Add qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5

.gitattributes

@@ -228,3 +228,8 @@ qwen3-4B-v1-if-partial-code_cont-legacy-system-1ep-muts-fnr-reflection-True-patc
 qwen3-4B-v1-if-partial-legacy-system-1ep-muts-fnr-reflection-True-patch-instruction-v1-classic-variant-legacy_lr_1e-06_bs_8_ga_8_pt_llm_muts_synth_muts_partial_synth_muts_git_understanding_as_muts_iim_system_checkpoint-2924/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 qwen3-4B-v4-if-partial-code_cont-legacy-system-1ep-muts-fnr-reflection-True-patch-instruction-v4-classic-variant-legacy-iim-system_lr_1e-06_bs_8_ga_8_pt_llm_muts_synth_muts_partial_synth_muts_git_understanding_as_muts_iim_system_checkpoint-2922/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 qwen3-4B-v4-if-partial-code_cont-legacy-system-2ep-muts-fnr-reflection-True-patch-instruction-v4-classic-variant-legacy-iim-system_lr_1e-06_bs_8_ga_8_ep_2_pt_llm_muts_synth_muts_partial_synth_muts_git_understanding_as_muts_iim_system_checkpoint-5844/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-1000/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-250/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-500/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-750/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/rewarded_completions.jsonl filter=lfs diff=lfs merge=lfs -text

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/README.md

@@ -0,0 +1,66 @@
+---
+library_name: transformers
+model_name: qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5
+tags:
+- generated_from_trainer
+- trl
+- grpo
+licence: license
+---
+
+# Model Card for qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5
+
+This model is a fine-tuned version of [None](https://huggingface.co/None).
+It has been trained using [TRL](https://github.com/huggingface/trl).
+
+## Quick start
+
+```python
+from transformers import pipeline
+
+question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
+generator = pipeline("text-generation", model="None", device="cuda")
+output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
+print(output["generated_text"])
+```
+
+## Training procedure
+
+[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://cohere.wandb.io/cohere/nathans-internship/runs/gf3tisqa) 
+
+
+
+This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
+
+### Framework versions
+
+- TRL: 1.3.0
+- Transformers: 4.57.6
+- Pytorch: 2.10.0+cu128
+- Datasets: 4.8.5
+- Tokenizers: 0.22.2
+
+## Citations
+
+Cite GRPO as:
+
+```bibtex
+@article{shao2024deepseekmath,
+    title        = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
+    author       = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
+    year         = 2024,
+    eprint       = {arXiv:2402.03300},
+}
+```
+
+Cite TRL as:
+    
+```bibtex
+@software{vonwerra2020trl,
+  title   = {{TRL: Transformers Reinforcement Learning}},
+  author  = {von Werra, Leandro and Belkada, Younes and Tunstall, Lewis and Beeching, Edward and Thrush, Tristan and Lambert, Nathan and Huang, Shengyi and Rasul, Kashif and Gallouédec, Quentin},
+  license = {Apache-2.0},
+  url     = {https://github.com/huggingface/trl},
+  year    = {2020}
+}
+```

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-1000/README.md

@@ -0,0 +1,209 @@
+---
+base_model: /root/repos/verl/new_models/qwen/qwen3-4B-v4-if-partial-code_cont-legacy-system-1ep-muts-fnr-reflection-True-patch-instruction-v4-classic-variant-legacy-iim-system_lr_1e-06_bs_8_ga_8_pt_llm_muts_synth_muts_partial_synth_muts_git_understanding_as_muts_iim_system_checkpoint-2922
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:/root/repos/verl/new_models/qwen/qwen3-4B-v4-if-partial-code_cont-legacy-system-1ep-muts-fnr-reflection-True-patch-instruction-v4-classic-variant-legacy-iim-system_lr_1e-06_bs_8_ga_8_pt_llm_muts_synth_muts_partial_synth_muts_git_understanding_as_muts_iim_system_checkpoint-2922
+- grpo
+- lora
+- transformers
+- trl
+---
+
+# 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 0.18.0

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-1000/adapter_config.json

@@ -0,0 +1,41 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "/root/repos/verl/new_models/qwen/qwen3-4B-v4-if-partial-code_cont-legacy-system-1ep-muts-fnr-reflection-True-patch-instruction-v4-classic-variant-legacy-iim-system_lr_1e-06_bs_8_ga_8_pt_llm_muts_synth_muts_partial_synth_muts_git_understanding_as_muts_iim_system_checkpoint-2922",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "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": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.0",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "v_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-1000/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3ede59ea521cebdd153c98bdabac533173386a278246b61c378b98220109f8a7
+size 23612264

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-1000/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
+}

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-1000/chat_template.jinja

@@ -0,0 +1,61 @@
+{%- 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 %}
+{%- 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" %}
+        {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- 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' }}
+{%- endif %}

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-1000/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:53f7ea6e0926860db568f67b35047067fa7c54403d0a866222062efb732c1b82
+size 47309579

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-1000/ref/adapter_config.json

@@ -0,0 +1,41 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "/root/repos/verl/new_models/qwen/qwen3-4B-v4-if-partial-code_cont-legacy-system-1ep-muts-fnr-reflection-True-patch-instruction-v4-classic-variant-legacy-iim-system_lr_1e-06_bs_8_ga_8_pt_llm_muts_synth_muts_partial_synth_muts_git_understanding_as_muts_iim_system_checkpoint-2922",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "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": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.0",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "v_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-1000/ref/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5b111c1638972db4a0a5619ba27132ed8cd78232ea1fdd336b96c7df928d1932
+size 11815792

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-1000/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:79306de783dedf356f21ea50f8def0b6ef22bf0e6b31fc9a0c56602236eba665
+size 15429

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-1000/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3f6908c372fb370be035d9e3446ef95acd845d2c67670ed86a53b7c33d27fe35
+size 15429

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-1000/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:81e1ab5885ea0b5cb31fe3736e6ce8ea8b7868fddc817f0b81928c09e42c02eb
+size 15429

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-1000/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:843c4387be0cbb3fd7d1d4a51b80d08aee55ddef3adf0a9c5ffa8a896ab25f63
+size 15365

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-1000/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4bfed19305e235c01f7b5a1dadc30bf1efa0576e8559c78b706a50a612005fd9
+size 1465

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-1000/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
+  }
+}

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-1000/tokenizer.json

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

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-1000/tokenizer_config.json

@@ -0,0 +1,239 @@
+{
+  "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": 1010000,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-1000/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:39626a1c4d97cde06662f17b3014e463f6a8f998ea251e23d77227c185b8a127
+size 8337

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-250/README.md

@@ -0,0 +1,209 @@
+---
+base_model: /root/repos/verl/new_models/qwen/qwen3-4B-v4-if-partial-code_cont-legacy-system-1ep-muts-fnr-reflection-True-patch-instruction-v4-classic-variant-legacy-iim-system_lr_1e-06_bs_8_ga_8_pt_llm_muts_synth_muts_partial_synth_muts_git_understanding_as_muts_iim_system_checkpoint-2922
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:/root/repos/verl/new_models/qwen/qwen3-4B-v4-if-partial-code_cont-legacy-system-1ep-muts-fnr-reflection-True-patch-instruction-v4-classic-variant-legacy-iim-system_lr_1e-06_bs_8_ga_8_pt_llm_muts_synth_muts_partial_synth_muts_git_understanding_as_muts_iim_system_checkpoint-2922
+- grpo
+- lora
+- transformers
+- trl
+---
+
+# 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 0.18.0

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-250/adapter_config.json

@@ -0,0 +1,41 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "/root/repos/verl/new_models/qwen/qwen3-4B-v4-if-partial-code_cont-legacy-system-1ep-muts-fnr-reflection-True-patch-instruction-v4-classic-variant-legacy-iim-system_lr_1e-06_bs_8_ga_8_pt_llm_muts_synth_muts_partial_synth_muts_git_understanding_as_muts_iim_system_checkpoint-2922",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "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": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.0",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "v_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-250/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:47db64de3da5d3eec5a2f11131c9b2e244ec4082d0ede494b5268adf03fd4f05
+size 23612264

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-250/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
+}

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-250/chat_template.jinja

@@ -0,0 +1,61 @@
+{%- 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 %}
+{%- 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" %}
+        {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- 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' }}
+{%- endif %}

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-250/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e6d83278b338744c41f3929480842746f569e4a80edb62150106406360bdce1f
+size 47309579

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-250/ref/adapter_config.json

@@ -0,0 +1,41 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "/root/repos/verl/new_models/qwen/qwen3-4B-v4-if-partial-code_cont-legacy-system-1ep-muts-fnr-reflection-True-patch-instruction-v4-classic-variant-legacy-iim-system_lr_1e-06_bs_8_ga_8_pt_llm_muts_synth_muts_partial_synth_muts_git_understanding_as_muts_iim_system_checkpoint-2922",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "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": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.0",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "v_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-250/ref/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5b111c1638972db4a0a5619ba27132ed8cd78232ea1fdd336b96c7df928d1932
+size 11815792

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-250/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0f591a7624037971b8954cc6d315d6d8ba72dd4aeba9b1cb3080eba9c4426db6
+size 15365

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-250/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b5896d7a9a237e18f0cefa4a29f3633c14092220fdb32c89f43b7cb37e5d8286
+size 15365

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-250/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2087f04ace980f03e3c1874df20da66a2846a98667b06feaa82252ed7cceead8
+size 15365

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-250/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b6ffaf3ddca2877d15bbb8308422c43e3d7bc7c0c78ced03cf01c570dd9ca9f2
+size 15365

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-250/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a20f50aeafb3dbc1a9811247df1ae3226b9a18969bf1359684550ab6ba528e96
+size 1465

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-250/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
+  }
+}

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-250/tokenizer.json

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

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-250/tokenizer_config.json

@@ -0,0 +1,239 @@
+{
+  "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": 1010000,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-250/trainer_state.json

@@ -0,0 +1,1212 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.05856172405715624,
+  "eval_steps": 50,
+  "global_step": 250,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": 0.51953125,
+      "completions/max_length": 768.0,
+      "completions/max_terminated_length": 128.0,
+      "completions/mean_length": 98.609375,
+      "completions/mean_terminated_length": 26.88617706298828,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "entropy": 0.35644982755184174,
+      "epoch": 0.00023424689622862497,
+      "frac_reward_zero_std": 0.34375,
+      "grad_norm": 0.5100324153900146,
+      "kl": 0.0,
+      "learning_rate": 0.0,
+      "loss": -0.0009,
+      "num_tokens": 322408.0,
+      "reward": 3.986328125,
+      "reward/patch_apply_success_pct_given_patch": 61.2275390625,
+      "reward/pct_prefill_correct_no_patch": 77.86458587646484,
+      "reward/pct_prefill_incorrect_with_patch": 69.86607360839844,
+      "reward/success_pct_prefill_correct": 74.47916412353516,
+      "reward/success_pct_prefill_incorrect": 44.01785659790039,
+      "reward_std": 8.538751602172852,
+      "rewards/oci_v2_harm_aware_reward_fn/mean": 3.986328125,
+      "rewards/oci_v2_harm_aware_reward_fn/std": 8.538751602172852,
+      "sampling/importance_sampling_ratio/max": 1.366683840751648,
+      "sampling/importance_sampling_ratio/mean": 0.49643030762672424,
+      "sampling/importance_sampling_ratio/min": 7.54401874303312e-09,
+      "sampling/sampling_logp_difference/max": 0.9534957408905029,
+      "sampling/sampling_logp_difference/mean": 0.024544060230255127,
+      "step": 1,
+      "step_time": 100.47959917550907
+    },
+    {
+      "clip_ratio/high_max": 0.0013864459680108768,
+      "clip_ratio/high_mean": 0.0006725406163039022,
+      "clip_ratio/low_mean": 0.00043979354935193743,
+      "clip_ratio/low_min": 3.110293821211801e-05,
+      "clip_ratio/region_mean": 0.0011123341650495098,
+      "completions/clipped_ratio": 0.5707465277777778,
+      "completions/max_length": 684.0,
+      "completions/max_terminated_length": 152.44444444444446,
+      "completions/mean_length": 115.08181423611111,
+      "completions/mean_terminated_length": 17.870292239718967,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "entropy": 0.3335921424958441,
+      "epoch": 0.0023424689622862497,
+      "frac_reward_zero_std": 0.2951388888888889,
+      "grad_norm": 0.447844922542572,
+      "kl": 0.0004705551647121966,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": -0.0002,
+      "num_tokens": 3380273.0,
+      "reward": 4.087022569444445,
+      "reward/patch_apply_success_pct_given_patch": 67.91800181070964,
+      "reward/pct_prefill_correct_no_patch": 83.95543840196397,
+      "reward/pct_prefill_incorrect_with_patch": 79.63603634304471,
+      "reward/success_pct_prefill_correct": 81.15740797254774,
+      "reward/success_pct_prefill_incorrect": 46.360367668999565,
+      "reward_std": 8.029874165852865,
+      "rewards/oci_v2_harm_aware_reward_fn/mean": 4.087022569444445,
+      "rewards/oci_v2_harm_aware_reward_fn/std": 8.029873900943333,
+      "sampling/importance_sampling_ratio/max": 1.52725268734826,
+      "sampling/importance_sampling_ratio/mean": 0.4651217857996623,
+      "sampling/importance_sampling_ratio/min": 4.9193822025015895e-06,
+      "sampling/sampling_logp_difference/max": 1.0352882146835327,
+      "sampling/sampling_logp_difference/mean": 0.023414300961626902,
+      "step": 10,
+      "step_time": 32.582850685498364
+    },
+    {
+      "clip_ratio/high_max": 0.0016945087958447403,
+      "clip_ratio/high_mean": 0.0008046950909374572,
+      "clip_ratio/low_mean": 0.0004977321016667701,
+      "clip_ratio/low_min": 7.139973422454204e-05,
+      "clip_ratio/region_mean": 0.0013024271991525893,
+      "completions/clipped_ratio": 0.6109375,
+      "completions/max_length": 722.5,
+      "completions/max_terminated_length": 220.8,
+      "completions/mean_length": 119.9419921875,
+      "completions/mean_terminated_length": 20.239689111709595,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "entropy": 0.35104990415275095,
+      "epoch": 0.004684937924572499,
+      "frac_reward_zero_std": 0.28125,
+      "grad_norm": 1.3310966491699219,
+      "kl": 0.0005461559168907116,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": -0.0055,
+      "num_tokens": 6740104.0,
+      "reward": 3.4971516847610475,
+      "reward/patch_apply_success_pct_given_patch": 65.30590858459473,
+      "reward/pct_prefill_correct_no_patch": 77.03385620117187,
+      "reward/pct_prefill_incorrect_with_patch": 79.48735313415527,
+      "reward/success_pct_prefill_correct": 73.99739685058594,
+      "reward/success_pct_prefill_incorrect": 47.143972396850586,
+      "reward_std": 8.33356432914734,
+      "rewards/oci_v2_harm_aware_reward_fn/mean": 3.4971516847610475,
+      "rewards/oci_v2_harm_aware_reward_fn/std": 8.333564186096192,
+      "sampling/importance_sampling_ratio/max": 1.4482219576835633,
+      "sampling/importance_sampling_ratio/mean": 0.4283165752887726,
+      "sampling/importance_sampling_ratio/min": 1.6469782855992807e-08,
+      "sampling/sampling_logp_difference/max": 0.8080385684967041,
+      "sampling/sampling_logp_difference/mean": 0.024413819052278997,
+      "step": 20,
+      "step_time": 33.25843693334609
+    },
+    {
+      "clip_ratio/high_max": 0.0015233877369610128,
+      "clip_ratio/high_mean": 0.0007453550670106779,
+      "clip_ratio/low_mean": 0.0005168871712157852,
+      "clip_ratio/low_min": 4.8378223800682464e-05,
+      "clip_ratio/region_mean": 0.0012622422393178568,
+      "completions/clipped_ratio": 0.6298828125,
+      "completions/max_length": 768.0,
+      "completions/max_terminated_length": 240.3,
+      "completions/mean_length": 128.2712890625,
+      "completions/mean_terminated_length": 24.723235177993775,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "entropy": 0.35342266485095025,
+      "epoch": 0.007027406886858749,
+      "frac_reward_zero_std": 0.265625,
+      "grad_norm": 0.5314885973930359,
+      "kl": 0.0005847927179274848,
+      "learning_rate": 2.9e-06,
+      "loss": 0.0133,
+      "num_tokens": 10234725.0,
+      "reward": 3.5868791580200194,
+      "reward/patch_apply_success_pct_given_patch": 62.882485961914064,
+      "reward/pct_prefill_correct_no_patch": 75.41927032470703,
+      "reward/pct_prefill_incorrect_with_patch": 80.59635391235352,
+      "reward/success_pct_prefill_correct": 71.62239608764648,
+      "reward/success_pct_prefill_incorrect": 45.64267120361328,
+      "reward_std": 8.274088811874389,
+      "rewards/oci_v2_harm_aware_reward_fn/mean": 3.5868791580200194,
+      "rewards/oci_v2_harm_aware_reward_fn/std": 8.274088430404664,
+      "sampling/importance_sampling_ratio/max": 1.6922070980072021,
+      "sampling/importance_sampling_ratio/mean": 0.41347462832927706,
+      "sampling/importance_sampling_ratio/min": 4.062776376690251e-06,
+      "sampling/sampling_logp_difference/max": 1.0166505098342895,
+      "sampling/sampling_logp_difference/mean": 0.024688040465116502,
+      "step": 30,
+      "step_time": 33.097395395627245
+    },
+    {
+      "clip_ratio/high_max": 0.0015834795682167168,
+      "clip_ratio/high_mean": 0.0007902724177256459,
+      "clip_ratio/low_mean": 0.0004727362485937192,
+      "clip_ratio/low_min": 2.8637799550779165e-05,
+      "clip_ratio/region_mean": 0.0012630086550416309,
+      "completions/clipped_ratio": 0.64609375,
+      "completions/max_length": 768.0,
+      "completions/max_terminated_length": 188.4,
+      "completions/mean_length": 126.687109375,
+      "completions/mean_terminated_length": 22.418661642074586,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "entropy": 0.3591292887926102,
+      "epoch": 0.009369875849144999,
+      "frac_reward_zero_std": 0.23125,
+      "grad_norm": 1.2263091802597046,
+      "kl": 0.0008070670657616575,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 0.0033,
+      "num_tokens": 13632051.0,
+      "reward": 3.718359375,
+      "reward/patch_apply_success_pct_given_patch": 68.43336563110351,
+      "reward/pct_prefill_correct_no_patch": 75.09375076293945,
+      "reward/pct_prefill_incorrect_with_patch": 80.73809776306152,
+      "reward/success_pct_prefill_correct": 72.55989646911621,
+      "reward/success_pct_prefill_incorrect": 49.25576667785644,
+      "reward_std": 8.253158855438233,
+      "rewards/oci_v2_harm_aware_reward_fn/mean": 3.718359375,
+      "rewards/oci_v2_harm_aware_reward_fn/std": 8.253158712387085,
+      "sampling/importance_sampling_ratio/max": 1.8009993076324462,
+      "sampling/importance_sampling_ratio/mean": 0.40669935643672944,
+      "sampling/importance_sampling_ratio/min": 4.5407724965067925e-10,
+      "sampling/sampling_logp_difference/max": 1.0855147659778595,
+      "sampling/sampling_logp_difference/mean": 0.025073571503162383,
+      "step": 40,
+      "step_time": 35.433205044595525
+    },
+    {
+      "clip_ratio/high_max": 0.0015883201300312066,
+      "clip_ratio/high_mean": 0.000773158567790233,
+      "clip_ratio/low_mean": 0.0005582512449109345,
+      "clip_ratio/low_min": 7.383882475551218e-05,
+      "clip_ratio/region_mean": 0.0013314098079717951,
+      "completions/clipped_ratio": 0.6916015625,
+      "completions/max_length": 768.0,
+      "completions/max_terminated_length": 193.6,
+      "completions/mean_length": 133.9275390625,
+      "completions/mean_terminated_length": 21.529980659484863,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "entropy": 0.3748301725834608,
+      "epoch": 0.011712344811431248,
+      "frac_reward_zero_std": 0.190625,
+      "grad_norm": 0.9505739808082581,
+      "kl": 0.0012336442589003128,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": -0.0539,
+      "num_tokens": 17056528.0,
+      "reward": 3.9427083015441893,
+      "reward/patch_apply_success_pct_given_patch": 66.34924964904785,
+      "reward/pct_prefill_correct_no_patch": 66.5703125,
+      "reward/pct_prefill_incorrect_with_patch": 88.64806594848633,
+      "reward/success_pct_prefill_correct": 65.23698005676269,
+      "reward/success_pct_prefill_incorrect": 53.74590721130371,
+      "reward_std": 8.083413982391358,
+      "rewards/oci_v2_harm_aware_reward_fn/mean": 3.9427083015441893,
+      "rewards/oci_v2_harm_aware_reward_fn/std": 8.083413934707641,
+      "sampling/importance_sampling_ratio/max": 1.737173867225647,
+      "sampling/importance_sampling_ratio/mean": 0.36960968673229216,
+      "sampling/importance_sampling_ratio/min": 2.6032340618981146e-09,
+      "sampling/sampling_logp_difference/max": 0.8185921430587768,
+      "sampling/sampling_logp_difference/mean": 0.026036406867206097,
+      "step": 50,
+      "step_time": 36.05194646338932
+    },
+    {
+      "epoch": 0.011712344811431248,
+      "eval_clip_ratio/high_max": 0.0,
+      "eval_clip_ratio/high_mean": 0.0,
+      "eval_clip_ratio/low_mean": 0.0,
+      "eval_clip_ratio/low_min": 0.0,
+      "eval_clip_ratio/region_mean": 0.0,
+      "eval_completions/clipped_ratio": 0.5804476351351351,
+      "eval_completions/max_length": 604.8108108108108,
+      "eval_completions/max_terminated_length": 115.32432432432432,
+      "eval_completions/mean_length": 148.9596706081081,
+      "eval_completions/mean_terminated_length": 43.3292588607685,
+      "eval_completions/min_length": 7.8108108108108105,
+      "eval_completions/min_terminated_length": 8.094594594594595,
+      "eval_entropy": 0.37274140441739884,
+      "eval_frac_reward_zero_std": 0.10472972972972973,
+      "eval_kl": 0.0017632326336162216,
+      "eval_loss": -0.15783905982971191,
+      "eval_num_tokens": 17056528.0,
+      "eval_reward": 3.0468046226211496,
+      "eval_reward/patch_apply_success_pct_given_patch": 37.25176849880734,
+      "eval_reward/pct_prefill_correct_no_patch": 39.14695945945946,
+      "eval_reward/pct_prefill_incorrect_with_patch": 28.969594594594593,
+      "eval_reward/success_pct_prefill_correct": 38.7035472972973,
+      "eval_reward/success_pct_prefill_incorrect": 14.949324324324325,
+      "eval_reward_std": 7.969465816343153,
+      "eval_rewards/oci_v2_harm_aware_reward_fn/mean": 3.0468046226211496,
+      "eval_rewards/oci_v2_harm_aware_reward_fn/std": 7.969466061205478,
+      "eval_runtime": 298.5492,
+      "eval_samples_per_second": 0.988,
+      "eval_sampling/importance_sampling_ratio/max": 1.104561196791159,
+      "eval_sampling/importance_sampling_ratio/mean": 0.4117495620975623,
+      "eval_sampling/importance_sampling_ratio/min": 0.00010188482712453056,
+      "eval_sampling/sampling_logp_difference/max": 0.5493728445993887,
+      "eval_sampling/sampling_logp_difference/mean": 0.02538555135598054,
+      "eval_steps_per_second": 0.017,
+      "step": 50
+    },
+    {
+      "clip_ratio/high_max": 0.0014512597961584106,
+      "clip_ratio/high_mean": 0.0007325455468162545,
+      "clip_ratio/low_mean": 0.0004296011052247195,
+      "clip_ratio/low_min": 2.8956899404875004e-05,
+      "clip_ratio/region_mean": 0.001162146647038753,
+      "completions/clipped_ratio": 0.7052734375,
+      "completions/max_length": 768.0,
+      "completions/max_terminated_length": 238.8,
+      "completions/mean_length": 141.196484375,
+      "completions/mean_terminated_length": 23.717469787597658,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "entropy": 0.3621661901473999,
+      "epoch": 0.014054813773717497,
+      "frac_reward_zero_std": 0.20625,
+      "grad_norm": 0.280788779258728,
+      "kl": 0.0014432421630772296,
+      "learning_rate": 5e-06,
+      "loss": -0.0471,
+      "num_tokens": 20528142.0,
+      "reward": 3.80390625,
+      "reward/patch_apply_success_pct_given_patch": 64.77769546508789,
+      "reward/pct_prefill_correct_no_patch": 62.50520782470703,
+      "reward/pct_prefill_incorrect_with_patch": 87.16257400512696,
+      "reward/success_pct_prefill_correct": 59.739583587646486,
+      "reward/success_pct_prefill_incorrect": 51.78962020874023,
+      "reward_std": 8.043180799484253,
+      "rewards/oci_v2_harm_aware_reward_fn/mean": 3.80390625,
+      "rewards/oci_v2_harm_aware_reward_fn/std": 8.043180561065673,
+      "sampling/importance_sampling_ratio/max": 1.6564959287643433,
+      "sampling/importance_sampling_ratio/mean": 0.3541590839624405,
+      "sampling/importance_sampling_ratio/min": 2.3992926228984417e-09,
+      "sampling/sampling_logp_difference/max": 0.9005755543708801,
+      "sampling/sampling_logp_difference/mean": 0.02513938508927822,
+      "step": 60,
+      "step_time": 36.51428347225301
+    },
+    {
+      "clip_ratio/high_max": 0.0015364812712505226,
+      "clip_ratio/high_mean": 0.0007954608877298597,
+      "clip_ratio/low_mean": 0.000561387133984681,
+      "clip_ratio/low_min": 0.00012351891564321704,
+      "clip_ratio/region_mean": 0.001356848014256684,
+      "completions/clipped_ratio": 0.7416015625,
+      "completions/max_length": 768.0,
+      "completions/max_terminated_length": 188.9,
+      "completions/mean_length": 150.3443359375,
+      "completions/mean_terminated_length": 34.43300151824951,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "entropy": 0.36352439820766447,
+      "epoch": 0.016397282736003747,
+      "frac_reward_zero_std": 0.18125,
+      "grad_norm": 0.8723386526107788,
+      "kl": 0.0015955766692059114,
+      "learning_rate": 5e-06,
+      "loss": -0.0594,
+      "num_tokens": 24052993.0,
+      "reward": 3.941650390625,
+      "reward/patch_apply_success_pct_given_patch": 66.87326087951661,
+      "reward/pct_prefill_correct_no_patch": 60.22842254638672,
+      "reward/pct_prefill_incorrect_with_patch": 91.54706039428712,
+      "reward/success_pct_prefill_correct": 57.530133438110354,
+      "reward/success_pct_prefill_incorrect": 52.73381652832031,
+      "reward_std": 7.884973955154419,
+      "rewards/oci_v2_harm_aware_reward_fn/mean": 3.941650390625,
+      "rewards/oci_v2_harm_aware_reward_fn/std": 7.884973812103271,
+      "sampling/importance_sampling_ratio/max": 1.7976397395133972,
+      "sampling/importance_sampling_ratio/mean": 0.3075092852115631,
+      "sampling/importance_sampling_ratio/min": 2.499234478492674e-10,
+      "sampling/sampling_logp_difference/max": 1.4041473269462585,
+      "sampling/sampling_logp_difference/mean": 0.025211768597364424,
+      "step": 70,
+      "step_time": 35.41440772935748
+    },
+    {
+      "clip_ratio/high_max": 0.001587765475778724,
+      "clip_ratio/high_mean": 0.0008066393937951943,
+      "clip_ratio/low_mean": 0.0005343625396562856,
+      "clip_ratio/low_min": 7.995337655302137e-05,
+      "clip_ratio/region_mean": 0.001341001929176855,
+      "completions/clipped_ratio": 0.7068359375,
+      "completions/max_length": 768.0,
+      "completions/max_terminated_length": 174.2,
+      "completions/mean_length": 138.1626953125,
+      "completions/mean_terminated_length": 17.698761558532716,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "entropy": 0.37485333159565926,
+      "epoch": 0.018739751698289998,
+      "frac_reward_zero_std": 0.146875,
+      "grad_norm": 0.5849360823631287,
+      "kl": 0.0015494659703108482,
+      "learning_rate": 5e-06,
+      "loss": -0.0632,
+      "num_tokens": 27552962.0,
+      "reward": 3.808837890625,
+      "reward/patch_apply_success_pct_given_patch": 67.21000442504882,
+      "reward/pct_prefill_correct_no_patch": 67.60677032470703,
+      "reward/pct_prefill_incorrect_with_patch": 88.28738937377929,
+      "reward/success_pct_prefill_correct": 65.94791717529297,
+      "reward/success_pct_prefill_incorrect": 49.17410774230957,
+      "reward_std": 7.960854721069336,
+      "rewards/oci_v2_harm_aware_reward_fn/mean": 3.808837890625,
+      "rewards/oci_v2_harm_aware_reward_fn/std": 7.960854578018188,
+      "sampling/importance_sampling_ratio/max": 1.6032965064048768,
+      "sampling/importance_sampling_ratio/mean": 0.3500533729791641,
+      "sampling/importance_sampling_ratio/min": 1.13169827123303e-07,
+      "sampling/sampling_logp_difference/max": 0.9364623248577117,
+      "sampling/sampling_logp_difference/mean": 0.026069757528603077,
+      "step": 80,
+      "step_time": 37.47133123772219
+    },
+    {
+      "clip_ratio/high_max": 0.0016020891736843624,
+      "clip_ratio/high_mean": 0.0007940935469378019,
+      "clip_ratio/low_mean": 0.00045407180650727243,
+      "clip_ratio/low_min": 4.186248988844454e-05,
+      "clip_ratio/region_mean": 0.0012481653444410767,
+      "completions/clipped_ratio": 0.6994140625,
+      "completions/max_length": 768.0,
+      "completions/max_terminated_length": 186.0,
+      "completions/mean_length": 136.425,
+      "completions/mean_terminated_length": 21.6198561668396,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "entropy": 0.3525923781096935,
+      "epoch": 0.02108222066057625,
+      "frac_reward_zero_std": 0.265625,
+      "grad_norm": 0.46370458602905273,
+      "kl": 0.001235544477822259,
+      "learning_rate": 5e-06,
+      "loss": -0.0369,
+      "num_tokens": 30982786.0,
+      "reward": 4.470458984375,
+      "reward/patch_apply_success_pct_given_patch": 68.86532058715821,
+      "reward/pct_prefill_correct_no_patch": 65.92708396911621,
+      "reward/pct_prefill_incorrect_with_patch": 87.17076034545899,
+      "reward/success_pct_prefill_correct": 65.73697967529297,
+      "reward/success_pct_prefill_incorrect": 56.66871223449707,
+      "reward_std": 7.92942180633545,
+      "rewards/oci_v2_harm_aware_reward_fn/mean": 4.470458984375,
+      "rewards/oci_v2_harm_aware_reward_fn/std": 7.92942180633545,
+      "sampling/importance_sampling_ratio/max": 1.6725685358047486,
+      "sampling/importance_sampling_ratio/mean": 0.35846699178218844,
+      "sampling/importance_sampling_ratio/min": 1.5536687862811627e-06,
+      "sampling/sampling_logp_difference/max": 1.0088762521743775,
+      "sampling/sampling_logp_difference/mean": 0.024435398168861866,
+      "step": 90,
+      "step_time": 33.930982740363106
+    },
+    {
+      "clip_ratio/high_max": 0.0016140311887284041,
+      "clip_ratio/high_mean": 0.0008042901145017822,
+      "clip_ratio/low_mean": 0.0005069966619885236,
+      "clip_ratio/low_min": 6.70985180477146e-05,
+      "clip_ratio/region_mean": 0.0013112867614836431,
+      "completions/clipped_ratio": 0.6908203125,
+      "completions/max_length": 768.0,
+      "completions/max_terminated_length": 178.3,
+      "completions/mean_length": 133.159765625,
+      "completions/mean_terminated_length": 19.256701946258545,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "entropy": 0.3675391260534525,
+      "epoch": 0.023424689622862496,
+      "frac_reward_zero_std": 0.2,
+      "grad_norm": 0.5126844048500061,
+      "kl": 0.0011453866405645385,
+      "learning_rate": 5e-06,
+      "loss": -0.0151,
+      "num_tokens": 34442340.0,
+      "reward": 3.9337890625,
+      "reward/patch_apply_success_pct_given_patch": 65.84646682739258,
+      "reward/pct_prefill_correct_no_patch": 72.34114608764648,
+      "reward/pct_prefill_incorrect_with_patch": 87.59970092773438,
+      "reward/success_pct_prefill_correct": 71.14062423706055,
+      "reward/success_pct_prefill_incorrect": 50.96205444335938,
+      "reward_std": 8.063340091705323,
+      "rewards/oci_v2_harm_aware_reward_fn/mean": 3.9337890625,
+      "rewards/oci_v2_harm_aware_reward_fn/std": 8.063339853286744,
+      "sampling/importance_sampling_ratio/max": 1.6369470715522767,
+      "sampling/importance_sampling_ratio/mean": 0.3648974269628525,
+      "sampling/importance_sampling_ratio/min": 1.0455670625957452e-09,
+      "sampling/sampling_logp_difference/max": 1.496512633562088,
+      "sampling/sampling_logp_difference/mean": 0.025561144202947618,
+      "step": 100,
+      "step_time": 37.93434434947558
+    },
+    {
+      "epoch": 0.023424689622862496,
+      "eval_clip_ratio/high_max": 0.0,
+      "eval_clip_ratio/high_mean": 0.0,
+      "eval_clip_ratio/low_mean": 0.0,
+      "eval_clip_ratio/low_min": 0.0,
+      "eval_clip_ratio/region_mean": 0.0,
+      "eval_completions/clipped_ratio": 0.4858530405405405,
+      "eval_completions/max_length": 545.2027027027027,
+      "eval_completions/max_terminated_length": 105.82432432432432,
+      "eval_completions/mean_length": 127.98944256756756,
+      "eval_completions/mean_terminated_length": 40.66666072768134,
+      "eval_completions/min_length": 8.04054054054054,
+      "eval_completions/min_terminated_length": 8.04054054054054,
+      "eval_entropy": 0.3296805012065011,
+      "eval_frac_reward_zero_std": 0.18243243243243243,
+      "eval_kl": 0.001050714309786673,
+      "eval_loss": -0.12259963899850845,
+      "eval_num_tokens": 34442340.0,
+      "eval_reward": 4.378730290644878,
+      "eval_reward/patch_apply_success_pct_given_patch": 36.607088826798105,
+      "eval_reward/pct_prefill_correct_no_patch": 47.508445945945944,
+      "eval_reward/pct_prefill_incorrect_with_patch": 28.399493243243242,
+      "eval_reward/success_pct_prefill_correct": 46.19932432432432,
+      "eval_reward/success_pct_prefill_incorrect": 15.202702702702704,
+      "eval_reward_std": 7.38360092768798,
+      "eval_rewards/oci_v2_harm_aware_reward_fn/mean": 4.378730290644878,
+      "eval_rewards/oci_v2_harm_aware_reward_fn/std": 7.383601101669106,
+      "eval_runtime": 255.0529,
+      "eval_samples_per_second": 1.157,
+      "eval_sampling/importance_sampling_ratio/max": 1.0465183572189227,
+      "eval_sampling/importance_sampling_ratio/mean": 0.4802864554363328,
+      "eval_sampling/importance_sampling_ratio/min": 0.0002676908223747319,
+      "eval_sampling/sampling_logp_difference/max": 0.4904198441151026,
+      "eval_sampling/sampling_logp_difference/mean": 0.0228210183842158,
+      "eval_steps_per_second": 0.02,
+      "step": 100
+    },
+    {
+      "clip_ratio/high_max": 0.0015675041857321048,
+      "clip_ratio/high_mean": 0.0007827833462215495,
+      "clip_ratio/low_mean": 0.00046802288024991865,
+      "clip_ratio/low_min": 2.960036290460266e-05,
+      "clip_ratio/region_mean": 0.0012508062263805186,
+      "completions/clipped_ratio": 0.678125,
+      "completions/max_length": 747.2,
+      "completions/max_terminated_length": 180.6,
+      "completions/mean_length": 127.1375,
+      "completions/mean_terminated_length": 19.610678672790527,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "entropy": 0.36924048140645027,
+      "epoch": 0.025767158585148747,
+      "frac_reward_zero_std": 0.21875,
+      "grad_norm": 0.12973232567310333,
+      "kl": 0.0013718604743189645,
+      "learning_rate": 5e-06,
+      "loss": -0.049,
+      "num_tokens": 37783732.0,
+      "reward": 4.370556640625,
+      "reward/patch_apply_success_pct_given_patch": 68.6411849975586,
+      "reward/pct_prefill_correct_no_patch": 67.61979255676269,
+      "reward/pct_prefill_incorrect_with_patch": 87.06994171142578,
+      "reward/success_pct_prefill_correct": 65.94791526794434,
+      "reward/success_pct_prefill_incorrect": 54.81733589172363,
+      "reward_std": 7.996706628799439,
+      "rewards/oci_v2_harm_aware_reward_fn/mean": 4.370556640625,
+      "rewards/oci_v2_harm_aware_reward_fn/std": 7.996706533432007,
+      "sampling/importance_sampling_ratio/max": 1.4522894620895386,
+      "sampling/importance_sampling_ratio/mean": 0.367797189950943,
+      "sampling/importance_sampling_ratio/min": 1.030247234340092e-09,
+      "sampling/sampling_logp_difference/max": 1.582759302854538,
+      "sampling/sampling_logp_difference/mean": 0.025711523182690144,
+      "step": 110,
+      "step_time": 35.57631639386527
+    },
+    {
+      "clip_ratio/high_max": 0.0015772733189805877,
+      "clip_ratio/high_mean": 0.000799956577429839,
+      "clip_ratio/low_mean": 0.00047146836595857167,
+      "clip_ratio/low_min": 6.006203984725289e-05,
+      "clip_ratio/region_mean": 0.0012714249431155622,
+      "completions/clipped_ratio": 0.697265625,
+      "completions/max_length": 742.9,
+      "completions/max_terminated_length": 187.0,
+      "completions/mean_length": 138.6630859375,
+      "completions/mean_terminated_length": 25.954986667633058,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "entropy": 0.35178216733038425,
+      "epoch": 0.028109627547434995,
+      "frac_reward_zero_std": 0.2625,
+      "grad_norm": 0.2743445932865143,
+      "kl": 0.0013355807575862854,
+      "learning_rate": 5e-06,
+      "loss": -0.0335,
+      "num_tokens": 41259655.0,
+      "reward": 4.68681640625,
+      "reward/patch_apply_success_pct_given_patch": 68.90146293640137,
+      "reward/pct_prefill_correct_no_patch": 75.35416793823242,
+      "reward/pct_prefill_incorrect_with_patch": 93.25613861083984,
+      "reward/success_pct_prefill_correct": 71.31250076293945,
+      "reward/success_pct_prefill_incorrect": 52.736793518066406,
+      "reward_std": 7.5017632961273195,
+      "rewards/oci_v2_harm_aware_reward_fn/mean": 4.68681640625,
+      "rewards/oci_v2_harm_aware_reward_fn/std": 7.501763105392456,
+      "sampling/importance_sampling_ratio/max": 1.285536026954651,
+      "sampling/importance_sampling_ratio/mean": 0.34532674252986906,
+      "sampling/importance_sampling_ratio/min": 5.670053868484995e-07,
+      "sampling/sampling_logp_difference/max": 1.3240800976753235,
+      "sampling/sampling_logp_difference/mean": 0.024372128956019878,
+      "step": 120,
+      "step_time": 34.42521113022231
+    },
+    {
+      "clip_ratio/high_max": 0.00171116842830088,
+      "clip_ratio/high_mean": 0.0008645262596473913,
+      "clip_ratio/low_mean": 0.0004899012997157115,
+      "clip_ratio/low_min": 6.280017660174054e-05,
+      "clip_ratio/region_mean": 0.0013544275538151852,
+      "completions/clipped_ratio": 0.7111328125,
+      "completions/max_length": 717.3,
+      "completions/max_terminated_length": 108.5,
+      "completions/mean_length": 132.801953125,
+      "completions/mean_terminated_length": 14.554699468612672,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "entropy": 0.36489393226802347,
+      "epoch": 0.030452096509721246,
+      "frac_reward_zero_std": 0.303125,
+      "grad_norm": 0.4577244520187378,
+      "kl": 0.0018485030472220388,
+      "learning_rate": 5e-06,
+      "loss": 0.0014,
+      "num_tokens": 44676881.0,
+      "reward": 4.101611328125,
+      "reward/patch_apply_success_pct_given_patch": 70.61904449462891,
+      "reward/pct_prefill_correct_no_patch": 69.28385314941406,
+      "reward/pct_prefill_incorrect_with_patch": 85.85751419067383,
+      "reward/success_pct_prefill_correct": 68.73697814941406,
+      "reward/success_pct_prefill_incorrect": 52.59319152832031,
+      "reward_std": 8.013053703308106,
+      "rewards/oci_v2_harm_aware_reward_fn/mean": 4.101611328125,
+      "rewards/oci_v2_harm_aware_reward_fn/std": 8.013053369522094,
+      "sampling/importance_sampling_ratio/max": 1.4449580669403077,
+      "sampling/importance_sampling_ratio/mean": 0.3434909701347351,
+      "sampling/importance_sampling_ratio/min": 2.563018692793923e-07,
+      "sampling/sampling_logp_difference/max": 1.2057989835739136,
+      "sampling/sampling_logp_difference/mean": 0.02540039997547865,
+      "step": 130,
+      "step_time": 33.50838142563589
+    },
+    {
+      "clip_ratio/high_max": 0.0016597176749201027,
+      "clip_ratio/high_mean": 0.0008639390482130694,
+      "clip_ratio/low_mean": 0.0005017756877350621,
+      "clip_ratio/low_min": 5.262866470729932e-05,
+      "clip_ratio/region_mean": 0.001365714739949908,
+      "completions/clipped_ratio": 0.7451171875,
+      "completions/max_length": 763.4,
+      "completions/max_terminated_length": 239.2,
+      "completions/mean_length": 149.3142578125,
+      "completions/mean_terminated_length": 21.67807569503784,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "entropy": 0.3566353913396597,
+      "epoch": 0.03279456547200749,
+      "frac_reward_zero_std": 0.221875,
+      "grad_norm": 0.28173503279685974,
+      "kl": 0.0018222495578811503,
+      "learning_rate": 5e-06,
+      "loss": -0.0364,
+      "num_tokens": 48221770.0,
+      "reward": 4.205144715309143,
+      "reward/patch_apply_success_pct_given_patch": 71.62333374023437,
+      "reward/pct_prefill_correct_no_patch": 56.796875381469725,
+      "reward/pct_prefill_incorrect_with_patch": 87.66480865478516,
+      "reward/success_pct_prefill_correct": 56.54947967529297,
+      "reward/success_pct_prefill_incorrect": 52.460380172729494,
+      "reward_std": 7.8182862281799315,
+      "rewards/oci_v2_harm_aware_reward_fn/mean": 4.205144715309143,
+      "rewards/oci_v2_harm_aware_reward_fn/std": 7.8182861328125,
+      "sampling/importance_sampling_ratio/max": 1.2473852515220643,
+      "sampling/importance_sampling_ratio/mean": 0.3019875317811966,
+      "sampling/importance_sampling_ratio/min": 1.2609959041450809e-09,
+      "sampling/sampling_logp_difference/max": 1.7118031144142152,
+      "sampling/sampling_logp_difference/mean": 0.024876235984265804,
+      "step": 140,
+      "step_time": 40.67678426941857
+    },
+    {
+      "clip_ratio/high_max": 0.0014939031188987428,
+      "clip_ratio/high_mean": 0.0007943237862491515,
+      "clip_ratio/low_mean": 0.0004927984671667218,
+      "clip_ratio/low_min": 3.9076757457223724e-05,
+      "clip_ratio/region_mean": 0.0012871222505054902,
+      "completions/clipped_ratio": 0.7115234375,
+      "completions/max_length": 749.3,
+      "completions/max_terminated_length": 155.1,
+      "completions/mean_length": 136.7783203125,
+      "completions/mean_terminated_length": 17.208456230163574,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "entropy": 0.3674771565943956,
+      "epoch": 0.035137034434293744,
+      "frac_reward_zero_std": 0.253125,
+      "grad_norm": 0.4556274116039276,
+      "kl": 0.0021282700858137106,
+      "learning_rate": 5e-06,
+      "loss": -0.0296,
+      "num_tokens": 51686411.0,
+      "reward": 3.927197265625,
+      "reward/patch_apply_success_pct_given_patch": 68.90301208496093,
+      "reward/pct_prefill_correct_no_patch": 70.19791564941406,
+      "reward/pct_prefill_incorrect_with_patch": 88.14267272949219,
+      "reward/success_pct_prefill_correct": 68.83854064941406,
+      "reward/success_pct_prefill_incorrect": 49.61960563659668,
+      "reward_std": 8.007440185546875,
+      "rewards/oci_v2_harm_aware_reward_fn/mean": 3.927197265625,
+      "rewards/oci_v2_harm_aware_reward_fn/std": 8.007440185546875,
+      "sampling/importance_sampling_ratio/max": 1.3250896513462067,
+      "sampling/importance_sampling_ratio/mean": 0.32481487691402433,
+      "sampling/importance_sampling_ratio/min": 7.723281968237336e-09,
+      "sampling/sampling_logp_difference/max": 1.3617030501365661,
+      "sampling/sampling_logp_difference/mean": 0.02578139118850231,
+      "step": 150,
+      "step_time": 34.89231479833834
+    },
+    {
+      "epoch": 0.035137034434293744,
+      "eval_clip_ratio/high_max": 0.0,
+      "eval_clip_ratio/high_mean": 0.0,
+      "eval_clip_ratio/low_mean": 0.0,
+      "eval_clip_ratio/low_min": 0.0,
+      "eval_clip_ratio/region_mean": 0.0,
+      "eval_completions/clipped_ratio": 0.5021114864864865,
+      "eval_completions/max_length": 564.7297297297297,
+      "eval_completions/max_terminated_length": 102.79729729729729,
+      "eval_completions/mean_length": 131.55679898648648,
+      "eval_completions/mean_terminated_length": 39.13790236292659,
+      "eval_completions/min_length": 9.68918918918919,
+      "eval_completions/min_terminated_length": 9.243243243243244,
+      "eval_entropy": 0.33011221019802867,
+      "eval_frac_reward_zero_std": 0.17229729729729729,
+      "eval_kl": 0.001551323342199966,
+      "eval_loss": -0.11743374168872833,
+      "eval_num_tokens": 51686411.0,
+      "eval_reward": 4.191278882928796,
+      "eval_reward/patch_apply_success_pct_given_patch": 34.181259970407226,
+      "eval_reward/pct_prefill_correct_no_patch": 46.5160472972973,
+      "eval_reward/pct_prefill_incorrect_with_patch": 28.65287162162162,
+      "eval_reward/success_pct_prefill_correct": 44.9535472972973,
+      "eval_reward/success_pct_prefill_incorrect": 15.266047297297296,
+      "eval_reward_std": 7.419620352822381,
+      "eval_rewards/oci_v2_harm_aware_reward_fn/mean": 4.191278882928796,
+      "eval_rewards/oci_v2_harm_aware_reward_fn/std": 7.419620539690997,
+      "eval_runtime": 261.4693,
+      "eval_samples_per_second": 1.128,
+      "eval_sampling/importance_sampling_ratio/max": 1.0105714910739176,
+      "eval_sampling/importance_sampling_ratio/mean": 0.4672904191286983,
+      "eval_sampling/importance_sampling_ratio/min": 0.00022436127598750633,
+      "eval_sampling/sampling_logp_difference/max": 0.5592713243252522,
+      "eval_sampling/sampling_logp_difference/mean": 0.023261879470098664,
+      "eval_steps_per_second": 0.019,
+      "step": 150
+    },
+    {
+      "clip_ratio/high_max": 0.0015072552330821054,
+      "clip_ratio/high_mean": 0.0007480308923732082,
+      "clip_ratio/low_mean": 0.00046035814530114293,
+      "clip_ratio/low_min": 4.659314690798055e-05,
+      "clip_ratio/region_mean": 0.0012083890383109974,
+      "completions/clipped_ratio": 0.645703125,
+      "completions/max_length": 768.0,
+      "completions/max_terminated_length": 135.7,
+      "completions/mean_length": 126.4341796875,
+      "completions/mean_terminated_length": 15.374606847763062,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "entropy": 0.3454160179942846,
+      "epoch": 0.037479503396579995,
+      "frac_reward_zero_std": 0.33125,
+      "grad_norm": 0.44730472564697266,
+      "kl": 0.0019512808983563445,
+      "learning_rate": 5e-06,
+      "loss": -0.0212,
+      "num_tokens": 55105338.0,
+      "reward": 4.60107421875,
+      "reward/patch_apply_success_pct_given_patch": 69.12932624816895,
+      "reward/pct_prefill_correct_no_patch": 81.05989646911621,
+      "reward/pct_prefill_incorrect_with_patch": 88.68526840209961,
+      "reward/success_pct_prefill_correct": 78.93750076293945,
+      "reward/success_pct_prefill_incorrect": 51.45628700256348,
+      "reward_std": 7.659944105148315,
+      "rewards/oci_v2_harm_aware_reward_fn/mean": 4.60107421875,
+      "rewards/oci_v2_harm_aware_reward_fn/std": 7.659944152832031,
+      "sampling/importance_sampling_ratio/max": 1.2753490328788757,
+      "sampling/importance_sampling_ratio/mean": 0.39345550537109375,
+      "sampling/importance_sampling_ratio/min": 6.3935438457677e-08,
+      "sampling/sampling_logp_difference/max": 1.3345935821533204,
+      "sampling/sampling_logp_difference/mean": 0.024168806709349157,
+      "step": 160,
+      "step_time": 34.51039165686816
+    },
+    {
+      "clip_ratio/high_max": 0.0017427360064175446,
+      "clip_ratio/high_mean": 0.0008446223799182917,
+      "clip_ratio/low_mean": 0.0004573430867822026,
+      "clip_ratio/low_min": 3.766135123441927e-05,
+      "clip_ratio/region_mean": 0.001301965461607324,
+      "completions/clipped_ratio": 0.701171875,
+      "completions/max_length": 750.4,
+      "completions/max_terminated_length": 126.2,
+      "completions/mean_length": 139.3912109375,
+      "completions/mean_terminated_length": 16.530626010894775,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "entropy": 0.3588992450386286,
+      "epoch": 0.039821972358866246,
+      "frac_reward_zero_std": 0.29375,
+      "grad_norm": 0.26454436779022217,
+      "kl": 0.001885227673483314,
+      "learning_rate": 5e-06,
+      "loss": -0.0325,
+      "num_tokens": 58586477.0,
+      "reward": 4.097314453125,
+      "reward/patch_apply_success_pct_given_patch": 67.74537048339843,
+      "reward/pct_prefill_correct_no_patch": 66.71354141235352,
+      "reward/pct_prefill_incorrect_with_patch": 90.23902587890625,
+      "reward/success_pct_prefill_correct": 64.35677032470703,
+      "reward/success_pct_prefill_incorrect": 50.78311004638672,
+      "reward_std": 7.915364122390747,
+      "rewards/oci_v2_harm_aware_reward_fn/mean": 4.097314453125,
+      "rewards/oci_v2_harm_aware_reward_fn/std": 7.915364122390747,
+      "sampling/importance_sampling_ratio/max": 1.254622995853424,
+      "sampling/importance_sampling_ratio/mean": 0.33434736728668213,
+      "sampling/importance_sampling_ratio/min": 7.279631671421331e-07,
+      "sampling/sampling_logp_difference/max": 1.2908679127693177,
+      "sampling/sampling_logp_difference/mean": 0.02497487124055624,
+      "step": 170,
+      "step_time": 35.27049569059163
+    },
+    {
+      "clip_ratio/high_max": 0.0017404656613507542,
+      "clip_ratio/high_mean": 0.0008687500426276529,
+      "clip_ratio/low_mean": 0.0004738669369544368,
+      "clip_ratio/low_min": 3.3239836193388327e-05,
+      "clip_ratio/region_mean": 0.001342616986585199,
+      "completions/clipped_ratio": 0.6466796875,
+      "completions/max_length": 728.9,
+      "completions/max_terminated_length": 146.8,
+      "completions/mean_length": 125.2267578125,
+      "completions/mean_terminated_length": 17.08717498779297,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "entropy": 0.35444816164672377,
+      "epoch": 0.0421644413211525,
+      "frac_reward_zero_std": 0.296875,
+      "grad_norm": 0.269290953874588,
+      "kl": 0.0018258504918776453,
+      "learning_rate": 5e-06,
+      "loss": -0.0136,
+      "num_tokens": 62020294.0,
+      "reward": 4.33388671875,
+      "reward/patch_apply_success_pct_given_patch": 68.74297790527343,
+      "reward/pct_prefill_correct_no_patch": 79.02083435058594,
+      "reward/pct_prefill_incorrect_with_patch": 83.8513786315918,
+      "reward/success_pct_prefill_correct": 78.68229217529297,
+      "reward/success_pct_prefill_incorrect": 49.30152549743652,
+      "reward_std": 7.767087364196778,
+      "rewards/oci_v2_harm_aware_reward_fn/mean": 4.33388671875,
+      "rewards/oci_v2_harm_aware_reward_fn/std": 7.767087364196778,
+      "sampling/importance_sampling_ratio/max": 1.271400785446167,
+      "sampling/importance_sampling_ratio/mean": 0.39874969720840453,
+      "sampling/importance_sampling_ratio/min": 1.2199684392053344e-05,
+      "sampling/sampling_logp_difference/max": 1.4641184091567994,
+      "sampling/sampling_logp_difference/mean": 0.02486763596534729,
+      "step": 180,
+      "step_time": 33.00146119841374
+    },
+    {
+      "clip_ratio/high_max": 0.0015795868832356064,
+      "clip_ratio/high_mean": 0.0007674965477235673,
+      "clip_ratio/low_mean": 0.0004804310066901962,
+      "clip_ratio/low_min": 4.5745386523776686e-05,
+      "clip_ratio/region_mean": 0.0012479275497753405,
+      "completions/clipped_ratio": 0.6447265625,
+      "completions/max_length": 710.8,
+      "completions/max_terminated_length": 241.1,
+      "completions/mean_length": 123.3158203125,
+      "completions/mean_terminated_length": 20.168214082717896,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "entropy": 0.36197232231497767,
+      "epoch": 0.04450691028343874,
+      "frac_reward_zero_std": 0.325,
+      "grad_norm": 0.4473939836025238,
+      "kl": 0.0019323770648043137,
+      "learning_rate": 5e-06,
+      "loss": -0.0301,
+      "num_tokens": 65397303.0,
+      "reward": 4.912451171875,
+      "reward/patch_apply_success_pct_given_patch": 69.71870727539063,
+      "reward/pct_prefill_correct_no_patch": 79.65624923706055,
+      "reward/pct_prefill_incorrect_with_patch": 89.52864608764648,
+      "reward/success_pct_prefill_correct": 77.54687461853027,
+      "reward/success_pct_prefill_incorrect": 54.71223907470703,
+      "reward_std": 7.583841943740845,
+      "rewards/oci_v2_harm_aware_reward_fn/mean": 4.912451171875,
+      "rewards/oci_v2_harm_aware_reward_fn/std": 7.583841943740845,
+      "sampling/importance_sampling_ratio/max": 1.2503548622131349,
+      "sampling/importance_sampling_ratio/mean": 0.39580638110637667,
+      "sampling/importance_sampling_ratio/min": 7.659829724077985e-07,
+      "sampling/sampling_logp_difference/max": 1.3984684586524962,
+      "sampling/sampling_logp_difference/mean": 0.025347075797617434,
+      "step": 190,
+      "step_time": 34.18269837521948
+    },
+    {
+      "clip_ratio/high_max": 0.0016514114551682724,
+      "clip_ratio/high_mean": 0.0008016838933144754,
+      "clip_ratio/low_mean": 0.0004561423882478266,
+      "clip_ratio/low_min": 4.083774074388202e-05,
+      "clip_ratio/region_mean": 0.0012578262760143844,
+      "completions/clipped_ratio": 0.616796875,
+      "completions/max_length": 736.5,
+      "completions/max_terminated_length": 159.9,
+      "completions/mean_length": 120.201953125,
+      "completions/mean_terminated_length": 19.540062618255615,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "entropy": 0.33935244772583245,
+      "epoch": 0.04684937924572499,
+      "frac_reward_zero_std": 0.384375,
+      "grad_norm": 0.534030020236969,
+      "kl": 0.0026403197101899424,
+      "learning_rate": 5e-06,
+      "loss": -0.0204,
+      "num_tokens": 68772001.0,
+      "reward": 4.7046061038970945,
+      "reward/patch_apply_success_pct_given_patch": 68.23201026916504,
+      "reward/pct_prefill_correct_no_patch": 80.64843826293945,
+      "reward/pct_prefill_incorrect_with_patch": 85.05059432983398,
+      "reward/success_pct_prefill_correct": 79.39583320617676,
+      "reward/success_pct_prefill_incorrect": 53.70907783508301,
+      "reward_std": 7.831330060958862,
+      "rewards/oci_v2_harm_aware_reward_fn/mean": 4.7046061038970945,
+      "rewards/oci_v2_harm_aware_reward_fn/std": 7.831329965591431,
+      "sampling/importance_sampling_ratio/max": 1.3202938795089723,
+      "sampling/importance_sampling_ratio/mean": 0.41943994760513303,
+      "sampling/importance_sampling_ratio/min": 2.1222655705084407e-06,
+      "sampling/sampling_logp_difference/max": 1.7484084129333497,
+      "sampling/sampling_logp_difference/mean": 0.02411667872220278,
+      "step": 200,
+      "step_time": 36.26669738506898
+    },
+    {
+      "epoch": 0.04684937924572499,
+      "eval_clip_ratio/high_max": 0.0,
+      "eval_clip_ratio/high_mean": 0.0,
+      "eval_clip_ratio/low_mean": 0.0,
+      "eval_clip_ratio/low_min": 0.0,
+      "eval_clip_ratio/region_mean": 0.0,
+      "eval_completions/clipped_ratio": 0.42081925675675674,
+      "eval_completions/max_length": 507.0135135135135,
+      "eval_completions/max_terminated_length": 108.27027027027027,
+      "eval_completions/mean_length": 116.3194679054054,
+      "eval_completions/mean_terminated_length": 39.49170837853406,
+      "eval_completions/min_length": 8.22972972972973,
+      "eval_completions/min_terminated_length": 8.675675675675675,
+      "eval_entropy": 0.2682908717039469,
+      "eval_frac_reward_zero_std": 0.30067567567567566,
+      "eval_kl": 0.0019366928398675235,
+      "eval_loss": -0.07588622719049454,
+      "eval_num_tokens": 68772001.0,
+      "eval_reward": 5.09135530768214,
+      "eval_reward/patch_apply_success_pct_given_patch": 32.234417124374495,
+      "eval_reward/pct_prefill_correct_no_patch": 53.589527027027025,
+      "eval_reward/pct_prefill_incorrect_with_patch": 27.82939189189189,
+      "eval_reward/success_pct_prefill_correct": 51.22466216216216,
+      "eval_reward/success_pct_prefill_incorrect": 14.611486486486486,
+      "eval_reward_std": 6.815579531966029,
+      "eval_rewards/oci_v2_harm_aware_reward_fn/mean": 5.09135530768214,
+      "eval_rewards/oci_v2_harm_aware_reward_fn/std": 6.815579738165881,
+      "eval_runtime": 241.3259,
+      "eval_samples_per_second": 1.222,
+      "eval_sampling/importance_sampling_ratio/max": 1.0236477892141085,
+      "eval_sampling/importance_sampling_ratio/mean": 0.5273233633991834,
+      "eval_sampling/importance_sampling_ratio/min": 0.0006832520327497961,
+      "eval_sampling/sampling_logp_difference/max": 0.5269765443093067,
+      "eval_sampling/sampling_logp_difference/mean": 0.019410189562099608,
+      "eval_steps_per_second": 0.021,
+      "step": 200
+    },
+    {
+      "clip_ratio/high_max": 0.001524216149118729,
+      "clip_ratio/high_mean": 0.0007171164044848411,
+      "clip_ratio/low_mean": 0.00047069367865333335,
+      "clip_ratio/low_min": 4.533111314231064e-05,
+      "clip_ratio/region_mean": 0.001187810087503749,
+      "completions/clipped_ratio": 0.6484375,
+      "completions/max_length": 715.6,
+      "completions/max_terminated_length": 156.9,
+      "completions/mean_length": 126.9662109375,
+      "completions/mean_terminated_length": 19.480879068374634,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "entropy": 0.35859104692935945,
+      "epoch": 0.049191848208011243,
+      "frac_reward_zero_std": 0.35625,
+      "grad_norm": 0.4524937570095062,
+      "kl": 0.0022339657152770087,
+      "learning_rate": 5e-06,
+      "loss": -0.0071,
+      "num_tokens": 72233444.0,
+      "reward": 4.220654296875,
+      "reward/patch_apply_success_pct_given_patch": 66.15447463989258,
+      "reward/pct_prefill_correct_no_patch": 78.52864608764648,
+      "reward/pct_prefill_incorrect_with_patch": 88.71521453857422,
+      "reward/success_pct_prefill_correct": 76.09375038146973,
+      "reward/success_pct_prefill_incorrect": 47.58779792785644,
+      "reward_std": 7.714002084732056,
+      "rewards/oci_v2_harm_aware_reward_fn/mean": 4.220654296875,
+      "rewards/oci_v2_harm_aware_reward_fn/std": 7.71400203704834,
+      "sampling/importance_sampling_ratio/max": 1.3017072200775146,
+      "sampling/importance_sampling_ratio/mean": 0.3829398512840271,
+      "sampling/importance_sampling_ratio/min": 7.960561056374287e-07,
+      "sampling/sampling_logp_difference/max": 1.1893439531326293,
+      "sampling/sampling_logp_difference/mean": 0.025247183069586752,
+      "step": 210,
+      "step_time": 34.08152010687627
+    },
+    {
+      "clip_ratio/high_max": 0.0016290366933390033,
+      "clip_ratio/high_mean": 0.0007606110095366603,
+      "clip_ratio/low_mean": 0.0005715575281101337,
+      "clip_ratio/low_min": 2.924471664300654e-05,
+      "clip_ratio/region_mean": 0.0013321685306436847,
+      "completions/clipped_ratio": 0.6185546875,
+      "completions/max_length": 767.1,
+      "completions/max_terminated_length": 125.8,
+      "completions/mean_length": 113.638671875,
+      "completions/mean_terminated_length": 18.631591701507567,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "entropy": 0.3554548051208258,
+      "epoch": 0.051534317170297494,
+      "frac_reward_zero_std": 0.40625,
+      "grad_norm": 0.33937129378318787,
+      "kl": 0.001615840035083238,
+      "learning_rate": 5e-06,
+      "loss": -0.0044,
+      "num_tokens": 75600586.0,
+      "reward": 4.85361328125,
+      "reward/patch_apply_success_pct_given_patch": 70.00259628295899,
+      "reward/pct_prefill_correct_no_patch": 87.33333358764648,
+      "reward/pct_prefill_incorrect_with_patch": 83.99311828613281,
+      "reward/success_pct_prefill_correct": 86.35677108764648,
+      "reward/success_pct_prefill_incorrect": 53.367373657226565,
+      "reward_std": 7.785676193237305,
+      "rewards/oci_v2_harm_aware_reward_fn/mean": 4.85361328125,
+      "rewards/oci_v2_harm_aware_reward_fn/std": 7.7856762409210205,
+      "sampling/importance_sampling_ratio/max": 1.2224095702171325,
+      "sampling/importance_sampling_ratio/mean": 0.41560044288635256,
+      "sampling/importance_sampling_ratio/min": 6.5917423597862015e-06,
+      "sampling/sampling_logp_difference/max": 0.8749272286891937,
+      "sampling/sampling_logp_difference/mean": 0.024940383993089198,
+      "step": 220,
+      "step_time": 35.16804402153939
+    },
+    {
+      "clip_ratio/high_max": 0.0018055139735224657,
+      "clip_ratio/high_mean": 0.0008565399515646277,
+      "clip_ratio/low_mean": 0.0004938499298987153,
+      "clip_ratio/low_min": 6.662547457381152e-05,
+      "clip_ratio/region_mean": 0.00135038988883025,
+      "completions/clipped_ratio": 0.62734375,
+      "completions/max_length": 768.0,
+      "completions/max_terminated_length": 309.5,
+      "completions/mean_length": 133.074609375,
+      "completions/mean_terminated_length": 23.312591361999512,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "entropy": 0.33703442458063365,
+      "epoch": 0.053876786132583745,
+      "frac_reward_zero_std": 0.31875,
+      "grad_norm": 0.19537745416164398,
+      "kl": 0.00198038191156229,
+      "learning_rate": 5e-06,
+      "loss": -0.0109,
+      "num_tokens": 79066552.0,
+      "reward": 4.357763671875,
+      "reward/patch_apply_success_pct_given_patch": 70.90412101745605,
+      "reward/pct_prefill_correct_no_patch": 84.83333358764648,
+      "reward/pct_prefill_incorrect_with_patch": 85.85267791748046,
+      "reward/success_pct_prefill_correct": 80.22656326293945,
+      "reward/success_pct_prefill_incorrect": 46.19587078094482,
+      "reward_std": 7.6421041011810305,
+      "rewards/oci_v2_harm_aware_reward_fn/mean": 4.357763671875,
+      "rewards/oci_v2_harm_aware_reward_fn/std": 7.642103910446167,
+      "sampling/importance_sampling_ratio/max": 1.4409372568130494,
+      "sampling/importance_sampling_ratio/mean": 0.4029893338680267,
+      "sampling/importance_sampling_ratio/min": 7.934131353631465e-09,
+      "sampling/sampling_logp_difference/max": 1.2338001132011414,
+      "sampling/sampling_logp_difference/mean": 0.023645965754985808,
+      "step": 230,
+      "step_time": 36.12927001840435
+    },
+    {
+      "clip_ratio/high_max": 0.001607694441190688,
+      "clip_ratio/high_mean": 0.000776769925505505,
+      "clip_ratio/low_mean": 0.00040344646076846403,
+      "clip_ratio/low_min": 2.9592250211862847e-05,
+      "clip_ratio/region_mean": 0.0011802163855463732,
+      "completions/clipped_ratio": 0.6642578125,
+      "completions/max_length": 768.0,
+      "completions/max_terminated_length": 180.3,
+      "completions/mean_length": 128.5185546875,
+      "completions/mean_terminated_length": 20.366122198104858,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "entropy": 0.3574300825595856,
+      "epoch": 0.05621925509486999,
+      "frac_reward_zero_std": 0.309375,
+      "grad_norm": 0.23202241957187653,
+      "kl": 0.0020687328178610187,
+      "learning_rate": 5e-06,
+      "loss": -0.0312,
+      "num_tokens": 82518679.0,
+      "reward": 5.126470279693604,
+      "reward/patch_apply_success_pct_given_patch": 72.32947463989258,
+      "reward/pct_prefill_correct_no_patch": 75.67187423706055,
+      "reward/pct_prefill_incorrect_with_patch": 87.48251342773438,
+      "reward/success_pct_prefill_correct": 74.88802032470703,
+      "reward/success_pct_prefill_incorrect": 59.09505195617676,
+      "reward_std": 7.667966508865357,
+      "rewards/oci_v2_harm_aware_reward_fn/mean": 5.126470279693604,
+      "rewards/oci_v2_harm_aware_reward_fn/std": 7.66796646118164,
+      "sampling/importance_sampling_ratio/max": 1.2375670552253724,
+      "sampling/importance_sampling_ratio/mean": 0.36907707154750824,
+      "sampling/importance_sampling_ratio/min": 4.612215892280765e-09,
+      "sampling/sampling_logp_difference/max": 1.3384527087211608,
+      "sampling/sampling_logp_difference/mean": 0.025057115592062473,
+      "step": 240,
+      "step_time": 35.637847752776
+    },
+    {
+      "clip_ratio/high_max": 0.0017825781749706948,
+      "clip_ratio/high_mean": 0.0008423530697655224,
+      "clip_ratio/low_mean": 0.0005241865664174838,
+      "clip_ratio/low_min": 5.9932574004051276e-05,
+      "clip_ratio/region_mean": 0.0013665396385476925,
+      "completions/clipped_ratio": 0.6330078125,
+      "completions/max_length": 717.7,
+      "completions/max_terminated_length": 209.9,
+      "completions/mean_length": 124.443359375,
+      "completions/mean_terminated_length": 19.373891258239745,
+      "completions/min_length": 3.0,
+      "completions/min_terminated_length": 3.0,
+      "entropy": 0.36250690259039403,
+      "epoch": 0.05856172405715624,
+      "frac_reward_zero_std": 0.328125,
+      "grad_norm": 0.43943753838539124,
+      "kl": 0.0024928284539782906,
+      "learning_rate": 5e-06,
+      "loss": -0.0364,
+      "num_tokens": 85921717.0,
+      "reward": 4.62705078125,
+      "reward/patch_apply_success_pct_given_patch": 65.50032081604004,
+      "reward/pct_prefill_correct_no_patch": 75.82291679382324,
+      "reward/pct_prefill_incorrect_with_patch": 84.63988189697265,
+      "reward/success_pct_prefill_correct": 75.33333320617676,
+      "reward/success_pct_prefill_incorrect": 53.19512748718262,
+      "reward_std": 7.864572048187256,
+      "rewards/oci_v2_harm_aware_reward_fn/mean": 4.62705078125,
+      "rewards/oci_v2_harm_aware_reward_fn/std": 7.864572143554687,
+      "sampling/importance_sampling_ratio/max": 1.3025658011436463,
+      "sampling/importance_sampling_ratio/mean": 0.3932421803474426,
+      "sampling/importance_sampling_ratio/min": 4.742778184447995e-07,
+      "sampling/sampling_logp_difference/max": 1.052924931049347,
+      "sampling/sampling_logp_difference/mean": 0.02553384527564049,
+      "step": 250,
+      "step_time": 32.904292047861965
+    },
+    {
+      "epoch": 0.05856172405715624,
+      "eval_clip_ratio/high_max": 0.0,
+      "eval_clip_ratio/high_mean": 0.0,
+      "eval_clip_ratio/low_mean": 0.0,
+      "eval_clip_ratio/low_min": 0.0,
+      "eval_clip_ratio/region_mean": 0.0,
+      "eval_completions/clipped_ratio": 0.47149493243243246,
+      "eval_completions/max_length": 524.2027027027027,
+      "eval_completions/max_terminated_length": 105.24324324324324,
+      "eval_completions/mean_length": 126.09586148648648,
+      "eval_completions/mean_terminated_length": 40.073304559733415,
+      "eval_completions/min_length": 9.68918918918919,
+      "eval_completions/min_terminated_length": 10.175675675675675,
+      "eval_entropy": 0.2999383756840551,
+      "eval_frac_reward_zero_std": 0.2635135135135135,
+      "eval_kl": 0.0032616897269444715,
+      "eval_loss": -0.09328581392765045,
+      "eval_num_tokens": 85921717.0,
+      "eval_reward": 4.793486569378827,
+      "eval_reward/patch_apply_success_pct_given_patch": 35.35875937101003,
+      "eval_reward/pct_prefill_correct_no_patch": 49.683277027027025,
+      "eval_reward/pct_prefill_incorrect_with_patch": 28.969594594594593,
+      "eval_reward/success_pct_prefill_correct": 47.99408783783784,
+      "eval_reward/success_pct_prefill_incorrect": 15.477195945945946,
+      "eval_reward_std": 6.969474463043986,
+      "eval_rewards/oci_v2_harm_aware_reward_fn/mean": 4.793486569378827,
+      "eval_rewards/oci_v2_harm_aware_reward_fn/std": 6.9694746660723075,
+      "eval_runtime": 248.8167,
+      "eval_samples_per_second": 1.186,
+      "eval_sampling/importance_sampling_ratio/max": 1.0139549906189378,
+      "eval_sampling/importance_sampling_ratio/mean": 0.4834826766840509,
+      "eval_sampling/importance_sampling_ratio/min": 0.00043338534439003667,
+      "eval_sampling/sampling_logp_difference/max": 0.6961322237510938,
+      "eval_sampling/sampling_logp_difference/mean": 0.02163152744037074,
+      "eval_steps_per_second": 0.02,
+      "step": 250
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 8000,
+  "num_input_tokens_seen": 85921717,
+  "num_train_epochs": 2,
+  "save_steps": 250,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-250/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:39626a1c4d97cde06662f17b3014e463f6a8f998ea251e23d77227c185b8a127
+size 8337

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-500/README.md

@@ -0,0 +1,209 @@
+---
+base_model: /root/repos/verl/new_models/qwen/qwen3-4B-v4-if-partial-code_cont-legacy-system-1ep-muts-fnr-reflection-True-patch-instruction-v4-classic-variant-legacy-iim-system_lr_1e-06_bs_8_ga_8_pt_llm_muts_synth_muts_partial_synth_muts_git_understanding_as_muts_iim_system_checkpoint-2922
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:/root/repos/verl/new_models/qwen/qwen3-4B-v4-if-partial-code_cont-legacy-system-1ep-muts-fnr-reflection-True-patch-instruction-v4-classic-variant-legacy-iim-system_lr_1e-06_bs_8_ga_8_pt_llm_muts_synth_muts_partial_synth_muts_git_understanding_as_muts_iim_system_checkpoint-2922
+- grpo
+- lora
+- transformers
+- trl
+---
+
+# 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 0.18.0

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-500/adapter_config.json

@@ -0,0 +1,41 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "/root/repos/verl/new_models/qwen/qwen3-4B-v4-if-partial-code_cont-legacy-system-1ep-muts-fnr-reflection-True-patch-instruction-v4-classic-variant-legacy-iim-system_lr_1e-06_bs_8_ga_8_pt_llm_muts_synth_muts_partial_synth_muts_git_understanding_as_muts_iim_system_checkpoint-2922",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "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": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.0",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "v_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-500/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4191fd2047a0562833d5f7d8e95a6cea4b447390495f50e04cb9435854474dc2
+size 23612264

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-500/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
+}

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-500/chat_template.jinja

@@ -0,0 +1,61 @@
+{%- 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 %}
+{%- 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" %}
+        {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- 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' }}
+{%- endif %}

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-500/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:128fad5e0bd082ea1de5a651ebf2f80d5ea76778692eced831663eecfd9611d7
+size 47309579

qwen3-4B-v4-if-partial-code_cont-legacy-system-grpo-harm-aware-three-types-b16g8-ng16-lr5e-06-v2_harm_aware-cwup-w50-c2i0.5/checkpoint-500/ref/adapter_config.json

@@ -0,0 +1,41 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "/root/repos/verl/new_models/qwen/qwen3-4B-v4-if-partial-code_cont-legacy-system-1ep-muts-fnr-reflection-True-patch-instruction-v4-classic-variant-legacy-iim-system_lr_1e-06_bs_8_ga_8_pt_llm_muts_synth_muts_partial_synth_muts_git_understanding_as_muts_iim_system_checkpoint-2922",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "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": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.0",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "v_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}