Science Question Answering on GPQA (Accuracy and Average Token Length)
51AccuracyDAST
Evaluation results
| Method | Links | ||
|---|---|---|---|
| DASTBase Model=DeepSeek-R1-Distill-Qwen-7B2026.05 | 51 | 6,273 | |
| SLATBase Model=DeepSeek-R1-Distill-Qwen-7B2026.05 | 50.6 | 3,234 | |
| LC-R1Base Model=DeepSeek-R1-Distill-Qwen-7B2026.05 | 49.5 | 4,249 | |
| TLMREBase Model=DeepSeek-R1-Distill-Qwen-7B2026.05 | 47.9 | 5,255 | |
| DeepSeek-R1-Distill-Qwen-7Bstatus=Original base model2026.05 | 47.5 | 6,663 | |
| L1-MaxBase Model=DeepSeek-R1-Distill-Qwen-7B2026.05 | 47 | 2,342 |