Science Question Answering on GPQA (Acceptance length, Delta %)
Evaluation results
| Method | Links | | |
|---|
| TTSTarget Model=Qwen/Qwen3-4B2026.05 | | 5.2 | 47.2 |
|---|
| TTSTarget Model=Qwen/Qwen3-8B2026.05 | | 4.7 | 47.3 |
|---|
| DFlashTarget Model=Qwen/Qwen3-4B2026.05 | | 3.6 | — |
|---|
| TTSTarget Model=Qwen/Qwen3.5-122B2026.05 | | 3.5 | 33.9 |
|---|
| DFlashTarget Model=Qwen/Qwen3-8B2026.05 | | 3.2 | — |
|---|
| DFlashTarget Model=Qwen/Qwen3.5-122B2026.05 | | 2.6 | — |
|---|