Upload webshop_sr_null SR bugfix checkpoint

.gitattributes

@@ -33,3 +33,5 @@ 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-2856/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

added_tokens.json

@@ -0,0 +1,28 @@
+{
+  "</think>": 151668,
+  "</tool_call>": 151658,
+  "</tool_response>": 151666,
+  "<think>": 151667,
+  "<tool_call>": 151657,
+  "<tool_response>": 151665,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

all_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 2.0,
+    "total_flos": 137777518706688.0,
+    "train_loss": 0.5963595523720696,
+    "train_runtime": 4888.1806,
+    "train_samples_per_second": 9.347,
+    "train_steps_per_second": 0.584
+}

chat_template.jinja

@@ -0,0 +1,89 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {{- messages[0].content + '\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' + messages[0].content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" and message.content is string and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}
+        {%- set ns.multi_step_tool = false %}
+        {%- set ns.last_query_index = index %}
+    {%- endif %}
+{%- endfor %}
+{%- for message in messages %}
+    {%- if message.content is string %}
+        {%- set content = message.content %}
+    {%- else %}
+        {%- set content = '' %}
+    {%- endif %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {%- if loop.last or (not loop.last and reasoning_content) %}
+                {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content.strip('\n') + '\n</think>\n\n' + content.lstrip('\n') }}
+            {%- else %}
+                {{- '<|im_start|>' + message.role + '\n' + content }}
+            {%- endif %}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and 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' }}
+        {{- content }}
+        {{- '\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' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}

checkpoint-2856/added_tokens.json

@@ -0,0 +1,28 @@
+{
+  "</think>": 151668,
+  "</tool_call>": 151658,
+  "</tool_response>": 151666,
+  "<think>": 151667,
+  "<tool_call>": 151657,
+  "<tool_response>": 151665,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

checkpoint-2856/chat_template.jinja

@@ -0,0 +1,89 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {{- messages[0].content + '\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' + messages[0].content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" and message.content is string and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}
+        {%- set ns.multi_step_tool = false %}
+        {%- set ns.last_query_index = index %}
+    {%- endif %}
+{%- endfor %}
+{%- for message in messages %}
+    {%- if message.content is string %}
+        {%- set content = message.content %}
+    {%- else %}
+        {%- set content = '' %}
+    {%- endif %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {%- if loop.last or (not loop.last and reasoning_content) %}
+                {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content.strip('\n') + '\n</think>\n\n' + content.lstrip('\n') }}
+            {%- else %}
+                {{- '<|im_start|>' + message.role + '\n' + content }}
+            {%- endif %}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and 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' }}
+        {{- content }}
+        {{- '\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' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}

checkpoint-2856/config.json

@@ -0,0 +1,30 @@
+{
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "eos_token_id": 151645,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 12288,
+  "max_position_embeddings": 40960,
+  "max_window_layers": 36,
+  "model_type": "qwen3",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 36,
+  "num_key_value_heads": 8,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000,
+  "sliding_window": null,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.52.4",
+  "use_cache": false,
+  "use_sliding_window": false,
+  "vocab_size": 151936
+}

checkpoint-2856/generation_config.json

@@ -0,0 +1,13 @@
+{
+  "bos_token_id": 151643,
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "pad_token_id": 151643,
+  "temperature": 0.6,
+  "top_k": 20,
+  "top_p": 0.95,
+  "transformers_version": "4.52.4"
+}

checkpoint-2856/model-00001-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2b0be61836f0b007bb5c5a491006a2ce49f3e66f9cfb97966ecdcd04a56bdd9f
+size 4902257696

checkpoint-2856/model-00002-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:93f965306bec10edef5a70fc848a366439e394f5a19060decf2ba944eabd210d
+size 4915960368

checkpoint-2856/model-00003-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8798485503eeb6b367126953d2070f4c2fcb539e7039345f0885810d650d76f3
+size 4983068496

checkpoint-2856/model-00004-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4d18f328ca8831716c919c95343c70004df1d8d9c5c1ef33c4005b8f4d47d571
+size 1580230264

checkpoint-2856/model.safetensors.index.json

@@ -0,0 +1,406 @@
+{
+  "metadata": {
+    "total_size": 16381470720
+  },
+  "weight_map": {
+    "lm_head.weight": "model-00004-of-00004.safetensors",
+    "model.embed_tokens.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.22.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.34.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.34.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.34.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.34.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.34.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.34.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.34.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.34.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.34.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.34.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.34.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.35.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.k_norm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.35.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.q_norm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.35.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.9.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.norm.weight": "model-00004-of-00004.safetensors"
+  }
+}

checkpoint-2856/special_tokens_map.json

@@ -0,0 +1,31 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

checkpoint-2856/tokenizer.json

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

checkpoint-2856/tokenizer_config.json

@@ -0,0 +1,240 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151665": {
+      "content": "<tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151666": {
+      "content": "</tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151667": {
+      "content": "<think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151668": {
+      "content": "</think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "padding_side": "right",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

checkpoint-2856/trainer_state.json

@@ -0,0 +1,2029 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 2856,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0070028011204481795,
+      "grad_norm": 54.822365459213835,
+      "learning_rate": 3.1468531468531473e-07,
+      "loss": 2.9142,
+      "step": 10
+    },
+    {
+      "epoch": 0.014005602240896359,
+      "grad_norm": 43.140784575015786,
+      "learning_rate": 6.643356643356644e-07,
+      "loss": 2.6364,
+      "step": 20
+    },
+    {
+      "epoch": 0.02100840336134454,
+      "grad_norm": 20.52836873662029,
+      "learning_rate": 1.013986013986014e-06,
+      "loss": 2.0887,
+      "step": 30
+    },
+    {
+      "epoch": 0.028011204481792718,
+      "grad_norm": 7.296293308558612,
+      "learning_rate": 1.3636363636363636e-06,
+      "loss": 1.593,
+      "step": 40
+    },
+    {
+      "epoch": 0.0350140056022409,
+      "grad_norm": 5.69690111508328,
+      "learning_rate": 1.7132867132867134e-06,
+      "loss": 1.4625,
+      "step": 50
+    },
+    {
+      "epoch": 0.04201680672268908,
+      "grad_norm": 4.407284596083797,
+      "learning_rate": 2.0629370629370634e-06,
+      "loss": 1.231,
+      "step": 60
+    },
+    {
+      "epoch": 0.049019607843137254,
+      "grad_norm": 4.514294110845002,
+      "learning_rate": 2.4125874125874128e-06,
+      "loss": 1.1051,
+      "step": 70
+    },
+    {
+      "epoch": 0.056022408963585436,
+      "grad_norm": 4.664107810659101,
+      "learning_rate": 2.762237762237762e-06,
+      "loss": 0.9814,
+      "step": 80
+    },
+    {
+      "epoch": 0.06302521008403361,
+      "grad_norm": 3.983542706015,
+      "learning_rate": 3.1118881118881123e-06,
+      "loss": 0.9429,
+      "step": 90
+    },
+    {
+      "epoch": 0.0700280112044818,
+      "grad_norm": 3.763711194865439,
+      "learning_rate": 3.4615384615384617e-06,
+      "loss": 0.8549,
+      "step": 100
+    },
+    {
+      "epoch": 0.07703081232492998,
+      "grad_norm": 5.299639736915794,
+      "learning_rate": 3.811188811188811e-06,
+      "loss": 0.9123,
+      "step": 110
+    },
+    {
+      "epoch": 0.08403361344537816,
+      "grad_norm": 5.376552790853205,
+      "learning_rate": 4.160839160839161e-06,
+      "loss": 0.7164,
+      "step": 120
+    },
+    {
+      "epoch": 0.09103641456582633,
+      "grad_norm": 3.979790265000713,
+      "learning_rate": 4.510489510489511e-06,
+      "loss": 0.8358,
+      "step": 130
+    },
+    {
+      "epoch": 0.09803921568627451,
+      "grad_norm": 3.2180158802612904,
+      "learning_rate": 4.86013986013986e-06,
+      "loss": 0.7863,
+      "step": 140
+    },
+    {
+      "epoch": 0.10504201680672269,
+      "grad_norm": 3.653854239138897,
+      "learning_rate": 5.20979020979021e-06,
+      "loss": 0.7784,
+      "step": 150
+    },
+    {
+      "epoch": 0.11204481792717087,
+      "grad_norm": 3.731091024980675,
+      "learning_rate": 5.5594405594405596e-06,
+      "loss": 0.696,
+      "step": 160
+    },
+    {
+      "epoch": 0.11904761904761904,
+      "grad_norm": 3.371981147503621,
+      "learning_rate": 5.90909090909091e-06,
+      "loss": 0.7731,
+      "step": 170
+    },
+    {
+      "epoch": 0.12605042016806722,
+      "grad_norm": 3.4539776878177544,
+      "learning_rate": 6.258741258741259e-06,
+      "loss": 0.7214,
+      "step": 180
+    },
+    {
+      "epoch": 0.1330532212885154,
+      "grad_norm": 4.764005243860567,
+      "learning_rate": 6.608391608391609e-06,
+      "loss": 0.7684,
+      "step": 190
+    },
+    {
+      "epoch": 0.1400560224089636,
+      "grad_norm": 3.4106980736508237,
+      "learning_rate": 6.958041958041959e-06,
+      "loss": 0.7845,
+      "step": 200
+    },
+    {
+      "epoch": 0.14705882352941177,
+      "grad_norm": 3.5675277403879213,
+      "learning_rate": 7.307692307692308e-06,
+      "loss": 0.6937,
+      "step": 210
+    },
+    {
+      "epoch": 0.15406162464985995,
+      "grad_norm": 3.6190398138061615,
+      "learning_rate": 7.657342657342658e-06,
+      "loss": 0.8302,
+      "step": 220
+    },
+    {
+      "epoch": 0.16106442577030813,
+      "grad_norm": 3.5569990424945925,
+      "learning_rate": 8.006993006993008e-06,
+      "loss": 0.7539,
+      "step": 230
+    },
+    {
+      "epoch": 0.16806722689075632,
+      "grad_norm": 2.957808482275135,
+      "learning_rate": 8.356643356643356e-06,
+      "loss": 0.7641,
+      "step": 240
+    },
+    {
+      "epoch": 0.17507002801120447,
+      "grad_norm": 2.701665888444842,
+      "learning_rate": 8.706293706293708e-06,
+      "loss": 0.7131,
+      "step": 250
+    },
+    {
+      "epoch": 0.18207282913165265,
+      "grad_norm": 3.582843229917075,
+      "learning_rate": 9.055944055944057e-06,
+      "loss": 0.595,
+      "step": 260
+    },
+    {
+      "epoch": 0.18907563025210083,
+      "grad_norm": 3.453144473645395,
+      "learning_rate": 9.405594405594406e-06,
+      "loss": 0.7676,
+      "step": 270
+    },
+    {
+      "epoch": 0.19607843137254902,
+      "grad_norm": 3.3312933939710114,
+      "learning_rate": 9.755244755244755e-06,
+      "loss": 0.7019,
+      "step": 280
+    },
+    {
+      "epoch": 0.2030812324929972,
+      "grad_norm": 3.7248118142373494,
+      "learning_rate": 9.999966378620396e-06,
+      "loss": 0.6575,
+      "step": 290
+    },
+    {
+      "epoch": 0.21008403361344538,
+      "grad_norm": 3.8896895986556688,
+      "learning_rate": 9.999368677783691e-06,
+      "loss": 0.6984,
+      "step": 300
+    },
+    {
+      "epoch": 0.21708683473389356,
+      "grad_norm": 3.182135009379796,
+      "learning_rate": 9.998023937980683e-06,
+      "loss": 0.8306,
+      "step": 310
+    },
+    {
+      "epoch": 0.22408963585434175,
+      "grad_norm": 2.8416958229620395,
+      "learning_rate": 9.995932360151348e-06,
+      "loss": 0.784,
+      "step": 320
+    },
+    {
+      "epoch": 0.23109243697478993,
+      "grad_norm": 3.150294362261809,
+      "learning_rate": 9.993094256833178e-06,
+      "loss": 0.7085,
+      "step": 330
+    },
+    {
+      "epoch": 0.23809523809523808,
+      "grad_norm": 3.0028637962745806,
+      "learning_rate": 9.989510052114473e-06,
+      "loss": 0.6222,
+      "step": 340
+    },
+    {
+      "epoch": 0.24509803921568626,
+      "grad_norm": 2.845426871309238,
+      "learning_rate": 9.985180281570976e-06,
+      "loss": 0.7224,
+      "step": 350
+    },
+    {
+      "epoch": 0.25210084033613445,
+      "grad_norm": 3.050098939646489,
+      "learning_rate": 9.980105592185838e-06,
+      "loss": 0.7097,
+      "step": 360
+    },
+    {
+      "epoch": 0.25910364145658266,
+      "grad_norm": 2.6020917053511763,
+      "learning_rate": 9.97428674225295e-06,
+      "loss": 0.7017,
+      "step": 370
+    },
+    {
+      "epoch": 0.2661064425770308,
+      "grad_norm": 3.6019215435633014,
+      "learning_rate": 9.967724601263624e-06,
+      "loss": 0.7097,
+      "step": 380
+    },
+    {
+      "epoch": 0.27310924369747897,
+      "grad_norm": 4.104445886424035,
+      "learning_rate": 9.960420149776674e-06,
+      "loss": 0.7022,
+      "step": 390
+    },
+    {
+      "epoch": 0.2801120448179272,
+      "grad_norm": 2.7297806144746204,
+      "learning_rate": 9.952374479271894e-06,
+      "loss": 0.7519,
+      "step": 400
+    },
+    {
+      "epoch": 0.28711484593837533,
+      "grad_norm": 3.4868126922293694,
+      "learning_rate": 9.943588791986956e-06,
+      "loss": 0.6911,
+      "step": 410
+    },
+    {
+      "epoch": 0.29411764705882354,
+      "grad_norm": 2.777255111697869,
+      "learning_rate": 9.934064400737776e-06,
+      "loss": 0.7083,
+      "step": 420
+    },
+    {
+      "epoch": 0.3011204481792717,
+      "grad_norm": 2.667581928295271,
+      "learning_rate": 9.923802728722326e-06,
+      "loss": 0.7803,
+      "step": 430
+    },
+    {
+      "epoch": 0.3081232492997199,
+      "grad_norm": 2.974529058609242,
+      "learning_rate": 9.912805309307994e-06,
+      "loss": 0.7277,
+      "step": 440
+    },
+    {
+      "epoch": 0.31512605042016806,
+      "grad_norm": 2.8420842979615855,
+      "learning_rate": 9.901073785802433e-06,
+      "loss": 0.6997,
+      "step": 450
+    },
+    {
+      "epoch": 0.32212885154061627,
+      "grad_norm": 2.9643793871710926,
+      "learning_rate": 9.888609911208024e-06,
+      "loss": 0.6917,
+      "step": 460
+    },
+    {
+      "epoch": 0.3291316526610644,
+      "grad_norm": 2.6029717609338254,
+      "learning_rate": 9.875415547959926e-06,
+      "loss": 0.7042,
+      "step": 470
+    },
+    {
+      "epoch": 0.33613445378151263,
+      "grad_norm": 2.6417095655684886,
+      "learning_rate": 9.861492667647776e-06,
+      "loss": 0.6818,
+      "step": 480
+    },
+    {
+      "epoch": 0.3431372549019608,
+      "grad_norm": 2.912733389632055,
+      "learning_rate": 9.846843350721084e-06,
+      "loss": 0.7252,
+      "step": 490
+    },
+    {
+      "epoch": 0.35014005602240894,
+      "grad_norm": 2.4477536058478804,
+      "learning_rate": 9.831469786178359e-06,
+      "loss": 0.7347,
+      "step": 500
+    },
+    {
+      "epoch": 0.35714285714285715,
+      "grad_norm": 2.6814440911113353,
+      "learning_rate": 9.81537427124001e-06,
+      "loss": 0.6773,
+      "step": 510
+    },
+    {
+      "epoch": 0.3641456582633053,
+      "grad_norm": 2.6474828003551125,
+      "learning_rate": 9.798559211005083e-06,
+      "loss": 0.6925,
+      "step": 520
+    },
+    {
+      "epoch": 0.3711484593837535,
+      "grad_norm": 2.96009182699851,
+      "learning_rate": 9.781027118091879e-06,
+      "loss": 0.6953,
+      "step": 530
+    },
+    {
+      "epoch": 0.37815126050420167,
+      "grad_norm": 3.047523864473747,
+      "learning_rate": 9.76278061226249e-06,
+      "loss": 0.7292,
+      "step": 540
+    },
+    {
+      "epoch": 0.3851540616246499,
+      "grad_norm": 2.991785788396449,
+      "learning_rate": 9.74382242003135e-06,
+      "loss": 0.7072,
+      "step": 550
+    },
+    {
+      "epoch": 0.39215686274509803,
+      "grad_norm": 2.369241516996275,
+      "learning_rate": 9.724155374257816e-06,
+      "loss": 0.6132,
+      "step": 560
+    },
+    {
+      "epoch": 0.39915966386554624,
+      "grad_norm": 2.8220170208873023,
+      "learning_rate": 9.703782413722856e-06,
+      "loss": 0.7118,
+      "step": 570
+    },
+    {
+      "epoch": 0.4061624649859944,
+      "grad_norm": 3.037044208560417,
+      "learning_rate": 9.682706582689932e-06,
+      "loss": 0.6837,
+      "step": 580
+    },
+    {
+      "epoch": 0.41316526610644255,
+      "grad_norm": 2.820941643724944,
+      "learning_rate": 9.660931030450092e-06,
+      "loss": 0.7782,
+      "step": 590
+    },
+    {
+      "epoch": 0.42016806722689076,
+      "grad_norm": 2.5666349293329107,
+      "learning_rate": 9.638459010851389e-06,
+      "loss": 0.6569,
+      "step": 600
+    },
+    {
+      "epoch": 0.4271708683473389,
+      "grad_norm": 2.903374170685687,
+      "learning_rate": 9.615293881812666e-06,
+      "loss": 0.6597,
+      "step": 610
+    },
+    {
+      "epoch": 0.4341736694677871,
+      "grad_norm": 2.4510273143556467,
+      "learning_rate": 9.591439104821795e-06,
+      "loss": 0.7102,
+      "step": 620
+    },
+    {
+      "epoch": 0.4411764705882353,
+      "grad_norm": 2.724792508265338,
+      "learning_rate": 9.566898244418443e-06,
+      "loss": 0.7036,
+      "step": 630
+    },
+    {
+      "epoch": 0.4481792717086835,
+      "grad_norm": 2.1578153197233183,
+      "learning_rate": 9.541674967661424e-06,
+      "loss": 0.6853,
+      "step": 640
+    },
+    {
+      "epoch": 0.45518207282913165,
+      "grad_norm": 2.2982556480501977,
+      "learning_rate": 9.515773043580754e-06,
+      "loss": 0.6587,
+      "step": 650
+    },
+    {
+      "epoch": 0.46218487394957986,
+      "grad_norm": 2.482971468008142,
+      "learning_rate": 9.489196342614447e-06,
+      "loss": 0.6941,
+      "step": 660
+    },
+    {
+      "epoch": 0.469187675070028,
+      "grad_norm": 2.792625017068267,
+      "learning_rate": 9.461948836030179e-06,
+      "loss": 0.6527,
+      "step": 670
+    },
+    {
+      "epoch": 0.47619047619047616,
+      "grad_norm": 2.8095289499726586,
+      "learning_rate": 9.434034595331863e-06,
+      "loss": 0.5537,
+      "step": 680
+    },
+    {
+      "epoch": 0.4831932773109244,
+      "grad_norm": 2.6013084541081577,
+      "learning_rate": 9.405457791651272e-06,
+      "loss": 0.7455,
+      "step": 690
+    },
+    {
+      "epoch": 0.49019607843137253,
+      "grad_norm": 2.323733631506617,
+      "learning_rate": 9.376222695124748e-06,
+      "loss": 0.6781,
+      "step": 700
+    },
+    {
+      "epoch": 0.49719887955182074,
+      "grad_norm": 2.533641065336669,
+      "learning_rate": 9.346333674255132e-06,
+      "loss": 0.6807,
+      "step": 710
+    },
+    {
+      "epoch": 0.5042016806722689,
+      "grad_norm": 2.310754494783324,
+      "learning_rate": 9.315795195259003e-06,
+      "loss": 0.7009,
+      "step": 720
+    },
+    {
+      "epoch": 0.511204481792717,
+      "grad_norm": 2.247769256472491,
+      "learning_rate": 9.284611821399295e-06,
+      "loss": 0.6949,
+      "step": 730
+    },
+    {
+      "epoch": 0.5182072829131653,
+      "grad_norm": 2.855866205125908,
+      "learning_rate": 9.252788212303426e-06,
+      "loss": 0.695,
+      "step": 740
+    },
+    {
+      "epoch": 0.5252100840336135,
+      "grad_norm": 2.2123260966081144,
+      "learning_rate": 9.220329123267031e-06,
+      "loss": 0.6498,
+      "step": 750
+    },
+    {
+      "epoch": 0.5322128851540616,
+      "grad_norm": 3.2793392899369955,
+      "learning_rate": 9.187239404543387e-06,
+      "loss": 0.6761,
+      "step": 760
+    },
+    {
+      "epoch": 0.5392156862745098,
+      "grad_norm": 2.9849801807546132,
+      "learning_rate": 9.153524000618663e-06,
+      "loss": 0.6895,
+      "step": 770
+    },
+    {
+      "epoch": 0.5462184873949579,
+      "grad_norm": 2.229602688156502,
+      "learning_rate": 9.119187949473075e-06,
+      "loss": 0.6221,
+      "step": 780
+    },
+    {
+      "epoch": 0.5532212885154062,
+      "grad_norm": 2.1724377907001164,
+      "learning_rate": 9.08423638182808e-06,
+      "loss": 0.6749,
+      "step": 790
+    },
+    {
+      "epoch": 0.5602240896358543,
+      "grad_norm": 2.4988277913931216,
+      "learning_rate": 9.048674520379715e-06,
+      "loss": 0.7179,
+      "step": 800
+    },
+    {
+      "epoch": 0.5672268907563025,
+      "grad_norm": 2.4743732152259117,
+      "learning_rate": 9.012507679018177e-06,
+      "loss": 0.6926,
+      "step": 810
+    },
+    {
+      "epoch": 0.5742296918767507,
+      "grad_norm": 2.3405306027107775,
+      "learning_rate": 8.975741262033793e-06,
+      "loss": 0.7389,
+      "step": 820
+    },
+    {
+      "epoch": 0.5812324929971989,
+      "grad_norm": 2.272548602465369,
+      "learning_rate": 8.93838076330947e-06,
+      "loss": 0.6424,
+      "step": 830
+    },
+    {
+      "epoch": 0.5882352941176471,
+      "grad_norm": 2.0400401153554935,
+      "learning_rate": 8.900431765499773e-06,
+      "loss": 0.6971,
+      "step": 840
+    },
+    {
+      "epoch": 0.5952380952380952,
+      "grad_norm": 2.084177383415346,
+      "learning_rate": 8.861899939196713e-06,
+      "loss": 0.6155,
+      "step": 850
+    },
+    {
+      "epoch": 0.6022408963585434,
+      "grad_norm": 3.794945397573399,
+      "learning_rate": 8.82279104208242e-06,
+      "loss": 0.6244,
+      "step": 860
+    },
+    {
+      "epoch": 0.6092436974789915,
+      "grad_norm": 3.49165664633522,
+      "learning_rate": 8.783110918068784e-06,
+      "loss": 0.7264,
+      "step": 870
+    },
+    {
+      "epoch": 0.6162464985994398,
+      "grad_norm": 2.039969801192167,
+      "learning_rate": 8.742865496424228e-06,
+      "loss": 0.6768,
+      "step": 880
+    },
+    {
+      "epoch": 0.623249299719888,
+      "grad_norm": 2.335677247223129,
+      "learning_rate": 8.70206079088771e-06,
+      "loss": 0.5889,
+      "step": 890
+    },
+    {
+      "epoch": 0.6302521008403361,
+      "grad_norm": 2.060225160885774,
+      "learning_rate": 8.660702898770113e-06,
+      "loss": 0.6492,
+      "step": 900
+    },
+    {
+      "epoch": 0.6372549019607843,
+      "grad_norm": 2.798146335622228,
+      "learning_rate": 8.618798000043142e-06,
+      "loss": 0.6209,
+      "step": 910
+    },
+    {
+      "epoch": 0.6442577030812325,
+      "grad_norm": 2.8734777136564493,
+      "learning_rate": 8.576352356415876e-06,
+      "loss": 0.6407,
+      "step": 920
+    },
+    {
+      "epoch": 0.6512605042016807,
+      "grad_norm": 2.6762821835109154,
+      "learning_rate": 8.533372310399093e-06,
+      "loss": 0.6735,
+      "step": 930
+    },
+    {
+      "epoch": 0.6582633053221288,
+      "grad_norm": 2.0693419886641125,
+      "learning_rate": 8.48986428435754e-06,
+      "loss": 0.6023,
+      "step": 940
+    },
+    {
+      "epoch": 0.665266106442577,
+      "grad_norm": 2.5455745068080304,
+      "learning_rate": 8.445834779550257e-06,
+      "loss": 0.5726,
+      "step": 950
+    },
+    {
+      "epoch": 0.6722689075630253,
+      "grad_norm": 2.8427810866745427,
+      "learning_rate": 8.40129037515911e-06,
+      "loss": 0.6442,
+      "step": 960
+    },
+    {
+      "epoch": 0.6792717086834734,
+      "grad_norm": 2.338498753658426,
+      "learning_rate": 8.356237727305695e-06,
+      "loss": 0.6775,
+      "step": 970
+    },
+    {
+      "epoch": 0.6862745098039216,
+      "grad_norm": 2.035099542215418,
+      "learning_rate": 8.310683568056725e-06,
+      "loss": 0.6076,
+      "step": 980
+    },
+    {
+      "epoch": 0.6932773109243697,
+      "grad_norm": 2.2530856849283394,
+      "learning_rate": 8.264634704418095e-06,
+      "loss": 0.6701,
+      "step": 990
+    },
+    {
+      "epoch": 0.7002801120448179,
+      "grad_norm": 2.124044857812194,
+      "learning_rate": 8.218098017317715e-06,
+      "loss": 0.6211,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7072829131652661,
+      "grad_norm": 2.3812764227997265,
+      "learning_rate": 8.171080460577337e-06,
+      "loss": 0.6853,
+      "step": 1010
+    },
+    {
+      "epoch": 0.7142857142857143,
+      "grad_norm": 2.125710699424273,
+      "learning_rate": 8.123589059873445e-06,
+      "loss": 0.6516,
+      "step": 1020
+    },
+    {
+      "epoch": 0.7212885154061625,
+      "grad_norm": 2.0639925369405163,
+      "learning_rate": 8.075630911687451e-06,
+      "loss": 0.5853,
+      "step": 1030
+    },
+    {
+      "epoch": 0.7282913165266106,
+      "grad_norm": 2.4862110603900676,
+      "learning_rate": 8.027213182245289e-06,
+      "loss": 0.6107,
+      "step": 1040
+    },
+    {
+      "epoch": 0.7352941176470589,
+      "grad_norm": 1.601475291312852,
+      "learning_rate": 7.978343106446575e-06,
+      "loss": 0.7112,
+      "step": 1050
+    },
+    {
+      "epoch": 0.742296918767507,
+      "grad_norm": 2.871165956851756,
+      "learning_rate": 7.929027986783538e-06,
+      "loss": 0.6573,
+      "step": 1060
+    },
+    {
+      "epoch": 0.7492997198879552,
+      "grad_norm": 2.502491985241327,
+      "learning_rate": 7.879275192249822e-06,
+      "loss": 0.6383,
+      "step": 1070
+    },
+    {
+      "epoch": 0.7563025210084033,
+      "grad_norm": 2.3534352305326656,
+      "learning_rate": 7.829092157239369e-06,
+      "loss": 0.6707,
+      "step": 1080
+    },
+    {
+      "epoch": 0.7633053221288515,
+      "grad_norm": 2.346735216001232,
+      "learning_rate": 7.778486380435512e-06,
+      "loss": 0.5754,
+      "step": 1090
+    },
+    {
+      "epoch": 0.7703081232492998,
+      "grad_norm": 1.7069335554420118,
+      "learning_rate": 7.727465423690487e-06,
+      "loss": 0.6674,
+      "step": 1100
+    },
+    {
+      "epoch": 0.7773109243697479,
+      "grad_norm": 2.857304704073804,
+      "learning_rate": 7.676036910895475e-06,
+      "loss": 0.6436,
+      "step": 1110
+    },
+    {
+      "epoch": 0.7843137254901961,
+      "grad_norm": 2.704621305833269,
+      "learning_rate": 7.624208526841399e-06,
+      "loss": 0.5633,
+      "step": 1120
+    },
+    {
+      "epoch": 0.7913165266106442,
+      "grad_norm": 2.3446023311507243,
+      "learning_rate": 7.571988016070611e-06,
+      "loss": 0.6534,
+      "step": 1130
+    },
+    {
+      "epoch": 0.7983193277310925,
+      "grad_norm": 2.655699735724774,
+      "learning_rate": 7.519383181719644e-06,
+      "loss": 0.6152,
+      "step": 1140
+    },
+    {
+      "epoch": 0.8053221288515406,
+      "grad_norm": 2.5608889463976983,
+      "learning_rate": 7.466401884353227e-06,
+      "loss": 0.6775,
+      "step": 1150
+    },
+    {
+      "epoch": 0.8123249299719888,
+      "grad_norm": 2.3142838348541965,
+      "learning_rate": 7.413052040789692e-06,
+      "loss": 0.6514,
+      "step": 1160
+    },
+    {
+      "epoch": 0.819327731092437,
+      "grad_norm": 2.134983409939985,
+      "learning_rate": 7.359341622918006e-06,
+      "loss": 0.6194,
+      "step": 1170
+    },
+    {
+      "epoch": 0.8263305322128851,
+      "grad_norm": 2.4655749339506086,
+      "learning_rate": 7.305278656506547e-06,
+      "loss": 0.6381,
+      "step": 1180
+    },
+    {
+      "epoch": 0.8333333333333334,
+      "grad_norm": 1.90146176033135,
+      "learning_rate": 7.2508712200038435e-06,
+      "loss": 0.6059,
+      "step": 1190
+    },
+    {
+      "epoch": 0.8403361344537815,
+      "grad_norm": 2.7581098466146834,
+      "learning_rate": 7.19612744333144e-06,
+      "loss": 0.7107,
+      "step": 1200
+    },
+    {
+      "epoch": 0.8473389355742297,
+      "grad_norm": 2.013181558142786,
+      "learning_rate": 7.141055506669072e-06,
+      "loss": 0.6062,
+      "step": 1210
+    },
+    {
+      "epoch": 0.8543417366946778,
+      "grad_norm": 2.153881730897074,
+      "learning_rate": 7.0856636392323205e-06,
+      "loss": 0.5956,
+      "step": 1220
+    },
+    {
+      "epoch": 0.8613445378151261,
+      "grad_norm": 2.527280124270099,
+      "learning_rate": 7.0299601180429615e-06,
+      "loss": 0.5936,
+      "step": 1230
+    },
+    {
+      "epoch": 0.8683473389355743,
+      "grad_norm": 1.858330034501666,
+      "learning_rate": 6.973953266692143e-06,
+      "loss": 0.6314,
+      "step": 1240
+    },
+    {
+      "epoch": 0.8753501400560224,
+      "grad_norm": 2.3746989063300186,
+      "learning_rate": 6.91765145409663e-06,
+      "loss": 0.6908,
+      "step": 1250
+    },
+    {
+      "epoch": 0.8823529411764706,
+      "grad_norm": 2.2513458203129963,
+      "learning_rate": 6.861063093248264e-06,
+      "loss": 0.5709,
+      "step": 1260
+    },
+    {
+      "epoch": 0.8893557422969187,
+      "grad_norm": 2.5609479562736106,
+      "learning_rate": 6.804196639956828e-06,
+      "loss": 0.627,
+      "step": 1270
+    },
+    {
+      "epoch": 0.896358543417367,
+      "grad_norm": 2.4301069707639233,
+      "learning_rate": 6.747060591586533e-06,
+      "loss": 0.5996,
+      "step": 1280
+    },
+    {
+      "epoch": 0.9033613445378151,
+      "grad_norm": 2.610417413605633,
+      "learning_rate": 6.689663485786287e-06,
+      "loss": 0.6943,
+      "step": 1290
+    },
+    {
+      "epoch": 0.9103641456582633,
+      "grad_norm": 2.240908654296222,
+      "learning_rate": 6.632013899213934e-06,
+      "loss": 0.6311,
+      "step": 1300
+    },
+    {
+      "epoch": 0.9173669467787114,
+      "grad_norm": 2.3215803208239394,
+      "learning_rate": 6.574120446254672e-06,
+      "loss": 0.6081,
+      "step": 1310
+    },
+    {
+      "epoch": 0.9243697478991597,
+      "grad_norm": 1.7958005235142336,
+      "learning_rate": 6.5159917777338466e-06,
+      "loss": 0.6685,
+      "step": 1320
+    },
+    {
+      "epoch": 0.9313725490196079,
+      "grad_norm": 2.042136754036796,
+      "learning_rate": 6.457636579624278e-06,
+      "loss": 0.5457,
+      "step": 1330
+    },
+    {
+      "epoch": 0.938375350140056,
+      "grad_norm": 2.0200483342599367,
+      "learning_rate": 6.3990635717483404e-06,
+      "loss": 0.6981,
+      "step": 1340
+    },
+    {
+      "epoch": 0.9453781512605042,
+      "grad_norm": 2.3485088870526827,
+      "learning_rate": 6.340281506475003e-06,
+      "loss": 0.6165,
+      "step": 1350
+    },
+    {
+      "epoch": 0.9523809523809523,
+      "grad_norm": 2.0775112823384383,
+      "learning_rate": 6.281299167411975e-06,
+      "loss": 0.5656,
+      "step": 1360
+    },
+    {
+      "epoch": 0.9593837535014006,
+      "grad_norm": 2.115129503123767,
+      "learning_rate": 6.222125368093213e-06,
+      "loss": 0.6473,
+      "step": 1370
+    },
+    {
+      "epoch": 0.9663865546218487,
+      "grad_norm": 2.203147650702632,
+      "learning_rate": 6.162768950661945e-06,
+      "loss": 0.6279,
+      "step": 1380
+    },
+    {
+      "epoch": 0.9733893557422969,
+      "grad_norm": 2.617800940269768,
+      "learning_rate": 6.103238784549404e-06,
+      "loss": 0.5426,
+      "step": 1390
+    },
+    {
+      "epoch": 0.9803921568627451,
+      "grad_norm": 2.1058865000385754,
+      "learning_rate": 6.043543765149514e-06,
+      "loss": 0.6551,
+      "step": 1400
+    },
+    {
+      "epoch": 0.9873949579831933,
+      "grad_norm": 2.262404955697173,
+      "learning_rate": 5.983692812489679e-06,
+      "loss": 0.6524,
+      "step": 1410
+    },
+    {
+      "epoch": 0.9943977591036415,
+      "grad_norm": 2.1193412453943803,
+      "learning_rate": 5.923694869897879e-06,
+      "loss": 0.5787,
+      "step": 1420
+    },
+    {
+      "epoch": 1.0014005602240896,
+      "grad_norm": 2.0659116339074166,
+      "learning_rate": 5.863558902666318e-06,
+      "loss": 0.6191,
+      "step": 1430
+    },
+    {
+      "epoch": 1.0084033613445378,
+      "grad_norm": 2.13662088721353,
+      "learning_rate": 5.803293896711756e-06,
+      "loss": 0.4869,
+      "step": 1440
+    },
+    {
+      "epoch": 1.015406162464986,
+      "grad_norm": 2.132520413876192,
+      "learning_rate": 5.74290885723278e-06,
+      "loss": 0.4959,
+      "step": 1450
+    },
+    {
+      "epoch": 1.022408963585434,
+      "grad_norm": 1.862348432644305,
+      "learning_rate": 5.682412807364187e-06,
+      "loss": 0.4611,
+      "step": 1460
+    },
+    {
+      "epoch": 1.0294117647058822,
+      "grad_norm": 2.1396458126762825,
+      "learning_rate": 5.621814786828683e-06,
+      "loss": 0.4828,
+      "step": 1470
+    },
+    {
+      "epoch": 1.0364145658263306,
+      "grad_norm": 2.23299330212776,
+      "learning_rate": 5.5611238505861094e-06,
+      "loss": 0.4362,
+      "step": 1480
+    },
+    {
+      "epoch": 1.0434173669467788,
+      "grad_norm": 2.336766894179504,
+      "learning_rate": 5.5003490674804e-06,
+      "loss": 0.4811,
+      "step": 1490
+    },
+    {
+      "epoch": 1.050420168067227,
+      "grad_norm": 2.2053948018429126,
+      "learning_rate": 5.439499518884433e-06,
+      "loss": 0.4525,
+      "step": 1500
+    },
+    {
+      "epoch": 1.057422969187675,
+      "grad_norm": 2.0455239012711806,
+      "learning_rate": 5.378584297343053e-06,
+      "loss": 0.4573,
+      "step": 1510
+    },
+    {
+      "epoch": 1.0644257703081232,
+      "grad_norm": 2.799894111126595,
+      "learning_rate": 5.3176125052143905e-06,
+      "loss": 0.4475,
+      "step": 1520
+    },
+    {
+      "epoch": 1.0714285714285714,
+      "grad_norm": 2.3537914083107743,
+      "learning_rate": 5.256593253309728e-06,
+      "loss": 0.4679,
+      "step": 1530
+    },
+    {
+      "epoch": 1.0784313725490196,
+      "grad_norm": 2.3121216793787283,
+      "learning_rate": 5.195535659532111e-06,
+      "loss": 0.4982,
+      "step": 1540
+    },
+    {
+      "epoch": 1.0854341736694677,
+      "grad_norm": 1.8801771013194575,
+      "learning_rate": 5.134448847513873e-06,
+      "loss": 0.4881,
+      "step": 1550
+    },
+    {
+      "epoch": 1.092436974789916,
+      "grad_norm": 2.117146997784492,
+      "learning_rate": 5.073341945253336e-06,
+      "loss": 0.4518,
+      "step": 1560
+    },
+    {
+      "epoch": 1.0994397759103642,
+      "grad_norm": 2.4512284660613592,
+      "learning_rate": 5.012224083750845e-06,
+      "loss": 0.4618,
+      "step": 1570
+    },
+    {
+      "epoch": 1.1064425770308124,
+      "grad_norm": 2.3001015988026747,
+      "learning_rate": 4.951104395644342e-06,
+      "loss": 0.4647,
+      "step": 1580
+    },
+    {
+      "epoch": 1.1134453781512605,
+      "grad_norm": 2.733492067730666,
+      "learning_rate": 4.889992013844711e-06,
+      "loss": 0.497,
+      "step": 1590
+    },
+    {
+      "epoch": 1.1204481792717087,
+      "grad_norm": 2.624090029655425,
+      "learning_rate": 4.828896070171084e-06,
+      "loss": 0.5142,
+      "step": 1600
+    },
+    {
+      "epoch": 1.1274509803921569,
+      "grad_norm": 2.3184684372082405,
+      "learning_rate": 4.767825693986295e-06,
+      "loss": 0.449,
+      "step": 1610
+    },
+    {
+      "epoch": 1.134453781512605,
+      "grad_norm": 2.5173043040744836,
+      "learning_rate": 4.706790010832714e-06,
+      "loss": 0.5264,
+      "step": 1620
+    },
+    {
+      "epoch": 1.1414565826330532,
+      "grad_norm": 2.088682951418163,
+      "learning_rate": 4.6457981410686524e-06,
+      "loss": 0.4551,
+      "step": 1630
+    },
+    {
+      "epoch": 1.1484593837535013,
+      "grad_norm": 1.9999194791787656,
+      "learning_rate": 4.584859198505526e-06,
+      "loss": 0.5141,
+      "step": 1640
+    },
+    {
+      "epoch": 1.1554621848739495,
+      "grad_norm": 2.0283863135066293,
+      "learning_rate": 4.5239822890460174e-06,
+      "loss": 0.4254,
+      "step": 1650
+    },
+    {
+      "epoch": 1.1624649859943978,
+      "grad_norm": 2.4761929786113246,
+      "learning_rate": 4.463176509323413e-06,
+      "loss": 0.4986,
+      "step": 1660
+    },
+    {
+      "epoch": 1.169467787114846,
+      "grad_norm": 1.6823258146384572,
+      "learning_rate": 4.402450945342317e-06,
+      "loss": 0.3929,
+      "step": 1670
+    },
+    {
+      "epoch": 1.1764705882352942,
+      "grad_norm": 2.3712484150543065,
+      "learning_rate": 4.341814671120965e-06,
+      "loss": 0.5022,
+      "step": 1680
+    },
+    {
+      "epoch": 1.1834733893557423,
+      "grad_norm": 2.1223027287039136,
+      "learning_rate": 4.2812767473353205e-06,
+      "loss": 0.5201,
+      "step": 1690
+    },
+    {
+      "epoch": 1.1904761904761905,
+      "grad_norm": 2.280910320108766,
+      "learning_rate": 4.220846219965168e-06,
+      "loss": 0.4627,
+      "step": 1700
+    },
+    {
+      "epoch": 1.1974789915966386,
+      "grad_norm": 1.850154690325662,
+      "learning_rate": 4.160532118942411e-06,
+      "loss": 0.4312,
+      "step": 1710
+    },
+    {
+      "epoch": 1.2044817927170868,
+      "grad_norm": 2.339402359759638,
+      "learning_rate": 4.100343456801747e-06,
+      "loss": 0.4317,
+      "step": 1720
+    },
+    {
+      "epoch": 1.211484593837535,
+      "grad_norm": 2.2500882330234604,
+      "learning_rate": 4.040289227333961e-06,
+      "loss": 0.4521,
+      "step": 1730
+    },
+    {
+      "epoch": 1.2184873949579833,
+      "grad_norm": 2.8132677454268857,
+      "learning_rate": 3.980378404242014e-06,
+      "loss": 0.4432,
+      "step": 1740
+    },
+    {
+      "epoch": 1.2254901960784315,
+      "grad_norm": 2.67999303980593,
+      "learning_rate": 3.920619939800131e-06,
+      "loss": 0.4613,
+      "step": 1750
+    },
+    {
+      "epoch": 1.2324929971988796,
+      "grad_norm": 1.8885239257708604,
+      "learning_rate": 3.861022763516091e-06,
+      "loss": 0.4774,
+      "step": 1760
+    },
+    {
+      "epoch": 1.2394957983193278,
+      "grad_norm": 2.815116777237025,
+      "learning_rate": 3.8015957807969247e-06,
+      "loss": 0.4879,
+      "step": 1770
+    },
+    {
+      "epoch": 1.246498599439776,
+      "grad_norm": 2.174587271539217,
+      "learning_rate": 3.7423478716182026e-06,
+      "loss": 0.456,
+      "step": 1780
+    },
+    {
+      "epoch": 1.253501400560224,
+      "grad_norm": 2.7567028224464183,
+      "learning_rate": 3.68328788919714e-06,
+      "loss": 0.4644,
+      "step": 1790
+    },
+    {
+      "epoch": 1.2605042016806722,
+      "grad_norm": 2.1568363923043266,
+      "learning_rate": 3.624424658669682e-06,
+      "loss": 0.5225,
+      "step": 1800
+    },
+    {
+      "epoch": 1.2675070028011204,
+      "grad_norm": 2.721189011538071,
+      "learning_rate": 3.5657669757717927e-06,
+      "loss": 0.3921,
+      "step": 1810
+    },
+    {
+      "epoch": 1.2745098039215685,
+      "grad_norm": 2.120385676401417,
+      "learning_rate": 3.5073236055251425e-06,
+      "loss": 0.4793,
+      "step": 1820
+    },
+    {
+      "epoch": 1.2815126050420167,
+      "grad_norm": 2.302768261718875,
+      "learning_rate": 3.4491032809273784e-06,
+      "loss": 0.4526,
+      "step": 1830
+    },
+    {
+      "epoch": 1.2885154061624648,
+      "grad_norm": 2.334008008597067,
+      "learning_rate": 3.3911147016471784e-06,
+      "loss": 0.4855,
+      "step": 1840
+    },
+    {
+      "epoch": 1.2955182072829132,
+      "grad_norm": 2.6089068847882744,
+      "learning_rate": 3.3333665327242965e-06,
+      "loss": 0.4522,
+      "step": 1850
+    },
+    {
+      "epoch": 1.3025210084033614,
+      "grad_norm": 2.8426511093929556,
+      "learning_rate": 3.275867403274763e-06,
+      "loss": 0.4965,
+      "step": 1860
+    },
+    {
+      "epoch": 1.3095238095238095,
+      "grad_norm": 2.515991574117609,
+      "learning_rate": 3.2186259052014797e-06,
+      "loss": 0.4635,
+      "step": 1870
+    },
+    {
+      "epoch": 1.3165266106442577,
+      "grad_norm": 2.410804360632545,
+      "learning_rate": 3.1616505919103446e-06,
+      "loss": 0.44,
+      "step": 1880
+    },
+    {
+      "epoch": 1.3235294117647058,
+      "grad_norm": 2.1144368698980234,
+      "learning_rate": 3.1049499770321572e-06,
+      "loss": 0.511,
+      "step": 1890
+    },
+    {
+      "epoch": 1.330532212885154,
+      "grad_norm": 2.3764720452535113,
+      "learning_rate": 3.0485325331504477e-06,
+      "loss": 0.4298,
+      "step": 1900
+    },
+    {
+      "epoch": 1.3375350140056024,
+      "grad_norm": 2.008815745845685,
+      "learning_rate": 2.992406690535453e-06,
+      "loss": 0.4353,
+      "step": 1910
+    },
+    {
+      "epoch": 1.3445378151260505,
+      "grad_norm": 2.5495322694247404,
+      "learning_rate": 2.936580835884404e-06,
+      "loss": 0.5351,
+      "step": 1920
+    },
+    {
+      "epoch": 1.3515406162464987,
+      "grad_norm": 2.6170532242550424,
+      "learning_rate": 2.8810633110683312e-06,
+      "loss": 0.4736,
+      "step": 1930
+    },
+    {
+      "epoch": 1.3585434173669468,
+      "grad_norm": 2.304939061322721,
+      "learning_rate": 2.8258624118855704e-06,
+      "loss": 0.497,
+      "step": 1940
+    },
+    {
+      "epoch": 1.365546218487395,
+      "grad_norm": 2.52028463522758,
+      "learning_rate": 2.770986386822145e-06,
+      "loss": 0.4561,
+      "step": 1950
+    },
+    {
+      "epoch": 1.3725490196078431,
+      "grad_norm": 2.1554513929621653,
+      "learning_rate": 2.7164434358192236e-06,
+      "loss": 0.4771,
+      "step": 1960
+    },
+    {
+      "epoch": 1.3795518207282913,
+      "grad_norm": 2.2864103571812247,
+      "learning_rate": 2.6622417090478304e-06,
+      "loss": 0.4244,
+      "step": 1970
+    },
+    {
+      "epoch": 1.3865546218487395,
+      "grad_norm": 2.3291887139186604,
+      "learning_rate": 2.6083893056909937e-06,
+      "loss": 0.4818,
+      "step": 1980
+    },
+    {
+      "epoch": 1.3935574229691876,
+      "grad_norm": 1.6186217281536261,
+      "learning_rate": 2.5548942727335046e-06,
+      "loss": 0.4473,
+      "step": 1990
+    },
+    {
+      "epoch": 1.4005602240896358,
+      "grad_norm": 1.9490552509158812,
+      "learning_rate": 2.5017646037594967e-06,
+      "loss": 0.4653,
+      "step": 2000
+    },
+    {
+      "epoch": 1.407563025210084,
+      "grad_norm": 2.4440857430217124,
+      "learning_rate": 2.4490082377579805e-06,
+      "loss": 0.4945,
+      "step": 2010
+    },
+    {
+      "epoch": 1.4145658263305323,
+      "grad_norm": 2.9756203221062476,
+      "learning_rate": 2.3966330579365436e-06,
+      "loss": 0.4371,
+      "step": 2020
+    },
+    {
+      "epoch": 1.4215686274509804,
+      "grad_norm": 2.202500602837109,
+      "learning_rate": 2.3446468905434093e-06,
+      "loss": 0.4495,
+      "step": 2030
+    },
+    {
+      "epoch": 1.4285714285714286,
+      "grad_norm": 1.9506623499545588,
+      "learning_rate": 2.2930575036979634e-06,
+      "loss": 0.4552,
+      "step": 2040
+    },
+    {
+      "epoch": 1.4355742296918768,
+      "grad_norm": 2.6310040966711776,
+      "learning_rate": 2.241872606229998e-06,
+      "loss": 0.4378,
+      "step": 2050
+    },
+    {
+      "epoch": 1.442577030812325,
+      "grad_norm": 2.4781025094081706,
+      "learning_rate": 2.191099846527813e-06,
+      "loss": 0.5055,
+      "step": 2060
+    },
+    {
+      "epoch": 1.449579831932773,
+      "grad_norm": 2.3262915560231217,
+      "learning_rate": 2.14074681139534e-06,
+      "loss": 0.4727,
+      "step": 2070
+    },
+    {
+      "epoch": 1.4565826330532212,
+      "grad_norm": 2.001470623484971,
+      "learning_rate": 2.090821024918462e-06,
+      "loss": 0.4826,
+      "step": 2080
+    },
+    {
+      "epoch": 1.4635854341736696,
+      "grad_norm": 2.8728529295622964,
+      "learning_rate": 2.0413299473407285e-06,
+      "loss": 0.4162,
+      "step": 2090
+    },
+    {
+      "epoch": 1.4705882352941178,
+      "grad_norm": 2.54674433384974,
+      "learning_rate": 1.9922809739485883e-06,
+      "loss": 0.4516,
+      "step": 2100
+    },
+    {
+      "epoch": 1.477591036414566,
+      "grad_norm": 1.7010786229876187,
+      "learning_rate": 1.943681433966338e-06,
+      "loss": 0.4727,
+      "step": 2110
+    },
+    {
+      "epoch": 1.484593837535014,
+      "grad_norm": 2.7445575800859747,
+      "learning_rate": 1.8955385894609414e-06,
+      "loss": 0.4835,
+      "step": 2120
+    },
+    {
+      "epoch": 1.4915966386554622,
+      "grad_norm": 2.5636924295564825,
+      "learning_rate": 1.8478596342568827e-06,
+      "loss": 0.5,
+      "step": 2130
+    },
+    {
+      "epoch": 1.4985994397759104,
+      "grad_norm": 2.6320112816326806,
+      "learning_rate": 1.8006516928612121e-06,
+      "loss": 0.4783,
+      "step": 2140
+    },
+    {
+      "epoch": 1.5056022408963585,
+      "grad_norm": 2.3130885237765653,
+      "learning_rate": 1.753921819398966e-06,
+      "loss": 0.4732,
+      "step": 2150
+    },
+    {
+      "epoch": 1.5126050420168067,
+      "grad_norm": 2.4566766430867855,
+      "learning_rate": 1.7076769965590834e-06,
+      "loss": 0.4846,
+      "step": 2160
+    },
+    {
+      "epoch": 1.5196078431372548,
+      "grad_norm": 2.1691439788909705,
+      "learning_rate": 1.6619241345510057e-06,
+      "loss": 0.5196,
+      "step": 2170
+    },
+    {
+      "epoch": 1.526610644257703,
+      "grad_norm": 1.9388105151944164,
+      "learning_rate": 1.6166700700721121e-06,
+      "loss": 0.3903,
+      "step": 2180
+    },
+    {
+      "epoch": 1.5336134453781511,
+      "grad_norm": 2.353752022853995,
+      "learning_rate": 1.571921565286139e-06,
+      "loss": 0.4923,
+      "step": 2190
+    },
+    {
+      "epoch": 1.5406162464985993,
+      "grad_norm": 2.572914747835839,
+      "learning_rate": 1.527685306812715e-06,
+      "loss": 0.4166,
+      "step": 2200
+    },
+    {
+      "epoch": 1.5476190476190477,
+      "grad_norm": 2.1724801712005792,
+      "learning_rate": 1.4839679047282206e-06,
+      "loss": 0.4504,
+      "step": 2210
+    },
+    {
+      "epoch": 1.5546218487394958,
+      "grad_norm": 2.594065623296099,
+      "learning_rate": 1.4407758915780578e-06,
+      "loss": 0.477,
+      "step": 2220
+    },
+    {
+      "epoch": 1.561624649859944,
+      "grad_norm": 2.208098426234172,
+      "learning_rate": 1.3981157214005098e-06,
+      "loss": 0.4586,
+      "step": 2230
+    },
+    {
+      "epoch": 1.5686274509803921,
+      "grad_norm": 2.139581629149966,
+      "learning_rate": 1.3559937687623458e-06,
+      "loss": 0.4276,
+      "step": 2240
+    },
+    {
+      "epoch": 1.5756302521008403,
+      "grad_norm": 2.538499089739133,
+      "learning_rate": 1.3144163278062848e-06,
+      "loss": 0.4206,
+      "step": 2250
+    },
+    {
+      "epoch": 1.5826330532212887,
+      "grad_norm": 2.041136800394131,
+      "learning_rate": 1.273389611310481e-06,
+      "loss": 0.4034,
+      "step": 2260
+    },
+    {
+      "epoch": 1.5896358543417368,
+      "grad_norm": 2.313241143682124,
+      "learning_rate": 1.2329197497601742e-06,
+      "loss": 0.4801,
+      "step": 2270
+    },
+    {
+      "epoch": 1.596638655462185,
+      "grad_norm": 2.309500171812923,
+      "learning_rate": 1.1930127904316286e-06,
+      "loss": 0.5045,
+      "step": 2280
+    },
+    {
+      "epoch": 1.6036414565826331,
+      "grad_norm": 2.485298750781323,
+      "learning_rate": 1.1536746964885065e-06,
+      "loss": 0.4301,
+      "step": 2290
+    },
+    {
+      "epoch": 1.6106442577030813,
+      "grad_norm": 2.591481445677265,
+      "learning_rate": 1.114911346090819e-06,
+      "loss": 0.4554,
+      "step": 2300
+    },
+    {
+      "epoch": 1.6176470588235294,
+      "grad_norm": 2.833792845787996,
+      "learning_rate": 1.0767285315165675e-06,
+      "loss": 0.4132,
+      "step": 2310
+    },
+    {
+      "epoch": 1.6246498599439776,
+      "grad_norm": 2.388415928161688,
+      "learning_rate": 1.0391319582962244e-06,
+      "loss": 0.4707,
+      "step": 2320
+    },
+    {
+      "epoch": 1.6316526610644257,
+      "grad_norm": 2.3441821249748798,
+      "learning_rate": 1.002127244360171e-06,
+      "loss": 0.422,
+      "step": 2330
+    },
+    {
+      "epoch": 1.638655462184874,
+      "grad_norm": 1.9731306455288935,
+      "learning_rate": 9.65719919199235e-07,
+      "loss": 0.3684,
+      "step": 2340
+    },
+    {
+      "epoch": 1.645658263305322,
+      "grad_norm": 3.7976656804390303,
+      "learning_rate": 9.299154230384383e-07,
+      "loss": 0.4887,
+      "step": 2350
+    },
+    {
+      "epoch": 1.6526610644257702,
+      "grad_norm": 2.29269452111851,
+      "learning_rate": 8.947191060240701e-07,
+      "loss": 0.4054,
+      "step": 2360
+    },
+    {
+      "epoch": 1.6596638655462184,
+      "grad_norm": 2.765208721785312,
+      "learning_rate": 8.601362274242465e-07,
+      "loss": 0.5191,
+      "step": 2370
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 2.637952088090675,
+      "learning_rate": 8.261719548430292e-07,
+      "loss": 0.479,
+      "step": 2380
+    },
+    {
+      "epoch": 1.673669467787115,
+      "grad_norm": 2.7259524316936496,
+      "learning_rate": 7.928313634482454e-07,
+      "loss": 0.4162,
+      "step": 2390
+    },
+    {
+      "epoch": 1.680672268907563,
+      "grad_norm": 2.783618319304046,
+      "learning_rate": 7.601194352131285e-07,
+      "loss": 0.4662,
+      "step": 2400
+    },
+    {
+      "epoch": 1.6876750700280112,
+      "grad_norm": 2.5896397818912957,
+      "learning_rate": 7.280410581718761e-07,
+      "loss": 0.466,
+      "step": 2410
+    },
+    {
+      "epoch": 1.6946778711484594,
+      "grad_norm": 2.313925409486687,
+      "learning_rate": 6.966010256892408e-07,
+      "loss": 0.421,
+      "step": 2420
+    },
+    {
+      "epoch": 1.7016806722689075,
+      "grad_norm": 2.2637352342234673,
+      "learning_rate": 6.658040357442874e-07,
+      "loss": 0.4384,
+      "step": 2430
+    },
+    {
+      "epoch": 1.708683473389356,
+      "grad_norm": 2.2403407432076836,
+      "learning_rate": 6.356546902283772e-07,
+      "loss": 0.4423,
+      "step": 2440
+    },
+    {
+      "epoch": 1.715686274509804,
+      "grad_norm": 2.7755062361806333,
+      "learning_rate": 6.061574942575266e-07,
+      "loss": 0.4507,
+      "step": 2450
+    },
+    {
+      "epoch": 1.7226890756302522,
+      "grad_norm": 2.3259900192767713,
+      "learning_rate": 5.773168554992248e-07,
+      "loss": 0.426,
+      "step": 2460
+    },
+    {
+      "epoch": 1.7296918767507004,
+      "grad_norm": 2.6341383732328456,
+      "learning_rate": 5.491370835138116e-07,
+      "loss": 0.4377,
+      "step": 2470
+    },
+    {
+      "epoch": 1.7366946778711485,
+      "grad_norm": 2.3924610495134933,
+      "learning_rate": 5.216223891105054e-07,
+      "loss": 0.4585,
+      "step": 2480
+    },
+    {
+      "epoch": 1.7436974789915967,
+      "grad_norm": 2.1758349972571507,
+      "learning_rate": 4.947768837182093e-07,
+      "loss": 0.4068,
+      "step": 2490
+    },
+    {
+      "epoch": 1.7507002801120448,
+      "grad_norm": 2.0421217828565448,
+      "learning_rate": 4.686045787711435e-07,
+      "loss": 0.4066,
+      "step": 2500
+    },
+    {
+      "epoch": 1.757703081232493,
+      "grad_norm": 2.115206308615413,
+      "learning_rate": 4.4310938510943533e-07,
+      "loss": 0.4522,
+      "step": 2510
+    },
+    {
+      "epoch": 1.7647058823529411,
+      "grad_norm": 2.066405782929325,
+      "learning_rate": 4.182951123947332e-07,
+      "loss": 0.3312,
+      "step": 2520
+    },
+    {
+      "epoch": 1.7717086834733893,
+      "grad_norm": 2.342484690456666,
+      "learning_rate": 3.9416546854094294e-07,
+      "loss": 0.4224,
+      "step": 2530
+    },
+    {
+      "epoch": 1.7787114845938374,
+      "grad_norm": 2.512724647352599,
+      "learning_rate": 3.7072405916016353e-07,
+      "loss": 0.445,
+      "step": 2540
+    },
+    {
+      "epoch": 1.7857142857142856,
+      "grad_norm": 2.470340765232829,
+      "learning_rate": 3.479743870239188e-07,
+      "loss": 0.4637,
+      "step": 2550
+    },
+    {
+      "epoch": 1.7927170868347337,
+      "grad_norm": 2.130736248713378,
+      "learning_rate": 3.2591985153974383e-07,
+      "loss": 0.387,
+      "step": 2560
+    },
+    {
+      "epoch": 1.7997198879551821,
+      "grad_norm": 2.1431789471135723,
+      "learning_rate": 3.0456374824322674e-07,
+      "loss": 0.4469,
+      "step": 2570
+    },
+    {
+      "epoch": 1.8067226890756303,
+      "grad_norm": 2.099234007419937,
+      "learning_rate": 2.839092683055622e-07,
+      "loss": 0.4039,
+      "step": 2580
+    },
+    {
+      "epoch": 1.8137254901960784,
+      "grad_norm": 2.459188631196229,
+      "learning_rate": 2.639594980567162e-07,
+      "loss": 0.3812,
+      "step": 2590
+    },
+    {
+      "epoch": 1.8207282913165266,
+      "grad_norm": 2.555886554006267,
+      "learning_rate": 2.447174185242324e-07,
+      "loss": 0.4212,
+      "step": 2600
+    },
+    {
+      "epoch": 1.8277310924369747,
+      "grad_norm": 2.2836207612772843,
+      "learning_rate": 2.2618590498779847e-07,
+      "loss": 0.4137,
+      "step": 2610
+    },
+    {
+      "epoch": 1.8347338935574231,
+      "grad_norm": 2.1026781532549936,
+      "learning_rate": 2.0836772654959647e-07,
+      "loss": 0.4213,
+      "step": 2620
+    },
+    {
+      "epoch": 1.8417366946778713,
+      "grad_norm": 2.4328664873281483,
+      "learning_rate": 1.9126554572052813e-07,
+      "loss": 0.4335,
+      "step": 2630
+    },
+    {
+      "epoch": 1.8487394957983194,
+      "grad_norm": 2.3974367367984857,
+      "learning_rate": 1.74881918022361e-07,
+      "loss": 0.4056,
+      "step": 2640
+    },
+    {
+      "epoch": 1.8557422969187676,
+      "grad_norm": 2.299811399188436,
+      "learning_rate": 1.5921929160587045e-07,
+      "loss": 0.4602,
+      "step": 2650
+    },
+    {
+      "epoch": 1.8627450980392157,
+      "grad_norm": 1.9124769037343121,
+      "learning_rate": 1.44280006885017e-07,
+      "loss": 0.4231,
+      "step": 2660
+    },
+    {
+      "epoch": 1.8697478991596639,
+      "grad_norm": 2.813578456729493,
+      "learning_rate": 1.3006629618722733e-07,
+      "loss": 0.4286,
+      "step": 2670
+    },
+    {
+      "epoch": 1.876750700280112,
+      "grad_norm": 2.4806383435943613,
+      "learning_rate": 1.1658028341982486e-07,
+      "loss": 0.4368,
+      "step": 2680
+    },
+    {
+      "epoch": 1.8837535014005602,
+      "grad_norm": 2.2681286986704534,
+      "learning_rate": 1.0382398375266111e-07,
+      "loss": 0.4691,
+      "step": 2690
+    },
+    {
+      "epoch": 1.8907563025210083,
+      "grad_norm": 1.9128855914355993,
+      "learning_rate": 9.179930331699615e-08,
+      "loss": 0.4508,
+      "step": 2700
+    },
+    {
+      "epoch": 1.8977591036414565,
+      "grad_norm": 2.1633704801409834,
+      "learning_rate": 8.050803892067139e-08,
+      "loss": 0.4422,
+      "step": 2710
+    },
+    {
+      "epoch": 1.9047619047619047,
+      "grad_norm": 2.8200396806385,
+      "learning_rate": 6.995187777961931e-08,
+      "loss": 0.4254,
+      "step": 2720
+    },
+    {
+      "epoch": 1.9117647058823528,
+      "grad_norm": 2.6445714035080887,
+      "learning_rate": 6.013239726574694e-08,
+      "loss": 0.4074,
+      "step": 2730
+    },
+    {
+      "epoch": 1.918767507002801,
+      "grad_norm": 2.0546751872771627,
+      "learning_rate": 5.105106467123477e-08,
+      "loss": 0.4262,
+      "step": 2740
+    },
+    {
+      "epoch": 1.9257703081232493,
+      "grad_norm": 2.7784776972549916,
+      "learning_rate": 4.2709236989287305e-08,
+      "loss": 0.4725,
+      "step": 2750
+    },
+    {
+      "epoch": 1.9327731092436975,
+      "grad_norm": 2.272829343457401,
+      "learning_rate": 3.510816071135681e-08,
+      "loss": 0.4409,
+      "step": 2760
+    },
+    {
+      "epoch": 1.9397759103641457,
+      "grad_norm": 2.4175361768734276,
+      "learning_rate": 2.8248971640887913e-08,
+      "loss": 0.393,
+      "step": 2770
+    },
+    {
+      "epoch": 1.9467787114845938,
+      "grad_norm": 2.2153647657315347,
+      "learning_rate": 2.213269472359836e-08,
+      "loss": 0.474,
+      "step": 2780
+    },
+    {
+      "epoch": 1.9537815126050422,
+      "grad_norm": 2.146439918476662,
+      "learning_rate": 1.6760243894321513e-08,
+      "loss": 0.4008,
+      "step": 2790
+    },
+    {
+      "epoch": 1.9607843137254903,
+      "grad_norm": 2.0928270476543522,
+      "learning_rate": 1.213242194044395e-08,
+      "loss": 0.4111,
+      "step": 2800
+    },
+    {
+      "epoch": 1.9677871148459385,
+      "grad_norm": 2.079556551258959,
+      "learning_rate": 8.249920381946387e-09,
+      "loss": 0.4999,
+      "step": 2810
+    },
+    {
+      "epoch": 1.9747899159663866,
+      "grad_norm": 2.0113264107881164,
+      "learning_rate": 5.113319368070801e-09,
+      "loss": 0.4397,
+      "step": 2820
+    },
+    {
+      "epoch": 1.9817927170868348,
+      "grad_norm": 2.3335982318192827,
+      "learning_rate": 2.723087590632556e-09,
+      "loss": 0.3984,
+      "step": 2830
+    },
+    {
+      "epoch": 1.988795518207283,
+      "grad_norm": 2.7347014131885015,
+      "learning_rate": 1.079582213983632e-09,
+      "loss": 0.4515,
+      "step": 2840
+    },
+    {
+      "epoch": 1.995798319327731,
+      "grad_norm": 1.8502359446882066,
+      "learning_rate": 1.830488216442161e-10,
+      "loss": 0.4036,
+      "step": 2850
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 2856,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "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": 137777518706688.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-2856/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bdc5f223219a2d86fa771859dc8e50b6963552678fb8dabcd99841c7bfe0e249
+size 8081

config.json

@@ -0,0 +1,30 @@
+{
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "eos_token_id": 151645,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 12288,
+  "max_position_embeddings": 40960,
+  "max_window_layers": 36,
+  "model_type": "qwen3",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 36,
+  "num_key_value_heads": 8,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000,
+  "sliding_window": null,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.52.4",
+  "use_cache": false,
+  "use_sliding_window": false,
+  "vocab_size": 151936
+}

generation_config.json

@@ -0,0 +1,13 @@
+{
+  "bos_token_id": 151643,
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "pad_token_id": 151643,
+  "temperature": 0.6,
+  "top_k": 20,
+  "top_p": 0.95,
+  "transformers_version": "4.52.4"
+}

model-00001-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2b0be61836f0b007bb5c5a491006a2ce49f3e66f9cfb97966ecdcd04a56bdd9f
+size 4902257696

model-00002-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:93f965306bec10edef5a70fc848a366439e394f5a19060decf2ba944eabd210d
+size 4915960368

model-00003-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8798485503eeb6b367126953d2070f4c2fcb539e7039345f0885810d650d76f3
+size 4983068496

model-00004-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4d18f328ca8831716c919c95343c70004df1d8d9c5c1ef33c4005b8f4d47d571
+size 1580230264

model.safetensors.index.json

@@ -0,0 +1,406 @@
+{
+  "metadata": {
+    "total_size": 16381470720
+  },
+  "weight_map": {
+    "lm_head.weight": "model-00004-of-00004.safetensors",
+    "model.embed_tokens.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.22.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.34.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.34.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.34.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.34.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.34.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.34.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.34.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.34.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.34.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.34.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.34.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.35.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.k_norm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.35.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.q_norm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.35.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.9.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.norm.weight": "model-00004-of-00004.safetensors"
+  }
+}

special_tokens_map.json

@@ -0,0 +1,31 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer.json

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

tokenizer_config.json

@@ -0,0 +1,240 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151665": {
+      "content": "<tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151666": {
+      "content": "</tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151667": {
+      "content": "<think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151668": {
+      "content": "</think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "padding_side": "right",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

train_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 2.0,
+    "total_flos": 137777518706688.0,
+    "train_loss": 0.5963595523720696,
+    "train_runtime": 4888.1806,
+    "train_samples_per_second": 9.347,
+    "train_steps_per_second": 0.584
+}

trainer_log.jsonl

@@ -0,0 +1,286 @@
+{"current_steps": 10, "total_steps": 2856, "loss": 2.9142, "lr": 3.1468531468531473e-07, "epoch": 0.0070028011204481795, "percentage": 0.35, "elapsed_time": "0:00:20", "remaining_time": "1:38:49"}
+{"current_steps": 20, "total_steps": 2856, "loss": 2.6364, "lr": 6.643356643356644e-07, "epoch": 0.014005602240896359, "percentage": 0.7, "elapsed_time": "0:00:39", "remaining_time": "1:32:29"}
+{"current_steps": 30, "total_steps": 2856, "loss": 2.0887, "lr": 1.013986013986014e-06, "epoch": 0.02100840336134454, "percentage": 1.05, "elapsed_time": "0:00:57", "remaining_time": "1:30:23"}
+{"current_steps": 40, "total_steps": 2856, "loss": 1.593, "lr": 1.3636363636363636e-06, "epoch": 0.028011204481792718, "percentage": 1.4, "elapsed_time": "0:01:15", "remaining_time": "1:28:57"}
+{"current_steps": 50, "total_steps": 2856, "loss": 1.4625, "lr": 1.7132867132867134e-06, "epoch": 0.0350140056022409, "percentage": 1.75, "elapsed_time": "0:01:34", "remaining_time": "1:28:00"}
+{"current_steps": 60, "total_steps": 2856, "loss": 1.231, "lr": 2.0629370629370634e-06, "epoch": 0.04201680672268908, "percentage": 2.1, "elapsed_time": "0:01:52", "remaining_time": "1:27:18"}
+{"current_steps": 70, "total_steps": 2856, "loss": 1.1051, "lr": 2.4125874125874128e-06, "epoch": 0.049019607843137254, "percentage": 2.45, "elapsed_time": "0:02:10", "remaining_time": "1:26:20"}
+{"current_steps": 80, "total_steps": 2856, "loss": 0.9814, "lr": 2.762237762237762e-06, "epoch": 0.056022408963585436, "percentage": 2.8, "elapsed_time": "0:02:28", "remaining_time": "1:25:48"}
+{"current_steps": 90, "total_steps": 2856, "loss": 0.9429, "lr": 3.1118881118881123e-06, "epoch": 0.06302521008403361, "percentage": 3.15, "elapsed_time": "0:02:46", "remaining_time": "1:25:02"}
+{"current_steps": 100, "total_steps": 2856, "loss": 0.8549, "lr": 3.4615384615384617e-06, "epoch": 0.0700280112044818, "percentage": 3.5, "elapsed_time": "0:03:04", "remaining_time": "1:24:38"}
+{"current_steps": 110, "total_steps": 2856, "loss": 0.9123, "lr": 3.811188811188811e-06, "epoch": 0.07703081232492998, "percentage": 3.85, "elapsed_time": "0:03:22", "remaining_time": "1:24:03"}
+{"current_steps": 120, "total_steps": 2856, "loss": 0.7164, "lr": 4.160839160839161e-06, "epoch": 0.08403361344537816, "percentage": 4.2, "elapsed_time": "0:03:39", "remaining_time": "1:23:27"}
+{"current_steps": 130, "total_steps": 2856, "loss": 0.8358, "lr": 4.510489510489511e-06, "epoch": 0.09103641456582633, "percentage": 4.55, "elapsed_time": "0:03:57", "remaining_time": "1:22:54"}
+{"current_steps": 140, "total_steps": 2856, "loss": 0.7863, "lr": 4.86013986013986e-06, "epoch": 0.09803921568627451, "percentage": 4.9, "elapsed_time": "0:04:14", "remaining_time": "1:22:20"}
+{"current_steps": 150, "total_steps": 2856, "loss": 0.7784, "lr": 5.20979020979021e-06, "epoch": 0.10504201680672269, "percentage": 5.25, "elapsed_time": "0:04:32", "remaining_time": "1:21:51"}
+{"current_steps": 160, "total_steps": 2856, "loss": 0.696, "lr": 5.5594405594405596e-06, "epoch": 0.11204481792717087, "percentage": 5.6, "elapsed_time": "0:04:49", "remaining_time": "1:21:19"}
+{"current_steps": 170, "total_steps": 2856, "loss": 0.7731, "lr": 5.90909090909091e-06, "epoch": 0.11904761904761904, "percentage": 5.95, "elapsed_time": "0:05:07", "remaining_time": "1:20:56"}
+{"current_steps": 180, "total_steps": 2856, "loss": 0.7214, "lr": 6.258741258741259e-06, "epoch": 0.12605042016806722, "percentage": 6.3, "elapsed_time": "0:05:25", "remaining_time": "1:20:32"}
+{"current_steps": 190, "total_steps": 2856, "loss": 0.7684, "lr": 6.608391608391609e-06, "epoch": 0.1330532212885154, "percentage": 6.65, "elapsed_time": "0:05:42", "remaining_time": "1:20:08"}
+{"current_steps": 200, "total_steps": 2856, "loss": 0.7845, "lr": 6.958041958041959e-06, "epoch": 0.1400560224089636, "percentage": 7.0, "elapsed_time": "0:05:59", "remaining_time": "1:19:39"}
+{"current_steps": 210, "total_steps": 2856, "loss": 0.6937, "lr": 7.307692307692308e-06, "epoch": 0.14705882352941177, "percentage": 7.35, "elapsed_time": "0:06:17", "remaining_time": "1:19:13"}
+{"current_steps": 220, "total_steps": 2856, "loss": 0.8302, "lr": 7.657342657342658e-06, "epoch": 0.15406162464985995, "percentage": 7.7, "elapsed_time": "0:06:34", "remaining_time": "1:18:46"}
+{"current_steps": 230, "total_steps": 2856, "loss": 0.7539, "lr": 8.006993006993008e-06, "epoch": 0.16106442577030813, "percentage": 8.05, "elapsed_time": "0:06:51", "remaining_time": "1:18:20"}
+{"current_steps": 240, "total_steps": 2856, "loss": 0.7641, "lr": 8.356643356643356e-06, "epoch": 0.16806722689075632, "percentage": 8.4, "elapsed_time": "0:07:09", "remaining_time": "1:17:56"}
+{"current_steps": 250, "total_steps": 2856, "loss": 0.7131, "lr": 8.706293706293708e-06, "epoch": 0.17507002801120447, "percentage": 8.75, "elapsed_time": "0:07:26", "remaining_time": "1:17:34"}
+{"current_steps": 260, "total_steps": 2856, "loss": 0.595, "lr": 9.055944055944057e-06, "epoch": 0.18207282913165265, "percentage": 9.1, "elapsed_time": "0:07:43", "remaining_time": "1:17:11"}
+{"current_steps": 270, "total_steps": 2856, "loss": 0.7676, "lr": 9.405594405594406e-06, "epoch": 0.18907563025210083, "percentage": 9.45, "elapsed_time": "0:08:00", "remaining_time": "1:16:45"}
+{"current_steps": 280, "total_steps": 2856, "loss": 0.7019, "lr": 9.755244755244755e-06, "epoch": 0.19607843137254902, "percentage": 9.8, "elapsed_time": "0:08:17", "remaining_time": "1:16:20"}
+{"current_steps": 290, "total_steps": 2856, "loss": 0.6575, "lr": 9.999966378620396e-06, "epoch": 0.2030812324929972, "percentage": 10.15, "elapsed_time": "0:08:34", "remaining_time": "1:15:55"}
+{"current_steps": 300, "total_steps": 2856, "loss": 0.6984, "lr": 9.999368677783691e-06, "epoch": 0.21008403361344538, "percentage": 10.5, "elapsed_time": "0:08:51", "remaining_time": "1:15:31"}
+{"current_steps": 310, "total_steps": 2856, "loss": 0.8306, "lr": 9.998023937980683e-06, "epoch": 0.21708683473389356, "percentage": 10.85, "elapsed_time": "0:09:09", "remaining_time": "1:15:10"}
+{"current_steps": 320, "total_steps": 2856, "loss": 0.784, "lr": 9.995932360151348e-06, "epoch": 0.22408963585434175, "percentage": 11.2, "elapsed_time": "0:09:26", "remaining_time": "1:14:50"}
+{"current_steps": 330, "total_steps": 2856, "loss": 0.7085, "lr": 9.993094256833178e-06, "epoch": 0.23109243697478993, "percentage": 11.55, "elapsed_time": "0:09:44", "remaining_time": "1:14:31"}
+{"current_steps": 340, "total_steps": 2856, "loss": 0.6222, "lr": 9.989510052114473e-06, "epoch": 0.23809523809523808, "percentage": 11.9, "elapsed_time": "0:10:01", "remaining_time": "1:14:09"}
+{"current_steps": 350, "total_steps": 2856, "loss": 0.7224, "lr": 9.985180281570976e-06, "epoch": 0.24509803921568626, "percentage": 12.25, "elapsed_time": "0:10:18", "remaining_time": "1:13:48"}
+{"current_steps": 360, "total_steps": 2856, "loss": 0.7097, "lr": 9.980105592185838e-06, "epoch": 0.25210084033613445, "percentage": 12.61, "elapsed_time": "0:10:35", "remaining_time": "1:13:26"}
+{"current_steps": 370, "total_steps": 2856, "loss": 0.7017, "lr": 9.97428674225295e-06, "epoch": 0.25910364145658266, "percentage": 12.96, "elapsed_time": "0:10:52", "remaining_time": "1:13:01"}
+{"current_steps": 380, "total_steps": 2856, "loss": 0.7097, "lr": 9.967724601263624e-06, "epoch": 0.2661064425770308, "percentage": 13.31, "elapsed_time": "0:11:09", "remaining_time": "1:12:41"}
+{"current_steps": 390, "total_steps": 2856, "loss": 0.7022, "lr": 9.960420149776674e-06, "epoch": 0.27310924369747897, "percentage": 13.66, "elapsed_time": "0:11:26", "remaining_time": "1:12:21"}
+{"current_steps": 400, "total_steps": 2856, "loss": 0.7519, "lr": 9.952374479271894e-06, "epoch": 0.2801120448179272, "percentage": 14.01, "elapsed_time": "0:11:43", "remaining_time": "1:11:59"}
+{"current_steps": 410, "total_steps": 2856, "loss": 0.6911, "lr": 9.943588791986956e-06, "epoch": 0.28711484593837533, "percentage": 14.36, "elapsed_time": "0:12:00", "remaining_time": "1:11:39"}
+{"current_steps": 420, "total_steps": 2856, "loss": 0.7083, "lr": 9.934064400737776e-06, "epoch": 0.29411764705882354, "percentage": 14.71, "elapsed_time": "0:12:17", "remaining_time": "1:11:18"}
+{"current_steps": 430, "total_steps": 2856, "loss": 0.7803, "lr": 9.923802728722326e-06, "epoch": 0.3011204481792717, "percentage": 15.06, "elapsed_time": "0:12:34", "remaining_time": "1:10:55"}
+{"current_steps": 440, "total_steps": 2856, "loss": 0.7277, "lr": 9.912805309307994e-06, "epoch": 0.3081232492997199, "percentage": 15.41, "elapsed_time": "0:12:51", "remaining_time": "1:10:36"}
+{"current_steps": 450, "total_steps": 2856, "loss": 0.6997, "lr": 9.901073785802433e-06, "epoch": 0.31512605042016806, "percentage": 15.76, "elapsed_time": "0:13:08", "remaining_time": "1:10:18"}
+{"current_steps": 460, "total_steps": 2856, "loss": 0.6917, "lr": 9.888609911208024e-06, "epoch": 0.32212885154061627, "percentage": 16.11, "elapsed_time": "0:13:25", "remaining_time": "1:09:55"}
+{"current_steps": 470, "total_steps": 2856, "loss": 0.7042, "lr": 9.875415547959926e-06, "epoch": 0.3291316526610644, "percentage": 16.46, "elapsed_time": "0:13:42", "remaining_time": "1:09:35"}
+{"current_steps": 480, "total_steps": 2856, "loss": 0.6818, "lr": 9.861492667647776e-06, "epoch": 0.33613445378151263, "percentage": 16.81, "elapsed_time": "0:13:59", "remaining_time": "1:09:15"}
+{"current_steps": 490, "total_steps": 2856, "loss": 0.7252, "lr": 9.846843350721084e-06, "epoch": 0.3431372549019608, "percentage": 17.16, "elapsed_time": "0:14:16", "remaining_time": "1:08:54"}
+{"current_steps": 500, "total_steps": 2856, "loss": 0.7347, "lr": 9.831469786178359e-06, "epoch": 0.35014005602240894, "percentage": 17.51, "elapsed_time": "0:14:33", "remaining_time": "1:08:34"}
+{"current_steps": 510, "total_steps": 2856, "loss": 0.6773, "lr": 9.81537427124001e-06, "epoch": 0.35714285714285715, "percentage": 17.86, "elapsed_time": "0:14:50", "remaining_time": "1:08:15"}
+{"current_steps": 520, "total_steps": 2856, "loss": 0.6925, "lr": 9.798559211005083e-06, "epoch": 0.3641456582633053, "percentage": 18.21, "elapsed_time": "0:15:07", "remaining_time": "1:07:56"}
+{"current_steps": 530, "total_steps": 2856, "loss": 0.6953, "lr": 9.781027118091879e-06, "epoch": 0.3711484593837535, "percentage": 18.56, "elapsed_time": "0:15:24", "remaining_time": "1:07:37"}
+{"current_steps": 540, "total_steps": 2856, "loss": 0.7292, "lr": 9.76278061226249e-06, "epoch": 0.37815126050420167, "percentage": 18.91, "elapsed_time": "0:15:41", "remaining_time": "1:07:15"}
+{"current_steps": 550, "total_steps": 2856, "loss": 0.7072, "lr": 9.74382242003135e-06, "epoch": 0.3851540616246499, "percentage": 19.26, "elapsed_time": "0:15:58", "remaining_time": "1:06:56"}
+{"current_steps": 560, "total_steps": 2856, "loss": 0.6132, "lr": 9.724155374257816e-06, "epoch": 0.39215686274509803, "percentage": 19.61, "elapsed_time": "0:16:15", "remaining_time": "1:06:37"}
+{"current_steps": 570, "total_steps": 2856, "loss": 0.7118, "lr": 9.703782413722856e-06, "epoch": 0.39915966386554624, "percentage": 19.96, "elapsed_time": "0:16:32", "remaining_time": "1:06:19"}
+{"current_steps": 580, "total_steps": 2856, "loss": 0.6837, "lr": 9.682706582689932e-06, "epoch": 0.4061624649859944, "percentage": 20.31, "elapsed_time": "0:16:49", "remaining_time": "1:06:01"}
+{"current_steps": 590, "total_steps": 2856, "loss": 0.7782, "lr": 9.660931030450092e-06, "epoch": 0.41316526610644255, "percentage": 20.66, "elapsed_time": "0:17:06", "remaining_time": "1:05:41"}
+{"current_steps": 600, "total_steps": 2856, "loss": 0.6569, "lr": 9.638459010851389e-06, "epoch": 0.42016806722689076, "percentage": 21.01, "elapsed_time": "0:17:23", "remaining_time": "1:05:23"}
+{"current_steps": 610, "total_steps": 2856, "loss": 0.6597, "lr": 9.615293881812666e-06, "epoch": 0.4271708683473389, "percentage": 21.36, "elapsed_time": "0:17:39", "remaining_time": "1:05:02"}
+{"current_steps": 620, "total_steps": 2856, "loss": 0.7102, "lr": 9.591439104821795e-06, "epoch": 0.4341736694677871, "percentage": 21.71, "elapsed_time": "0:17:56", "remaining_time": "1:04:42"}
+{"current_steps": 630, "total_steps": 2856, "loss": 0.7036, "lr": 9.566898244418443e-06, "epoch": 0.4411764705882353, "percentage": 22.06, "elapsed_time": "0:18:13", "remaining_time": "1:04:25"}
+{"current_steps": 640, "total_steps": 2856, "loss": 0.6853, "lr": 9.541674967661424e-06, "epoch": 0.4481792717086835, "percentage": 22.41, "elapsed_time": "0:18:31", "remaining_time": "1:04:07"}
+{"current_steps": 650, "total_steps": 2856, "loss": 0.6587, "lr": 9.515773043580754e-06, "epoch": 0.45518207282913165, "percentage": 22.76, "elapsed_time": "0:18:48", "remaining_time": "1:03:49"}
+{"current_steps": 660, "total_steps": 2856, "loss": 0.6941, "lr": 9.489196342614447e-06, "epoch": 0.46218487394957986, "percentage": 23.11, "elapsed_time": "0:19:05", "remaining_time": "1:03:29"}
+{"current_steps": 670, "total_steps": 2856, "loss": 0.6527, "lr": 9.461948836030179e-06, "epoch": 0.469187675070028, "percentage": 23.46, "elapsed_time": "0:19:22", "remaining_time": "1:03:12"}
+{"current_steps": 680, "total_steps": 2856, "loss": 0.5537, "lr": 9.434034595331863e-06, "epoch": 0.47619047619047616, "percentage": 23.81, "elapsed_time": "0:19:39", "remaining_time": "1:02:53"}
+{"current_steps": 690, "total_steps": 2856, "loss": 0.7455, "lr": 9.405457791651272e-06, "epoch": 0.4831932773109244, "percentage": 24.16, "elapsed_time": "0:19:56", "remaining_time": "1:02:35"}
+{"current_steps": 700, "total_steps": 2856, "loss": 0.6781, "lr": 9.376222695124748e-06, "epoch": 0.49019607843137253, "percentage": 24.51, "elapsed_time": "0:20:13", "remaining_time": "1:02:17"}
+{"current_steps": 710, "total_steps": 2856, "loss": 0.6807, "lr": 9.346333674255132e-06, "epoch": 0.49719887955182074, "percentage": 24.86, "elapsed_time": "0:20:30", "remaining_time": "1:01:59"}
+{"current_steps": 720, "total_steps": 2856, "loss": 0.7009, "lr": 9.315795195259003e-06, "epoch": 0.5042016806722689, "percentage": 25.21, "elapsed_time": "0:20:47", "remaining_time": "1:01:41"}
+{"current_steps": 730, "total_steps": 2856, "loss": 0.6949, "lr": 9.284611821399295e-06, "epoch": 0.511204481792717, "percentage": 25.56, "elapsed_time": "0:21:05", "remaining_time": "1:01:24"}
+{"current_steps": 740, "total_steps": 2856, "loss": 0.695, "lr": 9.252788212303426e-06, "epoch": 0.5182072829131653, "percentage": 25.91, "elapsed_time": "0:21:21", "remaining_time": "1:01:05"}
+{"current_steps": 750, "total_steps": 2856, "loss": 0.6498, "lr": 9.220329123267031e-06, "epoch": 0.5252100840336135, "percentage": 26.26, "elapsed_time": "0:21:38", "remaining_time": "1:00:47"}
+{"current_steps": 760, "total_steps": 2856, "loss": 0.6761, "lr": 9.187239404543387e-06, "epoch": 0.5322128851540616, "percentage": 26.61, "elapsed_time": "0:21:56", "remaining_time": "1:00:29"}
+{"current_steps": 770, "total_steps": 2856, "loss": 0.6895, "lr": 9.153524000618663e-06, "epoch": 0.5392156862745098, "percentage": 26.96, "elapsed_time": "0:22:12", "remaining_time": "1:00:10"}
+{"current_steps": 780, "total_steps": 2856, "loss": 0.6221, "lr": 9.119187949473075e-06, "epoch": 0.5462184873949579, "percentage": 27.31, "elapsed_time": "0:22:29", "remaining_time": "0:59:51"}
+{"current_steps": 790, "total_steps": 2856, "loss": 0.6749, "lr": 9.08423638182808e-06, "epoch": 0.5532212885154062, "percentage": 27.66, "elapsed_time": "0:22:46", "remaining_time": "0:59:33"}
+{"current_steps": 800, "total_steps": 2856, "loss": 0.7179, "lr": 9.048674520379715e-06, "epoch": 0.5602240896358543, "percentage": 28.01, "elapsed_time": "0:23:03", "remaining_time": "0:59:14"}
+{"current_steps": 810, "total_steps": 2856, "loss": 0.6926, "lr": 9.012507679018177e-06, "epoch": 0.5672268907563025, "percentage": 28.36, "elapsed_time": "0:23:20", "remaining_time": "0:58:56"}
+{"current_steps": 820, "total_steps": 2856, "loss": 0.7389, "lr": 8.975741262033793e-06, "epoch": 0.5742296918767507, "percentage": 28.71, "elapsed_time": "0:23:37", "remaining_time": "0:58:39"}
+{"current_steps": 830, "total_steps": 2856, "loss": 0.6424, "lr": 8.93838076330947e-06, "epoch": 0.5812324929971989, "percentage": 29.06, "elapsed_time": "0:23:54", "remaining_time": "0:58:20"}
+{"current_steps": 840, "total_steps": 2856, "loss": 0.6971, "lr": 8.900431765499773e-06, "epoch": 0.5882352941176471, "percentage": 29.41, "elapsed_time": "0:24:11", "remaining_time": "0:58:02"}
+{"current_steps": 850, "total_steps": 2856, "loss": 0.6155, "lr": 8.861899939196713e-06, "epoch": 0.5952380952380952, "percentage": 29.76, "elapsed_time": "0:24:27", "remaining_time": "0:57:44"}
+{"current_steps": 860, "total_steps": 2856, "loss": 0.6244, "lr": 8.82279104208242e-06, "epoch": 0.6022408963585434, "percentage": 30.11, "elapsed_time": "0:24:44", "remaining_time": "0:57:26"}
+{"current_steps": 870, "total_steps": 2856, "loss": 0.7264, "lr": 8.783110918068784e-06, "epoch": 0.6092436974789915, "percentage": 30.46, "elapsed_time": "0:25:01", "remaining_time": "0:57:08"}
+{"current_steps": 880, "total_steps": 2856, "loss": 0.6768, "lr": 8.742865496424228e-06, "epoch": 0.6162464985994398, "percentage": 30.81, "elapsed_time": "0:25:18", "remaining_time": "0:56:50"}
+{"current_steps": 890, "total_steps": 2856, "loss": 0.5889, "lr": 8.70206079088771e-06, "epoch": 0.623249299719888, "percentage": 31.16, "elapsed_time": "0:25:35", "remaining_time": "0:56:32"}
+{"current_steps": 900, "total_steps": 2856, "loss": 0.6492, "lr": 8.660702898770113e-06, "epoch": 0.6302521008403361, "percentage": 31.51, "elapsed_time": "0:25:52", "remaining_time": "0:56:14"}
+{"current_steps": 910, "total_steps": 2856, "loss": 0.6209, "lr": 8.618798000043142e-06, "epoch": 0.6372549019607843, "percentage": 31.86, "elapsed_time": "0:26:09", "remaining_time": "0:55:56"}
+{"current_steps": 920, "total_steps": 2856, "loss": 0.6407, "lr": 8.576352356415876e-06, "epoch": 0.6442577030812325, "percentage": 32.21, "elapsed_time": "0:26:26", "remaining_time": "0:55:38"}
+{"current_steps": 930, "total_steps": 2856, "loss": 0.6735, "lr": 8.533372310399093e-06, "epoch": 0.6512605042016807, "percentage": 32.56, "elapsed_time": "0:26:43", "remaining_time": "0:55:21"}
+{"current_steps": 940, "total_steps": 2856, "loss": 0.6023, "lr": 8.48986428435754e-06, "epoch": 0.6582633053221288, "percentage": 32.91, "elapsed_time": "0:27:00", "remaining_time": "0:55:03"}
+{"current_steps": 950, "total_steps": 2856, "loss": 0.5726, "lr": 8.445834779550257e-06, "epoch": 0.665266106442577, "percentage": 33.26, "elapsed_time": "0:27:17", "remaining_time": "0:54:45"}
+{"current_steps": 960, "total_steps": 2856, "loss": 0.6442, "lr": 8.40129037515911e-06, "epoch": 0.6722689075630253, "percentage": 33.61, "elapsed_time": "0:27:34", "remaining_time": "0:54:27"}
+{"current_steps": 970, "total_steps": 2856, "loss": 0.6775, "lr": 8.356237727305695e-06, "epoch": 0.6792717086834734, "percentage": 33.96, "elapsed_time": "0:27:51", "remaining_time": "0:54:10"}
+{"current_steps": 980, "total_steps": 2856, "loss": 0.6076, "lr": 8.310683568056725e-06, "epoch": 0.6862745098039216, "percentage": 34.31, "elapsed_time": "0:28:08", "remaining_time": "0:53:52"}
+{"current_steps": 990, "total_steps": 2856, "loss": 0.6701, "lr": 8.264634704418095e-06, "epoch": 0.6932773109243697, "percentage": 34.66, "elapsed_time": "0:28:25", "remaining_time": "0:53:34"}
+{"current_steps": 1000, "total_steps": 2856, "loss": 0.6211, "lr": 8.218098017317715e-06, "epoch": 0.7002801120448179, "percentage": 35.01, "elapsed_time": "0:28:42", "remaining_time": "0:53:17"}
+{"current_steps": 1010, "total_steps": 2856, "loss": 0.6853, "lr": 8.171080460577337e-06, "epoch": 0.7072829131652661, "percentage": 35.36, "elapsed_time": "0:29:00", "remaining_time": "0:53:00"}
+{"current_steps": 1020, "total_steps": 2856, "loss": 0.6516, "lr": 8.123589059873445e-06, "epoch": 0.7142857142857143, "percentage": 35.71, "elapsed_time": "0:29:16", "remaining_time": "0:52:42"}
+{"current_steps": 1030, "total_steps": 2856, "loss": 0.5853, "lr": 8.075630911687451e-06, "epoch": 0.7212885154061625, "percentage": 36.06, "elapsed_time": "0:29:33", "remaining_time": "0:52:24"}
+{"current_steps": 1040, "total_steps": 2856, "loss": 0.6107, "lr": 8.027213182245289e-06, "epoch": 0.7282913165266106, "percentage": 36.41, "elapsed_time": "0:29:50", "remaining_time": "0:52:07"}
+{"current_steps": 1050, "total_steps": 2856, "loss": 0.7112, "lr": 7.978343106446575e-06, "epoch": 0.7352941176470589, "percentage": 36.76, "elapsed_time": "0:30:08", "remaining_time": "0:51:49"}
+{"current_steps": 1060, "total_steps": 2856, "loss": 0.6573, "lr": 7.929027986783538e-06, "epoch": 0.742296918767507, "percentage": 37.11, "elapsed_time": "0:30:25", "remaining_time": "0:51:32"}
+{"current_steps": 1070, "total_steps": 2856, "loss": 0.6383, "lr": 7.879275192249822e-06, "epoch": 0.7492997198879552, "percentage": 37.46, "elapsed_time": "0:30:42", "remaining_time": "0:51:14"}
+{"current_steps": 1080, "total_steps": 2856, "loss": 0.6707, "lr": 7.829092157239369e-06, "epoch": 0.7563025210084033, "percentage": 37.82, "elapsed_time": "0:30:58", "remaining_time": "0:50:57"}
+{"current_steps": 1090, "total_steps": 2856, "loss": 0.5754, "lr": 7.778486380435512e-06, "epoch": 0.7633053221288515, "percentage": 38.17, "elapsed_time": "0:31:15", "remaining_time": "0:50:38"}
+{"current_steps": 1100, "total_steps": 2856, "loss": 0.6674, "lr": 7.727465423690487e-06, "epoch": 0.7703081232492998, "percentage": 38.52, "elapsed_time": "0:31:32", "remaining_time": "0:50:21"}
+{"current_steps": 1110, "total_steps": 2856, "loss": 0.6436, "lr": 7.676036910895475e-06, "epoch": 0.7773109243697479, "percentage": 38.87, "elapsed_time": "0:31:49", "remaining_time": "0:50:03"}
+{"current_steps": 1120, "total_steps": 2856, "loss": 0.5633, "lr": 7.624208526841399e-06, "epoch": 0.7843137254901961, "percentage": 39.22, "elapsed_time": "0:32:06", "remaining_time": "0:49:46"}
+{"current_steps": 1130, "total_steps": 2856, "loss": 0.6534, "lr": 7.571988016070611e-06, "epoch": 0.7913165266106442, "percentage": 39.57, "elapsed_time": "0:32:23", "remaining_time": "0:49:28"}
+{"current_steps": 1140, "total_steps": 2856, "loss": 0.6152, "lr": 7.519383181719644e-06, "epoch": 0.7983193277310925, "percentage": 39.92, "elapsed_time": "0:32:39", "remaining_time": "0:49:10"}
+{"current_steps": 1150, "total_steps": 2856, "loss": 0.6775, "lr": 7.466401884353227e-06, "epoch": 0.8053221288515406, "percentage": 40.27, "elapsed_time": "0:32:57", "remaining_time": "0:48:53"}
+{"current_steps": 1160, "total_steps": 2856, "loss": 0.6514, "lr": 7.413052040789692e-06, "epoch": 0.8123249299719888, "percentage": 40.62, "elapsed_time": "0:33:14", "remaining_time": "0:48:35"}
+{"current_steps": 1170, "total_steps": 2856, "loss": 0.6194, "lr": 7.359341622918006e-06, "epoch": 0.819327731092437, "percentage": 40.97, "elapsed_time": "0:33:31", "remaining_time": "0:48:17"}
+{"current_steps": 1180, "total_steps": 2856, "loss": 0.6381, "lr": 7.305278656506547e-06, "epoch": 0.8263305322128851, "percentage": 41.32, "elapsed_time": "0:33:48", "remaining_time": "0:48:00"}
+{"current_steps": 1190, "total_steps": 2856, "loss": 0.6059, "lr": 7.2508712200038435e-06, "epoch": 0.8333333333333334, "percentage": 41.67, "elapsed_time": "0:34:05", "remaining_time": "0:47:43"}
+{"current_steps": 1200, "total_steps": 2856, "loss": 0.7107, "lr": 7.19612744333144e-06, "epoch": 0.8403361344537815, "percentage": 42.02, "elapsed_time": "0:34:21", "remaining_time": "0:47:25"}
+{"current_steps": 1210, "total_steps": 2856, "loss": 0.6062, "lr": 7.141055506669072e-06, "epoch": 0.8473389355742297, "percentage": 42.37, "elapsed_time": "0:34:38", "remaining_time": "0:47:07"}
+{"current_steps": 1220, "total_steps": 2856, "loss": 0.5956, "lr": 7.0856636392323205e-06, "epoch": 0.8543417366946778, "percentage": 42.72, "elapsed_time": "0:34:55", "remaining_time": "0:46:50"}
+{"current_steps": 1230, "total_steps": 2856, "loss": 0.5936, "lr": 7.0299601180429615e-06, "epoch": 0.8613445378151261, "percentage": 43.07, "elapsed_time": "0:35:12", "remaining_time": "0:46:32"}
+{"current_steps": 1240, "total_steps": 2856, "loss": 0.6314, "lr": 6.973953266692143e-06, "epoch": 0.8683473389355743, "percentage": 43.42, "elapsed_time": "0:35:29", "remaining_time": "0:46:15"}
+{"current_steps": 1250, "total_steps": 2856, "loss": 0.6908, "lr": 6.91765145409663e-06, "epoch": 0.8753501400560224, "percentage": 43.77, "elapsed_time": "0:35:46", "remaining_time": "0:45:57"}
+{"current_steps": 1260, "total_steps": 2856, "loss": 0.5709, "lr": 6.861063093248264e-06, "epoch": 0.8823529411764706, "percentage": 44.12, "elapsed_time": "0:36:03", "remaining_time": "0:45:39"}
+{"current_steps": 1270, "total_steps": 2856, "loss": 0.627, "lr": 6.804196639956828e-06, "epoch": 0.8893557422969187, "percentage": 44.47, "elapsed_time": "0:36:20", "remaining_time": "0:45:22"}
+{"current_steps": 1280, "total_steps": 2856, "loss": 0.5996, "lr": 6.747060591586533e-06, "epoch": 0.896358543417367, "percentage": 44.82, "elapsed_time": "0:36:37", "remaining_time": "0:45:05"}
+{"current_steps": 1290, "total_steps": 2856, "loss": 0.6943, "lr": 6.689663485786287e-06, "epoch": 0.9033613445378151, "percentage": 45.17, "elapsed_time": "0:36:54", "remaining_time": "0:44:48"}
+{"current_steps": 1300, "total_steps": 2856, "loss": 0.6311, "lr": 6.632013899213934e-06, "epoch": 0.9103641456582633, "percentage": 45.52, "elapsed_time": "0:37:11", "remaining_time": "0:44:30"}
+{"current_steps": 1310, "total_steps": 2856, "loss": 0.6081, "lr": 6.574120446254672e-06, "epoch": 0.9173669467787114, "percentage": 45.87, "elapsed_time": "0:37:28", "remaining_time": "0:44:13"}
+{"current_steps": 1320, "total_steps": 2856, "loss": 0.6685, "lr": 6.5159917777338466e-06, "epoch": 0.9243697478991597, "percentage": 46.22, "elapsed_time": "0:37:45", "remaining_time": "0:43:56"}
+{"current_steps": 1330, "total_steps": 2856, "loss": 0.5457, "lr": 6.457636579624278e-06, "epoch": 0.9313725490196079, "percentage": 46.57, "elapsed_time": "0:38:02", "remaining_time": "0:43:38"}
+{"current_steps": 1340, "total_steps": 2856, "loss": 0.6981, "lr": 6.3990635717483404e-06, "epoch": 0.938375350140056, "percentage": 46.92, "elapsed_time": "0:38:19", "remaining_time": "0:43:21"}
+{"current_steps": 1350, "total_steps": 2856, "loss": 0.6165, "lr": 6.340281506475003e-06, "epoch": 0.9453781512605042, "percentage": 47.27, "elapsed_time": "0:38:36", "remaining_time": "0:43:04"}
+{"current_steps": 1360, "total_steps": 2856, "loss": 0.5656, "lr": 6.281299167411975e-06, "epoch": 0.9523809523809523, "percentage": 47.62, "elapsed_time": "0:38:53", "remaining_time": "0:42:46"}
+{"current_steps": 1370, "total_steps": 2856, "loss": 0.6473, "lr": 6.222125368093213e-06, "epoch": 0.9593837535014006, "percentage": 47.97, "elapsed_time": "0:39:10", "remaining_time": "0:42:29"}
+{"current_steps": 1380, "total_steps": 2856, "loss": 0.6279, "lr": 6.162768950661945e-06, "epoch": 0.9663865546218487, "percentage": 48.32, "elapsed_time": "0:39:27", "remaining_time": "0:42:11"}
+{"current_steps": 1390, "total_steps": 2856, "loss": 0.5426, "lr": 6.103238784549404e-06, "epoch": 0.9733893557422969, "percentage": 48.67, "elapsed_time": "0:39:44", "remaining_time": "0:41:54"}
+{"current_steps": 1400, "total_steps": 2856, "loss": 0.6551, "lr": 6.043543765149514e-06, "epoch": 0.9803921568627451, "percentage": 49.02, "elapsed_time": "0:40:00", "remaining_time": "0:41:37"}
+{"current_steps": 1410, "total_steps": 2856, "loss": 0.6524, "lr": 5.983692812489679e-06, "epoch": 0.9873949579831933, "percentage": 49.37, "elapsed_time": "0:40:17", "remaining_time": "0:41:19"}
+{"current_steps": 1420, "total_steps": 2856, "loss": 0.5787, "lr": 5.923694869897879e-06, "epoch": 0.9943977591036415, "percentage": 49.72, "elapsed_time": "0:40:35", "remaining_time": "0:41:02"}
+{"current_steps": 1430, "total_steps": 2856, "loss": 0.6191, "lr": 5.863558902666318e-06, "epoch": 1.0014005602240896, "percentage": 50.07, "elapsed_time": "0:41:02", "remaining_time": "0:40:55"}
+{"current_steps": 1440, "total_steps": 2856, "loss": 0.4869, "lr": 5.803293896711756e-06, "epoch": 1.0084033613445378, "percentage": 50.42, "elapsed_time": "0:41:19", "remaining_time": "0:40:38"}
+{"current_steps": 1450, "total_steps": 2856, "loss": 0.4959, "lr": 5.74290885723278e-06, "epoch": 1.015406162464986, "percentage": 50.77, "elapsed_time": "0:41:36", "remaining_time": "0:40:20"}
+{"current_steps": 1460, "total_steps": 2856, "loss": 0.4611, "lr": 5.682412807364187e-06, "epoch": 1.022408963585434, "percentage": 51.12, "elapsed_time": "0:41:52", "remaining_time": "0:40:02"}
+{"current_steps": 1470, "total_steps": 2856, "loss": 0.4828, "lr": 5.621814786828683e-06, "epoch": 1.0294117647058822, "percentage": 51.47, "elapsed_time": "0:42:09", "remaining_time": "0:39:44"}
+{"current_steps": 1480, "total_steps": 2856, "loss": 0.4362, "lr": 5.5611238505861094e-06, "epoch": 1.0364145658263306, "percentage": 51.82, "elapsed_time": "0:42:26", "remaining_time": "0:39:27"}
+{"current_steps": 1490, "total_steps": 2856, "loss": 0.4811, "lr": 5.5003490674804e-06, "epoch": 1.0434173669467788, "percentage": 52.17, "elapsed_time": "0:42:43", "remaining_time": "0:39:10"}
+{"current_steps": 1500, "total_steps": 2856, "loss": 0.4525, "lr": 5.439499518884433e-06, "epoch": 1.050420168067227, "percentage": 52.52, "elapsed_time": "0:43:01", "remaining_time": "0:38:53"}
+{"current_steps": 1510, "total_steps": 2856, "loss": 0.4573, "lr": 5.378584297343053e-06, "epoch": 1.057422969187675, "percentage": 52.87, "elapsed_time": "0:43:18", "remaining_time": "0:38:36"}
+{"current_steps": 1520, "total_steps": 2856, "loss": 0.4475, "lr": 5.3176125052143905e-06, "epoch": 1.0644257703081232, "percentage": 53.22, "elapsed_time": "0:43:35", "remaining_time": "0:38:19"}
+{"current_steps": 1530, "total_steps": 2856, "loss": 0.4679, "lr": 5.256593253309728e-06, "epoch": 1.0714285714285714, "percentage": 53.57, "elapsed_time": "0:43:53", "remaining_time": "0:38:01"}
+{"current_steps": 1540, "total_steps": 2856, "loss": 0.4982, "lr": 5.195535659532111e-06, "epoch": 1.0784313725490196, "percentage": 53.92, "elapsed_time": "0:44:09", "remaining_time": "0:37:44"}
+{"current_steps": 1550, "total_steps": 2856, "loss": 0.4881, "lr": 5.134448847513873e-06, "epoch": 1.0854341736694677, "percentage": 54.27, "elapsed_time": "0:44:26", "remaining_time": "0:37:27"}
+{"current_steps": 1560, "total_steps": 2856, "loss": 0.4518, "lr": 5.073341945253336e-06, "epoch": 1.092436974789916, "percentage": 54.62, "elapsed_time": "0:44:43", "remaining_time": "0:37:09"}
+{"current_steps": 1570, "total_steps": 2856, "loss": 0.4618, "lr": 5.012224083750845e-06, "epoch": 1.0994397759103642, "percentage": 54.97, "elapsed_time": "0:45:00", "remaining_time": "0:36:51"}
+{"current_steps": 1580, "total_steps": 2856, "loss": 0.4647, "lr": 4.951104395644342e-06, "epoch": 1.1064425770308124, "percentage": 55.32, "elapsed_time": "0:45:17", "remaining_time": "0:36:34"}
+{"current_steps": 1590, "total_steps": 2856, "loss": 0.497, "lr": 4.889992013844711e-06, "epoch": 1.1134453781512605, "percentage": 55.67, "elapsed_time": "0:45:33", "remaining_time": "0:36:16"}
+{"current_steps": 1600, "total_steps": 2856, "loss": 0.5142, "lr": 4.828896070171084e-06, "epoch": 1.1204481792717087, "percentage": 56.02, "elapsed_time": "0:45:51", "remaining_time": "0:35:59"}
+{"current_steps": 1610, "total_steps": 2856, "loss": 0.449, "lr": 4.767825693986295e-06, "epoch": 1.1274509803921569, "percentage": 56.37, "elapsed_time": "0:46:08", "remaining_time": "0:35:42"}
+{"current_steps": 1620, "total_steps": 2856, "loss": 0.5264, "lr": 4.706790010832714e-06, "epoch": 1.134453781512605, "percentage": 56.72, "elapsed_time": "0:46:25", "remaining_time": "0:35:24"}
+{"current_steps": 1630, "total_steps": 2856, "loss": 0.4551, "lr": 4.6457981410686524e-06, "epoch": 1.1414565826330532, "percentage": 57.07, "elapsed_time": "0:46:42", "remaining_time": "0:35:07"}
+{"current_steps": 1640, "total_steps": 2856, "loss": 0.5141, "lr": 4.584859198505526e-06, "epoch": 1.1484593837535013, "percentage": 57.42, "elapsed_time": "0:46:59", "remaining_time": "0:34:50"}
+{"current_steps": 1650, "total_steps": 2856, "loss": 0.4254, "lr": 4.5239822890460174e-06, "epoch": 1.1554621848739495, "percentage": 57.77, "elapsed_time": "0:47:15", "remaining_time": "0:34:32"}
+{"current_steps": 1660, "total_steps": 2856, "loss": 0.4986, "lr": 4.463176509323413e-06, "epoch": 1.1624649859943978, "percentage": 58.12, "elapsed_time": "0:47:32", "remaining_time": "0:34:15"}
+{"current_steps": 1670, "total_steps": 2856, "loss": 0.3929, "lr": 4.402450945342317e-06, "epoch": 1.169467787114846, "percentage": 58.47, "elapsed_time": "0:47:49", "remaining_time": "0:33:57"}
+{"current_steps": 1680, "total_steps": 2856, "loss": 0.5022, "lr": 4.341814671120965e-06, "epoch": 1.1764705882352942, "percentage": 58.82, "elapsed_time": "0:48:06", "remaining_time": "0:33:40"}
+{"current_steps": 1690, "total_steps": 2856, "loss": 0.5201, "lr": 4.2812767473353205e-06, "epoch": 1.1834733893557423, "percentage": 59.17, "elapsed_time": "0:48:23", "remaining_time": "0:33:23"}
+{"current_steps": 1700, "total_steps": 2856, "loss": 0.4627, "lr": 4.220846219965168e-06, "epoch": 1.1904761904761905, "percentage": 59.52, "elapsed_time": "0:48:40", "remaining_time": "0:33:05"}
+{"current_steps": 1710, "total_steps": 2856, "loss": 0.4312, "lr": 4.160532118942411e-06, "epoch": 1.1974789915966386, "percentage": 59.87, "elapsed_time": "0:48:57", "remaining_time": "0:32:48"}
+{"current_steps": 1720, "total_steps": 2856, "loss": 0.4317, "lr": 4.100343456801747e-06, "epoch": 1.2044817927170868, "percentage": 60.22, "elapsed_time": "0:49:14", "remaining_time": "0:32:31"}
+{"current_steps": 1730, "total_steps": 2856, "loss": 0.4521, "lr": 4.040289227333961e-06, "epoch": 1.211484593837535, "percentage": 60.57, "elapsed_time": "0:49:31", "remaining_time": "0:32:14"}
+{"current_steps": 1740, "total_steps": 2856, "loss": 0.4432, "lr": 3.980378404242014e-06, "epoch": 1.2184873949579833, "percentage": 60.92, "elapsed_time": "0:49:47", "remaining_time": "0:31:56"}
+{"current_steps": 1750, "total_steps": 2856, "loss": 0.4613, "lr": 3.920619939800131e-06, "epoch": 1.2254901960784315, "percentage": 61.27, "elapsed_time": "0:50:04", "remaining_time": "0:31:39"}
+{"current_steps": 1760, "total_steps": 2856, "loss": 0.4774, "lr": 3.861022763516091e-06, "epoch": 1.2324929971988796, "percentage": 61.62, "elapsed_time": "0:50:21", "remaining_time": "0:31:21"}
+{"current_steps": 1770, "total_steps": 2856, "loss": 0.4879, "lr": 3.8015957807969247e-06, "epoch": 1.2394957983193278, "percentage": 61.97, "elapsed_time": "0:50:38", "remaining_time": "0:31:04"}
+{"current_steps": 1780, "total_steps": 2856, "loss": 0.456, "lr": 3.7423478716182026e-06, "epoch": 1.246498599439776, "percentage": 62.32, "elapsed_time": "0:50:55", "remaining_time": "0:30:46"}
+{"current_steps": 1790, "total_steps": 2856, "loss": 0.4644, "lr": 3.68328788919714e-06, "epoch": 1.253501400560224, "percentage": 62.68, "elapsed_time": "0:51:12", "remaining_time": "0:30:29"}
+{"current_steps": 1800, "total_steps": 2856, "loss": 0.5225, "lr": 3.624424658669682e-06, "epoch": 1.2605042016806722, "percentage": 63.03, "elapsed_time": "0:51:28", "remaining_time": "0:30:12"}
+{"current_steps": 1810, "total_steps": 2856, "loss": 0.3921, "lr": 3.5657669757717927e-06, "epoch": 1.2675070028011204, "percentage": 63.38, "elapsed_time": "0:51:46", "remaining_time": "0:29:54"}
+{"current_steps": 1820, "total_steps": 2856, "loss": 0.4793, "lr": 3.5073236055251425e-06, "epoch": 1.2745098039215685, "percentage": 63.73, "elapsed_time": "0:52:03", "remaining_time": "0:29:37"}
+{"current_steps": 1830, "total_steps": 2856, "loss": 0.4526, "lr": 3.4491032809273784e-06, "epoch": 1.2815126050420167, "percentage": 64.08, "elapsed_time": "0:52:20", "remaining_time": "0:29:20"}
+{"current_steps": 1840, "total_steps": 2856, "loss": 0.4855, "lr": 3.3911147016471784e-06, "epoch": 1.2885154061624648, "percentage": 64.43, "elapsed_time": "0:52:37", "remaining_time": "0:29:03"}
+{"current_steps": 1850, "total_steps": 2856, "loss": 0.4522, "lr": 3.3333665327242965e-06, "epoch": 1.2955182072829132, "percentage": 64.78, "elapsed_time": "0:52:54", "remaining_time": "0:28:46"}
+{"current_steps": 1860, "total_steps": 2856, "loss": 0.4965, "lr": 3.275867403274763e-06, "epoch": 1.3025210084033614, "percentage": 65.13, "elapsed_time": "0:53:11", "remaining_time": "0:28:28"}
+{"current_steps": 1870, "total_steps": 2856, "loss": 0.4635, "lr": 3.2186259052014797e-06, "epoch": 1.3095238095238095, "percentage": 65.48, "elapsed_time": "0:53:28", "remaining_time": "0:28:11"}
+{"current_steps": 1880, "total_steps": 2856, "loss": 0.44, "lr": 3.1616505919103446e-06, "epoch": 1.3165266106442577, "percentage": 65.83, "elapsed_time": "0:53:45", "remaining_time": "0:27:54"}
+{"current_steps": 1890, "total_steps": 2856, "loss": 0.511, "lr": 3.1049499770321572e-06, "epoch": 1.3235294117647058, "percentage": 66.18, "elapsed_time": "0:54:03", "remaining_time": "0:27:37"}
+{"current_steps": 1900, "total_steps": 2856, "loss": 0.4298, "lr": 3.0485325331504477e-06, "epoch": 1.330532212885154, "percentage": 66.53, "elapsed_time": "0:54:20", "remaining_time": "0:27:20"}
+{"current_steps": 1910, "total_steps": 2856, "loss": 0.4353, "lr": 2.992406690535453e-06, "epoch": 1.3375350140056024, "percentage": 66.88, "elapsed_time": "0:54:37", "remaining_time": "0:27:03"}
+{"current_steps": 1920, "total_steps": 2856, "loss": 0.5351, "lr": 2.936580835884404e-06, "epoch": 1.3445378151260505, "percentage": 67.23, "elapsed_time": "0:54:54", "remaining_time": "0:26:45"}
+{"current_steps": 1930, "total_steps": 2856, "loss": 0.4736, "lr": 2.8810633110683312e-06, "epoch": 1.3515406162464987, "percentage": 67.58, "elapsed_time": "0:55:10", "remaining_time": "0:26:28"}
+{"current_steps": 1940, "total_steps": 2856, "loss": 0.497, "lr": 2.8258624118855704e-06, "epoch": 1.3585434173669468, "percentage": 67.93, "elapsed_time": "0:55:27", "remaining_time": "0:26:11"}
+{"current_steps": 1950, "total_steps": 2856, "loss": 0.4561, "lr": 2.770986386822145e-06, "epoch": 1.365546218487395, "percentage": 68.28, "elapsed_time": "0:55:44", "remaining_time": "0:25:53"}
+{"current_steps": 1960, "total_steps": 2856, "loss": 0.4771, "lr": 2.7164434358192236e-06, "epoch": 1.3725490196078431, "percentage": 68.63, "elapsed_time": "0:56:01", "remaining_time": "0:25:36"}
+{"current_steps": 1970, "total_steps": 2856, "loss": 0.4244, "lr": 2.6622417090478304e-06, "epoch": 1.3795518207282913, "percentage": 68.98, "elapsed_time": "0:56:18", "remaining_time": "0:25:19"}
+{"current_steps": 1980, "total_steps": 2856, "loss": 0.4818, "lr": 2.6083893056909937e-06, "epoch": 1.3865546218487395, "percentage": 69.33, "elapsed_time": "0:56:35", "remaining_time": "0:25:02"}
+{"current_steps": 1990, "total_steps": 2856, "loss": 0.4473, "lr": 2.5548942727335046e-06, "epoch": 1.3935574229691876, "percentage": 69.68, "elapsed_time": "0:56:52", "remaining_time": "0:24:44"}
+{"current_steps": 2000, "total_steps": 2856, "loss": 0.4653, "lr": 2.5017646037594967e-06, "epoch": 1.4005602240896358, "percentage": 70.03, "elapsed_time": "0:57:09", "remaining_time": "0:24:27"}
+{"current_steps": 2010, "total_steps": 2856, "loss": 0.4945, "lr": 2.4490082377579805e-06, "epoch": 1.407563025210084, "percentage": 70.38, "elapsed_time": "0:57:25", "remaining_time": "0:24:10"}
+{"current_steps": 2020, "total_steps": 2856, "loss": 0.4371, "lr": 2.3966330579365436e-06, "epoch": 1.4145658263305323, "percentage": 70.73, "elapsed_time": "0:57:42", "remaining_time": "0:23:53"}
+{"current_steps": 2030, "total_steps": 2856, "loss": 0.4495, "lr": 2.3446468905434093e-06, "epoch": 1.4215686274509804, "percentage": 71.08, "elapsed_time": "0:57:59", "remaining_time": "0:23:35"}
+{"current_steps": 2040, "total_steps": 2856, "loss": 0.4552, "lr": 2.2930575036979634e-06, "epoch": 1.4285714285714286, "percentage": 71.43, "elapsed_time": "0:58:16", "remaining_time": "0:23:18"}
+{"current_steps": 2050, "total_steps": 2856, "loss": 0.4378, "lr": 2.241872606229998e-06, "epoch": 1.4355742296918768, "percentage": 71.78, "elapsed_time": "0:58:33", "remaining_time": "0:23:01"}
+{"current_steps": 2060, "total_steps": 2856, "loss": 0.5055, "lr": 2.191099846527813e-06, "epoch": 1.442577030812325, "percentage": 72.13, "elapsed_time": "0:58:49", "remaining_time": "0:22:43"}
+{"current_steps": 2070, "total_steps": 2856, "loss": 0.4727, "lr": 2.14074681139534e-06, "epoch": 1.449579831932773, "percentage": 72.48, "elapsed_time": "0:59:07", "remaining_time": "0:22:26"}
+{"current_steps": 2080, "total_steps": 2856, "loss": 0.4826, "lr": 2.090821024918462e-06, "epoch": 1.4565826330532212, "percentage": 72.83, "elapsed_time": "0:59:23", "remaining_time": "0:22:09"}
+{"current_steps": 2090, "total_steps": 2856, "loss": 0.4162, "lr": 2.0413299473407285e-06, "epoch": 1.4635854341736696, "percentage": 73.18, "elapsed_time": "0:59:41", "remaining_time": "0:21:52"}
+{"current_steps": 2100, "total_steps": 2856, "loss": 0.4516, "lr": 1.9922809739485883e-06, "epoch": 1.4705882352941178, "percentage": 73.53, "elapsed_time": "0:59:58", "remaining_time": "0:21:35"}
+{"current_steps": 2110, "total_steps": 2856, "loss": 0.4727, "lr": 1.943681433966338e-06, "epoch": 1.477591036414566, "percentage": 73.88, "elapsed_time": "1:00:15", "remaining_time": "0:21:18"}
+{"current_steps": 2120, "total_steps": 2856, "loss": 0.4835, "lr": 1.8955385894609414e-06, "epoch": 1.484593837535014, "percentage": 74.23, "elapsed_time": "1:00:31", "remaining_time": "0:21:00"}
+{"current_steps": 2130, "total_steps": 2856, "loss": 0.5, "lr": 1.8478596342568827e-06, "epoch": 1.4915966386554622, "percentage": 74.58, "elapsed_time": "1:00:48", "remaining_time": "0:20:43"}
+{"current_steps": 2140, "total_steps": 2856, "loss": 0.4783, "lr": 1.8006516928612121e-06, "epoch": 1.4985994397759104, "percentage": 74.93, "elapsed_time": "1:01:05", "remaining_time": "0:20:26"}
+{"current_steps": 2150, "total_steps": 2856, "loss": 0.4732, "lr": 1.753921819398966e-06, "epoch": 1.5056022408963585, "percentage": 75.28, "elapsed_time": "1:01:22", "remaining_time": "0:20:09"}
+{"current_steps": 2160, "total_steps": 2856, "loss": 0.4846, "lr": 1.7076769965590834e-06, "epoch": 1.5126050420168067, "percentage": 75.63, "elapsed_time": "1:01:39", "remaining_time": "0:19:52"}
+{"current_steps": 2170, "total_steps": 2856, "loss": 0.5196, "lr": 1.6619241345510057e-06, "epoch": 1.5196078431372548, "percentage": 75.98, "elapsed_time": "1:01:56", "remaining_time": "0:19:34"}
+{"current_steps": 2180, "total_steps": 2856, "loss": 0.3903, "lr": 1.6166700700721121e-06, "epoch": 1.526610644257703, "percentage": 76.33, "elapsed_time": "1:02:13", "remaining_time": "0:19:17"}
+{"current_steps": 2190, "total_steps": 2856, "loss": 0.4923, "lr": 1.571921565286139e-06, "epoch": 1.5336134453781511, "percentage": 76.68, "elapsed_time": "1:02:30", "remaining_time": "0:19:00"}
+{"current_steps": 2200, "total_steps": 2856, "loss": 0.4166, "lr": 1.527685306812715e-06, "epoch": 1.5406162464985993, "percentage": 77.03, "elapsed_time": "1:02:47", "remaining_time": "0:18:43"}
+{"current_steps": 2210, "total_steps": 2856, "loss": 0.4504, "lr": 1.4839679047282206e-06, "epoch": 1.5476190476190477, "percentage": 77.38, "elapsed_time": "1:03:03", "remaining_time": "0:18:26"}
+{"current_steps": 2220, "total_steps": 2856, "loss": 0.477, "lr": 1.4407758915780578e-06, "epoch": 1.5546218487394958, "percentage": 77.73, "elapsed_time": "1:03:20", "remaining_time": "0:18:08"}
+{"current_steps": 2230, "total_steps": 2856, "loss": 0.4586, "lr": 1.3981157214005098e-06, "epoch": 1.561624649859944, "percentage": 78.08, "elapsed_time": "1:03:37", "remaining_time": "0:17:51"}
+{"current_steps": 2240, "total_steps": 2856, "loss": 0.4276, "lr": 1.3559937687623458e-06, "epoch": 1.5686274509803921, "percentage": 78.43, "elapsed_time": "1:03:54", "remaining_time": "0:17:34"}
+{"current_steps": 2250, "total_steps": 2856, "loss": 0.4206, "lr": 1.3144163278062848e-06, "epoch": 1.5756302521008403, "percentage": 78.78, "elapsed_time": "1:04:11", "remaining_time": "0:17:17"}
+{"current_steps": 2260, "total_steps": 2856, "loss": 0.4034, "lr": 1.273389611310481e-06, "epoch": 1.5826330532212887, "percentage": 79.13, "elapsed_time": "1:04:28", "remaining_time": "0:17:00"}
+{"current_steps": 2270, "total_steps": 2856, "loss": 0.4801, "lr": 1.2329197497601742e-06, "epoch": 1.5896358543417368, "percentage": 79.48, "elapsed_time": "1:04:45", "remaining_time": "0:16:43"}
+{"current_steps": 2280, "total_steps": 2856, "loss": 0.5045, "lr": 1.1930127904316286e-06, "epoch": 1.596638655462185, "percentage": 79.83, "elapsed_time": "1:05:03", "remaining_time": "0:16:26"}
+{"current_steps": 2290, "total_steps": 2856, "loss": 0.4301, "lr": 1.1536746964885065e-06, "epoch": 1.6036414565826331, "percentage": 80.18, "elapsed_time": "1:05:20", "remaining_time": "0:16:08"}
+{"current_steps": 2300, "total_steps": 2856, "loss": 0.4554, "lr": 1.114911346090819e-06, "epoch": 1.6106442577030813, "percentage": 80.53, "elapsed_time": "1:05:37", "remaining_time": "0:15:51"}
+{"current_steps": 2310, "total_steps": 2856, "loss": 0.4132, "lr": 1.0767285315165675e-06, "epoch": 1.6176470588235294, "percentage": 80.88, "elapsed_time": "1:05:54", "remaining_time": "0:15:34"}
+{"current_steps": 2320, "total_steps": 2856, "loss": 0.4707, "lr": 1.0391319582962244e-06, "epoch": 1.6246498599439776, "percentage": 81.23, "elapsed_time": "1:06:11", "remaining_time": "0:15:17"}
+{"current_steps": 2330, "total_steps": 2856, "loss": 0.422, "lr": 1.002127244360171e-06, "epoch": 1.6316526610644257, "percentage": 81.58, "elapsed_time": "1:06:28", "remaining_time": "0:15:00"}
+{"current_steps": 2340, "total_steps": 2856, "loss": 0.3684, "lr": 9.65719919199235e-07, "epoch": 1.638655462184874, "percentage": 81.93, "elapsed_time": "1:06:45", "remaining_time": "0:14:43"}
+{"current_steps": 2350, "total_steps": 2856, "loss": 0.4887, "lr": 9.299154230384383e-07, "epoch": 1.645658263305322, "percentage": 82.28, "elapsed_time": "1:07:02", "remaining_time": "0:14:26"}
+{"current_steps": 2360, "total_steps": 2856, "loss": 0.4054, "lr": 8.947191060240701e-07, "epoch": 1.6526610644257702, "percentage": 82.63, "elapsed_time": "1:07:19", "remaining_time": "0:14:09"}
+{"current_steps": 2370, "total_steps": 2856, "loss": 0.5191, "lr": 8.601362274242465e-07, "epoch": 1.6596638655462184, "percentage": 82.98, "elapsed_time": "1:07:36", "remaining_time": "0:13:51"}
+{"current_steps": 2380, "total_steps": 2856, "loss": 0.479, "lr": 8.261719548430292e-07, "epoch": 1.6666666666666665, "percentage": 83.33, "elapsed_time": "1:07:53", "remaining_time": "0:13:34"}
+{"current_steps": 2390, "total_steps": 2856, "loss": 0.4162, "lr": 7.928313634482454e-07, "epoch": 1.673669467787115, "percentage": 83.68, "elapsed_time": "1:08:10", "remaining_time": "0:13:17"}
+{"current_steps": 2400, "total_steps": 2856, "loss": 0.4662, "lr": 7.601194352131285e-07, "epoch": 1.680672268907563, "percentage": 84.03, "elapsed_time": "1:08:27", "remaining_time": "0:13:00"}
+{"current_steps": 2410, "total_steps": 2856, "loss": 0.466, "lr": 7.280410581718761e-07, "epoch": 1.6876750700280112, "percentage": 84.38, "elapsed_time": "1:08:44", "remaining_time": "0:12:43"}
+{"current_steps": 2420, "total_steps": 2856, "loss": 0.421, "lr": 6.966010256892408e-07, "epoch": 1.6946778711484594, "percentage": 84.73, "elapsed_time": "1:09:00", "remaining_time": "0:12:26"}
+{"current_steps": 2430, "total_steps": 2856, "loss": 0.4384, "lr": 6.658040357442874e-07, "epoch": 1.7016806722689075, "percentage": 85.08, "elapsed_time": "1:09:18", "remaining_time": "0:12:08"}
+{"current_steps": 2440, "total_steps": 2856, "loss": 0.4423, "lr": 6.356546902283772e-07, "epoch": 1.708683473389356, "percentage": 85.43, "elapsed_time": "1:09:35", "remaining_time": "0:11:51"}
+{"current_steps": 2450, "total_steps": 2856, "loss": 0.4507, "lr": 6.061574942575266e-07, "epoch": 1.715686274509804, "percentage": 85.78, "elapsed_time": "1:09:52", "remaining_time": "0:11:34"}
+{"current_steps": 2460, "total_steps": 2856, "loss": 0.426, "lr": 5.773168554992248e-07, "epoch": 1.7226890756302522, "percentage": 86.13, "elapsed_time": "1:10:09", "remaining_time": "0:11:17"}
+{"current_steps": 2470, "total_steps": 2856, "loss": 0.4377, "lr": 5.491370835138116e-07, "epoch": 1.7296918767507004, "percentage": 86.48, "elapsed_time": "1:10:25", "remaining_time": "0:11:00"}
+{"current_steps": 2480, "total_steps": 2856, "loss": 0.4585, "lr": 5.216223891105054e-07, "epoch": 1.7366946778711485, "percentage": 86.83, "elapsed_time": "1:10:42", "remaining_time": "0:10:43"}
+{"current_steps": 2490, "total_steps": 2856, "loss": 0.4068, "lr": 4.947768837182093e-07, "epoch": 1.7436974789915967, "percentage": 87.18, "elapsed_time": "1:10:59", "remaining_time": "0:10:26"}
+{"current_steps": 2500, "total_steps": 2856, "loss": 0.4066, "lr": 4.686045787711435e-07, "epoch": 1.7507002801120448, "percentage": 87.54, "elapsed_time": "1:11:16", "remaining_time": "0:10:09"}
+{"current_steps": 2510, "total_steps": 2856, "loss": 0.4522, "lr": 4.4310938510943533e-07, "epoch": 1.757703081232493, "percentage": 87.89, "elapsed_time": "1:11:34", "remaining_time": "0:09:51"}
+{"current_steps": 2520, "total_steps": 2856, "loss": 0.3312, "lr": 4.182951123947332e-07, "epoch": 1.7647058823529411, "percentage": 88.24, "elapsed_time": "1:11:51", "remaining_time": "0:09:34"}
+{"current_steps": 2530, "total_steps": 2856, "loss": 0.4224, "lr": 3.9416546854094294e-07, "epoch": 1.7717086834733893, "percentage": 88.59, "elapsed_time": "1:12:07", "remaining_time": "0:09:17"}
+{"current_steps": 2540, "total_steps": 2856, "loss": 0.445, "lr": 3.7072405916016353e-07, "epoch": 1.7787114845938374, "percentage": 88.94, "elapsed_time": "1:12:24", "remaining_time": "0:09:00"}
+{"current_steps": 2550, "total_steps": 2856, "loss": 0.4637, "lr": 3.479743870239188e-07, "epoch": 1.7857142857142856, "percentage": 89.29, "elapsed_time": "1:12:42", "remaining_time": "0:08:43"}
+{"current_steps": 2560, "total_steps": 2856, "loss": 0.387, "lr": 3.2591985153974383e-07, "epoch": 1.7927170868347337, "percentage": 89.64, "elapsed_time": "1:12:59", "remaining_time": "0:08:26"}
+{"current_steps": 2570, "total_steps": 2856, "loss": 0.4469, "lr": 3.0456374824322674e-07, "epoch": 1.7997198879551821, "percentage": 89.99, "elapsed_time": "1:13:15", "remaining_time": "0:08:09"}
+{"current_steps": 2580, "total_steps": 2856, "loss": 0.4039, "lr": 2.839092683055622e-07, "epoch": 1.8067226890756303, "percentage": 90.34, "elapsed_time": "1:13:32", "remaining_time": "0:07:52"}
+{"current_steps": 2590, "total_steps": 2856, "loss": 0.3812, "lr": 2.639594980567162e-07, "epoch": 1.8137254901960784, "percentage": 90.69, "elapsed_time": "1:13:49", "remaining_time": "0:07:34"}
+{"current_steps": 2600, "total_steps": 2856, "loss": 0.4212, "lr": 2.447174185242324e-07, "epoch": 1.8207282913165266, "percentage": 91.04, "elapsed_time": "1:14:06", "remaining_time": "0:07:17"}
+{"current_steps": 2610, "total_steps": 2856, "loss": 0.4137, "lr": 2.2618590498779847e-07, "epoch": 1.8277310924369747, "percentage": 91.39, "elapsed_time": "1:14:23", "remaining_time": "0:07:00"}
+{"current_steps": 2620, "total_steps": 2856, "loss": 0.4213, "lr": 2.0836772654959647e-07, "epoch": 1.8347338935574231, "percentage": 91.74, "elapsed_time": "1:14:40", "remaining_time": "0:06:43"}
+{"current_steps": 2630, "total_steps": 2856, "loss": 0.4335, "lr": 1.9126554572052813e-07, "epoch": 1.8417366946778713, "percentage": 92.09, "elapsed_time": "1:14:57", "remaining_time": "0:06:26"}
+{"current_steps": 2640, "total_steps": 2856, "loss": 0.4056, "lr": 1.74881918022361e-07, "epoch": 1.8487394957983194, "percentage": 92.44, "elapsed_time": "1:15:14", "remaining_time": "0:06:09"}
+{"current_steps": 2650, "total_steps": 2856, "loss": 0.4602, "lr": 1.5921929160587045e-07, "epoch": 1.8557422969187676, "percentage": 92.79, "elapsed_time": "1:15:31", "remaining_time": "0:05:52"}
+{"current_steps": 2660, "total_steps": 2856, "loss": 0.4231, "lr": 1.44280006885017e-07, "epoch": 1.8627450980392157, "percentage": 93.14, "elapsed_time": "1:15:47", "remaining_time": "0:05:35"}
+{"current_steps": 2670, "total_steps": 2856, "loss": 0.4286, "lr": 1.3006629618722733e-07, "epoch": 1.8697478991596639, "percentage": 93.49, "elapsed_time": "1:16:04", "remaining_time": "0:05:17"}
+{"current_steps": 2680, "total_steps": 2856, "loss": 0.4368, "lr": 1.1658028341982486e-07, "epoch": 1.876750700280112, "percentage": 93.84, "elapsed_time": "1:16:21", "remaining_time": "0:05:00"}
+{"current_steps": 2690, "total_steps": 2856, "loss": 0.4691, "lr": 1.0382398375266111e-07, "epoch": 1.8837535014005602, "percentage": 94.19, "elapsed_time": "1:16:38", "remaining_time": "0:04:43"}
+{"current_steps": 2700, "total_steps": 2856, "loss": 0.4508, "lr": 9.179930331699615e-08, "epoch": 1.8907563025210083, "percentage": 94.54, "elapsed_time": "1:16:55", "remaining_time": "0:04:26"}
+{"current_steps": 2710, "total_steps": 2856, "loss": 0.4422, "lr": 8.050803892067139e-08, "epoch": 1.8977591036414565, "percentage": 94.89, "elapsed_time": "1:17:12", "remaining_time": "0:04:09"}
+{"current_steps": 2720, "total_steps": 2856, "loss": 0.4254, "lr": 6.995187777961931e-08, "epoch": 1.9047619047619047, "percentage": 95.24, "elapsed_time": "1:17:28", "remaining_time": "0:03:52"}
+{"current_steps": 2730, "total_steps": 2856, "loss": 0.4074, "lr": 6.013239726574694e-08, "epoch": 1.9117647058823528, "percentage": 95.59, "elapsed_time": "1:17:45", "remaining_time": "0:03:35"}
+{"current_steps": 2740, "total_steps": 2856, "loss": 0.4262, "lr": 5.105106467123477e-08, "epoch": 1.918767507002801, "percentage": 95.94, "elapsed_time": "1:18:02", "remaining_time": "0:03:18"}
+{"current_steps": 2750, "total_steps": 2856, "loss": 0.4725, "lr": 4.2709236989287305e-08, "epoch": 1.9257703081232493, "percentage": 96.29, "elapsed_time": "1:18:19", "remaining_time": "0:03:01"}
+{"current_steps": 2760, "total_steps": 2856, "loss": 0.4409, "lr": 3.510816071135681e-08, "epoch": 1.9327731092436975, "percentage": 96.64, "elapsed_time": "1:18:35", "remaining_time": "0:02:44"}
+{"current_steps": 2770, "total_steps": 2856, "loss": 0.393, "lr": 2.8248971640887913e-08, "epoch": 1.9397759103641457, "percentage": 96.99, "elapsed_time": "1:18:52", "remaining_time": "0:02:26"}
+{"current_steps": 2780, "total_steps": 2856, "loss": 0.474, "lr": 2.213269472359836e-08, "epoch": 1.9467787114845938, "percentage": 97.34, "elapsed_time": "1:19:09", "remaining_time": "0:02:09"}
+{"current_steps": 2790, "total_steps": 2856, "loss": 0.4008, "lr": 1.6760243894321513e-08, "epoch": 1.9537815126050422, "percentage": 97.69, "elapsed_time": "1:19:26", "remaining_time": "0:01:52"}
+{"current_steps": 2800, "total_steps": 2856, "loss": 0.4111, "lr": 1.213242194044395e-08, "epoch": 1.9607843137254903, "percentage": 98.04, "elapsed_time": "1:19:43", "remaining_time": "0:01:35"}
+{"current_steps": 2810, "total_steps": 2856, "loss": 0.4999, "lr": 8.249920381946387e-09, "epoch": 1.9677871148459385, "percentage": 98.39, "elapsed_time": "1:20:00", "remaining_time": "0:01:18"}
+{"current_steps": 2820, "total_steps": 2856, "loss": 0.4397, "lr": 5.113319368070801e-09, "epoch": 1.9747899159663866, "percentage": 98.74, "elapsed_time": "1:20:17", "remaining_time": "0:01:01"}
+{"current_steps": 2830, "total_steps": 2856, "loss": 0.3984, "lr": 2.723087590632556e-09, "epoch": 1.9817927170868348, "percentage": 99.09, "elapsed_time": "1:20:34", "remaining_time": "0:00:44"}
+{"current_steps": 2840, "total_steps": 2856, "loss": 0.4515, "lr": 1.079582213983632e-09, "epoch": 1.988795518207283, "percentage": 99.44, "elapsed_time": "1:20:51", "remaining_time": "0:00:27"}
+{"current_steps": 2850, "total_steps": 2856, "loss": 0.4036, "lr": 1.830488216442161e-10, "epoch": 1.995798319327731, "percentage": 99.79, "elapsed_time": "1:21:08", "remaining_time": "0:00:10"}
+{"current_steps": 2856, "total_steps": 2856, "epoch": 2.0, "percentage": 100.0, "elapsed_time": "1:21:28", "remaining_time": "0:00:00"}

trainer_state.json

@@ -0,0 +1,2038 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 2856,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0070028011204481795,
+      "grad_norm": 54.822365459213835,
+      "learning_rate": 3.1468531468531473e-07,
+      "loss": 2.9142,
+      "step": 10
+    },
+    {
+      "epoch": 0.014005602240896359,
+      "grad_norm": 43.140784575015786,
+      "learning_rate": 6.643356643356644e-07,
+      "loss": 2.6364,
+      "step": 20
+    },
+    {
+      "epoch": 0.02100840336134454,
+      "grad_norm": 20.52836873662029,
+      "learning_rate": 1.013986013986014e-06,
+      "loss": 2.0887,
+      "step": 30
+    },
+    {
+      "epoch": 0.028011204481792718,
+      "grad_norm": 7.296293308558612,
+      "learning_rate": 1.3636363636363636e-06,
+      "loss": 1.593,
+      "step": 40
+    },
+    {
+      "epoch": 0.0350140056022409,
+      "grad_norm": 5.69690111508328,
+      "learning_rate": 1.7132867132867134e-06,
+      "loss": 1.4625,
+      "step": 50
+    },
+    {
+      "epoch": 0.04201680672268908,
+      "grad_norm": 4.407284596083797,
+      "learning_rate": 2.0629370629370634e-06,
+      "loss": 1.231,
+      "step": 60
+    },
+    {
+      "epoch": 0.049019607843137254,
+      "grad_norm": 4.514294110845002,
+      "learning_rate": 2.4125874125874128e-06,
+      "loss": 1.1051,
+      "step": 70
+    },
+    {
+      "epoch": 0.056022408963585436,
+      "grad_norm": 4.664107810659101,
+      "learning_rate": 2.762237762237762e-06,
+      "loss": 0.9814,
+      "step": 80
+    },
+    {
+      "epoch": 0.06302521008403361,
+      "grad_norm": 3.983542706015,
+      "learning_rate": 3.1118881118881123e-06,
+      "loss": 0.9429,
+      "step": 90
+    },
+    {
+      "epoch": 0.0700280112044818,
+      "grad_norm": 3.763711194865439,
+      "learning_rate": 3.4615384615384617e-06,
+      "loss": 0.8549,
+      "step": 100
+    },
+    {
+      "epoch": 0.07703081232492998,
+      "grad_norm": 5.299639736915794,
+      "learning_rate": 3.811188811188811e-06,
+      "loss": 0.9123,
+      "step": 110
+    },
+    {
+      "epoch": 0.08403361344537816,
+      "grad_norm": 5.376552790853205,
+      "learning_rate": 4.160839160839161e-06,
+      "loss": 0.7164,
+      "step": 120
+    },
+    {
+      "epoch": 0.09103641456582633,
+      "grad_norm": 3.979790265000713,
+      "learning_rate": 4.510489510489511e-06,
+      "loss": 0.8358,
+      "step": 130
+    },
+    {
+      "epoch": 0.09803921568627451,
+      "grad_norm": 3.2180158802612904,
+      "learning_rate": 4.86013986013986e-06,
+      "loss": 0.7863,
+      "step": 140
+    },
+    {
+      "epoch": 0.10504201680672269,
+      "grad_norm": 3.653854239138897,
+      "learning_rate": 5.20979020979021e-06,
+      "loss": 0.7784,
+      "step": 150
+    },
+    {
+      "epoch": 0.11204481792717087,
+      "grad_norm": 3.731091024980675,
+      "learning_rate": 5.5594405594405596e-06,
+      "loss": 0.696,
+      "step": 160
+    },
+    {
+      "epoch": 0.11904761904761904,
+      "grad_norm": 3.371981147503621,
+      "learning_rate": 5.90909090909091e-06,
+      "loss": 0.7731,
+      "step": 170
+    },
+    {
+      "epoch": 0.12605042016806722,
+      "grad_norm": 3.4539776878177544,
+      "learning_rate": 6.258741258741259e-06,
+      "loss": 0.7214,
+      "step": 180
+    },
+    {
+      "epoch": 0.1330532212885154,
+      "grad_norm": 4.764005243860567,
+      "learning_rate": 6.608391608391609e-06,
+      "loss": 0.7684,
+      "step": 190
+    },
+    {
+      "epoch": 0.1400560224089636,
+      "grad_norm": 3.4106980736508237,
+      "learning_rate": 6.958041958041959e-06,
+      "loss": 0.7845,
+      "step": 200
+    },
+    {
+      "epoch": 0.14705882352941177,
+      "grad_norm": 3.5675277403879213,
+      "learning_rate": 7.307692307692308e-06,
+      "loss": 0.6937,
+      "step": 210
+    },
+    {
+      "epoch": 0.15406162464985995,
+      "grad_norm": 3.6190398138061615,
+      "learning_rate": 7.657342657342658e-06,
+      "loss": 0.8302,
+      "step": 220
+    },
+    {
+      "epoch": 0.16106442577030813,
+      "grad_norm": 3.5569990424945925,
+      "learning_rate": 8.006993006993008e-06,
+      "loss": 0.7539,
+      "step": 230
+    },
+    {
+      "epoch": 0.16806722689075632,
+      "grad_norm": 2.957808482275135,
+      "learning_rate": 8.356643356643356e-06,
+      "loss": 0.7641,
+      "step": 240
+    },
+    {
+      "epoch": 0.17507002801120447,
+      "grad_norm": 2.701665888444842,
+      "learning_rate": 8.706293706293708e-06,
+      "loss": 0.7131,
+      "step": 250
+    },
+    {
+      "epoch": 0.18207282913165265,
+      "grad_norm": 3.582843229917075,
+      "learning_rate": 9.055944055944057e-06,
+      "loss": 0.595,
+      "step": 260
+    },
+    {
+      "epoch": 0.18907563025210083,
+      "grad_norm": 3.453144473645395,
+      "learning_rate": 9.405594405594406e-06,
+      "loss": 0.7676,
+      "step": 270
+    },
+    {
+      "epoch": 0.19607843137254902,
+      "grad_norm": 3.3312933939710114,
+      "learning_rate": 9.755244755244755e-06,
+      "loss": 0.7019,
+      "step": 280
+    },
+    {
+      "epoch": 0.2030812324929972,
+      "grad_norm": 3.7248118142373494,
+      "learning_rate": 9.999966378620396e-06,
+      "loss": 0.6575,
+      "step": 290
+    },
+    {
+      "epoch": 0.21008403361344538,
+      "grad_norm": 3.8896895986556688,
+      "learning_rate": 9.999368677783691e-06,
+      "loss": 0.6984,
+      "step": 300
+    },
+    {
+      "epoch": 0.21708683473389356,
+      "grad_norm": 3.182135009379796,
+      "learning_rate": 9.998023937980683e-06,
+      "loss": 0.8306,
+      "step": 310
+    },
+    {
+      "epoch": 0.22408963585434175,
+      "grad_norm": 2.8416958229620395,
+      "learning_rate": 9.995932360151348e-06,
+      "loss": 0.784,
+      "step": 320
+    },
+    {
+      "epoch": 0.23109243697478993,
+      "grad_norm": 3.150294362261809,
+      "learning_rate": 9.993094256833178e-06,
+      "loss": 0.7085,
+      "step": 330
+    },
+    {
+      "epoch": 0.23809523809523808,
+      "grad_norm": 3.0028637962745806,
+      "learning_rate": 9.989510052114473e-06,
+      "loss": 0.6222,
+      "step": 340
+    },
+    {
+      "epoch": 0.24509803921568626,
+      "grad_norm": 2.845426871309238,
+      "learning_rate": 9.985180281570976e-06,
+      "loss": 0.7224,
+      "step": 350
+    },
+    {
+      "epoch": 0.25210084033613445,
+      "grad_norm": 3.050098939646489,
+      "learning_rate": 9.980105592185838e-06,
+      "loss": 0.7097,
+      "step": 360
+    },
+    {
+      "epoch": 0.25910364145658266,
+      "grad_norm": 2.6020917053511763,
+      "learning_rate": 9.97428674225295e-06,
+      "loss": 0.7017,
+      "step": 370
+    },
+    {
+      "epoch": 0.2661064425770308,
+      "grad_norm": 3.6019215435633014,
+      "learning_rate": 9.967724601263624e-06,
+      "loss": 0.7097,
+      "step": 380
+    },
+    {
+      "epoch": 0.27310924369747897,
+      "grad_norm": 4.104445886424035,
+      "learning_rate": 9.960420149776674e-06,
+      "loss": 0.7022,
+      "step": 390
+    },
+    {
+      "epoch": 0.2801120448179272,
+      "grad_norm": 2.7297806144746204,
+      "learning_rate": 9.952374479271894e-06,
+      "loss": 0.7519,
+      "step": 400
+    },
+    {
+      "epoch": 0.28711484593837533,
+      "grad_norm": 3.4868126922293694,
+      "learning_rate": 9.943588791986956e-06,
+      "loss": 0.6911,
+      "step": 410
+    },
+    {
+      "epoch": 0.29411764705882354,
+      "grad_norm": 2.777255111697869,
+      "learning_rate": 9.934064400737776e-06,
+      "loss": 0.7083,
+      "step": 420
+    },
+    {
+      "epoch": 0.3011204481792717,
+      "grad_norm": 2.667581928295271,
+      "learning_rate": 9.923802728722326e-06,
+      "loss": 0.7803,
+      "step": 430
+    },
+    {
+      "epoch": 0.3081232492997199,
+      "grad_norm": 2.974529058609242,
+      "learning_rate": 9.912805309307994e-06,
+      "loss": 0.7277,
+      "step": 440
+    },
+    {
+      "epoch": 0.31512605042016806,
+      "grad_norm": 2.8420842979615855,
+      "learning_rate": 9.901073785802433e-06,
+      "loss": 0.6997,
+      "step": 450
+    },
+    {
+      "epoch": 0.32212885154061627,
+      "grad_norm": 2.9643793871710926,
+      "learning_rate": 9.888609911208024e-06,
+      "loss": 0.6917,
+      "step": 460
+    },
+    {
+      "epoch": 0.3291316526610644,
+      "grad_norm": 2.6029717609338254,
+      "learning_rate": 9.875415547959926e-06,
+      "loss": 0.7042,
+      "step": 470
+    },
+    {
+      "epoch": 0.33613445378151263,
+      "grad_norm": 2.6417095655684886,
+      "learning_rate": 9.861492667647776e-06,
+      "loss": 0.6818,
+      "step": 480
+    },
+    {
+      "epoch": 0.3431372549019608,
+      "grad_norm": 2.912733389632055,
+      "learning_rate": 9.846843350721084e-06,
+      "loss": 0.7252,
+      "step": 490
+    },
+    {
+      "epoch": 0.35014005602240894,
+      "grad_norm": 2.4477536058478804,
+      "learning_rate": 9.831469786178359e-06,
+      "loss": 0.7347,
+      "step": 500
+    },
+    {
+      "epoch": 0.35714285714285715,
+      "grad_norm": 2.6814440911113353,
+      "learning_rate": 9.81537427124001e-06,
+      "loss": 0.6773,
+      "step": 510
+    },
+    {
+      "epoch": 0.3641456582633053,
+      "grad_norm": 2.6474828003551125,
+      "learning_rate": 9.798559211005083e-06,
+      "loss": 0.6925,
+      "step": 520
+    },
+    {
+      "epoch": 0.3711484593837535,
+      "grad_norm": 2.96009182699851,
+      "learning_rate": 9.781027118091879e-06,
+      "loss": 0.6953,
+      "step": 530
+    },
+    {
+      "epoch": 0.37815126050420167,
+      "grad_norm": 3.047523864473747,
+      "learning_rate": 9.76278061226249e-06,
+      "loss": 0.7292,
+      "step": 540
+    },
+    {
+      "epoch": 0.3851540616246499,
+      "grad_norm": 2.991785788396449,
+      "learning_rate": 9.74382242003135e-06,
+      "loss": 0.7072,
+      "step": 550
+    },
+    {
+      "epoch": 0.39215686274509803,
+      "grad_norm": 2.369241516996275,
+      "learning_rate": 9.724155374257816e-06,
+      "loss": 0.6132,
+      "step": 560
+    },
+    {
+      "epoch": 0.39915966386554624,
+      "grad_norm": 2.8220170208873023,
+      "learning_rate": 9.703782413722856e-06,
+      "loss": 0.7118,
+      "step": 570
+    },
+    {
+      "epoch": 0.4061624649859944,
+      "grad_norm": 3.037044208560417,
+      "learning_rate": 9.682706582689932e-06,
+      "loss": 0.6837,
+      "step": 580
+    },
+    {
+      "epoch": 0.41316526610644255,
+      "grad_norm": 2.820941643724944,
+      "learning_rate": 9.660931030450092e-06,
+      "loss": 0.7782,
+      "step": 590
+    },
+    {
+      "epoch": 0.42016806722689076,
+      "grad_norm": 2.5666349293329107,
+      "learning_rate": 9.638459010851389e-06,
+      "loss": 0.6569,
+      "step": 600
+    },
+    {
+      "epoch": 0.4271708683473389,
+      "grad_norm": 2.903374170685687,
+      "learning_rate": 9.615293881812666e-06,
+      "loss": 0.6597,
+      "step": 610
+    },
+    {
+      "epoch": 0.4341736694677871,
+      "grad_norm": 2.4510273143556467,
+      "learning_rate": 9.591439104821795e-06,
+      "loss": 0.7102,
+      "step": 620
+    },
+    {
+      "epoch": 0.4411764705882353,
+      "grad_norm": 2.724792508265338,
+      "learning_rate": 9.566898244418443e-06,
+      "loss": 0.7036,
+      "step": 630
+    },
+    {
+      "epoch": 0.4481792717086835,
+      "grad_norm": 2.1578153197233183,
+      "learning_rate": 9.541674967661424e-06,
+      "loss": 0.6853,
+      "step": 640
+    },
+    {
+      "epoch": 0.45518207282913165,
+      "grad_norm": 2.2982556480501977,
+      "learning_rate": 9.515773043580754e-06,
+      "loss": 0.6587,
+      "step": 650
+    },
+    {
+      "epoch": 0.46218487394957986,
+      "grad_norm": 2.482971468008142,
+      "learning_rate": 9.489196342614447e-06,
+      "loss": 0.6941,
+      "step": 660
+    },
+    {
+      "epoch": 0.469187675070028,
+      "grad_norm": 2.792625017068267,
+      "learning_rate": 9.461948836030179e-06,
+      "loss": 0.6527,
+      "step": 670
+    },
+    {
+      "epoch": 0.47619047619047616,
+      "grad_norm": 2.8095289499726586,
+      "learning_rate": 9.434034595331863e-06,
+      "loss": 0.5537,
+      "step": 680
+    },
+    {
+      "epoch": 0.4831932773109244,
+      "grad_norm": 2.6013084541081577,
+      "learning_rate": 9.405457791651272e-06,
+      "loss": 0.7455,
+      "step": 690
+    },
+    {
+      "epoch": 0.49019607843137253,
+      "grad_norm": 2.323733631506617,
+      "learning_rate": 9.376222695124748e-06,
+      "loss": 0.6781,
+      "step": 700
+    },
+    {
+      "epoch": 0.49719887955182074,
+      "grad_norm": 2.533641065336669,
+      "learning_rate": 9.346333674255132e-06,
+      "loss": 0.6807,
+      "step": 710
+    },
+    {
+      "epoch": 0.5042016806722689,
+      "grad_norm": 2.310754494783324,
+      "learning_rate": 9.315795195259003e-06,
+      "loss": 0.7009,
+      "step": 720
+    },
+    {
+      "epoch": 0.511204481792717,
+      "grad_norm": 2.247769256472491,
+      "learning_rate": 9.284611821399295e-06,
+      "loss": 0.6949,
+      "step": 730
+    },
+    {
+      "epoch": 0.5182072829131653,
+      "grad_norm": 2.855866205125908,
+      "learning_rate": 9.252788212303426e-06,
+      "loss": 0.695,
+      "step": 740
+    },
+    {
+      "epoch": 0.5252100840336135,
+      "grad_norm": 2.2123260966081144,
+      "learning_rate": 9.220329123267031e-06,
+      "loss": 0.6498,
+      "step": 750
+    },
+    {
+      "epoch": 0.5322128851540616,
+      "grad_norm": 3.2793392899369955,
+      "learning_rate": 9.187239404543387e-06,
+      "loss": 0.6761,
+      "step": 760
+    },
+    {
+      "epoch": 0.5392156862745098,
+      "grad_norm": 2.9849801807546132,
+      "learning_rate": 9.153524000618663e-06,
+      "loss": 0.6895,
+      "step": 770
+    },
+    {
+      "epoch": 0.5462184873949579,
+      "grad_norm": 2.229602688156502,
+      "learning_rate": 9.119187949473075e-06,
+      "loss": 0.6221,
+      "step": 780
+    },
+    {
+      "epoch": 0.5532212885154062,
+      "grad_norm": 2.1724377907001164,
+      "learning_rate": 9.08423638182808e-06,
+      "loss": 0.6749,
+      "step": 790
+    },
+    {
+      "epoch": 0.5602240896358543,
+      "grad_norm": 2.4988277913931216,
+      "learning_rate": 9.048674520379715e-06,
+      "loss": 0.7179,
+      "step": 800
+    },
+    {
+      "epoch": 0.5672268907563025,
+      "grad_norm": 2.4743732152259117,
+      "learning_rate": 9.012507679018177e-06,
+      "loss": 0.6926,
+      "step": 810
+    },
+    {
+      "epoch": 0.5742296918767507,
+      "grad_norm": 2.3405306027107775,
+      "learning_rate": 8.975741262033793e-06,
+      "loss": 0.7389,
+      "step": 820
+    },
+    {
+      "epoch": 0.5812324929971989,
+      "grad_norm": 2.272548602465369,
+      "learning_rate": 8.93838076330947e-06,
+      "loss": 0.6424,
+      "step": 830
+    },
+    {
+      "epoch": 0.5882352941176471,
+      "grad_norm": 2.0400401153554935,
+      "learning_rate": 8.900431765499773e-06,
+      "loss": 0.6971,
+      "step": 840
+    },
+    {
+      "epoch": 0.5952380952380952,
+      "grad_norm": 2.084177383415346,
+      "learning_rate": 8.861899939196713e-06,
+      "loss": 0.6155,
+      "step": 850
+    },
+    {
+      "epoch": 0.6022408963585434,
+      "grad_norm": 3.794945397573399,
+      "learning_rate": 8.82279104208242e-06,
+      "loss": 0.6244,
+      "step": 860
+    },
+    {
+      "epoch": 0.6092436974789915,
+      "grad_norm": 3.49165664633522,
+      "learning_rate": 8.783110918068784e-06,
+      "loss": 0.7264,
+      "step": 870
+    },
+    {
+      "epoch": 0.6162464985994398,
+      "grad_norm": 2.039969801192167,
+      "learning_rate": 8.742865496424228e-06,
+      "loss": 0.6768,
+      "step": 880
+    },
+    {
+      "epoch": 0.623249299719888,
+      "grad_norm": 2.335677247223129,
+      "learning_rate": 8.70206079088771e-06,
+      "loss": 0.5889,
+      "step": 890
+    },
+    {
+      "epoch": 0.6302521008403361,
+      "grad_norm": 2.060225160885774,
+      "learning_rate": 8.660702898770113e-06,
+      "loss": 0.6492,
+      "step": 900
+    },
+    {
+      "epoch": 0.6372549019607843,
+      "grad_norm": 2.798146335622228,
+      "learning_rate": 8.618798000043142e-06,
+      "loss": 0.6209,
+      "step": 910
+    },
+    {
+      "epoch": 0.6442577030812325,
+      "grad_norm": 2.8734777136564493,
+      "learning_rate": 8.576352356415876e-06,
+      "loss": 0.6407,
+      "step": 920
+    },
+    {
+      "epoch": 0.6512605042016807,
+      "grad_norm": 2.6762821835109154,
+      "learning_rate": 8.533372310399093e-06,
+      "loss": 0.6735,
+      "step": 930
+    },
+    {
+      "epoch": 0.6582633053221288,
+      "grad_norm": 2.0693419886641125,
+      "learning_rate": 8.48986428435754e-06,
+      "loss": 0.6023,
+      "step": 940
+    },
+    {
+      "epoch": 0.665266106442577,
+      "grad_norm": 2.5455745068080304,
+      "learning_rate": 8.445834779550257e-06,
+      "loss": 0.5726,
+      "step": 950
+    },
+    {
+      "epoch": 0.6722689075630253,
+      "grad_norm": 2.8427810866745427,
+      "learning_rate": 8.40129037515911e-06,
+      "loss": 0.6442,
+      "step": 960
+    },
+    {
+      "epoch": 0.6792717086834734,
+      "grad_norm": 2.338498753658426,
+      "learning_rate": 8.356237727305695e-06,
+      "loss": 0.6775,
+      "step": 970
+    },
+    {
+      "epoch": 0.6862745098039216,
+      "grad_norm": 2.035099542215418,
+      "learning_rate": 8.310683568056725e-06,
+      "loss": 0.6076,
+      "step": 980
+    },
+    {
+      "epoch": 0.6932773109243697,
+      "grad_norm": 2.2530856849283394,
+      "learning_rate": 8.264634704418095e-06,
+      "loss": 0.6701,
+      "step": 990
+    },
+    {
+      "epoch": 0.7002801120448179,
+      "grad_norm": 2.124044857812194,
+      "learning_rate": 8.218098017317715e-06,
+      "loss": 0.6211,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7072829131652661,
+      "grad_norm": 2.3812764227997265,
+      "learning_rate": 8.171080460577337e-06,
+      "loss": 0.6853,
+      "step": 1010
+    },
+    {
+      "epoch": 0.7142857142857143,
+      "grad_norm": 2.125710699424273,
+      "learning_rate": 8.123589059873445e-06,
+      "loss": 0.6516,
+      "step": 1020
+    },
+    {
+      "epoch": 0.7212885154061625,
+      "grad_norm": 2.0639925369405163,
+      "learning_rate": 8.075630911687451e-06,
+      "loss": 0.5853,
+      "step": 1030
+    },
+    {
+      "epoch": 0.7282913165266106,
+      "grad_norm": 2.4862110603900676,
+      "learning_rate": 8.027213182245289e-06,
+      "loss": 0.6107,
+      "step": 1040
+    },
+    {
+      "epoch": 0.7352941176470589,
+      "grad_norm": 1.601475291312852,
+      "learning_rate": 7.978343106446575e-06,
+      "loss": 0.7112,
+      "step": 1050
+    },
+    {
+      "epoch": 0.742296918767507,
+      "grad_norm": 2.871165956851756,
+      "learning_rate": 7.929027986783538e-06,
+      "loss": 0.6573,
+      "step": 1060
+    },
+    {
+      "epoch": 0.7492997198879552,
+      "grad_norm": 2.502491985241327,
+      "learning_rate": 7.879275192249822e-06,
+      "loss": 0.6383,
+      "step": 1070
+    },
+    {
+      "epoch": 0.7563025210084033,
+      "grad_norm": 2.3534352305326656,
+      "learning_rate": 7.829092157239369e-06,
+      "loss": 0.6707,
+      "step": 1080
+    },
+    {
+      "epoch": 0.7633053221288515,
+      "grad_norm": 2.346735216001232,
+      "learning_rate": 7.778486380435512e-06,
+      "loss": 0.5754,
+      "step": 1090
+    },
+    {
+      "epoch": 0.7703081232492998,
+      "grad_norm": 1.7069335554420118,
+      "learning_rate": 7.727465423690487e-06,
+      "loss": 0.6674,
+      "step": 1100
+    },
+    {
+      "epoch": 0.7773109243697479,
+      "grad_norm": 2.857304704073804,
+      "learning_rate": 7.676036910895475e-06,
+      "loss": 0.6436,
+      "step": 1110
+    },
+    {
+      "epoch": 0.7843137254901961,
+      "grad_norm": 2.704621305833269,
+      "learning_rate": 7.624208526841399e-06,
+      "loss": 0.5633,
+      "step": 1120
+    },
+    {
+      "epoch": 0.7913165266106442,
+      "grad_norm": 2.3446023311507243,
+      "learning_rate": 7.571988016070611e-06,
+      "loss": 0.6534,
+      "step": 1130
+    },
+    {
+      "epoch": 0.7983193277310925,
+      "grad_norm": 2.655699735724774,
+      "learning_rate": 7.519383181719644e-06,
+      "loss": 0.6152,
+      "step": 1140
+    },
+    {
+      "epoch": 0.8053221288515406,
+      "grad_norm": 2.5608889463976983,
+      "learning_rate": 7.466401884353227e-06,
+      "loss": 0.6775,
+      "step": 1150
+    },
+    {
+      "epoch": 0.8123249299719888,
+      "grad_norm": 2.3142838348541965,
+      "learning_rate": 7.413052040789692e-06,
+      "loss": 0.6514,
+      "step": 1160
+    },
+    {
+      "epoch": 0.819327731092437,
+      "grad_norm": 2.134983409939985,
+      "learning_rate": 7.359341622918006e-06,
+      "loss": 0.6194,
+      "step": 1170
+    },
+    {
+      "epoch": 0.8263305322128851,
+      "grad_norm": 2.4655749339506086,
+      "learning_rate": 7.305278656506547e-06,
+      "loss": 0.6381,
+      "step": 1180
+    },
+    {
+      "epoch": 0.8333333333333334,
+      "grad_norm": 1.90146176033135,
+      "learning_rate": 7.2508712200038435e-06,
+      "loss": 0.6059,
+      "step": 1190
+    },
+    {
+      "epoch": 0.8403361344537815,
+      "grad_norm": 2.7581098466146834,
+      "learning_rate": 7.19612744333144e-06,
+      "loss": 0.7107,
+      "step": 1200
+    },
+    {
+      "epoch": 0.8473389355742297,
+      "grad_norm": 2.013181558142786,
+      "learning_rate": 7.141055506669072e-06,
+      "loss": 0.6062,
+      "step": 1210
+    },
+    {
+      "epoch": 0.8543417366946778,
+      "grad_norm": 2.153881730897074,
+      "learning_rate": 7.0856636392323205e-06,
+      "loss": 0.5956,
+      "step": 1220
+    },
+    {
+      "epoch": 0.8613445378151261,
+      "grad_norm": 2.527280124270099,
+      "learning_rate": 7.0299601180429615e-06,
+      "loss": 0.5936,
+      "step": 1230
+    },
+    {
+      "epoch": 0.8683473389355743,
+      "grad_norm": 1.858330034501666,
+      "learning_rate": 6.973953266692143e-06,
+      "loss": 0.6314,
+      "step": 1240
+    },
+    {
+      "epoch": 0.8753501400560224,
+      "grad_norm": 2.3746989063300186,
+      "learning_rate": 6.91765145409663e-06,
+      "loss": 0.6908,
+      "step": 1250
+    },
+    {
+      "epoch": 0.8823529411764706,
+      "grad_norm": 2.2513458203129963,
+      "learning_rate": 6.861063093248264e-06,
+      "loss": 0.5709,
+      "step": 1260
+    },
+    {
+      "epoch": 0.8893557422969187,
+      "grad_norm": 2.5609479562736106,
+      "learning_rate": 6.804196639956828e-06,
+      "loss": 0.627,
+      "step": 1270
+    },
+    {
+      "epoch": 0.896358543417367,
+      "grad_norm": 2.4301069707639233,
+      "learning_rate": 6.747060591586533e-06,
+      "loss": 0.5996,
+      "step": 1280
+    },
+    {
+      "epoch": 0.9033613445378151,
+      "grad_norm": 2.610417413605633,
+      "learning_rate": 6.689663485786287e-06,
+      "loss": 0.6943,
+      "step": 1290
+    },
+    {
+      "epoch": 0.9103641456582633,
+      "grad_norm": 2.240908654296222,
+      "learning_rate": 6.632013899213934e-06,
+      "loss": 0.6311,
+      "step": 1300
+    },
+    {
+      "epoch": 0.9173669467787114,
+      "grad_norm": 2.3215803208239394,
+      "learning_rate": 6.574120446254672e-06,
+      "loss": 0.6081,
+      "step": 1310
+    },
+    {
+      "epoch": 0.9243697478991597,
+      "grad_norm": 1.7958005235142336,
+      "learning_rate": 6.5159917777338466e-06,
+      "loss": 0.6685,
+      "step": 1320
+    },
+    {
+      "epoch": 0.9313725490196079,
+      "grad_norm": 2.042136754036796,
+      "learning_rate": 6.457636579624278e-06,
+      "loss": 0.5457,
+      "step": 1330
+    },
+    {
+      "epoch": 0.938375350140056,
+      "grad_norm": 2.0200483342599367,
+      "learning_rate": 6.3990635717483404e-06,
+      "loss": 0.6981,
+      "step": 1340
+    },
+    {
+      "epoch": 0.9453781512605042,
+      "grad_norm": 2.3485088870526827,
+      "learning_rate": 6.340281506475003e-06,
+      "loss": 0.6165,
+      "step": 1350
+    },
+    {
+      "epoch": 0.9523809523809523,
+      "grad_norm": 2.0775112823384383,
+      "learning_rate": 6.281299167411975e-06,
+      "loss": 0.5656,
+      "step": 1360
+    },
+    {
+      "epoch": 0.9593837535014006,
+      "grad_norm": 2.115129503123767,
+      "learning_rate": 6.222125368093213e-06,
+      "loss": 0.6473,
+      "step": 1370
+    },
+    {
+      "epoch": 0.9663865546218487,
+      "grad_norm": 2.203147650702632,
+      "learning_rate": 6.162768950661945e-06,
+      "loss": 0.6279,
+      "step": 1380
+    },
+    {
+      "epoch": 0.9733893557422969,
+      "grad_norm": 2.617800940269768,
+      "learning_rate": 6.103238784549404e-06,
+      "loss": 0.5426,
+      "step": 1390
+    },
+    {
+      "epoch": 0.9803921568627451,
+      "grad_norm": 2.1058865000385754,
+      "learning_rate": 6.043543765149514e-06,
+      "loss": 0.6551,
+      "step": 1400
+    },
+    {
+      "epoch": 0.9873949579831933,
+      "grad_norm": 2.262404955697173,
+      "learning_rate": 5.983692812489679e-06,
+      "loss": 0.6524,
+      "step": 1410
+    },
+    {
+      "epoch": 0.9943977591036415,
+      "grad_norm": 2.1193412453943803,
+      "learning_rate": 5.923694869897879e-06,
+      "loss": 0.5787,
+      "step": 1420
+    },
+    {
+      "epoch": 1.0014005602240896,
+      "grad_norm": 2.0659116339074166,
+      "learning_rate": 5.863558902666318e-06,
+      "loss": 0.6191,
+      "step": 1430
+    },
+    {
+      "epoch": 1.0084033613445378,
+      "grad_norm": 2.13662088721353,
+      "learning_rate": 5.803293896711756e-06,
+      "loss": 0.4869,
+      "step": 1440
+    },
+    {
+      "epoch": 1.015406162464986,
+      "grad_norm": 2.132520413876192,
+      "learning_rate": 5.74290885723278e-06,
+      "loss": 0.4959,
+      "step": 1450
+    },
+    {
+      "epoch": 1.022408963585434,
+      "grad_norm": 1.862348432644305,
+      "learning_rate": 5.682412807364187e-06,
+      "loss": 0.4611,
+      "step": 1460
+    },
+    {
+      "epoch": 1.0294117647058822,
+      "grad_norm": 2.1396458126762825,
+      "learning_rate": 5.621814786828683e-06,
+      "loss": 0.4828,
+      "step": 1470
+    },
+    {
+      "epoch": 1.0364145658263306,
+      "grad_norm": 2.23299330212776,
+      "learning_rate": 5.5611238505861094e-06,
+      "loss": 0.4362,
+      "step": 1480
+    },
+    {
+      "epoch": 1.0434173669467788,
+      "grad_norm": 2.336766894179504,
+      "learning_rate": 5.5003490674804e-06,
+      "loss": 0.4811,
+      "step": 1490
+    },
+    {
+      "epoch": 1.050420168067227,
+      "grad_norm": 2.2053948018429126,
+      "learning_rate": 5.439499518884433e-06,
+      "loss": 0.4525,
+      "step": 1500
+    },
+    {
+      "epoch": 1.057422969187675,
+      "grad_norm": 2.0455239012711806,
+      "learning_rate": 5.378584297343053e-06,
+      "loss": 0.4573,
+      "step": 1510
+    },
+    {
+      "epoch": 1.0644257703081232,
+      "grad_norm": 2.799894111126595,
+      "learning_rate": 5.3176125052143905e-06,
+      "loss": 0.4475,
+      "step": 1520
+    },
+    {
+      "epoch": 1.0714285714285714,
+      "grad_norm": 2.3537914083107743,
+      "learning_rate": 5.256593253309728e-06,
+      "loss": 0.4679,
+      "step": 1530
+    },
+    {
+      "epoch": 1.0784313725490196,
+      "grad_norm": 2.3121216793787283,
+      "learning_rate": 5.195535659532111e-06,
+      "loss": 0.4982,
+      "step": 1540
+    },
+    {
+      "epoch": 1.0854341736694677,
+      "grad_norm": 1.8801771013194575,
+      "learning_rate": 5.134448847513873e-06,
+      "loss": 0.4881,
+      "step": 1550
+    },
+    {
+      "epoch": 1.092436974789916,
+      "grad_norm": 2.117146997784492,
+      "learning_rate": 5.073341945253336e-06,
+      "loss": 0.4518,
+      "step": 1560
+    },
+    {
+      "epoch": 1.0994397759103642,
+      "grad_norm": 2.4512284660613592,
+      "learning_rate": 5.012224083750845e-06,
+      "loss": 0.4618,
+      "step": 1570
+    },
+    {
+      "epoch": 1.1064425770308124,
+      "grad_norm": 2.3001015988026747,
+      "learning_rate": 4.951104395644342e-06,
+      "loss": 0.4647,
+      "step": 1580
+    },
+    {
+      "epoch": 1.1134453781512605,
+      "grad_norm": 2.733492067730666,
+      "learning_rate": 4.889992013844711e-06,
+      "loss": 0.497,
+      "step": 1590
+    },
+    {
+      "epoch": 1.1204481792717087,
+      "grad_norm": 2.624090029655425,
+      "learning_rate": 4.828896070171084e-06,
+      "loss": 0.5142,
+      "step": 1600
+    },
+    {
+      "epoch": 1.1274509803921569,
+      "grad_norm": 2.3184684372082405,
+      "learning_rate": 4.767825693986295e-06,
+      "loss": 0.449,
+      "step": 1610
+    },
+    {
+      "epoch": 1.134453781512605,
+      "grad_norm": 2.5173043040744836,
+      "learning_rate": 4.706790010832714e-06,
+      "loss": 0.5264,
+      "step": 1620
+    },
+    {
+      "epoch": 1.1414565826330532,
+      "grad_norm": 2.088682951418163,
+      "learning_rate": 4.6457981410686524e-06,
+      "loss": 0.4551,
+      "step": 1630
+    },
+    {
+      "epoch": 1.1484593837535013,
+      "grad_norm": 1.9999194791787656,
+      "learning_rate": 4.584859198505526e-06,
+      "loss": 0.5141,
+      "step": 1640
+    },
+    {
+      "epoch": 1.1554621848739495,
+      "grad_norm": 2.0283863135066293,
+      "learning_rate": 4.5239822890460174e-06,
+      "loss": 0.4254,
+      "step": 1650
+    },
+    {
+      "epoch": 1.1624649859943978,
+      "grad_norm": 2.4761929786113246,
+      "learning_rate": 4.463176509323413e-06,
+      "loss": 0.4986,
+      "step": 1660
+    },
+    {
+      "epoch": 1.169467787114846,
+      "grad_norm": 1.6823258146384572,
+      "learning_rate": 4.402450945342317e-06,
+      "loss": 0.3929,
+      "step": 1670
+    },
+    {
+      "epoch": 1.1764705882352942,
+      "grad_norm": 2.3712484150543065,
+      "learning_rate": 4.341814671120965e-06,
+      "loss": 0.5022,
+      "step": 1680
+    },
+    {
+      "epoch": 1.1834733893557423,
+      "grad_norm": 2.1223027287039136,
+      "learning_rate": 4.2812767473353205e-06,
+      "loss": 0.5201,
+      "step": 1690
+    },
+    {
+      "epoch": 1.1904761904761905,
+      "grad_norm": 2.280910320108766,
+      "learning_rate": 4.220846219965168e-06,
+      "loss": 0.4627,
+      "step": 1700
+    },
+    {
+      "epoch": 1.1974789915966386,
+      "grad_norm": 1.850154690325662,
+      "learning_rate": 4.160532118942411e-06,
+      "loss": 0.4312,
+      "step": 1710
+    },
+    {
+      "epoch": 1.2044817927170868,
+      "grad_norm": 2.339402359759638,
+      "learning_rate": 4.100343456801747e-06,
+      "loss": 0.4317,
+      "step": 1720
+    },
+    {
+      "epoch": 1.211484593837535,
+      "grad_norm": 2.2500882330234604,
+      "learning_rate": 4.040289227333961e-06,
+      "loss": 0.4521,
+      "step": 1730
+    },
+    {
+      "epoch": 1.2184873949579833,
+      "grad_norm": 2.8132677454268857,
+      "learning_rate": 3.980378404242014e-06,
+      "loss": 0.4432,
+      "step": 1740
+    },
+    {
+      "epoch": 1.2254901960784315,
+      "grad_norm": 2.67999303980593,
+      "learning_rate": 3.920619939800131e-06,
+      "loss": 0.4613,
+      "step": 1750
+    },
+    {
+      "epoch": 1.2324929971988796,
+      "grad_norm": 1.8885239257708604,
+      "learning_rate": 3.861022763516091e-06,
+      "loss": 0.4774,
+      "step": 1760
+    },
+    {
+      "epoch": 1.2394957983193278,
+      "grad_norm": 2.815116777237025,
+      "learning_rate": 3.8015957807969247e-06,
+      "loss": 0.4879,
+      "step": 1770
+    },
+    {
+      "epoch": 1.246498599439776,
+      "grad_norm": 2.174587271539217,
+      "learning_rate": 3.7423478716182026e-06,
+      "loss": 0.456,
+      "step": 1780
+    },
+    {
+      "epoch": 1.253501400560224,
+      "grad_norm": 2.7567028224464183,
+      "learning_rate": 3.68328788919714e-06,
+      "loss": 0.4644,
+      "step": 1790
+    },
+    {
+      "epoch": 1.2605042016806722,
+      "grad_norm": 2.1568363923043266,
+      "learning_rate": 3.624424658669682e-06,
+      "loss": 0.5225,
+      "step": 1800
+    },
+    {
+      "epoch": 1.2675070028011204,
+      "grad_norm": 2.721189011538071,
+      "learning_rate": 3.5657669757717927e-06,
+      "loss": 0.3921,
+      "step": 1810
+    },
+    {
+      "epoch": 1.2745098039215685,
+      "grad_norm": 2.120385676401417,
+      "learning_rate": 3.5073236055251425e-06,
+      "loss": 0.4793,
+      "step": 1820
+    },
+    {
+      "epoch": 1.2815126050420167,
+      "grad_norm": 2.302768261718875,
+      "learning_rate": 3.4491032809273784e-06,
+      "loss": 0.4526,
+      "step": 1830
+    },
+    {
+      "epoch": 1.2885154061624648,
+      "grad_norm": 2.334008008597067,
+      "learning_rate": 3.3911147016471784e-06,
+      "loss": 0.4855,
+      "step": 1840
+    },
+    {
+      "epoch": 1.2955182072829132,
+      "grad_norm": 2.6089068847882744,
+      "learning_rate": 3.3333665327242965e-06,
+      "loss": 0.4522,
+      "step": 1850
+    },
+    {
+      "epoch": 1.3025210084033614,
+      "grad_norm": 2.8426511093929556,
+      "learning_rate": 3.275867403274763e-06,
+      "loss": 0.4965,
+      "step": 1860
+    },
+    {
+      "epoch": 1.3095238095238095,
+      "grad_norm": 2.515991574117609,
+      "learning_rate": 3.2186259052014797e-06,
+      "loss": 0.4635,
+      "step": 1870
+    },
+    {
+      "epoch": 1.3165266106442577,
+      "grad_norm": 2.410804360632545,
+      "learning_rate": 3.1616505919103446e-06,
+      "loss": 0.44,
+      "step": 1880
+    },
+    {
+      "epoch": 1.3235294117647058,
+      "grad_norm": 2.1144368698980234,
+      "learning_rate": 3.1049499770321572e-06,
+      "loss": 0.511,
+      "step": 1890
+    },
+    {
+      "epoch": 1.330532212885154,
+      "grad_norm": 2.3764720452535113,
+      "learning_rate": 3.0485325331504477e-06,
+      "loss": 0.4298,
+      "step": 1900
+    },
+    {
+      "epoch": 1.3375350140056024,
+      "grad_norm": 2.008815745845685,
+      "learning_rate": 2.992406690535453e-06,
+      "loss": 0.4353,
+      "step": 1910
+    },
+    {
+      "epoch": 1.3445378151260505,
+      "grad_norm": 2.5495322694247404,
+      "learning_rate": 2.936580835884404e-06,
+      "loss": 0.5351,
+      "step": 1920
+    },
+    {
+      "epoch": 1.3515406162464987,
+      "grad_norm": 2.6170532242550424,
+      "learning_rate": 2.8810633110683312e-06,
+      "loss": 0.4736,
+      "step": 1930
+    },
+    {
+      "epoch": 1.3585434173669468,
+      "grad_norm": 2.304939061322721,
+      "learning_rate": 2.8258624118855704e-06,
+      "loss": 0.497,
+      "step": 1940
+    },
+    {
+      "epoch": 1.365546218487395,
+      "grad_norm": 2.52028463522758,
+      "learning_rate": 2.770986386822145e-06,
+      "loss": 0.4561,
+      "step": 1950
+    },
+    {
+      "epoch": 1.3725490196078431,
+      "grad_norm": 2.1554513929621653,
+      "learning_rate": 2.7164434358192236e-06,
+      "loss": 0.4771,
+      "step": 1960
+    },
+    {
+      "epoch": 1.3795518207282913,
+      "grad_norm": 2.2864103571812247,
+      "learning_rate": 2.6622417090478304e-06,
+      "loss": 0.4244,
+      "step": 1970
+    },
+    {
+      "epoch": 1.3865546218487395,
+      "grad_norm": 2.3291887139186604,
+      "learning_rate": 2.6083893056909937e-06,
+      "loss": 0.4818,
+      "step": 1980
+    },
+    {
+      "epoch": 1.3935574229691876,
+      "grad_norm": 1.6186217281536261,
+      "learning_rate": 2.5548942727335046e-06,
+      "loss": 0.4473,
+      "step": 1990
+    },
+    {
+      "epoch": 1.4005602240896358,
+      "grad_norm": 1.9490552509158812,
+      "learning_rate": 2.5017646037594967e-06,
+      "loss": 0.4653,
+      "step": 2000
+    },
+    {
+      "epoch": 1.407563025210084,
+      "grad_norm": 2.4440857430217124,
+      "learning_rate": 2.4490082377579805e-06,
+      "loss": 0.4945,
+      "step": 2010
+    },
+    {
+      "epoch": 1.4145658263305323,
+      "grad_norm": 2.9756203221062476,
+      "learning_rate": 2.3966330579365436e-06,
+      "loss": 0.4371,
+      "step": 2020
+    },
+    {
+      "epoch": 1.4215686274509804,
+      "grad_norm": 2.202500602837109,
+      "learning_rate": 2.3446468905434093e-06,
+      "loss": 0.4495,
+      "step": 2030
+    },
+    {
+      "epoch": 1.4285714285714286,
+      "grad_norm": 1.9506623499545588,
+      "learning_rate": 2.2930575036979634e-06,
+      "loss": 0.4552,
+      "step": 2040
+    },
+    {
+      "epoch": 1.4355742296918768,
+      "grad_norm": 2.6310040966711776,
+      "learning_rate": 2.241872606229998e-06,
+      "loss": 0.4378,
+      "step": 2050
+    },
+    {
+      "epoch": 1.442577030812325,
+      "grad_norm": 2.4781025094081706,
+      "learning_rate": 2.191099846527813e-06,
+      "loss": 0.5055,
+      "step": 2060
+    },
+    {
+      "epoch": 1.449579831932773,
+      "grad_norm": 2.3262915560231217,
+      "learning_rate": 2.14074681139534e-06,
+      "loss": 0.4727,
+      "step": 2070
+    },
+    {
+      "epoch": 1.4565826330532212,
+      "grad_norm": 2.001470623484971,
+      "learning_rate": 2.090821024918462e-06,
+      "loss": 0.4826,
+      "step": 2080
+    },
+    {
+      "epoch": 1.4635854341736696,
+      "grad_norm": 2.8728529295622964,
+      "learning_rate": 2.0413299473407285e-06,
+      "loss": 0.4162,
+      "step": 2090
+    },
+    {
+      "epoch": 1.4705882352941178,
+      "grad_norm": 2.54674433384974,
+      "learning_rate": 1.9922809739485883e-06,
+      "loss": 0.4516,
+      "step": 2100
+    },
+    {
+      "epoch": 1.477591036414566,
+      "grad_norm": 1.7010786229876187,
+      "learning_rate": 1.943681433966338e-06,
+      "loss": 0.4727,
+      "step": 2110
+    },
+    {
+      "epoch": 1.484593837535014,
+      "grad_norm": 2.7445575800859747,
+      "learning_rate": 1.8955385894609414e-06,
+      "loss": 0.4835,
+      "step": 2120
+    },
+    {
+      "epoch": 1.4915966386554622,
+      "grad_norm": 2.5636924295564825,
+      "learning_rate": 1.8478596342568827e-06,
+      "loss": 0.5,
+      "step": 2130
+    },
+    {
+      "epoch": 1.4985994397759104,
+      "grad_norm": 2.6320112816326806,
+      "learning_rate": 1.8006516928612121e-06,
+      "loss": 0.4783,
+      "step": 2140
+    },
+    {
+      "epoch": 1.5056022408963585,
+      "grad_norm": 2.3130885237765653,
+      "learning_rate": 1.753921819398966e-06,
+      "loss": 0.4732,
+      "step": 2150
+    },
+    {
+      "epoch": 1.5126050420168067,
+      "grad_norm": 2.4566766430867855,
+      "learning_rate": 1.7076769965590834e-06,
+      "loss": 0.4846,
+      "step": 2160
+    },
+    {
+      "epoch": 1.5196078431372548,
+      "grad_norm": 2.1691439788909705,
+      "learning_rate": 1.6619241345510057e-06,
+      "loss": 0.5196,
+      "step": 2170
+    },
+    {
+      "epoch": 1.526610644257703,
+      "grad_norm": 1.9388105151944164,
+      "learning_rate": 1.6166700700721121e-06,
+      "loss": 0.3903,
+      "step": 2180
+    },
+    {
+      "epoch": 1.5336134453781511,
+      "grad_norm": 2.353752022853995,
+      "learning_rate": 1.571921565286139e-06,
+      "loss": 0.4923,
+      "step": 2190
+    },
+    {
+      "epoch": 1.5406162464985993,
+      "grad_norm": 2.572914747835839,
+      "learning_rate": 1.527685306812715e-06,
+      "loss": 0.4166,
+      "step": 2200
+    },
+    {
+      "epoch": 1.5476190476190477,
+      "grad_norm": 2.1724801712005792,
+      "learning_rate": 1.4839679047282206e-06,
+      "loss": 0.4504,
+      "step": 2210
+    },
+    {
+      "epoch": 1.5546218487394958,
+      "grad_norm": 2.594065623296099,
+      "learning_rate": 1.4407758915780578e-06,
+      "loss": 0.477,
+      "step": 2220
+    },
+    {
+      "epoch": 1.561624649859944,
+      "grad_norm": 2.208098426234172,
+      "learning_rate": 1.3981157214005098e-06,
+      "loss": 0.4586,
+      "step": 2230
+    },
+    {
+      "epoch": 1.5686274509803921,
+      "grad_norm": 2.139581629149966,
+      "learning_rate": 1.3559937687623458e-06,
+      "loss": 0.4276,
+      "step": 2240
+    },
+    {
+      "epoch": 1.5756302521008403,
+      "grad_norm": 2.538499089739133,
+      "learning_rate": 1.3144163278062848e-06,
+      "loss": 0.4206,
+      "step": 2250
+    },
+    {
+      "epoch": 1.5826330532212887,
+      "grad_norm": 2.041136800394131,
+      "learning_rate": 1.273389611310481e-06,
+      "loss": 0.4034,
+      "step": 2260
+    },
+    {
+      "epoch": 1.5896358543417368,
+      "grad_norm": 2.313241143682124,
+      "learning_rate": 1.2329197497601742e-06,
+      "loss": 0.4801,
+      "step": 2270
+    },
+    {
+      "epoch": 1.596638655462185,
+      "grad_norm": 2.309500171812923,
+      "learning_rate": 1.1930127904316286e-06,
+      "loss": 0.5045,
+      "step": 2280
+    },
+    {
+      "epoch": 1.6036414565826331,
+      "grad_norm": 2.485298750781323,
+      "learning_rate": 1.1536746964885065e-06,
+      "loss": 0.4301,
+      "step": 2290
+    },
+    {
+      "epoch": 1.6106442577030813,
+      "grad_norm": 2.591481445677265,
+      "learning_rate": 1.114911346090819e-06,
+      "loss": 0.4554,
+      "step": 2300
+    },
+    {
+      "epoch": 1.6176470588235294,
+      "grad_norm": 2.833792845787996,
+      "learning_rate": 1.0767285315165675e-06,
+      "loss": 0.4132,
+      "step": 2310
+    },
+    {
+      "epoch": 1.6246498599439776,
+      "grad_norm": 2.388415928161688,
+      "learning_rate": 1.0391319582962244e-06,
+      "loss": 0.4707,
+      "step": 2320
+    },
+    {
+      "epoch": 1.6316526610644257,
+      "grad_norm": 2.3441821249748798,
+      "learning_rate": 1.002127244360171e-06,
+      "loss": 0.422,
+      "step": 2330
+    },
+    {
+      "epoch": 1.638655462184874,
+      "grad_norm": 1.9731306455288935,
+      "learning_rate": 9.65719919199235e-07,
+      "loss": 0.3684,
+      "step": 2340
+    },
+    {
+      "epoch": 1.645658263305322,
+      "grad_norm": 3.7976656804390303,
+      "learning_rate": 9.299154230384383e-07,
+      "loss": 0.4887,
+      "step": 2350
+    },
+    {
+      "epoch": 1.6526610644257702,
+      "grad_norm": 2.29269452111851,
+      "learning_rate": 8.947191060240701e-07,
+      "loss": 0.4054,
+      "step": 2360
+    },
+    {
+      "epoch": 1.6596638655462184,
+      "grad_norm": 2.765208721785312,
+      "learning_rate": 8.601362274242465e-07,
+      "loss": 0.5191,
+      "step": 2370
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 2.637952088090675,
+      "learning_rate": 8.261719548430292e-07,
+      "loss": 0.479,
+      "step": 2380
+    },
+    {
+      "epoch": 1.673669467787115,
+      "grad_norm": 2.7259524316936496,
+      "learning_rate": 7.928313634482454e-07,
+      "loss": 0.4162,
+      "step": 2390
+    },
+    {
+      "epoch": 1.680672268907563,
+      "grad_norm": 2.783618319304046,
+      "learning_rate": 7.601194352131285e-07,
+      "loss": 0.4662,
+      "step": 2400
+    },
+    {
+      "epoch": 1.6876750700280112,
+      "grad_norm": 2.5896397818912957,
+      "learning_rate": 7.280410581718761e-07,
+      "loss": 0.466,
+      "step": 2410
+    },
+    {
+      "epoch": 1.6946778711484594,
+      "grad_norm": 2.313925409486687,
+      "learning_rate": 6.966010256892408e-07,
+      "loss": 0.421,
+      "step": 2420
+    },
+    {
+      "epoch": 1.7016806722689075,
+      "grad_norm": 2.2637352342234673,
+      "learning_rate": 6.658040357442874e-07,
+      "loss": 0.4384,
+      "step": 2430
+    },
+    {
+      "epoch": 1.708683473389356,
+      "grad_norm": 2.2403407432076836,
+      "learning_rate": 6.356546902283772e-07,
+      "loss": 0.4423,
+      "step": 2440
+    },
+    {
+      "epoch": 1.715686274509804,
+      "grad_norm": 2.7755062361806333,
+      "learning_rate": 6.061574942575266e-07,
+      "loss": 0.4507,
+      "step": 2450
+    },
+    {
+      "epoch": 1.7226890756302522,
+      "grad_norm": 2.3259900192767713,
+      "learning_rate": 5.773168554992248e-07,
+      "loss": 0.426,
+      "step": 2460
+    },
+    {
+      "epoch": 1.7296918767507004,
+      "grad_norm": 2.6341383732328456,
+      "learning_rate": 5.491370835138116e-07,
+      "loss": 0.4377,
+      "step": 2470
+    },
+    {
+      "epoch": 1.7366946778711485,
+      "grad_norm": 2.3924610495134933,
+      "learning_rate": 5.216223891105054e-07,
+      "loss": 0.4585,
+      "step": 2480
+    },
+    {
+      "epoch": 1.7436974789915967,
+      "grad_norm": 2.1758349972571507,
+      "learning_rate": 4.947768837182093e-07,
+      "loss": 0.4068,
+      "step": 2490
+    },
+    {
+      "epoch": 1.7507002801120448,
+      "grad_norm": 2.0421217828565448,
+      "learning_rate": 4.686045787711435e-07,
+      "loss": 0.4066,
+      "step": 2500
+    },
+    {
+      "epoch": 1.757703081232493,
+      "grad_norm": 2.115206308615413,
+      "learning_rate": 4.4310938510943533e-07,
+      "loss": 0.4522,
+      "step": 2510
+    },
+    {
+      "epoch": 1.7647058823529411,
+      "grad_norm": 2.066405782929325,
+      "learning_rate": 4.182951123947332e-07,
+      "loss": 0.3312,
+      "step": 2520
+    },
+    {
+      "epoch": 1.7717086834733893,
+      "grad_norm": 2.342484690456666,
+      "learning_rate": 3.9416546854094294e-07,
+      "loss": 0.4224,
+      "step": 2530
+    },
+    {
+      "epoch": 1.7787114845938374,
+      "grad_norm": 2.512724647352599,
+      "learning_rate": 3.7072405916016353e-07,
+      "loss": 0.445,
+      "step": 2540
+    },
+    {
+      "epoch": 1.7857142857142856,
+      "grad_norm": 2.470340765232829,
+      "learning_rate": 3.479743870239188e-07,
+      "loss": 0.4637,
+      "step": 2550
+    },
+    {
+      "epoch": 1.7927170868347337,
+      "grad_norm": 2.130736248713378,
+      "learning_rate": 3.2591985153974383e-07,
+      "loss": 0.387,
+      "step": 2560
+    },
+    {
+      "epoch": 1.7997198879551821,
+      "grad_norm": 2.1431789471135723,
+      "learning_rate": 3.0456374824322674e-07,
+      "loss": 0.4469,
+      "step": 2570
+    },
+    {
+      "epoch": 1.8067226890756303,
+      "grad_norm": 2.099234007419937,
+      "learning_rate": 2.839092683055622e-07,
+      "loss": 0.4039,
+      "step": 2580
+    },
+    {
+      "epoch": 1.8137254901960784,
+      "grad_norm": 2.459188631196229,
+      "learning_rate": 2.639594980567162e-07,
+      "loss": 0.3812,
+      "step": 2590
+    },
+    {
+      "epoch": 1.8207282913165266,
+      "grad_norm": 2.555886554006267,
+      "learning_rate": 2.447174185242324e-07,
+      "loss": 0.4212,
+      "step": 2600
+    },
+    {
+      "epoch": 1.8277310924369747,
+      "grad_norm": 2.2836207612772843,
+      "learning_rate": 2.2618590498779847e-07,
+      "loss": 0.4137,
+      "step": 2610
+    },
+    {
+      "epoch": 1.8347338935574231,
+      "grad_norm": 2.1026781532549936,
+      "learning_rate": 2.0836772654959647e-07,
+      "loss": 0.4213,
+      "step": 2620
+    },
+    {
+      "epoch": 1.8417366946778713,
+      "grad_norm": 2.4328664873281483,
+      "learning_rate": 1.9126554572052813e-07,
+      "loss": 0.4335,
+      "step": 2630
+    },
+    {
+      "epoch": 1.8487394957983194,
+      "grad_norm": 2.3974367367984857,
+      "learning_rate": 1.74881918022361e-07,
+      "loss": 0.4056,
+      "step": 2640
+    },
+    {
+      "epoch": 1.8557422969187676,
+      "grad_norm": 2.299811399188436,
+      "learning_rate": 1.5921929160587045e-07,
+      "loss": 0.4602,
+      "step": 2650
+    },
+    {
+      "epoch": 1.8627450980392157,
+      "grad_norm": 1.9124769037343121,
+      "learning_rate": 1.44280006885017e-07,
+      "loss": 0.4231,
+      "step": 2660
+    },
+    {
+      "epoch": 1.8697478991596639,
+      "grad_norm": 2.813578456729493,
+      "learning_rate": 1.3006629618722733e-07,
+      "loss": 0.4286,
+      "step": 2670
+    },
+    {
+      "epoch": 1.876750700280112,
+      "grad_norm": 2.4806383435943613,
+      "learning_rate": 1.1658028341982486e-07,
+      "loss": 0.4368,
+      "step": 2680
+    },
+    {
+      "epoch": 1.8837535014005602,
+      "grad_norm": 2.2681286986704534,
+      "learning_rate": 1.0382398375266111e-07,
+      "loss": 0.4691,
+      "step": 2690
+    },
+    {
+      "epoch": 1.8907563025210083,
+      "grad_norm": 1.9128855914355993,
+      "learning_rate": 9.179930331699615e-08,
+      "loss": 0.4508,
+      "step": 2700
+    },
+    {
+      "epoch": 1.8977591036414565,
+      "grad_norm": 2.1633704801409834,
+      "learning_rate": 8.050803892067139e-08,
+      "loss": 0.4422,
+      "step": 2710
+    },
+    {
+      "epoch": 1.9047619047619047,
+      "grad_norm": 2.8200396806385,
+      "learning_rate": 6.995187777961931e-08,
+      "loss": 0.4254,
+      "step": 2720
+    },
+    {
+      "epoch": 1.9117647058823528,
+      "grad_norm": 2.6445714035080887,
+      "learning_rate": 6.013239726574694e-08,
+      "loss": 0.4074,
+      "step": 2730
+    },
+    {
+      "epoch": 1.918767507002801,
+      "grad_norm": 2.0546751872771627,
+      "learning_rate": 5.105106467123477e-08,
+      "loss": 0.4262,
+      "step": 2740
+    },
+    {
+      "epoch": 1.9257703081232493,
+      "grad_norm": 2.7784776972549916,
+      "learning_rate": 4.2709236989287305e-08,
+      "loss": 0.4725,
+      "step": 2750
+    },
+    {
+      "epoch": 1.9327731092436975,
+      "grad_norm": 2.272829343457401,
+      "learning_rate": 3.510816071135681e-08,
+      "loss": 0.4409,
+      "step": 2760
+    },
+    {
+      "epoch": 1.9397759103641457,
+      "grad_norm": 2.4175361768734276,
+      "learning_rate": 2.8248971640887913e-08,
+      "loss": 0.393,
+      "step": 2770
+    },
+    {
+      "epoch": 1.9467787114845938,
+      "grad_norm": 2.2153647657315347,
+      "learning_rate": 2.213269472359836e-08,
+      "loss": 0.474,
+      "step": 2780
+    },
+    {
+      "epoch": 1.9537815126050422,
+      "grad_norm": 2.146439918476662,
+      "learning_rate": 1.6760243894321513e-08,
+      "loss": 0.4008,
+      "step": 2790
+    },
+    {
+      "epoch": 1.9607843137254903,
+      "grad_norm": 2.0928270476543522,
+      "learning_rate": 1.213242194044395e-08,
+      "loss": 0.4111,
+      "step": 2800
+    },
+    {
+      "epoch": 1.9677871148459385,
+      "grad_norm": 2.079556551258959,
+      "learning_rate": 8.249920381946387e-09,
+      "loss": 0.4999,
+      "step": 2810
+    },
+    {
+      "epoch": 1.9747899159663866,
+      "grad_norm": 2.0113264107881164,
+      "learning_rate": 5.113319368070801e-09,
+      "loss": 0.4397,
+      "step": 2820
+    },
+    {
+      "epoch": 1.9817927170868348,
+      "grad_norm": 2.3335982318192827,
+      "learning_rate": 2.723087590632556e-09,
+      "loss": 0.3984,
+      "step": 2830
+    },
+    {
+      "epoch": 1.988795518207283,
+      "grad_norm": 2.7347014131885015,
+      "learning_rate": 1.079582213983632e-09,
+      "loss": 0.4515,
+      "step": 2840
+    },
+    {
+      "epoch": 1.995798319327731,
+      "grad_norm": 1.8502359446882066,
+      "learning_rate": 1.830488216442161e-10,
+      "loss": 0.4036,
+      "step": 2850
+    },
+    {
+      "epoch": 2.0,
+      "step": 2856,
+      "total_flos": 137777518706688.0,
+      "train_loss": 0.5963595523720696,
+      "train_runtime": 4888.1806,
+      "train_samples_per_second": 9.347,
+      "train_steps_per_second": 0.584
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 2856,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "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": 137777518706688.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bdc5f223219a2d86fa771859dc8e50b6963552678fb8dabcd99841c7bfe0e249
+size 8081