Science Reasoning on GPQA (Pass@1 and Pass@16)
52Pass@1Base Model
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Base ModelModel Scale=Qwen3-4B, Thinking Mode=Thinking2026.04 | 52 | 84.4 | |
| MIXED-CUTSModel Scale=Qwen3-4B, Thinking Mode=Non-thinking2026.04 | 50.1 | 84.5 | |
| GRPOModel Scale=Qwen3-4B, Thinking Mode=Non-thinking2026.04 | 48.1 | 84.6 | |
| Base ModelModel Scale=Qwen3-4B, Thinking Mode=Non-thinking2026.04 | 45.3 | 83.9 | |
| MIXED-CUTSModel Scale=Qwen3-1.7B, Thinking Mode=Non-thinking2026.04 | 36 | 79.4 | |
| Base ModelModel Scale=Qwen3-1.7B, Thinking Mode=Thinking2026.04 | 34.9 | 75.3 | |
| GRPOModel Scale=Qwen3-1.7B, Thinking Mode=Non-thinking2026.04 | 34.2 | 80.1 | |
| Base ModelModel Scale=Qwen3-1.7B, Thinking Mode=Non-thinking2026.04 | 32.1 | 80.3 |