Upload folder using huggingface_hub

.gitattributes

@@ -33,3 +33,7 @@ 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
+Direction/Ageis_Danger/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+Direction/Beaver-Danger/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+Direction/PKURLHF-10K_Safety/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+initial-state/dolly_ckpt_5850/tokenizer.json filter=lfs diff=lfs merge=lfs -text

Direction/Ageis_Danger/README.md

@@ -0,0 +1,208 @@
+---
+base_model: Qwen/Qwen3-8B
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:/workspace/jzc/model/Qwen3-8B
+- llama-factory
+- lora
+- transformers
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.17.1

Direction/Ageis_Danger/adapter_config.json

@@ -0,0 +1,42 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "/workspace/jzc/model/Qwen3-8B",
+  "bias": "none",
+  "corda_config": null,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 16,
+  "lora_bias": false,
+  "lora_dropout": 0.0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "q_proj",
+    "gate_proj",
+    "v_proj",
+    "o_proj",
+    "up_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

Direction/Ageis_Danger/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6264a511500b07de2227578730dd4701a9b3a53cf51032eb0aec99594ca44751
+size 87360584

Direction/Ageis_Danger/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
+}

Direction/Ageis_Danger/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 %}

Direction/Ageis_Danger/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
+  }
+}

Direction/Ageis_Danger/tokenizer.json

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

Direction/Ageis_Danger/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
+}

Direction/Ageis_Danger/trainer_state.json

@@ -0,0 +1,2659 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 10.0,
+  "eval_steps": 500,
+  "global_step": 3750,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02666666666666667,
+      "grad_norm": 0.6864556074142456,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 2.0431,
+      "step": 10
+    },
+    {
+      "epoch": 0.05333333333333334,
+      "grad_norm": 0.6834467649459839,
+      "learning_rate": 2.533333333333333e-07,
+      "loss": 2.0222,
+      "step": 20
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.7390895485877991,
+      "learning_rate": 3.8666666666666674e-07,
+      "loss": 2.0474,
+      "step": 30
+    },
+    {
+      "epoch": 0.10666666666666667,
+      "grad_norm": 0.8008406758308411,
+      "learning_rate": 5.2e-07,
+      "loss": 2.087,
+      "step": 40
+    },
+    {
+      "epoch": 0.13333333333333333,
+      "grad_norm": 0.7216496467590332,
+      "learning_rate": 6.533333333333334e-07,
+      "loss": 2.0767,
+      "step": 50
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7768703699111938,
+      "learning_rate": 7.866666666666667e-07,
+      "loss": 2.0302,
+      "step": 60
+    },
+    {
+      "epoch": 0.18666666666666668,
+      "grad_norm": 0.9261518120765686,
+      "learning_rate": 9.200000000000001e-07,
+      "loss": 2.0068,
+      "step": 70
+    },
+    {
+      "epoch": 0.21333333333333335,
+      "grad_norm": 0.6796293258666992,
+      "learning_rate": 1.0533333333333333e-06,
+      "loss": 2.0384,
+      "step": 80
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.0748271942138672,
+      "learning_rate": 1.1866666666666668e-06,
+      "loss": 2.104,
+      "step": 90
+    },
+    {
+      "epoch": 0.26666666666666666,
+      "grad_norm": 0.9115467667579651,
+      "learning_rate": 1.32e-06,
+      "loss": 2.0462,
+      "step": 100
+    },
+    {
+      "epoch": 0.29333333333333333,
+      "grad_norm": 0.915414035320282,
+      "learning_rate": 1.4533333333333335e-06,
+      "loss": 2.0701,
+      "step": 110
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.0734838247299194,
+      "learning_rate": 1.586666666666667e-06,
+      "loss": 2.066,
+      "step": 120
+    },
+    {
+      "epoch": 0.3466666666666667,
+      "grad_norm": 1.0868602991104126,
+      "learning_rate": 1.72e-06,
+      "loss": 2.0059,
+      "step": 130
+    },
+    {
+      "epoch": 0.37333333333333335,
+      "grad_norm": 1.1686838865280151,
+      "learning_rate": 1.8533333333333333e-06,
+      "loss": 2.0448,
+      "step": 140
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.0780514478683472,
+      "learning_rate": 1.9866666666666666e-06,
+      "loss": 1.9928,
+      "step": 150
+    },
+    {
+      "epoch": 0.4266666666666667,
+      "grad_norm": 1.1489983797073364,
+      "learning_rate": 2.12e-06,
+      "loss": 2.0029,
+      "step": 160
+    },
+    {
+      "epoch": 0.4533333333333333,
+      "grad_norm": 1.2736639976501465,
+      "learning_rate": 2.2533333333333335e-06,
+      "loss": 1.9858,
+      "step": 170
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.3319426774978638,
+      "learning_rate": 2.386666666666667e-06,
+      "loss": 1.9551,
+      "step": 180
+    },
+    {
+      "epoch": 0.5066666666666667,
+      "grad_norm": 1.349578619003296,
+      "learning_rate": 2.52e-06,
+      "loss": 1.872,
+      "step": 190
+    },
+    {
+      "epoch": 0.5333333333333333,
+      "grad_norm": 0.9017946720123291,
+      "learning_rate": 2.6533333333333335e-06,
+      "loss": 1.8316,
+      "step": 200
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.002862811088562,
+      "learning_rate": 2.786666666666667e-06,
+      "loss": 1.8138,
+      "step": 210
+    },
+    {
+      "epoch": 0.5866666666666667,
+      "grad_norm": 0.8344107866287231,
+      "learning_rate": 2.92e-06,
+      "loss": 1.7294,
+      "step": 220
+    },
+    {
+      "epoch": 0.6133333333333333,
+      "grad_norm": 0.9120360016822815,
+      "learning_rate": 3.053333333333334e-06,
+      "loss": 1.7006,
+      "step": 230
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.883114218711853,
+      "learning_rate": 3.186666666666667e-06,
+      "loss": 1.6236,
+      "step": 240
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 0.8164843916893005,
+      "learning_rate": 3.3200000000000004e-06,
+      "loss": 1.6641,
+      "step": 250
+    },
+    {
+      "epoch": 0.6933333333333334,
+      "grad_norm": 0.8954416513442993,
+      "learning_rate": 3.4533333333333334e-06,
+      "loss": 1.6328,
+      "step": 260
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.8396301865577698,
+      "learning_rate": 3.5866666666666673e-06,
+      "loss": 1.6047,
+      "step": 270
+    },
+    {
+      "epoch": 0.7466666666666667,
+      "grad_norm": 0.9113221168518066,
+      "learning_rate": 3.7200000000000004e-06,
+      "loss": 1.6098,
+      "step": 280
+    },
+    {
+      "epoch": 0.7733333333333333,
+      "grad_norm": 0.7408691048622131,
+      "learning_rate": 3.853333333333334e-06,
+      "loss": 1.5339,
+      "step": 290
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.9837501645088196,
+      "learning_rate": 3.986666666666667e-06,
+      "loss": 1.592,
+      "step": 300
+    },
+    {
+      "epoch": 0.8266666666666667,
+      "grad_norm": 0.9414945840835571,
+      "learning_rate": 4.12e-06,
+      "loss": 1.5275,
+      "step": 310
+    },
+    {
+      "epoch": 0.8533333333333334,
+      "grad_norm": 0.6915170550346375,
+      "learning_rate": 4.253333333333334e-06,
+      "loss": 1.4722,
+      "step": 320
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.0188401937484741,
+      "learning_rate": 4.3866666666666665e-06,
+      "loss": 1.4897,
+      "step": 330
+    },
+    {
+      "epoch": 0.9066666666666666,
+      "grad_norm": 0.7740457057952881,
+      "learning_rate": 4.520000000000001e-06,
+      "loss": 1.377,
+      "step": 340
+    },
+    {
+      "epoch": 0.9333333333333333,
+      "grad_norm": 0.6730086207389832,
+      "learning_rate": 4.653333333333333e-06,
+      "loss": 1.392,
+      "step": 350
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.855743944644928,
+      "learning_rate": 4.786666666666667e-06,
+      "loss": 1.3936,
+      "step": 360
+    },
+    {
+      "epoch": 0.9866666666666667,
+      "grad_norm": 0.718235969543457,
+      "learning_rate": 4.92e-06,
+      "loss": 1.3313,
+      "step": 370
+    },
+    {
+      "epoch": 1.0133333333333334,
+      "grad_norm": 0.6589620113372803,
+      "learning_rate": 4.999982670673473e-06,
+      "loss": 1.3231,
+      "step": 380
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.7204256653785706,
+      "learning_rate": 4.999787718509086e-06,
+      "loss": 1.312,
+      "step": 390
+    },
+    {
+      "epoch": 1.0666666666666667,
+      "grad_norm": 0.7311373949050903,
+      "learning_rate": 4.999376169470306e-06,
+      "loss": 1.3119,
+      "step": 400
+    },
+    {
+      "epoch": 1.0933333333333333,
+      "grad_norm": 0.6310704946517944,
+      "learning_rate": 4.998748059216254e-06,
+      "loss": 1.2798,
+      "step": 410
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.7810817360877991,
+      "learning_rate": 4.997903442170241e-06,
+      "loss": 1.2968,
+      "step": 420
+    },
+    {
+      "epoch": 1.1466666666666667,
+      "grad_norm": 0.7063048481941223,
+      "learning_rate": 4.996842391515045e-06,
+      "loss": 1.277,
+      "step": 430
+    },
+    {
+      "epoch": 1.1733333333333333,
+      "grad_norm": 0.6194319725036621,
+      "learning_rate": 4.995564999186572e-06,
+      "loss": 1.3146,
+      "step": 440
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.8911550045013428,
+      "learning_rate": 4.994071375865898e-06,
+      "loss": 1.2271,
+      "step": 450
+    },
+    {
+      "epoch": 1.2266666666666666,
+      "grad_norm": 0.7139999866485596,
+      "learning_rate": 4.992361650969668e-06,
+      "loss": 1.2009,
+      "step": 460
+    },
+    {
+      "epoch": 1.2533333333333334,
+      "grad_norm": 0.7921785712242126,
+      "learning_rate": 4.990435972638888e-06,
+      "loss": 1.241,
+      "step": 470
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.6499159336090088,
+      "learning_rate": 4.988294507726089e-06,
+      "loss": 1.2606,
+      "step": 480
+    },
+    {
+      "epoch": 1.3066666666666666,
+      "grad_norm": 0.8065305352210999,
+      "learning_rate": 4.98593744178087e-06,
+      "loss": 1.2408,
+      "step": 490
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 0.603751003742218,
+      "learning_rate": 4.983364979033819e-06,
+      "loss": 1.1539,
+      "step": 500
+    },
+    {
+      "epoch": 1.3599999999999999,
+      "grad_norm": 0.6962315440177917,
+      "learning_rate": 4.980577342378818e-06,
+      "loss": 1.2015,
+      "step": 510
+    },
+    {
+      "epoch": 1.3866666666666667,
+      "grad_norm": 0.7914028763771057,
+      "learning_rate": 4.977574773353732e-06,
+      "loss": 1.2067,
+      "step": 520
+    },
+    {
+      "epoch": 1.4133333333333333,
+      "grad_norm": 0.7412983775138855,
+      "learning_rate": 4.974357532119478e-06,
+      "loss": 1.1334,
+      "step": 530
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.8885224461555481,
+      "learning_rate": 4.970925897437484e-06,
+      "loss": 1.2074,
+      "step": 540
+    },
+    {
+      "epoch": 1.4666666666666668,
+      "grad_norm": 0.6707415580749512,
+      "learning_rate": 4.967280166645538e-06,
+      "loss": 1.2794,
+      "step": 550
+    },
+    {
+      "epoch": 1.4933333333333334,
+      "grad_norm": 1.029408574104309,
+      "learning_rate": 4.9634206556320186e-06,
+      "loss": 1.1519,
+      "step": 560
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.7895164489746094,
+      "learning_rate": 4.959347698808532e-06,
+      "loss": 1.1693,
+      "step": 570
+    },
+    {
+      "epoch": 1.5466666666666666,
+      "grad_norm": 0.7517342567443848,
+      "learning_rate": 4.95506164908093e-06,
+      "loss": 1.1478,
+      "step": 580
+    },
+    {
+      "epoch": 1.5733333333333333,
+      "grad_norm": 0.7386269569396973,
+      "learning_rate": 4.9505628778187365e-06,
+      "loss": 1.154,
+      "step": 590
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.7774485349655151,
+      "learning_rate": 4.94585177482297e-06,
+      "loss": 1.197,
+      "step": 600
+    },
+    {
+      "epoch": 1.6266666666666667,
+      "grad_norm": 0.6724669337272644,
+      "learning_rate": 4.940928748292363e-06,
+      "loss": 1.1248,
+      "step": 610
+    },
+    {
+      "epoch": 1.6533333333333333,
+      "grad_norm": 0.8252294659614563,
+      "learning_rate": 4.9357942247879995e-06,
+      "loss": 1.1685,
+      "step": 620
+    },
+    {
+      "epoch": 1.6800000000000002,
+      "grad_norm": 0.8395477533340454,
+      "learning_rate": 4.930448649196356e-06,
+      "loss": 1.1423,
+      "step": 630
+    },
+    {
+      "epoch": 1.7066666666666666,
+      "grad_norm": 0.797327995300293,
+      "learning_rate": 4.924892484690744e-06,
+      "loss": 1.195,
+      "step": 640
+    },
+    {
+      "epoch": 1.7333333333333334,
+      "grad_norm": 0.8169471621513367,
+      "learning_rate": 4.91912621269119e-06,
+      "loss": 1.1516,
+      "step": 650
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.9324586391448975,
+      "learning_rate": 4.913150332822716e-06,
+      "loss": 1.1535,
+      "step": 660
+    },
+    {
+      "epoch": 1.7866666666666666,
+      "grad_norm": 0.9258119463920593,
+      "learning_rate": 4.906965362872048e-06,
+      "loss": 1.14,
+      "step": 670
+    },
+    {
+      "epoch": 1.8133333333333335,
+      "grad_norm": 0.7775595188140869,
+      "learning_rate": 4.9005718387427546e-06,
+      "loss": 1.1633,
+      "step": 680
+    },
+    {
+      "epoch": 1.8399999999999999,
+      "grad_norm": 0.9818838238716125,
+      "learning_rate": 4.893970314408813e-06,
+      "loss": 1.1303,
+      "step": 690
+    },
+    {
+      "epoch": 1.8666666666666667,
+      "grad_norm": 0.7248987555503845,
+      "learning_rate": 4.887161361866608e-06,
+      "loss": 1.116,
+      "step": 700
+    },
+    {
+      "epoch": 1.8933333333333333,
+      "grad_norm": 0.7499696612358093,
+      "learning_rate": 4.880145571085369e-06,
+      "loss": 1.1252,
+      "step": 710
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.9208846092224121,
+      "learning_rate": 4.872923549956058e-06,
+      "loss": 1.1746,
+      "step": 720
+    },
+    {
+      "epoch": 1.9466666666666668,
+      "grad_norm": 0.8718836903572083,
+      "learning_rate": 4.86549592423869e-06,
+      "loss": 1.1291,
+      "step": 730
+    },
+    {
+      "epoch": 1.9733333333333334,
+      "grad_norm": 0.8388645052909851,
+      "learning_rate": 4.857863337508119e-06,
+      "loss": 1.1033,
+      "step": 740
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.8666885495185852,
+      "learning_rate": 4.850026451098271e-06,
+      "loss": 1.1277,
+      "step": 750
+    },
+    {
+      "epoch": 2.026666666666667,
+      "grad_norm": 1.0474607944488525,
+      "learning_rate": 4.841985944044845e-06,
+      "loss": 1.1568,
+      "step": 760
+    },
+    {
+      "epoch": 2.0533333333333332,
+      "grad_norm": 0.9431724548339844,
+      "learning_rate": 4.833742513026478e-06,
+      "loss": 1.0784,
+      "step": 770
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.7262744307518005,
+      "learning_rate": 4.825296872304377e-06,
+      "loss": 1.1638,
+      "step": 780
+    },
+    {
+      "epoch": 2.1066666666666665,
+      "grad_norm": 0.9374427795410156,
+      "learning_rate": 4.816649753660431e-06,
+      "loss": 1.0557,
+      "step": 790
+    },
+    {
+      "epoch": 2.1333333333333333,
+      "grad_norm": 0.8942416310310364,
+      "learning_rate": 4.807801906333809e-06,
+      "loss": 1.0435,
+      "step": 800
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.8669484257698059,
+      "learning_rate": 4.7987540969560385e-06,
+      "loss": 1.0561,
+      "step": 810
+    },
+    {
+      "epoch": 2.1866666666666665,
+      "grad_norm": 0.9552891850471497,
+      "learning_rate": 4.789507109484579e-06,
+      "loss": 1.1086,
+      "step": 820
+    },
+    {
+      "epoch": 2.2133333333333334,
+      "grad_norm": 0.9328513145446777,
+      "learning_rate": 4.7800617451348974e-06,
+      "loss": 1.1033,
+      "step": 830
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.8567037582397461,
+      "learning_rate": 4.770418822311046e-06,
+      "loss": 1.0873,
+      "step": 840
+    },
+    {
+      "epoch": 2.2666666666666666,
+      "grad_norm": 0.7724719047546387,
+      "learning_rate": 4.760579176534747e-06,
+      "loss": 1.0785,
+      "step": 850
+    },
+    {
+      "epoch": 2.2933333333333334,
+      "grad_norm": 0.8977177143096924,
+      "learning_rate": 4.750543660373004e-06,
+      "loss": 1.0793,
+      "step": 860
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 1.0932681560516357,
+      "learning_rate": 4.7403131433642226e-06,
+      "loss": 1.0835,
+      "step": 870
+    },
+    {
+      "epoch": 2.3466666666666667,
+      "grad_norm": 0.922770619392395,
+      "learning_rate": 4.729888511942877e-06,
+      "loss": 1.0805,
+      "step": 880
+    },
+    {
+      "epoch": 2.3733333333333335,
+      "grad_norm": 0.9030175805091858,
+      "learning_rate": 4.719270669362699e-06,
+      "loss": 1.0933,
+      "step": 890
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 1.004112958908081,
+      "learning_rate": 4.708460535618411e-06,
+      "loss": 1.1069,
+      "step": 900
+    },
+    {
+      "epoch": 2.4266666666666667,
+      "grad_norm": 1.0625816583633423,
+      "learning_rate": 4.697459047366022e-06,
+      "loss": 1.0471,
+      "step": 910
+    },
+    {
+      "epoch": 2.453333333333333,
+      "grad_norm": 0.8528280258178711,
+      "learning_rate": 4.68626715784166e-06,
+      "loss": 1.0602,
+      "step": 920
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.1137285232543945,
+      "learning_rate": 4.674885836778983e-06,
+      "loss": 1.089,
+      "step": 930
+    },
+    {
+      "epoch": 2.506666666666667,
+      "grad_norm": 1.0087735652923584,
+      "learning_rate": 4.6633160703251556e-06,
+      "loss": 1.1693,
+      "step": 940
+    },
+    {
+      "epoch": 2.533333333333333,
+      "grad_norm": 0.829055666923523,
+      "learning_rate": 4.6515588609554006e-06,
+      "loss": 1.0238,
+      "step": 950
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 1.0512988567352295,
+      "learning_rate": 4.639615227386141e-06,
+      "loss": 1.1054,
+      "step": 960
+    },
+    {
+      "epoch": 2.586666666666667,
+      "grad_norm": 0.7199288606643677,
+      "learning_rate": 4.62748620448673e-06,
+      "loss": 1.1123,
+      "step": 970
+    },
+    {
+      "epoch": 2.6133333333333333,
+      "grad_norm": 0.8040386438369751,
+      "learning_rate": 4.615172843189785e-06,
+      "loss": 1.0573,
+      "step": 980
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.8797423839569092,
+      "learning_rate": 4.602676210400126e-06,
+      "loss": 1.0704,
+      "step": 990
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 0.9572004079818726,
+      "learning_rate": 4.589997388902339e-06,
+      "loss": 1.0475,
+      "step": 1000
+    },
+    {
+      "epoch": 2.6933333333333334,
+      "grad_norm": 0.9472495913505554,
+      "learning_rate": 4.577137477266948e-06,
+      "loss": 1.0591,
+      "step": 1010
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "grad_norm": 0.8334089517593384,
+      "learning_rate": 4.564097589755233e-06,
+      "loss": 1.062,
+      "step": 1020
+    },
+    {
+      "epoch": 2.7466666666666666,
+      "grad_norm": 1.1869418621063232,
+      "learning_rate": 4.550878856222684e-06,
+      "loss": 1.1608,
+      "step": 1030
+    },
+    {
+      "epoch": 2.7733333333333334,
+      "grad_norm": 1.0386947393417358,
+      "learning_rate": 4.537482422021105e-06,
+      "loss": 1.1039,
+      "step": 1040
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.9129137992858887,
+      "learning_rate": 4.523909447899365e-06,
+      "loss": 1.1069,
+      "step": 1050
+    },
+    {
+      "epoch": 2.8266666666666667,
+      "grad_norm": 1.2377030849456787,
+      "learning_rate": 4.510161109902837e-06,
+      "loss": 1.0762,
+      "step": 1060
+    },
+    {
+      "epoch": 2.8533333333333335,
+      "grad_norm": 0.881824254989624,
+      "learning_rate": 4.496238599271485e-06,
+      "loss": 1.0351,
+      "step": 1070
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.8841559886932373,
+      "learning_rate": 4.482143122336658e-06,
+      "loss": 1.0592,
+      "step": 1080
+    },
+    {
+      "epoch": 2.9066666666666667,
+      "grad_norm": 0.8864892721176147,
+      "learning_rate": 4.467875900416558e-06,
+      "loss": 1.0976,
+      "step": 1090
+    },
+    {
+      "epoch": 2.9333333333333336,
+      "grad_norm": 1.1504698991775513,
+      "learning_rate": 4.4534381697104255e-06,
+      "loss": 1.0186,
+      "step": 1100
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 1.1625890731811523,
+      "learning_rate": 4.438831181191422e-06,
+      "loss": 1.0585,
+      "step": 1110
+    },
+    {
+      "epoch": 2.986666666666667,
+      "grad_norm": 0.9680048227310181,
+      "learning_rate": 4.424056200498237e-06,
+      "loss": 1.1027,
+      "step": 1120
+    },
+    {
+      "epoch": 3.013333333333333,
+      "grad_norm": 1.0091793537139893,
+      "learning_rate": 4.409114507825431e-06,
+      "loss": 1.0871,
+      "step": 1130
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 0.887014627456665,
+      "learning_rate": 4.394007397812509e-06,
+      "loss": 1.1032,
+      "step": 1140
+    },
+    {
+      "epoch": 3.066666666666667,
+      "grad_norm": 1.0513815879821777,
+      "learning_rate": 4.3787361794317405e-06,
+      "loss": 1.0047,
+      "step": 1150
+    },
+    {
+      "epoch": 3.0933333333333333,
+      "grad_norm": 1.031087040901184,
+      "learning_rate": 4.363302175874751e-06,
+      "loss": 1.0925,
+      "step": 1160
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.9170109033584595,
+      "learning_rate": 4.347706724437865e-06,
+      "loss": 1.0359,
+      "step": 1170
+    },
+    {
+      "epoch": 3.1466666666666665,
+      "grad_norm": 1.0121595859527588,
+      "learning_rate": 4.33195117640624e-06,
+      "loss": 1.0739,
+      "step": 1180
+    },
+    {
+      "epoch": 3.1733333333333333,
+      "grad_norm": 1.2087076902389526,
+      "learning_rate": 4.316036896936774e-06,
+      "loss": 1.0599,
+      "step": 1190
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 1.2129995822906494,
+      "learning_rate": 4.299965264939834e-06,
+      "loss": 1.0306,
+      "step": 1200
+    },
+    {
+      "epoch": 3.2266666666666666,
+      "grad_norm": 1.1269701719284058,
+      "learning_rate": 4.283737672959766e-06,
+      "loss": 1.0243,
+      "step": 1210
+    },
+    {
+      "epoch": 3.2533333333333334,
+      "grad_norm": 1.1129047870635986,
+      "learning_rate": 4.267355527054243e-06,
+      "loss": 1.0502,
+      "step": 1220
+    },
+    {
+      "epoch": 3.2800000000000002,
+      "grad_norm": 0.9930374026298523,
+      "learning_rate": 4.250820246672433e-06,
+      "loss": 1.0108,
+      "step": 1230
+    },
+    {
+      "epoch": 3.3066666666666666,
+      "grad_norm": 1.027233600616455,
+      "learning_rate": 4.234133264532012e-06,
+      "loss": 1.0999,
+      "step": 1240
+    },
+    {
+      "epoch": 3.3333333333333335,
+      "grad_norm": 1.2046165466308594,
+      "learning_rate": 4.217296026495022e-06,
+      "loss": 1.0455,
+      "step": 1250
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 1.0823917388916016,
+      "learning_rate": 4.200309991442591e-06,
+      "loss": 1.0443,
+      "step": 1260
+    },
+    {
+      "epoch": 3.3866666666666667,
+      "grad_norm": 1.0402946472167969,
+      "learning_rate": 4.1831766311485345e-06,
+      "loss": 1.0554,
+      "step": 1270
+    },
+    {
+      "epoch": 3.413333333333333,
+      "grad_norm": 1.111549735069275,
+      "learning_rate": 4.165897430151822e-06,
+      "loss": 1.0947,
+      "step": 1280
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 1.0272332429885864,
+      "learning_rate": 4.148473885627952e-06,
+      "loss": 1.0399,
+      "step": 1290
+    },
+    {
+      "epoch": 3.466666666666667,
+      "grad_norm": 1.1737327575683594,
+      "learning_rate": 4.130907507259233e-06,
+      "loss": 1.0306,
+      "step": 1300
+    },
+    {
+      "epoch": 3.493333333333333,
+      "grad_norm": 1.2049609422683716,
+      "learning_rate": 4.113199817103964e-06,
+      "loss": 1.0165,
+      "step": 1310
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 1.266038179397583,
+      "learning_rate": 4.095352349464564e-06,
+      "loss": 1.0218,
+      "step": 1320
+    },
+    {
+      "epoch": 3.546666666666667,
+      "grad_norm": 1.0571839809417725,
+      "learning_rate": 4.077366650754624e-06,
+      "loss": 1.0092,
+      "step": 1330
+    },
+    {
+      "epoch": 3.5733333333333333,
+      "grad_norm": 1.0605683326721191,
+      "learning_rate": 4.059244279364923e-06,
+      "loss": 1.0095,
+      "step": 1340
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.9361685514450073,
+      "learning_rate": 4.040986805528392e-06,
+      "loss": 1.0308,
+      "step": 1350
+    },
+    {
+      "epoch": 3.626666666666667,
+      "grad_norm": 0.9611579179763794,
+      "learning_rate": 4.022595811184064e-06,
+      "loss": 1.0537,
+      "step": 1360
+    },
+    {
+      "epoch": 3.6533333333333333,
+      "grad_norm": 1.002272367477417,
+      "learning_rate": 4.004072889840006e-06,
+      "loss": 1.0158,
+      "step": 1370
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 1.0818594694137573,
+      "learning_rate": 3.985419646435244e-06,
+      "loss": 1.0203,
+      "step": 1380
+    },
+    {
+      "epoch": 3.7066666666666666,
+      "grad_norm": 1.0322810411453247,
+      "learning_rate": 3.966637697200704e-06,
+      "loss": 0.9659,
+      "step": 1390
+    },
+    {
+      "epoch": 3.7333333333333334,
+      "grad_norm": 1.035426378250122,
+      "learning_rate": 3.94772866951917e-06,
+      "loss": 1.0147,
+      "step": 1400
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 1.0693508386611938,
+      "learning_rate": 3.928694201784282e-06,
+      "loss": 0.9923,
+      "step": 1410
+    },
+    {
+      "epoch": 3.7866666666666666,
+      "grad_norm": 1.0520384311676025,
+      "learning_rate": 3.909535943258567e-06,
+      "loss": 1.0362,
+      "step": 1420
+    },
+    {
+      "epoch": 3.8133333333333335,
+      "grad_norm": 1.161549687385559,
+      "learning_rate": 3.890255553930548e-06,
+      "loss": 1.0248,
+      "step": 1430
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 1.1647064685821533,
+      "learning_rate": 3.870854704370902e-06,
+      "loss": 1.0184,
+      "step": 1440
+    },
+    {
+      "epoch": 3.8666666666666667,
+      "grad_norm": 1.0443631410598755,
+      "learning_rate": 3.851335075587717e-06,
+      "loss": 1.0075,
+      "step": 1450
+    },
+    {
+      "epoch": 3.8933333333333335,
+      "grad_norm": 1.2016109228134155,
+      "learning_rate": 3.831698358880843e-06,
+      "loss": 0.9735,
+      "step": 1460
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 1.184370994567871,
+      "learning_rate": 3.8119462556953358e-06,
+      "loss": 1.0525,
+      "step": 1470
+    },
+    {
+      "epoch": 3.9466666666666668,
+      "grad_norm": 1.0538806915283203,
+      "learning_rate": 3.7920804774740427e-06,
+      "loss": 1.0021,
+      "step": 1480
+    },
+    {
+      "epoch": 3.9733333333333336,
+      "grad_norm": 1.1907788515090942,
+      "learning_rate": 3.772102745509313e-06,
+      "loss": 1.0103,
+      "step": 1490
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.9973454475402832,
+      "learning_rate": 3.75201479079385e-06,
+      "loss": 1.0488,
+      "step": 1500
+    },
+    {
+      "epoch": 4.026666666666666,
+      "grad_norm": 1.0266902446746826,
+      "learning_rate": 3.731818353870729e-06,
+      "loss": 1.0314,
+      "step": 1510
+    },
+    {
+      "epoch": 4.053333333333334,
+      "grad_norm": 1.2079659700393677,
+      "learning_rate": 3.7115151846825874e-06,
+      "loss": 0.9852,
+      "step": 1520
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 1.3338674306869507,
+      "learning_rate": 3.6911070424199967e-06,
+      "loss": 1.0028,
+      "step": 1530
+    },
+    {
+      "epoch": 4.1066666666666665,
+      "grad_norm": 1.3352152109146118,
+      "learning_rate": 3.6705956953690364e-06,
+      "loss": 1.0209,
+      "step": 1540
+    },
+    {
+      "epoch": 4.133333333333334,
+      "grad_norm": 1.1906076669692993,
+      "learning_rate": 3.649982920758082e-06,
+      "loss": 0.9795,
+      "step": 1550
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 1.0354583263397217,
+      "learning_rate": 3.6292705046038077e-06,
+      "loss": 0.978,
+      "step": 1560
+    },
+    {
+      "epoch": 4.1866666666666665,
+      "grad_norm": 1.252071738243103,
+      "learning_rate": 3.608460241556443e-06,
+      "loss": 1.0051,
+      "step": 1570
+    },
+    {
+      "epoch": 4.213333333333333,
+      "grad_norm": 0.9613018035888672,
+      "learning_rate": 3.5875539347442694e-06,
+      "loss": 1.0369,
+      "step": 1580
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 1.083121657371521,
+      "learning_rate": 3.5665533956173857e-06,
+      "loss": 1.045,
+      "step": 1590
+    },
+    {
+      "epoch": 4.266666666666667,
+      "grad_norm": 1.428909182548523,
+      "learning_rate": 3.5454604437907535e-06,
+      "loss": 1.0239,
+      "step": 1600
+    },
+    {
+      "epoch": 4.293333333333333,
+      "grad_norm": 1.4894375801086426,
+      "learning_rate": 3.5242769068865375e-06,
+      "loss": 1.0049,
+      "step": 1610
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 1.0027519464492798,
+      "learning_rate": 3.503004620375744e-06,
+      "loss": 0.9904,
+      "step": 1620
+    },
+    {
+      "epoch": 4.346666666666667,
+      "grad_norm": 1.0072699785232544,
+      "learning_rate": 3.481645427419188e-06,
+      "loss": 0.9837,
+      "step": 1630
+    },
+    {
+      "epoch": 4.373333333333333,
+      "grad_norm": 1.1597809791564941,
+      "learning_rate": 3.460201178707791e-06,
+      "loss": 0.9199,
+      "step": 1640
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 1.1061170101165771,
+      "learning_rate": 3.438673732302223e-06,
+      "loss": 1.0196,
+      "step": 1650
+    },
+    {
+      "epoch": 4.426666666666667,
+      "grad_norm": 1.3632824420928955,
+      "learning_rate": 3.417064953471911e-06,
+      "loss": 0.987,
+      "step": 1660
+    },
+    {
+      "epoch": 4.453333333333333,
+      "grad_norm": 0.9218217134475708,
+      "learning_rate": 3.395376714533419e-06,
+      "loss": 1.0263,
+      "step": 1670
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 1.3146958351135254,
+      "learning_rate": 3.3736108946882233e-06,
+      "loss": 1.0308,
+      "step": 1680
+    },
+    {
+      "epoch": 4.506666666666667,
+      "grad_norm": 1.104790210723877,
+      "learning_rate": 3.35176937985988e-06,
+      "loss": 0.9673,
+      "step": 1690
+    },
+    {
+      "epoch": 4.533333333333333,
+      "grad_norm": 1.3057141304016113,
+      "learning_rate": 3.329854062530621e-06,
+      "loss": 1.09,
+      "step": 1700
+    },
+    {
+      "epoch": 4.5600000000000005,
+      "grad_norm": 1.2161588668823242,
+      "learning_rate": 3.307866841577381e-06,
+      "loss": 1.0211,
+      "step": 1710
+    },
+    {
+      "epoch": 4.586666666666667,
+      "grad_norm": 1.0284647941589355,
+      "learning_rate": 3.2858096221072605e-06,
+      "loss": 1.0146,
+      "step": 1720
+    },
+    {
+      "epoch": 4.613333333333333,
+      "grad_norm": 1.2662749290466309,
+      "learning_rate": 3.2636843152924595e-06,
+      "loss": 1.0113,
+      "step": 1730
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 1.086302638053894,
+      "learning_rate": 3.241492838204684e-06,
+      "loss": 0.9937,
+      "step": 1740
+    },
+    {
+      "epoch": 4.666666666666667,
+      "grad_norm": 1.2689603567123413,
+      "learning_rate": 3.2192371136490325e-06,
+      "loss": 0.9909,
+      "step": 1750
+    },
+    {
+      "epoch": 4.693333333333333,
+      "grad_norm": 1.1881645917892456,
+      "learning_rate": 3.1969190699973985e-06,
+      "loss": 0.9918,
+      "step": 1760
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 1.037034511566162,
+      "learning_rate": 3.174540641021384e-06,
+      "loss": 0.9945,
+      "step": 1770
+    },
+    {
+      "epoch": 4.746666666666667,
+      "grad_norm": 1.156396746635437,
+      "learning_rate": 3.152103765724743e-06,
+      "loss": 0.9715,
+      "step": 1780
+    },
+    {
+      "epoch": 4.773333333333333,
+      "grad_norm": 1.177418828010559,
+      "learning_rate": 3.129610388175373e-06,
+      "loss": 0.9705,
+      "step": 1790
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 1.1656800508499146,
+      "learning_rate": 3.1070624573368772e-06,
+      "loss": 1.0075,
+      "step": 1800
+    },
+    {
+      "epoch": 4.826666666666666,
+      "grad_norm": 1.6069332361221313,
+      "learning_rate": 3.0844619268996845e-06,
+      "loss": 1.0066,
+      "step": 1810
+    },
+    {
+      "epoch": 4.8533333333333335,
+      "grad_norm": 1.326254963874817,
+      "learning_rate": 3.061810755111776e-06,
+      "loss": 0.9984,
+      "step": 1820
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 1.3211278915405273,
+      "learning_rate": 3.0391109046090082e-06,
+      "loss": 1.0083,
+      "step": 1830
+    },
+    {
+      "epoch": 4.906666666666666,
+      "grad_norm": 1.1794722080230713,
+      "learning_rate": 3.016364342245059e-06,
+      "loss": 1.0047,
+      "step": 1840
+    },
+    {
+      "epoch": 4.933333333333334,
+      "grad_norm": 1.3180665969848633,
+      "learning_rate": 2.9935730389210076e-06,
+      "loss": 0.9606,
+      "step": 1850
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 1.159301996231079,
+      "learning_rate": 2.970738969414563e-06,
+      "loss": 1.0171,
+      "step": 1860
+    },
+    {
+      "epoch": 4.986666666666666,
+      "grad_norm": 1.176303505897522,
+      "learning_rate": 2.9478641122089563e-06,
+      "loss": 1.0312,
+      "step": 1870
+    },
+    {
+      "epoch": 5.013333333333334,
+      "grad_norm": 1.5347254276275635,
+      "learning_rate": 2.924950449321515e-06,
+      "loss": 0.9875,
+      "step": 1880
+    },
+    {
+      "epoch": 5.04,
+      "grad_norm": 1.2689859867095947,
+      "learning_rate": 2.9019999661319296e-06,
+      "loss": 0.9468,
+      "step": 1890
+    },
+    {
+      "epoch": 5.066666666666666,
+      "grad_norm": 1.233862280845642,
+      "learning_rate": 2.8790146512102228e-06,
+      "loss": 0.9642,
+      "step": 1900
+    },
+    {
+      "epoch": 5.093333333333334,
+      "grad_norm": 1.0744117498397827,
+      "learning_rate": 2.8559964961444533e-06,
+      "loss": 0.9939,
+      "step": 1910
+    },
+    {
+      "epoch": 5.12,
+      "grad_norm": 1.403465986251831,
+      "learning_rate": 2.8329474953681506e-06,
+      "loss": 0.9849,
+      "step": 1920
+    },
+    {
+      "epoch": 5.1466666666666665,
+      "grad_norm": 1.2598626613616943,
+      "learning_rate": 2.8098696459875048e-06,
+      "loss": 0.9666,
+      "step": 1930
+    },
+    {
+      "epoch": 5.173333333333334,
+      "grad_norm": 1.214621663093567,
+      "learning_rate": 2.786764947608324e-06,
+      "loss": 0.9845,
+      "step": 1940
+    },
+    {
+      "epoch": 5.2,
+      "grad_norm": 1.134233832359314,
+      "learning_rate": 2.7636354021627802e-06,
+      "loss": 1.0092,
+      "step": 1950
+    },
+    {
+      "epoch": 5.226666666666667,
+      "grad_norm": 1.0942606925964355,
+      "learning_rate": 2.7404830137359445e-06,
+      "loss": 1.0145,
+      "step": 1960
+    },
+    {
+      "epoch": 5.253333333333333,
+      "grad_norm": 1.3385876417160034,
+      "learning_rate": 2.717309788392144e-06,
+      "loss": 0.9876,
+      "step": 1970
+    },
+    {
+      "epoch": 5.28,
+      "grad_norm": 1.1661790609359741,
+      "learning_rate": 2.694117734001143e-06,
+      "loss": 1.0215,
+      "step": 1980
+    },
+    {
+      "epoch": 5.306666666666667,
+      "grad_norm": 1.1474223136901855,
+      "learning_rate": 2.670908860064172e-06,
+      "loss": 0.9624,
+      "step": 1990
+    },
+    {
+      "epoch": 5.333333333333333,
+      "grad_norm": 1.262390375137329,
+      "learning_rate": 2.6476851775398073e-06,
+      "loss": 0.9575,
+      "step": 2000
+    },
+    {
+      "epoch": 5.36,
+      "grad_norm": 1.2726715803146362,
+      "learning_rate": 2.624448698669731e-06,
+      "loss": 0.9906,
+      "step": 2010
+    },
+    {
+      "epoch": 5.386666666666667,
+      "grad_norm": 1.4422998428344727,
+      "learning_rate": 2.6012014368043813e-06,
+      "loss": 1.0085,
+      "step": 2020
+    },
+    {
+      "epoch": 5.413333333333333,
+      "grad_norm": 1.073887825012207,
+      "learning_rate": 2.5779454062285e-06,
+      "loss": 1.0043,
+      "step": 2030
+    },
+    {
+      "epoch": 5.44,
+      "grad_norm": 1.2706611156463623,
+      "learning_rate": 2.5546826219866018e-06,
+      "loss": 0.99,
+      "step": 2040
+    },
+    {
+      "epoch": 5.466666666666667,
+      "grad_norm": 1.2542794942855835,
+      "learning_rate": 2.531415099708382e-06,
+      "loss": 0.9931,
+      "step": 2050
+    },
+    {
+      "epoch": 5.493333333333333,
+      "grad_norm": 1.207595705986023,
+      "learning_rate": 2.5081448554340688e-06,
+      "loss": 1.0107,
+      "step": 2060
+    },
+    {
+      "epoch": 5.52,
+      "grad_norm": 1.1939449310302734,
+      "learning_rate": 2.484873905439739e-06,
+      "loss": 1.0018,
+      "step": 2070
+    },
+    {
+      "epoch": 5.546666666666667,
+      "grad_norm": 1.145877718925476,
+      "learning_rate": 2.4616042660626176e-06,
+      "loss": 0.9923,
+      "step": 2080
+    },
+    {
+      "epoch": 5.573333333333333,
+      "grad_norm": 1.1784632205963135,
+      "learning_rate": 2.4383379535263725e-06,
+      "loss": 0.9456,
+      "step": 2090
+    },
+    {
+      "epoch": 5.6,
+      "grad_norm": 1.2335214614868164,
+      "learning_rate": 2.4150769837664102e-06,
+      "loss": 0.9523,
+      "step": 2100
+    },
+    {
+      "epoch": 5.626666666666667,
+      "grad_norm": 1.4584623575210571,
+      "learning_rate": 2.391823372255208e-06,
+      "loss": 0.9789,
+      "step": 2110
+    },
+    {
+      "epoch": 5.653333333333333,
+      "grad_norm": 1.1800681352615356,
+      "learning_rate": 2.368579133827679e-06,
+      "loss": 1.0049,
+      "step": 2120
+    },
+    {
+      "epoch": 5.68,
+      "grad_norm": 1.2724323272705078,
+      "learning_rate": 2.3453462825065966e-06,
+      "loss": 0.9843,
+      "step": 2130
+    },
+    {
+      "epoch": 5.706666666666667,
+      "grad_norm": 1.3822221755981445,
+      "learning_rate": 2.3221268313280836e-06,
+      "loss": 0.9483,
+      "step": 2140
+    },
+    {
+      "epoch": 5.733333333333333,
+      "grad_norm": 1.4377061128616333,
+      "learning_rate": 2.2989227921671935e-06,
+      "loss": 0.9726,
+      "step": 2150
+    },
+    {
+      "epoch": 5.76,
+      "grad_norm": 1.4040346145629883,
+      "learning_rate": 2.27573617556359e-06,
+      "loss": 0.9877,
+      "step": 2160
+    },
+    {
+      "epoch": 5.786666666666667,
+      "grad_norm": 1.2673684358596802,
+      "learning_rate": 2.2525689905473377e-06,
+      "loss": 0.9995,
+      "step": 2170
+    },
+    {
+      "epoch": 5.8133333333333335,
+      "grad_norm": 1.2852178812026978,
+      "learning_rate": 2.2294232444648316e-06,
+      "loss": 0.9519,
+      "step": 2180
+    },
+    {
+      "epoch": 5.84,
+      "grad_norm": 1.2947932481765747,
+      "learning_rate": 2.206300942804865e-06,
+      "loss": 0.968,
+      "step": 2190
+    },
+    {
+      "epoch": 5.866666666666667,
+      "grad_norm": 1.0892807245254517,
+      "learning_rate": 2.183204089024864e-06,
+      "loss": 0.9982,
+      "step": 2200
+    },
+    {
+      "epoch": 5.8933333333333335,
+      "grad_norm": 1.2275118827819824,
+      "learning_rate": 2.160134684377295e-06,
+      "loss": 0.9891,
+      "step": 2210
+    },
+    {
+      "epoch": 5.92,
+      "grad_norm": 1.333308458328247,
+      "learning_rate": 2.1370947277362646e-06,
+      "loss": 0.977,
+      "step": 2220
+    },
+    {
+      "epoch": 5.946666666666666,
+      "grad_norm": 1.2039783000946045,
+      "learning_rate": 2.1140862154243223e-06,
+      "loss": 0.9839,
+      "step": 2230
+    },
+    {
+      "epoch": 5.973333333333334,
+      "grad_norm": 1.297869324684143,
+      "learning_rate": 2.0911111410394915e-06,
+      "loss": 0.9947,
+      "step": 2240
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 1.3803035020828247,
+      "learning_rate": 2.0681714952825274e-06,
+      "loss": 0.9492,
+      "step": 2250
+    },
+    {
+      "epoch": 6.026666666666666,
+      "grad_norm": 1.5339772701263428,
+      "learning_rate": 2.0452692657844333e-06,
+      "loss": 1.0125,
+      "step": 2260
+    },
+    {
+      "epoch": 6.053333333333334,
+      "grad_norm": 1.5292214155197144,
+      "learning_rate": 2.0224064369342388e-06,
+      "loss": 0.9575,
+      "step": 2270
+    },
+    {
+      "epoch": 6.08,
+      "grad_norm": 1.2730904817581177,
+      "learning_rate": 1.9995849897070616e-06,
+      "loss": 0.9329,
+      "step": 2280
+    },
+    {
+      "epoch": 6.1066666666666665,
+      "grad_norm": 1.529963731765747,
+      "learning_rate": 1.9768069014924622e-06,
+      "loss": 0.9253,
+      "step": 2290
+    },
+    {
+      "epoch": 6.133333333333334,
+      "grad_norm": 1.3134512901306152,
+      "learning_rate": 1.9540741459231124e-06,
+      "loss": 0.9921,
+      "step": 2300
+    },
+    {
+      "epoch": 6.16,
+      "grad_norm": 1.308095932006836,
+      "learning_rate": 1.9313886927037843e-06,
+      "loss": 0.9777,
+      "step": 2310
+    },
+    {
+      "epoch": 6.1866666666666665,
+      "grad_norm": 1.4481947422027588,
+      "learning_rate": 1.908752507440689e-06,
+      "loss": 0.9695,
+      "step": 2320
+    },
+    {
+      "epoch": 6.213333333333333,
+      "grad_norm": 1.304056167602539,
+      "learning_rate": 1.8861675514711572e-06,
+      "loss": 0.921,
+      "step": 2330
+    },
+    {
+      "epoch": 6.24,
+      "grad_norm": 1.9694091081619263,
+      "learning_rate": 1.863635781693705e-06,
+      "loss": 0.9746,
+      "step": 2340
+    },
+    {
+      "epoch": 6.266666666666667,
+      "grad_norm": 1.4358716011047363,
+      "learning_rate": 1.8411591503984687e-06,
+      "loss": 0.9828,
+      "step": 2350
+    },
+    {
+      "epoch": 6.293333333333333,
+      "grad_norm": 1.2268424034118652,
+      "learning_rate": 1.818739605098051e-06,
+      "loss": 0.9686,
+      "step": 2360
+    },
+    {
+      "epoch": 6.32,
+      "grad_norm": 1.4600015878677368,
+      "learning_rate": 1.796379088358775e-06,
+      "loss": 0.968,
+      "step": 2370
+    },
+    {
+      "epoch": 6.346666666666667,
+      "grad_norm": 1.2927517890930176,
+      "learning_rate": 1.774079537632369e-06,
+      "loss": 0.9727,
+      "step": 2380
+    },
+    {
+      "epoch": 6.373333333333333,
+      "grad_norm": 1.3452156782150269,
+      "learning_rate": 1.7518428850880928e-06,
+      "loss": 1.0324,
+      "step": 2390
+    },
+    {
+      "epoch": 6.4,
+      "grad_norm": 1.2633044719696045,
+      "learning_rate": 1.7296710574453262e-06,
+      "loss": 0.9418,
+      "step": 2400
+    },
+    {
+      "epoch": 6.426666666666667,
+      "grad_norm": 1.2620983123779297,
+      "learning_rate": 1.7075659758066207e-06,
+      "loss": 0.9838,
+      "step": 2410
+    },
+    {
+      "epoch": 6.453333333333333,
+      "grad_norm": 1.704779863357544,
+      "learning_rate": 1.6855295554912477e-06,
+      "loss": 0.9787,
+      "step": 2420
+    },
+    {
+      "epoch": 6.48,
+      "grad_norm": 1.4947396516799927,
+      "learning_rate": 1.6635637058692417e-06,
+      "loss": 0.9441,
+      "step": 2430
+    },
+    {
+      "epoch": 6.506666666666667,
+      "grad_norm": 1.3541390895843506,
+      "learning_rate": 1.6416703301959622e-06,
+      "loss": 0.9573,
+      "step": 2440
+    },
+    {
+      "epoch": 6.533333333333333,
+      "grad_norm": 1.5069646835327148,
+      "learning_rate": 1.619851325447182e-06,
+      "loss": 0.9402,
+      "step": 2450
+    },
+    {
+      "epoch": 6.5600000000000005,
+      "grad_norm": 1.052452802658081,
+      "learning_rate": 1.5981085821547237e-06,
+      "loss": 1.0117,
+      "step": 2460
+    },
+    {
+      "epoch": 6.586666666666667,
+      "grad_norm": 1.3735320568084717,
+      "learning_rate": 1.5764439842426516e-06,
+      "loss": 0.9713,
+      "step": 2470
+    },
+    {
+      "epoch": 6.613333333333333,
+      "grad_norm": 1.4487781524658203,
+      "learning_rate": 1.5548594088640368e-06,
+      "loss": 0.9591,
+      "step": 2480
+    },
+    {
+      "epoch": 6.64,
+      "grad_norm": 1.3759781122207642,
+      "learning_rate": 1.5333567262383086e-06,
+      "loss": 1.002,
+      "step": 2490
+    },
+    {
+      "epoch": 6.666666666666667,
+      "grad_norm": 1.5115113258361816,
+      "learning_rate": 1.5119377994892095e-06,
+      "loss": 0.9722,
+      "step": 2500
+    },
+    {
+      "epoch": 6.693333333333333,
+      "grad_norm": 1.2209426164627075,
+      "learning_rate": 1.4906044844833605e-06,
+      "loss": 0.9731,
+      "step": 2510
+    },
+    {
+      "epoch": 6.72,
+      "grad_norm": 1.4368799924850464,
+      "learning_rate": 1.4693586296694574e-06,
+      "loss": 0.997,
+      "step": 2520
+    },
+    {
+      "epoch": 6.746666666666667,
+      "grad_norm": 1.284833312034607,
+      "learning_rate": 1.4482020759181136e-06,
+      "loss": 0.9805,
+      "step": 2530
+    },
+    {
+      "epoch": 6.773333333333333,
+      "grad_norm": 1.4915701150894165,
+      "learning_rate": 1.4271366563623512e-06,
+      "loss": 0.9859,
+      "step": 2540
+    },
+    {
+      "epoch": 6.8,
+      "grad_norm": 1.3779624700546265,
+      "learning_rate": 1.406164196238768e-06,
+      "loss": 0.9346,
+      "step": 2550
+    },
+    {
+      "epoch": 6.826666666666666,
+      "grad_norm": 1.6257760524749756,
+      "learning_rate": 1.3852865127293901e-06,
+      "loss": 0.9563,
+      "step": 2560
+    },
+    {
+      "epoch": 6.8533333333333335,
+      "grad_norm": 1.3327535390853882,
+      "learning_rate": 1.364505414804221e-06,
+      "loss": 0.9809,
+      "step": 2570
+    },
+    {
+      "epoch": 6.88,
+      "grad_norm": 1.8340835571289062,
+      "learning_rate": 1.3438227030644946e-06,
+      "loss": 0.9424,
+      "step": 2580
+    },
+    {
+      "epoch": 6.906666666666666,
+      "grad_norm": 1.3357032537460327,
+      "learning_rate": 1.3232401695866686e-06,
+      "loss": 0.9846,
+      "step": 2590
+    },
+    {
+      "epoch": 6.933333333333334,
+      "grad_norm": 1.2755762338638306,
+      "learning_rate": 1.3027595977671443e-06,
+      "loss": 0.94,
+      "step": 2600
+    },
+    {
+      "epoch": 6.96,
+      "grad_norm": 1.5766125917434692,
+      "learning_rate": 1.282382762167739e-06,
+      "loss": 0.9797,
+      "step": 2610
+    },
+    {
+      "epoch": 6.986666666666666,
+      "grad_norm": 1.2762746810913086,
+      "learning_rate": 1.2621114283619345e-06,
+      "loss": 0.9921,
+      "step": 2620
+    },
+    {
+      "epoch": 7.013333333333334,
+      "grad_norm": 1.5844758749008179,
+      "learning_rate": 1.241947352781889e-06,
+      "loss": 0.9246,
+      "step": 2630
+    },
+    {
+      "epoch": 7.04,
+      "grad_norm": 1.1820189952850342,
+      "learning_rate": 1.2218922825662558e-06,
+      "loss": 0.989,
+      "step": 2640
+    },
+    {
+      "epoch": 7.066666666666666,
+      "grad_norm": 1.3478128910064697,
+      "learning_rate": 1.2019479554087964e-06,
+      "loss": 0.9711,
+      "step": 2650
+    },
+    {
+      "epoch": 7.093333333333334,
+      "grad_norm": 1.191152572631836,
+      "learning_rate": 1.182116099407819e-06,
+      "loss": 0.9399,
+      "step": 2660
+    },
+    {
+      "epoch": 7.12,
+      "grad_norm": 1.3069545030593872,
+      "learning_rate": 1.1623984329164413e-06,
+      "loss": 1.037,
+      "step": 2670
+    },
+    {
+      "epoch": 7.1466666666666665,
+      "grad_norm": 1.3280954360961914,
+      "learning_rate": 1.142796664393707e-06,
+      "loss": 0.9824,
+      "step": 2680
+    },
+    {
+      "epoch": 7.173333333333334,
+      "grad_norm": 1.5953059196472168,
+      "learning_rate": 1.1233124922565494e-06,
+      "loss": 0.9549,
+      "step": 2690
+    },
+    {
+      "epoch": 7.2,
+      "grad_norm": 1.1579573154449463,
+      "learning_rate": 1.1039476047326352e-06,
+      "loss": 0.9626,
+      "step": 2700
+    },
+    {
+      "epoch": 7.226666666666667,
+      "grad_norm": 1.3610824346542358,
+      "learning_rate": 1.0847036797140832e-06,
+      "loss": 0.9968,
+      "step": 2710
+    },
+    {
+      "epoch": 7.253333333333333,
+      "grad_norm": 1.4503854513168335,
+      "learning_rate": 1.065582384612082e-06,
+      "loss": 0.8981,
+      "step": 2720
+    },
+    {
+      "epoch": 7.28,
+      "grad_norm": 1.4366077184677124,
+      "learning_rate": 1.0465853762124134e-06,
+      "loss": 0.9544,
+      "step": 2730
+    },
+    {
+      "epoch": 7.306666666666667,
+      "grad_norm": 1.6409127712249756,
+      "learning_rate": 1.0277143005319038e-06,
+      "loss": 0.9835,
+      "step": 2740
+    },
+    {
+      "epoch": 7.333333333333333,
+      "grad_norm": 1.6304514408111572,
+      "learning_rate": 1.0089707926757954e-06,
+      "loss": 0.9667,
+      "step": 2750
+    },
+    {
+      "epoch": 7.36,
+      "grad_norm": 1.4634387493133545,
+      "learning_rate": 9.9035647669608e-07,
+      "loss": 0.968,
+      "step": 2760
+    },
+    {
+      "epoch": 7.386666666666667,
+      "grad_norm": 1.411413550376892,
+      "learning_rate": 9.718729654507713e-07,
+      "loss": 0.9681,
+      "step": 2770
+    },
+    {
+      "epoch": 7.413333333333333,
+      "grad_norm": 1.3875097036361694,
+      "learning_rate": 9.535218604641652e-07,
+      "loss": 0.9676,
+      "step": 2780
+    },
+    {
+      "epoch": 7.44,
+      "grad_norm": 1.4006201028823853,
+      "learning_rate": 9.353047517880709e-07,
+      "loss": 0.9615,
+      "step": 2790
+    },
+    {
+      "epoch": 7.466666666666667,
+      "grad_norm": 1.580459713935852,
+      "learning_rate": 9.172232178640361e-07,
+      "loss": 0.9638,
+      "step": 2800
+    },
+    {
+      "epoch": 7.493333333333333,
+      "grad_norm": 1.4033180475234985,
+      "learning_rate": 8.992788253865872e-07,
+      "loss": 0.9885,
+      "step": 2810
+    },
+    {
+      "epoch": 7.52,
+      "grad_norm": 1.7243731021881104,
+      "learning_rate": 8.814731291674756e-07,
+      "loss": 0.9464,
+      "step": 2820
+    },
+    {
+      "epoch": 7.546666666666667,
+      "grad_norm": 1.453606367111206,
+      "learning_rate": 8.63807672000963e-07,
+      "loss": 0.9594,
+      "step": 2830
+    },
+    {
+      "epoch": 7.573333333333333,
+      "grad_norm": 1.6155478954315186,
+      "learning_rate": 8.462839845301438e-07,
+      "loss": 0.9609,
+      "step": 2840
+    },
+    {
+      "epoch": 7.6,
+      "grad_norm": 1.4838684797286987,
+      "learning_rate": 8.289035851143198e-07,
+      "loss": 0.9643,
+      "step": 2850
+    },
+    {
+      "epoch": 7.626666666666667,
+      "grad_norm": 1.3259801864624023,
+      "learning_rate": 8.116679796974389e-07,
+      "loss": 0.9646,
+      "step": 2860
+    },
+    {
+      "epoch": 7.653333333333333,
+      "grad_norm": 1.5047879219055176,
+      "learning_rate": 7.945786616776157e-07,
+      "loss": 0.9062,
+      "step": 2870
+    },
+    {
+      "epoch": 7.68,
+      "grad_norm": 1.5503374338150024,
+      "learning_rate": 7.776371117777276e-07,
+      "loss": 0.9242,
+      "step": 2880
+    },
+    {
+      "epoch": 7.706666666666667,
+      "grad_norm": 1.6389449834823608,
+      "learning_rate": 7.60844797917123e-07,
+      "loss": 0.9476,
+      "step": 2890
+    },
+    {
+      "epoch": 7.733333333333333,
+      "grad_norm": 1.2944670915603638,
+      "learning_rate": 7.442031750844269e-07,
+      "loss": 0.974,
+      "step": 2900
+    },
+    {
+      "epoch": 7.76,
+      "grad_norm": 1.4704028367996216,
+      "learning_rate": 7.277136852114747e-07,
+      "loss": 0.9126,
+      "step": 2910
+    },
+    {
+      "epoch": 7.786666666666667,
+      "grad_norm": 1.8540769815444946,
+      "learning_rate": 7.113777570483701e-07,
+      "loss": 0.9673,
+      "step": 2920
+    },
+    {
+      "epoch": 7.8133333333333335,
+      "grad_norm": 1.45095694065094,
+      "learning_rate": 6.951968060396963e-07,
+      "loss": 0.9677,
+      "step": 2930
+    },
+    {
+      "epoch": 7.84,
+      "grad_norm": 1.4591078758239746,
+      "learning_rate": 6.791722342018656e-07,
+      "loss": 1.0081,
+      "step": 2940
+    },
+    {
+      "epoch": 7.866666666666667,
+      "grad_norm": 1.3265269994735718,
+      "learning_rate": 6.633054300016464e-07,
+      "loss": 0.9352,
+      "step": 2950
+    },
+    {
+      "epoch": 7.8933333333333335,
+      "grad_norm": 1.5188606977462769,
+      "learning_rate": 6.475977682358558e-07,
+      "loss": 0.9725,
+      "step": 2960
+    },
+    {
+      "epoch": 7.92,
+      "grad_norm": 1.5498340129852295,
+      "learning_rate": 6.320506099122359e-07,
+      "loss": 0.9883,
+      "step": 2970
+    },
+    {
+      "epoch": 7.946666666666666,
+      "grad_norm": 1.5423794984817505,
+      "learning_rate": 6.166653021315336e-07,
+      "loss": 0.9092,
+      "step": 2980
+    },
+    {
+      "epoch": 7.973333333333334,
+      "grad_norm": 1.4549285173416138,
+      "learning_rate": 6.01443177970773e-07,
+      "loss": 0.9351,
+      "step": 2990
+    },
+    {
+      "epoch": 8.0,
+      "grad_norm": 1.2420357465744019,
+      "learning_rate": 5.863855563677559e-07,
+      "loss": 0.9546,
+      "step": 3000
+    },
+    {
+      "epoch": 8.026666666666667,
+      "grad_norm": 1.233181357383728,
+      "learning_rate": 5.714937420067746e-07,
+      "loss": 0.916,
+      "step": 3010
+    },
+    {
+      "epoch": 8.053333333333333,
+      "grad_norm": 1.1143083572387695,
+      "learning_rate": 5.567690252055738e-07,
+      "loss": 0.9752,
+      "step": 3020
+    },
+    {
+      "epoch": 8.08,
+      "grad_norm": 1.7582933902740479,
+      "learning_rate": 5.422126818035403e-07,
+      "loss": 0.907,
+      "step": 3030
+    },
+    {
+      "epoch": 8.106666666666667,
+      "grad_norm": 1.5106295347213745,
+      "learning_rate": 5.278259730511651e-07,
+      "loss": 0.9487,
+      "step": 3040
+    },
+    {
+      "epoch": 8.133333333333333,
+      "grad_norm": 1.3198901414871216,
+      "learning_rate": 5.136101455007541e-07,
+      "loss": 0.922,
+      "step": 3050
+    },
+    {
+      "epoch": 8.16,
+      "grad_norm": 1.433039665222168,
+      "learning_rate": 4.995664308984254e-07,
+      "loss": 0.912,
+      "step": 3060
+    },
+    {
+      "epoch": 8.186666666666667,
+      "grad_norm": 1.372050404548645,
+      "learning_rate": 4.856960460773766e-07,
+      "loss": 0.9788,
+      "step": 3070
+    },
+    {
+      "epoch": 8.213333333333333,
+      "grad_norm": 1.4133652448654175,
+      "learning_rate": 4.7200019285245867e-07,
+      "loss": 0.9483,
+      "step": 3080
+    },
+    {
+      "epoch": 8.24,
+      "grad_norm": 1.8541817665100098,
+      "learning_rate": 4.5848005791603533e-07,
+      "loss": 0.9734,
+      "step": 3090
+    },
+    {
+      "epoch": 8.266666666666667,
+      "grad_norm": 1.7887778282165527,
+      "learning_rate": 4.451368127351674e-07,
+      "loss": 1.0254,
+      "step": 3100
+    },
+    {
+      "epoch": 8.293333333333333,
+      "grad_norm": 1.3291224241256714,
+      "learning_rate": 4.319716134501051e-07,
+      "loss": 0.969,
+      "step": 3110
+    },
+    {
+      "epoch": 8.32,
+      "grad_norm": 1.4450148344039917,
+      "learning_rate": 4.1898560077411664e-07,
+      "loss": 0.9059,
+      "step": 3120
+    },
+    {
+      "epoch": 8.346666666666668,
+      "grad_norm": 1.4905790090560913,
+      "learning_rate": 4.061798998946459e-07,
+      "loss": 0.9379,
+      "step": 3130
+    },
+    {
+      "epoch": 8.373333333333333,
+      "grad_norm": 1.5517672300338745,
+      "learning_rate": 3.935556203758237e-07,
+      "loss": 0.9874,
+      "step": 3140
+    },
+    {
+      "epoch": 8.4,
+      "grad_norm": 1.5111827850341797,
+      "learning_rate": 3.8111385606232565e-07,
+      "loss": 0.9313,
+      "step": 3150
+    },
+    {
+      "epoch": 8.426666666666666,
+      "grad_norm": 1.3383586406707764,
+      "learning_rate": 3.6885568498459395e-07,
+      "loss": 0.9489,
+      "step": 3160
+    },
+    {
+      "epoch": 8.453333333333333,
+      "grad_norm": 1.7002395391464233,
+      "learning_rate": 3.5678216926543384e-07,
+      "loss": 0.9954,
+      "step": 3170
+    },
+    {
+      "epoch": 8.48,
+      "grad_norm": 1.522956371307373,
+      "learning_rate": 3.448943550279804e-07,
+      "loss": 0.9239,
+      "step": 3180
+    },
+    {
+      "epoch": 8.506666666666666,
+      "grad_norm": 1.3977686166763306,
+      "learning_rate": 3.331932723050596e-07,
+      "loss": 0.9494,
+      "step": 3190
+    },
+    {
+      "epoch": 8.533333333333333,
+      "grad_norm": 1.385392427444458,
+      "learning_rate": 3.216799349499383e-07,
+      "loss": 1.0081,
+      "step": 3200
+    },
+    {
+      "epoch": 8.56,
+      "grad_norm": 1.3390395641326904,
+      "learning_rate": 3.1035534054847884e-07,
+      "loss": 0.9623,
+      "step": 3210
+    },
+    {
+      "epoch": 8.586666666666666,
+      "grad_norm": 1.3952502012252808,
+      "learning_rate": 2.992204703326995e-07,
+      "loss": 0.953,
+      "step": 3220
+    },
+    {
+      "epoch": 8.613333333333333,
+      "grad_norm": 1.281001329421997,
+      "learning_rate": 2.882762890957586e-07,
+      "loss": 1.0032,
+      "step": 3230
+    },
+    {
+      "epoch": 8.64,
+      "grad_norm": 1.4301480054855347,
+      "learning_rate": 2.7752374510835456e-07,
+      "loss": 0.9094,
+      "step": 3240
+    },
+    {
+      "epoch": 8.666666666666666,
+      "grad_norm": 1.2691582441329956,
+      "learning_rate": 2.6696377003656654e-07,
+      "loss": 0.9169,
+      "step": 3250
+    },
+    {
+      "epoch": 8.693333333333333,
+      "grad_norm": 1.4293104410171509,
+      "learning_rate": 2.565972788611243e-07,
+      "loss": 0.9849,
+      "step": 3260
+    },
+    {
+      "epoch": 8.72,
+      "grad_norm": 1.213094711303711,
+      "learning_rate": 2.46425169798134e-07,
+      "loss": 0.9507,
+      "step": 3270
+    },
+    {
+      "epoch": 8.746666666666666,
+      "grad_norm": 1.4834156036376953,
+      "learning_rate": 2.3644832422124565e-07,
+      "loss": 0.9814,
+      "step": 3280
+    },
+    {
+      "epoch": 8.773333333333333,
+      "grad_norm": 1.3371623754501343,
+      "learning_rate": 2.2666760658529103e-07,
+      "loss": 0.952,
+      "step": 3290
+    },
+    {
+      "epoch": 8.8,
+      "grad_norm": 1.2769241333007812,
+      "learning_rate": 2.1708386435137647e-07,
+      "loss": 0.9713,
+      "step": 3300
+    },
+    {
+      "epoch": 8.826666666666666,
+      "grad_norm": 1.410171627998352,
+      "learning_rate": 2.0769792791345945e-07,
+      "loss": 0.9523,
+      "step": 3310
+    },
+    {
+      "epoch": 8.853333333333333,
+      "grad_norm": 1.3789598941802979,
+      "learning_rate": 1.9851061052639202e-07,
+      "loss": 1.0252,
+      "step": 3320
+    },
+    {
+      "epoch": 8.88,
+      "grad_norm": 1.4115532636642456,
+      "learning_rate": 1.8952270823546253e-07,
+      "loss": 0.9217,
+      "step": 3330
+    },
+    {
+      "epoch": 8.906666666666666,
+      "grad_norm": 1.686996340751648,
+      "learning_rate": 1.8073499980741426e-07,
+      "loss": 0.9918,
+      "step": 3340
+    },
+    {
+      "epoch": 8.933333333333334,
+      "grad_norm": 1.3518773317337036,
+      "learning_rate": 1.721482466629737e-07,
+      "loss": 0.9355,
+      "step": 3350
+    },
+    {
+      "epoch": 8.96,
+      "grad_norm": 1.308510184288025,
+      "learning_rate": 1.637631928108721e-07,
+      "loss": 0.9804,
+      "step": 3360
+    },
+    {
+      "epoch": 8.986666666666666,
+      "grad_norm": 1.519740104675293,
+      "learning_rate": 1.5558056478338523e-07,
+      "loss": 0.9214,
+      "step": 3370
+    },
+    {
+      "epoch": 9.013333333333334,
+      "grad_norm": 1.4635462760925293,
+      "learning_rate": 1.476010715733761e-07,
+      "loss": 0.9473,
+      "step": 3380
+    },
+    {
+      "epoch": 9.04,
+      "grad_norm": 1.6289703845977783,
+      "learning_rate": 1.3982540457286893e-07,
+      "loss": 1.0251,
+      "step": 3390
+    },
+    {
+      "epoch": 9.066666666666666,
+      "grad_norm": 1.1922096014022827,
+      "learning_rate": 1.3225423751313942e-07,
+      "loss": 0.9078,
+      "step": 3400
+    },
+    {
+      "epoch": 9.093333333333334,
+      "grad_norm": 1.4207053184509277,
+      "learning_rate": 1.248882264063389e-07,
+      "loss": 0.9616,
+      "step": 3410
+    },
+    {
+      "epoch": 9.12,
+      "grad_norm": 1.662922739982605,
+      "learning_rate": 1.1772800948865542e-07,
+      "loss": 0.9826,
+      "step": 3420
+    },
+    {
+      "epoch": 9.146666666666667,
+      "grad_norm": 1.499869704246521,
+      "learning_rate": 1.1077420716501031e-07,
+      "loss": 0.9619,
+      "step": 3430
+    },
+    {
+      "epoch": 9.173333333333334,
+      "grad_norm": 1.6088509559631348,
+      "learning_rate": 1.0402742195530501e-07,
+      "loss": 0.9319,
+      "step": 3440
+    },
+    {
+      "epoch": 9.2,
+      "grad_norm": 1.6875691413879395,
+      "learning_rate": 9.748823844221239e-08,
+      "loss": 0.9762,
+      "step": 3450
+    },
+    {
+      "epoch": 9.226666666666667,
+      "grad_norm": 1.240744709968567,
+      "learning_rate": 9.115722322052878e-08,
+      "loss": 0.9661,
+      "step": 3460
+    },
+    {
+      "epoch": 9.253333333333334,
+      "grad_norm": 1.5115891695022583,
+      "learning_rate": 8.503492484807613e-08,
+      "loss": 0.9694,
+      "step": 3470
+    },
+    {
+      "epoch": 9.28,
+      "grad_norm": 1.4953974485397339,
+      "learning_rate": 7.912187379817582e-08,
+      "loss": 0.9815,
+      "step": 3480
+    },
+    {
+      "epoch": 9.306666666666667,
+      "grad_norm": 1.2569257020950317,
+      "learning_rate": 7.341858241368182e-08,
+      "loss": 0.9463,
+      "step": 3490
+    },
+    {
+      "epoch": 9.333333333333334,
+      "grad_norm": 1.4589368104934692,
+      "learning_rate": 6.79255448625904e-08,
+      "loss": 0.9505,
+      "step": 3500
+    },
+    {
+      "epoch": 9.36,
+      "grad_norm": 1.5550317764282227,
+      "learning_rate": 6.264323709522125e-08,
+      "loss": 0.9576,
+      "step": 3510
+    },
+    {
+      "epoch": 9.386666666666667,
+      "grad_norm": 1.359791874885559,
+      "learning_rate": 5.7572116802979695e-08,
+      "loss": 0.9569,
+      "step": 3520
+    },
+    {
+      "epoch": 9.413333333333334,
+      "grad_norm": 1.360062837600708,
+      "learning_rate": 5.2712623378697035e-08,
+      "loss": 0.9108,
+      "step": 3530
+    },
+    {
+      "epoch": 9.44,
+      "grad_norm": 1.419716238975525,
+      "learning_rate": 4.806517787856152e-08,
+      "loss": 0.999,
+      "step": 3540
+    },
+    {
+      "epoch": 9.466666666666667,
+      "grad_norm": 1.3147867918014526,
+      "learning_rate": 4.3630182985633093e-08,
+      "loss": 0.9516,
+      "step": 3550
+    },
+    {
+      "epoch": 9.493333333333334,
+      "grad_norm": 1.3970048427581787,
+      "learning_rate": 3.940802297495466e-08,
+      "loss": 0.9381,
+      "step": 3560
+    },
+    {
+      "epoch": 9.52,
+      "grad_norm": 1.278246283531189,
+      "learning_rate": 3.539906368025453e-08,
+      "loss": 0.9293,
+      "step": 3570
+    },
+    {
+      "epoch": 9.546666666666667,
+      "grad_norm": 1.3652037382125854,
+      "learning_rate": 3.1603652462249e-08,
+      "loss": 0.9309,
+      "step": 3580
+    },
+    {
+      "epoch": 9.573333333333334,
+      "grad_norm": 1.4681710004806519,
+      "learning_rate": 2.802211817854561e-08,
+      "loss": 0.9415,
+      "step": 3590
+    },
+    {
+      "epoch": 9.6,
+      "grad_norm": 1.4077391624450684,
+      "learning_rate": 2.465477115514675e-08,
+      "loss": 0.9555,
+      "step": 3600
+    },
+    {
+      "epoch": 9.626666666666667,
+      "grad_norm": 1.8016074895858765,
+      "learning_rate": 2.1501903159563688e-08,
+      "loss": 0.9258,
+      "step": 3610
+    },
+    {
+      "epoch": 9.653333333333332,
+      "grad_norm": 1.37924325466156,
+      "learning_rate": 1.856378737553427e-08,
+      "loss": 0.9798,
+      "step": 3620
+    },
+    {
+      "epoch": 9.68,
+      "grad_norm": 1.2986595630645752,
+      "learning_rate": 1.584067837935327e-08,
+      "loss": 0.9444,
+      "step": 3630
+    },
+    {
+      "epoch": 9.706666666666667,
+      "grad_norm": 1.3296507596969604,
+      "learning_rate": 1.3332812117814731e-08,
+      "loss": 0.9421,
+      "step": 3640
+    },
+    {
+      "epoch": 9.733333333333333,
+      "grad_norm": 1.47772216796875,
+      "learning_rate": 1.1040405887767503e-08,
+      "loss": 0.8807,
+      "step": 3650
+    },
+    {
+      "epoch": 9.76,
+      "grad_norm": 1.6291658878326416,
+      "learning_rate": 8.963658317286684e-09,
+      "loss": 0.9037,
+      "step": 3660
+    },
+    {
+      "epoch": 9.786666666666667,
+      "grad_norm": 1.5793346166610718,
+      "learning_rate": 7.102749348465166e-09,
+      "loss": 0.9839,
+      "step": 3670
+    },
+    {
+      "epoch": 9.813333333333333,
+      "grad_norm": 1.3573871850967407,
+      "learning_rate": 5.457840221820831e-09,
+      "loss": 0.9405,
+      "step": 3680
+    },
+    {
+      "epoch": 9.84,
+      "grad_norm": 1.3204113245010376,
+      "learning_rate": 4.029073462325506e-09,
+      "loss": 0.9589,
+      "step": 3690
+    },
+    {
+      "epoch": 9.866666666666667,
+      "grad_norm": 1.4250832796096802,
+      "learning_rate": 2.8165728670573324e-09,
+      "loss": 0.8948,
+      "step": 3700
+    },
+    {
+      "epoch": 9.893333333333333,
+      "grad_norm": 1.9013235569000244,
+      "learning_rate": 1.8204434944729676e-09,
+      "loss": 1.0401,
+      "step": 3710
+    },
+    {
+      "epoch": 9.92,
+      "grad_norm": 1.3507270812988281,
+      "learning_rate": 1.0407716553040291e-09,
+      "loss": 0.9874,
+      "step": 3720
+    },
+    {
+      "epoch": 9.946666666666667,
+      "grad_norm": 1.3722429275512695,
+      "learning_rate": 4.77624905080576e-10,
+      "loss": 0.9466,
+      "step": 3730
+    },
+    {
+      "epoch": 9.973333333333333,
+      "grad_norm": 1.249128818511963,
+      "learning_rate": 1.3105203827634693e-10,
+      "loss": 0.9351,
+      "step": 3740
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 1.5156923532485962,
+      "learning_rate": 1.0830840810327482e-12,
+      "loss": 0.9953,
+      "step": 3750
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 3750,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.3564794130805555e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}

Direction/Ageis_Danger/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9f95f06ecbe5d43fe2559731e992e8aee53d0efb541c4475216a0fb6c1b74b34
+size 5880

Direction/Beaver-Danger/README.md

@@ -0,0 +1,208 @@
+---
+base_model: /workspace/jzc/model/Qwen3-8B
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:/workspace/jzc/model/Qwen3-8B
+- llama-factory
+- lora
+- transformers
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.17.1

Direction/Beaver-Danger/adapter_config.json

@@ -0,0 +1,42 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "/workspace/jzc/model/Qwen3-8B",
+  "bias": "none",
+  "corda_config": null,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 16,
+  "lora_bias": false,
+  "lora_dropout": 0.0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "v_proj",
+    "k_proj",
+    "o_proj",
+    "up_proj",
+    "gate_proj",
+    "q_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

Direction/Beaver-Danger/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c4497dc97b43f46af20be10a5eaad6f94816cf24913c270af3e3c584a8bf66e0
+size 87360584

Direction/Beaver-Danger/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
+}

Direction/Beaver-Danger/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 %}

Direction/Beaver-Danger/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
+  }
+}

Direction/Beaver-Danger/tokenizer.json

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

Direction/Beaver-Danger/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
+}

Direction/Beaver-Danger/trainer_state.json

@@ -0,0 +1,2659 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 10.0,
+  "eval_steps": 500,
+  "global_step": 3750,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02666666666666667,
+      "grad_norm": 1.3260700702667236,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 2.9661,
+      "step": 10
+    },
+    {
+      "epoch": 0.05333333333333334,
+      "grad_norm": 1.2190521955490112,
+      "learning_rate": 2.533333333333333e-07,
+      "loss": 2.8546,
+      "step": 20
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.063557744026184,
+      "learning_rate": 3.8666666666666674e-07,
+      "loss": 2.7771,
+      "step": 30
+    },
+    {
+      "epoch": 0.10666666666666667,
+      "grad_norm": 1.1022443771362305,
+      "learning_rate": 5.2e-07,
+      "loss": 2.7986,
+      "step": 40
+    },
+    {
+      "epoch": 0.13333333333333333,
+      "grad_norm": 1.0045348405838013,
+      "learning_rate": 6.533333333333334e-07,
+      "loss": 2.7848,
+      "step": 50
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.1520053148269653,
+      "learning_rate": 7.866666666666667e-07,
+      "loss": 2.7392,
+      "step": 60
+    },
+    {
+      "epoch": 0.18666666666666668,
+      "grad_norm": 1.1911686658859253,
+      "learning_rate": 9.200000000000001e-07,
+      "loss": 2.8294,
+      "step": 70
+    },
+    {
+      "epoch": 0.21333333333333335,
+      "grad_norm": 0.8869664072990417,
+      "learning_rate": 1.0533333333333333e-06,
+      "loss": 2.7817,
+      "step": 80
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.3061726093292236,
+      "learning_rate": 1.1866666666666668e-06,
+      "loss": 2.8212,
+      "step": 90
+    },
+    {
+      "epoch": 0.26666666666666666,
+      "grad_norm": 1.6410006284713745,
+      "learning_rate": 1.32e-06,
+      "loss": 2.8196,
+      "step": 100
+    },
+    {
+      "epoch": 0.29333333333333333,
+      "grad_norm": 1.4250684976577759,
+      "learning_rate": 1.4533333333333335e-06,
+      "loss": 2.6571,
+      "step": 110
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.6860215663909912,
+      "learning_rate": 1.586666666666667e-06,
+      "loss": 2.7758,
+      "step": 120
+    },
+    {
+      "epoch": 0.3466666666666667,
+      "grad_norm": 1.3881759643554688,
+      "learning_rate": 1.72e-06,
+      "loss": 2.7617,
+      "step": 130
+    },
+    {
+      "epoch": 0.37333333333333335,
+      "grad_norm": 1.5089608430862427,
+      "learning_rate": 1.8533333333333333e-06,
+      "loss": 2.6495,
+      "step": 140
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.2898715734481812,
+      "learning_rate": 1.9866666666666666e-06,
+      "loss": 2.7944,
+      "step": 150
+    },
+    {
+      "epoch": 0.4266666666666667,
+      "grad_norm": 2.510946273803711,
+      "learning_rate": 2.12e-06,
+      "loss": 2.7703,
+      "step": 160
+    },
+    {
+      "epoch": 0.4533333333333333,
+      "grad_norm": 1.4257193803787231,
+      "learning_rate": 2.2533333333333335e-06,
+      "loss": 2.6653,
+      "step": 170
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9228178262710571,
+      "learning_rate": 2.386666666666667e-06,
+      "loss": 2.6569,
+      "step": 180
+    },
+    {
+      "epoch": 0.5066666666666667,
+      "grad_norm": 1.9167159795761108,
+      "learning_rate": 2.52e-06,
+      "loss": 2.5694,
+      "step": 190
+    },
+    {
+      "epoch": 0.5333333333333333,
+      "grad_norm": 1.9402084350585938,
+      "learning_rate": 2.6533333333333335e-06,
+      "loss": 2.4804,
+      "step": 200
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.3673012256622314,
+      "learning_rate": 2.786666666666667e-06,
+      "loss": 2.3965,
+      "step": 210
+    },
+    {
+      "epoch": 0.5866666666666667,
+      "grad_norm": 1.1642200946807861,
+      "learning_rate": 2.92e-06,
+      "loss": 2.4015,
+      "step": 220
+    },
+    {
+      "epoch": 0.6133333333333333,
+      "grad_norm": 0.8972470760345459,
+      "learning_rate": 3.053333333333334e-06,
+      "loss": 2.2293,
+      "step": 230
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.0226929187774658,
+      "learning_rate": 3.186666666666667e-06,
+      "loss": 2.2138,
+      "step": 240
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 1.28902006149292,
+      "learning_rate": 3.3200000000000004e-06,
+      "loss": 2.1027,
+      "step": 250
+    },
+    {
+      "epoch": 0.6933333333333334,
+      "grad_norm": 1.0202113389968872,
+      "learning_rate": 3.4533333333333334e-06,
+      "loss": 2.0698,
+      "step": 260
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.9459338188171387,
+      "learning_rate": 3.5866666666666673e-06,
+      "loss": 2.0861,
+      "step": 270
+    },
+    {
+      "epoch": 0.7466666666666667,
+      "grad_norm": 1.0297719240188599,
+      "learning_rate": 3.7200000000000004e-06,
+      "loss": 2.0712,
+      "step": 280
+    },
+    {
+      "epoch": 0.7733333333333333,
+      "grad_norm": 1.0611679553985596,
+      "learning_rate": 3.853333333333334e-06,
+      "loss": 1.9712,
+      "step": 290
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.9117859601974487,
+      "learning_rate": 3.986666666666667e-06,
+      "loss": 2.0212,
+      "step": 300
+    },
+    {
+      "epoch": 0.8266666666666667,
+      "grad_norm": 0.820168673992157,
+      "learning_rate": 4.12e-06,
+      "loss": 1.9746,
+      "step": 310
+    },
+    {
+      "epoch": 0.8533333333333334,
+      "grad_norm": 0.7669106125831604,
+      "learning_rate": 4.253333333333334e-06,
+      "loss": 1.9602,
+      "step": 320
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7456678748130798,
+      "learning_rate": 4.3866666666666665e-06,
+      "loss": 1.9342,
+      "step": 330
+    },
+    {
+      "epoch": 0.9066666666666666,
+      "grad_norm": 0.8794134855270386,
+      "learning_rate": 4.520000000000001e-06,
+      "loss": 1.9663,
+      "step": 340
+    },
+    {
+      "epoch": 0.9333333333333333,
+      "grad_norm": 0.6996933221817017,
+      "learning_rate": 4.653333333333333e-06,
+      "loss": 1.9244,
+      "step": 350
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.4994454383850098,
+      "learning_rate": 4.786666666666667e-06,
+      "loss": 1.9133,
+      "step": 360
+    },
+    {
+      "epoch": 0.9866666666666667,
+      "grad_norm": 0.8297287225723267,
+      "learning_rate": 4.92e-06,
+      "loss": 1.9561,
+      "step": 370
+    },
+    {
+      "epoch": 1.0133333333333334,
+      "grad_norm": 0.793429970741272,
+      "learning_rate": 4.999982670673473e-06,
+      "loss": 1.8335,
+      "step": 380
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.2577004432678223,
+      "learning_rate": 4.999787718509086e-06,
+      "loss": 1.9429,
+      "step": 390
+    },
+    {
+      "epoch": 1.0666666666666667,
+      "grad_norm": 0.8015560507774353,
+      "learning_rate": 4.999376169470306e-06,
+      "loss": 1.9304,
+      "step": 400
+    },
+    {
+      "epoch": 1.0933333333333333,
+      "grad_norm": 0.6889002323150635,
+      "learning_rate": 4.998748059216254e-06,
+      "loss": 1.7845,
+      "step": 410
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.9721826314926147,
+      "learning_rate": 4.997903442170241e-06,
+      "loss": 1.8076,
+      "step": 420
+    },
+    {
+      "epoch": 1.1466666666666667,
+      "grad_norm": 1.0830353498458862,
+      "learning_rate": 4.996842391515045e-06,
+      "loss": 1.8498,
+      "step": 430
+    },
+    {
+      "epoch": 1.1733333333333333,
+      "grad_norm": 0.6819325089454651,
+      "learning_rate": 4.995564999186572e-06,
+      "loss": 1.8069,
+      "step": 440
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.8205446004867554,
+      "learning_rate": 4.994071375865898e-06,
+      "loss": 1.8426,
+      "step": 450
+    },
+    {
+      "epoch": 1.2266666666666666,
+      "grad_norm": 0.7085703611373901,
+      "learning_rate": 4.992361650969668e-06,
+      "loss": 1.7295,
+      "step": 460
+    },
+    {
+      "epoch": 1.2533333333333334,
+      "grad_norm": 0.7688453197479248,
+      "learning_rate": 4.990435972638888e-06,
+      "loss": 1.8634,
+      "step": 470
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.7346873879432678,
+      "learning_rate": 4.988294507726089e-06,
+      "loss": 1.7841,
+      "step": 480
+    },
+    {
+      "epoch": 1.3066666666666666,
+      "grad_norm": 0.6709712743759155,
+      "learning_rate": 4.98593744178087e-06,
+      "loss": 1.7151,
+      "step": 490
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 0.7021077871322632,
+      "learning_rate": 4.983364979033819e-06,
+      "loss": 1.7789,
+      "step": 500
+    },
+    {
+      "epoch": 1.3599999999999999,
+      "grad_norm": 0.6195208430290222,
+      "learning_rate": 4.980577342378818e-06,
+      "loss": 1.8095,
+      "step": 510
+    },
+    {
+      "epoch": 1.3866666666666667,
+      "grad_norm": 0.8497361540794373,
+      "learning_rate": 4.977574773353732e-06,
+      "loss": 1.7106,
+      "step": 520
+    },
+    {
+      "epoch": 1.4133333333333333,
+      "grad_norm": 0.9590965509414673,
+      "learning_rate": 4.974357532119478e-06,
+      "loss": 1.7598,
+      "step": 530
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.6314051151275635,
+      "learning_rate": 4.970925897437484e-06,
+      "loss": 1.7866,
+      "step": 540
+    },
+    {
+      "epoch": 1.4666666666666668,
+      "grad_norm": 0.6002180576324463,
+      "learning_rate": 4.967280166645538e-06,
+      "loss": 1.7957,
+      "step": 550
+    },
+    {
+      "epoch": 1.4933333333333334,
+      "grad_norm": 0.9273757934570312,
+      "learning_rate": 4.9634206556320186e-06,
+      "loss": 1.7368,
+      "step": 560
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.9748147130012512,
+      "learning_rate": 4.959347698808532e-06,
+      "loss": 1.7901,
+      "step": 570
+    },
+    {
+      "epoch": 1.5466666666666666,
+      "grad_norm": 0.7918932437896729,
+      "learning_rate": 4.95506164908093e-06,
+      "loss": 1.7963,
+      "step": 580
+    },
+    {
+      "epoch": 1.5733333333333333,
+      "grad_norm": 0.823905348777771,
+      "learning_rate": 4.9505628778187365e-06,
+      "loss": 1.8262,
+      "step": 590
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.7708048820495605,
+      "learning_rate": 4.94585177482297e-06,
+      "loss": 1.784,
+      "step": 600
+    },
+    {
+      "epoch": 1.6266666666666667,
+      "grad_norm": 0.6546101570129395,
+      "learning_rate": 4.940928748292363e-06,
+      "loss": 1.7299,
+      "step": 610
+    },
+    {
+      "epoch": 1.6533333333333333,
+      "grad_norm": 0.9508903622627258,
+      "learning_rate": 4.9357942247879995e-06,
+      "loss": 1.7814,
+      "step": 620
+    },
+    {
+      "epoch": 1.6800000000000002,
+      "grad_norm": 0.8177071213722229,
+      "learning_rate": 4.930448649196356e-06,
+      "loss": 1.7693,
+      "step": 630
+    },
+    {
+      "epoch": 1.7066666666666666,
+      "grad_norm": 0.8351238369941711,
+      "learning_rate": 4.924892484690744e-06,
+      "loss": 1.7546,
+      "step": 640
+    },
+    {
+      "epoch": 1.7333333333333334,
+      "grad_norm": 0.6392200589179993,
+      "learning_rate": 4.91912621269119e-06,
+      "loss": 1.7813,
+      "step": 650
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.7682125568389893,
+      "learning_rate": 4.913150332822716e-06,
+      "loss": 1.705,
+      "step": 660
+    },
+    {
+      "epoch": 1.7866666666666666,
+      "grad_norm": 0.853183925151825,
+      "learning_rate": 4.906965362872048e-06,
+      "loss": 1.7138,
+      "step": 670
+    },
+    {
+      "epoch": 1.8133333333333335,
+      "grad_norm": 0.6856978535652161,
+      "learning_rate": 4.9005718387427546e-06,
+      "loss": 1.7742,
+      "step": 680
+    },
+    {
+      "epoch": 1.8399999999999999,
+      "grad_norm": 0.7499570846557617,
+      "learning_rate": 4.893970314408813e-06,
+      "loss": 1.8262,
+      "step": 690
+    },
+    {
+      "epoch": 1.8666666666666667,
+      "grad_norm": 0.9084373712539673,
+      "learning_rate": 4.887161361866608e-06,
+      "loss": 1.7991,
+      "step": 700
+    },
+    {
+      "epoch": 1.8933333333333333,
+      "grad_norm": 0.7706170082092285,
+      "learning_rate": 4.880145571085369e-06,
+      "loss": 1.7892,
+      "step": 710
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.8223916292190552,
+      "learning_rate": 4.872923549956058e-06,
+      "loss": 1.7989,
+      "step": 720
+    },
+    {
+      "epoch": 1.9466666666666668,
+      "grad_norm": 0.8382287621498108,
+      "learning_rate": 4.86549592423869e-06,
+      "loss": 1.8062,
+      "step": 730
+    },
+    {
+      "epoch": 1.9733333333333334,
+      "grad_norm": 0.562559187412262,
+      "learning_rate": 4.857863337508119e-06,
+      "loss": 1.7391,
+      "step": 740
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.9999620914459229,
+      "learning_rate": 4.850026451098271e-06,
+      "loss": 1.7954,
+      "step": 750
+    },
+    {
+      "epoch": 2.026666666666667,
+      "grad_norm": 0.7311544418334961,
+      "learning_rate": 4.841985944044845e-06,
+      "loss": 1.7463,
+      "step": 760
+    },
+    {
+      "epoch": 2.0533333333333332,
+      "grad_norm": 0.8479844331741333,
+      "learning_rate": 4.833742513026478e-06,
+      "loss": 1.7205,
+      "step": 770
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.9184486865997314,
+      "learning_rate": 4.825296872304377e-06,
+      "loss": 1.7573,
+      "step": 780
+    },
+    {
+      "epoch": 2.1066666666666665,
+      "grad_norm": 0.8509571552276611,
+      "learning_rate": 4.816649753660431e-06,
+      "loss": 1.7615,
+      "step": 790
+    },
+    {
+      "epoch": 2.1333333333333333,
+      "grad_norm": 0.6607502698898315,
+      "learning_rate": 4.807801906333809e-06,
+      "loss": 1.7398,
+      "step": 800
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.8238988518714905,
+      "learning_rate": 4.7987540969560385e-06,
+      "loss": 1.7306,
+      "step": 810
+    },
+    {
+      "epoch": 2.1866666666666665,
+      "grad_norm": 0.868171215057373,
+      "learning_rate": 4.789507109484579e-06,
+      "loss": 1.7441,
+      "step": 820
+    },
+    {
+      "epoch": 2.2133333333333334,
+      "grad_norm": 0.7437669038772583,
+      "learning_rate": 4.7800617451348974e-06,
+      "loss": 1.7433,
+      "step": 830
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.7135257720947266,
+      "learning_rate": 4.770418822311046e-06,
+      "loss": 1.789,
+      "step": 840
+    },
+    {
+      "epoch": 2.2666666666666666,
+      "grad_norm": 0.8468518853187561,
+      "learning_rate": 4.760579176534747e-06,
+      "loss": 1.7478,
+      "step": 850
+    },
+    {
+      "epoch": 2.2933333333333334,
+      "grad_norm": 1.2122632265090942,
+      "learning_rate": 4.750543660373004e-06,
+      "loss": 1.6655,
+      "step": 860
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.7517653107643127,
+      "learning_rate": 4.7403131433642226e-06,
+      "loss": 1.6733,
+      "step": 870
+    },
+    {
+      "epoch": 2.3466666666666667,
+      "grad_norm": 0.9318462014198303,
+      "learning_rate": 4.729888511942877e-06,
+      "loss": 1.7519,
+      "step": 880
+    },
+    {
+      "epoch": 2.3733333333333335,
+      "grad_norm": 0.854913592338562,
+      "learning_rate": 4.719270669362699e-06,
+      "loss": 1.7217,
+      "step": 890
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.67042475938797,
+      "learning_rate": 4.708460535618411e-06,
+      "loss": 1.7709,
+      "step": 900
+    },
+    {
+      "epoch": 2.4266666666666667,
+      "grad_norm": 0.7077845931053162,
+      "learning_rate": 4.697459047366022e-06,
+      "loss": 1.7217,
+      "step": 910
+    },
+    {
+      "epoch": 2.453333333333333,
+      "grad_norm": 0.7887906432151794,
+      "learning_rate": 4.68626715784166e-06,
+      "loss": 1.7028,
+      "step": 920
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.8714291453361511,
+      "learning_rate": 4.674885836778983e-06,
+      "loss": 1.824,
+      "step": 930
+    },
+    {
+      "epoch": 2.506666666666667,
+      "grad_norm": 0.8725156784057617,
+      "learning_rate": 4.6633160703251556e-06,
+      "loss": 1.7337,
+      "step": 940
+    },
+    {
+      "epoch": 2.533333333333333,
+      "grad_norm": 0.6795626282691956,
+      "learning_rate": 4.6515588609554006e-06,
+      "loss": 1.7776,
+      "step": 950
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.8653692603111267,
+      "learning_rate": 4.639615227386141e-06,
+      "loss": 1.6206,
+      "step": 960
+    },
+    {
+      "epoch": 2.586666666666667,
+      "grad_norm": 0.7805237770080566,
+      "learning_rate": 4.62748620448673e-06,
+      "loss": 1.7503,
+      "step": 970
+    },
+    {
+      "epoch": 2.6133333333333333,
+      "grad_norm": 0.7651466727256775,
+      "learning_rate": 4.615172843189785e-06,
+      "loss": 1.7778,
+      "step": 980
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.8041864633560181,
+      "learning_rate": 4.602676210400126e-06,
+      "loss": 1.6354,
+      "step": 990
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 0.7719329595565796,
+      "learning_rate": 4.589997388902339e-06,
+      "loss": 1.7719,
+      "step": 1000
+    },
+    {
+      "epoch": 2.6933333333333334,
+      "grad_norm": 0.9812129735946655,
+      "learning_rate": 4.577137477266948e-06,
+      "loss": 1.7822,
+      "step": 1010
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "grad_norm": 0.7337241172790527,
+      "learning_rate": 4.564097589755233e-06,
+      "loss": 1.7204,
+      "step": 1020
+    },
+    {
+      "epoch": 2.7466666666666666,
+      "grad_norm": 0.5740780234336853,
+      "learning_rate": 4.550878856222684e-06,
+      "loss": 1.7386,
+      "step": 1030
+    },
+    {
+      "epoch": 2.7733333333333334,
+      "grad_norm": 0.7582064270973206,
+      "learning_rate": 4.537482422021105e-06,
+      "loss": 1.7177,
+      "step": 1040
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.775303304195404,
+      "learning_rate": 4.523909447899365e-06,
+      "loss": 1.7284,
+      "step": 1050
+    },
+    {
+      "epoch": 2.8266666666666667,
+      "grad_norm": 0.7809900641441345,
+      "learning_rate": 4.510161109902837e-06,
+      "loss": 1.7085,
+      "step": 1060
+    },
+    {
+      "epoch": 2.8533333333333335,
+      "grad_norm": 0.8396899700164795,
+      "learning_rate": 4.496238599271485e-06,
+      "loss": 1.684,
+      "step": 1070
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.7145466208457947,
+      "learning_rate": 4.482143122336658e-06,
+      "loss": 1.7753,
+      "step": 1080
+    },
+    {
+      "epoch": 2.9066666666666667,
+      "grad_norm": 0.7297652959823608,
+      "learning_rate": 4.467875900416558e-06,
+      "loss": 1.6807,
+      "step": 1090
+    },
+    {
+      "epoch": 2.9333333333333336,
+      "grad_norm": 0.7068011164665222,
+      "learning_rate": 4.4534381697104255e-06,
+      "loss": 1.7068,
+      "step": 1100
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.7846283316612244,
+      "learning_rate": 4.438831181191422e-06,
+      "loss": 1.6906,
+      "step": 1110
+    },
+    {
+      "epoch": 2.986666666666667,
+      "grad_norm": 0.8811623454093933,
+      "learning_rate": 4.424056200498237e-06,
+      "loss": 1.6248,
+      "step": 1120
+    },
+    {
+      "epoch": 3.013333333333333,
+      "grad_norm": 0.8921287059783936,
+      "learning_rate": 4.409114507825431e-06,
+      "loss": 1.6849,
+      "step": 1130
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 0.6005886793136597,
+      "learning_rate": 4.394007397812509e-06,
+      "loss": 1.7756,
+      "step": 1140
+    },
+    {
+      "epoch": 3.066666666666667,
+      "grad_norm": 0.8570132851600647,
+      "learning_rate": 4.3787361794317405e-06,
+      "loss": 1.6584,
+      "step": 1150
+    },
+    {
+      "epoch": 3.0933333333333333,
+      "grad_norm": 0.7680262923240662,
+      "learning_rate": 4.363302175874751e-06,
+      "loss": 1.7428,
+      "step": 1160
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.9048510789871216,
+      "learning_rate": 4.347706724437865e-06,
+      "loss": 1.6959,
+      "step": 1170
+    },
+    {
+      "epoch": 3.1466666666666665,
+      "grad_norm": 0.7438698410987854,
+      "learning_rate": 4.33195117640624e-06,
+      "loss": 1.7442,
+      "step": 1180
+    },
+    {
+      "epoch": 3.1733333333333333,
+      "grad_norm": 0.7802926301956177,
+      "learning_rate": 4.316036896936774e-06,
+      "loss": 1.6798,
+      "step": 1190
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 0.8764957785606384,
+      "learning_rate": 4.299965264939834e-06,
+      "loss": 1.7208,
+      "step": 1200
+    },
+    {
+      "epoch": 3.2266666666666666,
+      "grad_norm": 0.8330620527267456,
+      "learning_rate": 4.283737672959766e-06,
+      "loss": 1.6817,
+      "step": 1210
+    },
+    {
+      "epoch": 3.2533333333333334,
+      "grad_norm": 0.905133068561554,
+      "learning_rate": 4.267355527054243e-06,
+      "loss": 1.7485,
+      "step": 1220
+    },
+    {
+      "epoch": 3.2800000000000002,
+      "grad_norm": 0.6527189612388611,
+      "learning_rate": 4.250820246672433e-06,
+      "loss": 1.7212,
+      "step": 1230
+    },
+    {
+      "epoch": 3.3066666666666666,
+      "grad_norm": 0.7526225447654724,
+      "learning_rate": 4.234133264532012e-06,
+      "loss": 1.7012,
+      "step": 1240
+    },
+    {
+      "epoch": 3.3333333333333335,
+      "grad_norm": 0.7123927474021912,
+      "learning_rate": 4.217296026495022e-06,
+      "loss": 1.6295,
+      "step": 1250
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 0.8093245625495911,
+      "learning_rate": 4.200309991442591e-06,
+      "loss": 1.7663,
+      "step": 1260
+    },
+    {
+      "epoch": 3.3866666666666667,
+      "grad_norm": 1.3336246013641357,
+      "learning_rate": 4.1831766311485345e-06,
+      "loss": 1.717,
+      "step": 1270
+    },
+    {
+      "epoch": 3.413333333333333,
+      "grad_norm": 0.7733551263809204,
+      "learning_rate": 4.165897430151822e-06,
+      "loss": 1.725,
+      "step": 1280
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 0.8533732295036316,
+      "learning_rate": 4.148473885627952e-06,
+      "loss": 1.7301,
+      "step": 1290
+    },
+    {
+      "epoch": 3.466666666666667,
+      "grad_norm": 0.9067566990852356,
+      "learning_rate": 4.130907507259233e-06,
+      "loss": 1.7135,
+      "step": 1300
+    },
+    {
+      "epoch": 3.493333333333333,
+      "grad_norm": 0.7603811621665955,
+      "learning_rate": 4.113199817103964e-06,
+      "loss": 1.7212,
+      "step": 1310
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.8485473990440369,
+      "learning_rate": 4.095352349464564e-06,
+      "loss": 1.6639,
+      "step": 1320
+    },
+    {
+      "epoch": 3.546666666666667,
+      "grad_norm": 0.7045131921768188,
+      "learning_rate": 4.077366650754624e-06,
+      "loss": 1.7154,
+      "step": 1330
+    },
+    {
+      "epoch": 3.5733333333333333,
+      "grad_norm": 1.1325947046279907,
+      "learning_rate": 4.059244279364923e-06,
+      "loss": 1.7172,
+      "step": 1340
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.9101923704147339,
+      "learning_rate": 4.040986805528392e-06,
+      "loss": 1.7554,
+      "step": 1350
+    },
+    {
+      "epoch": 3.626666666666667,
+      "grad_norm": 0.758923351764679,
+      "learning_rate": 4.022595811184064e-06,
+      "loss": 1.6887,
+      "step": 1360
+    },
+    {
+      "epoch": 3.6533333333333333,
+      "grad_norm": 0.8347360491752625,
+      "learning_rate": 4.004072889840006e-06,
+      "loss": 1.6517,
+      "step": 1370
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 0.678141713142395,
+      "learning_rate": 3.985419646435244e-06,
+      "loss": 1.6396,
+      "step": 1380
+    },
+    {
+      "epoch": 3.7066666666666666,
+      "grad_norm": 0.6728235483169556,
+      "learning_rate": 3.966637697200704e-06,
+      "loss": 1.6285,
+      "step": 1390
+    },
+    {
+      "epoch": 3.7333333333333334,
+      "grad_norm": 0.882037878036499,
+      "learning_rate": 3.94772866951917e-06,
+      "loss": 1.6916,
+      "step": 1400
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 0.8103038668632507,
+      "learning_rate": 3.928694201784282e-06,
+      "loss": 1.7442,
+      "step": 1410
+    },
+    {
+      "epoch": 3.7866666666666666,
+      "grad_norm": 0.8916019201278687,
+      "learning_rate": 3.909535943258567e-06,
+      "loss": 1.7152,
+      "step": 1420
+    },
+    {
+      "epoch": 3.8133333333333335,
+      "grad_norm": 0.8124565482139587,
+      "learning_rate": 3.890255553930548e-06,
+      "loss": 1.584,
+      "step": 1430
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 1.175717830657959,
+      "learning_rate": 3.870854704370902e-06,
+      "loss": 1.6622,
+      "step": 1440
+    },
+    {
+      "epoch": 3.8666666666666667,
+      "grad_norm": 0.7688944339752197,
+      "learning_rate": 3.851335075587717e-06,
+      "loss": 1.6617,
+      "step": 1450
+    },
+    {
+      "epoch": 3.8933333333333335,
+      "grad_norm": 0.8198382258415222,
+      "learning_rate": 3.831698358880843e-06,
+      "loss": 1.7615,
+      "step": 1460
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 0.9463858008384705,
+      "learning_rate": 3.8119462556953358e-06,
+      "loss": 1.7056,
+      "step": 1470
+    },
+    {
+      "epoch": 3.9466666666666668,
+      "grad_norm": 0.860659658908844,
+      "learning_rate": 3.7920804774740427e-06,
+      "loss": 1.7146,
+      "step": 1480
+    },
+    {
+      "epoch": 3.9733333333333336,
+      "grad_norm": 0.6901930570602417,
+      "learning_rate": 3.772102745509313e-06,
+      "loss": 1.6673,
+      "step": 1490
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.8208654522895813,
+      "learning_rate": 3.75201479079385e-06,
+      "loss": 1.6727,
+      "step": 1500
+    },
+    {
+      "epoch": 4.026666666666666,
+      "grad_norm": 0.8625465035438538,
+      "learning_rate": 3.731818353870729e-06,
+      "loss": 1.7461,
+      "step": 1510
+    },
+    {
+      "epoch": 4.053333333333334,
+      "grad_norm": 0.9363060593605042,
+      "learning_rate": 3.7115151846825874e-06,
+      "loss": 1.6987,
+      "step": 1520
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 0.85496985912323,
+      "learning_rate": 3.6911070424199967e-06,
+      "loss": 1.7011,
+      "step": 1530
+    },
+    {
+      "epoch": 4.1066666666666665,
+      "grad_norm": 0.7028411626815796,
+      "learning_rate": 3.6705956953690364e-06,
+      "loss": 1.7364,
+      "step": 1540
+    },
+    {
+      "epoch": 4.133333333333334,
+      "grad_norm": 0.7788967490196228,
+      "learning_rate": 3.649982920758082e-06,
+      "loss": 1.6497,
+      "step": 1550
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 0.7477676272392273,
+      "learning_rate": 3.6292705046038077e-06,
+      "loss": 1.6234,
+      "step": 1560
+    },
+    {
+      "epoch": 4.1866666666666665,
+      "grad_norm": 0.9469152092933655,
+      "learning_rate": 3.608460241556443e-06,
+      "loss": 1.7088,
+      "step": 1570
+    },
+    {
+      "epoch": 4.213333333333333,
+      "grad_norm": 0.71943598985672,
+      "learning_rate": 3.5875539347442694e-06,
+      "loss": 1.6717,
+      "step": 1580
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 1.0844371318817139,
+      "learning_rate": 3.5665533956173857e-06,
+      "loss": 1.6755,
+      "step": 1590
+    },
+    {
+      "epoch": 4.266666666666667,
+      "grad_norm": 0.6315621137619019,
+      "learning_rate": 3.5454604437907535e-06,
+      "loss": 1.7006,
+      "step": 1600
+    },
+    {
+      "epoch": 4.293333333333333,
+      "grad_norm": 1.314296007156372,
+      "learning_rate": 3.5242769068865375e-06,
+      "loss": 1.6445,
+      "step": 1610
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 0.684222400188446,
+      "learning_rate": 3.503004620375744e-06,
+      "loss": 1.7615,
+      "step": 1620
+    },
+    {
+      "epoch": 4.346666666666667,
+      "grad_norm": 0.9206071496009827,
+      "learning_rate": 3.481645427419188e-06,
+      "loss": 1.6997,
+      "step": 1630
+    },
+    {
+      "epoch": 4.373333333333333,
+      "grad_norm": 0.8316521048545837,
+      "learning_rate": 3.460201178707791e-06,
+      "loss": 1.6902,
+      "step": 1640
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 0.8833271861076355,
+      "learning_rate": 3.438673732302223e-06,
+      "loss": 1.7635,
+      "step": 1650
+    },
+    {
+      "epoch": 4.426666666666667,
+      "grad_norm": 1.076093316078186,
+      "learning_rate": 3.417064953471911e-06,
+      "loss": 1.669,
+      "step": 1660
+    },
+    {
+      "epoch": 4.453333333333333,
+      "grad_norm": 0.9731333255767822,
+      "learning_rate": 3.395376714533419e-06,
+      "loss": 1.6524,
+      "step": 1670
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 0.9789080619812012,
+      "learning_rate": 3.3736108946882233e-06,
+      "loss": 1.6656,
+      "step": 1680
+    },
+    {
+      "epoch": 4.506666666666667,
+      "grad_norm": 0.8803356289863586,
+      "learning_rate": 3.35176937985988e-06,
+      "loss": 1.7349,
+      "step": 1690
+    },
+    {
+      "epoch": 4.533333333333333,
+      "grad_norm": 0.8241089582443237,
+      "learning_rate": 3.329854062530621e-06,
+      "loss": 1.6017,
+      "step": 1700
+    },
+    {
+      "epoch": 4.5600000000000005,
+      "grad_norm": 0.9876794815063477,
+      "learning_rate": 3.307866841577381e-06,
+      "loss": 1.7662,
+      "step": 1710
+    },
+    {
+      "epoch": 4.586666666666667,
+      "grad_norm": 0.8717515468597412,
+      "learning_rate": 3.2858096221072605e-06,
+      "loss": 1.6321,
+      "step": 1720
+    },
+    {
+      "epoch": 4.613333333333333,
+      "grad_norm": 0.8869647979736328,
+      "learning_rate": 3.2636843152924595e-06,
+      "loss": 1.6662,
+      "step": 1730
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 0.9849333763122559,
+      "learning_rate": 3.241492838204684e-06,
+      "loss": 1.6656,
+      "step": 1740
+    },
+    {
+      "epoch": 4.666666666666667,
+      "grad_norm": 1.0160856246948242,
+      "learning_rate": 3.2192371136490325e-06,
+      "loss": 1.7322,
+      "step": 1750
+    },
+    {
+      "epoch": 4.693333333333333,
+      "grad_norm": 1.1271207332611084,
+      "learning_rate": 3.1969190699973985e-06,
+      "loss": 1.635,
+      "step": 1760
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 0.8037768006324768,
+      "learning_rate": 3.174540641021384e-06,
+      "loss": 1.6201,
+      "step": 1770
+    },
+    {
+      "epoch": 4.746666666666667,
+      "grad_norm": 0.8098615407943726,
+      "learning_rate": 3.152103765724743e-06,
+      "loss": 1.6724,
+      "step": 1780
+    },
+    {
+      "epoch": 4.773333333333333,
+      "grad_norm": 0.9140426516532898,
+      "learning_rate": 3.129610388175373e-06,
+      "loss": 1.7291,
+      "step": 1790
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 0.8616933822631836,
+      "learning_rate": 3.1070624573368772e-06,
+      "loss": 1.7201,
+      "step": 1800
+    },
+    {
+      "epoch": 4.826666666666666,
+      "grad_norm": 0.8575533032417297,
+      "learning_rate": 3.0844619268996845e-06,
+      "loss": 1.6765,
+      "step": 1810
+    },
+    {
+      "epoch": 4.8533333333333335,
+      "grad_norm": 1.0969172716140747,
+      "learning_rate": 3.061810755111776e-06,
+      "loss": 1.7236,
+      "step": 1820
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 0.8472016453742981,
+      "learning_rate": 3.0391109046090082e-06,
+      "loss": 1.6881,
+      "step": 1830
+    },
+    {
+      "epoch": 4.906666666666666,
+      "grad_norm": 0.8848344683647156,
+      "learning_rate": 3.016364342245059e-06,
+      "loss": 1.6624,
+      "step": 1840
+    },
+    {
+      "epoch": 4.933333333333334,
+      "grad_norm": 0.8401445150375366,
+      "learning_rate": 2.9935730389210076e-06,
+      "loss": 1.6553,
+      "step": 1850
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 0.8503660559654236,
+      "learning_rate": 2.970738969414563e-06,
+      "loss": 1.6752,
+      "step": 1860
+    },
+    {
+      "epoch": 4.986666666666666,
+      "grad_norm": 0.9966988563537598,
+      "learning_rate": 2.9478641122089563e-06,
+      "loss": 1.7338,
+      "step": 1870
+    },
+    {
+      "epoch": 5.013333333333334,
+      "grad_norm": 0.602899968624115,
+      "learning_rate": 2.924950449321515e-06,
+      "loss": 1.5855,
+      "step": 1880
+    },
+    {
+      "epoch": 5.04,
+      "grad_norm": 1.0888887643814087,
+      "learning_rate": 2.9019999661319296e-06,
+      "loss": 1.7572,
+      "step": 1890
+    },
+    {
+      "epoch": 5.066666666666666,
+      "grad_norm": 0.95545893907547,
+      "learning_rate": 2.8790146512102228e-06,
+      "loss": 1.6945,
+      "step": 1900
+    },
+    {
+      "epoch": 5.093333333333334,
+      "grad_norm": 0.7938442826271057,
+      "learning_rate": 2.8559964961444533e-06,
+      "loss": 1.6246,
+      "step": 1910
+    },
+    {
+      "epoch": 5.12,
+      "grad_norm": 0.9190901517868042,
+      "learning_rate": 2.8329474953681506e-06,
+      "loss": 1.7174,
+      "step": 1920
+    },
+    {
+      "epoch": 5.1466666666666665,
+      "grad_norm": 0.8581053614616394,
+      "learning_rate": 2.8098696459875048e-06,
+      "loss": 1.6249,
+      "step": 1930
+    },
+    {
+      "epoch": 5.173333333333334,
+      "grad_norm": 1.1097462177276611,
+      "learning_rate": 2.786764947608324e-06,
+      "loss": 1.7614,
+      "step": 1940
+    },
+    {
+      "epoch": 5.2,
+      "grad_norm": 0.8244250416755676,
+      "learning_rate": 2.7636354021627802e-06,
+      "loss": 1.6954,
+      "step": 1950
+    },
+    {
+      "epoch": 5.226666666666667,
+      "grad_norm": 1.0831646919250488,
+      "learning_rate": 2.7404830137359445e-06,
+      "loss": 1.7215,
+      "step": 1960
+    },
+    {
+      "epoch": 5.253333333333333,
+      "grad_norm": 0.8918278217315674,
+      "learning_rate": 2.717309788392144e-06,
+      "loss": 1.6235,
+      "step": 1970
+    },
+    {
+      "epoch": 5.28,
+      "grad_norm": 0.8835996389389038,
+      "learning_rate": 2.694117734001143e-06,
+      "loss": 1.7047,
+      "step": 1980
+    },
+    {
+      "epoch": 5.306666666666667,
+      "grad_norm": 0.8840324282646179,
+      "learning_rate": 2.670908860064172e-06,
+      "loss": 1.6767,
+      "step": 1990
+    },
+    {
+      "epoch": 5.333333333333333,
+      "grad_norm": 1.132769227027893,
+      "learning_rate": 2.6476851775398073e-06,
+      "loss": 1.6917,
+      "step": 2000
+    },
+    {
+      "epoch": 5.36,
+      "grad_norm": 0.9361376762390137,
+      "learning_rate": 2.624448698669731e-06,
+      "loss": 1.7446,
+      "step": 2010
+    },
+    {
+      "epoch": 5.386666666666667,
+      "grad_norm": 0.9744315147399902,
+      "learning_rate": 2.6012014368043813e-06,
+      "loss": 1.7049,
+      "step": 2020
+    },
+    {
+      "epoch": 5.413333333333333,
+      "grad_norm": 0.9135685563087463,
+      "learning_rate": 2.5779454062285e-06,
+      "loss": 1.6978,
+      "step": 2030
+    },
+    {
+      "epoch": 5.44,
+      "grad_norm": 0.8381025791168213,
+      "learning_rate": 2.5546826219866018e-06,
+      "loss": 1.7194,
+      "step": 2040
+    },
+    {
+      "epoch": 5.466666666666667,
+      "grad_norm": 1.0132782459259033,
+      "learning_rate": 2.531415099708382e-06,
+      "loss": 1.6255,
+      "step": 2050
+    },
+    {
+      "epoch": 5.493333333333333,
+      "grad_norm": 0.836195170879364,
+      "learning_rate": 2.5081448554340688e-06,
+      "loss": 1.6757,
+      "step": 2060
+    },
+    {
+      "epoch": 5.52,
+      "grad_norm": 1.2027889490127563,
+      "learning_rate": 2.484873905439739e-06,
+      "loss": 1.6886,
+      "step": 2070
+    },
+    {
+      "epoch": 5.546666666666667,
+      "grad_norm": 1.023413896560669,
+      "learning_rate": 2.4616042660626176e-06,
+      "loss": 1.6878,
+      "step": 2080
+    },
+    {
+      "epoch": 5.573333333333333,
+      "grad_norm": 0.9862648248672485,
+      "learning_rate": 2.4383379535263725e-06,
+      "loss": 1.6724,
+      "step": 2090
+    },
+    {
+      "epoch": 5.6,
+      "grad_norm": 0.9263085126876831,
+      "learning_rate": 2.4150769837664102e-06,
+      "loss": 1.6894,
+      "step": 2100
+    },
+    {
+      "epoch": 5.626666666666667,
+      "grad_norm": 1.2808797359466553,
+      "learning_rate": 2.391823372255208e-06,
+      "loss": 1.6914,
+      "step": 2110
+    },
+    {
+      "epoch": 5.653333333333333,
+      "grad_norm": 1.2096583843231201,
+      "learning_rate": 2.368579133827679e-06,
+      "loss": 1.6269,
+      "step": 2120
+    },
+    {
+      "epoch": 5.68,
+      "grad_norm": 0.9128383994102478,
+      "learning_rate": 2.3453462825065966e-06,
+      "loss": 1.6455,
+      "step": 2130
+    },
+    {
+      "epoch": 5.706666666666667,
+      "grad_norm": 0.9094266295433044,
+      "learning_rate": 2.3221268313280836e-06,
+      "loss": 1.674,
+      "step": 2140
+    },
+    {
+      "epoch": 5.733333333333333,
+      "grad_norm": 1.0370837450027466,
+      "learning_rate": 2.2989227921671935e-06,
+      "loss": 1.7358,
+      "step": 2150
+    },
+    {
+      "epoch": 5.76,
+      "grad_norm": 1.0183449983596802,
+      "learning_rate": 2.27573617556359e-06,
+      "loss": 1.6758,
+      "step": 2160
+    },
+    {
+      "epoch": 5.786666666666667,
+      "grad_norm": 0.7834766507148743,
+      "learning_rate": 2.2525689905473377e-06,
+      "loss": 1.64,
+      "step": 2170
+    },
+    {
+      "epoch": 5.8133333333333335,
+      "grad_norm": 0.9842936992645264,
+      "learning_rate": 2.2294232444648316e-06,
+      "loss": 1.5864,
+      "step": 2180
+    },
+    {
+      "epoch": 5.84,
+      "grad_norm": 0.9197008609771729,
+      "learning_rate": 2.206300942804865e-06,
+      "loss": 1.6242,
+      "step": 2190
+    },
+    {
+      "epoch": 5.866666666666667,
+      "grad_norm": 0.9002187252044678,
+      "learning_rate": 2.183204089024864e-06,
+      "loss": 1.7396,
+      "step": 2200
+    },
+    {
+      "epoch": 5.8933333333333335,
+      "grad_norm": 0.894535481929779,
+      "learning_rate": 2.160134684377295e-06,
+      "loss": 1.6573,
+      "step": 2210
+    },
+    {
+      "epoch": 5.92,
+      "grad_norm": 0.7890879511833191,
+      "learning_rate": 2.1370947277362646e-06,
+      "loss": 1.6712,
+      "step": 2220
+    },
+    {
+      "epoch": 5.946666666666666,
+      "grad_norm": 0.7837865352630615,
+      "learning_rate": 2.1140862154243223e-06,
+      "loss": 1.6761,
+      "step": 2230
+    },
+    {
+      "epoch": 5.973333333333334,
+      "grad_norm": 0.9894525408744812,
+      "learning_rate": 2.0911111410394915e-06,
+      "loss": 1.5695,
+      "step": 2240
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 0.9099172353744507,
+      "learning_rate": 2.0681714952825274e-06,
+      "loss": 1.6609,
+      "step": 2250
+    },
+    {
+      "epoch": 6.026666666666666,
+      "grad_norm": 0.8880329728126526,
+      "learning_rate": 2.0452692657844333e-06,
+      "loss": 1.6532,
+      "step": 2260
+    },
+    {
+      "epoch": 6.053333333333334,
+      "grad_norm": 1.021958351135254,
+      "learning_rate": 2.0224064369342388e-06,
+      "loss": 1.7392,
+      "step": 2270
+    },
+    {
+      "epoch": 6.08,
+      "grad_norm": 1.2252111434936523,
+      "learning_rate": 1.9995849897070616e-06,
+      "loss": 1.6888,
+      "step": 2280
+    },
+    {
+      "epoch": 6.1066666666666665,
+      "grad_norm": 1.0892808437347412,
+      "learning_rate": 1.9768069014924622e-06,
+      "loss": 1.569,
+      "step": 2290
+    },
+    {
+      "epoch": 6.133333333333334,
+      "grad_norm": 1.123616099357605,
+      "learning_rate": 1.9540741459231124e-06,
+      "loss": 1.6742,
+      "step": 2300
+    },
+    {
+      "epoch": 6.16,
+      "grad_norm": 1.00949227809906,
+      "learning_rate": 1.9313886927037843e-06,
+      "loss": 1.6555,
+      "step": 2310
+    },
+    {
+      "epoch": 6.1866666666666665,
+      "grad_norm": 0.8746801018714905,
+      "learning_rate": 1.908752507440689e-06,
+      "loss": 1.7283,
+      "step": 2320
+    },
+    {
+      "epoch": 6.213333333333333,
+      "grad_norm": 0.9742733836174011,
+      "learning_rate": 1.8861675514711572e-06,
+      "loss": 1.6918,
+      "step": 2330
+    },
+    {
+      "epoch": 6.24,
+      "grad_norm": 1.311090111732483,
+      "learning_rate": 1.863635781693705e-06,
+      "loss": 1.6135,
+      "step": 2340
+    },
+    {
+      "epoch": 6.266666666666667,
+      "grad_norm": 1.0751314163208008,
+      "learning_rate": 1.8411591503984687e-06,
+      "loss": 1.6877,
+      "step": 2350
+    },
+    {
+      "epoch": 6.293333333333333,
+      "grad_norm": 0.9745664000511169,
+      "learning_rate": 1.818739605098051e-06,
+      "loss": 1.7167,
+      "step": 2360
+    },
+    {
+      "epoch": 6.32,
+      "grad_norm": 1.391249656677246,
+      "learning_rate": 1.796379088358775e-06,
+      "loss": 1.6697,
+      "step": 2370
+    },
+    {
+      "epoch": 6.346666666666667,
+      "grad_norm": 1.0126644372940063,
+      "learning_rate": 1.774079537632369e-06,
+      "loss": 1.6313,
+      "step": 2380
+    },
+    {
+      "epoch": 6.373333333333333,
+      "grad_norm": 0.9063668847084045,
+      "learning_rate": 1.7518428850880928e-06,
+      "loss": 1.6423,
+      "step": 2390
+    },
+    {
+      "epoch": 6.4,
+      "grad_norm": 0.9129550457000732,
+      "learning_rate": 1.7296710574453262e-06,
+      "loss": 1.6806,
+      "step": 2400
+    },
+    {
+      "epoch": 6.426666666666667,
+      "grad_norm": 1.0306205749511719,
+      "learning_rate": 1.7075659758066207e-06,
+      "loss": 1.8324,
+      "step": 2410
+    },
+    {
+      "epoch": 6.453333333333333,
+      "grad_norm": 0.9289770126342773,
+      "learning_rate": 1.6855295554912477e-06,
+      "loss": 1.7036,
+      "step": 2420
+    },
+    {
+      "epoch": 6.48,
+      "grad_norm": 1.1734378337860107,
+      "learning_rate": 1.6635637058692417e-06,
+      "loss": 1.7271,
+      "step": 2430
+    },
+    {
+      "epoch": 6.506666666666667,
+      "grad_norm": 1.1899135112762451,
+      "learning_rate": 1.6416703301959622e-06,
+      "loss": 1.6116,
+      "step": 2440
+    },
+    {
+      "epoch": 6.533333333333333,
+      "grad_norm": 0.9925697445869446,
+      "learning_rate": 1.619851325447182e-06,
+      "loss": 1.65,
+      "step": 2450
+    },
+    {
+      "epoch": 6.5600000000000005,
+      "grad_norm": 1.0551037788391113,
+      "learning_rate": 1.5981085821547237e-06,
+      "loss": 1.6553,
+      "step": 2460
+    },
+    {
+      "epoch": 6.586666666666667,
+      "grad_norm": 0.9644522666931152,
+      "learning_rate": 1.5764439842426516e-06,
+      "loss": 1.6487,
+      "step": 2470
+    },
+    {
+      "epoch": 6.613333333333333,
+      "grad_norm": 1.1285290718078613,
+      "learning_rate": 1.5548594088640368e-06,
+      "loss": 1.692,
+      "step": 2480
+    },
+    {
+      "epoch": 6.64,
+      "grad_norm": 1.2168961763381958,
+      "learning_rate": 1.5333567262383086e-06,
+      "loss": 1.6596,
+      "step": 2490
+    },
+    {
+      "epoch": 6.666666666666667,
+      "grad_norm": 1.0515251159667969,
+      "learning_rate": 1.5119377994892095e-06,
+      "loss": 1.603,
+      "step": 2500
+    },
+    {
+      "epoch": 6.693333333333333,
+      "grad_norm": 1.0192151069641113,
+      "learning_rate": 1.4906044844833605e-06,
+      "loss": 1.6846,
+      "step": 2510
+    },
+    {
+      "epoch": 6.72,
+      "grad_norm": 1.1747512817382812,
+      "learning_rate": 1.4693586296694574e-06,
+      "loss": 1.6895,
+      "step": 2520
+    },
+    {
+      "epoch": 6.746666666666667,
+      "grad_norm": 1.0625276565551758,
+      "learning_rate": 1.4482020759181136e-06,
+      "loss": 1.6625,
+      "step": 2530
+    },
+    {
+      "epoch": 6.773333333333333,
+      "grad_norm": 1.2509487867355347,
+      "learning_rate": 1.4271366563623512e-06,
+      "loss": 1.6851,
+      "step": 2540
+    },
+    {
+      "epoch": 6.8,
+      "grad_norm": 1.12450110912323,
+      "learning_rate": 1.406164196238768e-06,
+      "loss": 1.6029,
+      "step": 2550
+    },
+    {
+      "epoch": 6.826666666666666,
+      "grad_norm": 1.1328859329223633,
+      "learning_rate": 1.3852865127293901e-06,
+      "loss": 1.5987,
+      "step": 2560
+    },
+    {
+      "epoch": 6.8533333333333335,
+      "grad_norm": 1.0585845708847046,
+      "learning_rate": 1.364505414804221e-06,
+      "loss": 1.6181,
+      "step": 2570
+    },
+    {
+      "epoch": 6.88,
+      "grad_norm": 1.2313337326049805,
+      "learning_rate": 1.3438227030644946e-06,
+      "loss": 1.7003,
+      "step": 2580
+    },
+    {
+      "epoch": 6.906666666666666,
+      "grad_norm": 1.1627655029296875,
+      "learning_rate": 1.3232401695866686e-06,
+      "loss": 1.6519,
+      "step": 2590
+    },
+    {
+      "epoch": 6.933333333333334,
+      "grad_norm": 1.4945528507232666,
+      "learning_rate": 1.3027595977671443e-06,
+      "loss": 1.6431,
+      "step": 2600
+    },
+    {
+      "epoch": 6.96,
+      "grad_norm": 1.0905581712722778,
+      "learning_rate": 1.282382762167739e-06,
+      "loss": 1.6311,
+      "step": 2610
+    },
+    {
+      "epoch": 6.986666666666666,
+      "grad_norm": 1.1594579219818115,
+      "learning_rate": 1.2621114283619345e-06,
+      "loss": 1.7169,
+      "step": 2620
+    },
+    {
+      "epoch": 7.013333333333334,
+      "grad_norm": 1.0197468996047974,
+      "learning_rate": 1.241947352781889e-06,
+      "loss": 1.6843,
+      "step": 2630
+    },
+    {
+      "epoch": 7.04,
+      "grad_norm": 0.9002560973167419,
+      "learning_rate": 1.2218922825662558e-06,
+      "loss": 1.6436,
+      "step": 2640
+    },
+    {
+      "epoch": 7.066666666666666,
+      "grad_norm": 1.0222526788711548,
+      "learning_rate": 1.2019479554087964e-06,
+      "loss": 1.6276,
+      "step": 2650
+    },
+    {
+      "epoch": 7.093333333333334,
+      "grad_norm": 1.2399083375930786,
+      "learning_rate": 1.182116099407819e-06,
+      "loss": 1.6515,
+      "step": 2660
+    },
+    {
+      "epoch": 7.12,
+      "grad_norm": 1.0906753540039062,
+      "learning_rate": 1.1623984329164413e-06,
+      "loss": 1.6438,
+      "step": 2670
+    },
+    {
+      "epoch": 7.1466666666666665,
+      "grad_norm": 1.0210061073303223,
+      "learning_rate": 1.142796664393707e-06,
+      "loss": 1.705,
+      "step": 2680
+    },
+    {
+      "epoch": 7.173333333333334,
+      "grad_norm": 0.9715026617050171,
+      "learning_rate": 1.1233124922565494e-06,
+      "loss": 1.7126,
+      "step": 2690
+    },
+    {
+      "epoch": 7.2,
+      "grad_norm": 0.8580982685089111,
+      "learning_rate": 1.1039476047326352e-06,
+      "loss": 1.6729,
+      "step": 2700
+    },
+    {
+      "epoch": 7.226666666666667,
+      "grad_norm": 1.1537833213806152,
+      "learning_rate": 1.0847036797140832e-06,
+      "loss": 1.7293,
+      "step": 2710
+    },
+    {
+      "epoch": 7.253333333333333,
+      "grad_norm": 0.998429536819458,
+      "learning_rate": 1.065582384612082e-06,
+      "loss": 1.7036,
+      "step": 2720
+    },
+    {
+      "epoch": 7.28,
+      "grad_norm": 1.0259299278259277,
+      "learning_rate": 1.0465853762124134e-06,
+      "loss": 1.6303,
+      "step": 2730
+    },
+    {
+      "epoch": 7.306666666666667,
+      "grad_norm": 1.121588110923767,
+      "learning_rate": 1.0277143005319038e-06,
+      "loss": 1.6615,
+      "step": 2740
+    },
+    {
+      "epoch": 7.333333333333333,
+      "grad_norm": 0.9042307138442993,
+      "learning_rate": 1.0089707926757954e-06,
+      "loss": 1.6152,
+      "step": 2750
+    },
+    {
+      "epoch": 7.36,
+      "grad_norm": 0.9918031692504883,
+      "learning_rate": 9.9035647669608e-07,
+      "loss": 1.5828,
+      "step": 2760
+    },
+    {
+      "epoch": 7.386666666666667,
+      "grad_norm": 1.0638200044631958,
+      "learning_rate": 9.718729654507713e-07,
+      "loss": 1.6624,
+      "step": 2770
+    },
+    {
+      "epoch": 7.413333333333333,
+      "grad_norm": 1.013146996498108,
+      "learning_rate": 9.535218604641652e-07,
+      "loss": 1.7115,
+      "step": 2780
+    },
+    {
+      "epoch": 7.44,
+      "grad_norm": 1.302118182182312,
+      "learning_rate": 9.353047517880709e-07,
+      "loss": 1.638,
+      "step": 2790
+    },
+    {
+      "epoch": 7.466666666666667,
+      "grad_norm": 0.7931357622146606,
+      "learning_rate": 9.172232178640361e-07,
+      "loss": 1.7482,
+      "step": 2800
+    },
+    {
+      "epoch": 7.493333333333333,
+      "grad_norm": 0.9321092367172241,
+      "learning_rate": 8.992788253865872e-07,
+      "loss": 1.6079,
+      "step": 2810
+    },
+    {
+      "epoch": 7.52,
+      "grad_norm": 1.0876185894012451,
+      "learning_rate": 8.814731291674756e-07,
+      "loss": 1.6749,
+      "step": 2820
+    },
+    {
+      "epoch": 7.546666666666667,
+      "grad_norm": 1.0826046466827393,
+      "learning_rate": 8.63807672000963e-07,
+      "loss": 1.6491,
+      "step": 2830
+    },
+    {
+      "epoch": 7.573333333333333,
+      "grad_norm": 1.0835968255996704,
+      "learning_rate": 8.462839845301438e-07,
+      "loss": 1.6671,
+      "step": 2840
+    },
+    {
+      "epoch": 7.6,
+      "grad_norm": 1.2509957551956177,
+      "learning_rate": 8.289035851143198e-07,
+      "loss": 1.67,
+      "step": 2850
+    },
+    {
+      "epoch": 7.626666666666667,
+      "grad_norm": 1.1528269052505493,
+      "learning_rate": 8.116679796974389e-07,
+      "loss": 1.6247,
+      "step": 2860
+    },
+    {
+      "epoch": 7.653333333333333,
+      "grad_norm": 0.9256828427314758,
+      "learning_rate": 7.945786616776157e-07,
+      "loss": 1.6268,
+      "step": 2870
+    },
+    {
+      "epoch": 7.68,
+      "grad_norm": 1.0600520372390747,
+      "learning_rate": 7.776371117777276e-07,
+      "loss": 1.6902,
+      "step": 2880
+    },
+    {
+      "epoch": 7.706666666666667,
+      "grad_norm": 0.9527045488357544,
+      "learning_rate": 7.60844797917123e-07,
+      "loss": 1.5988,
+      "step": 2890
+    },
+    {
+      "epoch": 7.733333333333333,
+      "grad_norm": 1.0828256607055664,
+      "learning_rate": 7.442031750844269e-07,
+      "loss": 1.6704,
+      "step": 2900
+    },
+    {
+      "epoch": 7.76,
+      "grad_norm": 1.1773947477340698,
+      "learning_rate": 7.277136852114747e-07,
+      "loss": 1.6827,
+      "step": 2910
+    },
+    {
+      "epoch": 7.786666666666667,
+      "grad_norm": 1.3546745777130127,
+      "learning_rate": 7.113777570483701e-07,
+      "loss": 1.6406,
+      "step": 2920
+    },
+    {
+      "epoch": 7.8133333333333335,
+      "grad_norm": 1.0221235752105713,
+      "learning_rate": 6.951968060396963e-07,
+      "loss": 1.6239,
+      "step": 2930
+    },
+    {
+      "epoch": 7.84,
+      "grad_norm": 0.8836374282836914,
+      "learning_rate": 6.791722342018656e-07,
+      "loss": 1.6505,
+      "step": 2940
+    },
+    {
+      "epoch": 7.866666666666667,
+      "grad_norm": 1.4861336946487427,
+      "learning_rate": 6.633054300016464e-07,
+      "loss": 1.6439,
+      "step": 2950
+    },
+    {
+      "epoch": 7.8933333333333335,
+      "grad_norm": 1.0975189208984375,
+      "learning_rate": 6.475977682358558e-07,
+      "loss": 1.6943,
+      "step": 2960
+    },
+    {
+      "epoch": 7.92,
+      "grad_norm": 0.9906185269355774,
+      "learning_rate": 6.320506099122359e-07,
+      "loss": 1.6972,
+      "step": 2970
+    },
+    {
+      "epoch": 7.946666666666666,
+      "grad_norm": 0.9991128444671631,
+      "learning_rate": 6.166653021315336e-07,
+      "loss": 1.7144,
+      "step": 2980
+    },
+    {
+      "epoch": 7.973333333333334,
+      "grad_norm": 0.7832310795783997,
+      "learning_rate": 6.01443177970773e-07,
+      "loss": 1.64,
+      "step": 2990
+    },
+    {
+      "epoch": 8.0,
+      "grad_norm": 1.045725703239441,
+      "learning_rate": 5.863855563677559e-07,
+      "loss": 1.6638,
+      "step": 3000
+    },
+    {
+      "epoch": 8.026666666666667,
+      "grad_norm": 0.9321222901344299,
+      "learning_rate": 5.714937420067746e-07,
+      "loss": 1.6576,
+      "step": 3010
+    },
+    {
+      "epoch": 8.053333333333333,
+      "grad_norm": 1.1314470767974854,
+      "learning_rate": 5.567690252055738e-07,
+      "loss": 1.634,
+      "step": 3020
+    },
+    {
+      "epoch": 8.08,
+      "grad_norm": 1.1661452054977417,
+      "learning_rate": 5.422126818035403e-07,
+      "loss": 1.6576,
+      "step": 3030
+    },
+    {
+      "epoch": 8.106666666666667,
+      "grad_norm": 1.023749589920044,
+      "learning_rate": 5.278259730511651e-07,
+      "loss": 1.6985,
+      "step": 3040
+    },
+    {
+      "epoch": 8.133333333333333,
+      "grad_norm": 0.8616606593132019,
+      "learning_rate": 5.136101455007541e-07,
+      "loss": 1.7006,
+      "step": 3050
+    },
+    {
+      "epoch": 8.16,
+      "grad_norm": 1.0434175729751587,
+      "learning_rate": 4.995664308984254e-07,
+      "loss": 1.6621,
+      "step": 3060
+    },
+    {
+      "epoch": 8.186666666666667,
+      "grad_norm": 1.206966519355774,
+      "learning_rate": 4.856960460773766e-07,
+      "loss": 1.6407,
+      "step": 3070
+    },
+    {
+      "epoch": 8.213333333333333,
+      "grad_norm": 0.9314172863960266,
+      "learning_rate": 4.7200019285245867e-07,
+      "loss": 1.7501,
+      "step": 3080
+    },
+    {
+      "epoch": 8.24,
+      "grad_norm": 1.2337085008621216,
+      "learning_rate": 4.5848005791603533e-07,
+      "loss": 1.6169,
+      "step": 3090
+    },
+    {
+      "epoch": 8.266666666666667,
+      "grad_norm": 0.9701104164123535,
+      "learning_rate": 4.451368127351674e-07,
+      "loss": 1.6666,
+      "step": 3100
+    },
+    {
+      "epoch": 8.293333333333333,
+      "grad_norm": 1.1220892667770386,
+      "learning_rate": 4.319716134501051e-07,
+      "loss": 1.5962,
+      "step": 3110
+    },
+    {
+      "epoch": 8.32,
+      "grad_norm": 1.0029473304748535,
+      "learning_rate": 4.1898560077411664e-07,
+      "loss": 1.6603,
+      "step": 3120
+    },
+    {
+      "epoch": 8.346666666666668,
+      "grad_norm": 1.011228084564209,
+      "learning_rate": 4.061798998946459e-07,
+      "loss": 1.6106,
+      "step": 3130
+    },
+    {
+      "epoch": 8.373333333333333,
+      "grad_norm": 1.3020498752593994,
+      "learning_rate": 3.935556203758237e-07,
+      "loss": 1.669,
+      "step": 3140
+    },
+    {
+      "epoch": 8.4,
+      "grad_norm": 1.105768084526062,
+      "learning_rate": 3.8111385606232565e-07,
+      "loss": 1.672,
+      "step": 3150
+    },
+    {
+      "epoch": 8.426666666666666,
+      "grad_norm": 1.1611669063568115,
+      "learning_rate": 3.6885568498459395e-07,
+      "loss": 1.6565,
+      "step": 3160
+    },
+    {
+      "epoch": 8.453333333333333,
+      "grad_norm": 0.9635624289512634,
+      "learning_rate": 3.5678216926543384e-07,
+      "loss": 1.6181,
+      "step": 3170
+    },
+    {
+      "epoch": 8.48,
+      "grad_norm": 1.1056032180786133,
+      "learning_rate": 3.448943550279804e-07,
+      "loss": 1.6407,
+      "step": 3180
+    },
+    {
+      "epoch": 8.506666666666666,
+      "grad_norm": 1.136650800704956,
+      "learning_rate": 3.331932723050596e-07,
+      "loss": 1.6685,
+      "step": 3190
+    },
+    {
+      "epoch": 8.533333333333333,
+      "grad_norm": 0.9246675968170166,
+      "learning_rate": 3.216799349499383e-07,
+      "loss": 1.6401,
+      "step": 3200
+    },
+    {
+      "epoch": 8.56,
+      "grad_norm": 1.0271546840667725,
+      "learning_rate": 3.1035534054847884e-07,
+      "loss": 1.6806,
+      "step": 3210
+    },
+    {
+      "epoch": 8.586666666666666,
+      "grad_norm": 1.0452704429626465,
+      "learning_rate": 2.992204703326995e-07,
+      "loss": 1.6791,
+      "step": 3220
+    },
+    {
+      "epoch": 8.613333333333333,
+      "grad_norm": 1.186958909034729,
+      "learning_rate": 2.882762890957586e-07,
+      "loss": 1.7309,
+      "step": 3230
+    },
+    {
+      "epoch": 8.64,
+      "grad_norm": 1.1407443284988403,
+      "learning_rate": 2.7752374510835456e-07,
+      "loss": 1.6452,
+      "step": 3240
+    },
+    {
+      "epoch": 8.666666666666666,
+      "grad_norm": 1.0465816259384155,
+      "learning_rate": 2.6696377003656654e-07,
+      "loss": 1.6748,
+      "step": 3250
+    },
+    {
+      "epoch": 8.693333333333333,
+      "grad_norm": 1.1659401655197144,
+      "learning_rate": 2.565972788611243e-07,
+      "loss": 1.6536,
+      "step": 3260
+    },
+    {
+      "epoch": 8.72,
+      "grad_norm": 1.190261960029602,
+      "learning_rate": 2.46425169798134e-07,
+      "loss": 1.7099,
+      "step": 3270
+    },
+    {
+      "epoch": 8.746666666666666,
+      "grad_norm": 1.4275482892990112,
+      "learning_rate": 2.3644832422124565e-07,
+      "loss": 1.6508,
+      "step": 3280
+    },
+    {
+      "epoch": 8.773333333333333,
+      "grad_norm": 0.7873063087463379,
+      "learning_rate": 2.2666760658529103e-07,
+      "loss": 1.5954,
+      "step": 3290
+    },
+    {
+      "epoch": 8.8,
+      "grad_norm": 1.0107383728027344,
+      "learning_rate": 2.1708386435137647e-07,
+      "loss": 1.6384,
+      "step": 3300
+    },
+    {
+      "epoch": 8.826666666666666,
+      "grad_norm": 1.151199221611023,
+      "learning_rate": 2.0769792791345945e-07,
+      "loss": 1.5958,
+      "step": 3310
+    },
+    {
+      "epoch": 8.853333333333333,
+      "grad_norm": 0.9328808188438416,
+      "learning_rate": 1.9851061052639202e-07,
+      "loss": 1.7356,
+      "step": 3320
+    },
+    {
+      "epoch": 8.88,
+      "grad_norm": 1.2401057481765747,
+      "learning_rate": 1.8952270823546253e-07,
+      "loss": 1.6644,
+      "step": 3330
+    },
+    {
+      "epoch": 8.906666666666666,
+      "grad_norm": 1.0410566329956055,
+      "learning_rate": 1.8073499980741426e-07,
+      "loss": 1.6447,
+      "step": 3340
+    },
+    {
+      "epoch": 8.933333333333334,
+      "grad_norm": 0.9945191144943237,
+      "learning_rate": 1.721482466629737e-07,
+      "loss": 1.63,
+      "step": 3350
+    },
+    {
+      "epoch": 8.96,
+      "grad_norm": 1.384772539138794,
+      "learning_rate": 1.637631928108721e-07,
+      "loss": 1.5961,
+      "step": 3360
+    },
+    {
+      "epoch": 8.986666666666666,
+      "grad_norm": 1.1313170194625854,
+      "learning_rate": 1.5558056478338523e-07,
+      "loss": 1.6825,
+      "step": 3370
+    },
+    {
+      "epoch": 9.013333333333334,
+      "grad_norm": 1.0066853761672974,
+      "learning_rate": 1.476010715733761e-07,
+      "loss": 1.6951,
+      "step": 3380
+    },
+    {
+      "epoch": 9.04,
+      "grad_norm": 1.199820637702942,
+      "learning_rate": 1.3982540457286893e-07,
+      "loss": 1.6796,
+      "step": 3390
+    },
+    {
+      "epoch": 9.066666666666666,
+      "grad_norm": 1.135715126991272,
+      "learning_rate": 1.3225423751313942e-07,
+      "loss": 1.6329,
+      "step": 3400
+    },
+    {
+      "epoch": 9.093333333333334,
+      "grad_norm": 1.0266982316970825,
+      "learning_rate": 1.248882264063389e-07,
+      "loss": 1.6754,
+      "step": 3410
+    },
+    {
+      "epoch": 9.12,
+      "grad_norm": 1.192221999168396,
+      "learning_rate": 1.1772800948865542e-07,
+      "loss": 1.7027,
+      "step": 3420
+    },
+    {
+      "epoch": 9.146666666666667,
+      "grad_norm": 1.3425045013427734,
+      "learning_rate": 1.1077420716501031e-07,
+      "loss": 1.6439,
+      "step": 3430
+    },
+    {
+      "epoch": 9.173333333333334,
+      "grad_norm": 1.2292110919952393,
+      "learning_rate": 1.0402742195530501e-07,
+      "loss": 1.6746,
+      "step": 3440
+    },
+    {
+      "epoch": 9.2,
+      "grad_norm": 0.9197198152542114,
+      "learning_rate": 9.748823844221239e-08,
+      "loss": 1.6584,
+      "step": 3450
+    },
+    {
+      "epoch": 9.226666666666667,
+      "grad_norm": 0.9650914669036865,
+      "learning_rate": 9.115722322052878e-08,
+      "loss": 1.6752,
+      "step": 3460
+    },
+    {
+      "epoch": 9.253333333333334,
+      "grad_norm": 1.2544925212860107,
+      "learning_rate": 8.503492484807613e-08,
+      "loss": 1.6303,
+      "step": 3470
+    },
+    {
+      "epoch": 9.28,
+      "grad_norm": 1.1284185647964478,
+      "learning_rate": 7.912187379817582e-08,
+      "loss": 1.6722,
+      "step": 3480
+    },
+    {
+      "epoch": 9.306666666666667,
+      "grad_norm": 1.0232893228530884,
+      "learning_rate": 7.341858241368182e-08,
+      "loss": 1.7343,
+      "step": 3490
+    },
+    {
+      "epoch": 9.333333333333334,
+      "grad_norm": 1.286456823348999,
+      "learning_rate": 6.79255448625904e-08,
+      "loss": 1.6559,
+      "step": 3500
+    },
+    {
+      "epoch": 9.36,
+      "grad_norm": 0.9723238348960876,
+      "learning_rate": 6.264323709522125e-08,
+      "loss": 1.7105,
+      "step": 3510
+    },
+    {
+      "epoch": 9.386666666666667,
+      "grad_norm": 0.915547251701355,
+      "learning_rate": 5.7572116802979695e-08,
+      "loss": 1.6813,
+      "step": 3520
+    },
+    {
+      "epoch": 9.413333333333334,
+      "grad_norm": 1.073350191116333,
+      "learning_rate": 5.2712623378697035e-08,
+      "loss": 1.662,
+      "step": 3530
+    },
+    {
+      "epoch": 9.44,
+      "grad_norm": 1.4046480655670166,
+      "learning_rate": 4.806517787856152e-08,
+      "loss": 1.6464,
+      "step": 3540
+    },
+    {
+      "epoch": 9.466666666666667,
+      "grad_norm": 0.9930236339569092,
+      "learning_rate": 4.3630182985633093e-08,
+      "loss": 1.5746,
+      "step": 3550
+    },
+    {
+      "epoch": 9.493333333333334,
+      "grad_norm": 0.9176751375198364,
+      "learning_rate": 3.940802297495466e-08,
+      "loss": 1.6134,
+      "step": 3560
+    },
+    {
+      "epoch": 9.52,
+      "grad_norm": 1.1889771223068237,
+      "learning_rate": 3.539906368025453e-08,
+      "loss": 1.6769,
+      "step": 3570
+    },
+    {
+      "epoch": 9.546666666666667,
+      "grad_norm": 1.149104118347168,
+      "learning_rate": 3.1603652462249e-08,
+      "loss": 1.6366,
+      "step": 3580
+    },
+    {
+      "epoch": 9.573333333333334,
+      "grad_norm": 0.9914628863334656,
+      "learning_rate": 2.802211817854561e-08,
+      "loss": 1.6296,
+      "step": 3590
+    },
+    {
+      "epoch": 9.6,
+      "grad_norm": 0.9465372562408447,
+      "learning_rate": 2.465477115514675e-08,
+      "loss": 1.6666,
+      "step": 3600
+    },
+    {
+      "epoch": 9.626666666666667,
+      "grad_norm": 1.336161494255066,
+      "learning_rate": 2.1501903159563688e-08,
+      "loss": 1.5738,
+      "step": 3610
+    },
+    {
+      "epoch": 9.653333333333332,
+      "grad_norm": 0.9720616340637207,
+      "learning_rate": 1.856378737553427e-08,
+      "loss": 1.6266,
+      "step": 3620
+    },
+    {
+      "epoch": 9.68,
+      "grad_norm": 0.8966802358627319,
+      "learning_rate": 1.584067837935327e-08,
+      "loss": 1.6879,
+      "step": 3630
+    },
+    {
+      "epoch": 9.706666666666667,
+      "grad_norm": 1.0030333995819092,
+      "learning_rate": 1.3332812117814731e-08,
+      "loss": 1.6272,
+      "step": 3640
+    },
+    {
+      "epoch": 9.733333333333333,
+      "grad_norm": 0.9756541848182678,
+      "learning_rate": 1.1040405887767503e-08,
+      "loss": 1.6611,
+      "step": 3650
+    },
+    {
+      "epoch": 9.76,
+      "grad_norm": 1.1682424545288086,
+      "learning_rate": 8.963658317286684e-09,
+      "loss": 1.6519,
+      "step": 3660
+    },
+    {
+      "epoch": 9.786666666666667,
+      "grad_norm": 0.93636554479599,
+      "learning_rate": 7.102749348465166e-09,
+      "loss": 1.669,
+      "step": 3670
+    },
+    {
+      "epoch": 9.813333333333333,
+      "grad_norm": 1.0292489528656006,
+      "learning_rate": 5.457840221820831e-09,
+      "loss": 1.5997,
+      "step": 3680
+    },
+    {
+      "epoch": 9.84,
+      "grad_norm": 1.1855809688568115,
+      "learning_rate": 4.029073462325506e-09,
+      "loss": 1.6465,
+      "step": 3690
+    },
+    {
+      "epoch": 9.866666666666667,
+      "grad_norm": 0.9122327566146851,
+      "learning_rate": 2.8165728670573324e-09,
+      "loss": 1.591,
+      "step": 3700
+    },
+    {
+      "epoch": 9.893333333333333,
+      "grad_norm": 1.2053117752075195,
+      "learning_rate": 1.8204434944729676e-09,
+      "loss": 1.6682,
+      "step": 3710
+    },
+    {
+      "epoch": 9.92,
+      "grad_norm": 1.091283917427063,
+      "learning_rate": 1.0407716553040291e-09,
+      "loss": 1.6886,
+      "step": 3720
+    },
+    {
+      "epoch": 9.946666666666667,
+      "grad_norm": 0.9393439292907715,
+      "learning_rate": 4.77624905080576e-10,
+      "loss": 1.5919,
+      "step": 3730
+    },
+    {
+      "epoch": 9.973333333333333,
+      "grad_norm": 1.102495551109314,
+      "learning_rate": 1.3105203827634693e-10,
+      "loss": 1.7873,
+      "step": 3740
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 1.1831676959991455,
+      "learning_rate": 1.0830840810327482e-12,
+      "loss": 1.6486,
+      "step": 3750
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 3750,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.442270942438359e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}

Direction/Beaver-Danger/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2a9c00fc227fdce226a6e1d78baf7b9dc7392718c01908736a1625727530bd28
+size 5880

Direction/PKURLHF-10K_Safety/README.md

@@ -0,0 +1,202 @@
+---
+base_model: /datas/huggingface/Qwen3-8B
+library_name: peft
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.15.2

Direction/PKURLHF-10K_Safety/adapter_config.json

@@ -0,0 +1,39 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "/datas/huggingface/Qwen3-8B",
+  "bias": "none",
+  "corda_config": null,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 16,
+  "lora_bias": false,
+  "lora_dropout": 0.0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "q_proj",
+    "up_proj",
+    "gate_proj",
+    "k_proj",
+    "down_proj",
+    "o_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

Direction/PKURLHF-10K_Safety/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:379899c81f155487b924fdc186c6637795057834e4ae33fb91f8aa59a5c9c03e
+size 87360584

Direction/PKURLHF-10K_Safety/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
+}

Direction/PKURLHF-10K_Safety/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 %}

Direction/PKURLHF-10K_Safety/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4b2506283f0c8d04d91a1cb444336a19e1336b1a7ac5ca0f06755c310b6daa62
+size 175012683

Direction/PKURLHF-10K_Safety/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:01f9a0f7843a37be87edd23f4e88aa93b38b95cc2c07503eeb1cf2e4632453a2
+size 14645

Direction/PKURLHF-10K_Safety/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:91da9de8e9bdb9fbad87610906a20f7120bcf6de355ceee649e94c94e3f4ad25
+size 1465

Direction/PKURLHF-10K_Safety/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
+  }
+}

Direction/PKURLHF-10K_Safety/tokenizer.json

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

Direction/PKURLHF-10K_Safety/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
+}

Direction/PKURLHF-10K_Safety/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c50dc44de7f4ac5941a9f80302f3613586626bba2cfc7aae255c69e4a3432eec
+size 6225

README.md

@@ -1,3 +1,46 @@
----
-license: apache-2.0
----
+## Introduction
+
+This repository provides the weight files required for computing sample-level **SQSD** scores based on **Qwen3-8B**, as used in the paper **"From Parameter Dynamics to Risk Scoring: Quantifying Sample-Level Safety Degradation in LLM Fine-tuning"**.
+
+Two types of weights are needed to compute **SQSD**:
+
+- **Parameter shift direction weights (Direction)**: Encode safety-relevant directions in the model's parameter space, used to measure how individual fine-tuning samples affect model safety.
+- **Model initialization weights (initial-state)**: Serve as the starting point for SQSD computation. **Note**: These weights are only required when computing **Danger-Projection**. For details, please refer to **Section 4.3 Parameter Initialization** of the paper.
+
+## Links
+
+- Paper: https://arxiv.org/abs/2605.04572
+- GitHub: https://github.com/Jason-wx/SQSD
+
+## Directory Structure
+
+```
+hf_upload/
+├── Direction/               # Parameter shift direction weights
+│   ├── Ageis_Danger/        # Danger direction weights
+│   ├── Beaver-Danger/       # Danger direction weights
+│   └── PKURLHF-10K_Safety/  # Safety direction weights
+├── initial-state/           # Model initialization weights
+│   └── dolly_ckpt_5850/     # Initial weights (initialized via Danger-Projection)
+└── README.md
+```
+
+## Direction Folder
+
+The Direction folder contains three sets of direction weights, each extracted from a different dataset, encoding either a safety or danger direction in parameter space:
+
+| Name | Type | Description |
+|------|------|-------------|
+| Ageis_Danger | Danger | Danger direction weights extracted from the Aegis dataset |
+| Beaver-Danger | Danger | Danger direction weights extracted from the BeaverTails dataset |
+| PKURLHF-10K_Safety | Safety | Safety direction weights extracted from the PKU-RLHF dataset |
+
+These direction weights encode safety-relevant parameter shift directions and are a core dependency for computing SQSD scores.
+
+## initial-state Folder
+
+The weights in `initial-state` (`dolly_ckpt_5850`) represent the model initialization state derived via the **Danger-Projection** method — specifically, the parameter point obtained by projecting the base model weights along the danger direction. This serves as the reference starting point for subsequent SQSD computation.
+
+> ⚠️ The paper defines two initialization strategies depending on the projection direction (see **Section 4.3 Parameter Initialization**):
+> - **Danger direction** (drift-enhanced sensitivity): θ_initial = θ_t, initialized from a fine-tuning checkpoint that exhibits high directional sensitivity. The weights provided here (`dolly_ckpt_5850`) serve this purpose.
+> - **Safety direction** (linear-path sensitivity): θ_initial = θ_0 + α\*V_safety, initialized by interpolating from the base model along the safety direction vector. **No additional checkpoint is required** — only the base model weights and the safety direction weights from the `Direction` folder are needed.

initial-state/dolly_ckpt_5850/README.md

@@ -0,0 +1,202 @@
+---
+base_model: /datas/huggingface/Qwen3-8B
+library_name: peft
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.15.2

initial-state/dolly_ckpt_5850/adapter_config.json

@@ -0,0 +1,39 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "/datas/huggingface/Qwen3-8B",
+  "bias": "none",
+  "corda_config": null,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 16,
+  "lora_bias": false,
+  "lora_dropout": 0.0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "gate_proj",
+    "o_proj",
+    "q_proj",
+    "v_proj",
+    "up_proj",
+    "k_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_rslora": false
+}

initial-state/dolly_ckpt_5850/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1bc3091a9232876ccb3178f80a6f4db587910dbd5d06f68a1c35ad8f9c1c5563
+size 87360584

initial-state/dolly_ckpt_5850/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
+}

initial-state/dolly_ckpt_5850/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 %}

initial-state/dolly_ckpt_5850/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6af0d5ba04007eaac0243d7edd05961fb86105689ed8cf59c5fec806d9816cef
+size 175012683

initial-state/dolly_ckpt_5850/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c8707ffebfdb9126b866080be7f7c305b76f99b38d4f05102f060b97c427f18c
+size 14645

initial-state/dolly_ckpt_5850/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ae1b709f0013d3dc90d317a90857e87d052caed139bc77cec5ca4ccd7ebe2b80
+size 1465