Code on MBPP (Average Performance)
67.19Average PerformanceAIPO
Evaluation Results
| Method | Links | |
|---|---|---|
| AIPOStudent LLM=DeepSeek-R1-Distill-Qwen-1.5B, Teacher LLM=Qwen3-30B-A3B-Instruct-2507, Runs (k)=42026.05 | 67.19 | |
| GRPOStudent LLM=DeepSeek-R1-Distill-Qwen-1.5B, Teacher LLM=Qwen3-30B-A3B-Instruct-2507, Runs (k)=42026.05 | 65.05 | |
| OriginalStudent LLM=DeepSeek-R1-Distill-Qwen-1.5B, Teacher LLM=Qwen3-30B-A3B-Instruct-2507, Runs (k)=42026.05 | 60.12 | |
| SFTStudent LLM=DeepSeek-R1-Distill-Qwen-1.5B, Teacher LLM=Qwen3-30B-A3B-Instruct-2507, Runs (k)=42026.05 | 55.51 | |
| DARTBase Model=Qwen3-0.6B2026.05 | 34.91 | |
| STaRBase Model=Qwen3-0.6B2026.05 | 34.6 | |
| GRPOBase Model=Qwen3-0.6B2026.05 | 34.52 | |
| BaseBase Model=Qwen3-0.6B2026.05 | 34.01 | |
| DARTBase Model=Qwen2.5-0.5B-Instruct2026.05 | 34.01 | |
| TESSYBase Model=Qwen3-0.6B2026.05 | 31.11 | |
| Original-SFTBase Model=Qwen3-0.6B2026.05 | 30.52 | |
| GRPOBase Model=Qwen2.5-0.5B-Instruct2026.05 | 29.98 | |
| STaRBase Model=Qwen2.5-0.5B-Instruct2026.05 | 29.88 | |
| BaseBase Model=Qwen2.5-0.5B-Instruct2026.05 | 29.72 | |
| TESSYBase Model=Qwen2.5-0.5B-Instruct2026.05 | 28.88 | |
| Original-SFTBase Model=Qwen2.5-0.5B-Instruct2026.05 | 19.75 |