Puzzle on Reasoning Gym
43.98Avg@4AIPO
Evaluation Results
| Method | Links | |
|---|---|---|
| AIPOModel=Qwen2.5-32B-Instruct2026.05 | 43.98 | |
| GRPOModel=Qwen2.5-32B-Instruct2026.05 | 34.24 | |
| SFTModel=Qwen2.5-32B-Instruct2026.05 | 31.13 | |
| OriginalModel=Qwen2.5-32B-Instruct2026.05 | 28.56 | |
| AIPOStudent LLM=DeepSeek-R1-Distill-Qwen-1.5B, Teacher LLM=Qwen3-30B-A3B-Instruct-2507, Runs (k)=42026.05 | 16.44 | |
| GRPOStudent LLM=DeepSeek-R1-Distill-Qwen-1.5B, Teacher LLM=Qwen3-30B-A3B-Instruct-2507, Runs (k)=42026.05 | 13.53 | |
| SFTStudent LLM=DeepSeek-R1-Distill-Qwen-1.5B, Teacher LLM=Qwen3-30B-A3B-Instruct-2507, Runs (k)=42026.05 | 10.98 | |
| OriginalStudent LLM=DeepSeek-R1-Distill-Qwen-1.5B, Teacher LLM=Qwen3-30B-A3B-Instruct-2507, Runs (k)=42026.05 | 3.33 |