Knowledge-intensive Question Answering on TriviaQA
71.7Pass@1 AccuracyBaseline
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Baselinebackbone=QwQ-32B2026.06 | 71.7 | 630 | |
| DyConbackbone=QwQ-32B2026.06 | 71.7 | 592 | |
| DyConbackbone=Qwen3-14B2026.06 | 66.8 | 440 | |
| Baselinebackbone=Qwen3-14B2026.06 | 66.4 | 508 | |
| DyConbackbone=Qwen3-4B-Thinking-25072026.06 | 34.1 | 729 | |
| Baselinebackbone=Qwen3-4B-Thinking-25072026.06 | 33.5 | 1,041 | |
| DyConbackbone=DeepSeek-R1-Distill-Qwen-7B2026.06 | 19.6 | 619 | |
| Baselinebackbone=DeepSeek-R1-Distill-Qwen-7B2026.06 | 19.2 | 1,295 |