Question Answering on GPQA diamond (Pass@1, TCN, WTN)
50Pass@1Qwen3-8B + PRUNETIR
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Qwen3-8B + PRUNETIRModel=Qwen3-8B, Method=PRUNETIR2026.05 | 50 | 2.2 | 5,300 | |
| Qwen3-8BModel=Qwen3-8B2026.05 | 47.1 | 4.3 | 6,600 |
| Method | Links | |||
|---|---|---|---|---|
| Qwen3-8B + PRUNETIRModel=Qwen3-8B, Method=PRUNETIR2026.05 | 50 | 2.2 | 5,300 | |
| Qwen3-8BModel=Qwen3-8B2026.05 | 47.1 | 4.3 | 6,600 |