sail
/

Llama-3.2-3B-Oat-Zero

@@ -1,199 +1,129 @@
 ---
 library_name: transformers
-tags: []
 ---
-# 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. -->
-This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
-- **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]

 ---
 library_name: transformers
+license: apache-2.0
 ---
+# Llama-3.2-3B-Oat-Zero
+## Links
+- 📜 [Paper](https://github.com/sail-sg/understand-r1-zero/blob/main/understand-r1-zero.pdf)
+- 💻 [GitHub](https://github.com/sail-sg/understand-r1-zero)
+- 🤗 [Oat-Zero Collection](https://huggingface.co/collections/sail/oat-zero-understanding-r1-zero-like-training-67dcdb07b9f3eb05f1501c4a)
+## Introduction
+To understand R1-Zero-like training, we critically examine two core components: **base models**
+and **reinforcement learning**. We highlight our findings below.
+### On base models:
+1. **DeepSeek-V3-Base already exhibit "Aha moment"**.
+<p align="center">
+  <img src="https://raw.githubusercontent.com/sail-sg/understand-r1-zero/refs/heads/main/assets/deepseek-base-aha.png" width=65%/>
+</p>
+2. As the popular choice for R1-Zero-like training, Qwen2.5 base models demonstrate strong reasoning capabilities
+even **without** prompt templates: the average benchmark scores improve by **~60%** (compared to the traditional 4-shot prompting)!
+<p align="center">
+  <img src="https://raw.githubusercontent.com/sail-sg/understand-r1-zero/refs/heads/main/assets/qwen-math-base-scores.png" width=65%/>
+</p>
+### On reinforcement learning:
+3. GRPO leads to **biased** optimization! We propose a simple fix that improves token efficiency
+while maintaining reasoning performance, termed as Dr. GRPO (GRPO **D**one **R**ight).
+<p align="center">
+  <img src="https://raw.githubusercontent.com/sail-sg/understand-r1-zero/refs/heads/main/assets/drgrpo.png" width=65%/>
+</p>
+4. In R1-Zero-like training, the template and the question set perform a duet to affect the RL dynamics
+   * (Left Plot) For Qwen2.5-Math-1.5B, a mismatched template (e.g., R1 template) in fact **destructs the reasoning capabilities before RL reconstructing it**. This makes the improvement impressive on the surface.
+   * (Middle Plot) However, if a template does not deviate from the pretraining distribution too far, even a small and completely o.o.d. question set (e.g., GSM8K) could induce the reasoning ability equally well, by reinforcing correct reasoning behaviors instead of infusing new knowledge.
+<p align="center">
+  <img src="https://raw.githubusercontent.com/sail-sg/understand-r1-zero/refs/heads/main/assets/template-data-duet.png" width=65%/>
+</p>
+5. Beyond Qwen, Llama can also be RL-tuned from base models. In this case, domain-specific pretraining will improves RL ceiling.
+   * (Right Plot) GRPO can even make Llama with math knowledge "Aha" by increasing the output length; however, it is likely due to its length bias, which can be removed by Dr. GRPO.
+ <p align="center">
+  <img src="https://raw.githubusercontent.com/sail-sg/understand-r1-zero/refs/heads/main/assets/llama-r1-zero.png" width=65%/>
+</p>
+### Our minimalist R1-Zero recipe:
+Our analysis suggests a minimalist recipe for R1-Zero-like training:
+We RL-tune Qwen2.5-
+Math-7B using the (unbiased) Dr. GRPO algorithm on MATH level 3-5 questions with the Qwen-Math template, and achieve state-of-the-art performance with only 27 hours compute on 8× A100 GPUs
+ <p align="center">
+  <img src="https://raw.githubusercontent.com/sail-sg/understand-r1-zero/refs/heads/main/assets/benchmark.png" width=90%/>
+</p>
+If you are interested in more details, please check out our [paper](https://github.com/sail-sg/understand-r1-zero/blob/main/understand-r1-zero.pdf)!
+## Usage
+```python
+import time
+import vllm
+def apply_qwen_math_template(question: str):
+    return (
+        "<|im_start|>system\nPlease reason step by step, and put your final answer within \\boxed{}.<|im_end|>\n<|im_start|>user\n"
+        + question
+        + "<|im_end|>\n<|im_start|>assistant\n"
+    )
+def apply_r1_template(question: str):
+    return (
+        "A conversation between User and Assistant. The User asks a question, and the Assistant solves it. The Assistant first thinks about the reasoning process in the mind and then provides the User with the answer. "
+        "The reasoning process is enclosed within <think> </think> and answer is enclosed within <answer> </answer> tags, respectively, i.e., <think> reasoning process here </think> <answer> answer here </answer>.\nUser: "
+        + question
+        + "\nAssistant: <think>"
+    )
+model_name = "sail/Llama-3.2-3B-Oat-Zero"
+sampling_params = vllm.SamplingParams(
+    n=1,
+    temperature=0,
+    top_p=1,
+    max_tokens=3000,
+    logprobs=2,
+    seed=int(time.time_ns()),
+)
+model = vllm.LLM(
+    model_name,
+    swap_space=32,
+    max_model_len=4096,
+    dtype="bfloat16",
+    enable_prefix_caching=True,
+)
+if "Llama-3.2-3B-Oat-Zero" in model_name:
+    apply_template = apply_r1_template
+else:
+    apply_template = apply_qwen_math_template
+prompts = [
+    "How many positive whole-number divisors does 196 have?"
+]
+prompts = list(map(apply_template, prompts))
+outputs = model.generate(prompts, sampling_params)
+print(outputs)
+```
+## Citation
+```latex
+@misc{liu2025understanding,
+  title={Understanding R1-Zero-Like Training: A Critical Perspective},
+  author={Zichen Liu and Changyu Chen and Wenjun Li and Penghui Qi and Tianyu Pang and Chao Du and Wee Sun Lee and Min Lin},
+  year={2025},
+  howpublished={\url{https://github.com/sail-sg/understand-r1-zero}},
+}
+```