Upload README.md with huggingface_hub

README.md

@@ -1,18 +1,37 @@
 # OPSD Experiment Results
 
-Reproduction of [OPSD (On-Policy Self-Distillation)](https://github.com/siyan-zhao/OPSD) experiments on **Qwen3-4B**.
+Reproduction of [OPSD (On-Policy Self-Distillation)](https://github.com/siyan-zhao/OPSD) on Qwen3-1.7B, 4B, and 8B.
 
-## Methods Compared
-- **Base**: Qwen3-4B without fine-tuning
-- **SFT**: Supervised fine-tuning on OpenThoughts Math 30K
-- **GRPO**: Group Relative Policy Optimization with correctness reward
-- **OPSD**: On-Policy Self-Distillation (proposed method)
+## Results (Avg@12)
 
-## Shared Config
-- LoRA: r=64, alpha=128
-- Learning rate: 5e-6
+### Qwen3-1.7B
+| Method | AIME24 | AIME25 | HMMT25 |
+|--------|:------:|:------:|:------:|
+| Base | 47.2% | 35.3% | 21.9% |
+| OPSD (best) | **49.2%** | **37.5%** | **24.4%** |
+| SFT (best) | 37.5% | 30.8% | 19.2% |
+| GRPO (best) | 47.8% | 35.0% | 22.8% |
+
+### Qwen3-4B
+| Method | AIME24 | AIME25 | HMMT25 |
+|--------|:------:|:------:|:------:|
+| Base | **71.1%** | 60.0% | 38.6% |
+| OPSD (best) | 62.2% | 57.2% | 34.2% |
+| SFT (best) | 62.5% | 58.1% | 33.3% |
+| GRPO (best) | 68.9% | **65.0%** | **41.9%** |
+
+### Qwen3-8B
+| Method | AIME24 | AIME25 | HMMT25 |
+|--------|:------:|:------:|:------:|
+| Base | **72.8%** | 61.7% | 38.6% |
+| OPSD (best) | 69.4% | 63.3% | 38.6% |
+| SFT (best) | 69.2% | 60.3% | 36.1% |
+| GRPO (best) | 72.2% | **65.8%** | **40.8%** |
+
+## Setup
+- All methods: lr=5e-6, BS=32, LoRA r=64 alpha=128, 200 steps
+- Eval: val_n=12, temperature=1.0, thinking mode enabled
 - Data: siyanzhao/Openthoughts_math_30k_opsd
-- Eval: temperature=1.0, thinking mode, val_n=12
 
 ## Reference
-Paper: [Self-Distilled Reasoner](https://arxiv.org/pdf/2601.18734v3)
+[Self-Distilled Reasoner: On-Policy Self-Distillation for LLMs](https://arxiv.org/pdf/2601.18734v3)