+      "learning_rate": 9.958843964532683e-06,
+      "loss": 6.431344604492187,
+      "step": 295
+    },
+    {
+      "epoch": 0.5594405594405595,
+      "grad_norm": 15.72003173828125,
+      "learning_rate": 9.953477827325229e-06,
+      "loss": 6.5927978515625,
+      "step": 300
+    },
+    {
+      "epoch": 0.5594405594405595,
+      "eval_com_accuracy": 0.148,
+      "eval_rew_accuracy": 0.368,
+      "step": 300
+    },
+    {
+      "epoch": 0.5687645687645687,
+      "grad_norm": 16.416725158691406,
+      "learning_rate": 9.947784587605678e-06,
+      "loss": 6.411885833740234,
+      "step": 305
+    },
+    {
+      "epoch": 0.578088578088578,
+      "grad_norm": 14.880790710449219,
+      "learning_rate": 9.941764621326655e-06,
+      "loss": 6.5298927307128904,
+      "step": 310
+    },
+    {
+      "epoch": 0.5874125874125874,
+      "grad_norm": 17.828296661376953,
+      "learning_rate": 9.935418326016153e-06,
+      "loss": 6.045618057250977,
+      "step": 315
+    },
+    {
+      "epoch": 0.5967365967365967,
+      "grad_norm": 13.30245590209961,
+      "learning_rate": 9.928746120751275e-06,
+      "loss": 6.1650840759277346,
+      "step": 320
+    },
+    {
+      "epoch": 0.6060606060606061,
+      "grad_norm": 13.527432441711426,
+      "learning_rate": 9.921748446130564e-06,
+      "loss": 6.349945831298828,
+      "step": 325
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 14.491817474365234,
+      "learning_rate": 9.9144257642449e-06,
+      "loss": 6.232943725585938,
+      "step": 330
+    },
+    {
+      "epoch": 0.6247086247086248,
+      "grad_norm": 15.7533597946167,
+      "learning_rate": 9.906778558647e-06,
+      "loss": 6.564260864257813,
+      "step": 335
+    },
+    {
+      "epoch": 0.634032634032634,
+      "grad_norm": 16.538602828979492,
+      "learning_rate": 9.898807334319471e-06,
+      "loss": 6.122506332397461,
+      "step": 340
+    },
+    {
+      "epoch": 0.6433566433566433,
+      "grad_norm": 16.459688186645508,
+      "learning_rate": 9.890512617641474e-06,
+      "loss": 6.332431030273438,
+      "step": 345
+    },
+    {
+      "epoch": 0.6526806526806527,
+      "grad_norm": 14.879058837890625,
+      "learning_rate": 9.881894956353963e-06,
+      "loss": 6.118003082275391,
+      "step": 350
+    },
+    {
+      "epoch": 0.6526806526806527,
+      "eval_com_accuracy": 0.164,
+      "eval_rew_accuracy": 0.364,
+      "step": 350
+    },
+    {
+      "epoch": 0.662004662004662,
+      "grad_norm": 16.02516746520996,
+      "learning_rate": 9.87295491952351e-06,
+      "loss": 6.241584014892578,
+      "step": 355
+    },
+    {
+      "epoch": 0.6713286713286714,
+      "grad_norm": 14.9837646484375,
+      "learning_rate": 9.863693097504733e-06,
+      "loss": 6.083467864990235,
+      "step": 360
+    },
+    {
+      "epoch": 0.6806526806526807,
+      "grad_norm": 14.138517379760742,
+      "learning_rate": 9.854110101901308e-06,
+      "loss": 6.143560791015625,
+      "step": 365
+    },
+    {
+      "epoch": 0.6899766899766899,
+      "grad_norm": 12.696243286132812,
+      "learning_rate": 9.844206565525585e-06,
+      "loss": 6.061418914794922,
+      "step": 370
+    },
+    {
+      "epoch": 0.6993006993006993,
+      "grad_norm": 15.22846508026123,
+      "learning_rate": 9.833983142356792e-06,
+      "loss": 6.193332672119141,
+      "step": 375
+    },
+    {
+      "epoch": 0.7086247086247086,
+      "grad_norm": 15.284857749938965,
+      "learning_rate": 9.823440507497863e-06,
+      "loss": 6.0868377685546875,
+      "step": 380
+    },
+    {
+      "epoch": 0.717948717948718,
+      "grad_norm": 14.117541313171387,
+      "learning_rate": 9.812579357130848e-06,
+      "loss": 6.289948272705078,
+      "step": 385
+    },
+    {
+      "epoch": 0.7272727272727273,
+      "grad_norm": 12.648367881774902,
+      "learning_rate": 9.801400408470943e-06,
+      "loss": 6.194795227050781,
+      "step": 390
+    },
+    {
+      "epoch": 0.7365967365967366,
+      "grad_norm": 12.394021034240723,
+      "learning_rate": 9.789904399719124e-06,
+      "loss": 6.031880950927734,
+      "step": 395
+    },
+    {
+      "epoch": 0.745920745920746,
+      "grad_norm": 15.566425323486328,
+      "learning_rate": 9.778092090013416e-06,
+      "loss": 6.0966644287109375,
+      "step": 400
+    },
+    {
+      "epoch": 0.745920745920746,
+      "eval_com_accuracy": 0.184,
+      "eval_rew_accuracy": 0.364,
+      "step": 400
+    },
+    {
+      "epoch": 0.7552447552447552,
+      "grad_norm": 14.885455131530762,
+      "learning_rate": 9.76596425937874e-06,
+      "loss": 6.476510620117187,
+      "step": 405
+    },
+    {
+      "epoch": 0.7645687645687645,
+      "grad_norm": 16.627117156982422,
+      "learning_rate": 9.753521708675426e-06,
+      "loss": 5.9605262756347654,
+      "step": 410
+    },
+    {
+      "epoch": 0.7738927738927739,
+      "grad_norm": 14.517923355102539,
+      "learning_rate": 9.740765259546312e-06,
+      "loss": 6.1104694366455075,
+      "step": 415
+    },
+    {
+      "epoch": 0.7832167832167832,
+      "grad_norm": 12.007637977600098,
+      "learning_rate": 9.727695754362498e-06,
+      "loss": 5.964785766601563,
+      "step": 420
+    },
+    {
+      "epoch": 0.7925407925407926,
+      "grad_norm": 16.17314338684082,
+      "learning_rate": 9.714314056167711e-06,
+      "loss": 5.976337432861328,
+      "step": 425
+    },
+    {
+      "epoch": 0.8018648018648019,
+      "grad_norm": 13.246196746826172,
+      "learning_rate": 9.700621048621322e-06,
+      "loss": 5.936331939697266,
+      "step": 430
+    },
+    {
+      "epoch": 0.8111888111888111,
+      "grad_norm": 15.030324935913086,
+      "learning_rate": 9.686617635939988e-06,
+      "loss": 6.209032821655273,
+      "step": 435
+    },
+    {
+      "epoch": 0.8205128205128205,
+      "grad_norm": 13.853086471557617,
+      "learning_rate": 9.672304742837945e-06,
+      "loss": 5.985638427734375,
+      "step": 440
+    },
+    {
+      "epoch": 0.8298368298368298,
+      "grad_norm": 13.534188270568848,
+      "learning_rate": 9.657683314465948e-06,
+      "loss": 5.7971649169921875,
+      "step": 445
+    },
+    {
+      "epoch": 0.8391608391608392,
+      "grad_norm": 26.1879825592041,
+      "learning_rate": 9.642754316348846e-06,
+      "loss": 5.949985504150391,
+      "step": 450
+    },
+    {
+      "epoch": 0.8391608391608392,
+      "eval_com_accuracy": 0.176,
+      "eval_rew_accuracy": 0.34,
+      "step": 450
+    },
+    {
+      "epoch": 0.8484848484848485,
+      "grad_norm": 16.848834991455078,
+      "learning_rate": 9.627518734321837e-06,
+      "loss": 6.1193115234375,
+      "step": 455
+    },
+    {
+      "epoch": 0.8578088578088578,
+      "grad_norm": 30.765796661376953,
+      "learning_rate": 9.61197757446536e-06,
+      "loss": 5.74780158996582,
+      "step": 460
+    },
+    {
+      "epoch": 0.8671328671328671,
+      "grad_norm": 13.27775764465332,
+      "learning_rate": 9.596131863038664e-06,
+      "loss": 5.994970703125,
+      "step": 465
+    },
+    {
+      "epoch": 0.8764568764568764,
+      "grad_norm": 14.923792839050293,
+      "learning_rate": 9.579982646412039e-06,
+      "loss": 5.810473251342773,
+      "step": 470
+    },
+    {
+      "epoch": 0.8857808857808858,
+      "grad_norm": 14.351722717285156,
+      "learning_rate": 9.563530990997707e-06,
+      "loss": 6.09692268371582,
+      "step": 475
+    },
+    {
+      "epoch": 0.8951048951048951,
+      "grad_norm": 11.83583927154541,
+      "learning_rate": 9.546777983179421e-06,
+      "loss": 5.867227935791016,
+      "step": 480
+    },
+    {
+      "epoch": 0.9044289044289044,
+      "grad_norm": 19.415884017944336,
+      "learning_rate": 9.529724729240712e-06,
+      "loss": 5.834615707397461,
+      "step": 485
+    },
+    {
+      "epoch": 0.9137529137529138,
+      "grad_norm": 12.147261619567871,
+      "learning_rate": 9.512372355291838e-06,
+      "loss": 6.039286041259766,
+      "step": 490
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 12.706372261047363,
+      "learning_rate": 9.494722007195427e-06,
+      "loss": 6.06304702758789,
+      "step": 495
+    },
+    {
+      "epoch": 0.9324009324009324,
+      "grad_norm": 12.642041206359863,
+      "learning_rate": 9.476774850490803e-06,
+      "loss": 5.9729866027832035,
+      "step": 500
+    },
+    {
+      "epoch": 0.9324009324009324,
+      "eval_com_accuracy": 0.204,
+      "eval_rew_accuracy": 0.328,
+      "step": 500
+    },
+    {
+      "epoch": 0.9417249417249417,
+      "grad_norm": 12.66169548034668,
+      "learning_rate": 9.458532070317021e-06,
+      "loss": 5.8281097412109375,
+      "step": 505
+    },
+    {
+      "epoch": 0.951048951048951,
+      "grad_norm": 13.191854476928711,
+      "learning_rate": 9.439994871334614e-06,
+      "loss": 6.183546447753907,
+      "step": 510
+    },
+    {
+      "epoch": 0.9603729603729604,
+      "grad_norm": 12.52676010131836,
+      "learning_rate": 9.421164477646031e-06,
+      "loss": 5.785601043701172,
+      "step": 515
+    },
+    {
+      "epoch": 0.9696969696969697,
+      "grad_norm": 13.225418090820312,
+      "learning_rate": 9.402042132714817e-06,
+      "loss": 5.784825134277344,
+      "step": 520
+    },
+    {
+      "epoch": 0.9790209790209791,
+      "grad_norm": 12.035611152648926,
+      "learning_rate": 9.382629099283486e-06,
+      "loss": 5.9122272491455075,
+      "step": 525
+    },
+    {
+      "epoch": 0.9883449883449883,
+      "grad_norm": 12.166351318359375,
+      "learning_rate": 9.362926659290149e-06,
+      "loss": 5.790053939819336,
+      "step": 530
+    },
+    {
+      "epoch": 0.9976689976689976,
+      "grad_norm": 13.22154426574707,
+      "learning_rate": 9.342936113783855e-06,
+      "loss": 5.745006561279297,
+      "step": 535
+    },
+    {
+      "epoch": 1.0055944055944055,
+      "grad_norm": 13.12752628326416,
+      "learning_rate": 9.32265878283868e-06,
+      "loss": 5.1531005859375,
+      "step": 540
+    },
+    {
+      "epoch": 1.014918414918415,
+      "grad_norm": 15.964373588562012,
+      "learning_rate": 9.302096005466547e-06,
+      "loss": 5.7917236328125,
+      "step": 545
+    },
+    {
+      "epoch": 1.0242424242424242,
+      "grad_norm": 18.454797744750977,
+      "learning_rate": 9.281249139528816e-06,
+      "loss": 5.967070007324219,
+      "step": 550
+    },
+    {
+      "epoch": 1.0242424242424242,
+      "eval_com_accuracy": 0.212,
+      "eval_rew_accuracy": 0.352,
+      "step": 550
+    },
+    {
+      "epoch": 1.0335664335664336,
+      "grad_norm": 15.802419662475586,
+      "learning_rate": 9.260119561646614e-06,
+      "loss": 6.093529129028321,
+      "step": 555
+    },
+    {
+      "epoch": 1.0428904428904429,
+      "grad_norm": 21.95537757873535,
+      "learning_rate": 9.238708667109924e-06,
+      "loss": 5.489861297607422,
+      "step": 560
+    },
+    {
+      "epoch": 1.0522144522144523,
+      "grad_norm": 13.373199462890625,
+      "learning_rate": 9.217017869785453e-06,
+      "loss": 5.871435928344726,
+      "step": 565
+    },
+    {
+      "epoch": 1.0615384615384615,
+      "grad_norm": 13.36340618133545,
+      "learning_rate": 9.19504860202327e-06,
+      "loss": 5.805611038208008,
+      "step": 570
+    },
+    {
+      "epoch": 1.0708624708624708,
+      "grad_norm": 14.099553108215332,
+      "learning_rate": 9.172802314562214e-06,
+      "loss": 5.600634765625,
+      "step": 575
+    },
+    {
+      "epoch": 1.0801864801864802,
+      "grad_norm": 15.012040138244629,
+      "learning_rate": 9.150280476434098e-06,
+      "loss": 5.793732452392578,
+      "step": 580
+    },
+    {
+      "epoch": 1.0895104895104895,
+      "grad_norm": 15.98738956451416,
+      "learning_rate": 9.127484574866699e-06,
+      "loss": 5.8642620086669925,
+      "step": 585
+    },
+    {
+      "epoch": 1.098834498834499,
+      "grad_norm": 13.785572052001953,
+      "learning_rate": 9.104416115185557e-06,
+      "loss": 5.770995330810547,
+      "step": 590
+    },
+    {
+      "epoch": 1.1081585081585081,
+      "grad_norm": 12.145087242126465,
+      "learning_rate": 9.08107662071456e-06,
+      "loss": 5.850731277465821,
+      "step": 595
+    },
+    {
+      "epoch": 1.1174825174825176,
+      "grad_norm": 13.681862831115723,
+      "learning_rate": 9.057467632675357e-06,
+      "loss": 5.782034301757813,
+      "step": 600
+    },
+    {
+      "epoch": 1.1174825174825176,
+      "eval_com_accuracy": 0.22,
+      "eval_rew_accuracy": 0.348,
+      "step": 600
+    },
+    {
+      "epoch": 1.1268065268065268,
+      "grad_norm": 13.870027542114258,
+      "learning_rate": 9.033590710085584e-06,
+      "loss": 5.723427963256836,
+      "step": 605
+    },
+    {
+      "epoch": 1.136130536130536,
+      "grad_norm": 16.20020294189453,
+      "learning_rate": 9.00944742965592e-06,
+      "loss": 5.953633117675781,
+      "step": 610
+    },
+    {
+      "epoch": 1.1454545454545455,
+      "grad_norm": 15.48993968963623,
+      "learning_rate": 8.985039385685952e-06,
+      "loss": 5.809123992919922,
+      "step": 615
+    },
+    {
+      "epoch": 1.1547785547785547,
+      "grad_norm": 13.686211585998535,
+      "learning_rate": 8.960368189958913e-06,
+      "loss": 5.813122177124024,
+      "step": 620
+    },
+    {
+      "epoch": 1.1641025641025642,
+      "grad_norm": 13.733199119567871,
+      "learning_rate": 8.935435471635238e-06,
+      "loss": 5.6748512268066404,
+      "step": 625
+    },
+    {
+      "epoch": 1.1734265734265734,
+      "grad_norm": 12.450380325317383,
+      "learning_rate": 8.91024287714499e-06,
+      "loss": 5.6834667205810545,
+      "step": 630
+    },
+    {
+      "epoch": 1.1827505827505829,
+      "grad_norm": 14.111934661865234,
+      "learning_rate": 8.884792070079128e-06,
+      "loss": 5.680038833618164,
+      "step": 635
+    },
+    {
+      "epoch": 1.192074592074592,
+      "grad_norm": 13.175325393676758,
+      "learning_rate": 8.859084731079664e-06,
+      "loss": 5.445759201049805,
+      "step": 640
+    },
+    {
+      "epoch": 1.2013986013986013,
+      "grad_norm": 14.951041221618652,
+      "learning_rate": 8.833122557728667e-06,
+      "loss": 5.673246765136719,
+      "step": 645
+    },
+    {
+      "epoch": 1.2107226107226108,
+      "grad_norm": 12.364238739013672,
+      "learning_rate": 8.806907264436183e-06,
+      "loss": 5.234909439086914,
+      "step": 650
+    },
+    {
+      "epoch": 1.2107226107226108,
+      "eval_com_accuracy": 0.224,
+      "eval_rew_accuracy": 0.348,
+      "step": 650
+    },
+    {
+      "epoch": 1.22004662004662,
+      "grad_norm": 13.065054893493652,
+      "learning_rate": 8.780440582327005e-06,
+      "loss": 5.442026519775391,
+      "step": 655
+    },
+    {
+      "epoch": 1.2293706293706295,
+      "grad_norm": 13.872124671936035,
+      "learning_rate": 8.75372425912637e-06,
+      "loss": 5.736603546142578,
+      "step": 660
+    },
+    {
+      "epoch": 1.2386946386946387,
+      "grad_norm": 13.677570343017578,
+      "learning_rate": 8.726760059044542e-06,
+      "loss": 5.951699447631836,
+      "step": 665
+    },
+    {
+      "epoch": 1.248018648018648,
+      "grad_norm": 15.023405075073242,
+      "learning_rate": 8.699549762660318e-06,
+      "loss": 5.936520004272461,
+      "step": 670
+    },
+    {
+      "epoch": 1.2573426573426574,
+      "grad_norm": 12.804817199707031,
+      "learning_rate": 8.672095166803445e-06,
+      "loss": 5.778074645996094,
+      "step": 675
+    },
+    {
+      "epoch": 1.2666666666666666,
+      "grad_norm": 13.148968696594238,
+      "learning_rate": 8.644398084435959e-06,
+      "loss": 5.7417655944824215,
+      "step": 680
+    },
+    {
+      "epoch": 1.275990675990676,
+      "grad_norm": 14.131410598754883,
+      "learning_rate": 8.616460344532483e-06,
+      "loss": 5.58878173828125,
+      "step": 685
+    },
+    {
+      "epoch": 1.2853146853146853,
+      "grad_norm": 14.100456237792969,
+      "learning_rate": 8.588283791959437e-06,
+      "loss": 5.546084976196289,
+      "step": 690
+    },
+    {
+      "epoch": 1.2946386946386945,
+      "grad_norm": 13.94593620300293,
+      "learning_rate": 8.559870287353214e-06,
+      "loss": 6.0555156707763675,
+      "step": 695
+    },
+    {
+      "epoch": 1.303962703962704,
+      "grad_norm": 18.369823455810547,
+      "learning_rate": 8.531221706997316e-06,
+      "loss": 5.930012512207031,
+      "step": 700
+    },
+    {
+      "epoch": 1.303962703962704,
+      "eval_com_accuracy": 0.24,
+      "eval_rew_accuracy": 0.364,
+      "step": 700
+    },
+    {
+      "epoch": 1.3132867132867134,
+      "grad_norm": 14.022618293762207,
+      "learning_rate": 8.502339942698463e-06,
+      "loss": 5.831372833251953,
+      "step": 705
+    },
+    {
+      "epoch": 1.3226107226107227,
+      "grad_norm": 12.844898223876953,
+      "learning_rate": 8.473226901661643e-06,
+      "loss": 5.503532409667969,
+      "step": 710
+    },
+    {
+      "epoch": 1.3319347319347319,
+      "grad_norm": 13.991181373596191,
+      "learning_rate": 8.443884506364192e-06,
+      "loss": 5.344602966308594,
+      "step": 715
+    },
+    {
+      "epoch": 1.3412587412587413,
+      "grad_norm": 13.290801048278809,
+      "learning_rate": 8.414314694428842e-06,
+      "loss": 5.7142791748046875,
+      "step": 720
+    },
+    {
+      "epoch": 1.3505827505827506,
+      "grad_norm": 13.986117362976074,
+      "learning_rate": 8.384519418495755e-06,
+      "loss": 5.509880065917969,
+      "step": 725
+    },
+    {
+      "epoch": 1.3599067599067598,
+      "grad_norm": 17.483638763427734,
+      "learning_rate": 8.354500646093592e-06,
+      "loss": 5.769047546386719,
+      "step": 730
+    },
+    {
+      "epoch": 1.3692307692307693,
+      "grad_norm": 12.383395195007324,
+      "learning_rate": 8.324260359509594e-06,
+      "loss": 5.693521881103516,
+      "step": 735
+    },
+    {
+      "epoch": 1.3785547785547785,
+      "grad_norm": 12.225541114807129,
+      "learning_rate": 8.29380055565866e-06,
+      "loss": 5.618255233764648,
+      "step": 740
+    },
+    {
+      "epoch": 1.387878787878788,
+      "grad_norm": 11.737737655639648,
+      "learning_rate": 8.263123245951504e-06,
+      "loss": 5.702220916748047,
+      "step": 745
+    },
+    {
+      "epoch": 1.3972027972027972,
+      "grad_norm": 14.235527992248535,
+      "learning_rate": 8.232230456161819e-06,
+      "loss": 5.744013977050781,
+      "step": 750
+    },
+    {
+      "epoch": 1.3972027972027972,
+      "eval_com_accuracy": 0.24,
+      "eval_rew_accuracy": 0.368,
+      "step": 750
+    },
+    {
+      "epoch": 1.4065268065268066,
+      "grad_norm": 12.433022499084473,
+      "learning_rate": 8.201124226292505e-06,
+      "loss": 5.6122283935546875,
+      "step": 755
+    },
+    {
+      "epoch": 1.4158508158508158,
+      "grad_norm": 13.874029159545898,
+      "learning_rate": 8.169806610440966e-06,
+      "loss": 5.530184936523438,
+      "step": 760
+    },
+    {
+      "epoch": 1.425174825174825,
+      "grad_norm": 12.640862464904785,
+      "learning_rate": 8.138279676663458e-06,
+      "loss": 5.602608489990234,
+      "step": 765
+    },
+    {
+      "epoch": 1.4344988344988345,
+      "grad_norm": 13.672120094299316,
+      "learning_rate": 8.106545506838533e-06,
+      "loss": 5.971489715576172,
+      "step": 770
+    },
+    {
+      "epoch": 1.4438228438228438,
+      "grad_norm": 12.973994255065918,
+      "learning_rate": 8.074606196529554e-06,
+      "loss": 5.967403411865234,
+      "step": 775
+    },
+    {
+      "epoch": 1.4531468531468532,
+      "grad_norm": 12.31808090209961,
+      "learning_rate": 8.042463854846325e-06,
+      "loss": 5.71468505859375,
+      "step": 780
+    },
+    {
+      "epoch": 1.4624708624708624,
+      "grad_norm": 12.416245460510254,
+      "learning_rate": 8.010120604305806e-06,
+      "loss": 5.836771392822266,
+      "step": 785
+    },
+    {
+      "epoch": 1.471794871794872,
+      "grad_norm": 13.020674705505371,
+      "learning_rate": 7.977578580691963e-06,
+      "loss": 5.71131591796875,
+      "step": 790
+    },
+    {
+      "epoch": 1.4811188811188811,
+      "grad_norm": 13.946771621704102,
+      "learning_rate": 7.944839932914718e-06,
+      "loss": 5.564648818969727,
+      "step": 795
+    },
+    {
+      "epoch": 1.4904428904428904,
+      "grad_norm": 12.287398338317871,
+      "learning_rate": 7.91190682286806e-06,
+      "loss": 5.753592300415039,
+      "step": 800
+    },
+    {
+      "epoch": 1.4904428904428904,
+      "eval_com_accuracy": 0.236,
+      "eval_rew_accuracy": 0.384,
+      "step": 800
+    },
+    {
+      "epoch": 1.4997668997668998,
+      "grad_norm": 20.467731475830078,
+      "learning_rate": 7.878781425287277e-06,
+      "loss": 5.64927978515625,
+      "step": 805
+    },
+    {
+      "epoch": 1.509090909090909,
+      "grad_norm": 13.811100006103516,
+      "learning_rate": 7.84546592760535e-06,
+      "loss": 5.415856170654297,
+      "step": 810
+    },
+    {
+      "epoch": 1.5184149184149183,
+      "grad_norm": 12.4107666015625,
+      "learning_rate": 7.811962529808499e-06,
+      "loss": 5.767454147338867,
+      "step": 815
+    },
+    {
+      "epoch": 1.5277389277389277,
+      "grad_norm": 14.148603439331055,
+      "learning_rate": 7.778273444290921e-06,
+      "loss": 5.463053131103516,
+      "step": 820
+    },
+    {
+      "epoch": 1.5370629370629372,
+      "grad_norm": 13.375722885131836,
+      "learning_rate": 7.744400895708683e-06,
+      "loss": 5.625373077392578,
+      "step": 825
+    },
+    {
+      "epoch": 1.5463869463869464,
+      "grad_norm": 12.991719245910645,
+      "learning_rate": 7.710347120832821e-06,
+      "loss": 5.6271202087402346,
+      "step": 830
+    },
+    {
+      "epoch": 1.5557109557109556,
+      "grad_norm": 12.906146049499512,
+      "learning_rate": 7.676114368401635e-06,
+      "loss": 5.5160057067871096,
+      "step": 835
+    },
+    {
+      "epoch": 1.565034965034965,
+      "grad_norm": 12.344596862792969,
+      "learning_rate": 7.641704898972194e-06,
+      "loss": 5.506198120117188,
+      "step": 840
+    },
+    {
+      "epoch": 1.5743589743589743,
+      "grad_norm": 13.285745620727539,
+      "learning_rate": 7.607120984771058e-06,
+      "loss": 5.7151634216308596,
+      "step": 845
+    },
+    {
+      "epoch": 1.5836829836829835,
+      "grad_norm": 12.432035446166992,
+      "learning_rate": 7.572364909544235e-06,
+      "loss": 5.852275085449219,
+      "step": 850
+    },
+    {
+      "epoch": 1.5836829836829835,
+      "eval_com_accuracy": 0.228,
+      "eval_rew_accuracy": 0.332,
+      "step": 850
+    },
+    {
+      "epoch": 1.593006993006993,
+      "grad_norm": 12.581634521484375,
+      "learning_rate": 7.537438968406372e-06,
+      "loss": 5.735688018798828,
+      "step": 855
+    },
+    {
+      "epoch": 1.6023310023310025,
+      "grad_norm": 13.161237716674805,
+      "learning_rate": 7.502345467689202e-06,
+      "loss": 5.648350524902344,
+      "step": 860
+    },
+    {
+      "epoch": 1.6116550116550117,
+      "grad_norm": 12.48270034790039,
+      "learning_rate": 7.4670867247892346e-06,
+      "loss": 5.635168075561523,
+      "step": 865
+    },
+    {
+      "epoch": 1.620979020979021,
+      "grad_norm": 13.835477828979492,
+      "learning_rate": 7.431665068014737e-06,
+      "loss": 5.548571014404297,
+      "step": 870
+    },
+    {
+      "epoch": 1.6303030303030304,
+      "grad_norm": 12.403617858886719,
+      "learning_rate": 7.396082836431981e-06,
+      "loss": 5.7486827850341795,
+      "step": 875
+    },
+    {
+      "epoch": 1.6396270396270396,
+      "grad_norm": 13.730851173400879,
+      "learning_rate": 7.3603423797107845e-06,
+      "loss": 5.288154602050781,
+      "step": 880
+    },
+    {
+      "epoch": 1.6489510489510488,
+      "grad_norm": 14.688888549804688,
+      "learning_rate": 7.324446057969346e-06,
+      "loss": 5.648311614990234,
+      "step": 885
+    },
+    {
+      "epoch": 1.6582750582750583,
+      "grad_norm": 13.311734199523926,
+      "learning_rate": 7.288396241618401e-06,
+      "loss": 5.636775207519531,
+      "step": 890
+    },
+    {
+      "epoch": 1.6675990675990677,
+      "grad_norm": 12.776606559753418,
+      "learning_rate": 7.252195311204689e-06,
+      "loss": 5.666929626464844,
+      "step": 895
+    },
+    {
+      "epoch": 1.676923076923077,
+      "grad_norm": 14.130080223083496,
+      "learning_rate": 7.215845657253755e-06,
+      "loss": 5.534748840332031,
+      "step": 900
+    },
+    {
+      "epoch": 1.676923076923077,
+      "eval_com_accuracy": 0.264,
+      "eval_rew_accuracy": 0.364,
+      "step": 900
+    },
+    {
+      "epoch": 1.6862470862470862,
+      "grad_norm": 14.159326553344727,
+      "learning_rate": 7.1793496801120885e-06,
+      "loss": 5.652375793457031,
+      "step": 905
+    },
+    {
+      "epoch": 1.6955710955710956,
+      "grad_norm": 13.113655090332031,
+      "learning_rate": 7.1427097897886225e-06,
+      "loss": 5.383832168579102,
+      "step": 910
+    },
+    {
+      "epoch": 1.7048951048951049,
+      "grad_norm": 14.203001022338867,
+      "learning_rate": 7.105928405795584e-06,
+      "loss": 5.6499076843261715,
+      "step": 915
+    },
+    {
+      "epoch": 1.714219114219114,
+      "grad_norm": 12.1084623336792,
+      "learning_rate": 7.069007956988718e-06,
+      "loss": 5.560025787353515,
+      "step": 920
+    },
+    {
+      "epoch": 1.7235431235431236,
+      "grad_norm": 13.819426536560059,
+      "learning_rate": 7.031950881406913e-06,
+      "loss": 5.535141372680664,
+      "step": 925
+    },
+    {
+      "epoch": 1.732867132867133,
+      "grad_norm": 17.341110229492188,
+      "learning_rate": 6.994759626111189e-06,
+      "loss": 5.697111511230469,
+      "step": 930
+    },
+    {
+      "epoch": 1.7421911421911422,
+      "grad_norm": 16.404672622680664,
+      "learning_rate": 6.957436647023117e-06,
+      "loss": 5.4155632019042965,
+      "step": 935
+    },
+    {
+      "epoch": 1.7515151515151515,
+      "grad_norm": 13.115015029907227,
+      "learning_rate": 6.919984408762632e-06,
+      "loss": 5.758349990844726,
+      "step": 940
+    },
+    {
+      "epoch": 1.760839160839161,
+      "grad_norm": 13.850481986999512,
+      "learning_rate": 6.882405384485294e-06,
+      "loss": 5.69927864074707,
+      "step": 945
+    },
+    {
+      "epoch": 1.7701631701631702,
+      "grad_norm": 12.921819686889648,
+      "learning_rate": 6.844702055718964e-06,
+      "loss": 5.631483459472657,
+      "step": 950
+    },
+    {
+      "epoch": 1.7701631701631702,
+      "eval_com_accuracy": 0.26,
+      "eval_rew_accuracy": 0.352,
+      "step": 950
+    },
+    {
+      "epoch": 1.7794871794871794,
+      "grad_norm": 13.401888847351074,
+      "learning_rate": 6.806876912199945e-06,
+      "loss": 5.824296569824218,
+      "step": 955
+    },
+    {
+      "epoch": 1.7888111888111888,
+      "grad_norm": 13.481993675231934,
+      "learning_rate": 6.768932451708557e-06,
+      "loss": 5.752721786499023,
+      "step": 960
+    },
+    {
+      "epoch": 1.7981351981351983,
+      "grad_norm": 14.022424697875977,
+      "learning_rate": 6.730871179904218e-06,
+      "loss": 5.453369522094727,
+      "step": 965
+    },
+    {
+      "epoch": 1.8074592074592073,
+      "grad_norm": 13.432122230529785,
+      "learning_rate": 6.692695610159966e-06,
+      "loss": 5.664297103881836,
+      "step": 970
+    },
+    {
+      "epoch": 1.8167832167832167,
+      "grad_norm": 14.292946815490723,
+      "learning_rate": 6.6544082633964955e-06,
+      "loss": 5.576699066162109,
+      "step": 975
+    },
+    {
+      "epoch": 1.8261072261072262,
+      "grad_norm": 14.256867408752441,
+      "learning_rate": 6.6160116679156874e-06,
+      "loss": 5.551004791259766,
+      "step": 980
+    },
+    {
+      "epoch": 1.8354312354312354,
+      "grad_norm": 14.11919116973877,
+      "learning_rate": 6.577508359233653e-06,
+      "loss": 5.580959320068359,
+      "step": 985
+    },
+    {
+      "epoch": 1.8447552447552447,
+      "grad_norm": 14.006019592285156,
+      "learning_rate": 6.538900879913301e-06,
+      "loss": 5.443265533447265,
+      "step": 990
+    },
+    {
+      "epoch": 1.8540792540792541,
+      "grad_norm": 11.976597785949707,
+      "learning_rate": 6.500191779396439e-06,
+      "loss": 5.410481643676758,
+      "step": 995
+    },
+    {
+      "epoch": 1.8634032634032636,
+      "grad_norm": 14.439790725708008,
+      "learning_rate": 6.461383613835427e-06,
+      "loss": 5.3311511993408205,
+      "step": 1000
+    },
+    {
+      "epoch": 1.8634032634032636,
+      "eval_com_accuracy": 0.24,
+      "eval_rew_accuracy": 0.356,
+      "step": 1000
+    },
+    {
+      "epoch": 1.8727272727272726,
+      "grad_norm": 15.29392147064209,
+      "learning_rate": 6.4224789459243705e-06,
+      "loss": 5.477576446533203,
+      "step": 1005
+    },
+    {
+      "epoch": 1.882051282051282,
+      "grad_norm": 14.032716751098633,
+      "learning_rate": 6.383480344729903e-06,
+      "loss": 5.458187103271484,
+      "step": 1010
+    },
+    {
+      "epoch": 1.8913752913752915,
+      "grad_norm": 11.861150741577148,
+      "learning_rate": 6.344390385521534e-06,
+      "loss": 5.806562423706055,
+      "step": 1015
+    },
+    {
+      "epoch": 1.9006993006993007,
+      "grad_norm": 14.015438079833984,
+      "learning_rate": 6.305211649601595e-06,
+      "loss": 5.2158203125,
+      "step": 1020
+    },
+    {
+      "epoch": 1.91002331002331,
+      "grad_norm": 15.672555923461914,
+      "learning_rate": 6.265946724134782e-06,
+      "loss": 5.469319915771484,
+      "step": 1025
+    },
+    {
+      "epoch": 1.9193473193473194,
+      "grad_norm": 13.891338348388672,
+      "learning_rate": 6.226598201977299e-06,
+      "loss": 5.495826721191406,
+      "step": 1030
+    },
+    {
+      "epoch": 1.9286713286713286,
+      "grad_norm": 12.441986083984375,
+      "learning_rate": 6.187168681505666e-06,
+      "loss": 5.466391372680664,
+      "step": 1035
+    },
+    {
+      "epoch": 1.9379953379953379,
+      "grad_norm": 13.905447006225586,
+      "learning_rate": 6.1476607664451105e-06,
+      "loss": 5.558423233032227,
+      "step": 1040
+    },
+    {
+      "epoch": 1.9473193473193473,
+      "grad_norm": 14.817802429199219,
+      "learning_rate": 6.1080770656976444e-06,
+      "loss": 5.50462646484375,
+      "step": 1045
+    },
+    {
+      "epoch": 1.9566433566433568,
+      "grad_norm": 14.293425559997559,
+      "learning_rate": 6.068420193169779e-06,
+      "loss": 5.567475128173828,
+      "step": 1050
+    },
+    {
+      "epoch": 1.9566433566433568,
+      "eval_com_accuracy": 0.26,
+      "eval_rew_accuracy": 0.384,
+      "step": 1050
+    },
+    {
+      "epoch": 1.965967365967366,
+      "grad_norm": 13.285664558410645,
+      "learning_rate": 6.0286927675999205e-06,
+      "loss": 5.3831031799316404,
+      "step": 1055
+    },
+    {
+      "epoch": 1.9752913752913752,
+      "grad_norm": 12.634699821472168,
+      "learning_rate": 5.98889741238544e-06,
+      "loss": 5.4307403564453125,
+      "step": 1060
+    },
+    {
+      "epoch": 1.9846153846153847,
+      "grad_norm": 13.824172019958496,
+      "learning_rate": 5.949036755409432e-06,
+      "loss": 5.559865570068359,
+      "step": 1065
+    },
+    {
+      "epoch": 1.993939393939394,
+      "grad_norm": 14.028116226196289,
+      "learning_rate": 5.909113428867195e-06,
+      "loss": 5.506551361083984,
+      "step": 1070
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 2148,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-1074/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b6ca094dca60606350f327809d196d6faa38b81cdd561be5d924e82822f09231
+size 5201

checkpoint-1611/chat_template.jinja

@@ -0,0 +1,120 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {%- if messages[0].content is string %}
+            {{- messages[0].content }}
+        {%- else %}
+            {%- for content in messages[0].content %}
+                {%- if 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '\n\n' }}
+    {%- endif %}
+    {{- "# 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>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\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" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' }}
+        {%- if messages[0].content is string %}
+            {{- messages[0].content }}
+        {%- else %}
+            {%- for content in messages[0].content %}
+                {%- if 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- for message in messages %}
+    {%- if message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- else %}
+            {%- for content in message.content %}
+                {%- if content.type == 'image' or 'image' in content or 'image_url' in content %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                    {%- if add_vision_id %}Picture {{ image_count.value }}: {% endif -%}
+                    <|vision_start|><|image_pad|><|vision_end|>
+                {%- elif content.type == 'video' or 'video' in content %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                    {%- if add_vision_id %}Video {{ video_count.value }}: {% endif -%}
+                    <|vision_start|><|video_pad|><|vision_end|>
+                {%- elif 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role + '\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- else %}
+            {%- for content_item in message.content %}
+                {%- if 'text' in content_item %}
+                    {{- content_item.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and message.content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- else %}
+            {%- for content in message.content %}
+                {%- if content.type == 'image' or 'image' in content or 'image_url' in content %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                    {%- if add_vision_id %}Picture {{ image_count.value }}: {% endif -%}
+                    <|vision_start|><|image_pad|><|vision_end|>
+                {%- elif content.type == 'video' or 'video' in content %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                    {%- if add_vision_id %}Video {{ video_count.value }}: {% endif -%}
+                    <|vision_start|><|video_pad|><|vision_end|>
+                {%- elif 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}

checkpoint-1611/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6e3149394644aa537fa2ade1cc32141f45f23bf2f62008931aa4dc5d1228e108
+size 4885870044

checkpoint-1611/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:338233802a6165498d51ce8433326c5847271032e988b86ee66ad9b84b19532c
+size 8527467778

checkpoint-1611/processor_config.json

@@ -0,0 +1,63 @@
+{
+  "image_processor": {
+    "data_format": "channels_first",
+    "do_convert_rgb": true,
+    "do_normalize": true,
+    "do_rescale": true,
+    "do_resize": true,
+    "image_mean": [
+      0.5,
+      0.5,
+      0.5
+    ],
+    "image_processor_type": "Qwen2VLImageProcessorFast",
+    "image_std": [
+      0.5,
+      0.5,
+      0.5
+    ],
+    "merge_size": 2,
+    "patch_size": 16,
+    "resample": 3,
+    "rescale_factor": 0.00392156862745098,
+    "size": {
+      "longest_edge": 16777216,
+      "shortest_edge": 65536
+    },
+    "temporal_patch_size": 2
+  },
+  "processor_class": "Qwen3VLProcessor",
+  "video_processor": {
+    "data_format": "channels_first",
+    "default_to_square": true,
+    "do_convert_rgb": true,
+    "do_normalize": true,
+    "do_rescale": true,
+    "do_resize": true,
+    "do_sample_frames": true,
+    "fps": 2,
+    "image_mean": [
+      0.5,
+      0.5,
+      0.5
+    ],
+    "image_std": [
+      0.5,
+      0.5,
+      0.5
+    ],
+    "max_frames": 768,
+    "merge_size": 2,
+    "min_frames": 4,
+    "patch_size": 16,
+    "resample": 3,
+    "rescale_factor": 0.00392156862745098,
+    "return_metadata": false,
+    "size": {
+      "longest_edge": 25165824,
+      "shortest_edge": 4096
+    },
+    "temporal_patch_size": 2,
+    "video_processor_type": "Qwen3VLVideoProcessor"
+  }
+}

checkpoint-1611/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:659b1cdee2219458dd84ce6a632a595465680b8080e5c44bd600ff97eca8d752
+size 15429

checkpoint-1611/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:86accf27064cdd503053e90476a6bd10de333d4ff0594535ad55ea13a473c91d
+size 15429

checkpoint-1611/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:18ca8d714ef40be035404c1957b5a4dee84e1f43980408393f8aa710552ee6f6
+size 15429

checkpoint-1611/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2cfdebe99e40accc9c9d8f09c63136a14abda997d9b501969ec8e16e9d183179
+size 15429

checkpoint-1611/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6b84dfa67996ae63f368c737005548dbc693f441e40d5fb4a97bef193cc88674
+size 1465

checkpoint-1611/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8579e1ca7cc5d82a9e0202eed555529996f4ffe7f563c2979a0290cf3db452d3
+size 11422818

checkpoint-1611/tokenizer_config.json

@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_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|>"
+  ],
+  "is_local": false,
+  "model_max_length": 262144,
+  "pad_token": "<|endoftext|>",
+  "processor_class": "Qwen3VLProcessor",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

checkpoint-1611/trainer_state.json

@@ -0,0 +1,2480 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 50,
+  "global_step": 1611,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.009324009324009324,
+      "grad_norm": 269.7618103027344,
+      "learning_rate": 1.8604651162790698e-07,
+      "loss": 13.883856201171875,
+      "step": 5
+    },
+    {
+      "epoch": 0.018648018648018648,
+      "grad_norm": 256.5934143066406,
+      "learning_rate": 4.186046511627907e-07,
+      "loss": 13.883564758300782,
+      "step": 10
+    },
+    {
+      "epoch": 0.027972027972027972,
+      "grad_norm": 247.4810028076172,
+      "learning_rate": 6.511627906976745e-07,
+      "loss": 13.845947265625,
+      "step": 15
+    },
+    {
+      "epoch": 0.037296037296037296,
+      "grad_norm": 243.70272827148438,
+      "learning_rate": 8.837209302325582e-07,
+      "loss": 13.705046081542969,
+      "step": 20
+    },
+    {
+      "epoch": 0.046620046620046623,
+      "grad_norm": 238.2322235107422,
+      "learning_rate": 1.116279069767442e-06,
+      "loss": 13.556683349609376,
+      "step": 25
+    },
+    {
+      "epoch": 0.055944055944055944,
+      "grad_norm": 241.71214294433594,
+      "learning_rate": 1.3488372093023258e-06,
+      "loss": 13.285650634765625,
+      "step": 30
+    },
+    {
+      "epoch": 0.06526806526806526,
+      "grad_norm": 221.19960021972656,
+      "learning_rate": 1.5813953488372093e-06,
+      "loss": 13.05435791015625,
+      "step": 35
+    },
+    {
+      "epoch": 0.07459207459207459,
+      "grad_norm": 205.08326721191406,
+      "learning_rate": 1.8139534883720933e-06,
+      "loss": 12.770874786376954,
+      "step": 40
+    },
+    {
+      "epoch": 0.08391608391608392,
+      "grad_norm": 170.5838623046875,
+      "learning_rate": 2.0465116279069768e-06,
+      "loss": 12.280790710449219,
+      "step": 45
+    },
+    {
+      "epoch": 0.09324009324009325,
+      "grad_norm": 96.90670776367188,
+      "learning_rate": 2.2790697674418607e-06,
+      "loss": 11.730733489990234,
+      "step": 50
+    },
+    {
+      "epoch": 0.09324009324009325,
+      "eval_com_accuracy": 0.2,
+      "eval_rew_accuracy": 0.276,
+      "step": 50
+    },
+    {
+      "epoch": 0.10256410256410256,
+      "grad_norm": 61.14415740966797,
+      "learning_rate": 2.5116279069767446e-06,
+      "loss": 11.25308609008789,
+      "step": 55
+    },
+    {
+      "epoch": 0.11188811188811189,
+      "grad_norm": 56.027095794677734,
+      "learning_rate": 2.744186046511628e-06,
+      "loss": 11.197045135498048,
+      "step": 60
+    },
+    {
+      "epoch": 0.12121212121212122,
+      "grad_norm": 46.54409408569336,
+      "learning_rate": 2.9767441860465116e-06,
+      "loss": 10.874276733398437,
+      "step": 65
+    },
+    {
+      "epoch": 0.13053613053613053,
+      "grad_norm": 40.62468719482422,
+      "learning_rate": 3.2093023255813956e-06,
+      "loss": 10.69852294921875,
+      "step": 70
+    },
+    {
+      "epoch": 0.13986013986013987,
+      "grad_norm": 37.44975662231445,
+      "learning_rate": 3.4418604651162795e-06,
+      "loss": 10.560739135742187,
+      "step": 75
+    },
+    {
+      "epoch": 0.14918414918414918,
+      "grad_norm": 32.19042205810547,
+      "learning_rate": 3.674418604651163e-06,
+      "loss": 10.408053588867187,
+      "step": 80
+    },
+    {
+      "epoch": 0.1585081585081585,
+      "grad_norm": 27.208942413330078,
+      "learning_rate": 3.906976744186047e-06,
+      "loss": 10.108213806152344,
+      "step": 85
+    },
+    {
+      "epoch": 0.16783216783216784,
+      "grad_norm": 24.76529312133789,
+      "learning_rate": 4.1395348837209304e-06,
+      "loss": 9.929393005371093,
+      "step": 90
+    },
+    {
+      "epoch": 0.17715617715617715,
+      "grad_norm": 23.917495727539062,
+      "learning_rate": 4.372093023255815e-06,
+      "loss": 9.677659606933593,
+      "step": 95
+    },
+    {
+      "epoch": 0.1864801864801865,
+      "grad_norm": 22.98892593383789,
+      "learning_rate": 4.604651162790698e-06,
+      "loss": 9.481817626953125,
+      "step": 100
+    },
+    {
+      "epoch": 0.1864801864801865,
+      "eval_com_accuracy": 0.176,
+      "eval_rew_accuracy": 0.264,
+      "step": 100
+    },
+    {
+      "epoch": 0.1958041958041958,
+      "grad_norm": 25.177200317382812,
+      "learning_rate": 4.837209302325582e-06,
+      "loss": 9.42150650024414,
+      "step": 105
+    },
+    {
+      "epoch": 0.20512820512820512,
+      "grad_norm": 18.95603370666504,
+      "learning_rate": 5.069767441860466e-06,
+      "loss": 9.267645263671875,
+      "step": 110
+    },
+    {
+      "epoch": 0.21445221445221446,
+      "grad_norm": 22.155452728271484,
+      "learning_rate": 5.302325581395349e-06,
+      "loss": 9.06854248046875,
+      "step": 115
+    },
+    {
+      "epoch": 0.22377622377622378,
+      "grad_norm": 20.26909065246582,
+      "learning_rate": 5.534883720930233e-06,
+      "loss": 9.045303344726562,
+      "step": 120
+    },
+    {
+      "epoch": 0.2331002331002331,
+      "grad_norm": 26.49457359313965,
+      "learning_rate": 5.7674418604651175e-06,
+      "loss": 8.91327362060547,
+      "step": 125
+    },
+    {
+      "epoch": 0.24242424242424243,
+      "grad_norm": 19.886178970336914,
+      "learning_rate": 6e-06,
+      "loss": 8.950289154052735,
+      "step": 130
+    },
+    {
+      "epoch": 0.2517482517482518,
+      "grad_norm": 18.22088050842285,
+      "learning_rate": 6.2325581395348845e-06,
+      "loss": 8.710862731933593,
+      "step": 135
+    },
+    {
+      "epoch": 0.26107226107226106,
+      "grad_norm": 21.09984016418457,
+      "learning_rate": 6.465116279069767e-06,
+      "loss": 8.762718200683594,
+      "step": 140
+    },
+    {
+      "epoch": 0.2703962703962704,
+      "grad_norm": 17.96813201904297,
+      "learning_rate": 6.6976744186046515e-06,
+      "loss": 8.533232116699219,
+      "step": 145
+    },
+    {
+      "epoch": 0.27972027972027974,
+      "grad_norm": 20.52909278869629,
+      "learning_rate": 6.930232558139536e-06,
+      "loss": 8.279258728027344,
+      "step": 150
+    },
+    {
+      "epoch": 0.27972027972027974,
+      "eval_com_accuracy": 0.152,
+      "eval_rew_accuracy": 0.252,
+      "step": 150
+    },
+    {
+      "epoch": 0.289044289044289,
+      "grad_norm": 19.475040435791016,
+      "learning_rate": 7.1627906976744185e-06,
+      "loss": 8.369435882568359,
+      "step": 155
+    },
+    {
+      "epoch": 0.29836829836829837,
+      "grad_norm": 16.741973876953125,
+      "learning_rate": 7.395348837209303e-06,
+      "loss": 8.280664825439453,
+      "step": 160
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 19.79235076904297,
+      "learning_rate": 7.627906976744187e-06,
+      "loss": 8.200827026367188,
+      "step": 165
+    },
+    {
+      "epoch": 0.317016317016317,
+      "grad_norm": 32.991554260253906,
+      "learning_rate": 7.86046511627907e-06,
+      "loss": 7.973581695556641,
+      "step": 170
+    },
+    {
+      "epoch": 0.32634032634032634,
+      "grad_norm": 17.949445724487305,
+      "learning_rate": 8.093023255813955e-06,
+      "loss": 8.141377258300782,
+      "step": 175
+    },
+    {
+      "epoch": 0.3356643356643357,
+      "grad_norm": 16.19559097290039,
+      "learning_rate": 8.325581395348837e-06,
+      "loss": 7.837094116210937,
+      "step": 180
+    },
+    {
+      "epoch": 0.34498834498834496,
+      "grad_norm": 23.260622024536133,
+      "learning_rate": 8.558139534883722e-06,
+      "loss": 7.763467407226562,
+      "step": 185
+    },
+    {
+      "epoch": 0.3543123543123543,
+      "grad_norm": 24.858335494995117,
+      "learning_rate": 8.790697674418606e-06,
+      "loss": 7.886874389648438,
+      "step": 190
+    },
+    {
+      "epoch": 0.36363636363636365,
+      "grad_norm": 18.59360122680664,
+      "learning_rate": 9.023255813953489e-06,
+      "loss": 7.7624969482421875,
+      "step": 195
+    },
+    {
+      "epoch": 0.372960372960373,
+      "grad_norm": 18.056495666503906,
+      "learning_rate": 9.255813953488373e-06,
+      "loss": 7.4621437072753904,
+      "step": 200
+    },
+    {
+      "epoch": 0.372960372960373,
+      "eval_com_accuracy": 0.144,
+      "eval_rew_accuracy": 0.332,
+      "step": 200
+    },
+    {
+      "epoch": 0.3822843822843823,
+      "grad_norm": 16.496261596679688,
+      "learning_rate": 9.488372093023258e-06,
+      "loss": 7.526261901855468,
+      "step": 205
+    },
+    {
+      "epoch": 0.3916083916083916,
+      "grad_norm": 16.77888298034668,
+      "learning_rate": 9.72093023255814e-06,
+      "loss": 7.624967193603515,
+      "step": 210
+    },
+    {
+      "epoch": 0.40093240093240096,
+      "grad_norm": 17.096778869628906,
+      "learning_rate": 9.953488372093025e-06,
+      "loss": 7.337849426269531,
+      "step": 215
+    },
+    {
+      "epoch": 0.41025641025641024,
+      "grad_norm": 18.3544979095459,
+      "learning_rate": 9.999894343918674e-06,
+      "loss": 7.354197692871094,
+      "step": 220
+    },
+    {
+      "epoch": 0.4195804195804196,
+      "grad_norm": 15.287137985229492,
+      "learning_rate": 9.999465123741172e-06,
+      "loss": 7.085916137695312,
+      "step": 225
+    },
+    {
+      "epoch": 0.4289044289044289,
+      "grad_norm": 17.03432273864746,
+      "learning_rate": 9.998705764284132e-06,
+      "loss": 7.059596252441406,
+      "step": 230
+    },
+    {
+      "epoch": 0.4382284382284382,
+      "grad_norm": 15.470928192138672,
+      "learning_rate": 9.99761631569179e-06,
+      "loss": 7.118296813964844,
+      "step": 235
+    },
+    {
+      "epoch": 0.44755244755244755,
+      "grad_norm": 15.732117652893066,
+      "learning_rate": 9.9961968499058e-06,
+      "loss": 6.981436157226563,
+      "step": 240
+    },
+    {
+      "epoch": 0.4568764568764569,
+      "grad_norm": 20.92630958557129,
+      "learning_rate": 9.994447460660473e-06,
+      "loss": 6.96953125,
+      "step": 245
+    },
+    {
+      "epoch": 0.4662004662004662,
+      "grad_norm": 14.848095893859863,
+      "learning_rate": 9.992368263476585e-06,
+      "loss": 6.831211090087891,
+      "step": 250
+    },
+    {
+      "epoch": 0.4662004662004662,
+      "eval_com_accuracy": 0.132,
+      "eval_rew_accuracy": 0.312,
+      "step": 250
+    },
+    {
+      "epoch": 0.4755244755244755,
+      "grad_norm": 18.59755516052246,
+      "learning_rate": 9.989959395653756e-06,
+      "loss": 6.873024749755859,
+      "step": 255
+    },
+    {
+      "epoch": 0.48484848484848486,
+      "grad_norm": 15.705364227294922,
+      "learning_rate": 9.98722101626138e-06,
+      "loss": 6.633563232421875,
+      "step": 260
+    },
+    {
+      "epoch": 0.49417249417249415,
+      "grad_norm": 39.2294807434082,
+      "learning_rate": 9.984153306128124e-06,
+      "loss": 6.63948974609375,
+      "step": 265
+    },
+    {
+      "epoch": 0.5034965034965035,
+      "grad_norm": 13.209358215332031,
+      "learning_rate": 9.980756467829977e-06,
+      "loss": 6.688297271728516,
+      "step": 270
+    },
+    {
+      "epoch": 0.5128205128205128,
+      "grad_norm": 14.645393371582031,
+      "learning_rate": 9.977030725676887e-06,
+      "loss": 6.4885505676269535,
+      "step": 275
+    },
+    {
+      "epoch": 0.5221445221445221,
+      "grad_norm": 17.008819580078125,
+      "learning_rate": 9.972976325697938e-06,
+      "loss": 6.78935546875,
+      "step": 280
+    },
+    {
+      "epoch": 0.5314685314685315,
+      "grad_norm": 16.029518127441406,
+      "learning_rate": 9.96859353562511e-06,
+      "loss": 6.681609344482422,
+      "step": 285
+    },
+    {
+      "epoch": 0.5407925407925408,
+      "grad_norm": 13.878348350524902,
+      "learning_rate": 9.963882644875594e-06,
+      "loss": 6.528868103027344,
+      "step": 290
+    },
+    {
+      "epoch": 0.5501165501165501,
+      "grad_norm": 13.996464729309082,
+      "learning_rate": 9.958843964532683e-06,
+      "loss": 6.431344604492187,
+      "step": 295
+    },
+    {
+      "epoch": 0.5594405594405595,
+      "grad_norm": 15.72003173828125,
+      "learning_rate": 9.953477827325229e-06,
+      "loss": 6.5927978515625,
+      "step": 300
+    },
+    {
+      "epoch": 0.5594405594405595,
+      "eval_com_accuracy": 0.148,
+      "eval_rew_accuracy": 0.368,
+      "step": 300
+    },
+    {
+      "epoch": 0.5687645687645687,
+      "grad_norm": 16.416725158691406,
+      "learning_rate": 9.947784587605678e-06,
+      "loss": 6.411885833740234,
+      "step": 305
+    },
+    {
+      "epoch": 0.578088578088578,
+      "grad_norm": 14.880790710449219,
+      "learning_rate": 9.941764621326655e-06,
+      "loss": 6.5298927307128904,
+      "step": 310
+    },
+    {
+      "epoch": 0.5874125874125874,
+      "grad_norm": 17.828296661376953,
+      "learning_rate": 9.935418326016153e-06,
+      "loss": 6.045618057250977,
+      "step": 315
+    },
+    {
+      "epoch": 0.5967365967365967,
+      "grad_norm": 13.30245590209961,
+      "learning_rate": 9.928746120751275e-06,
+      "loss": 6.1650840759277346,
+      "step": 320
+    },
+    {
+      "epoch": 0.6060606060606061,
+      "grad_norm": 13.527432441711426,
+      "learning_rate": 9.921748446130564e-06,
+      "loss": 6.349945831298828,
+      "step": 325
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 14.491817474365234,
+      "learning_rate": 9.9144257642449e-06,
+      "loss": 6.232943725585938,
+      "step": 330
+    },
+    {
+      "epoch": 0.6247086247086248,
+      "grad_norm": 15.7533597946167,
+      "learning_rate": 9.906778558647e-06,
+      "loss": 6.564260864257813,
+      "step": 335
+    },
+    {
+      "epoch": 0.634032634032634,
+      "grad_norm": 16.538602828979492,
+      "learning_rate": 9.898807334319471e-06,
+      "loss": 6.122506332397461,
+      "step": 340
+    },
+    {
+      "epoch": 0.6433566433566433,
+      "grad_norm": 16.459688186645508,
+      "learning_rate": 9.890512617641474e-06,
+      "loss": 6.332431030273438,
+      "step": 345
+    },
+    {
+      "epoch": 0.6526806526806527,
+      "grad_norm": 14.879058837890625,
+      "learning_rate": 9.881894956353963e-06,
+      "loss": 6.118003082275391,
+      "step": 350
+    },
+    {
+      "epoch": 0.6526806526806527,
+      "eval_com_accuracy": 0.164,
+      "eval_rew_accuracy": 0.364,
+      "step": 350
+    },
+    {
+      "epoch": 0.662004662004662,
+      "grad_norm": 16.02516746520996,
+      "learning_rate": 9.87295491952351e-06,
+      "loss": 6.241584014892578,
+      "step": 355
+    },
+    {
+      "epoch": 0.6713286713286714,
+      "grad_norm": 14.9837646484375,
+      "learning_rate": 9.863693097504733e-06,
+      "loss": 6.083467864990235,
+      "step": 360
+    },
+    {
+      "epoch": 0.6806526806526807,
+      "grad_norm": 14.138517379760742,
+      "learning_rate": 9.854110101901308e-06,
+      "loss": 6.143560791015625,
+      "step": 365
+    },
+    {
+      "epoch": 0.6899766899766899,
+      "grad_norm": 12.696243286132812,
+      "learning_rate": 9.844206565525585e-06,
+      "loss": 6.061418914794922,
+      "step": 370
+    },
+    {
+      "epoch": 0.6993006993006993,
+      "grad_norm": 15.22846508026123,
+      "learning_rate": 9.833983142356792e-06,
+      "loss": 6.193332672119141,
+      "step": 375
+    },
+    {
+      "epoch": 0.7086247086247086,
+      "grad_norm": 15.284857749938965,
+      "learning_rate": 9.823440507497863e-06,
+      "loss": 6.0868377685546875,
+      "step": 380
+    },
+    {
+      "epoch": 0.717948717948718,
+      "grad_norm": 14.117541313171387,
+      "learning_rate": 9.812579357130848e-06,
+      "loss": 6.289948272705078,
+      "step": 385
+    },
+    {
+      "epoch": 0.7272727272727273,
+      "grad_norm": 12.648367881774902,
+      "learning_rate": 9.801400408470943e-06,
+      "loss": 6.194795227050781,
+      "step": 390
+    },
+    {
+      "epoch": 0.7365967365967366,
+      "grad_norm": 12.394021034240723,
+      "learning_rate": 9.789904399719124e-06,
+      "loss": 6.031880950927734,
+      "step": 395
+    },
+    {
+      "epoch": 0.745920745920746,
+      "grad_norm": 15.566425323486328,
+      "learning_rate": 9.778092090013416e-06,
+      "loss": 6.0966644287109375,
+      "step": 400
+    },
+    {
+      "epoch": 0.745920745920746,
+      "eval_com_accuracy": 0.184,
+      "eval_rew_accuracy": 0.364,
+      "step": 400
+    },
+    {
+      "epoch": 0.7552447552447552,
+      "grad_norm": 14.885455131530762,
+      "learning_rate": 9.76596425937874e-06,
+      "loss": 6.476510620117187,
+      "step": 405
+    },
+    {
+      "epoch": 0.7645687645687645,
+      "grad_norm": 16.627117156982422,
+      "learning_rate": 9.753521708675426e-06,
+      "loss": 5.9605262756347654,
+      "step": 410
+    },
+    {
+      "epoch": 0.7738927738927739,
+      "grad_norm": 14.517923355102539,
+      "learning_rate": 9.740765259546312e-06,
+      "loss": 6.1104694366455075,
+      "step": 415
+    },
+    {
+      "epoch": 0.7832167832167832,
+      "grad_norm": 12.007637977600098,
+      "learning_rate": 9.727695754362498e-06,
+      "loss": 5.964785766601563,
+      "step": 420
+    },
+    {
+      "epoch": 0.7925407925407926,
+      "grad_norm": 16.17314338684082,
+      "learning_rate": 9.714314056167711e-06,
+      "loss": 5.976337432861328,
+      "step": 425
+    },
+    {
+      "epoch": 0.8018648018648019,
+      "grad_norm": 13.246196746826172,
+      "learning_rate": 9.700621048621322e-06,
+      "loss": 5.936331939697266,
+      "step": 430
+    },
+    {
+      "epoch": 0.8111888111888111,
+      "grad_norm": 15.030324935913086,
+      "learning_rate": 9.686617635939988e-06,
+      "loss": 6.209032821655273,
+      "step": 435
+    },
+    {
+      "epoch": 0.8205128205128205,
+      "grad_norm": 13.853086471557617,
+      "learning_rate": 9.672304742837945e-06,
+      "loss": 5.985638427734375,
+      "step": 440
+    },
+    {
+      "epoch": 0.8298368298368298,
+      "grad_norm": 13.534188270568848,
+      "learning_rate": 9.657683314465948e-06,
+      "loss": 5.7971649169921875,
+      "step": 445
+    },
+    {
+      "epoch": 0.8391608391608392,
+      "grad_norm": 26.1879825592041,
+      "learning_rate": 9.642754316348846e-06,
+      "loss": 5.949985504150391,
+      "step": 450
+    },
+    {
+      "epoch": 0.8391608391608392,
+      "eval_com_accuracy": 0.176,
+      "eval_rew_accuracy": 0.34,
+      "step": 450
+    },
+    {
+      "epoch": 0.8484848484848485,
+      "grad_norm": 16.848834991455078,
+      "learning_rate": 9.627518734321837e-06,
+      "loss": 6.1193115234375,
+      "step": 455
+    },
+    {
+      "epoch": 0.8578088578088578,
+      "grad_norm": 30.765796661376953,
+      "learning_rate": 9.61197757446536e-06,
+      "loss": 5.74780158996582,
+      "step": 460
+    },
+    {
+      "epoch": 0.8671328671328671,
+      "grad_norm": 13.27775764465332,
+      "learning_rate": 9.596131863038664e-06,
+      "loss": 5.994970703125,
+      "step": 465
+    },
+    {
+      "epoch": 0.8764568764568764,
+      "grad_norm": 14.923792839050293,
+      "learning_rate": 9.579982646412039e-06,
+      "loss": 5.810473251342773,
+      "step": 470
+    },
+    {
+      "epoch": 0.8857808857808858,
+      "grad_norm": 14.351722717285156,
+      "learning_rate": 9.563530990997707e-06,
+      "loss": 6.09692268371582,
+      "step": 475
+    },
+    {
+      "epoch": 0.8951048951048951,
+      "grad_norm": 11.83583927154541,
+      "learning_rate": 9.546777983179421e-06,
+      "loss": 5.867227935791016,
+      "step": 480
+    },
+    {
+      "epoch": 0.9044289044289044,
+      "grad_norm": 19.415884017944336,
+      "learning_rate": 9.529724729240712e-06,
+      "loss": 5.834615707397461,
+      "step": 485
+    },
+    {
+      "epoch": 0.9137529137529138,
+      "grad_norm": 12.147261619567871,
+      "learning_rate": 9.512372355291838e-06,
+      "loss": 6.039286041259766,
+      "step": 490
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 12.706372261047363,
+      "learning_rate": 9.494722007195427e-06,
+      "loss": 6.06304702758789,
+      "step": 495
+    },
+    {
+      "epoch": 0.9324009324009324,
+      "grad_norm": 12.642041206359863,
+      "learning_rate": 9.476774850490803e-06,
+      "loss": 5.9729866027832035,
+      "step": 500
+    },
+    {
+      "epoch": 0.9324009324009324,
+      "eval_com_accuracy": 0.204,
+      "eval_rew_accuracy": 0.328,
+      "step": 500
+    },
+    {
+      "epoch": 0.9417249417249417,
+      "grad_norm": 12.66169548034668,
+      "learning_rate": 9.458532070317021e-06,
+      "loss": 5.8281097412109375,
+      "step": 505
+    },
+    {
+      "epoch": 0.951048951048951,
+      "grad_norm": 13.191854476928711,
+      "learning_rate": 9.439994871334614e-06,
+      "loss": 6.183546447753907,
+      "step": 510
+    },
+    {
+      "epoch": 0.9603729603729604,
+      "grad_norm": 12.52676010131836,
+      "learning_rate": 9.421164477646031e-06,
+      "loss": 5.785601043701172,
+      "step": 515
+    },
+    {
+      "epoch": 0.9696969696969697,
+      "grad_norm": 13.225418090820312,
+      "learning_rate": 9.402042132714817e-06,
+      "loss": 5.784825134277344,
+      "step": 520
+    },
+    {
+      "epoch": 0.9790209790209791,
+      "grad_norm": 12.035611152648926,
+      "learning_rate": 9.382629099283486e-06,
+      "loss": 5.9122272491455075,
+      "step": 525
+    },
+    {
+      "epoch": 0.9883449883449883,
+      "grad_norm": 12.166351318359375,
+      "learning_rate": 9.362926659290149e-06,
+      "loss": 5.790053939819336,
+      "step": 530
+    },
+    {
+      "epoch": 0.9976689976689976,
+      "grad_norm": 13.22154426574707,
+      "learning_rate": 9.342936113783855e-06,
+      "loss": 5.745006561279297,
+      "step": 535
+    },
+    {
+      "epoch": 1.0055944055944055,
+      "grad_norm": 13.12752628326416,
+      "learning_rate": 9.32265878283868e-06,
+      "loss": 5.1531005859375,
+      "step": 540
+    },
+    {
+      "epoch": 1.014918414918415,
+      "grad_norm": 15.964373588562012,
+      "learning_rate": 9.302096005466547e-06,
+      "loss": 5.7917236328125,
+      "step": 545
+    },
+    {
+      "epoch": 1.0242424242424242,
+      "grad_norm": 18.454797744750977,
+      "learning_rate": 9.281249139528816e-06,
+      "loss": 5.967070007324219,
+      "step": 550
+    },
+    {
+      "epoch": 1.0242424242424242,
+      "eval_com_accuracy": 0.212,
+      "eval_rew_accuracy": 0.352,
+      "step": 550
+    },
+    {
+      "epoch": 1.0335664335664336,
+      "grad_norm": 15.802419662475586,
+      "learning_rate": 9.260119561646614e-06,
+      "loss": 6.093529129028321,
+      "step": 555
+    },
+    {
+      "epoch": 1.0428904428904429,
+      "grad_norm": 21.95537757873535,
+      "learning_rate": 9.238708667109924e-06,
+      "loss": 5.489861297607422,
+      "step": 560
+    },
+    {
+      "epoch": 1.0522144522144523,
+      "grad_norm": 13.373199462890625,
+      "learning_rate": 9.217017869785453e-06,
+      "loss": 5.871435928344726,
+      "step": 565
+    },
+    {
+      "epoch": 1.0615384615384615,
+      "grad_norm": 13.36340618133545,
+      "learning_rate": 9.19504860202327e-06,
+      "loss": 5.805611038208008,
+      "step": 570
+    },
+    {
+      "epoch": 1.0708624708624708,
+      "grad_norm": 14.099553108215332,
+      "learning_rate": 9.172802314562214e-06,
+      "loss": 5.600634765625,
+      "step": 575
+    },
+    {
+      "epoch": 1.0801864801864802,
+      "grad_norm": 15.012040138244629,
+      "learning_rate": 9.150280476434098e-06,
+      "loss": 5.793732452392578,
+      "step": 580
+    },
+    {
+      "epoch": 1.0895104895104895,
+      "grad_norm": 15.98738956451416,
+      "learning_rate": 9.127484574866699e-06,
+      "loss": 5.8642620086669925,
+      "step": 585
+    },
+    {
+      "epoch": 1.098834498834499,
+      "grad_norm": 13.785572052001953,
+      "learning_rate": 9.104416115185557e-06,
+      "loss": 5.770995330810547,
+      "step": 590
+    },
+    {
+      "epoch": 1.1081585081585081,
+      "grad_norm": 12.145087242126465,
+      "learning_rate": 9.08107662071456e-06,
+      "loss": 5.850731277465821,
+      "step": 595
+    },
+    {
+      "epoch": 1.1174825174825176,
+      "grad_norm": 13.681862831115723,
+      "learning_rate": 9.057467632675357e-06,
+      "loss": 5.782034301757813,
+      "step": 600
+    },
+    {
+      "epoch": 1.1174825174825176,
+      "eval_com_accuracy": 0.22,
+      "eval_rew_accuracy": 0.348,
+      "step": 600
+    },
+    {
+      "epoch": 1.1268065268065268,
+      "grad_norm": 13.870027542114258,
+      "learning_rate": 9.033590710085584e-06,
+      "loss": 5.723427963256836,
+      "step": 605
+    },
+    {
+      "epoch": 1.136130536130536,
+      "grad_norm": 16.20020294189453,
+      "learning_rate": 9.00944742965592e-06,
+      "loss": 5.953633117675781,
+      "step": 610
+    },
+    {
+      "epoch": 1.1454545454545455,
+      "grad_norm": 15.48993968963623,
+      "learning_rate": 8.985039385685952e-06,
+      "loss": 5.809123992919922,
+      "step": 615
+    },
+    {
+      "epoch": 1.1547785547785547,
+      "grad_norm": 13.686211585998535,
+      "learning_rate": 8.960368189958913e-06,
+      "loss": 5.813122177124024,
+      "step": 620
+    },
+    {
+      "epoch": 1.1641025641025642,
+      "grad_norm": 13.733199119567871,
+      "learning_rate": 8.935435471635238e-06,
+      "loss": 5.6748512268066404,
+      "step": 625
+    },
+    {
+      "epoch": 1.1734265734265734,
+      "grad_norm": 12.450380325317383,
+      "learning_rate": 8.91024287714499e-06,
+      "loss": 5.6834667205810545,
+      "step": 630
+    },
+    {
+      "epoch": 1.1827505827505829,
+      "grad_norm": 14.111934661865234,
+      "learning_rate": 8.884792070079128e-06,
+      "loss": 5.680038833618164,
+      "step": 635
+    },
+    {
+      "epoch": 1.192074592074592,
+      "grad_norm": 13.175325393676758,
+      "learning_rate": 8.859084731079664e-06,
+      "loss": 5.445759201049805,
+      "step": 640
+    },
+    {
+      "epoch": 1.2013986013986013,
+      "grad_norm": 14.951041221618652,
+      "learning_rate": 8.833122557728667e-06,
+      "loss": 5.673246765136719,
+      "step": 645
+    },
+    {
+      "epoch": 1.2107226107226108,
+      "grad_norm": 12.364238739013672,
+      "learning_rate": 8.806907264436183e-06,
+      "loss": 5.234909439086914,
+      "step": 650
+    },
+    {
+      "epoch": 1.2107226107226108,
+      "eval_com_accuracy": 0.224,
+      "eval_rew_accuracy": 0.348,
+      "step": 650
+    },
+    {
+      "epoch": 1.22004662004662,
+      "grad_norm": 13.065054893493652,
+      "learning_rate": 8.780440582327005e-06,
+      "loss": 5.442026519775391,
+      "step": 655
+    },
+    {
+      "epoch": 1.2293706293706295,
+      "grad_norm": 13.872124671936035,
+      "learning_rate": 8.75372425912637e-06,
+      "loss": 5.736603546142578,
+      "step": 660
+    },
+    {
+      "epoch": 1.2386946386946387,
+      "grad_norm": 13.677570343017578,
+      "learning_rate": 8.726760059044542e-06,
+      "loss": 5.951699447631836,
+      "step": 665
+    },
+    {
+      "epoch": 1.248018648018648,
+      "grad_norm": 15.023405075073242,
+      "learning_rate": 8.699549762660318e-06,
+      "loss": 5.936520004272461,
+      "step": 670
+    },
+    {
+      "epoch": 1.2573426573426574,
+      "grad_norm": 12.804817199707031,
+      "learning_rate": 8.672095166803445e-06,
+      "loss": 5.778074645996094,
+      "step": 675
+    },
+    {
+      "epoch": 1.2666666666666666,
+      "grad_norm": 13.148968696594238,
+      "learning_rate": 8.644398084435959e-06,
+      "loss": 5.7417655944824215,
+      "step": 680
+    },
+    {
+      "epoch": 1.275990675990676,
+      "grad_norm": 14.131410598754883,
+      "learning_rate": 8.616460344532483e-06,
+      "loss": 5.58878173828125,
+      "step": 685
+    },
+    {
+      "epoch": 1.2853146853146853,
+      "grad_norm": 14.100456237792969,
+      "learning_rate": 8.588283791959437e-06,
+      "loss": 5.546084976196289,
+      "step": 690
+    },
+    {
+      "epoch": 1.2946386946386945,
+      "grad_norm": 13.94593620300293,
+      "learning_rate": 8.559870287353214e-06,
+      "loss": 6.0555156707763675,
+      "step": 695
+    },
+    {
+      "epoch": 1.303962703962704,
+      "grad_norm": 18.369823455810547,
+      "learning_rate": 8.531221706997316e-06,
+      "loss": 5.930012512207031,
+      "step": 700
+    },
+    {
+      "epoch": 1.303962703962704,
+      "eval_com_accuracy": 0.24,
+      "eval_rew_accuracy": 0.364,
+      "step": 700
+    },
+    {
+      "epoch": 1.3132867132867134,
+      "grad_norm": 14.022618293762207,
+      "learning_rate": 8.502339942698463e-06,
+      "loss": 5.831372833251953,
+      "step": 705
+    },
+    {
+      "epoch": 1.3226107226107227,
+      "grad_norm": 12.844898223876953,
+      "learning_rate": 8.473226901661643e-06,
+      "loss": 5.503532409667969,
+      "step": 710
+    },
+    {
+      "epoch": 1.3319347319347319,
+      "grad_norm": 13.991181373596191,
+      "learning_rate": 8.443884506364192e-06,
+      "loss": 5.344602966308594,
+      "step": 715
+    },
+    {
+      "epoch": 1.3412587412587413,
+      "grad_norm": 13.290801048278809,
+      "learning_rate": 8.414314694428842e-06,
+      "loss": 5.7142791748046875,
+      "step": 720
+    },
+    {
+      "epoch": 1.3505827505827506,
+      "grad_norm": 13.986117362976074,
+      "learning_rate": 8.384519418495755e-06,
+      "loss": 5.509880065917969,
+      "step": 725
+    },
+    {
+      "epoch": 1.3599067599067598,
+      "grad_norm": 17.483638763427734,
+      "learning_rate": 8.354500646093592e-06,
+      "loss": 5.769047546386719,
+      "step": 730
+    },
+    {
+      "epoch": 1.3692307692307693,
+      "grad_norm": 12.383395195007324,
+      "learning_rate": 8.324260359509594e-06,
+      "loss": 5.693521881103516,
+      "step": 735
+    },
+    {
+      "epoch": 1.3785547785547785,
+      "grad_norm": 12.225541114807129,
+      "learning_rate": 8.29380055565866e-06,
+      "loss": 5.618255233764648,
+      "step": 740
+    },
+    {
+      "epoch": 1.387878787878788,
+      "grad_norm": 11.737737655639648,
+      "learning_rate": 8.263123245951504e-06,
+      "loss": 5.702220916748047,
+      "step": 745
+    },
+    {
+      "epoch": 1.3972027972027972,
+      "grad_norm": 14.235527992248535,
+      "learning_rate": 8.232230456161819e-06,
+      "loss": 5.744013977050781,
+      "step": 750
+    },
+    {
+      "epoch": 1.3972027972027972,
+      "eval_com_accuracy": 0.24,
+      "eval_rew_accuracy": 0.368,
+      "step": 750
+    },
+    {
+      "epoch": 1.4065268065268066,
+      "grad_norm": 12.433022499084473,
+      "learning_rate": 8.201124226292505e-06,
+      "loss": 5.6122283935546875,
+      "step": 755
+    },
+    {
+      "epoch": 1.4158508158508158,
+      "grad_norm": 13.874029159545898,
+      "learning_rate": 8.169806610440966e-06,
+      "loss": 5.530184936523438,
+      "step": 760
+    },
+    {
+      "epoch": 1.425174825174825,
+      "grad_norm": 12.640862464904785,
+      "learning_rate": 8.138279676663458e-06,
+      "loss": 5.602608489990234,
+      "step": 765
+    },
+    {
+      "epoch": 1.4344988344988345,
+      "grad_norm": 13.672120094299316,
+      "learning_rate": 8.106545506838533e-06,
+      "loss": 5.971489715576172,
+      "step": 770
+    },
+    {
+      "epoch": 1.4438228438228438,
+      "grad_norm": 12.973994255065918,
+      "learning_rate": 8.074606196529554e-06,
+      "loss": 5.967403411865234,
+      "step": 775
+    },
+    {
+      "epoch": 1.4531468531468532,
+      "grad_norm": 12.31808090209961,
+      "learning_rate": 8.042463854846325e-06,
+      "loss": 5.71468505859375,
+      "step": 780
+    },
+    {
+      "epoch": 1.4624708624708624,
+      "grad_norm": 12.416245460510254,
+      "learning_rate": 8.010120604305806e-06,
+      "loss": 5.836771392822266,
+      "step": 785
+    },
+    {
+      "epoch": 1.471794871794872,
+      "grad_norm": 13.020674705505371,
+      "learning_rate": 7.977578580691963e-06,
+      "loss": 5.71131591796875,
+      "step": 790
+    },
+    {
+      "epoch": 1.4811188811188811,
+      "grad_norm": 13.946771621704102,
+      "learning_rate": 7.944839932914718e-06,
+      "loss": 5.564648818969727,
+      "step": 795
+    },
+    {
+      "epoch": 1.4904428904428904,
+      "grad_norm": 12.287398338317871,
+      "learning_rate": 7.91190682286806e-06,
+      "loss": 5.753592300415039,
+      "step": 800
+    },
+    {
+      "epoch": 1.4904428904428904,
+      "eval_com_accuracy": 0.236,
+      "eval_rew_accuracy": 0.384,
+      "step": 800
+    },
+    {
+      "epoch": 1.4997668997668998,
+      "grad_norm": 20.467731475830078,
+      "learning_rate": 7.878781425287277e-06,
+      "loss": 5.64927978515625,
+      "step": 805
+    },
+    {
+      "epoch": 1.509090909090909,
+      "grad_norm": 13.811100006103516,
+      "learning_rate": 7.84546592760535e-06,
+      "loss": 5.415856170654297,
+      "step": 810
+    },
+    {
+      "epoch": 1.5184149184149183,
+      "grad_norm": 12.4107666015625,
+      "learning_rate": 7.811962529808499e-06,
+      "loss": 5.767454147338867,
+      "step": 815
+    },
+    {
+      "epoch": 1.5277389277389277,
+      "grad_norm": 14.148603439331055,
+      "learning_rate": 7.778273444290921e-06,
+      "loss": 5.463053131103516,
+      "step": 820
+    },
+    {
+      "epoch": 1.5370629370629372,
+      "grad_norm": 13.375722885131836,
+      "learning_rate": 7.744400895708683e-06,
+      "loss": 5.625373077392578,
+      "step": 825
+    },
+    {
+      "epoch": 1.5463869463869464,
+      "grad_norm": 12.991719245910645,
+      "learning_rate": 7.710347120832821e-06,
+      "loss": 5.6271202087402346,
+      "step": 830
+    },
+    {
+      "epoch": 1.5557109557109556,
+      "grad_norm": 12.906146049499512,
+      "learning_rate": 7.676114368401635e-06,
+      "loss": 5.5160057067871096,
+      "step": 835
+    },
+    {
+      "epoch": 1.565034965034965,
+      "grad_norm": 12.344596862792969,
+      "learning_rate": 7.641704898972194e-06,
+      "loss": 5.506198120117188,
+      "step": 840
+    },
+    {
+      "epoch": 1.5743589743589743,
+      "grad_norm": 13.285745620727539,
+      "learning_rate": 7.607120984771058e-06,
+      "loss": 5.7151634216308596,
+      "step": 845
+    },
+    {
+      "epoch": 1.5836829836829835,
+      "grad_norm": 12.432035446166992,
+      "learning_rate": 7.572364909544235e-06,
+      "loss": 5.852275085449219,
+      "step": 850
+    },
+    {
+      "epoch": 1.5836829836829835,
+      "eval_com_accuracy": 0.228,
+      "eval_rew_accuracy": 0.332,
+      "step": 850
+    },
+    {
+      "epoch": 1.593006993006993,
+      "grad_norm": 12.581634521484375,
+      "learning_rate": 7.537438968406372e-06,
+      "loss": 5.735688018798828,
+      "step": 855
+    },
+    {
+      "epoch": 1.6023310023310025,
+      "grad_norm": 13.161237716674805,
+      "learning_rate": 7.502345467689202e-06,
+      "loss": 5.648350524902344,
+      "step": 860
+    },
+    {
+      "epoch": 1.6116550116550117,
+      "grad_norm": 12.48270034790039,
+      "learning_rate": 7.4670867247892346e-06,
+      "loss": 5.635168075561523,
+      "step": 865
+    },
+    {
+      "epoch": 1.620979020979021,
+      "grad_norm": 13.835477828979492,
+      "learning_rate": 7.431665068014737e-06,
+      "loss": 5.548571014404297,
+      "step": 870
+    },
+    {
+      "epoch": 1.6303030303030304,
+      "grad_norm": 12.403617858886719,
+      "learning_rate": 7.396082836431981e-06,
+      "loss": 5.7486827850341795,
+      "step": 875
+    },
+    {
+      "epoch": 1.6396270396270396,
+      "grad_norm": 13.730851173400879,
+      "learning_rate": 7.3603423797107845e-06,
+      "loss": 5.288154602050781,
+      "step": 880
+    },
+    {
+      "epoch": 1.6489510489510488,
+      "grad_norm": 14.688888549804688,
+      "learning_rate": 7.324446057969346e-06,
+      "loss": 5.648311614990234,
+      "step": 885
+    },
+    {
+      "epoch": 1.6582750582750583,
+      "grad_norm": 13.311734199523926,
+      "learning_rate": 7.288396241618401e-06,
+      "loss": 5.636775207519531,
+      "step": 890
+    },
+    {
+      "epoch": 1.6675990675990677,
+      "grad_norm": 12.776606559753418,
+      "learning_rate": 7.252195311204689e-06,
+      "loss": 5.666929626464844,
+      "step": 895
+    },
+    {
+      "epoch": 1.676923076923077,
+      "grad_norm": 14.130080223083496,
+      "learning_rate": 7.215845657253755e-06,
+      "loss": 5.534748840332031,
+      "step": 900
+    },
+    {
+      "epoch": 1.676923076923077,
+      "eval_com_accuracy": 0.264,
+      "eval_rew_accuracy": 0.364,
+      "step": 900
+    },
+    {
+      "epoch": 1.6862470862470862,
+      "grad_norm": 14.159326553344727,
+      "learning_rate": 7.1793496801120885e-06,
+      "loss": 5.652375793457031,
+      "step": 905
+    },
+    {
+      "epoch": 1.6955710955710956,
+      "grad_norm": 13.113655090332031,
+      "learning_rate": 7.1427097897886225e-06,
+      "loss": 5.383832168579102,
+      "step": 910
+    },
+    {
+      "epoch": 1.7048951048951049,
+      "grad_norm": 14.203001022338867,
+      "learning_rate": 7.105928405795584e-06,
+      "loss": 5.6499076843261715,
+      "step": 915
+    },
+    {
+      "epoch": 1.714219114219114,
+      "grad_norm": 12.1084623336792,
+      "learning_rate": 7.069007956988718e-06,
+      "loss": 5.560025787353515,
+      "step": 920
+    },
+    {
+      "epoch": 1.7235431235431236,
+      "grad_norm": 13.819426536560059,
+      "learning_rate": 7.031950881406913e-06,
+      "loss": 5.535141372680664,
+      "step": 925
+    },
+    {
+      "epoch": 1.732867132867133,
+      "grad_norm": 17.341110229492188,
+      "learning_rate": 6.994759626111189e-06,
+      "loss": 5.697111511230469,
+      "step": 930
+    },
+    {
+      "epoch": 1.7421911421911422,
+      "grad_norm": 16.404672622680664,
+      "learning_rate": 6.957436647023117e-06,
+      "loss": 5.4155632019042965,
+      "step": 935
+    },
+    {
+      "epoch": 1.7515151515151515,
+      "grad_norm": 13.115015029907227,
+      "learning_rate": 6.919984408762632e-06,
+      "loss": 5.758349990844726,
+      "step": 940
+    },
+    {
+      "epoch": 1.760839160839161,
+      "grad_norm": 13.850481986999512,
+      "learning_rate": 6.882405384485294e-06,
+      "loss": 5.69927864074707,
+      "step": 945
+    },
+    {
+      "epoch": 1.7701631701631702,
+      "grad_norm": 12.921819686889648,
+      "learning_rate": 6.844702055718964e-06,
+      "loss": 5.631483459472657,
+      "step": 950
+    },
+    {
+      "epoch": 1.7701631701631702,
+      "eval_com_accuracy": 0.26,
+      "eval_rew_accuracy": 0.352,
+      "step": 950
+    },
+    {
+      "epoch": 1.7794871794871794,
+      "grad_norm": 13.401888847351074,
+      "learning_rate": 6.806876912199945e-06,
+      "loss": 5.824296569824218,
+      "step": 955
+    },
+    {
+      "epoch": 1.7888111888111888,
+      "grad_norm": 13.481993675231934,
+      "learning_rate": 6.768932451708557e-06,
+      "loss": 5.752721786499023,
+      "step": 960
+    },
+    {
+      "epoch": 1.7981351981351983,
+      "grad_norm": 14.022424697875977,
+      "learning_rate": 6.730871179904218e-06,
+      "loss": 5.453369522094727,
+      "step": 965
+    },
+    {
+      "epoch": 1.8074592074592073,
+      "grad_norm": 13.432122230529785,
+      "learning_rate": 6.692695610159966e-06,
+      "loss": 5.664297103881836,
+      "step": 970
+    },
+    {
+      "epoch": 1.8167832167832167,
+      "grad_norm": 14.292946815490723,
+      "learning_rate": 6.6544082633964955e-06,
+      "loss": 5.576699066162109,
+      "step": 975
+    },
+    {
+      "epoch": 1.8261072261072262,
+      "grad_norm": 14.256867408752441,
+      "learning_rate": 6.6160116679156874e-06,
+      "loss": 5.551004791259766,
+      "step": 980
+    },
+    {
+      "epoch": 1.8354312354312354,
+      "grad_norm": 14.11919116973877,
+      "learning_rate": 6.577508359233653e-06,
+      "loss": 5.580959320068359,
+      "step": 985
+    },
+    {
+      "epoch": 1.8447552447552447,
+      "grad_norm": 14.006019592285156,
+      "learning_rate": 6.538900879913301e-06,
+      "loss": 5.443265533447265,
+      "step": 990
+    },
+    {
+      "epoch": 1.8540792540792541,
+      "grad_norm": 11.976597785949707,
+      "learning_rate": 6.500191779396439e-06,
+      "loss": 5.410481643676758,
+      "step": 995
+    },
+    {
+      "epoch": 1.8634032634032636,
+      "grad_norm": 14.439790725708008,
+      "learning_rate": 6.461383613835427e-06,
+      "loss": 5.3311511993408205,
+      "step": 1000
+    },
+    {
+      "epoch": 1.8634032634032636,
+      "eval_com_accuracy": 0.24,
+      "eval_rew_accuracy": 0.356,
+      "step": 1000
+    },
+    {
+      "epoch": 1.8727272727272726,
+      "grad_norm": 15.29392147064209,
+      "learning_rate": 6.4224789459243705e-06,
+      "loss": 5.477576446533203,
+      "step": 1005
+    },
+    {
+      "epoch": 1.882051282051282,
+      "grad_norm": 14.032716751098633,
+      "learning_rate": 6.383480344729903e-06,
+      "loss": 5.458187103271484,
+      "step": 1010
+    },
+    {
+      "epoch": 1.8913752913752915,
+      "grad_norm": 11.861150741577148,
+      "learning_rate": 6.344390385521534e-06,
+      "loss": 5.806562423706055,
+      "step": 1015
+    },
+    {
+      "epoch": 1.9006993006993007,
+      "grad_norm": 14.015438079833984,
+      "learning_rate": 6.305211649601595e-06,
+      "loss": 5.2158203125,
+      "step": 1020
+    },
+    {
+      "epoch": 1.91002331002331,
+      "grad_norm": 15.672555923461914,
+      "learning_rate": 6.265946724134782e-06,
+      "loss": 5.469319915771484,
+      "step": 1025
+    },
+    {
+      "epoch": 1.9193473193473194,
+      "grad_norm": 13.891338348388672,
+      "learning_rate": 6.226598201977299e-06,
+      "loss": 5.495826721191406,
+      "step": 1030
+    },
+    {
+      "epoch": 1.9286713286713286,
+      "grad_norm": 12.441986083984375,
+      "learning_rate": 6.187168681505666e-06,
+      "loss": 5.466391372680664,
+      "step": 1035
+    },
+    {
+      "epoch": 1.9379953379953379,
+      "grad_norm": 13.905447006225586,
+      "learning_rate": 6.1476607664451105e-06,
+      "loss": 5.558423233032227,
+      "step": 1040
+    },
+    {
+      "epoch": 1.9473193473193473,
+      "grad_norm": 14.817802429199219,
+      "learning_rate": 6.1080770656976444e-06,
+      "loss": 5.50462646484375,
+      "step": 1045
+    },
+    {
+      "epoch": 1.9566433566433568,
+      "grad_norm": 14.293425559997559,
+      "learning_rate": 6.068420193169779e-06,
+      "loss": 5.567475128173828,
+      "step": 1050
+    },
+    {
+      "epoch": 1.9566433566433568,
+      "eval_com_accuracy": 0.26,
+      "eval_rew_accuracy": 0.384,
+      "step": 1050
+    },
+    {
+      "epoch": 1.965967365967366,
+      "grad_norm": 13.285664558410645,
+      "learning_rate": 6.0286927675999205e-06,
+      "loss": 5.3831031799316404,
+      "step": 1055
+    },
+    {
+      "epoch": 1.9752913752913752,
+      "grad_norm": 12.634699821472168,
+      "learning_rate": 5.98889741238544e-06,
+      "loss": 5.4307403564453125,
+      "step": 1060
+    },
+    {
+      "epoch": 1.9846153846153847,
+      "grad_norm": 13.824172019958496,
+      "learning_rate": 5.949036755409432e-06,
+      "loss": 5.559865570068359,
+      "step": 1065
+    },
+    {
+      "epoch": 1.993939393939394,
+      "grad_norm": 14.028116226196289,
+      "learning_rate": 5.909113428867195e-06,
+      "loss": 5.506551361083984,
+      "step": 1070
+    },
+    {
+      "epoch": 2.001864801864802,
+      "grad_norm": 15.495360374450684,
+      "learning_rate": 5.869130069092401e-06,
+      "loss": 4.655789947509765,
+      "step": 1075
+    },
+    {
+      "epoch": 2.011188811188811,
+      "grad_norm": 14.094182014465332,
+      "learning_rate": 5.829089316383018e-06,
+      "loss": 5.403445434570313,
+      "step": 1080
+    },
+    {
+      "epoch": 2.0205128205128204,
+      "grad_norm": 13.709051132202148,
+      "learning_rate": 5.7889938148269445e-06,
+      "loss": 5.5024055480957035,
+      "step": 1085
+    },
+    {
+      "epoch": 2.02983682983683,
+      "grad_norm": 13.718189239501953,
+      "learning_rate": 5.748846212127421e-06,
+      "loss": 5.404336547851562,
+      "step": 1090
+    },
+    {
+      "epoch": 2.0391608391608393,
+      "grad_norm": 14.198376655578613,
+      "learning_rate": 5.708649159428181e-06,
+      "loss": 5.498197937011719,
+      "step": 1095
+    },
+    {
+      "epoch": 2.0484848484848484,
+      "grad_norm": 12.810367584228516,
+      "learning_rate": 5.668405311138382e-06,
+      "loss": 5.639860534667969,
+      "step": 1100
+    },
+    {
+      "epoch": 2.0484848484848484,
+      "eval_com_accuracy": 0.264,
+      "eval_rew_accuracy": 0.392,
+      "step": 1100
+    },
+    {
+      "epoch": 2.057808857808858,
+      "grad_norm": 13.649062156677246,
+      "learning_rate": 5.628117324757326e-06,
+      "loss": 5.567802429199219,
+      "step": 1105
+    },
+    {
+      "epoch": 2.0671328671328673,
+      "grad_norm": 15.412623405456543,
+      "learning_rate": 5.587787860698975e-06,
+      "loss": 5.585782623291015,
+      "step": 1110
+    },
+    {
+      "epoch": 2.0764568764568763,
+      "grad_norm": 13.465250968933105,
+      "learning_rate": 5.547419582116259e-06,
+      "loss": 5.357003021240234,
+      "step": 1115
+    },
+    {
+      "epoch": 2.0857808857808857,
+      "grad_norm": 14.019501686096191,
+      "learning_rate": 5.507015154725226e-06,
+      "loss": 5.621536636352539,
+      "step": 1120
+    },
+    {
+      "epoch": 2.095104895104895,
+      "grad_norm": 15.479727745056152,
+      "learning_rate": 5.466577246629006e-06,
+      "loss": 5.355072784423828,
+      "step": 1125
+    },
+    {
+      "epoch": 2.1044289044289046,
+      "grad_norm": 13.079503059387207,
+      "learning_rate": 5.426108528141627e-06,
+      "loss": 5.450039672851562,
+      "step": 1130
+    },
+    {
+      "epoch": 2.1137529137529136,
+      "grad_norm": 16.310596466064453,
+      "learning_rate": 5.385611671611676e-06,
+      "loss": 5.359495162963867,
+      "step": 1135
+    },
+    {
+      "epoch": 2.123076923076923,
+      "grad_norm": 15.107529640197754,
+      "learning_rate": 5.345089351245834e-06,
+      "loss": 5.5543464660644535,
+      "step": 1140
+    },
+    {
+      "epoch": 2.1324009324009325,
+      "grad_norm": 14.484854698181152,
+      "learning_rate": 5.304544242932288e-06,
+      "loss": 5.393303680419922,
+      "step": 1145
+    },
+    {
+      "epoch": 2.1417249417249415,
+      "grad_norm": 13.286919593811035,
+      "learning_rate": 5.26397902406402e-06,
+      "loss": 5.56396369934082,
+      "step": 1150
+    },
+    {
+      "epoch": 2.1417249417249415,
+      "eval_com_accuracy": 0.272,
+      "eval_rew_accuracy": 0.372,
+      "step": 1150
+    },
+    {
+      "epoch": 2.151048951048951,
+      "grad_norm": 14.213571548461914,
+      "learning_rate": 5.223396373362013e-06,
+      "loss": 5.63990478515625,
+      "step": 1155
+    },
+    {
+      "epoch": 2.1603729603729604,
+      "grad_norm": 14.445123672485352,
+      "learning_rate": 5.182798970698361e-06,
+      "loss": 5.836676788330078,
+      "step": 1160
+    },
+    {
+      "epoch": 2.16969696969697,
+      "grad_norm": 14.63541030883789,
+      "learning_rate": 5.142189496919302e-06,
+      "loss": 5.595640182495117,
+      "step": 1165
+    },
+    {
+      "epoch": 2.179020979020979,
+      "grad_norm": 13.398268699645996,
+      "learning_rate": 5.101570633668185e-06,
+      "loss": 5.657070159912109,
+      "step": 1170
+    },
+    {
+      "epoch": 2.1883449883449884,
+      "grad_norm": 13.70476245880127,
+      "learning_rate": 5.060945063208399e-06,
+      "loss": 5.337051010131836,
+      "step": 1175
+    },
+    {
+      "epoch": 2.197668997668998,
+      "grad_norm": 13.812256813049316,
+      "learning_rate": 5.02031546824624e-06,
+      "loss": 5.393893051147461,
+      "step": 1180
+    },
+    {
+      "epoch": 2.206993006993007,
+      "grad_norm": 14.151784896850586,
+      "learning_rate": 4.979684531753761e-06,
+      "loss": 5.489350128173828,
+      "step": 1185
+    },
+    {
+      "epoch": 2.2163170163170163,
+      "grad_norm": 14.830500602722168,
+      "learning_rate": 4.9390549367916004e-06,
+      "loss": 5.4119518280029295,
+      "step": 1190
+    },
+    {
+      "epoch": 2.2256410256410257,
+      "grad_norm": 14.537983894348145,
+      "learning_rate": 4.898429366331815e-06,
+      "loss": 5.393406677246094,
+      "step": 1195
+    },
+    {
+      "epoch": 2.234965034965035,
+      "grad_norm": 12.427031517028809,
+      "learning_rate": 4.857810503080701e-06,
+      "loss": 5.359439849853516,
+      "step": 1200
+    },
+    {
+      "epoch": 2.234965034965035,
+      "eval_com_accuracy": 0.256,
+      "eval_rew_accuracy": 0.368,
+      "step": 1200
+    },
+    {
+      "epoch": 2.244289044289044,
+      "grad_norm": 13.743568420410156,
+      "learning_rate": 4.81720102930164e-06,
+      "loss": 5.248456954956055,
+      "step": 1205
+    },
+    {
+      "epoch": 2.2536130536130536,
+      "grad_norm": 15.313255310058594,
+      "learning_rate": 4.776603626637988e-06,
+      "loss": 5.48780517578125,
+      "step": 1210
+    },
+    {
+      "epoch": 2.262937062937063,
+      "grad_norm": 14.26356029510498,
+      "learning_rate": 4.736020975935981e-06,
+      "loss": 5.3895729064941404,
+      "step": 1215
+    },
+    {
+      "epoch": 2.272261072261072,
+      "grad_norm": 14.363275527954102,
+      "learning_rate": 4.695455757067712e-06,
+      "loss": 5.294971084594726,
+      "step": 1220
+    },
+    {
+      "epoch": 2.2815850815850816,
+      "grad_norm": 14.772239685058594,
+      "learning_rate": 4.6549106487541666e-06,
+      "loss": 5.744273376464844,
+      "step": 1225
+    },
+    {
+      "epoch": 2.290909090909091,
+      "grad_norm": 13.530016899108887,
+      "learning_rate": 4.614388328388327e-06,
+      "loss": 5.28902587890625,
+      "step": 1230
+    },
+    {
+      "epoch": 2.3002331002331005,
+      "grad_norm": 13.482331275939941,
+      "learning_rate": 4.573891471858375e-06,
+      "loss": 5.445513153076172,
+      "step": 1235
+    },
+    {
+      "epoch": 2.3095571095571095,
+      "grad_norm": 13.693358421325684,
+      "learning_rate": 4.533422753370995e-06,
+      "loss": 5.312195205688477,
+      "step": 1240
+    },
+    {
+      "epoch": 2.318881118881119,
+      "grad_norm": 14.169803619384766,
+      "learning_rate": 4.492984845274774e-06,
+      "loss": 5.5245819091796875,
+      "step": 1245
+    },
+    {
+      "epoch": 2.3282051282051284,
+      "grad_norm": 26.758739471435547,
+      "learning_rate": 4.4525804178837425e-06,
+      "loss": 5.587922668457031,
+      "step": 1250
+    },
+    {
+      "epoch": 2.3282051282051284,
+      "eval_com_accuracy": 0.26,
+      "eval_rew_accuracy": 0.336,
+      "step": 1250
+    },
+    {
+      "epoch": 2.3375291375291374,
+      "grad_norm": 15.485888481140137,
+      "learning_rate": 4.412212139301027e-06,
+      "loss": 5.402701187133789,
+      "step": 1255
+    },
+    {
+      "epoch": 2.346853146853147,
+      "grad_norm": 16.002370834350586,
+      "learning_rate": 4.371882675242674e-06,
+      "loss": 5.409712219238282,
+      "step": 1260
+    },
+    {
+      "epoch": 2.3561771561771563,
+      "grad_norm": 16.42519760131836,
+      "learning_rate": 4.331594688861619e-06,
+      "loss": 5.624014282226563,
+      "step": 1265
+    },
+    {
+      "epoch": 2.3655011655011657,
+      "grad_norm": 16.416702270507812,
+      "learning_rate": 4.291350840571821e-06,
+      "loss": 5.3733673095703125,
+      "step": 1270
+    },
+    {
+      "epoch": 2.3748251748251747,
+      "grad_norm": 13.191758155822754,
+      "learning_rate": 4.251153787872579e-06,
+      "loss": 5.316907501220703,
+      "step": 1275
+    },
+    {
+      "epoch": 2.384149184149184,
+      "grad_norm": 14.480209350585938,
+      "learning_rate": 4.211006185173056e-06,
+      "loss": 5.500685501098633,
+      "step": 1280
+    },
+    {
+      "epoch": 2.3934731934731936,
+      "grad_norm": 15.625597953796387,
+      "learning_rate": 4.170910683616985e-06,
+      "loss": 5.321009826660156,
+      "step": 1285
+    },
+    {
+      "epoch": 2.4027972027972027,
+      "grad_norm": 13.146830558776855,
+      "learning_rate": 4.130869930907599e-06,
+      "loss": 5.352607345581054,
+      "step": 1290
+    },
+    {
+      "epoch": 2.412121212121212,
+      "grad_norm": 13.852392196655273,
+      "learning_rate": 4.090886571132807e-06,
+      "loss": 5.265444946289063,
+      "step": 1295
+    },
+    {
+      "epoch": 2.4214452214452216,
+      "grad_norm": 15.4178466796875,
+      "learning_rate": 4.050963244590571e-06,
+      "loss": 5.396755599975586,
+      "step": 1300
+    },
+    {
+      "epoch": 2.4214452214452216,
+      "eval_com_accuracy": 0.26,
+      "eval_rew_accuracy": 0.372,
+      "step": 1300
+    },
+    {
+      "epoch": 2.430769230769231,
+      "grad_norm": 14.469849586486816,
+      "learning_rate": 4.011102587614563e-06,
+      "loss": 5.381956100463867,
+      "step": 1305
+    },
+    {
+      "epoch": 2.44009324009324,
+      "grad_norm": 14.326638221740723,
+      "learning_rate": 3.97130723240008e-06,
+      "loss": 5.4776451110839846,
+      "step": 1310
+    },
+    {
+      "epoch": 2.4494172494172495,
+      "grad_norm": 13.75566577911377,
+      "learning_rate": 3.9315798068302214e-06,
+      "loss": 5.416980743408203,
+      "step": 1315
+    },
+    {
+      "epoch": 2.458741258741259,
+      "grad_norm": 13.778778076171875,
+      "learning_rate": 3.891922934302356e-06,
+      "loss": 5.254080200195313,
+      "step": 1320
+    },
+    {
+      "epoch": 2.468065268065268,
+      "grad_norm": 13.458685874938965,
+      "learning_rate": 3.852339233554891e-06,
+      "loss": 5.272119140625,
+      "step": 1325
+    },
+    {
+      "epoch": 2.4773892773892774,
+      "grad_norm": 15.419684410095215,
+      "learning_rate": 3.812831318494335e-06,
+      "loss": 5.448992538452148,
+      "step": 1330
+    },
+    {
+      "epoch": 2.486713286713287,
+      "grad_norm": 13.448263168334961,
+      "learning_rate": 3.773401798022701e-06,
+      "loss": 5.6390220642089846,
+      "step": 1335
+    },
+    {
+      "epoch": 2.496037296037296,
+      "grad_norm": 18.160816192626953,
+      "learning_rate": 3.7340532758652217e-06,
+      "loss": 5.633776092529297,
+      "step": 1340
+    },
+    {
+      "epoch": 2.5053613053613053,
+      "grad_norm": 20.26373863220215,
+      "learning_rate": 3.6947883503984037e-06,
+      "loss": 5.412459182739258,
+      "step": 1345
+    },
+    {
+      "epoch": 2.5146853146853148,
+      "grad_norm": 15.404874801635742,
+      "learning_rate": 3.655609614478467e-06,
+      "loss": 5.253081893920898,
+      "step": 1350
+    },
+    {
+      "epoch": 2.5146853146853148,
+      "eval_com_accuracy": 0.26,
+      "eval_rew_accuracy": 0.36,
+      "step": 1350
+    },
+    {
+      "epoch": 2.5240093240093238,
+      "grad_norm": 14.836410522460938,
+      "learning_rate": 3.6165196552701e-06,
+      "loss": 5.615703582763672,
+      "step": 1355
+    },
+    {
+      "epoch": 2.533333333333333,
+      "grad_norm": 13.233349800109863,
+      "learning_rate": 3.577521054075631e-06,
+      "loss": 5.440980911254883,
+      "step": 1360
+    },
+    {
+      "epoch": 2.5426573426573427,
+      "grad_norm": 14.456931114196777,
+      "learning_rate": 3.538616386164575e-06,
+      "loss": 5.664167022705078,
+      "step": 1365
+    },
+    {
+      "epoch": 2.551981351981352,
+      "grad_norm": 14.075837135314941,
+      "learning_rate": 3.4998082206035606e-06,
+      "loss": 5.418301010131836,
+      "step": 1370
+    },
+    {
+      "epoch": 2.5613053613053616,
+      "grad_norm": 13.708165168762207,
+      "learning_rate": 3.4610991200867006e-06,
+      "loss": 5.395311737060547,
+      "step": 1375
+    },
+    {
+      "epoch": 2.5706293706293706,
+      "grad_norm": 13.784794807434082,
+      "learning_rate": 3.4224916407663484e-06,
+      "loss": 5.168692016601563,
+      "step": 1380
+    },
+    {
+      "epoch": 2.57995337995338,
+      "grad_norm": 13.721598625183105,
+      "learning_rate": 3.3839883320843125e-06,
+      "loss": 5.642356109619141,
+      "step": 1385
+    },
+    {
+      "epoch": 2.589277389277389,
+      "grad_norm": 14.451462745666504,
+      "learning_rate": 3.3455917366035058e-06,
+      "loss": 5.229490661621094,
+      "step": 1390
+    },
+    {
+      "epoch": 2.5986013986013985,
+      "grad_norm": 13.549321174621582,
+      "learning_rate": 3.307304389840036e-06,
+      "loss": 5.413081359863281,
+      "step": 1395
+    },
+    {
+      "epoch": 2.607925407925408,
+      "grad_norm": 14.325213432312012,
+      "learning_rate": 3.2691288200957826e-06,
+      "loss": 5.3188224792480465,
+      "step": 1400
+    },
+    {
+      "epoch": 2.607925407925408,
+      "eval_com_accuracy": 0.26,
+      "eval_rew_accuracy": 0.348,
+      "step": 1400
+    },
+    {
+      "epoch": 2.6172494172494174,
+      "grad_norm": 13.765854835510254,
+      "learning_rate": 3.2310675482914444e-06,
+      "loss": 5.554658508300781,
+      "step": 1405
+    },
+    {
+      "epoch": 2.626573426573427,
+      "grad_norm": 14.936039924621582,
+      "learning_rate": 3.1931230878000586e-06,
+      "loss": 5.512041473388672,
+      "step": 1410
+    },
+    {
+      "epoch": 2.635897435897436,
+      "grad_norm": 16.414018630981445,
+      "learning_rate": 3.155297944281036e-06,
+      "loss": 5.256567001342773,
+      "step": 1415
+    },
+    {
+      "epoch": 2.6452214452214453,
+      "grad_norm": 13.871015548706055,
+      "learning_rate": 3.1175946155147064e-06,
+      "loss": 5.480235290527344,
+      "step": 1420
+    },
+    {
+      "epoch": 2.6545454545454543,
+      "grad_norm": 13.673002243041992,
+      "learning_rate": 3.0800155912373696e-06,
+      "loss": 5.272347259521484,
+      "step": 1425
+    },
+    {
+      "epoch": 2.6638694638694638,
+      "grad_norm": 15.32669734954834,
+      "learning_rate": 3.042563352976884e-06,
+      "loss": 5.1510368347167965,
+      "step": 1430
+    },
+    {
+      "epoch": 2.6731934731934732,
+      "grad_norm": 13.43411922454834,
+      "learning_rate": 3.005240373888812e-06,
+      "loss": 5.218346405029297,
+      "step": 1435
+    },
+    {
+      "epoch": 2.6825174825174827,
+      "grad_norm": 13.674540519714355,
+      "learning_rate": 2.9680491185930877e-06,
+      "loss": 5.514698791503906,
+      "step": 1440
+    },
+    {
+      "epoch": 2.6918414918414917,
+      "grad_norm": 14.424562454223633,
+      "learning_rate": 2.9309920430112825e-06,
+      "loss": 5.328684997558594,
+      "step": 1445
+    },
+    {
+      "epoch": 2.701165501165501,
+      "grad_norm": 14.4567232131958,
+      "learning_rate": 2.8940715942044204e-06,
+      "loss": 5.470043182373047,
+      "step": 1450
+    },
+    {
+      "epoch": 2.701165501165501,
+      "eval_com_accuracy": 0.244,
+      "eval_rew_accuracy": 0.364,
+      "step": 1450
+    },
+    {
+      "epoch": 2.7104895104895106,
+      "grad_norm": 14.073286056518555,
+      "learning_rate": 2.8572902102113788e-06,
+      "loss": 5.592926025390625,
+      "step": 1455
+    },
+    {
+      "epoch": 2.7198135198135196,
+      "grad_norm": 15.481256484985352,
+      "learning_rate": 2.820650319887911e-06,
+      "loss": 5.42064094543457,
+      "step": 1460
+    },
+    {
+      "epoch": 2.729137529137529,
+      "grad_norm": 13.80669116973877,
+      "learning_rate": 2.784154342746246e-06,
+      "loss": 5.514656066894531,
+      "step": 1465
+    },
+    {
+      "epoch": 2.7384615384615385,
+      "grad_norm": 15.876877784729004,
+      "learning_rate": 2.747804688795311e-06,
+      "loss": 5.645168304443359,
+      "step": 1470
+    },
+    {
+      "epoch": 2.747785547785548,
+      "grad_norm": 14.138867378234863,
+      "learning_rate": 2.7116037583816e-06,
+      "loss": 5.482178497314453,
+      "step": 1475
+    },
+    {
+      "epoch": 2.757109557109557,
+      "grad_norm": 13.498207092285156,
+      "learning_rate": 2.6755539420306565e-06,
+      "loss": 5.522509384155273,
+      "step": 1480
+    },
+    {
+      "epoch": 2.7664335664335664,
+      "grad_norm": 12.95719051361084,
+      "learning_rate": 2.6396576202892176e-06,
+      "loss": 5.203734970092773,
+      "step": 1485
+    },
+    {
+      "epoch": 2.775757575757576,
+      "grad_norm": 14.70195198059082,
+      "learning_rate": 2.603917163568021e-06,
+      "loss": 5.4792030334472654,
+      "step": 1490
+    },
+    {
+      "epoch": 2.785081585081585,
+      "grad_norm": 16.21117401123047,
+      "learning_rate": 2.5683349319852647e-06,
+      "loss": 5.337226867675781,
+      "step": 1495
+    },
+    {
+      "epoch": 2.7944055944055943,
+      "grad_norm": 15.80867862701416,
+      "learning_rate": 2.5329132752107675e-06,
+      "loss": 5.441407012939453,
+      "step": 1500
+    },
+    {
+      "epoch": 2.7944055944055943,
+      "eval_com_accuracy": 0.252,
+      "eval_rew_accuracy": 0.364,
+      "step": 1500
+    },
+    {
+      "epoch": 2.803729603729604,
+      "grad_norm": 17.126665115356445,
+      "learning_rate": 2.497654532310799e-06,
+      "loss": 5.372947692871094,
+      "step": 1505
+    },
+    {
+      "epoch": 2.8130536130536132,
+      "grad_norm": 14.965190887451172,
+      "learning_rate": 2.4625610315936267e-06,
+      "loss": 5.591331481933594,
+      "step": 1510
+    },
+    {
+      "epoch": 2.8223776223776222,
+      "grad_norm": 13.920053482055664,
+      "learning_rate": 2.427635090455766e-06,
+      "loss": 5.381235504150391,
+      "step": 1515
+    },
+    {
+      "epoch": 2.8317016317016317,
+      "grad_norm": 14.707354545593262,
+      "learning_rate": 2.3928790152289443e-06,
+      "loss": 5.375308990478516,
+      "step": 1520
+    },
+    {
+      "epoch": 2.841025641025641,
+      "grad_norm": 13.846839904785156,
+      "learning_rate": 2.358295101027807e-06,
+      "loss": 5.24262580871582,
+      "step": 1525
+    },
+    {
+      "epoch": 2.85034965034965,
+      "grad_norm": 17.17586898803711,
+      "learning_rate": 2.323885631598366e-06,
+      "loss": 5.628063583374024,
+      "step": 1530
+    },
+    {
+      "epoch": 2.8596736596736596,
+      "grad_norm": 14.309005737304688,
+      "learning_rate": 2.2896528791671807e-06,
+      "loss": 5.600907897949218,
+      "step": 1535
+    },
+    {
+      "epoch": 2.868997668997669,
+      "grad_norm": 18.62157440185547,
+      "learning_rate": 2.2555991042913177e-06,
+      "loss": 5.567650604248047,
+      "step": 1540
+    },
+    {
+      "epoch": 2.8783216783216785,
+      "grad_norm": 14.592391967773438,
+      "learning_rate": 2.221726555709079e-06,
+      "loss": 5.777060699462891,
+      "step": 1545
+    },
+    {
+      "epoch": 2.8876456876456875,
+      "grad_norm": 13.5873441696167,
+      "learning_rate": 2.188037470191502e-06,
+      "loss": 5.495726776123047,
+      "step": 1550
+    },
+    {
+      "epoch": 2.8876456876456875,
+      "eval_com_accuracy": 0.26,
+      "eval_rew_accuracy": 0.344,
+      "step": 1550
+    },
+    {
+      "epoch": 2.896969696969697,
+      "grad_norm": 13.62496280670166,
+      "learning_rate": 2.154534072394651e-06,
+      "loss": 5.313283920288086,
+      "step": 1555
+    },
+    {
+      "epoch": 2.9062937062937064,
+      "grad_norm": 14.277817726135254,
+      "learning_rate": 2.1212185747127235e-06,
+      "loss": 5.38345947265625,
+      "step": 1560
+    },
+    {
+      "epoch": 2.9156177156177154,
+      "grad_norm": 13.139540672302246,
+      "learning_rate": 2.0880931771319395e-06,
+      "loss": 5.48463134765625,
+      "step": 1565
+    },
+    {
+      "epoch": 2.924941724941725,
+      "grad_norm": 15.709213256835938,
+      "learning_rate": 2.055160067085283e-06,
+      "loss": 5.492628479003907,
+      "step": 1570
+    },
+    {
+      "epoch": 2.9342657342657343,
+      "grad_norm": 14.224997520446777,
+      "learning_rate": 2.0224214193080394e-06,
+      "loss": 5.611021423339844,
+      "step": 1575
+    },
+    {
+      "epoch": 2.943589743589744,
+      "grad_norm": 14.523746490478516,
+      "learning_rate": 1.989879395694194e-06,
+      "loss": 5.288306427001953,
+      "step": 1580
+    },
+    {
+      "epoch": 2.952913752913753,
+      "grad_norm": 14.50571346282959,
+      "learning_rate": 1.9575361451536772e-06,
+      "loss": 5.350891876220703,
+      "step": 1585
+    },
+    {
+      "epoch": 2.9622377622377623,
+      "grad_norm": 15.541993141174316,
+      "learning_rate": 1.925393803470447e-06,
+      "loss": 5.3829093933105465,
+      "step": 1590
+    },
+    {
+      "epoch": 2.9715617715617717,
+      "grad_norm": 14.223262786865234,
+      "learning_rate": 1.893454493161468e-06,
+      "loss": 5.267041397094727,
+      "step": 1595
+    },
+    {
+      "epoch": 2.9808857808857807,
+      "grad_norm": 16.193429946899414,
+      "learning_rate": 1.8617203233365427e-06,
+      "loss": 5.335537719726562,
+      "step": 1600
+    },
+    {
+      "epoch": 2.9808857808857807,
+      "eval_com_accuracy": 0.26,
+      "eval_rew_accuracy": 0.356,
+      "step": 1600
+    },
+    {
+      "epoch": 2.99020979020979,
+      "grad_norm": 13.891239166259766,
+      "learning_rate": 1.8301933895590362e-06,
+      "loss": 5.594977569580078,
+      "step": 1605
+    },
+    {
+      "epoch": 2.9995337995337996,
+      "grad_norm": 14.599571228027344,
+      "learning_rate": 1.7988757737074959e-06,
+      "loss": 5.364975357055664,
+      "step": 1610
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 2148,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-1611/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b6ca094dca60606350f327809d196d6faa38b81cdd561be5d924e82822f09231
+size 5201

checkpoint-2148/chat_template.jinja

@@ -0,0 +1,120 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {%- if messages[0].content is string %}
+            {{- messages[0].content }}
+        {%- else %}
+            {%- for content in messages[0].content %}
+                {%- if 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '\n\n' }}
+    {%- endif %}
+    {{- "# 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>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\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" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' }}
+        {%- if messages[0].content is string %}
+            {{- messages[0].content }}
+        {%- else %}
+            {%- for content in messages[0].content %}
+                {%- if 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- for message in messages %}
+    {%- if message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- else %}
+            {%- for content in message.content %}
+                {%- if content.type == 'image' or 'image' in content or 'image_url' in content %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                    {%- if add_vision_id %}Picture {{ image_count.value }}: {% endif -%}
+                    <|vision_start|><|image_pad|><|vision_end|>
+                {%- elif content.type == 'video' or 'video' in content %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                    {%- if add_vision_id %}Video {{ video_count.value }}: {% endif -%}
+                    <|vision_start|><|video_pad|><|vision_end|>
+                {%- elif 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role + '\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- else %}
+            {%- for content_item in message.content %}
+                {%- if 'text' in content_item %}
+                    {{- content_item.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and message.content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- else %}
+            {%- for content in message.content %}
+                {%- if content.type == 'image' or 'image' in content or 'image_url' in content %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                    {%- if add_vision_id %}Picture {{ image_count.value }}: {% endif -%}
+                    <|vision_start|><|image_pad|><|vision_end|>
+                {%- elif content.type == 'video' or 'video' in content %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                    {%- if add_vision_id %}Video {{ video_count.value }}: {% endif -%}
+                    <|vision_start|><|video_pad|><|vision_end|>
+                {%- elif 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}

checkpoint-2148/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6a6c776684566ca7d3f365368ab34e3caca855fd649ad7fdc68dc39174b2ec57
+size 4885870044

checkpoint-2148/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e04d17c502e7b6494cb8f097e1882687b6bd69314ebafff62978a2801e2cc7e9
+size 8527467778

checkpoint-2148/processor_config.json

@@ -0,0 +1,63 @@
+{
+  "image_processor": {
+    "data_format": "channels_first",
+    "do_convert_rgb": true,
+    "do_normalize": true,
+    "do_rescale": true,
+    "do_resize": true,
+    "image_mean": [
+      0.5,
+      0.5,
+      0.5
+    ],
+    "image_processor_type": "Qwen2VLImageProcessorFast",
+    "image_std": [
+      0.5,
+      0.5,
+      0.5
+    ],
+    "merge_size": 2,
+    "patch_size": 16,
+    "resample": 3,
+    "rescale_factor": 0.00392156862745098,
+    "size": {
+      "longest_edge": 16777216,
+      "shortest_edge": 65536
+    },
+    "temporal_patch_size": 2
+  },
+  "processor_class": "Qwen3VLProcessor",
+  "video_processor": {
+    "data_format": "channels_first",
+    "default_to_square": true,
+    "do_convert_rgb": true,
+    "do_normalize": true,
+    "do_rescale": true,
+    "do_resize": true,
+    "do_sample_frames": true,
+    "fps": 2,
+    "image_mean": [
+      0.5,
+      0.5,
+      0.5
+    ],
+    "image_std": [
+      0.5,
+      0.5,
+      0.5
+    ],
+    "max_frames": 768,
+    "merge_size": 2,
+    "min_frames": 4,
+    "patch_size": 16,
+    "resample": 3,
+    "rescale_factor": 0.00392156862745098,
+    "return_metadata": false,
+    "size": {
+      "longest_edge": 25165824,
+      "shortest_edge": 4096
+    },
+    "temporal_patch_size": 2,
+    "video_processor_type": "Qwen3VLVideoProcessor"
+  }
+}

checkpoint-2148/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:659b1cdee2219458dd84ce6a632a595465680b8080e5c44bd600ff97eca8d752
+size 15429

checkpoint-2148/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:86accf27064cdd503053e90476a6bd10de333d4ff0594535ad55ea13a473c91d
+size 15429

checkpoint-2148/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:18ca8d714ef40be035404c1957b5a4dee84e1f43980408393f8aa710552ee6f6
+size 15429

checkpoint-2148/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2cfdebe99e40accc9c9d8f09c63136a14abda997d9b501969ec8e16e9d183179
+size 15429

checkpoint-2148/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f0648e142616ff430dea19a2fe2dd565cd4ea971c97c5d1264c371e82312a23b
+size 1465

checkpoint-2148/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8579e1ca7cc5d82a9e0202eed555529996f4ffe7f563c2979a0290cf3db452d3
+size 11422818

checkpoint-2148/tokenizer_config.json

@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_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|>"
+  ],
+  "is_local": false,
+  "model_max_length": 262144,
+  "pad_token": "<|endoftext|>",
+  "processor_class": "Qwen3VLProcessor",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

checkpoint-2148/trainer_state.json

@@ -0,0 +1,3289 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "eval_steps": 50,
+  "global_step": 2148,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.009324009324009324,
+      "grad_norm": 269.7618103027344,
+      "learning_rate": 1.8604651162790698e-07,
+      "loss": 13.883856201171875,
+      "step": 5
+    },
+    {
+      "epoch": 0.018648018648018648,
+      "grad_norm": 256.5934143066406,
+      "learning_rate": 4.186046511627907e-07,
+      "loss": 13.883564758300782,
+      "step": 10
+    },
+    {
+      "epoch": 0.027972027972027972,
+      "grad_norm": 247.4810028076172,
+      "learning_rate": 6.511627906976745e-07,
+      "loss": 13.845947265625,
+      "step": 15
+    },
+    {
+      "epoch": 0.037296037296037296,
+      "grad_norm": 243.70272827148438,
+      "learning_rate": 8.837209302325582e-07,
+      "loss": 13.705046081542969,
+      "step": 20
+    },
+    {
+      "epoch": 0.046620046620046623,
+      "grad_norm": 238.2322235107422,
+      "learning_rate": 1.116279069767442e-06,
+      "loss": 13.556683349609376,
+      "step": 25
+    },
+    {
+      "epoch": 0.055944055944055944,
+      "grad_norm": 241.71214294433594,
+      "learning_rate": 1.3488372093023258e-06,
+      "loss": 13.285650634765625,
+      "step": 30
+    },
+    {
+      "epoch": 0.06526806526806526,
+      "grad_norm": 221.19960021972656,
+      "learning_rate": 1.5813953488372093e-06,
+      "loss": 13.05435791015625,
+      "step": 35
+    },
+    {
+      "epoch": 0.07459207459207459,
+      "grad_norm": 205.08326721191406,
+      "learning_rate": 1.8139534883720933e-06,
+      "loss": 12.770874786376954,
+      "step": 40
+    },
+    {
+      "epoch": 0.08391608391608392,
+      "grad_norm": 170.5838623046875,
+      "learning_rate": 2.0465116279069768e-06,
+      "loss": 12.280790710449219,
+      "step": 45
+    },
+    {
+      "epoch": 0.09324009324009325,
+      "grad_norm": 96.90670776367188,
+      "learning_rate": 2.2790697674418607e-06,
+      "loss": 11.730733489990234,
+      "step": 50
+    },
+    {
+      "epoch": 0.09324009324009325,
+      "eval_com_accuracy": 0.2,
+      "eval_rew_accuracy": 0.276,
+      "step": 50
+    },
+    {
+      "epoch": 0.10256410256410256,
+      "grad_norm": 61.14415740966797,
+      "learning_rate": 2.5116279069767446e-06,
+      "loss": 11.25308609008789,
+      "step": 55
+    },
+    {
+      "epoch": 0.11188811188811189,
+      "grad_norm": 56.027095794677734,
+      "learning_rate": 2.744186046511628e-06,
+      "loss": 11.197045135498048,
+      "step": 60
+    },
+    {
+      "epoch": 0.12121212121212122,
+      "grad_norm": 46.54409408569336,
+      "learning_rate": 2.9767441860465116e-06,
+      "loss": 10.874276733398437,
+      "step": 65
+    },
+    {
+      "epoch": 0.13053613053613053,
+      "grad_norm": 40.62468719482422,
+      "learning_rate": 3.2093023255813956e-06,
+      "loss": 10.69852294921875,
+      "step": 70
+    },
+    {
+      "epoch": 0.13986013986013987,
+      "grad_norm": 37.44975662231445,
+      "learning_rate": 3.4418604651162795e-06,
+      "loss": 10.560739135742187,
+      "step": 75
+    },
+    {
+      "epoch": 0.14918414918414918,
+      "grad_norm": 32.19042205810547,
+      "learning_rate": 3.674418604651163e-06,
+      "loss": 10.408053588867187,
+      "step": 80
+    },
+    {
+      "epoch": 0.1585081585081585,
+      "grad_norm": 27.208942413330078,
+      "learning_rate": 3.906976744186047e-06,
+      "loss": 10.108213806152344,
+      "step": 85
+    },
+    {
+      "epoch": 0.16783216783216784,
+      "grad_norm": 24.76529312133789,
+      "learning_rate": 4.1395348837209304e-06,
+      "loss": 9.929393005371093,
+      "step": 90
+    },
+    {
+      "epoch": 0.17715617715617715,
+      "grad_norm": 23.917495727539062,
+      "learning_rate": 4.372093023255815e-06,
+      "loss": 9.677659606933593,
+      "step": 95
+    },
+    {
+      "epoch": 0.1864801864801865,
+      "grad_norm": 22.98892593383789,
+      "learning_rate": 4.604651162790698e-06,
+      "loss": 9.481817626953125,
+      "step": 100
+    },
+    {
+      "epoch": 0.1864801864801865,
+      "eval_com_accuracy": 0.176,
+      "eval_rew_accuracy": 0.264,
+      "step": 100
+    },
+    {
+      "epoch": 0.1958041958041958,
+      "grad_norm": 25.177200317382812,
+      "learning_rate": 4.837209302325582e-06,
+      "loss": 9.42150650024414,
+      "step": 105
+    },
+    {
+      "epoch": 0.20512820512820512,
+      "grad_norm": 18.95603370666504,
+      "learning_rate": 5.069767441860466e-06,
+      "loss": 9.267645263671875,
+      "step": 110
+    },
+    {
+      "epoch": 0.21445221445221446,
+      "grad_norm": 22.155452728271484,
+      "learning_rate": 5.302325581395349e-06,
+      "loss": 9.06854248046875,
+      "step": 115
+    },
+    {
+      "epoch": 0.22377622377622378,
+      "grad_norm": 20.26909065246582,
+      "learning_rate": 5.534883720930233e-06,
+      "loss": 9.045303344726562,
+      "step": 120
+    },
+    {
+      "epoch": 0.2331002331002331,
+      "grad_norm": 26.49457359313965,
+      "learning_rate": 5.7674418604651175e-06,
+      "loss": 8.91327362060547,
+      "step": 125
+    },
+    {
+      "epoch": 0.24242424242424243,
+      "grad_norm": 19.886178970336914,
+      "learning_rate": 6e-06,
+      "loss": 8.950289154052735,
+      "step": 130
+    },
+    {
+      "epoch": 0.2517482517482518,
+      "grad_norm": 18.22088050842285,
+      "learning_rate": 6.2325581395348845e-06,
+      "loss": 8.710862731933593,
+      "step": 135
+    },
+    {
+      "epoch": 0.26107226107226106,
+      "grad_norm": 21.09984016418457,
+      "learning_rate": 6.465116279069767e-06,
+      "loss": 8.762718200683594,
+      "step": 140
+    },
+    {
+      "epoch": 0.2703962703962704,
+      "grad_norm": 17.96813201904297,
+      "learning_rate": 6.6976744186046515e-06,
+      "loss": 8.533232116699219,
+      "step": 145
+    },
+    {
+      "epoch": 0.27972027972027974,
+      "grad_norm": 20.52909278869629,
+      "learning_rate": 6.930232558139536e-06,
+      "loss": 8.279258728027344,
+      "step": 150
+    },
+    {
+      "epoch": 0.27972027972027974,
+      "eval_com_accuracy": 0.152,
+      "eval_rew_accuracy": 0.252,
+      "step": 150
+    },
+    {
+      "epoch": 0.289044289044289,
+      "grad_norm": 19.475040435791016,
+      "learning_rate": 7.1627906976744185e-06,
+      "loss": 8.369435882568359,
+      "step": 155
+    },
+    {
+      "epoch": 0.29836829836829837,
+      "grad_norm": 16.741973876953125,
+      "learning_rate": 7.395348837209303e-06,
+      "loss": 8.280664825439453,
+      "step": 160
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 19.79235076904297,
+      "learning_rate": 7.627906976744187e-06,
+      "loss": 8.200827026367188,
+      "step": 165
+    },
+    {
+      "epoch": 0.317016317016317,
+      "grad_norm": 32.991554260253906,
+      "learning_rate": 7.86046511627907e-06,
+      "loss": 7.973581695556641,
+      "step": 170
+    },
+    {
+      "epoch": 0.32634032634032634,
+      "grad_norm": 17.949445724487305,
+      "learning_rate": 8.093023255813955e-06,
+      "loss": 8.141377258300782,
+      "step": 175
+    },
+    {
+      "epoch": 0.3356643356643357,
+      "grad_norm": 16.19559097290039,
+      "learning_rate": 8.325581395348837e-06,
+      "loss": 7.837094116210937,
+      "step": 180
+    },
+    {
+      "epoch": 0.34498834498834496,
+      "grad_norm": 23.260622024536133,
+      "learning_rate": 8.558139534883722e-06,
+      "loss": 7.763467407226562,
+      "step": 185
+    },
+    {
+      "epoch": 0.3543123543123543,
+      "grad_norm": 24.858335494995117,
+      "learning_rate": 8.790697674418606e-06,
+      "loss": 7.886874389648438,
+      "step": 190
+    },
+    {
+      "epoch": 0.36363636363636365,
+      "grad_norm": 18.59360122680664,
+      "learning_rate": 9.023255813953489e-06,
+      "loss": 7.7624969482421875,
+      "step": 195
+    },
+    {
+      "epoch": 0.372960372960373,
+      "grad_norm": 18.056495666503906,
+      "learning_rate": 9.255813953488373e-06,
+      "loss": 7.4621437072753904,
+      "step": 200
+    },
+    {
+      "epoch": 0.372960372960373,
+      "eval_com_accuracy": 0.144,
+      "eval_rew_accuracy": 0.332,
+      "step": 200
+    },
+    {
+      "epoch": 0.3822843822843823,
+      "grad_norm": 16.496261596679688,
+      "learning_rate": 9.488372093023258e-06,
+      "loss": 7.526261901855468,
+      "step": 205
+    },
+    {
+      "epoch": 0.3916083916083916,
+      "grad_norm": 16.77888298034668,
+      "learning_rate": 9.72093023255814e-06,
+      "loss": 7.624967193603515,
+      "step": 210
+    },
+    {
+      "epoch": 0.40093240093240096,
+      "grad_norm": 17.096778869628906,
+      "learning_rate": 9.953488372093025e-06,
+      "loss": 7.337849426269531,
+      "step": 215
+    },
+    {
+      "epoch": 0.41025641025641024,
+      "grad_norm": 18.3544979095459,
+      "learning_rate": 9.999894343918674e-06,
+      "loss": 7.354197692871094,
+      "step": 220
+    },
+    {
+      "epoch": 0.4195804195804196,
+      "grad_norm": 15.287137985229492,
+      "learning_rate": 9.999465123741172e-06,
+      "loss": 7.085916137695312,
+      "step": 225
+    },
+    {
+      "epoch": 0.4289044289044289,
+      "grad_norm": 17.03432273864746,
+      "learning_rate": 9.998705764284132e-06,
+      "loss": 7.059596252441406,
+      "step": 230
+    },
+    {
+      "epoch": 0.4382284382284382,
+      "grad_norm": 15.470928192138672,
+      "learning_rate": 9.99761631569179e-06,
+      "loss": 7.118296813964844,
+      "step": 235
+    },
+    {
+      "epoch": 0.44755244755244755,
+      "grad_norm": 15.732117652893066,
+      "learning_rate": 9.9961968499058e-06,
+      "loss": 6.981436157226563,
+      "step": 240
+    },
+    {
+      "epoch": 0.4568764568764569,
+      "grad_norm": 20.92630958557129,
+      "learning_rate": 9.994447460660473e-06,
+      "loss": 6.96953125,
+      "step": 245
+    },
+    {
+      "epoch": 0.4662004662004662,
+      "grad_norm": 14.848095893859863,
+      "learning_rate": 9.992368263476585e-06,
+      "loss": 6.831211090087891,
+      "step": 250
+    },
+    {
+      "epoch": 0.4662004662004662,
+      "eval_com_accuracy": 0.132,
+      "eval_rew_accuracy": 0.312,
+      "step": 250
+    },
+    {
+      "epoch": 0.4755244755244755,
+      "grad_norm": 18.59755516052246,
+      "learning_rate": 9.989959395653756e-06,
+      "loss": 6.873024749755859,
+      "step": 255
+    },
+    {
+      "epoch": 0.48484848484848486,
+      "grad_norm": 15.705364227294922,
+      "learning_rate": 9.98722101626138e-06,
+      "loss": 6.633563232421875,
+      "step": 260
+    },
+    {
+      "epoch": 0.49417249417249415,
+      "grad_norm": 39.2294807434082,
+      "learning_rate": 9.984153306128124e-06,
+      "loss": 6.63948974609375,
+      "step": 265
+    },
+    {
+      "epoch": 0.5034965034965035,
+      "grad_norm": 13.209358215332031,
+      "learning_rate": 9.980756467829977e-06,
+      "loss": 6.688297271728516,
+      "step": 270
+    },
+    {
+      "epoch": 0.5128205128205128,
+      "grad_norm": 14.645393371582031,
+      "learning_rate": 9.977030725676887e-06,
+      "loss": 6.4885505676269535,
+      "step": 275
+    },
+    {
+      "epoch": 0.5221445221445221,
+      "grad_norm": 17.008819580078125,
+      "learning_rate": 9.972976325697938e-06,
+      "loss": 6.78935546875,
+      "step": 280
+    },
+    {
+      "epoch": 0.5314685314685315,
+      "grad_norm": 16.029518127441406,
+      "learning_rate": 9.96859353562511e-06,
+      "loss": 6.681609344482422,
+      "step": 285
+    },
+    {
+      "epoch": 0.5407925407925408,
+      "grad_norm": 13.878348350524902,
+      "learning_rate": 9.963882644875594e-06,
+      "loss": 6.528868103027344,
+      "step": 290
+    },
+    {
+      "epoch": 0.5501165501165501,
+      "grad_norm": 13.996464729309082,
+      "learning_rate": 9.958843964532683e-06,
+      "loss": 6.431344604492187,
+      "step": 295
+    },
+    {
+      "epoch": 0.5594405594405595,
+      "grad_norm": 15.72003173828125,
+      "learning_rate": 9.953477827325229e-06,
+      "loss": 6.5927978515625,
+      "step": 300
+    },
+    {
+      "epoch": 0.5594405594405595,
+      "eval_com_accuracy": 0.148,
+      "eval_rew_accuracy": 0.368,
+      "step": 300
+    },
+    {
+      "epoch": 0.5687645687645687,
+      "grad_norm": 16.416725158691406,
+      "learning_rate": 9.947784587605678e-06,
+      "loss": 6.411885833740234,
+      "step": 305
+    },
+    {
+      "epoch": 0.578088578088578,
+      "grad_norm": 14.880790710449219,
+      "learning_rate": 9.941764621326655e-06,
+      "loss": 6.5298927307128904,
+      "step": 310
+    },
+    {
+      "epoch": 0.5874125874125874,
+      "grad_norm": 17.828296661376953,
+      "learning_rate": 9.935418326016153e-06,
+      "loss": 6.045618057250977,
+      "step": 315
+    },
+    {
+      "epoch": 0.5967365967365967,
+      "grad_norm": 13.30245590209961,
+      "learning_rate": 9.928746120751275e-06,
+      "loss": 6.1650840759277346,
+      "step": 320
+    },
+    {
+      "epoch": 0.6060606060606061,
+      "grad_norm": 13.527432441711426,
+      "learning_rate": 9.921748446130564e-06,
+      "loss": 6.349945831298828,
+      "step": 325
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 14.491817474365234,
+      "learning_rate": 9.9144257642449e-06,
+      "loss": 6.232943725585938,
+      "step": 330
+    },
+    {
+      "epoch": 0.6247086247086248,
+      "grad_norm": 15.7533597946167,
+      "learning_rate": 9.906778558647e-06,
+      "loss": 6.564260864257813,
+      "step": 335
+    },
+    {
+      "epoch": 0.634032634032634,
+      "grad_norm": 16.538602828979492,
+      "learning_rate": 9.898807334319471e-06,
+      "loss": 6.122506332397461,
+      "step": 340
+    },
+    {
+      "epoch": 0.6433566433566433,
+      "grad_norm": 16.459688186645508,
+      "learning_rate": 9.890512617641474e-06,
+      "loss": 6.332431030273438,
+      "step": 345
+    },
+    {
+      "epoch": 0.6526806526806527,
+      "grad_norm": 14.879058837890625,
+      "learning_rate": 9.881894956353963e-06,
+      "loss": 6.118003082275391,
+      "step": 350
+    },
+    {
+      "epoch": 0.6526806526806527,
+      "eval_com_accuracy": 0.164,
+      "eval_rew_accuracy": 0.364,
+      "step": 350
+    },
+    {
+      "epoch": 0.662004662004662,
+      "grad_norm": 16.02516746520996,
+      "learning_rate": 9.87295491952351e-06,
+      "loss": 6.241584014892578,
+      "step": 355
+    },
+    {
+      "epoch": 0.6713286713286714,
+      "grad_norm": 14.9837646484375,
+      "learning_rate": 9.863693097504733e-06,
+      "loss": 6.083467864990235,
+      "step": 360
+    },
+    {
+      "epoch": 0.6806526806526807,
+      "grad_norm": 14.138517379760742,
+      "learning_rate": 9.854110101901308e-06,
+      "loss": 6.143560791015625,
+      "step": 365
+    },
+    {
+      "epoch": 0.6899766899766899,
+      "grad_norm": 12.696243286132812,
+      "learning_rate": 9.844206565525585e-06,
+      "loss": 6.061418914794922,
+      "step": 370
+    },
+    {
+      "epoch": 0.6993006993006993,
+      "grad_norm": 15.22846508026123,
+      "learning_rate": 9.833983142356792e-06,
+      "loss": 6.193332672119141,
+      "step": 375
+    },
+    {
+      "epoch": 0.7086247086247086,
+      "grad_norm": 15.284857749938965,
+      "learning_rate": 9.823440507497863e-06,
+      "loss": 6.0868377685546875,
+      "step": 380
+    },
+    {
+      "epoch": 0.717948717948718,
+      "grad_norm": 14.117541313171387,
+      "learning_rate": 9.812579357130848e-06,
+      "loss": 6.289948272705078,
+      "step": 385
+    },
+    {
+      "epoch": 0.7272727272727273,
+      "grad_norm": 12.648367881774902,
+      "learning_rate": 9.801400408470943e-06,
+      "loss": 6.194795227050781,
+      "step": 390
+    },
+    {
+      "epoch": 0.7365967365967366,
+      "grad_norm": 12.394021034240723,
+      "learning_rate": 9.789904399719124e-06,
+      "loss": 6.031880950927734,
+      "step": 395
+    },
+    {
+      "epoch": 0.745920745920746,
+      "grad_norm": 15.566425323486328,
+      "learning_rate": 9.778092090013416e-06,
+      "loss": 6.0966644287109375,
+      "step": 400
+    },
+    {
+      "epoch": 0.745920745920746,
+      "eval_com_accuracy": 0.184,
+      "eval_rew_accuracy": 0.364,
+      "step": 400
+    },
+    {
+      "epoch": 0.7552447552447552,
+      "grad_norm": 14.885455131530762,
+      "learning_rate": 9.76596425937874e-06,
+      "loss": 6.476510620117187,
+      "step": 405
+    },
+    {
+      "epoch": 0.7645687645687645,
+      "grad_norm": 16.627117156982422,
+      "learning_rate": 9.753521708675426e-06,
+      "loss": 5.9605262756347654,
+      "step": 410
+    },
+    {
+      "epoch": 0.7738927738927739,
+      "grad_norm": 14.517923355102539,
+      "learning_rate": 9.740765259546312e-06,
+      "loss": 6.1104694366455075,
+      "step": 415
+    },
+    {
+      "epoch": 0.7832167832167832,
+      "grad_norm": 12.007637977600098,
+      "learning_rate": 9.727695754362498e-06,
+      "loss": 5.964785766601563,
+      "step": 420
+    },
+    {
+      "epoch": 0.7925407925407926,
+      "grad_norm": 16.17314338684082,
+      "learning_rate": 9.714314056167711e-06,
+      "loss": 5.976337432861328,
+      "step": 425
+    },
+    {
+      "epoch": 0.8018648018648019,
+      "grad_norm": 13.246196746826172,
+      "learning_rate": 9.700621048621322e-06,
+      "loss": 5.936331939697266,
+      "step": 430
+    },
+    {
+      "epoch": 0.8111888111888111,
+      "grad_norm": 15.030324935913086,
+      "learning_rate": 9.686617635939988e-06,
+      "loss": 6.209032821655273,
+      "step": 435
+    },
+    {
+      "epoch": 0.8205128205128205,
+      "grad_norm": 13.853086471557617,
+      "learning_rate": 9.672304742837945e-06,
+      "loss": 5.985638427734375,
+      "step": 440
+    },
+    {
+      "epoch": 0.8298368298368298,
+      "grad_norm": 13.534188270568848,
+      "learning_rate": 9.657683314465948e-06,
+      "loss": 5.7971649169921875,
+      "step": 445
+    },
+    {
+      "epoch": 0.8391608391608392,
+      "grad_norm": 26.1879825592041,
+      "learning_rate": 9.642754316348846e-06,
+      "loss": 5.949985504150391,
+      "step": 450
+    },
+    {
+      "epoch": 0.8391608391608392,
+      "eval_com_accuracy": 0.176,
+      "eval_rew_accuracy": 0.34,
+      "step": 450
+    },
+    {
+      "epoch": 0.8484848484848485,
+      "grad_norm": 16.848834991455078,
+      "learning_rate": 9.627518734321837e-06,
+      "loss": 6.1193115234375,
+      "step": 455
+    },
+    {
+      "epoch": 0.8578088578088578,
+      "grad_norm": 30.765796661376953,
+      "learning_rate": 9.61197757446536e-06,
+      "loss": 5.74780158996582,
+      "step": 460
+    },
+    {
+      "epoch": 0.8671328671328671,
+      "grad_norm": 13.27775764465332,
+      "learning_rate": 9.596131863038664e-06,
+      "loss": 5.994970703125,
+      "step": 465
+    },
+    {
+      "epoch": 0.8764568764568764,
+      "grad_norm": 14.923792839050293,
+      "learning_rate": 9.579982646412039e-06,
+      "loss": 5.810473251342773,
+      "step": 470
+    },
+    {
+      "epoch": 0.8857808857808858,
+      "grad_norm": 14.351722717285156,
+      "learning_rate": 9.563530990997707e-06,
+      "loss": 6.09692268371582,
+      "step": 475
+    },
+    {
+      "epoch": 0.8951048951048951,
+      "grad_norm": 11.83583927154541,
+      "learning_rate": 9.546777983179421e-06,
+      "loss": 5.867227935791016,
+      "step": 480
+    },
+    {
+      "epoch": 0.9044289044289044,
+      "grad_norm": 19.415884017944336,
+      "learning_rate": 9.529724729240712e-06,
+      "loss": 5.834615707397461,
+      "step": 485
+    },
+    {
+      "epoch": 0.9137529137529138,
+      "grad_norm": 12.147261619567871,
+      "learning_rate": 9.512372355291838e-06,
+      "loss": 6.039286041259766,
+      "step": 490
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 12.706372261047363,
+      "learning_rate": 9.494722007195427e-06,
+      "loss": 6.06304702758789,
+      "step": 495
+    },
+    {
+      "epoch": 0.9324009324009324,
+      "grad_norm": 12.642041206359863,
+      "learning_rate": 9.476774850490803e-06,
+      "loss": 5.9729866027832035,
+      "step": 500
+    },
+    {
+      "epoch": 0.9324009324009324,
+      "eval_com_accuracy": 0.204,
+      "eval_rew_accuracy": 0.328,
+      "step": 500
+    },
+    {
+      "epoch": 0.9417249417249417,
+      "grad_norm": 12.66169548034668,
+      "learning_rate": 9.458532070317021e-06,
+      "loss": 5.8281097412109375,
+      "step": 505
+    },
+    {
+      "epoch": 0.951048951048951,
+      "grad_norm": 13.191854476928711,
+      "learning_rate": 9.439994871334614e-06,
+      "loss": 6.183546447753907,
+      "step": 510
+    },
+    {
+      "epoch": 0.9603729603729604,
+      "grad_norm": 12.52676010131836,
+      "learning_rate": 9.421164477646031e-06,
+      "loss": 5.785601043701172,
+      "step": 515
+    },
+    {
+      "epoch": 0.9696969696969697,
+      "grad_norm": 13.225418090820312,
+      "learning_rate": 9.402042132714817e-06,
+      "loss": 5.784825134277344,
+      "step": 520
+    },
+    {
+      "epoch": 0.9790209790209791,
+      "grad_norm": 12.035611152648926,
+      "learning_rate": 9.382629099283486e-06,
+      "loss": 5.9122272491455075,
+      "step": 525
+    },
+    {
+      "epoch": 0.9883449883449883,
+      "grad_norm": 12.166351318359375,
+      "learning_rate": 9.362926659290149e-06,
+      "loss": 5.790053939819336,
+      "step": 530
+    },
+    {
+      "epoch": 0.9976689976689976,
+      "grad_norm": 13.22154426574707,
+      "learning_rate": 9.342936113783855e-06,
+      "loss": 5.745006561279297,
+      "step": 535
+    },
+    {
+      "epoch": 1.0055944055944055,
+      "grad_norm": 13.12752628326416,
+      "learning_rate": 9.32265878283868e-06,
+      "loss": 5.1531005859375,
+      "step": 540
+    },
+    {
+      "epoch": 1.014918414918415,
+      "grad_norm": 15.964373588562012,
+      "learning_rate": 9.302096005466547e-06,
+      "loss": 5.7917236328125,
+      "step": 545
+    },
+    {
+      "epoch": 1.0242424242424242,
+      "grad_norm": 18.454797744750977,
+      "learning_rate": 9.281249139528816e-06,
+      "loss": 5.967070007324219,
+      "step": 550
+    },
+    {
+      "epoch": 1.0242424242424242,
+      "eval_com_accuracy": 0.212,
+      "eval_rew_accuracy": 0.352,
+      "step": 550
+    },
+    {
+      "epoch": 1.0335664335664336,
+      "grad_norm": 15.802419662475586,
+      "learning_rate": 9.260119561646614e-06,
+      "loss": 6.093529129028321,
+      "step": 555
+    },
+    {
+      "epoch": 1.0428904428904429,
+      "grad_norm": 21.95537757873535,
+      "learning_rate": 9.238708667109924e-06,
+      "loss": 5.489861297607422,
+      "step": 560
+    },
+    {
+      "epoch": 1.0522144522144523,
+      "grad_norm": 13.373199462890625,
+      "learning_rate": 9.217017869785453e-06,
+      "loss": 5.871435928344726,
+      "step": 565
+    },
+    {
+      "epoch": 1.0615384615384615,
+      "grad_norm": 13.36340618133545,
+      "learning_rate": 9.19504860202327e-06,
+      "loss": 5.805611038208008,
+      "step": 570
+    },
+    {
+      "epoch": 1.0708624708624708,
+      "grad_norm": 14.099553108215332,
+      "learning_rate": 9.172802314562214e-06,
+      "loss": 5.600634765625,
+      "step": 575
+    },
+    {
+      "epoch": 1.0801864801864802,
+      "grad_norm": 15.012040138244629,
+      "learning_rate": 9.150280476434098e-06,
+      "loss": 5.793732452392578,
+      "step": 580
+    },
+    {
+      "epoch": 1.0895104895104895,
+      "grad_norm": 15.98738956451416,
+      "learning_rate": 9.127484574866699e-06,
+      "loss": 5.8642620086669925,
+      "step": 585
+    },
+    {
+      "epoch": 1.098834498834499,
+      "grad_norm": 13.785572052001953,
+      "learning_rate": 9.104416115185557e-06,
+      "loss": 5.770995330810547,
+      "step": 590
+    },
+    {
+      "epoch": 1.1081585081585081,
+      "grad_norm": 12.145087242126465,
+      "learning_rate": 9.08107662071456e-06,
+      "loss": 5.850731277465821,
+      "step": 595
+    },
+    {
+      "epoch": 1.1174825174825176,
+      "grad_norm": 13.681862831115723,
+      "learning_rate": 9.057467632675357e-06,
+      "loss": 5.782034301757813,
+      "step": 600
+    },
+    {
+      "epoch": 1.1174825174825176,
+      "eval_com_accuracy": 0.22,
+      "eval_rew_accuracy": 0.348,
+      "step": 600
+    },
+    {
+      "epoch": 1.1268065268065268,
+      "grad_norm": 13.870027542114258,
+      "learning_rate": 9.033590710085584e-06,
+      "loss": 5.723427963256836,
+      "step": 605
+    },
+    {
+      "epoch": 1.136130536130536,
+      "grad_norm": 16.20020294189453,
+      "learning_rate": 9.00944742965592e-06,
+      "loss": 5.953633117675781,
+      "step": 610
+    },
+    {
+      "epoch": 1.1454545454545455,
+      "grad_norm": 15.48993968963623,
+      "learning_rate": 8.985039385685952e-06,
+      "loss": 5.809123992919922,
+      "step": 615
+    },
+    {
+      "epoch": 1.1547785547785547,
+      "grad_norm": 13.686211585998535,
+      "learning_rate": 8.960368189958913e-06,
+      "loss": 5.813122177124024,
+      "step": 620
+    },
+    {
+      "epoch": 1.1641025641025642,
+      "grad_norm": 13.733199119567871,
+      "learning_rate": 8.935435471635238e-06,
+      "loss": 5.6748512268066404,
+      "step": 625
+    },
+    {
+      "epoch": 1.1734265734265734,
+      "grad_norm": 12.450380325317383,
+      "learning_rate": 8.91024287714499e-06,
+      "loss": 5.6834667205810545,
+      "step": 630
+    },
+    {
+      "epoch": 1.1827505827505829,
+      "grad_norm": 14.111934661865234,
+      "learning_rate": 8.884792070079128e-06,
+      "loss": 5.680038833618164,
+      "step": 635
+    },
+    {
+      "epoch": 1.192074592074592,
+      "grad_norm": 13.175325393676758,
+      "learning_rate": 8.859084731079664e-06,
+      "loss": 5.445759201049805,
+      "step": 640
+    },
+    {
+      "epoch": 1.2013986013986013,
+      "grad_norm": 14.951041221618652,
+      "learning_rate": 8.833122557728667e-06,
+      "loss": 5.673246765136719,
+      "step": 645
+    },
+    {
+      "epoch": 1.2107226107226108,
+      "grad_norm": 12.364238739013672,
+      "learning_rate": 8.806907264436183e-06,
+      "loss": 5.234909439086914,
+      "step": 650
+    },
+    {
+      "epoch": 1.2107226107226108,
+      "eval_com_accuracy": 0.224,
+      "eval_rew_accuracy": 0.348,
+      "step": 650
+    },
+    {
+      "epoch": 1.22004662004662,
+      "grad_norm": 13.065054893493652,
+      "learning_rate": 8.780440582327005e-06,
+      "loss": 5.442026519775391,
+      "step": 655
+    },
+    {
+      "epoch": 1.2293706293706295,
+      "grad_norm": 13.872124671936035,
+      "learning_rate": 8.75372425912637e-06,
+      "loss": 5.736603546142578,
+      "step": 660
+    },
+    {
+      "epoch": 1.2386946386946387,
+      "grad_norm": 13.677570343017578,
+      "learning_rate": 8.726760059044542e-06,
+      "loss": 5.951699447631836,
+      "step": 665
+    },
+    {
+      "epoch": 1.248018648018648,
+      "grad_norm": 15.023405075073242,
+      "learning_rate": 8.699549762660318e-06,
+      "loss": 5.936520004272461,
+      "step": 670
+    },
+    {
+      "epoch": 1.2573426573426574,
+      "grad_norm": 12.804817199707031,
+      "learning_rate": 8.672095166803445e-06,
+      "loss": 5.778074645996094,
+      "step": 675
+    },
+    {
+      "epoch": 1.2666666666666666,
+      "grad_norm": 13.148968696594238,
+      "learning_rate": 8.644398084435959e-06,
+      "loss": 5.7417655944824215,
+      "step": 680
+    },
+    {
+      "epoch": 1.275990675990676,
+      "grad_norm": 14.131410598754883,
+      "learning_rate": 8.616460344532483e-06,
+      "loss": 5.58878173828125,
+      "step": 685
+    },
+    {
+      "epoch": 1.2853146853146853,
+      "grad_norm": 14.100456237792969,
+      "learning_rate": 8.588283791959437e-06,
+      "loss": 5.546084976196289,
+      "step": 690
+    },
+    {
+      "epoch": 1.2946386946386945,
+      "grad_norm": 13.94593620300293,
+      "learning_rate": 8.559870287353214e-06,
+      "loss": 6.0555156707763675,
+      "step": 695
+    },
+    {
+      "epoch": 1.303962703962704,
+      "grad_norm": 18.369823455810547,
+      "learning_rate": 8.531221706997316e-06,
+      "loss": 5.930012512207031,
+      "step": 700
+    },
+    {
+      "epoch": 1.303962703962704,
+      "eval_com_accuracy": 0.24,
+      "eval_rew_accuracy": 0.364,
+      "step": 700
+    },
+    {
+      "epoch": 1.3132867132867134,
+      "grad_norm": 14.022618293762207,
+      "learning_rate": 8.502339942698463e-06,
+      "loss": 5.831372833251953,
+      "step": 705
+    },
+    {
+      "epoch": 1.3226107226107227,
+      "grad_norm": 12.844898223876953,
+      "learning_rate": 8.473226901661643e-06,
+      "loss": 5.503532409667969,
+      "step": 710
+    },
+    {
+      "epoch": 1.3319347319347319,
+      "grad_norm": 13.991181373596191,
+      "learning_rate": 8.443884506364192e-06,
+      "loss": 5.344602966308594,
+      "step": 715
+    },
+    {
+      "epoch": 1.3412587412587413,
+      "grad_norm": 13.290801048278809,
+      "learning_rate": 8.414314694428842e-06,
+      "loss": 5.7142791748046875,
+      "step": 720
+    },
+    {
+      "epoch": 1.3505827505827506,
+      "grad_norm": 13.986117362976074,
+      "learning_rate": 8.384519418495755e-06,
+      "loss": 5.509880065917969,
+      "step": 725
+    },
+    {
+      "epoch": 1.3599067599067598,
+      "grad_norm": 17.483638763427734,
+      "learning_rate": 8.354500646093592e-06,
+      "loss": 5.769047546386719,
+      "step": 730
+    },
+    {
+      "epoch": 1.3692307692307693,
+      "grad_norm": 12.383395195007324,
+      "learning_rate": 8.324260359509594e-06,
+      "loss": 5.693521881103516,
+      "step": 735
+    },
+    {
+      "epoch": 1.3785547785547785,
+      "grad_norm": 12.225541114807129,
+      "learning_rate": 8.29380055565866e-06,
+      "loss": 5.618255233764648,
+      "step": 740
+    },
+    {
+      "epoch": 1.387878787878788,
+      "grad_norm": 11.737737655639648,
+      "learning_rate": 8.263123245951504e-06,
+      "loss": 5.702220916748047,
+      "step": 745
+    },
+    {
+      "epoch": 1.3972027972027972,
+      "grad_norm": 14.235527992248535,
+      "learning_rate": 8.232230456161819e-06,
+      "loss": 5.744013977050781,
+      "step": 750
+    },
+    {
+      "epoch": 1.3972027972027972,
+      "eval_com_accuracy": 0.24,
+      "eval_rew_accuracy": 0.368,
+      "step": 750
+    },
+    {
+      "epoch": 1.4065268065268066,
+      "grad_norm": 12.433022499084473,
+      "learning_rate": 8.201124226292505e-06,
+      "loss": 5.6122283935546875,
+      "step": 755
+    },
+    {
+      "epoch": 1.4158508158508158,
+      "grad_norm": 13.874029159545898,
+      "learning_rate": 8.169806610440966e-06,
+      "loss": 5.530184936523438,
+      "step": 760
+    },
+    {
+      "epoch": 1.425174825174825,
+      "grad_norm": 12.640862464904785,
+      "learning_rate": 8.138279676663458e-06,
+      "loss": 5.602608489990234,
+      "step": 765
+    },
+    {
+      "epoch": 1.4344988344988345,
+      "grad_norm": 13.672120094299316,
+      "learning_rate": 8.106545506838533e-06,
+      "loss": 5.971489715576172,
+      "step": 770
+    },
+    {
+      "epoch": 1.4438228438228438,
+      "grad_norm": 12.973994255065918,
+      "learning_rate": 8.074606196529554e-06,
+      "loss": 5.967403411865234,
+      "step": 775
+    },
+    {
+      "epoch": 1.4531468531468532,
+      "grad_norm": 12.31808090209961,
+      "learning_rate": 8.042463854846325e-06,
+      "loss": 5.71468505859375,
+      "step": 780
+    },
+    {
+      "epoch": 1.4624708624708624,
+      "grad_norm": 12.416245460510254,
+      "learning_rate": 8.010120604305806e-06,
+      "loss": 5.836771392822266,
+      "step": 785
+    },
+    {
+      "epoch": 1.471794871794872,
+      "grad_norm": 13.020674705505371,
+      "learning_rate": 7.977578580691963e-06,
+      "loss": 5.71131591796875,
+      "step": 790
+    },
+    {
+      "epoch": 1.4811188811188811,
+      "grad_norm": 13.946771621704102,
+      "learning_rate": 7.944839932914718e-06,
+      "loss": 5.564648818969727,
+      "step": 795
+    },
+    {
+      "epoch": 1.4904428904428904,
+      "grad_norm": 12.287398338317871,
+      "learning_rate": 7.91190682286806e-06,
+      "loss": 5.753592300415039,
+      "step": 800
+    },
+    {
+      "epoch": 1.4904428904428904,
+      "eval_com_accuracy": 0.236,
+      "eval_rew_accuracy": 0.384,
+      "step": 800
+    },
+    {
+      "epoch": 1.4997668997668998,
+      "grad_norm": 20.467731475830078,
+      "learning_rate": 7.878781425287277e-06,
+      "loss": 5.64927978515625,
+      "step": 805
+    },
+    {
+      "epoch": 1.509090909090909,
+      "grad_norm": 13.811100006103516,
+      "learning_rate": 7.84546592760535e-06,
+      "loss": 5.415856170654297,
+      "step": 810
+    },
+    {
+      "epoch": 1.5184149184149183,
+      "grad_norm": 12.4107666015625,
+      "learning_rate": 7.811962529808499e-06,
+      "loss": 5.767454147338867,
+      "step": 815
+    },
+    {
+      "epoch": 1.5277389277389277,
+      "grad_norm": 14.148603439331055,
+      "learning_rate": 7.778273444290921e-06,
+      "loss": 5.463053131103516,
+      "step": 820
+    },
+    {
+      "epoch": 1.5370629370629372,
+      "grad_norm": 13.375722885131836,
+      "learning_rate": 7.744400895708683e-06,
+      "loss": 5.625373077392578,
+      "step": 825
+    },
+    {
+      "epoch": 1.5463869463869464,
+      "grad_norm": 12.991719245910645,
+      "learning_rate": 7.710347120832821e-06,
+      "loss": 5.6271202087402346,
+      "step": 830
+    },
+    {
+      "epoch": 1.5557109557109556,
+      "grad_norm": 12.906146049499512,
+      "learning_rate": 7.676114368401635e-06,
+      "loss": 5.5160057067871096,
+      "step": 835
+    },
+    {
+      "epoch": 1.565034965034965,
+      "grad_norm": 12.344596862792969,
+      "learning_rate": 7.641704898972194e-06,
+      "loss": 5.506198120117188,
+      "step": 840
+    },
+    {
+      "epoch": 1.5743589743589743,
+      "grad_norm": 13.285745620727539,
+      "learning_rate": 7.607120984771058e-06,
+      "loss": 5.7151634216308596,
+      "step": 845
+    },
+    {
+      "epoch": 1.5836829836829835,
+      "grad_norm": 12.432035446166992,
+      "learning_rate": 7.572364909544235e-06,
+      "loss": 5.852275085449219,
+      "step": 850
+    },
+    {
+      "epoch": 1.5836829836829835,
+      "eval_com_accuracy": 0.228,
+      "eval_rew_accuracy": 0.332,
+      "step": 850
+    },
+    {
+      "epoch": 1.593006993006993,
+      "grad_norm": 12.581634521484375,
+      "learning_rate": 7.537438968406372e-06,
+      "loss": 5.735688018798828,
+      "step": 855
+    },
+    {
+      "epoch": 1.6023310023310025,
+      "grad_norm": 13.161237716674805,
+      "learning_rate": 7.502345467689202e-06,
+      "loss": 5.648350524902344,
+      "step": 860
+    },
+    {
+      "epoch": 1.6116550116550117,
+      "grad_norm": 12.48270034790039,
+      "learning_rate": 7.4670867247892346e-06,
+      "loss": 5.635168075561523,
+      "step": 865
+    },
+    {
+      "epoch": 1.620979020979021,
+      "grad_norm": 13.835477828979492,
+      "learning_rate": 7.431665068014737e-06,
+      "loss": 5.548571014404297,
+      "step": 870
+    },
+    {
+      "epoch": 1.6303030303030304,
+      "grad_norm": 12.403617858886719,
+      "learning_rate": 7.396082836431981e-06,
+      "loss": 5.7486827850341795,
+      "step": 875
+    },
+    {
+      "epoch": 1.6396270396270396,
+      "grad_norm": 13.730851173400879,
+      "learning_rate": 7.3603423797107845e-06,
+      "loss": 5.288154602050781,
+      "step": 880
+    },
+    {
+      "epoch": 1.6489510489510488,
+      "grad_norm": 14.688888549804688,
+      "learning_rate": 7.324446057969346e-06,
+      "loss": 5.648311614990234,
+      "step": 885
+    },
+    {
+      "epoch": 1.6582750582750583,
+      "grad_norm": 13.311734199523926,
+      "learning_rate": 7.288396241618401e-06,
+      "loss": 5.636775207519531,
+      "step": 890
+    },
+    {
+      "epoch": 1.6675990675990677,
+      "grad_norm": 12.776606559753418,
+      "learning_rate": 7.252195311204689e-06,
+      "loss": 5.666929626464844,
+      "step": 895
+    },
+    {
+      "epoch": 1.676923076923077,
+      "grad_norm": 14.130080223083496,
+      "learning_rate": 7.215845657253755e-06,
+      "loss": 5.534748840332031,
+      "step": 900
+    },
+    {
+      "epoch": 1.676923076923077,
+      "eval_com_accuracy": 0.264,
+      "eval_rew_accuracy": 0.364,
+      "step": 900
+    },
+    {
+      "epoch": 1.6862470862470862,
+      "grad_norm": 14.159326553344727,
+      "learning_rate": 7.1793496801120885e-06,
+      "loss": 5.652375793457031,
+      "step": 905
+    },
+    {
+      "epoch": 1.6955710955710956,
+      "grad_norm": 13.113655090332031,
+      "learning_rate": 7.1427097897886225e-06,
+      "loss": 5.383832168579102,
+      "step": 910
+    },
+    {
+      "epoch": 1.7048951048951049,
+      "grad_norm": 14.203001022338867,
+      "learning_rate": 7.105928405795584e-06,
+      "loss": 5.6499076843261715,
+      "step": 915
+    },
+    {
+      "epoch": 1.714219114219114,
+      "grad_norm": 12.1084623336792,
+      "learning_rate": 7.069007956988718e-06,
+      "loss": 5.560025787353515,
+      "step": 920
+    },
+    {
+      "epoch": 1.7235431235431236,
+      "grad_norm": 13.819426536560059,
+      "learning_rate": 7.031950881406913e-06,
+      "loss": 5.535141372680664,
+      "step": 925
+    },
+    {
+      "epoch": 1.732867132867133,
+      "grad_norm": 17.341110229492188,
+      "learning_rate": 6.994759626111189e-06,
+      "loss": 5.697111511230469,
+      "step": 930
+    },
+    {
+      "epoch": 1.7421911421911422,
+      "grad_norm": 16.404672622680664,
+      "learning_rate": 6.957436647023117e-06,
+      "loss": 5.4155632019042965,
+      "step": 935
+    },
+    {
+      "epoch": 1.7515151515151515,
+      "grad_norm": 13.115015029907227,
+      "learning_rate": 6.919984408762632e-06,
+      "loss": 5.758349990844726,
+      "step": 940
+    },
+    {
+      "epoch": 1.760839160839161,
+      "grad_norm": 13.850481986999512,
+      "learning_rate": 6.882405384485294e-06,
+      "loss": 5.69927864074707,
+      "step": 945
+    },
+    {
+      "epoch": 1.7701631701631702,
+      "grad_norm": 12.921819686889648,
+      "learning_rate": 6.844702055718964e-06,
+      "loss": 5.631483459472657,
+      "step": 950
+    },
+    {
+      "epoch": 1.7701631701631702,
+      "eval_com_accuracy": 0.26,
+      "eval_rew_accuracy": 0.352,
+      "step": 950
+    },
+    {
+      "epoch": 1.7794871794871794,
+      "grad_norm": 13.401888847351074,
+      "learning_rate": 6.806876912199945e-06,
+      "loss": 5.824296569824218,
+      "step": 955
+    },
+    {
+      "epoch": 1.7888111888111888,
+      "grad_norm": 13.481993675231934,
+      "learning_rate": 6.768932451708557e-06,
+      "loss": 5.752721786499023,
+      "step": 960
+    },
+    {
+      "epoch": 1.7981351981351983,
+      "grad_norm": 14.022424697875977,
+      "learning_rate": 6.730871179904218e-06,
+      "loss": 5.453369522094727,
+      "step": 965
+    },
+    {
+      "epoch": 1.8074592074592073,
+      "grad_norm": 13.432122230529785,
+      "learning_rate": 6.692695610159966e-06,
+      "loss": 5.664297103881836,
+      "step": 970
+    },
+    {
+      "epoch": 1.8167832167832167,
+      "grad_norm": 14.292946815490723,
+      "learning_rate": 6.6544082633964955e-06,
+      "loss": 5.576699066162109,
+      "step": 975
+    },
+    {
+      "epoch": 1.8261072261072262,
+      "grad_norm": 14.256867408752441,
+      "learning_rate": 6.6160116679156874e-06,
+      "loss": 5.551004791259766,
+      "step": 980
+    },
+    {
+      "epoch": 1.8354312354312354,
+      "grad_norm": 14.11919116973877,
+      "learning_rate": 6.577508359233653e-06,
+      "loss": 5.580959320068359,
+      "step": 985
+    },
+    {
+      "epoch": 1.8447552447552447,
+      "grad_norm": 14.006019592285156,
+      "learning_rate": 6.538900879913301e-06,
+      "loss": 5.443265533447265,
+      "step": 990
+    },
+    {
+      "epoch": 1.8540792540792541,
+      "grad_norm": 11.976597785949707,
+      "learning_rate": 6.500191779396439e-06,
+      "loss": 5.410481643676758,
+      "step": 995
+    },
+    {
+      "epoch": 1.8634032634032636,
+      "grad_norm": 14.439790725708008,
+      "learning_rate": 6.461383613835427e-06,
+      "loss": 5.3311511993408205,
+      "step": 1000
+    },
+    {
+      "epoch": 1.8634032634032636,
+      "eval_com_accuracy": 0.24,
+      "eval_rew_accuracy": 0.356,
+      "step": 1000
+    },
+    {
+      "epoch": 1.8727272727272726,
+      "grad_norm": 15.29392147064209,
+      "learning_rate": 6.4224789459243705e-06,
+      "loss": 5.477576446533203,
+      "step": 1005
+    },
+    {
+      "epoch": 1.882051282051282,
+      "grad_norm": 14.032716751098633,
+      "learning_rate": 6.383480344729903e-06,
+      "loss": 5.458187103271484,
+      "step": 1010
+    },
+    {
+      "epoch": 1.8913752913752915,
+      "grad_norm": 11.861150741577148,
+      "learning_rate": 6.344390385521534e-06,
+      "loss": 5.806562423706055,
+      "step": 1015
+    },
+    {
+      "epoch": 1.9006993006993007,
+      "grad_norm": 14.015438079833984,
+      "learning_rate": 6.305211649601595e-06,
+      "loss": 5.2158203125,
+      "step": 1020
+    },
+    {
+      "epoch": 1.91002331002331,
+      "grad_norm": 15.672555923461914,
+      "learning_rate": 6.265946724134782e-06,
+      "loss": 5.469319915771484,
+      "step": 1025
+    },
+    {
+      "epoch": 1.9193473193473194,
+      "grad_norm": 13.891338348388672,
+      "learning_rate": 6.226598201977299e-06,
+      "loss": 5.495826721191406,
+      "step": 1030
+    },
+    {
+      "epoch": 1.9286713286713286,
+      "grad_norm": 12.441986083984375,
+      "learning_rate": 6.187168681505666e-06,
+      "loss": 5.466391372680664,
+      "step": 1035
+    },
+    {
+      "epoch": 1.9379953379953379,
+      "grad_norm": 13.905447006225586,
+      "learning_rate": 6.1476607664451105e-06,
+      "loss": 5.558423233032227,
+      "step": 1040
+    },
+    {
+      "epoch": 1.9473193473193473,
+      "grad_norm": 14.817802429199219,
+      "learning_rate": 6.1080770656976444e-06,
+      "loss": 5.50462646484375,
+      "step": 1045
+    },
+    {
+      "epoch": 1.9566433566433568,
+      "grad_norm": 14.293425559997559,
+      "learning_rate": 6.068420193169779e-06,
+      "loss": 5.567475128173828,
+      "step": 1050
+    },
+    {
+      "epoch": 1.9566433566433568,
+      "eval_com_accuracy": 0.26,
+      "eval_rew_accuracy": 0.384,
+      "step": 1050
+    },
+    {
+      "epoch": 1.965967365967366,
+      "grad_norm": 13.285664558410645,
+      "learning_rate": 6.0286927675999205e-06,
+      "loss": 5.3831031799316404,
+      "step": 1055
+    },
+    {
+      "epoch": 1.9752913752913752,
+      "grad_norm": 12.634699821472168,
+      "learning_rate": 5.98889741238544e-06,
+      "loss": 5.4307403564453125,
+      "step": 1060
+    },
+    {
+      "epoch": 1.9846153846153847,
+      "grad_norm": 13.824172019958496,
+      "learning_rate": 5.949036755409432e-06,
+      "loss": 5.559865570068359,
+      "step": 1065
+    },
+    {
+      "epoch": 1.993939393939394,
+      "grad_norm": 14.028116226196289,
+      "learning_rate": 5.909113428867195e-06,
+      "loss": 5.506551361083984,
+      "step": 1070
+    },
+    {
+      "epoch": 2.001864801864802,
+      "grad_norm": 15.495360374450684,
+      "learning_rate": 5.869130069092401e-06,
+      "loss": 4.655789947509765,
+      "step": 1075
+    },
+    {
+      "epoch": 2.011188811188811,
+      "grad_norm": 14.094182014465332,
+      "learning_rate": 5.829089316383018e-06,
+      "loss": 5.403445434570313,
+      "step": 1080
+    },
+    {
+      "epoch": 2.0205128205128204,
+      "grad_norm": 13.709051132202148,
+      "learning_rate": 5.7889938148269445e-06,
+      "loss": 5.5024055480957035,
+      "step": 1085
+    },
+    {
+      "epoch": 2.02983682983683,
+      "grad_norm": 13.718189239501953,
+      "learning_rate": 5.748846212127421e-06,
+      "loss": 5.404336547851562,
+      "step": 1090
+    },
+    {
+      "epoch": 2.0391608391608393,
+      "grad_norm": 14.198376655578613,
+      "learning_rate": 5.708649159428181e-06,
+      "loss": 5.498197937011719,
+      "step": 1095
+    },
+    {
+      "epoch": 2.0484848484848484,
+      "grad_norm": 12.810367584228516,
+      "learning_rate": 5.668405311138382e-06,
+      "loss": 5.639860534667969,
+      "step": 1100
+    },
+    {
+      "epoch": 2.0484848484848484,
+      "eval_com_accuracy": 0.264,
+      "eval_rew_accuracy": 0.392,
+      "step": 1100
+    },
+    {
+      "epoch": 2.057808857808858,
+      "grad_norm": 13.649062156677246,
+      "learning_rate": 5.628117324757326e-06,
+      "loss": 5.567802429199219,
+      "step": 1105
+    },
+    {
+      "epoch": 2.0671328671328673,
+      "grad_norm": 15.412623405456543,
+      "learning_rate": 5.587787860698975e-06,
+      "loss": 5.585782623291015,
+      "step": 1110
+    },
+    {
+      "epoch": 2.0764568764568763,
+      "grad_norm": 13.465250968933105,
+      "learning_rate": 5.547419582116259e-06,
+      "loss": 5.357003021240234,
+      "step": 1115
+    },
+    {
+      "epoch": 2.0857808857808857,
+      "grad_norm": 14.019501686096191,
+      "learning_rate": 5.507015154725226e-06,
+      "loss": 5.621536636352539,
+      "step": 1120
+    },
+    {
+      "epoch": 2.095104895104895,
+      "grad_norm": 15.479727745056152,
+      "learning_rate": 5.466577246629006e-06,
+      "loss": 5.355072784423828,
+      "step": 1125
+    },
+    {
+      "epoch": 2.1044289044289046,
+      "grad_norm": 13.079503059387207,
+      "learning_rate": 5.426108528141627e-06,
+      "loss": 5.450039672851562,
+      "step": 1130
+    },
+    {
+      "epoch": 2.1137529137529136,
+      "grad_norm": 16.310596466064453,
+      "learning_rate": 5.385611671611676e-06,
+      "loss": 5.359495162963867,
+      "step": 1135
+    },
+    {
+      "epoch": 2.123076923076923,
+      "grad_norm": 15.107529640197754,
+      "learning_rate": 5.345089351245834e-06,
+      "loss": 5.5543464660644535,
+      "step": 1140
+    },
+    {
+      "epoch": 2.1324009324009325,
+      "grad_norm": 14.484854698181152,
+      "learning_rate": 5.304544242932288e-06,
+      "loss": 5.393303680419922,
+      "step": 1145
+    },
+    {
+      "epoch": 2.1417249417249415,
+      "grad_norm": 13.286919593811035,
+      "learning_rate": 5.26397902406402e-06,
+      "loss": 5.56396369934082,
+      "step": 1150
+    },
+    {
+      "epoch": 2.1417249417249415,
+      "eval_com_accuracy": 0.272,
+      "eval_rew_accuracy": 0.372,
+      "step": 1150
+    },
+    {
+      "epoch": 2.151048951048951,
+      "grad_norm": 14.213571548461914,
+      "learning_rate": 5.223396373362013e-06,
+      "loss": 5.63990478515625,
+      "step": 1155
+    },
+    {
+      "epoch": 2.1603729603729604,
+      "grad_norm": 14.445123672485352,
+      "learning_rate": 5.182798970698361e-06,
+      "loss": 5.836676788330078,
+      "step": 1160
+    },
+    {
+      "epoch": 2.16969696969697,
+      "grad_norm": 14.63541030883789,
+      "learning_rate": 5.142189496919302e-06,
+      "loss": 5.595640182495117,
+      "step": 1165
+    },
+    {
+      "epoch": 2.179020979020979,
+      "grad_norm": 13.398268699645996,
+      "learning_rate": 5.101570633668185e-06,
+      "loss": 5.657070159912109,
+      "step": 1170
+    },
+    {
+      "epoch": 2.1883449883449884,
+      "grad_norm": 13.70476245880127,
+      "learning_rate": 5.060945063208399e-06,
+      "loss": 5.337051010131836,
+      "step": 1175
+    },
+    {
+      "epoch": 2.197668997668998,
+      "grad_norm": 13.812256813049316,
+      "learning_rate": 5.02031546824624e-06,
+      "loss": 5.393893051147461,
+      "step": 1180
+    },
+    {
+      "epoch": 2.206993006993007,
+      "grad_norm": 14.151784896850586,
+      "learning_rate": 4.979684531753761e-06,
+      "loss": 5.489350128173828,
+      "step": 1185
+    },
+    {
+      "epoch": 2.2163170163170163,
+      "grad_norm": 14.830500602722168,
+      "learning_rate": 4.9390549367916004e-06,
+      "loss": 5.4119518280029295,
+      "step": 1190
+    },
+    {
+      "epoch": 2.2256410256410257,
+      "grad_norm": 14.537983894348145,
+      "learning_rate": 4.898429366331815e-06,
+      "loss": 5.393406677246094,
+      "step": 1195
+    },
+    {
+      "epoch": 2.234965034965035,
+      "grad_norm": 12.427031517028809,
+      "learning_rate": 4.857810503080701e-06,
+      "loss": 5.359439849853516,
+      "step": 1200
+    },
+    {
+      "epoch": 2.234965034965035,
+      "eval_com_accuracy": 0.256,
+      "eval_rew_accuracy": 0.368,
+      "step": 1200
+    },
+    {
+      "epoch": 2.244289044289044,
+      "grad_norm": 13.743568420410156,
+      "learning_rate": 4.81720102930164e-06,
+      "loss": 5.248456954956055,
+      "step": 1205
+    },
+    {
+      "epoch": 2.2536130536130536,
+      "grad_norm": 15.313255310058594,
+      "learning_rate": 4.776603626637988e-06,
+      "loss": 5.48780517578125,
+      "step": 1210
+    },
+    {
+      "epoch": 2.262937062937063,
+      "grad_norm": 14.26356029510498,
+      "learning_rate": 4.736020975935981e-06,
+      "loss": 5.3895729064941404,
+      "step": 1215
+    },
+    {
+      "epoch": 2.272261072261072,
+      "grad_norm": 14.363275527954102,
+      "learning_rate": 4.695455757067712e-06,
+      "loss": 5.294971084594726,
+      "step": 1220
+    },
+    {
+      "epoch": 2.2815850815850816,
+      "grad_norm": 14.772239685058594,
+      "learning_rate": 4.6549106487541666e-06,
+      "loss": 5.744273376464844,
+      "step": 1225
+    },
+    {
+      "epoch": 2.290909090909091,
+      "grad_norm": 13.530016899108887,
+      "learning_rate": 4.614388328388327e-06,
+      "loss": 5.28902587890625,
+      "step": 1230
+    },
+    {
+      "epoch": 2.3002331002331005,
+      "grad_norm": 13.482331275939941,
+      "learning_rate": 4.573891471858375e-06,
+      "loss": 5.445513153076172,
+      "step": 1235
+    },
+    {
+      "epoch": 2.3095571095571095,
+      "grad_norm": 13.693358421325684,
+      "learning_rate": 4.533422753370995e-06,
+      "loss": 5.312195205688477,
+      "step": 1240
+    },
+    {
+      "epoch": 2.318881118881119,
+      "grad_norm": 14.169803619384766,
+      "learning_rate": 4.492984845274774e-06,
+      "loss": 5.5245819091796875,
+      "step": 1245
+    },
+    {
+      "epoch": 2.3282051282051284,
+      "grad_norm": 26.758739471435547,
+      "learning_rate": 4.4525804178837425e-06,
+      "loss": 5.587922668457031,
+      "step": 1250
+    },
+    {
+      "epoch": 2.3282051282051284,
+      "eval_com_accuracy": 0.26,
+      "eval_rew_accuracy": 0.336,
+      "step": 1250
+    },
+    {
+      "epoch": 2.3375291375291374,
+      "grad_norm": 15.485888481140137,
+      "learning_rate": 4.412212139301027e-06,
+      "loss": 5.402701187133789,
+      "step": 1255
+    },
+    {
+      "epoch": 2.346853146853147,
+      "grad_norm": 16.002370834350586,
+      "learning_rate": 4.371882675242674e-06,
+      "loss": 5.409712219238282,
+      "step": 1260
+    },
+    {
+      "epoch": 2.3561771561771563,
+      "grad_norm": 16.42519760131836,
+      "learning_rate": 4.331594688861619e-06,
+      "loss": 5.624014282226563,
+      "step": 1265
+    },
+    {
+      "epoch": 2.3655011655011657,
+      "grad_norm": 16.416702270507812,
+      "learning_rate": 4.291350840571821e-06,
+      "loss": 5.3733673095703125,
+      "step": 1270
+    },
+    {
+      "epoch": 2.3748251748251747,
+      "grad_norm": 13.191758155822754,
+      "learning_rate": 4.251153787872579e-06,
+      "loss": 5.316907501220703,
+      "step": 1275
+    },
+    {
+      "epoch": 2.384149184149184,
+      "grad_norm": 14.480209350585938,
+      "learning_rate": 4.211006185173056e-06,
+      "loss": 5.500685501098633,
+      "step": 1280
+    },
+    {
+      "epoch": 2.3934731934731936,
+      "grad_norm": 15.625597953796387,
+      "learning_rate": 4.170910683616985e-06,
+      "loss": 5.321009826660156,
+      "step": 1285
+    },
+    {
+      "epoch": 2.4027972027972027,
+      "grad_norm": 13.146830558776855,
+      "learning_rate": 4.130869930907599e-06,
+      "loss": 5.352607345581054,
+      "step": 1290
+    },
+    {
+      "epoch": 2.412121212121212,
+      "grad_norm": 13.852392196655273,
+      "learning_rate": 4.090886571132807e-06,
+      "loss": 5.265444946289063,
+      "step": 1295
+    },
+    {
+      "epoch": 2.4214452214452216,
+      "grad_norm": 15.4178466796875,
+      "learning_rate": 4.050963244590571e-06,
+      "loss": 5.396755599975586,
+      "step": 1300
+    },
+    {
+      "epoch": 2.4214452214452216,
+      "eval_com_accuracy": 0.26,
+      "eval_rew_accuracy": 0.372,
+      "step": 1300
+    },
+    {
+      "epoch": 2.430769230769231,
+      "grad_norm": 14.469849586486816,
+      "learning_rate": 4.011102587614563e-06,
+      "loss": 5.381956100463867,
+      "step": 1305
+    },
+    {
+      "epoch": 2.44009324009324,
+      "grad_norm": 14.326638221740723,
+      "learning_rate": 3.97130723240008e-06,
+      "loss": 5.4776451110839846,
+      "step": 1310
+    },
+    {
+      "epoch": 2.4494172494172495,
+      "grad_norm": 13.75566577911377,
+      "learning_rate": 3.9315798068302214e-06,
+      "loss": 5.416980743408203,
+      "step": 1315
+    },
+    {
+      "epoch": 2.458741258741259,
+      "grad_norm": 13.778778076171875,
+      "learning_rate": 3.891922934302356e-06,
+      "loss": 5.254080200195313,
+      "step": 1320
+    },
+    {
+      "epoch": 2.468065268065268,
+      "grad_norm": 13.458685874938965,
+      "learning_rate": 3.852339233554891e-06,
+      "loss": 5.272119140625,
+      "step": 1325
+    },
+    {
+      "epoch": 2.4773892773892774,
+      "grad_norm": 15.419684410095215,
+      "learning_rate": 3.812831318494335e-06,
+      "loss": 5.448992538452148,
+      "step": 1330
+    },
+    {
+      "epoch": 2.486713286713287,
+      "grad_norm": 13.448263168334961,
+      "learning_rate": 3.773401798022701e-06,
+      "loss": 5.6390220642089846,
+      "step": 1335
+    },
+    {
+      "epoch": 2.496037296037296,
+      "grad_norm": 18.160816192626953,
+      "learning_rate": 3.7340532758652217e-06,
+      "loss": 5.633776092529297,
+      "step": 1340
+    },
+    {
+      "epoch": 2.5053613053613053,
+      "grad_norm": 20.26373863220215,
+      "learning_rate": 3.6947883503984037e-06,
+      "loss": 5.412459182739258,
+      "step": 1345
+    },
+    {
+      "epoch": 2.5146853146853148,
+      "grad_norm": 15.404874801635742,
+      "learning_rate": 3.655609614478467e-06,
+      "loss": 5.253081893920898,
+      "step": 1350
+    },
+    {
+      "epoch": 2.5146853146853148,
+      "eval_com_accuracy": 0.26,
+      "eval_rew_accuracy": 0.36,
+      "step": 1350
+    },
+    {
+      "epoch": 2.5240093240093238,
+      "grad_norm": 14.836410522460938,
+      "learning_rate": 3.6165196552701e-06,
+      "loss": 5.615703582763672,
+      "step": 1355
+    },
+    {
+      "epoch": 2.533333333333333,
+      "grad_norm": 13.233349800109863,
+      "learning_rate": 3.577521054075631e-06,
+      "loss": 5.440980911254883,
+      "step": 1360
+    },
+    {
+      "epoch": 2.5426573426573427,
+      "grad_norm": 14.456931114196777,
+      "learning_rate": 3.538616386164575e-06,
+      "loss": 5.664167022705078,
+      "step": 1365
+    },
+    {
+      "epoch": 2.551981351981352,
+      "grad_norm": 14.075837135314941,
+      "learning_rate": 3.4998082206035606e-06,
+      "loss": 5.418301010131836,
+      "step": 1370
+    },
+    {
+      "epoch": 2.5613053613053616,
+      "grad_norm": 13.708165168762207,
+      "learning_rate": 3.4610991200867006e-06,
+      "loss": 5.395311737060547,
+      "step": 1375
+    },
+    {
+      "epoch": 2.5706293706293706,
+      "grad_norm": 13.784794807434082,
+      "learning_rate": 3.4224916407663484e-06,
+      "loss": 5.168692016601563,
+      "step": 1380
+    },
+    {
+      "epoch": 2.57995337995338,
+      "grad_norm": 13.721598625183105,
+      "learning_rate": 3.3839883320843125e-06,
+      "loss": 5.642356109619141,
+      "step": 1385
+    },
+    {
+      "epoch": 2.589277389277389,
+      "grad_norm": 14.451462745666504,
+      "learning_rate": 3.3455917366035058e-06,
+      "loss": 5.229490661621094,
+      "step": 1390
+    },
+    {
+      "epoch": 2.5986013986013985,
+      "grad_norm": 13.549321174621582,
+      "learning_rate": 3.307304389840036e-06,
+      "loss": 5.413081359863281,
+      "step": 1395
+    },
+    {
+      "epoch": 2.607925407925408,
+      "grad_norm": 14.325213432312012,
+      "learning_rate": 3.2691288200957826e-06,
+      "loss": 5.3188224792480465,
+      "step": 1400
+    },
+    {
+      "epoch": 2.607925407925408,
+      "eval_com_accuracy": 0.26,
+      "eval_rew_accuracy": 0.348,
+      "step": 1400
+    },
+    {
+      "epoch": 2.6172494172494174,
+      "grad_norm": 13.765854835510254,
+      "learning_rate": 3.2310675482914444e-06,
+      "loss": 5.554658508300781,
+      "step": 1405
+    },
+    {
+      "epoch": 2.626573426573427,
+      "grad_norm": 14.936039924621582,
+      "learning_rate": 3.1931230878000586e-06,
+      "loss": 5.512041473388672,
+      "step": 1410
+    },
+    {
+      "epoch": 2.635897435897436,
+      "grad_norm": 16.414018630981445,
+      "learning_rate": 3.155297944281036e-06,
+      "loss": 5.256567001342773,
+      "step": 1415
+    },
+    {
+      "epoch": 2.6452214452214453,
+      "grad_norm": 13.871015548706055,
+      "learning_rate": 3.1175946155147064e-06,
+      "loss": 5.480235290527344,
+      "step": 1420
+    },
+    {
+      "epoch": 2.6545454545454543,
+      "grad_norm": 13.673002243041992,
+      "learning_rate": 3.0800155912373696e-06,
+      "loss": 5.272347259521484,
+      "step": 1425
+    },
+    {
+      "epoch": 2.6638694638694638,
+      "grad_norm": 15.32669734954834,
+      "learning_rate": 3.042563352976884e-06,
+      "loss": 5.1510368347167965,
+      "step": 1430
+    },
+    {
+      "epoch": 2.6731934731934732,
+      "grad_norm": 13.43411922454834,
+      "learning_rate": 3.005240373888812e-06,
+      "loss": 5.218346405029297,
+      "step": 1435
+    },
+    {
+      "epoch": 2.6825174825174827,
+      "grad_norm": 13.674540519714355,
+      "learning_rate": 2.9680491185930877e-06,
+      "loss": 5.514698791503906,
+      "step": 1440
+    },
+    {
+      "epoch": 2.6918414918414917,
+      "grad_norm": 14.424562454223633,
+      "learning_rate": 2.9309920430112825e-06,
+      "loss": 5.328684997558594,
+      "step": 1445
+    },
+    {
+      "epoch": 2.701165501165501,
+      "grad_norm": 14.4567232131958,
+      "learning_rate": 2.8940715942044204e-06,
+      "loss": 5.470043182373047,
+      "step": 1450
+    },
+    {
+      "epoch": 2.701165501165501,
+      "eval_com_accuracy": 0.244,
+      "eval_rew_accuracy": 0.364,
+      "step": 1450
+    },
+    {
+      "epoch": 2.7104895104895106,
+      "grad_norm": 14.073286056518555,
+      "learning_rate": 2.8572902102113788e-06,
+      "loss": 5.592926025390625,
+      "step": 1455
+    },
+    {
+      "epoch": 2.7198135198135196,
+      "grad_norm": 15.481256484985352,
+      "learning_rate": 2.820650319887911e-06,
+      "loss": 5.42064094543457,
+      "step": 1460
+    },
+    {
+      "epoch": 2.729137529137529,
+      "grad_norm": 13.80669116973877,
+      "learning_rate": 2.784154342746246e-06,
+      "loss": 5.514656066894531,
+      "step": 1465
+    },
+    {
+      "epoch": 2.7384615384615385,
+      "grad_norm": 15.876877784729004,
+      "learning_rate": 2.747804688795311e-06,
+      "loss": 5.645168304443359,
+      "step": 1470
+    },
+    {
+      "epoch": 2.747785547785548,
+      "grad_norm": 14.138867378234863,
+      "learning_rate": 2.7116037583816e-06,
+      "loss": 5.482178497314453,
+      "step": 1475
+    },
+    {
+      "epoch": 2.757109557109557,
+      "grad_norm": 13.498207092285156,
+      "learning_rate": 2.6755539420306565e-06,
+      "loss": 5.522509384155273,
+      "step": 1480
+    },
+    {
+      "epoch": 2.7664335664335664,
+      "grad_norm": 12.95719051361084,
+      "learning_rate": 2.6396576202892176e-06,
+      "loss": 5.203734970092773,
+      "step": 1485
+    },
+    {
+      "epoch": 2.775757575757576,
+      "grad_norm": 14.70195198059082,
+      "learning_rate": 2.603917163568021e-06,
+      "loss": 5.4792030334472654,
+      "step": 1490
+    },
+    {
+      "epoch": 2.785081585081585,
+      "grad_norm": 16.21117401123047,
+      "learning_rate": 2.5683349319852647e-06,
+      "loss": 5.337226867675781,
+      "step": 1495
+    },
+    {
+      "epoch": 2.7944055944055943,
+      "grad_norm": 15.80867862701416,
+      "learning_rate": 2.5329132752107675e-06,
+      "loss": 5.441407012939453,
+      "step": 1500
+    },
+    {
+      "epoch": 2.7944055944055943,
+      "eval_com_accuracy": 0.252,
+      "eval_rew_accuracy": 0.364,
+      "step": 1500
+    },
+    {
+      "epoch": 2.803729603729604,
+      "grad_norm": 17.126665115356445,
+      "learning_rate": 2.497654532310799e-06,
+      "loss": 5.372947692871094,
+      "step": 1505
+    },
+    {
+      "epoch": 2.8130536130536132,
+      "grad_norm": 14.965190887451172,
+      "learning_rate": 2.4625610315936267e-06,
+      "loss": 5.591331481933594,
+      "step": 1510
+    },
+    {
+      "epoch": 2.8223776223776222,
+      "grad_norm": 13.920053482055664,
+      "learning_rate": 2.427635090455766e-06,
+      "loss": 5.381235504150391,
+      "step": 1515
+    },
+    {
+      "epoch": 2.8317016317016317,
+      "grad_norm": 14.707354545593262,
+      "learning_rate": 2.3928790152289443e-06,
+      "loss": 5.375308990478516,
+      "step": 1520
+    },
+    {
+      "epoch": 2.841025641025641,
+      "grad_norm": 13.846839904785156,
+      "learning_rate": 2.358295101027807e-06,
+      "loss": 5.24262580871582,
+      "step": 1525
+    },
+    {
+      "epoch": 2.85034965034965,
+      "grad_norm": 17.17586898803711,
+      "learning_rate": 2.323885631598366e-06,
+      "loss": 5.628063583374024,
+      "step": 1530
+    },
+    {
+      "epoch": 2.8596736596736596,
+      "grad_norm": 14.309005737304688,
+      "learning_rate": 2.2896528791671807e-06,
+      "loss": 5.600907897949218,
+      "step": 1535
+    },
+    {
+      "epoch": 2.868997668997669,
+      "grad_norm": 18.62157440185547,
+      "learning_rate": 2.2555991042913177e-06,
+      "loss": 5.567650604248047,
+      "step": 1540
+    },
+    {
+      "epoch": 2.8783216783216785,
+      "grad_norm": 14.592391967773438,
+      "learning_rate": 2.221726555709079e-06,
+      "loss": 5.777060699462891,
+      "step": 1545
+    },
+    {
+      "epoch": 2.8876456876456875,
+      "grad_norm": 13.5873441696167,
+      "learning_rate": 2.188037470191502e-06,
+      "loss": 5.495726776123047,
+      "step": 1550
+    },
+    {
+      "epoch": 2.8876456876456875,
+      "eval_com_accuracy": 0.26,
+      "eval_rew_accuracy": 0.344,
+      "step": 1550
+    },
+    {
+      "epoch": 2.896969696969697,
+      "grad_norm": 13.62496280670166,
+      "learning_rate": 2.154534072394651e-06,
+      "loss": 5.313283920288086,
+      "step": 1555
+    },
+    {
+      "epoch": 2.9062937062937064,
+      "grad_norm": 14.277817726135254,
+      "learning_rate": 2.1212185747127235e-06,
+      "loss": 5.38345947265625,
+      "step": 1560
+    },
+    {
+      "epoch": 2.9156177156177154,
+      "grad_norm": 13.139540672302246,
+      "learning_rate": 2.0880931771319395e-06,
+      "loss": 5.48463134765625,
+      "step": 1565
+    },
+    {
+      "epoch": 2.924941724941725,
+      "grad_norm": 15.709213256835938,
+      "learning_rate": 2.055160067085283e-06,
+      "loss": 5.492628479003907,
+      "step": 1570
+    },
+    {
+      "epoch": 2.9342657342657343,
+      "grad_norm": 14.224997520446777,
+      "learning_rate": 2.0224214193080394e-06,
+      "loss": 5.611021423339844,
+      "step": 1575
+    },
+    {
+      "epoch": 2.943589743589744,
+      "grad_norm": 14.523746490478516,
+      "learning_rate": 1.989879395694194e-06,
+      "loss": 5.288306427001953,
+      "step": 1580
+    },
+    {
+      "epoch": 2.952913752913753,
+      "grad_norm": 14.50571346282959,
+      "learning_rate": 1.9575361451536772e-06,
+      "loss": 5.350891876220703,
+      "step": 1585
+    },
+    {
+      "epoch": 2.9622377622377623,
+      "grad_norm": 15.541993141174316,
+      "learning_rate": 1.925393803470447e-06,
+      "loss": 5.3829093933105465,
+      "step": 1590
+    },
+    {
+      "epoch": 2.9715617715617717,
+      "grad_norm": 14.223262786865234,
+      "learning_rate": 1.893454493161468e-06,
+      "loss": 5.267041397094727,
+      "step": 1595
+    },
+    {
+      "epoch": 2.9808857808857807,
+      "grad_norm": 16.193429946899414,
+      "learning_rate": 1.8617203233365427e-06,
+      "loss": 5.335537719726562,
+      "step": 1600
+    },
+    {
+      "epoch": 2.9808857808857807,
+      "eval_com_accuracy": 0.26,
+      "eval_rew_accuracy": 0.356,
+      "step": 1600
+    },
+    {
+      "epoch": 2.99020979020979,
+      "grad_norm": 13.891239166259766,
+      "learning_rate": 1.8301933895590362e-06,
+      "loss": 5.594977569580078,
+      "step": 1605
+    },
+    {
+      "epoch": 2.9995337995337996,
+      "grad_norm": 14.599571228027344,
+      "learning_rate": 1.7988757737074959e-06,
+      "loss": 5.364975357055664,
+      "step": 1610
+    },
+    {
+      "epoch": 3.0074592074592075,
+      "grad_norm": 14.290489196777344,
+      "learning_rate": 1.7677695438381831e-06,
+      "loss": 4.734904479980469,
+      "step": 1615
+    },
+    {
+      "epoch": 3.016783216783217,
+      "grad_norm": 17.73172950744629,
+      "learning_rate": 1.7368767540484965e-06,
+      "loss": 5.451268005371094,
+      "step": 1620
+    },
+    {
+      "epoch": 3.026107226107226,
+      "grad_norm": 16.327598571777344,
+      "learning_rate": 1.706199444341341e-06,
+      "loss": 5.246796035766602,
+      "step": 1625
+    },
+    {
+      "epoch": 3.0354312354312354,
+      "grad_norm": 14.45212459564209,
+      "learning_rate": 1.6757396404904087e-06,
+      "loss": 5.129561996459961,
+      "step": 1630
+    },
+    {
+      "epoch": 3.044755244755245,
+      "grad_norm": 19.204496383666992,
+      "learning_rate": 1.6454993539064075e-06,
+      "loss": 5.312553787231446,
+      "step": 1635
+    },
+    {
+      "epoch": 3.0540792540792543,
+      "grad_norm": 15.517951011657715,
+      "learning_rate": 1.6154805815042457e-06,
+      "loss": 5.416378402709961,
+      "step": 1640
+    },
+    {
+      "epoch": 3.0634032634032633,
+      "grad_norm": 15.498330116271973,
+      "learning_rate": 1.585685305571159e-06,
+      "loss": 5.478387451171875,
+      "step": 1645
+    },
+    {
+      "epoch": 3.0727272727272728,
+      "grad_norm": 13.686599731445312,
+      "learning_rate": 1.5561154936358069e-06,
+      "loss": 5.502795028686523,
+      "step": 1650
+    },
+    {
+      "epoch": 3.0727272727272728,
+      "eval_com_accuracy": 0.256,
+      "eval_rew_accuracy": 0.348,
+      "step": 1650
+    },
+    {
+      "epoch": 3.082051282051282,
+      "grad_norm": 15.222505569458008,
+      "learning_rate": 1.526773098338359e-06,
+      "loss": 5.323673248291016,
+      "step": 1655
+    },
+    {
+      "epoch": 3.091375291375291,
+      "grad_norm": 14.830018043518066,
+      "learning_rate": 1.4976600573015398e-06,
+      "loss": 5.360482406616211,
+      "step": 1660
+    },
+    {
+      "epoch": 3.1006993006993007,
+      "grad_norm": 19.069103240966797,
+      "learning_rate": 1.4687782930026833e-06,
+      "loss": 5.443345642089843,
+      "step": 1665
+    },
+    {
+      "epoch": 3.11002331002331,
+      "grad_norm": 25.908740997314453,
+      "learning_rate": 1.4401297126467884e-06,
+      "loss": 5.528006362915039,
+      "step": 1670
+    },
+    {
+      "epoch": 3.1193473193473196,
+      "grad_norm": 17.149354934692383,
+      "learning_rate": 1.411716208040566e-06,
+      "loss": 5.359912872314453,
+      "step": 1675
+    },
+    {
+      "epoch": 3.1286713286713286,
+      "grad_norm": 13.994576454162598,
+      "learning_rate": 1.3835396554675179e-06,
+      "loss": 5.480604553222657,
+      "step": 1680
+    },
+    {
+      "epoch": 3.137995337995338,
+      "grad_norm": 13.94813346862793,
+      "learning_rate": 1.3556019155640416e-06,
+      "loss": 5.426166534423828,
+      "step": 1685
+    },
+    {
+      "epoch": 3.1473193473193475,
+      "grad_norm": 18.871957778930664,
+      "learning_rate": 1.327904833196556e-06,
+      "loss": 5.287055969238281,
+      "step": 1690
+    },
+    {
+      "epoch": 3.1566433566433565,
+      "grad_norm": 14.017885208129883,
+      "learning_rate": 1.3004502373396821e-06,
+      "loss": 5.530019378662109,
+      "step": 1695
+    },
+    {
+      "epoch": 3.165967365967366,
+      "grad_norm": 14.46150016784668,
+      "learning_rate": 1.273239940955459e-06,
+      "loss": 5.436450958251953,
+      "step": 1700
+    },
+    {
+      "epoch": 3.165967365967366,
+      "eval_com_accuracy": 0.252,
+      "eval_rew_accuracy": 0.344,
+      "step": 1700
+    },
+    {
+      "epoch": 3.1752913752913754,
+      "grad_norm": 14.199810028076172,
+      "learning_rate": 1.246275740873631e-06,
+      "loss": 5.5229228973388675,
+      "step": 1705
+    },
+    {
+      "epoch": 3.184615384615385,
+      "grad_norm": 14.207246780395508,
+      "learning_rate": 1.2195594176729963e-06,
+      "loss": 5.065603637695313,
+      "step": 1710
+    },
+    {
+      "epoch": 3.193939393939394,
+      "grad_norm": 14.264204978942871,
+      "learning_rate": 1.1930927355638189e-06,
+      "loss": 5.147642517089844,
+      "step": 1715
+    },
+    {
+      "epoch": 3.2032634032634033,
+      "grad_norm": 15.168913841247559,
+      "learning_rate": 1.1668774422713336e-06,
+      "loss": 5.345111846923828,
+      "step": 1720
+    },
+    {
+      "epoch": 3.2125874125874128,
+      "grad_norm": 13.536885261535645,
+      "learning_rate": 1.140915268920339e-06,
+      "loss": 5.612071228027344,
+      "step": 1725
+    },
+    {
+      "epoch": 3.2219114219114218,
+      "grad_norm": 14.979183197021484,
+      "learning_rate": 1.1152079299208724e-06,
+      "loss": 5.372085189819336,
+      "step": 1730
+    },
+    {
+      "epoch": 3.231235431235431,
+      "grad_norm": 15.19270133972168,
+      "learning_rate": 1.0897571228550097e-06,
+      "loss": 5.345587158203125,
+      "step": 1735
+    },
+    {
+      "epoch": 3.2405594405594407,
+      "grad_norm": 13.973429679870605,
+      "learning_rate": 1.0645645283647616e-06,
+      "loss": 5.468335723876953,
+      "step": 1740
+    },
+    {
+      "epoch": 3.2498834498834497,
+      "grad_norm": 14.033141136169434,
+      "learning_rate": 1.0396318100410868e-06,
+      "loss": 5.231144714355469,
+      "step": 1745
+    },
+    {
+      "epoch": 3.259207459207459,
+      "grad_norm": 13.720804214477539,
+      "learning_rate": 1.0149606143140484e-06,
+      "loss": 5.193565750122071,
+      "step": 1750
+    },
+    {
+      "epoch": 3.259207459207459,
+      "eval_com_accuracy": 0.26,
+      "eval_rew_accuracy": 0.356,
+      "step": 1750
+    },
+    {
+      "epoch": 3.2685314685314686,
+      "grad_norm": 14.3617582321167,
+      "learning_rate": 9.905525703440815e-07,
+      "loss": 5.266670989990234,
+      "step": 1755
+    },
+    {
+      "epoch": 3.277855477855478,
+      "grad_norm": 14.0305757522583,
+      "learning_rate": 9.664092899144156e-07,
+      "loss": 5.1834564208984375,
+      "step": 1760
+    },
+    {
+      "epoch": 3.287179487179487,
+      "grad_norm": 13.830121040344238,
+      "learning_rate": 9.425323673246461e-07,
+      "loss": 5.506520462036133,
+      "step": 1765
+    },
+    {
+      "epoch": 3.2965034965034965,
+      "grad_norm": 14.764183044433594,
+      "learning_rate": 9.189233792854424e-07,
+      "loss": 5.4712566375732425,
+      "step": 1770
+    },
+    {
+      "epoch": 3.305827505827506,
+      "grad_norm": 13.076425552368164,
+      "learning_rate": 8.955838848144449e-07,
+      "loss": 5.39610481262207,
+      "step": 1775
+    },
+    {
+      "epoch": 3.315151515151515,
+      "grad_norm": 24.025074005126953,
+      "learning_rate": 8.725154251333012e-07,
+      "loss": 5.462095642089844,
+      "step": 1780
+    },
+    {
+      "epoch": 3.3244755244755244,
+      "grad_norm": 15.28011417388916,
+      "learning_rate": 8.49719523565904e-07,
+      "loss": 5.121075057983399,
+      "step": 1785
+    },
+    {
+      "epoch": 3.333799533799534,
+      "grad_norm": 13.63133716583252,
+      "learning_rate": 8.271976854377861e-07,
+      "loss": 5.500334930419922,
+      "step": 1790
+    },
+    {
+      "epoch": 3.3431235431235433,
+      "grad_norm": 15.102256774902344,
+      "learning_rate": 8.049513979767304e-07,
+      "loss": 5.435873794555664,
+      "step": 1795
+    },
+    {
+      "epoch": 3.3524475524475523,
+      "grad_norm": 16.562448501586914,
+      "learning_rate": 7.829821302145485e-07,
+      "loss": 5.118262100219726,
+      "step": 1800
+    },
+    {
+      "epoch": 3.3524475524475523,
+      "eval_com_accuracy": 0.236,
+      "eval_rew_accuracy": 0.344,
+      "step": 1800
+    },
+    {
+      "epoch": 3.361771561771562,
+      "grad_norm": 15.783401489257812,
+      "learning_rate": 7.612913328900784e-07,
+      "loss": 5.411731719970703,
+      "step": 1805
+    },
+    {
+      "epoch": 3.3710955710955712,
+      "grad_norm": 15.205412864685059,
+      "learning_rate": 7.398804383533886e-07,
+      "loss": 5.4931282043457035,
+      "step": 1810
+    },
+    {
+      "epoch": 3.3804195804195802,
+      "grad_norm": 15.897871971130371,
+      "learning_rate": 7.187508604711851e-07,
+      "loss": 5.3865409851074215,
+      "step": 1815
+    },
+    {
+      "epoch": 3.3897435897435897,
+      "grad_norm": 14.67023754119873,
+      "learning_rate": 6.979039945334543e-07,
+      "loss": 5.469716644287109,
+      "step": 1820
+    },
+    {
+      "epoch": 3.399067599067599,
+      "grad_norm": 14.824798583984375,
+      "learning_rate": 6.77341217161322e-07,
+      "loss": 5.260548782348633,
+      "step": 1825
+    },
+    {
+      "epoch": 3.408391608391608,
+      "grad_norm": 15.626496315002441,
+      "learning_rate": 6.570638862161449e-07,
+      "loss": 5.215817642211914,
+      "step": 1830
+    },
+    {
+      "epoch": 3.4177156177156176,
+      "grad_norm": 97.59706115722656,
+      "learning_rate": 6.370733407098517e-07,
+      "loss": 5.424036026000977,
+      "step": 1835
+    },
+    {
+      "epoch": 3.427039627039627,
+      "grad_norm": 14.15798282623291,
+      "learning_rate": 6.173709007165158e-07,
+      "loss": 5.179172515869141,
+      "step": 1840
+    },
+    {
+      "epoch": 3.4363636363636365,
+      "grad_norm": 13.659977912902832,
+      "learning_rate": 5.979578672851843e-07,
+      "loss": 5.505921936035156,
+      "step": 1845
+    },
+    {
+      "epoch": 3.4456876456876455,
+      "grad_norm": 13.775196075439453,
+      "learning_rate": 5.788355223539698e-07,
+      "loss": 5.567048645019531,
+      "step": 1850
+    },
+    {
+      "epoch": 3.4456876456876455,
+      "eval_com_accuracy": 0.248,
+      "eval_rew_accuracy": 0.348,
+      "step": 1850
+    },
+    {
+      "epoch": 3.455011655011655,
+      "grad_norm": 13.840899467468262,
+      "learning_rate": 5.600051286653884e-07,
+      "loss": 5.582321548461914,
+      "step": 1855
+    },
+    {
+      "epoch": 3.4643356643356644,
+      "grad_norm": 17.72808837890625,
+      "learning_rate": 5.414679296829806e-07,
+      "loss": 5.374624633789063,
+      "step": 1860
+    },
+    {
+      "epoch": 3.4736596736596734,
+      "grad_norm": 13.779251098632812,
+      "learning_rate": 5.232251495091989e-07,
+      "loss": 5.559833908081055,
+      "step": 1865
+    },
+    {
+      "epoch": 3.482983682983683,
+      "grad_norm": 15.004316329956055,
+      "learning_rate": 5.052779928045737e-07,
+      "loss": 5.440023422241211,
+      "step": 1870
+    },
+    {
+      "epoch": 3.4923076923076923,
+      "grad_norm": 14.249396324157715,
+      "learning_rate": 4.87627644708163e-07,
+      "loss": 5.40099868774414,
+      "step": 1875
+    },
+    {
+      "epoch": 3.501631701631702,
+      "grad_norm": 14.191717147827148,
+      "learning_rate": 4.7027527075929e-07,
+      "loss": 5.396835708618164,
+      "step": 1880
+    },
+    {
+      "epoch": 3.510955710955711,
+      "grad_norm": 14.725996971130371,
+      "learning_rate": 4.532220168205798e-07,
+      "loss": 5.4339447021484375,
+      "step": 1885
+    },
+    {
+      "epoch": 3.5202797202797202,
+      "grad_norm": 13.92712688446045,
+      "learning_rate": 4.364690090022938e-07,
+      "loss": 5.301970291137695,
+      "step": 1890
+    },
+    {
+      "epoch": 3.5296037296037297,
+      "grad_norm": 16.015522003173828,
+      "learning_rate": 4.2001735358796316e-07,
+      "loss": 5.519898986816406,
+      "step": 1895
+    },
+    {
+      "epoch": 3.5389277389277387,
+      "grad_norm": 13.642130851745605,
+      "learning_rate": 4.0386813696133564e-07,
+      "loss": 5.369457244873047,
+      "step": 1900
+    },
+    {
+      "epoch": 3.5389277389277387,
+      "eval_com_accuracy": 0.252,
+      "eval_rew_accuracy": 0.348,
+      "step": 1900
+    },
+    {
+      "epoch": 3.548251748251748,
+      "grad_norm": 14.768367767333984,
+      "learning_rate": 3.8802242553464096e-07,
+      "loss": 5.300347900390625,
+      "step": 1905
+    },
+    {
+      "epoch": 3.5575757575757576,
+      "grad_norm": 13.686610221862793,
+      "learning_rate": 3.7248126567816454e-07,
+      "loss": 5.458771514892578,
+      "step": 1910
+    },
+    {
+      "epoch": 3.566899766899767,
+      "grad_norm": 14.974004745483398,
+      "learning_rate": 3.572456836511551e-07,
+      "loss": 5.505842208862305,
+      "step": 1915
+    },
+    {
+      "epoch": 3.576223776223776,
+      "grad_norm": 19.55851936340332,
+      "learning_rate": 3.4231668553405316e-07,
+      "loss": 5.585261535644531,
+      "step": 1920
+    },
+    {
+      "epoch": 3.5855477855477855,
+      "grad_norm": 13.967267990112305,
+      "learning_rate": 3.276952571620556e-07,
+      "loss": 5.267148971557617,
+      "step": 1925
+    },
+    {
+      "epoch": 3.594871794871795,
+      "grad_norm": 16.652421951293945,
+      "learning_rate": 3.133823640600137e-07,
+      "loss": 5.403562927246094,
+      "step": 1930
+    },
+    {
+      "epoch": 3.604195804195804,
+      "grad_norm": 14.056427955627441,
+      "learning_rate": 2.9937895137868046e-07,
+      "loss": 5.545249176025391,
+      "step": 1935
+    },
+    {
+      "epoch": 3.6135198135198134,
+      "grad_norm": 14.697785377502441,
+      "learning_rate": 2.8568594383229067e-07,
+      "loss": 5.691058349609375,
+      "step": 1940
+    },
+    {
+      "epoch": 3.622843822843823,
+      "grad_norm": 14.570009231567383,
+      "learning_rate": 2.723042456375036e-07,
+      "loss": 5.311951446533203,
+      "step": 1945
+    },
+    {
+      "epoch": 3.6321678321678323,
+      "grad_norm": 14.328276634216309,
+      "learning_rate": 2.592347404536888e-07,
+      "loss": 5.5065868377685545,
+      "step": 1950
+    },
+    {
+      "epoch": 3.6321678321678323,
+      "eval_com_accuracy": 0.26,
+      "eval_rew_accuracy": 0.348,
+      "step": 1950
+    },
+    {
+      "epoch": 3.6414918414918414,
+      "grad_norm": 16.094568252563477,
+      "learning_rate": 2.4647829132457446e-07,
+      "loss": 5.409658813476563,
+      "step": 1955
+    },
+    {
+      "epoch": 3.650815850815851,
+      "grad_norm": 13.347336769104004,
+      "learning_rate": 2.340357406212601e-07,
+      "loss": 5.344178009033203,
+      "step": 1960
+    },
+    {
+      "epoch": 3.6601398601398603,
+      "grad_norm": 13.901823043823242,
+      "learning_rate": 2.2190790998658561e-07,
+      "loss": 5.574573135375976,
+      "step": 1965
+    },
+    {
+      "epoch": 3.6694638694638693,
+      "grad_norm": 12.91339111328125,
+      "learning_rate": 2.1009560028087627e-07,
+      "loss": 5.605550384521484,
+      "step": 1970
+    },
+    {
+      "epoch": 3.6787878787878787,
+      "grad_norm": 14.216018676757812,
+      "learning_rate": 1.985995915290595e-07,
+      "loss": 5.501739501953125,
+      "step": 1975
+    },
+    {
+      "epoch": 3.688111888111888,
+      "grad_norm": 13.566910743713379,
+      "learning_rate": 1.8742064286915329e-07,
+      "loss": 5.589977645874024,
+      "step": 1980
+    },
+    {
+      "epoch": 3.6974358974358976,
+      "grad_norm": 13.63309383392334,
+      "learning_rate": 1.7655949250213743e-07,
+      "loss": 5.244425201416016,
+      "step": 1985
+    },
+    {
+      "epoch": 3.7067599067599066,
+      "grad_norm": 14.691123008728027,
+      "learning_rate": 1.660168576432092e-07,
+      "loss": 5.428138351440429,
+      "step": 1990
+    },
+    {
+      "epoch": 3.716083916083916,
+      "grad_norm": 13.5573148727417,
+      "learning_rate": 1.5579343447441663e-07,
+      "loss": 5.335743713378906,
+      "step": 1995
+    },
+    {
+      "epoch": 3.7254079254079255,
+      "grad_norm": 16.859336853027344,
+      "learning_rate": 1.458898980986917e-07,
+      "loss": 5.435336303710938,
+      "step": 2000
+    },
+    {
+      "epoch": 3.7254079254079255,
+      "eval_com_accuracy": 0.26,
+      "eval_rew_accuracy": 0.352,
+      "step": 2000
+    },
+    {
+      "epoch": 3.7347319347319345,
+      "grad_norm": 14.824188232421875,
+      "learning_rate": 1.363069024952668e-07,
+      "loss": 5.184574890136719,
+      "step": 2005
+    },
+    {
+      "epoch": 3.744055944055944,
+      "grad_norm": 14.50694465637207,
+      "learning_rate": 1.2704508047649e-07,
+      "loss": 5.404220962524414,
+      "step": 2010
+    },
+    {
+      "epoch": 3.7533799533799534,
+      "grad_norm": 13.273224830627441,
+      "learning_rate": 1.1810504364603737e-07,
+      "loss": 5.528971099853516,
+      "step": 2015
+    },
+    {
+      "epoch": 3.762703962703963,
+      "grad_norm": 14.738325119018555,
+      "learning_rate": 1.094873823585263e-07,
+      "loss": 5.249575424194336,
+      "step": 2020
+    },
+    {
+      "epoch": 3.772027972027972,
+      "grad_norm": 27.92183494567871,
+      "learning_rate": 1.011926656805301e-07,
+      "loss": 5.55212631225586,
+      "step": 2025
+    },
+    {
+      "epoch": 3.7813519813519814,
+      "grad_norm": 14.745790481567383,
+      "learning_rate": 9.322144135300137e-08,
+      "loss": 5.203295135498047,
+      "step": 2030
+    },
+    {
+      "epoch": 3.790675990675991,
+      "grad_norm": 14.700698852539062,
+      "learning_rate": 8.557423575510037e-08,
+      "loss": 5.222881317138672,
+      "step": 2035
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 14.061659812927246,
+      "learning_rate": 7.825155386943784e-08,
+      "loss": 5.428164291381836,
+      "step": 2040
+    },
+    {
+      "epoch": 3.8093240093240093,
+      "grad_norm": 14.425374984741211,
+      "learning_rate": 7.125387924872552e-08,
+      "loss": 5.418885040283203,
+      "step": 2045
+    },
+    {
+      "epoch": 3.8186480186480187,
+      "grad_norm": 14.355052947998047,
+      "learning_rate": 6.458167398384896e-08,
+      "loss": 5.363260650634766,
+      "step": 2050
+    },
+    {
+      "epoch": 3.8186480186480187,
+      "eval_com_accuracy": 0.264,
+      "eval_rew_accuracy": 0.328,
+      "step": 2050
+    },
+    {
+      "epoch": 3.827972027972028,
+      "grad_norm": 14.03409481048584,
+      "learning_rate": 5.823537867334694e-08,
+      "loss": 5.503321838378906,
+      "step": 2055
+    },
+    {
+      "epoch": 3.837296037296037,
+      "grad_norm": 14.262378692626953,
+      "learning_rate": 5.221541239432415e-08,
+      "loss": 5.388937759399414,
+      "step": 2060
+    },
+    {
+      "epoch": 3.8466200466200466,
+      "grad_norm": 14.72974967956543,
+      "learning_rate": 4.65221726747711e-08,
+      "loss": 5.465239334106445,
+      "step": 2065
+    },
+    {
+      "epoch": 3.855944055944056,
+      "grad_norm": 20.170591354370117,
+      "learning_rate": 4.11560354673185e-08,
+      "loss": 5.374723052978515,
+      "step": 2070
+    },
+    {
+      "epoch": 3.865268065268065,
+      "grad_norm": 14.229021072387695,
+      "learning_rate": 3.611735512440706e-08,
+      "loss": 5.41649169921875,
+      "step": 2075
+    },
+    {
+      "epoch": 3.8745920745920746,
+      "grad_norm": 14.37116527557373,
+      "learning_rate": 3.1406464374890144e-08,
+      "loss": 5.885035705566406,
+      "step": 2080
+    },
+    {
+      "epoch": 3.883916083916084,
+      "grad_norm": 13.708508491516113,
+      "learning_rate": 2.7023674302061875e-08,
+      "loss": 5.252231597900391,
+      "step": 2085
+    },
+    {
+      "epoch": 3.8932400932400935,
+      "grad_norm": 13.693293571472168,
+      "learning_rate": 2.296927432311358e-08,
+      "loss": 5.494184494018555,
+      "step": 2090
+    },
+    {
+      "epoch": 3.9025641025641025,
+      "grad_norm": 14.291114807128906,
+      "learning_rate": 1.9243532170023504e-08,
+      "loss": 5.161873245239258,
+      "step": 2095
+    },
+    {
+      "epoch": 3.911888111888112,
+      "grad_norm": 15.632975578308105,
+      "learning_rate": 1.584669387187765e-08,
+      "loss": 5.218763732910157,
+      "step": 2100
+    },
+    {
+      "epoch": 3.911888111888112,
+      "eval_com_accuracy": 0.256,
+      "eval_rew_accuracy": 0.348,
+      "step": 2100
+    },
+    {
+      "epoch": 3.9212121212121214,
+      "grad_norm": 13.507603645324707,
+      "learning_rate": 1.2778983738620521e-08,
+      "loss": 5.341087341308594,
+      "step": 2105
+    },
+    {
+      "epoch": 3.9305361305361304,
+      "grad_norm": 15.44629955291748,
+      "learning_rate": 1.0040604346245319e-08,
+      "loss": 5.3995521545410154,
+      "step": 2110
+    },
+    {
+      "epoch": 3.93986013986014,
+      "grad_norm": 16.452760696411133,
+      "learning_rate": 7.631736523416867e-09,
+      "loss": 5.533244323730469,
+      "step": 2115
+    },
+    {
+      "epoch": 3.9491841491841493,
+      "grad_norm": 13.974261283874512,
+      "learning_rate": 5.552539339528373e-09,
+      "loss": 5.49418830871582,
+      "step": 2120
+    },
+    {
+      "epoch": 3.9585081585081587,
+      "grad_norm": 14.339764595031738,
+      "learning_rate": 3.803150094200403e-09,
+      "loss": 5.264930343627929,
+      "step": 2125
+    },
+    {
+      "epoch": 3.9678321678321677,
+      "grad_norm": 15.614706993103027,
+      "learning_rate": 2.3836843082108987e-09,
+      "loss": 5.345928192138672,
+      "step": 2130
+    },
+    {
+      "epoch": 3.977156177156177,
+      "grad_norm": 13.78399658203125,
+      "learning_rate": 1.2942357158701734e-09,
+      "loss": 5.354546356201172,
+      "step": 2135
+    },
+    {
+      "epoch": 3.986480186480186,
+      "grad_norm": 15.413246154785156,
+      "learning_rate": 5.348762588286427e-10,
+      "loss": 5.361167907714844,
+      "step": 2140
+    },
+    {
+      "epoch": 3.9958041958041957,
+      "grad_norm": 13.466837882995605,
+      "learning_rate": 1.0565608132728778e-10,
+      "loss": 5.534800338745117,
+      "step": 2145
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 2148,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "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": 0.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-2148/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b6ca094dca60606350f327809d196d6faa38b81cdd561be5d924e82822f09231
+size 5201

checkpoint-537/chat_template.jinja

@@ -0,0 +1,120 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {%- if messages[0].content is string %}
+            {{- messages[0].content }}
+        {%- else %}
+            {%- for content in messages[0].content %}
+                {%- if 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '\n\n' }}
+    {%- endif %}
+    {{- "# 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>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\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" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' }}
+        {%- if messages[0].content is string %}
+            {{- messages[0].content }}
+        {%- else %}
+            {%- for content in messages[0].content %}
+                {%- if 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- for message in messages %}
+    {%- if message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- else %}
+            {%- for content in message.content %}
+                {%- if content.type == 'image' or 'image' in content or 'image_url' in content %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                    {%- if add_vision_id %}Picture {{ image_count.value }}: {% endif -%}
+                    <|vision_start|><|image_pad|><|vision_end|>
+                {%- elif content.type == 'video' or 'video' in content %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                    {%- if add_vision_id %}Video {{ video_count.value }}: {% endif -%}
+                    <|vision_start|><|video_pad|><|vision_end|>
+                {%- elif 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role + '\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- else %}
+            {%- for content_item in message.content %}
+                {%- if 'text' in content_item %}
+                    {{- content_item.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and message.content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- else %}
+            {%- for content in message.content %}
+                {%- if content.type == 'image' or 'image' in content or 'image_url' in content %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                    {%- if add_vision_id %}Picture {{ image_count.value }}: {% endif -%}
+                    <|vision_start|><|image_pad|><|vision_end|>
+                {%- elif content.type == 'video' or 'video' in content %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                    {%- if add_vision_id %}Video {{ video_count.value }}: {% endif -%}
+                    <|vision_start|><|video_pad|><|vision_end|>
+                {%- elif 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}

checkpoint-537/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4229d15b27ff28c9f42527317e0829421f5cdb985fc8f221b1eb944128d6f59e
+size 4885870044

checkpoint-537/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4cbcf2233432062b1fa0687277ea1a8485b14d8ad6abcd77c80693607bebb7f7
+size 8527467778

checkpoint-537/processor_config.json

@@ -0,0 +1,63 @@
+{
+  "image_processor": {
+    "data_format": "channels_first",
+    "do_convert_rgb": true,
+    "do_normalize": true,
+    "do_rescale": true,
+    "do_resize": true,
+    "image_mean": [
+      0.5,
+      0.5,
+      0.5
+    ],
+    "image_processor_type": "Qwen2VLImageProcessorFast",
+    "image_std": [
+      0.5,
+      0.5,
+      0.5
+    ],
+    "merge_size": 2,
+    "patch_size": 16,
+    "resample": 3,
+    "rescale_factor": 0.00392156862745098,
+    "size": {
+      "longest_edge": 16777216,
+      "shortest_edge": 65536
+    },
+    "temporal_patch_size": 2
+  },
+  "processor_class": "Qwen3VLProcessor",
+  "video_processor": {
+    "data_format": "channels_first",
+    "default_to_square": true,
+    "do_convert_rgb": true,
+    "do_normalize": true,
+    "do_rescale": true,
+    "do_resize": true,
+    "do_sample_frames": true,
+    "fps": 2,
+    "image_mean": [
+      0.5,
+      0.5,
+      0.5
+    ],
+    "image_std": [
+      0.5,
+      0.5,
+      0.5
+    ],
+    "max_frames": 768,
+    "merge_size": 2,
+    "min_frames": 4,
+    "patch_size": 16,
+    "resample": 3,
+    "rescale_factor": 0.00392156862745098,
+    "return_metadata": false,
+    "size": {
+      "longest_edge": 25165824,
+      "shortest_edge": 4096
+    },
+    "temporal_patch_size": 2,
+    "video_processor_type": "Qwen3VLVideoProcessor"
+  }
+}

checkpoint-537/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:659b1cdee2219458dd84ce6a632a595465680b8080e5c44bd600ff97eca8d752
+size 15429

checkpoint-537/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:86accf27064cdd503053e90476a6bd10de333d4ff0594535ad55ea13a473c91d
+size 15429

checkpoint-537/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:18ca8d714ef40be035404c1957b5a4dee84e1f43980408393f8aa710552ee6f6
+size 15429

checkpoint-537/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2cfdebe99e40accc9c9d8f09c63136a14abda997d9b501969ec8e16e9d183179
+size 15429

checkpoint-537/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1f7fe019d7b5610635791b26be1530136e4922aac9ed7000c5d71f6ed8f98dac
+size 1465