Science Question Answering on GPQA Diamond (avg@4, pass@4)
68.56Average@4Regular
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| RegularBackbone=Qwen3.5-9B, Response Budget=32K-slot2026.05 | 68.56 | 78.79 | |
| PIPO-SFT + OPDBackbone=Qwen3.5-9B, Response Budget=32K-slot2026.05 | 67.17 | 82.32 | |
| PIPO-SFTBackbone=Qwen3.5-9B, Response Budget=32K-slot2026.05 | 63.76 | 81.31 | |
| Eagle-2Backbone=Qwen3.5-9B, Response Budget=32K-slot2026.05 | 62.63 | 73.23 | |
| RegularBackbone=Qwen3.5-4B, Response Budget=32K-slot2026.05 | 59.72 | 72.73 | |
| PIPO-SFTBackbone=Qwen3.5-4B, Response Budget=32K-slot2026.05 | 59.47 | 79.29 | |
| MTPBackbone=Qwen3.5-9B, Response Budget=32K-slot2026.05 | 55.3 | 71.21 | |
| PIPO-SFT + OPDBackbone=Qwen3.5-4B, Response Budget=32K-slot2026.05 | 54.17 | 72.73 | |
| Eagle-2Backbone=Qwen3.5-4B, Response Budget=32K-slot2026.05 | 53.66 | 66.16 | |
| MTPBackbone=Qwen3.5-4B, Response Budget=32K-slot2026.05 | 52.27 | 69.19 |