Upload model artifacts

.gitattributes

@@ -33,3 +33,7 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+checkpoint-1000/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-1497/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-500/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,72 @@
+---
+base_model: mistralai/Ministral-3-3B-Instruct-2512-BF16
+library_name: peft
+model_name: mistral-grpo
+tags:
+- base_model:adapter:mistralai/Ministral-3-3B-Instruct-2512-BF16
+- grpo
+- lora
+- transformers
+- trl
+licence: license
+pipeline_tag: text-generation
+---
+
+# Model Card for mistral-grpo
+
+This model is a fine-tuned version of [mistralai/Ministral-3-3B-Instruct-2512-BF16](https://huggingface.co/mistralai/Ministral-3-3B-Instruct-2512-BF16).
+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://wandb.ai/evalonlabs/mistral-rl/runs/wehkefcs) 
+
+
+
+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
+
+- PEFT 0.18.1
+- TRL: 0.29.0
+- Transformers: 5.2.0
+- Pytorch: 2.6.0+cu124
+- Datasets: 4.6.1
+- 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}
+}
+```

adapter_config.json

@@ -0,0 +1,46 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "mistralai/Ministral-3-3B-Instruct-2512-BF16",
+  "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": 64,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 32,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "v_proj",
+    "down_proj",
+    "gate_proj",
+    "up_proj",
+    "k_proj",
+    "o_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7c442e995e1108b636c351205f914dbc3f4bd6da2c2a3b8bf4793070403e6ee5
+size 135113744

chat_template.jinja

@@ -0,0 +1,121 @@
+{#- Default system message if no system prompt is passed. #}
+{%- set default_system_message = 'You are Ministral-3-3B-Instruct-2512, a Large Language Model (LLM) created by Mistral AI, a French startup headquartered in Paris.\nYou power an AI assistant called Le Chat.\nYour knowledge base was last updated on 2023-10-01.\nThe current date is {today}.\n\nWhen you\'re not sure about some information or when the user\'s request requires up-to-date or specific data, you must use the available tools to fetch the information. Do not hesitate to use tools whenever they can provide a more accurate or complete response. If no relevant tools are available, then clearly state that you don\'t have the information and avoid making up anything.\nIf the user\'s question is not clear, ambiguous, or does not provide enough context for you to accurately answer the question, you do not try to answer it right away and you rather ask the user to clarify their request (e.g. "What are some good restaurants around me?" => "Where are you?" or "When is the next flight to Tokyo" => "Where do you travel from?").\nYou are always very attentive to dates, in particular you try to resolve dates (e.g. "yesterday" is {yesterday}) and when asked about information at specific dates, you discard information that is at another date.\nYou follow these instructions in all languages, and always respond to the user in the language they use or request.\nNext sections describe the capabilities that you have.\n\n# WEB BROWSING INSTRUCTIONS\n\nYou cannot perform any web search or access internet to open URLs, links etc. If it seems like the user is expecting you to do so, you clarify the situation and ask the user to copy paste the text directly in the chat.\n\n# MULTI-MODAL INSTRUCTIONS\n\nYou have the ability to read images, but you cannot generate images. You also cannot transcribe audio files or videos.\nYou cannot read nor transcribe audio files or videos.\n\n# TOOL CALLING INSTRUCTIONS\n\nYou may have access to tools that you can use to fetch information or perform actions. You must use these tools in the following situations:\n\n1. When the request requires up-to-date information.\n2. When the request requires specific data that you do not have in your knowledge base.\n3. When the request involves actions that you cannot perform without tools.\n\nAlways prioritize using tools to provide the most accurate and helpful response. If tools are not available, inform the user that you cannot perform the requested action at the moment.' %}
+
+{#- Begin of sequence token. #}
+{{- bos_token }}
+
+{#- Handle system prompt if it exists. #}
+{#- System prompt supports text content or text chunks. #}
+{%- if messages[0]['role'] == 'system' %}
+    {{- '[SYSTEM_PROMPT]' -}}
+    {%- if messages[0]['content'] is string %}
+        {{- messages[0]['content'] -}}
+    {%- else %}        
+        {%- for block in messages[0]['content'] %}
+            {%- if block['type'] == 'text' %}
+                {{- block['text'] }}
+            {%- else %}
+                {{- raise_exception('Only text chunks are supported in system message contents.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- endif %}
+    {{- '[/SYSTEM_PROMPT]' -}}
+    {%- set loop_messages = messages[1:] %}
+{%- else %}
+    {%- set loop_messages = messages %}
+    {%- if default_system_message != '' %}
+        {{- '[SYSTEM_PROMPT]' + default_system_message + '[/SYSTEM_PROMPT]' }}
+    {%- endif %}
+{%- endif %}
+
+
+{#- Tools definition #}
+{%- set tools_definition = '' %}
+{%- set has_tools = false %}
+{%- if tools is defined and tools is not none and tools|length > 0 %}
+    {%- set has_tools = true %}
+    {%- set tools_definition = '[AVAILABLE_TOOLS]' + (tools| tojson) + '[/AVAILABLE_TOOLS]' %}
+    {{- tools_definition }}
+{%- endif %}
+
+{#- Checks for alternating user/assistant messages. #}
+{%- set ns = namespace(index=0) %}
+{%- for message in loop_messages %}
+    {%- if message.role == 'user' or (message.role == 'assistant' and (message.tool_calls is not defined or message.tool_calls is none or message.tool_calls | length == 0)) %}
+        {%- if (message['role'] == 'user') != (ns.index % 2 == 0) %}
+            {{- raise_exception('After the optional system message, conversation roles must alternate user and assistant roles except for tool calls and results.') }}
+        {%- endif %}
+        {%- set ns.index = ns.index + 1 %}
+    {%- endif %}
+{%- endfor %}
+
+{#- Handle conversation messages. #}
+{%- for message in loop_messages %}
+
+    {#- User messages supports text content or text and image chunks. #}
+    {%- if message['role'] == 'user' %}
+        {%- if message['content'] is string %}
+            {{- '[INST]' + message['content'] + '[/INST]' }}
+        {%- elif message['content'] | length > 0 %}
+            {{- '[INST]' }}
+            {%- if message['content'] | length == 2 %}
+                {%- set blocks = message['content'] | sort(attribute='type') %}
+            {%- else %}
+                {%- set blocks = message['content'] %}
+            {%- endif %}
+            {%- for block in blocks %}
+                {%- if block['type'] == 'text' %}
+                    {{- block['text'] }}
+                {%- elif block['type'] in ['image', 'image_url'] %}
+                    {{- '[IMG]' }}
+                {%- else %}
+                    {{- raise_exception('Only text, image and image_url chunks are supported in user message content.') }}
+                {%- endif %}
+            {%- endfor %}
+            {{- '[/INST]' }}
+        {%- else %}
+            {{- raise_exception('User message must have a string or a list of chunks in content') }}
+        {%- endif %}
+
+    {#- Assistant messages supports text content or text and image chunks. #}
+    {%- elif message['role'] == 'assistant' %}
+        {%- if (message['content'] is none or message['content'] == '' or message['content']|length == 0) and (message['tool_calls'] is not defined or message['tool_calls'] is none or message['tool_calls']|length == 0) %}
+            {{- raise_exception('Assistant message must have a string or a list of chunks in content or a list of tool calls.') }}
+        {%- endif %}
+
+        {%- if message['content'] is string %}
+            {{- message['content'] }}
+        {%- elif message['content'] | length > 0 %}
+            {%- for block in message['content'] %}
+                {%- if block['type'] == 'text' %}
+                    {{- block['text'] }}
+                {%- else %}
+                    {{- raise_exception('Only text chunks are supported in assistant message contents.') }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        
+        {%- if message['tool_calls'] is defined and message['tool_calls'] is not none and message['tool_calls']|length > 0 %}
+            {%- for tool in message['tool_calls'] %}
+                {%- set arguments = tool['function']['arguments'] %}
+                {%- if arguments is not string %}
+                    {%- set arguments = arguments|tojson|safe %}
+                {%- elif arguments == '' %}
+                    {%- set arguments = '{}' %}
+                {%- endif %}
+                {{- '[TOOL_CALLS]' + tool['function']['name'] + '[ARGS]' + arguments }}
+            {%- endfor %}
+        {%- endif %}
+
+        {#- End of sequence token for each assistant messages. #}
+        {{- eos_token }}
+
+    {#- Tool messages only supports text content. #}
+    {%- elif message['role'] == 'tool' %}
+        {{- '[TOOL_RESULTS]' + message['content']|string + '[/TOOL_RESULTS]' }}
+
+    {#- Raise exception for unsupported roles. #}
+    {%- else %}
+        {{- raise_exception('Only user, assistant and tool roles are supported, got ' + message['role']) }}
+    {%- endif %}
+{%- endfor %}

checkpoint-1000/README.md

@@ -0,0 +1,209 @@
+---
+base_model: mistralai/Ministral-3-3B-Instruct-2512-BF16
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:mistralai/Ministral-3-3B-Instruct-2512-BF16
+- 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 0.18.1

checkpoint-1000/adapter_config.json

@@ -0,0 +1,46 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "mistralai/Ministral-3-3B-Instruct-2512-BF16",
+  "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": 64,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 32,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "v_proj",
+    "down_proj",
+    "gate_proj",
+    "up_proj",
+    "k_proj",
+    "o_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

checkpoint-1000/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3f6c63ba0e1a8246aedca0b97d31e9e4e27b41a5a1ac3ce0af2652f127b4a68c
+size 135113744

checkpoint-1000/chat_template.jinja

@@ -0,0 +1,121 @@
+{#- Default system message if no system prompt is passed. #}
+{%- set default_system_message = 'You are Ministral-3-3B-Instruct-2512, a Large Language Model (LLM) created by Mistral AI, a French startup headquartered in Paris.\nYou power an AI assistant called Le Chat.\nYour knowledge base was last updated on 2023-10-01.\nThe current date is {today}.\n\nWhen you\'re not sure about some information or when the user\'s request requires up-to-date or specific data, you must use the available tools to fetch the information. Do not hesitate to use tools whenever they can provide a more accurate or complete response. If no relevant tools are available, then clearly state that you don\'t have the information and avoid making up anything.\nIf the user\'s question is not clear, ambiguous, or does not provide enough context for you to accurately answer the question, you do not try to answer it right away and you rather ask the user to clarify their request (e.g. "What are some good restaurants around me?" => "Where are you?" or "When is the next flight to Tokyo" => "Where do you travel from?").\nYou are always very attentive to dates, in particular you try to resolve dates (e.g. "yesterday" is {yesterday}) and when asked about information at specific dates, you discard information that is at another date.\nYou follow these instructions in all languages, and always respond to the user in the language they use or request.\nNext sections describe the capabilities that you have.\n\n# WEB BROWSING INSTRUCTIONS\n\nYou cannot perform any web search or access internet to open URLs, links etc. If it seems like the user is expecting you to do so, you clarify the situation and ask the user to copy paste the text directly in the chat.\n\n# MULTI-MODAL INSTRUCTIONS\n\nYou have the ability to read images, but you cannot generate images. You also cannot transcribe audio files or videos.\nYou cannot read nor transcribe audio files or videos.\n\n# TOOL CALLING INSTRUCTIONS\n\nYou may have access to tools that you can use to fetch information or perform actions. You must use these tools in the following situations:\n\n1. When the request requires up-to-date information.\n2. When the request requires specific data that you do not have in your knowledge base.\n3. When the request involves actions that you cannot perform without tools.\n\nAlways prioritize using tools to provide the most accurate and helpful response. If tools are not available, inform the user that you cannot perform the requested action at the moment.' %}
+
+{#- Begin of sequence token. #}
+{{- bos_token }}
+
+{#- Handle system prompt if it exists. #}
+{#- System prompt supports text content or text chunks. #}
+{%- if messages[0]['role'] == 'system' %}
+    {{- '[SYSTEM_PROMPT]' -}}
+    {%- if messages[0]['content'] is string %}
+        {{- messages[0]['content'] -}}
+    {%- else %}        
+        {%- for block in messages[0]['content'] %}
+            {%- if block['type'] == 'text' %}
+                {{- block['text'] }}
+            {%- else %}
+                {{- raise_exception('Only text chunks are supported in system message contents.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- endif %}
+    {{- '[/SYSTEM_PROMPT]' -}}
+    {%- set loop_messages = messages[1:] %}
+{%- else %}
+    {%- set loop_messages = messages %}
+    {%- if default_system_message != '' %}
+        {{- '[SYSTEM_PROMPT]' + default_system_message + '[/SYSTEM_PROMPT]' }}
+    {%- endif %}
+{%- endif %}
+
+
+{#- Tools definition #}
+{%- set tools_definition = '' %}
+{%- set has_tools = false %}
+{%- if tools is defined and tools is not none and tools|length > 0 %}
+    {%- set has_tools = true %}
+    {%- set tools_definition = '[AVAILABLE_TOOLS]' + (tools| tojson) + '[/AVAILABLE_TOOLS]' %}
+    {{- tools_definition }}
+{%- endif %}
+
+{#- Checks for alternating user/assistant messages. #}
+{%- set ns = namespace(index=0) %}
+{%- for message in loop_messages %}
+    {%- if message.role == 'user' or (message.role == 'assistant' and (message.tool_calls is not defined or message.tool_calls is none or message.tool_calls | length == 0)) %}
+        {%- if (message['role'] == 'user') != (ns.index % 2 == 0) %}
+            {{- raise_exception('After the optional system message, conversation roles must alternate user and assistant roles except for tool calls and results.') }}
+        {%- endif %}
+        {%- set ns.index = ns.index + 1 %}
+    {%- endif %}
+{%- endfor %}
+
+{#- Handle conversation messages. #}
+{%- for message in loop_messages %}
+
+    {#- User messages supports text content or text and image chunks. #}
+    {%- if message['role'] == 'user' %}
+        {%- if message['content'] is string %}
+            {{- '[INST]' + message['content'] + '[/INST]' }}
+        {%- elif message['content'] | length > 0 %}
+            {{- '[INST]' }}
+            {%- if message['content'] | length == 2 %}
+                {%- set blocks = message['content'] | sort(attribute='type') %}
+            {%- else %}
+                {%- set blocks = message['content'] %}
+            {%- endif %}
+            {%- for block in blocks %}
+                {%- if block['type'] == 'text' %}
+                    {{- block['text'] }}
+                {%- elif block['type'] in ['image', 'image_url'] %}
+                    {{- '[IMG]' }}
+                {%- else %}
+                    {{- raise_exception('Only text, image and image_url chunks are supported in user message content.') }}
+                {%- endif %}
+            {%- endfor %}
+            {{- '[/INST]' }}
+        {%- else %}
+            {{- raise_exception('User message must have a string or a list of chunks in content') }}
+        {%- endif %}
+
+    {#- Assistant messages supports text content or text and image chunks. #}
+    {%- elif message['role'] == 'assistant' %}
+        {%- if (message['content'] is none or message['content'] == '' or message['content']|length == 0) and (message['tool_calls'] is not defined or message['tool_calls'] is none or message['tool_calls']|length == 0) %}
+            {{- raise_exception('Assistant message must have a string or a list of chunks in content or a list of tool calls.') }}
+        {%- endif %}
+
+        {%- if message['content'] is string %}
+            {{- message['content'] }}
+        {%- elif message['content'] | length > 0 %}
+            {%- for block in message['content'] %}
+                {%- if block['type'] == 'text' %}
+                    {{- block['text'] }}
+                {%- else %}
+                    {{- raise_exception('Only text chunks are supported in assistant message contents.') }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        
+        {%- if message['tool_calls'] is defined and message['tool_calls'] is not none and message['tool_calls']|length > 0 %}
+            {%- for tool in message['tool_calls'] %}
+                {%- set arguments = tool['function']['arguments'] %}
+                {%- if arguments is not string %}
+                    {%- set arguments = arguments|tojson|safe %}
+                {%- elif arguments == '' %}
+                    {%- set arguments = '{}' %}
+                {%- endif %}
+                {{- '[TOOL_CALLS]' + tool['function']['name'] + '[ARGS]' + arguments }}
+            {%- endfor %}
+        {%- endif %}
+
+        {#- End of sequence token for each assistant messages. #}
+        {{- eos_token }}
+
+    {#- Tool messages only supports text content. #}
+    {%- elif message['role'] == 'tool' %}
+        {{- '[TOOL_RESULTS]' + message['content']|string + '[/TOOL_RESULTS]' }}
+
+    {#- Raise exception for unsupported roles. #}
+    {%- else %}
+        {{- raise_exception('Only user, assistant and tool roles are supported, got ' + message['role']) }}
+    {%- endif %}
+{%- endfor %}

checkpoint-1000/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c1fcbb2bef58e10219ac75da0d2ddc2aeaf8b755720d55d5f3c630c82da161cb
+size 197964694

checkpoint-1000/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f162ce076d12440e0277da5c150f3edde30ecefa385a835ce60f7fb38194e3d0
+size 14244

checkpoint-1000/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c5233508910fea36061cadbd717b379e5d4b9b72d4eb35439d12d0e0933e6481
+size 1064

checkpoint-1000/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8fbe698063980a09a4487ec5bbcc545aab380d686f9d918cad649bb49d257f83
+size 17078265

checkpoint-1000/tokenizer_config.json

@@ -0,0 +1,15 @@
+{
+  "add_prefix_space": null,
+  "backend": "tokenizers",
+  "bos_token": "<s>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "</s>",
+  "is_local": true,
+  "legacy": true,
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<pad>",
+  "processor_class": "PixtralProcessor",
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}

checkpoint-1000/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ee28670c78b80e386d67006aee3880339b84b20bda4953e363dae2a4315c97a5
+size 6520

checkpoint-1497/README.md

@@ -0,0 +1,209 @@
+---
+base_model: mistralai/Ministral-3-3B-Instruct-2512-BF16
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:mistralai/Ministral-3-3B-Instruct-2512-BF16
+- 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 0.18.1

checkpoint-1497/adapter_config.json

@@ -0,0 +1,46 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "mistralai/Ministral-3-3B-Instruct-2512-BF16",
+  "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": 64,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 32,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "v_proj",
+    "down_proj",
+    "gate_proj",
+    "up_proj",
+    "k_proj",
+    "o_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

checkpoint-1497/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7c442e995e1108b636c351205f914dbc3f4bd6da2c2a3b8bf4793070403e6ee5
+size 135113744

checkpoint-1497/chat_template.jinja

@@ -0,0 +1,121 @@
+{#- Default system message if no system prompt is passed. #}
+{%- set default_system_message = 'You are Ministral-3-3B-Instruct-2512, a Large Language Model (LLM) created by Mistral AI, a French startup headquartered in Paris.\nYou power an AI assistant called Le Chat.\nYour knowledge base was last updated on 2023-10-01.\nThe current date is {today}.\n\nWhen you\'re not sure about some information or when the user\'s request requires up-to-date or specific data, you must use the available tools to fetch the information. Do not hesitate to use tools whenever they can provide a more accurate or complete response. If no relevant tools are available, then clearly state that you don\'t have the information and avoid making up anything.\nIf the user\'s question is not clear, ambiguous, or does not provide enough context for you to accurately answer the question, you do not try to answer it right away and you rather ask the user to clarify their request (e.g. "What are some good restaurants around me?" => "Where are you?" or "When is the next flight to Tokyo" => "Where do you travel from?").\nYou are always very attentive to dates, in particular you try to resolve dates (e.g. "yesterday" is {yesterday}) and when asked about information at specific dates, you discard information that is at another date.\nYou follow these instructions in all languages, and always respond to the user in the language they use or request.\nNext sections describe the capabilities that you have.\n\n# WEB BROWSING INSTRUCTIONS\n\nYou cannot perform any web search or access internet to open URLs, links etc. If it seems like the user is expecting you to do so, you clarify the situation and ask the user to copy paste the text directly in the chat.\n\n# MULTI-MODAL INSTRUCTIONS\n\nYou have the ability to read images, but you cannot generate images. You also cannot transcribe audio files or videos.\nYou cannot read nor transcribe audio files or videos.\n\n# TOOL CALLING INSTRUCTIONS\n\nYou may have access to tools that you can use to fetch information or perform actions. You must use these tools in the following situations:\n\n1. When the request requires up-to-date information.\n2. When the request requires specific data that you do not have in your knowledge base.\n3. When the request involves actions that you cannot perform without tools.\n\nAlways prioritize using tools to provide the most accurate and helpful response. If tools are not available, inform the user that you cannot perform the requested action at the moment.' %}
+
+{#- Begin of sequence token. #}
+{{- bos_token }}
+
+{#- Handle system prompt if it exists. #}
+{#- System prompt supports text content or text chunks. #}
+{%- if messages[0]['role'] == 'system' %}
+    {{- '[SYSTEM_PROMPT]' -}}
+    {%- if messages[0]['content'] is string %}
+        {{- messages[0]['content'] -}}
+    {%- else %}        
+        {%- for block in messages[0]['content'] %}
+            {%- if block['type'] == 'text' %}
+                {{- block['text'] }}
+            {%- else %}
+                {{- raise_exception('Only text chunks are supported in system message contents.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- endif %}
+    {{- '[/SYSTEM_PROMPT]' -}}
+    {%- set loop_messages = messages[1:] %}
+{%- else %}
+    {%- set loop_messages = messages %}
+    {%- if default_system_message != '' %}
+        {{- '[SYSTEM_PROMPT]' + default_system_message + '[/SYSTEM_PROMPT]' }}
+    {%- endif %}
+{%- endif %}
+
+
+{#- Tools definition #}
+{%- set tools_definition = '' %}
+{%- set has_tools = false %}
+{%- if tools is defined and tools is not none and tools|length > 0 %}
+    {%- set has_tools = true %}
+    {%- set tools_definition = '[AVAILABLE_TOOLS]' + (tools| tojson) + '[/AVAILABLE_TOOLS]' %}
+    {{- tools_definition }}
+{%- endif %}
+
+{#- Checks for alternating user/assistant messages. #}
+{%- set ns = namespace(index=0) %}
+{%- for message in loop_messages %}
+    {%- if message.role == 'user' or (message.role == 'assistant' and (message.tool_calls is not defined or message.tool_calls is none or message.tool_calls | length == 0)) %}
+        {%- if (message['role'] == 'user') != (ns.index % 2 == 0) %}
+            {{- raise_exception('After the optional system message, conversation roles must alternate user and assistant roles except for tool calls and results.') }}
+        {%- endif %}
+        {%- set ns.index = ns.index + 1 %}
+    {%- endif %}
+{%- endfor %}
+
+{#- Handle conversation messages. #}
+{%- for message in loop_messages %}
+
+    {#- User messages supports text content or text and image chunks. #}
+    {%- if message['role'] == 'user' %}
+        {%- if message['content'] is string %}
+            {{- '[INST]' + message['content'] + '[/INST]' }}
+        {%- elif message['content'] | length > 0 %}
+            {{- '[INST]' }}
+            {%- if message['content'] | length == 2 %}
+                {%- set blocks = message['content'] | sort(attribute='type') %}
+            {%- else %}
+                {%- set blocks = message['content'] %}
+            {%- endif %}
+            {%- for block in blocks %}
+                {%- if block['type'] == 'text' %}
+                    {{- block['text'] }}
+                {%- elif block['type'] in ['image', 'image_url'] %}
+                    {{- '[IMG]' }}
+                {%- else %}
+                    {{- raise_exception('Only text, image and image_url chunks are supported in user message content.') }}
+                {%- endif %}
+            {%- endfor %}
+            {{- '[/INST]' }}
+        {%- else %}
+            {{- raise_exception('User message must have a string or a list of chunks in content') }}
+        {%- endif %}
+
+    {#- Assistant messages supports text content or text and image chunks. #}
+    {%- elif message['role'] == 'assistant' %}
+        {%- if (message['content'] is none or message['content'] == '' or message['content']|length == 0) and (message['tool_calls'] is not defined or message['tool_calls'] is none or message['tool_calls']|length == 0) %}
+            {{- raise_exception('Assistant message must have a string or a list of chunks in content or a list of tool calls.') }}
+        {%- endif %}
+
+        {%- if message['content'] is string %}
+            {{- message['content'] }}
+        {%- elif message['content'] | length > 0 %}
+            {%- for block in message['content'] %}
+                {%- if block['type'] == 'text' %}
+                    {{- block['text'] }}
+                {%- else %}
+                    {{- raise_exception('Only text chunks are supported in assistant message contents.') }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        
+        {%- if message['tool_calls'] is defined and message['tool_calls'] is not none and message['tool_calls']|length > 0 %}
+            {%- for tool in message['tool_calls'] %}
+                {%- set arguments = tool['function']['arguments'] %}
+                {%- if arguments is not string %}
+                    {%- set arguments = arguments|tojson|safe %}
+                {%- elif arguments == '' %}
+                    {%- set arguments = '{}' %}
+                {%- endif %}
+                {{- '[TOOL_CALLS]' + tool['function']['name'] + '[ARGS]' + arguments }}
+            {%- endfor %}
+        {%- endif %}
+
+        {#- End of sequence token for each assistant messages. #}
+        {{- eos_token }}
+
+    {#- Tool messages only supports text content. #}
+    {%- elif message['role'] == 'tool' %}
+        {{- '[TOOL_RESULTS]' + message['content']|string + '[/TOOL_RESULTS]' }}
+
+    {#- Raise exception for unsupported roles. #}
+    {%- else %}
+        {{- raise_exception('Only user, assistant and tool roles are supported, got ' + message['role']) }}
+    {%- endif %}
+{%- endfor %}

checkpoint-1497/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d4f147f46ba382b41fb77da9b98ae40ec5b1486abb974c9d5ac8273ca2279e20
+size 197964694

checkpoint-1497/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8afd33cc5160947cad77a8f0c2339139b5795af18fb88001ef3939e90cb5c0bd
+size 14244

checkpoint-1497/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1cd8f8c0e7f4f45006b76eec3c635a5757261af3c8e489b87fc7e84f1fd48b86
+size 1064

checkpoint-1497/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8fbe698063980a09a4487ec5bbcc545aab380d686f9d918cad649bb49d257f83
+size 17078265

checkpoint-1497/tokenizer_config.json

@@ -0,0 +1,15 @@
+{
+  "add_prefix_space": null,
+  "backend": "tokenizers",
+  "bos_token": "<s>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "</s>",
+  "is_local": true,
+  "legacy": true,
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<pad>",
+  "processor_class": "PixtralProcessor",
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}

checkpoint-1497/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ee28670c78b80e386d67006aee3880339b84b20bda4953e363dae2a4315c97a5
+size 6520

checkpoint-500/README.md

@@ -0,0 +1,209 @@
+---
+base_model: mistralai/Ministral-3-3B-Instruct-2512-BF16
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:mistralai/Ministral-3-3B-Instruct-2512-BF16
+- 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 0.18.1

checkpoint-500/adapter_config.json

@@ -0,0 +1,46 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "mistralai/Ministral-3-3B-Instruct-2512-BF16",
+  "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": 64,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 32,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "v_proj",
+    "down_proj",
+    "gate_proj",
+    "up_proj",
+    "k_proj",
+    "o_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

checkpoint-500/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8ef4cc2bbf1c483aa9f2043d0053f20209468eadfa8346ddd1f5c34d0da6d5d4
+size 135113744

checkpoint-500/chat_template.jinja

@@ -0,0 +1,121 @@
+{#- Default system message if no system prompt is passed. #}
+{%- set default_system_message = 'You are Ministral-3-3B-Instruct-2512, a Large Language Model (LLM) created by Mistral AI, a French startup headquartered in Paris.\nYou power an AI assistant called Le Chat.\nYour knowledge base was last updated on 2023-10-01.\nThe current date is {today}.\n\nWhen you\'re not sure about some information or when the user\'s request requires up-to-date or specific data, you must use the available tools to fetch the information. Do not hesitate to use tools whenever they can provide a more accurate or complete response. If no relevant tools are available, then clearly state that you don\'t have the information and avoid making up anything.\nIf the user\'s question is not clear, ambiguous, or does not provide enough context for you to accurately answer the question, you do not try to answer it right away and you rather ask the user to clarify their request (e.g. "What are some good restaurants around me?" => "Where are you?" or "When is the next flight to Tokyo" => "Where do you travel from?").\nYou are always very attentive to dates, in particular you try to resolve dates (e.g. "yesterday" is {yesterday}) and when asked about information at specific dates, you discard information that is at another date.\nYou follow these instructions in all languages, and always respond to the user in the language they use or request.\nNext sections describe the capabilities that you have.\n\n# WEB BROWSING INSTRUCTIONS\n\nYou cannot perform any web search or access internet to open URLs, links etc. If it seems like the user is expecting you to do so, you clarify the situation and ask the user to copy paste the text directly in the chat.\n\n# MULTI-MODAL INSTRUCTIONS\n\nYou have the ability to read images, but you cannot generate images. You also cannot transcribe audio files or videos.\nYou cannot read nor transcribe audio files or videos.\n\n# TOOL CALLING INSTRUCTIONS\n\nYou may have access to tools that you can use to fetch information or perform actions. You must use these tools in the following situations:\n\n1. When the request requires up-to-date information.\n2. When the request requires specific data that you do not have in your knowledge base.\n3. When the request involves actions that you cannot perform without tools.\n\nAlways prioritize using tools to provide the most accurate and helpful response. If tools are not available, inform the user that you cannot perform the requested action at the moment.' %}
+
+{#- Begin of sequence token. #}
+{{- bos_token }}
+
+{#- Handle system prompt if it exists. #}
+{#- System prompt supports text content or text chunks. #}
+{%- if messages[0]['role'] == 'system' %}
+    {{- '[SYSTEM_PROMPT]' -}}
+    {%- if messages[0]['content'] is string %}
+        {{- messages[0]['content'] -}}
+    {%- else %}        
+        {%- for block in messages[0]['content'] %}
+            {%- if block['type'] == 'text' %}
+                {{- block['text'] }}
+            {%- else %}
+                {{- raise_exception('Only text chunks are supported in system message contents.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- endif %}
+    {{- '[/SYSTEM_PROMPT]' -}}
+    {%- set loop_messages = messages[1:] %}
+{%- else %}
+    {%- set loop_messages = messages %}
+    {%- if default_system_message != '' %}
+        {{- '[SYSTEM_PROMPT]' + default_system_message + '[/SYSTEM_PROMPT]' }}
+    {%- endif %}
+{%- endif %}
+
+
+{#- Tools definition #}
+{%- set tools_definition = '' %}
+{%- set has_tools = false %}
+{%- if tools is defined and tools is not none and tools|length > 0 %}
+    {%- set has_tools = true %}
+    {%- set tools_definition = '[AVAILABLE_TOOLS]' + (tools| tojson) + '[/AVAILABLE_TOOLS]' %}
+    {{- tools_definition }}
+{%- endif %}
+
+{#- Checks for alternating user/assistant messages. #}
+{%- set ns = namespace(index=0) %}
+{%- for message in loop_messages %}
+    {%- if message.role == 'user' or (message.role == 'assistant' and (message.tool_calls is not defined or message.tool_calls is none or message.tool_calls | length == 0)) %}
+        {%- if (message['role'] == 'user') != (ns.index % 2 == 0) %}
+            {{- raise_exception('After the optional system message, conversation roles must alternate user and assistant roles except for tool calls and results.') }}
+        {%- endif %}
+        {%- set ns.index = ns.index + 1 %}
+    {%- endif %}
+{%- endfor %}
+
+{#- Handle conversation messages. #}
+{%- for message in loop_messages %}
+
+    {#- User messages supports text content or text and image chunks. #}
+    {%- if message['role'] == 'user' %}
+        {%- if message['content'] is string %}
+            {{- '[INST]' + message['content'] + '[/INST]' }}
+        {%- elif message['content'] | length > 0 %}
+            {{- '[INST]' }}
+            {%- if message['content'] | length == 2 %}
+                {%- set blocks = message['content'] | sort(attribute='type') %}
+            {%- else %}
+                {%- set blocks = message['content'] %}
+            {%- endif %}
+            {%- for block in blocks %}
+                {%- if block['type'] == 'text' %}
+                    {{- block['text'] }}
+                {%- elif block['type'] in ['image', 'image_url'] %}
+                    {{- '[IMG]' }}
+                {%- else %}
+                    {{- raise_exception('Only text, image and image_url chunks are supported in user message content.') }}
+                {%- endif %}
+            {%- endfor %}
+            {{- '[/INST]' }}
+        {%- else %}
+            {{- raise_exception('User message must have a string or a list of chunks in content') }}
+        {%- endif %}
+
+    {#- Assistant messages supports text content or text and image chunks. #}
+    {%- elif message['role'] == 'assistant' %}
+        {%- if (message['content'] is none or message['content'] == '' or message['content']|length == 0) and (message['tool_calls'] is not defined or message['tool_calls'] is none or message['tool_calls']|length == 0) %}
+            {{- raise_exception('Assistant message must have a string or a list of chunks in content or a list of tool calls.') }}
+        {%- endif %}
+
+        {%- if message['content'] is string %}
+            {{- message['content'] }}
+        {%- elif message['content'] | length > 0 %}
+            {%- for block in message['content'] %}
+                {%- if block['type'] == 'text' %}
+                    {{- block['text'] }}
+                {%- else %}
+                    {{- raise_exception('Only text chunks are supported in assistant message contents.') }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        
+        {%- if message['tool_calls'] is defined and message['tool_calls'] is not none and message['tool_calls']|length > 0 %}
+            {%- for tool in message['tool_calls'] %}
+                {%- set arguments = tool['function']['arguments'] %}
+                {%- if arguments is not string %}
+                    {%- set arguments = arguments|tojson|safe %}
+                {%- elif arguments == '' %}
+                    {%- set arguments = '{}' %}
+                {%- endif %}
+                {{- '[TOOL_CALLS]' + tool['function']['name'] + '[ARGS]' + arguments }}
+            {%- endfor %}
+        {%- endif %}
+
+        {#- End of sequence token for each assistant messages. #}
+        {{- eos_token }}
+
+    {#- Tool messages only supports text content. #}
+    {%- elif message['role'] == 'tool' %}
+        {{- '[TOOL_RESULTS]' + message['content']|string + '[/TOOL_RESULTS]' }}
+
+    {#- Raise exception for unsupported roles. #}
+    {%- else %}
+        {{- raise_exception('Only user, assistant and tool roles are supported, got ' + message['role']) }}
+    {%- endif %}
+{%- endfor %}

checkpoint-500/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8d8e7f906a146ca6d24a0c842d34a984ab4357a80a8fb139ddff08e6adce44a0
+size 197964694

checkpoint-500/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1b92b213b211bbe05e560e8d90e951a0b83e2f44d0e9f9780e747467b7bbd275
+size 14244

checkpoint-500/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:27a642dc07201e7ca576aadd4a094f00aaeed9d3641c508b04f3fa3420d29831
+size 1064

checkpoint-500/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8fbe698063980a09a4487ec5bbcc545aab380d686f9d918cad649bb49d257f83
+size 17078265

checkpoint-500/tokenizer_config.json

@@ -0,0 +1,15 @@
+{
+  "add_prefix_space": null,
+  "backend": "tokenizers",
+  "bos_token": "<s>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "</s>",
+  "is_local": true,
+  "legacy": true,
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<pad>",
+  "processor_class": "PixtralProcessor",
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}

checkpoint-500/trainer_state.json

@@ -0,0 +1,1411 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.33400133600534404,
+  "eval_steps": 500,
+  "global_step": 500,
+  "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": 0.26875,
+      "completions/max_length": 181.2,
+      "completions/max_terminated_length": 141.0,
+      "completions/mean_length": 90.075,
+      "completions/mean_terminated_length": 53.20902862548828,
+      "completions/min_length": 10.6,
+      "completions/min_terminated_length": 10.6,
+      "entropy": 2.674410171434283,
+      "epoch": 0.006680026720106881,
+      "frac_reward_zero_std": 0.125,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 1.6875e-07,
+      "loss": -0.012859511375427245,
+      "num_tokens": 25700.0,
+      "reward": 0.35619792342185974,
+      "reward_std": 0.8869030147790908,
+      "rewards/reward_fn/mean": 0.35619792342185974,
+      "rewards/reward_fn/std": 0.8869030192494393,
+      "step": 10,
+      "step_time": 15.245318131498788
+    },
+    {
+      "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": 0.2625,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 139.6,
+      "completions/mean_length": 85.9875,
+      "completions/mean_terminated_length": 48.72976589202881,
+      "completions/min_length": 4.3,
+      "completions/min_terminated_length": 4.3,
+      "entropy": 2.678321486338973,
+      "epoch": 0.013360053440213761,
+      "frac_reward_zero_std": 0.075,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 3.5625e-07,
+      "loss": -0.11072454452514649,
+      "num_tokens": 49302.0,
+      "reward": 0.2132708251476288,
+      "reward_std": 1.0340060368180275,
+      "rewards/reward_fn/mean": 0.2132708251476288,
+      "rewards/reward_fn/std": 1.0340060606598853,
+      "step": 20,
+      "step_time": 15.825530114897628
+    },
+    {
+      "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": 0.40625,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 115.4,
+      "completions/mean_length": 107.34375,
+      "completions/mean_terminated_length": 49.45969867706299,
+      "completions/min_length": 10.7,
+      "completions/min_terminated_length": 10.7,
+      "entropy": 2.920446590706706,
+      "epoch": 0.02004008016032064,
+      "frac_reward_zero_std": 0.3,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 5.4375e-07,
+      "loss": -0.018171514570713043,
+      "num_tokens": 75689.0,
+      "reward": 0.3708645820617676,
+      "reward_std": 1.0961956143379212,
+      "rewards/reward_fn/mean": 0.3708645820617676,
+      "rewards/reward_fn/std": 1.096195685863495,
+      "step": 30,
+      "step_time": 15.974439114202687
+    },
+    {
+      "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": 0.3625,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 127.0,
+      "completions/mean_length": 100.55,
+      "completions/mean_terminated_length": 49.24266414642334,
+      "completions/min_length": 4.4,
+      "completions/min_terminated_length": 4.4,
+      "entropy": 2.910760021582246,
+      "epoch": 0.026720106880427523,
+      "frac_reward_zero_std": 0.125,
+      "grad_norm": 0.08203125,
+      "learning_rate": 7.3125e-07,
+      "loss": -0.10931088924407958,
+      "num_tokens": 100649.0,
+      "reward": 0.49843747913837433,
+      "reward_std": 1.0061400771141051,
+      "rewards/reward_fn/mean": 0.49843747913837433,
+      "rewards/reward_fn/std": 1.00614013671875,
+      "step": 40,
+      "step_time": 15.9474278408994
+    },
+    {
+      "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": 0.40625,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 121.7,
+      "completions/mean_length": 107.1375,
+      "completions/mean_terminated_length": 47.574611282348634,
+      "completions/min_length": 8.3,
+      "completions/min_terminated_length": 8.3,
+      "entropy": 2.712150690332055,
+      "epoch": 0.033400133600534405,
+      "frac_reward_zero_std": 0.1,
+      "grad_norm": 0.134765625,
+      "learning_rate": 9.187500000000001e-07,
+      "loss": -0.034616786241531375,
+      "num_tokens": 127139.0,
+      "reward": 0.3191354088485241,
+      "reward_std": 1.0306348025798797,
+      "rewards/reward_fn/mean": 0.3191354088485241,
+      "rewards/reward_fn/std": 1.0306348025798797,
+      "step": 50,
+      "step_time": 15.74354040830185
+    },
+    {
+      "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": 0.35,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 153.8,
+      "completions/mean_length": 107.3125,
+      "completions/mean_terminated_length": 61.44384765625,
+      "completions/min_length": 11.9,
+      "completions/min_terminated_length": 11.9,
+      "entropy": 2.746824205853045,
+      "epoch": 0.04008016032064128,
+      "frac_reward_zero_std": 0.1,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 1.10625e-06,
+      "loss": -0.04829469323158264,
+      "num_tokens": 156385.0,
+      "reward": 0.38029164373874663,
+      "reward_std": 1.0828113079071044,
+      "rewards/reward_fn/mean": 0.38029164373874663,
+      "rewards/reward_fn/std": 1.0828113496303557,
+      "step": 60,
+      "step_time": 15.773239352899328
+    },
+    {
+      "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": 0.29375,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 163.9,
+      "completions/mean_length": 97.20625,
+      "completions/mean_terminated_length": 60.003815460205075,
+      "completions/min_length": 11.2,
+      "completions/min_terminated_length": 11.2,
+      "entropy": 2.6204975010827183,
+      "epoch": 0.04676018704074816,
+      "frac_reward_zero_std": 0.2,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 1.2937500000000001e-06,
+      "loss": -0.07032976150512696,
+      "num_tokens": 180486.0,
+      "reward": 0.4105520874261856,
+      "reward_std": 1.023577058315277,
+      "rewards/reward_fn/mean": 0.4105520874261856,
+      "rewards/reward_fn/std": 1.0235771000385285,
+      "step": 70,
+      "step_time": 15.689090457798375
+    },
+    {
+      "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": 0.33125,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 122.5,
+      "completions/mean_length": 94.375,
+      "completions/mean_terminated_length": 47.72081985473633,
+      "completions/min_length": 9.2,
+      "completions/min_terminated_length": 9.2,
+      "entropy": 2.62342791412957,
+      "epoch": 0.053440213760855046,
+      "frac_reward_zero_std": 0.175,
+      "grad_norm": 0.15625,
+      "learning_rate": 1.48125e-06,
+      "loss": -0.025419986248016356,
+      "num_tokens": 205518.0,
+      "reward": 0.19294790625572206,
+      "reward_std": 1.1096562147140503,
+      "rewards/reward_fn/mean": 0.19294790625572206,
+      "rewards/reward_fn/std": 1.1096562445163727,
+      "step": 80,
+      "step_time": 15.946593014401515
+    },
+    {
+      "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": 0.31875,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 134.0,
+      "completions/mean_length": 99.7625,
+      "completions/mean_terminated_length": 55.72392253875732,
+      "completions/min_length": 6.1,
+      "completions/min_terminated_length": 6.1,
+      "entropy": 2.642461443692446,
+      "epoch": 0.06012024048096192,
+      "frac_reward_zero_std": 0.2,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 1.4904728299223714e-06,
+      "loss": 0.02786439061164856,
+      "num_tokens": 232576.0,
+      "reward": 0.20244790613651276,
+      "reward_std": 1.146572184562683,
+      "rewards/reward_fn/mean": 0.20244790613651276,
+      "rewards/reward_fn/std": 1.146572232246399,
+      "step": 90,
+      "step_time": 15.867746153000917
+    },
+    {
+      "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": 0.33125,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 143.8,
+      "completions/mean_length": 93.9375,
+      "completions/mean_terminated_length": 47.145199394226076,
+      "completions/min_length": 2.4,
+      "completions/min_terminated_length": 2.4,
+      "entropy": 2.976877874881029,
+      "epoch": 0.06680026720106881,
+      "frac_reward_zero_std": 0.05,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 1.4798870853916728e-06,
+      "loss": -0.14498503208160402,
+      "num_tokens": 257582.0,
+      "reward": 0.14554165303707123,
+      "reward_std": 1.1003411173820496,
+      "rewards/reward_fn/mean": 0.14554165303707123,
+      "rewards/reward_fn/std": 1.100341135263443,
+      "step": 100,
+      "step_time": 15.938584135997008
+    },
+    {
+      "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": 0.275,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 155.4,
+      "completions/mean_length": 90.33125,
+      "completions/mean_terminated_length": 53.50349349975586,
+      "completions/min_length": 1.7,
+      "completions/min_terminated_length": 1.7,
+      "entropy": 2.8531898425891997,
+      "epoch": 0.07348029392117568,
+      "frac_reward_zero_std": 0.075,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 1.469301340860974e-06,
+      "loss": -0.12909494638442992,
+      "num_tokens": 284727.0,
+      "reward": 0.0315208375453949,
+      "reward_std": 1.1703906178474426,
+      "rewards/reward_fn/mean": 0.0315208375453949,
+      "rewards/reward_fn/std": 1.1703906536102295,
+      "step": 110,
+      "step_time": 16.003242681801204
+    },
+    {
+      "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": 0.23125,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 143.8,
+      "completions/mean_length": 83.6125,
+      "completions/mean_terminated_length": 51.575297355651855,
+      "completions/min_length": 1.5,
+      "completions/min_terminated_length": 1.5,
+      "entropy": 2.7128803845494986,
+      "epoch": 0.08016032064128256,
+      "frac_reward_zero_std": 0.175,
+      "grad_norm": 0.07275390625,
+      "learning_rate": 1.4587155963302754e-06,
+      "loss": -0.06841488480567932,
+      "num_tokens": 310509.0,
+      "reward": 0.25534376204013826,
+      "reward_std": 1.08806232213974,
+      "rewards/reward_fn/mean": 0.25534376204013826,
+      "rewards/reward_fn/std": 1.0880623579025268,
+      "step": 120,
+      "step_time": 16.015716333096133
+    },
+    {
+      "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": 0.2625,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 145.1,
+      "completions/mean_length": 85.6625,
+      "completions/mean_terminated_length": 48.96218070983887,
+      "completions/min_length": 6.1,
+      "completions/min_terminated_length": 6.1,
+      "entropy": 2.6464409498497843,
+      "epoch": 0.08684034736138944,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.095703125,
+      "learning_rate": 1.4481298517995768e-06,
+      "loss": 0.07086914777755737,
+      "num_tokens": 334067.0,
+      "reward": 0.17779166698455812,
+      "reward_std": 1.0640078246593476,
+      "rewards/reward_fn/mean": 0.17779166698455812,
+      "rewards/reward_fn/std": 1.064007866382599,
+      "step": 130,
+      "step_time": 15.856749649995617
+    },
+    {
+      "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": 0.34375,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 134.3,
+      "completions/mean_length": 95.675,
+      "completions/mean_terminated_length": 45.25946407318115,
+      "completions/min_length": 6.9,
+      "completions/min_terminated_length": 6.9,
+      "entropy": 2.8085414838045835,
+      "epoch": 0.09352037408149633,
+      "frac_reward_zero_std": 0.125,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 1.4375441072688781e-06,
+      "loss": -0.04502600431442261,
+      "num_tokens": 360675.0,
+      "reward": 0.07731249704957008,
+      "reward_std": 1.0695590198040008,
+      "rewards/reward_fn/mean": 0.07731249704957008,
+      "rewards/reward_fn/std": 1.0695590734481812,
+      "step": 140,
+      "step_time": 15.795801008101625
+    },
+    {
+      "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": 0.3125,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 121.1,
+      "completions/mean_length": 91.0625,
+      "completions/mean_terminated_length": 47.073708152771,
+      "completions/min_length": 6.0,
+      "completions/min_terminated_length": 6.0,
+      "entropy": 2.8401680497452615,
+      "epoch": 0.10020040080160321,
+      "frac_reward_zero_std": 0.075,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 1.4269583627381793e-06,
+      "loss": -0.07802374362945556,
+      "num_tokens": 385949.0,
+      "reward": 0.03469790667295456,
+      "reward_std": 1.0852368116378783,
+      "rewards/reward_fn/mean": 0.03469790667295456,
+      "rewards/reward_fn/std": 1.0852368474006653,
+      "step": 150,
+      "step_time": 15.794295253699238
+    },
+    {
+      "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": 0.35,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 148.5,
+      "completions/mean_length": 104.26875,
+      "completions/mean_terminated_length": 59.56922569274902,
+      "completions/min_length": 12.2,
+      "completions/min_terminated_length": 12.2,
+      "entropy": 2.7122420724481344,
+      "epoch": 0.10688042752171009,
+      "frac_reward_zero_std": 0.125,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 1.4163726182074806e-06,
+      "loss": -0.05119783282279968,
+      "num_tokens": 412428.0,
+      "reward": 0.17459374517202378,
+      "reward_std": 1.0888913512229919,
+      "rewards/reward_fn/mean": 0.17459374517202378,
+      "rewards/reward_fn/std": 1.0888914048671723,
+      "step": 160,
+      "step_time": 15.665904332897117
+    },
+    {
+      "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": 0.3625,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 144.4,
+      "completions/mean_length": 103.55625,
+      "completions/mean_terminated_length": 54.23816108703613,
+      "completions/min_length": 3.3,
+      "completions/min_terminated_length": 3.3,
+      "entropy": 2.6974239563569427,
+      "epoch": 0.11356045424181696,
+      "frac_reward_zero_std": 0.1,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 1.405786873676782e-06,
+      "loss": -0.047921139001846316,
+      "num_tokens": 440797.0,
+      "reward": 0.14482289999723436,
+      "reward_std": 1.1464063763618468,
+      "rewards/reward_fn/mean": 0.14482289999723436,
+      "rewards/reward_fn/std": 1.146406388282776,
+      "step": 170,
+      "step_time": 15.801852633895397
+    },
+    {
+      "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": 0.3125,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 144.6,
+      "completions/mean_length": 93.09375,
+      "completions/mean_terminated_length": 49.29199752807617,
+      "completions/min_length": 6.5,
+      "completions/min_terminated_length": 6.5,
+      "entropy": 2.9900201810523868,
+      "epoch": 0.12024048096192384,
+      "frac_reward_zero_std": 0.075,
+      "grad_norm": 0.146484375,
+      "learning_rate": 1.3952011291460833e-06,
+      "loss": -0.12824782133102416,
+      "num_tokens": 463188.0,
+      "reward": 0.12326040863990784,
+      "reward_std": 1.0692023456096649,
+      "rewards/reward_fn/mean": 0.12326040863990784,
+      "rewards/reward_fn/std": 1.0692023813724518,
+      "step": 180,
+      "step_time": 15.808984902502198
+    },
+    {
+      "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": 0.38125,
+      "completions/max_length": 187.6,
+      "completions/max_terminated_length": 143.9,
+      "completions/mean_length": 103.05,
+      "completions/mean_terminated_length": 49.861496162414554,
+      "completions/min_length": 5.7,
+      "completions/min_terminated_length": 5.7,
+      "entropy": 2.757765106856823,
+      "epoch": 0.12692050768203073,
+      "frac_reward_zero_std": 0.075,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 1.3846153846153846e-06,
+      "loss": -0.05187699794769287,
+      "num_tokens": 492296.0,
+      "reward": 0.14946874752640724,
+      "reward_std": 1.1358891725540161,
+      "rewards/reward_fn/mean": 0.14946874752640724,
+      "rewards/reward_fn/std": 1.135889208316803,
+      "step": 190,
+      "step_time": 15.582480278503136
+    },
+    {
+      "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": 0.26875,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 154.9,
+      "completions/mean_length": 89.9,
+      "completions/mean_terminated_length": 51.897169494628905,
+      "completions/min_length": 4.1,
+      "completions/min_terminated_length": 4.1,
+      "entropy": 2.6228192538022994,
+      "epoch": 0.13360053440213762,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.158203125,
+      "learning_rate": 1.374029640084686e-06,
+      "loss": -0.08793725967407226,
+      "num_tokens": 516356.0,
+      "reward": 0.3006145887076855,
+      "reward_std": 1.1243340432643891,
+      "rewards/reward_fn/mean": 0.3006145887076855,
+      "rewards/reward_fn/std": 1.1243340492248535,
+      "step": 200,
+      "step_time": 15.928501791401505
+    },
+    {
+      "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": 0.33125,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 133.1,
+      "completions/mean_length": 98.15625,
+      "completions/mean_terminated_length": 52.321857643127444,
+      "completions/min_length": 6.0,
+      "completions/min_terminated_length": 6.0,
+      "entropy": 2.698680349998176,
+      "epoch": 0.1402805611222445,
+      "frac_reward_zero_std": 0.075,
+      "grad_norm": 0.130859375,
+      "learning_rate": 1.3634438955539873e-06,
+      "loss": -0.034938347339630124,
+      "num_tokens": 541793.0,
+      "reward": 0.21104166135191918,
+      "reward_std": 1.0634925305843352,
+      "rewards/reward_fn/mean": 0.21104166135191918,
+      "rewards/reward_fn/std": 1.063492578268051,
+      "step": 210,
+      "step_time": 15.917452003002108
+    },
+    {
+      "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": 0.2875,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 123.2,
+      "completions/mean_length": 88.0875,
+      "completions/mean_terminated_length": 47.874640464782715,
+      "completions/min_length": 2.6,
+      "completions/min_terminated_length": 2.6,
+      "entropy": 2.815653139166534,
+      "epoch": 0.14696058784235136,
+      "frac_reward_zero_std": 0.175,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 1.3528581510232887e-06,
+      "loss": -0.07358459830284118,
+      "num_tokens": 566039.0,
+      "reward": 0.3140833258628845,
+      "reward_std": 1.054498302936554,
+      "rewards/reward_fn/mean": 0.3140833258628845,
+      "rewards/reward_fn/std": 1.0544983208179475,
+      "step": 220,
+      "step_time": 15.865132569494017
+    },
+    {
+      "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": 0.33125,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 140.9,
+      "completions/mean_length": 99.8,
+      "completions/mean_terminated_length": 56.69819946289063,
+      "completions/min_length": 3.5,
+      "completions/min_terminated_length": 3.5,
+      "entropy": 2.598500589840114,
+      "epoch": 0.15364061456245826,
+      "frac_reward_zero_std": 0.125,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 1.34227240649259e-06,
+      "loss": 0.06172789335250854,
+      "num_tokens": 593343.0,
+      "reward": 0.4534791558980942,
+      "reward_std": 0.9242997214198112,
+      "rewards/reward_fn/mean": 0.4534791558980942,
+      "rewards/reward_fn/std": 0.9242997393012047,
+      "step": 230,
+      "step_time": 15.902188144398679
+    },
+    {
+      "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": 0.2875,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 167.9,
+      "completions/mean_length": 96.125,
+      "completions/mean_terminated_length": 60.7716682434082,
+      "completions/min_length": 9.8,
+      "completions/min_terminated_length": 9.8,
+      "entropy": 2.4708725826814772,
+      "epoch": 0.16032064128256512,
+      "frac_reward_zero_std": 0.15,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 1.3316866619618913e-06,
+      "loss": 0.05918135643005371,
+      "num_tokens": 619827.0,
+      "reward": -0.016895844042301177,
+      "reward_std": 1.1622604727745056,
+      "rewards/reward_fn/mean": -0.016895844042301177,
+      "rewards/reward_fn/std": 1.1622605085372926,
+      "step": 240,
+      "step_time": 15.943352174599568
+    },
+    {
+      "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": 0.35,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 124.2,
+      "completions/mean_length": 97.46875,
+      "completions/mean_terminated_length": 48.5021728515625,
+      "completions/min_length": 10.3,
+      "completions/min_terminated_length": 10.3,
+      "entropy": 2.844438760727644,
+      "epoch": 0.16700066800267202,
+      "frac_reward_zero_std": 0.125,
+      "grad_norm": 0.1484375,
+      "learning_rate": 1.3211009174311927e-06,
+      "loss": -0.03755800724029541,
+      "num_tokens": 646586.0,
+      "reward": 0.28715626522898674,
+      "reward_std": 0.9988754749298095,
+      "rewards/reward_fn/mean": 0.28715626522898674,
+      "rewards/reward_fn/std": 0.9988755285739899,
+      "step": 250,
+      "step_time": 15.891257435495936
+    },
+    {
+      "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": 0.4375,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 155.9,
+      "completions/mean_length": 116.375,
+      "completions/mean_terminated_length": 56.563998794555665,
+      "completions/min_length": 6.3,
+      "completions/min_terminated_length": 6.3,
+      "entropy": 2.863031363673508,
+      "epoch": 0.1736806947227789,
+      "frac_reward_zero_std": 0.2,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 1.310515172900494e-06,
+      "loss": -0.029664209485054015,
+      "num_tokens": 672174.0,
+      "reward": 0.5583645716309548,
+      "reward_std": 0.9518099159002305,
+      "rewards/reward_fn/mean": 0.5583645716309548,
+      "rewards/reward_fn/std": 0.9518099725246429,
+      "step": 260,
+      "step_time": 15.782717786495777
+    },
+    {
+      "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": 0.41875,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 124.8,
+      "completions/mean_length": 108.55,
+      "completions/mean_terminated_length": 48.82083435058594,
+      "completions/min_length": 10.2,
+      "completions/min_terminated_length": 10.2,
+      "entropy": 3.0812535148113964,
+      "epoch": 0.18036072144288579,
+      "frac_reward_zero_std": 0.075,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 1.2999294283697954e-06,
+      "loss": -0.060126584768295285,
+      "num_tokens": 702254.0,
+      "reward": 0.08691668063402176,
+      "reward_std": 1.124253910779953,
+      "rewards/reward_fn/mean": 0.08691668063402176,
+      "rewards/reward_fn/std": 1.1242539584636688,
+      "step": 270,
+      "step_time": 15.878148274200793
+    },
+    {
+      "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": 0.2625,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 142.3,
+      "completions/mean_length": 93.35,
+      "completions/mean_terminated_length": 58.149365234375,
+      "completions/min_length": 6.3,
+      "completions/min_terminated_length": 6.3,
+      "entropy": 2.482130799628794,
+      "epoch": 0.18704074816299265,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.14453125,
+      "learning_rate": 1.2893436838390967e-06,
+      "loss": -0.10814018249511718,
+      "num_tokens": 731318.0,
+      "reward": 0.40751040875911715,
+      "reward_std": 1.029860746860504,
+      "rewards/reward_fn/mean": 0.40751040875911715,
+      "rewards/reward_fn/std": 1.029860782623291,
+      "step": 280,
+      "step_time": 15.845484773800127
+    },
+    {
+      "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": 0.4375,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 134.3,
+      "completions/mean_length": 113.9,
+      "completions/mean_terminated_length": 55.71571807861328,
+      "completions/min_length": 12.9,
+      "completions/min_terminated_length": 12.9,
+      "entropy": 3.084173835068941,
+      "epoch": 0.19372077488309952,
+      "frac_reward_zero_std": 0.175,
+      "grad_norm": 0.162109375,
+      "learning_rate": 1.278757939308398e-06,
+      "loss": -0.030109062790870667,
+      "num_tokens": 756718.0,
+      "reward": 0.09826041013002396,
+      "reward_std": 1.0646479148417711,
+      "rewards/reward_fn/mean": 0.09826041013002396,
+      "rewards/reward_fn/std": 1.064647940546274,
+      "step": 290,
+      "step_time": 15.918558336896968
+    },
+    {
+      "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": 0.36875,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 142.2,
+      "completions/mean_length": 106.1,
+      "completions/mean_terminated_length": 57.426274108886716,
+      "completions/min_length": 5.1,
+      "completions/min_terminated_length": 5.1,
+      "entropy": 2.7191914174705745,
+      "epoch": 0.20040080160320642,
+      "frac_reward_zero_std": 0.1,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 1.2681721947776994e-06,
+      "loss": -0.05046193599700928,
+      "num_tokens": 788326.0,
+      "reward": 0.27198959290981295,
+      "reward_std": 1.0938624262809753,
+      "rewards/reward_fn/mean": 0.27198959290981295,
+      "rewards/reward_fn/std": 1.0938624918460846,
+      "step": 300,
+      "step_time": 15.950138682403486
+    },
+    {
+      "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": 0.4,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 130.5,
+      "completions/mean_length": 110.725,
+      "completions/mean_terminated_length": 55.914433288574216,
+      "completions/min_length": 12.6,
+      "completions/min_terminated_length": 12.6,
+      "entropy": 2.7512730032205583,
+      "epoch": 0.2070808283233133,
+      "frac_reward_zero_std": 0.125,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 1.2575864502470007e-06,
+      "loss": 0.004815106093883514,
+      "num_tokens": 815894.0,
+      "reward": 0.2241875022649765,
+      "reward_std": 1.0307687520980835,
+      "rewards/reward_fn/mean": 0.2241875022649765,
+      "rewards/reward_fn/std": 1.0307688117027283,
+      "step": 310,
+      "step_time": 15.88405146629375
+    },
+    {
+      "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": 0.36875,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 134.4,
+      "completions/mean_length": 104.75625,
+      "completions/mean_terminated_length": 54.86628341674805,
+      "completions/min_length": 4.7,
+      "completions/min_terminated_length": 4.7,
+      "entropy": 2.9182289419695735,
+      "epoch": 0.21376085504342018,
+      "frac_reward_zero_std": 0.025,
+      "grad_norm": 0.291015625,
+      "learning_rate": 1.247000705716302e-06,
+      "loss": -0.013232138752937318,
+      "num_tokens": 846603.0,
+      "reward": 0.07972916066646576,
+      "reward_std": 1.1528971672058106,
+      "rewards/reward_fn/mean": 0.07972916066646576,
+      "rewards/reward_fn/std": 1.152897197008133,
+      "step": 320,
+      "step_time": 15.879019064803288
+    },
+    {
+      "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": 0.3375,
+      "completions/max_length": 180.1,
+      "completions/max_terminated_length": 115.8,
+      "completions/mean_length": 97.09375,
+      "completions/mean_terminated_length": 48.5493278503418,
+      "completions/min_length": 7.0,
+      "completions/min_terminated_length": 7.0,
+      "entropy": 2.6609113456681373,
+      "epoch": 0.22044088176352705,
+      "frac_reward_zero_std": 0.1,
+      "grad_norm": 0.111328125,
+      "learning_rate": 1.2364149611856034e-06,
+      "loss": -0.09263116121292114,
+      "num_tokens": 869686.0,
+      "reward": 0.22312501221895217,
+      "reward_std": 1.1098669052124024,
+      "rewards/reward_fn/mean": 0.22312501221895217,
+      "rewards/reward_fn/std": 1.1098668992519378,
+      "step": 330,
+      "step_time": 15.147548897802334
+    },
+    {
+      "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": 0.28125,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 163.0,
+      "completions/mean_length": 94.7875,
+      "completions/mean_terminated_length": 57.240910339355466,
+      "completions/min_length": 3.3,
+      "completions/min_terminated_length": 3.3,
+      "entropy": 2.5511690624058248,
+      "epoch": 0.22712090848363392,
+      "frac_reward_zero_std": 0.15,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 1.2258292166549048e-06,
+      "loss": -0.0902456521987915,
+      "num_tokens": 894584.0,
+      "reward": 0.4284583508968353,
+      "reward_std": 1.0353572845458985,
+      "rewards/reward_fn/mean": 0.4284583508968353,
+      "rewards/reward_fn/std": 1.0353573381900787,
+      "step": 340,
+      "step_time": 15.964643920902745
+    },
+    {
+      "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": 0.33125,
+      "completions/max_length": 184.2,
+      "completions/max_terminated_length": 122.0,
+      "completions/mean_length": 91.99375,
+      "completions/mean_terminated_length": 44.41522560119629,
+      "completions/min_length": 8.1,
+      "completions/min_terminated_length": 8.1,
+      "entropy": 2.6959754671901464,
+      "epoch": 0.23380093520374082,
+      "frac_reward_zero_std": 0.1,
+      "grad_norm": 0.119140625,
+      "learning_rate": 1.2152434721242061e-06,
+      "loss": 0.010559720546007156,
+      "num_tokens": 924415.0,
+      "reward": 0.12816667780280114,
+      "reward_std": 1.1092212915420532,
+      "rewards/reward_fn/mean": 0.12816667780280114,
+      "rewards/reward_fn/std": 1.1092213094234467,
+      "step": 350,
+      "step_time": 15.440228611296334
+    },
+    {
+      "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": 0.34375,
+      "completions/max_length": 187.5,
+      "completions/max_terminated_length": 153.3,
+      "completions/mean_length": 98.63125,
+      "completions/mean_terminated_length": 49.16804809570313,
+      "completions/min_length": 5.4,
+      "completions/min_terminated_length": 5.4,
+      "entropy": 3.016822088509798,
+      "epoch": 0.24048096192384769,
+      "frac_reward_zero_std": 0.175,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 1.2046577275935075e-06,
+      "loss": -0.009723667800426484,
+      "num_tokens": 950400.0,
+      "reward": 0.2331979066133499,
+      "reward_std": 1.0595607578754425,
+      "rewards/reward_fn/mean": 0.2331979066133499,
+      "rewards/reward_fn/std": 1.059560775756836,
+      "step": 360,
+      "step_time": 15.58997365470059
+    },
+    {
+      "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": 0.3,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 140.7,
+      "completions/mean_length": 98.74375,
+      "completions/mean_terminated_length": 59.88666248321533,
+      "completions/min_length": 10.4,
+      "completions/min_terminated_length": 10.4,
+      "entropy": 2.751246392726898,
+      "epoch": 0.24716098864395458,
+      "frac_reward_zero_std": 0.075,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 1.1940719830628088e-06,
+      "loss": -0.08575952053070068,
+      "num_tokens": 978815.0,
+      "reward": 0.15954167246818543,
+      "reward_std": 1.111803650856018,
+      "rewards/reward_fn/mean": 0.15954167246818543,
+      "rewards/reward_fn/std": 1.1118036925792694,
+      "step": 370,
+      "step_time": 16.093606699199153
+    },
+    {
+      "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": 0.2875,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 138.3,
+      "completions/mean_length": 96.49375,
+      "completions/mean_terminated_length": 57.97613124847412,
+      "completions/min_length": 7.6,
+      "completions/min_terminated_length": 7.6,
+      "entropy": 2.564889958500862,
+      "epoch": 0.25384101536406145,
+      "frac_reward_zero_std": 0.075,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 1.1834862385321102e-06,
+      "loss": -0.006212610751390457,
+      "num_tokens": 1005498.0,
+      "reward": 0.3756875067949295,
+      "reward_std": 1.0436641812324523,
+      "rewards/reward_fn/mean": 0.3756875067949295,
+      "rewards/reward_fn/std": 1.0436642050743103,
+      "step": 380,
+      "step_time": 16.12421152170209
+    },
+    {
+      "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": 0.30625,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 151.9,
+      "completions/mean_length": 99.2875,
+      "completions/mean_terminated_length": 60.877184295654295,
+      "completions/min_length": 11.8,
+      "completions/min_terminated_length": 11.8,
+      "entropy": 2.5775781558826565,
+      "epoch": 0.2605210420841683,
+      "frac_reward_zero_std": 0.075,
+      "grad_norm": 0.189453125,
+      "learning_rate": 1.1729004940014115e-06,
+      "loss": -0.0865351676940918,
+      "num_tokens": 1034348.0,
+      "reward": 0.25194790959358215,
+      "reward_std": 1.0172604084014893,
+      "rewards/reward_fn/mean": 0.25194790959358215,
+      "rewards/reward_fn/std": 1.0172604203224183,
+      "step": 390,
+      "step_time": 15.996389424398512
+    },
+    {
+      "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": 0.3625,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 120.7,
+      "completions/mean_length": 101.30625,
+      "completions/mean_terminated_length": 49.48569221496582,
+      "completions/min_length": 3.5,
+      "completions/min_terminated_length": 3.5,
+      "entropy": 2.823385216109455,
+      "epoch": 0.26720106880427524,
+      "frac_reward_zero_std": 0.1,
+      "grad_norm": 0.111328125,
+      "learning_rate": 1.1623147494707128e-06,
+      "loss": -0.0563660979270935,
+      "num_tokens": 1062181.0,
+      "reward": 0.3629791595041752,
+      "reward_std": 1.0256887555122376,
+      "rewards/reward_fn/mean": 0.3629791595041752,
+      "rewards/reward_fn/std": 1.0256887912750243,
+      "step": 400,
+      "step_time": 16.02646716330637
+    },
+    {
+      "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": 0.35,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 144.8,
+      "completions/mean_length": 106.41875,
+      "completions/mean_terminated_length": 60.527869415283206,
+      "completions/min_length": 11.5,
+      "completions/min_terminated_length": 11.5,
+      "entropy": 2.6901804924011232,
+      "epoch": 0.2738810955243821,
+      "frac_reward_zero_std": 0.05,
+      "grad_norm": 0.140625,
+      "learning_rate": 1.1517290049400142e-06,
+      "loss": -0.0425216406583786,
+      "num_tokens": 1086396.0,
+      "reward": 0.2832083344459534,
+      "reward_std": 1.1361640512943267,
+      "rewards/reward_fn/mean": 0.2832083344459534,
+      "rewards/reward_fn/std": 1.1361640512943267,
+      "step": 410,
+      "step_time": 16.024322779294017
+    },
+    {
+      "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": 0.2875,
+      "completions/max_length": 185.4,
+      "completions/max_terminated_length": 154.8,
+      "completions/mean_length": 93.74375,
+      "completions/mean_terminated_length": 55.715553283691406,
+      "completions/min_length": 4.4,
+      "completions/min_terminated_length": 4.4,
+      "entropy": 2.5821650655008854,
+      "epoch": 0.280561122244489,
+      "frac_reward_zero_std": 0.15,
+      "grad_norm": 0.142578125,
+      "learning_rate": 1.1411432604093155e-06,
+      "loss": -0.061094099283218385,
+      "num_tokens": 1111451.0,
+      "reward": 0.10051042139530182,
+      "reward_std": 1.105427885055542,
+      "rewards/reward_fn/mean": 0.10051042139530182,
+      "rewards/reward_fn/std": 1.1054279208183289,
+      "step": 420,
+      "step_time": 15.741778691902436
+    },
+    {
+      "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": 0.3875,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 142.5,
+      "completions/mean_length": 111.3375,
+      "completions/mean_terminated_length": 60.72189846038818,
+      "completions/min_length": 19.3,
+      "completions/min_terminated_length": 19.3,
+      "entropy": 2.8669227328151465,
+      "epoch": 0.28724114896459585,
+      "frac_reward_zero_std": 0.125,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 1.1305575158786169e-06,
+      "loss": -0.03572247922420502,
+      "num_tokens": 1138769.0,
+      "reward": 0.16211459040641785,
+      "reward_std": 1.073091048002243,
+      "rewards/reward_fn/mean": 0.16211459040641785,
+      "rewards/reward_fn/std": 1.073091071844101,
+      "step": 430,
+      "step_time": 15.911773163802717
+    },
+    {
+      "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": 0.3625,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 161.2,
+      "completions/mean_length": 109.19375,
+      "completions/mean_terminated_length": 63.74071998596192,
+      "completions/min_length": 11.9,
+      "completions/min_terminated_length": 11.9,
+      "entropy": 2.515946627780795,
+      "epoch": 0.2939211756847027,
+      "frac_reward_zero_std": 0.125,
+      "grad_norm": 0.158203125,
+      "learning_rate": 1.1199717713479182e-06,
+      "loss": -0.013631168007850646,
+      "num_tokens": 1169404.0,
+      "reward": 0.15146875381469727,
+      "reward_std": 1.121659505367279,
+      "rewards/reward_fn/mean": 0.15146875381469727,
+      "rewards/reward_fn/std": 1.1216595470905304,
+      "step": 440,
+      "step_time": 16.11628591810295
+    },
+    {
+      "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": 0.38125,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 152.7,
+      "completions/mean_length": 110.65625,
+      "completions/mean_terminated_length": 64.46861190795899,
+      "completions/min_length": 13.4,
+      "completions/min_terminated_length": 13.4,
+      "entropy": 2.85670913644135,
+      "epoch": 0.30060120240480964,
+      "frac_reward_zero_std": 0.125,
+      "grad_norm": 0.07470703125,
+      "learning_rate": 1.1093860268172196e-06,
+      "loss": -0.010854866355657578,
+      "num_tokens": 1198457.0,
+      "reward": -0.0358958289027214,
+      "reward_std": 1.0866289496421815,
+      "rewards/reward_fn/mean": -0.0358958289027214,
+      "rewards/reward_fn/std": 1.0866289854049682,
+      "step": 450,
+      "step_time": 16.119690579098823
+    },
+    {
+      "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": 0.325,
+      "completions/max_length": 188.7,
+      "completions/max_terminated_length": 137.8,
+      "completions/mean_length": 98.55625,
+      "completions/mean_terminated_length": 53.62260208129883,
+      "completions/min_length": 6.1,
+      "completions/min_terminated_length": 6.1,
+      "entropy": 2.561788680776954,
+      "epoch": 0.3072812291249165,
+      "frac_reward_zero_std": 0.1,
+      "grad_norm": 0.14453125,
+      "learning_rate": 1.098800282286521e-06,
+      "loss": -0.03874463438987732,
+      "num_tokens": 1224202.0,
+      "reward": 0.16965624839067459,
+      "reward_std": 1.059354692697525,
+      "rewards/reward_fn/mean": 0.16965624839067459,
+      "rewards/reward_fn/std": 1.059354728460312,
+      "step": 460,
+      "step_time": 15.739375571500569
+    },
+    {
+      "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": 0.2625,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 158.4,
+      "completions/mean_length": 95.03125,
+      "completions/mean_terminated_length": 61.78505249023438,
+      "completions/min_length": 12.4,
+      "completions/min_terminated_length": 12.4,
+      "entropy": 2.6588732454925776,
+      "epoch": 0.3139612558450234,
+      "frac_reward_zero_std": 0.125,
+      "grad_norm": 0.134765625,
+      "learning_rate": 1.0882145377558222e-06,
+      "loss": -0.05535932779312134,
+      "num_tokens": 1246923.0,
+      "reward": 0.19603125378489494,
+      "reward_std": 1.0779352009296417,
+      "rewards/reward_fn/mean": 0.19603125378489494,
+      "rewards/reward_fn/std": 1.0779352366924286,
+      "step": 470,
+      "step_time": 15.847925305603713
+    },
+    {
+      "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": 0.23125,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 143.8,
+      "completions/mean_length": 79.23125,
+      "completions/mean_terminated_length": 45.72937412261963,
+      "completions/min_length": 1.1,
+      "completions/min_terminated_length": 1.1,
+      "entropy": 2.690891198255122,
+      "epoch": 0.32064128256513025,
+      "frac_reward_zero_std": 0.1,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 1.0776287932251236e-06,
+      "loss": -0.12146193981170654,
+      "num_tokens": 1274408.0,
+      "reward": 0.044479166716337205,
+      "reward_std": 1.0702742159366607,
+      "rewards/reward_fn/mean": 0.044479166716337205,
+      "rewards/reward_fn/std": 1.0702742516994477,
+      "step": 480,
+      "step_time": 15.923641223995947
+    },
+    {
+      "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": 0.40625,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 150.8,
+      "completions/mean_length": 113.54375,
+      "completions/mean_terminated_length": 58.41944580078125,
+      "completions/min_length": 9.3,
+      "completions/min_terminated_length": 9.3,
+      "entropy": 2.7810823559761046,
+      "epoch": 0.3273213092852371,
+      "frac_reward_zero_std": 0.075,
+      "grad_norm": 0.15625,
+      "learning_rate": 1.067043048694425e-06,
+      "loss": -0.08153985142707824,
+      "num_tokens": 1302683.0,
+      "reward": 0.47086457461118697,
+      "reward_std": 0.9505439102649689,
+      "rewards/reward_fn/mean": 0.47086457461118697,
+      "rewards/reward_fn/std": 0.9505439460277557,
+      "step": 490,
+      "step_time": 15.770985922595719
+    },
+    {
+      "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": 0.40625,
+      "completions/max_length": 192.0,
+      "completions/max_terminated_length": 116.3,
+      "completions/mean_length": 109.33125,
+      "completions/mean_terminated_length": 55.47568225860596,
+      "completions/min_length": 7.3,
+      "completions/min_terminated_length": 7.3,
+      "entropy": 2.960219758935273,
+      "epoch": 0.33400133600534404,
+      "frac_reward_zero_std": 0.175,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 1.0564573041637263e-06,
+      "loss": -0.053396403789520264,
+      "num_tokens": 1330108.0,
+      "reward": 0.1449270747601986,
+      "reward_std": 1.1236790299415589,
+      "rewards/reward_fn/mean": 0.1449270747601986,
+      "rewards/reward_fn/std": 1.1236790895462037,
+      "step": 500,
+      "step_time": 15.827259837596648
+    },
+    {
+      "epoch": 0.33400133600534404,
+      "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": 0.3165322580645161,
+      "eval_completions/max_length": 180.2741935483871,
+      "eval_completions/max_terminated_length": 113.79032258064517,
+      "eval_completions/mean_length": 98.66129032258064,
+      "eval_completions/mean_terminated_length": 60.57563466410483,
+      "eval_completions/min_length": 22.741935483870968,
+      "eval_completions/min_terminated_length": 22.741935483870968,
+      "eval_entropy": 2.8856024588308027,
+      "eval_frac_reward_zero_std": 0.13709677419354838,
+      "eval_loss": -0.05357525870203972,
+      "eval_num_tokens": 1330108.0,
+      "eval_reward": 0.19787298455353705,
+      "eval_reward_std": 0.9424176115182138,
+      "eval_rewards/reward_fn/mean": 0.19787298455353705,
+      "eval_rewards/reward_fn/std": 0.9424176405995123,
+      "eval_runtime": 632.0076,
+      "eval_samples_per_second": 0.195,
+      "eval_steps_per_second": 0.025,
+      "step": 500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 1497,
+  "num_input_tokens_seen": 1330108,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-500/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ee28670c78b80e386d67006aee3880339b84b20bda4953e363dae2a4315c97a5
+size 6520

tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8fbe698063980a09a4487ec5bbcc545aab380d686f9d918cad649bb49d257f83
+size 17078265

tokenizer_config.json

@@ -0,0 +1,15 @@
+{
+  "add_prefix_space": null,
+  "backend": "tokenizers",
+  "bos_token": "<s>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "</s>",
+  "is_local": true,
+  "legacy": true,
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<pad>",
+  "processor_class": "PixtralProcessor",
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ee28670c78b80e386d67006aee3880339b84b20bda4953e363dae2a4315c97a5
+size 6520