omrisap
/

Qwen2.5-Math-1.5B-TreeRPO

@@ -1,199 +1,70 @@
 ---
 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
+datasets:
+- AI-MO/NuminaMath-CoT
+base_model:
+- Qwen/Qwen2.5-Math-1.5B
+pipeline_tag: text-generation
+tags:
+- reinforcement-learning
+- grpo
+- hierarchical
+- reasoning
+- math
+- tree-based
+model_name: TreeRPO-Qwen2.5-Math-1.5B
 ---
+# TreeRPO-Qwen2.5-Math-1.5B
+**Short summary:** A 1.5B parameter math reasoning model fine-tuned with *TreeRPO*, a hierarchical extension of GRPO that assigns rewards to “thought” nodes instead of whole sequences—achieving higher GSM8K accuracy with **~10K total** supervised + RL examples and **no reward model**.
+🔎 **Full write-up (method, math, analysis):**
+https://omrisapir.substack.com/publish/post/167273414
 ## Model Details
+- **Base model:** `Qwen/Qwen2.5-Math-1.5B`
+- **Method:** TreeRPO (tree-structured GRPO: depth ≤ 7, branching by entropy + length)
+- **Reward signal:** Deterministic exact-match checker (binary). Interior node reward = average of descendant leaf rewards.
+- **No reference policy / KL:** β = 0 (stability from clipping + relative baseline)
+- **Data efficiency:** 5K SFT CoT examples + 5K RL prompts (vs. multi-million-scale baselines)
+- **Intended domain:** Grade-school & intermediate math word problems (GSM8K style)
+## Intended Use
+Research on hierarchical RL for reasoning; math tutoring prototypes with human oversight; experimentation in deterministic pass/fail domains (e.g., potential extension to code with unit tests).
+**Not intended for:** Open-ended unsafe dialogue, factual QA outside math, high‑stakes decision making.
+## Training Summary
+| Phase | Data | Epochs | Notes |
+|-------|------|--------|-------|
+| SFT | 5K CoT examples (NuminaMath-CoT subset) | 1 | Standard causal LM fine-tune |
+| RL (TreeRPO) | 5K prompts (disjoint) | 1 | Max depth 7; typical branch factor 2 |
+Key hyperparameters: segment length threshold `L_min = 150`, entropy threshold over top‑20 logits `H_th = 1.0`, sampling (temp=0.6, top-p=0.85, top-k=25), PPO/GRPO clip ε=0.2, β=0. Trained on a single 48GB GPU (~18h RL phase).
+## Evaluation (GSM8K Test Set, 1,319 problems)
+| Model | Greedy (%) | Maj@8 (%) | Notes |
+|-------|------------|-----------|-------|
+| Qwen2.5-Math-1.5B-Instruct | 84.8 | 89.5 | Reported settings |
+| **TreeRPO-Qwen2.5-Math-1.5B** | **86.4** | **89.6** | Same decoding (temp 0 / (0.7, top-p 0.8)) |
+- **Greedy** = temperature 0 deterministic decoding.
+- **Maj@8** = 8 sampled completions (temp 0.7, top-p 0.8) majority vote on final boxed answer. Ties / missing boxed answer → incorrect. Single-run numbers (no multi-seed variance).
+## How to Use
+```python
+from transformers import AutoModelForCausalLM, AutoTokenizer
+import torch
+model_name = "your-namespace/TreeRPO-Qwen2.5-Math-1.5B"
+tok = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype=torch.bfloat16, device_map="auto")
+prompt = "Solve step by step: If 3x + 5 = 17, what is x? Put final answer in \\boxed{}."
+inputs = tok(prompt, return_tensors="pt").to(model.device)
+out = model.generate(**inputs, max_new_tokens=256, temperature=0.0)
+print(tok.decode(out[0], skip_special_tokens=True))