Multitask Language Understanding on MMLU (Accuracy and Token Usage)
65.1AccuracyLaser
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| LaserBackbone=DeepSeek-R1-Distill-Qwen-7B2026.05 | 65.1 | 781.1 | |
| ExpThinkBackbone=DeepSeek-R1-Distill-Qwen-7B2026.05 | 65.1 | 448.5 | |
| JETBackbone=DeepSeek-R1-Distill-Qwen-7B2026.05 | 64.6 | 1,036 | |
| VanillaBackbone=DeepSeek-R1-Distill-Qwen-7B2026.05 | 63.9 | 1,170.8 | |
| LC-R1Backbone=DeepSeek-R1-Distill-Qwen-7B2026.05 | 63.4 | 649.6 | |
| AdaptThinkBackbone=DeepSeek-R1-Distill-Qwen-7B2026.05 | 63.2 | 902.6 | |
| AutoThinkBackbone=DeepSeek-R1-Distill-Qwen-7B2026.05 | 57.4 | 861 | |
| JETBackbone=DeepSeek-R1-Distill-Qwen-1.5B2026.05 | 46.7 | 1,027.1 | |
| AdaptThinkBackbone=DeepSeek-R1-Distill-Qwen-1.5B2026.05 | 44.7 | 995.2 | |
| ExpThinkBackbone=DeepSeek-R1-Distill-Qwen-1.5B2026.05 | 44.2 | 622.1 | |
| LaserBackbone=DeepSeek-R1-Distill-Qwen-1.5B2026.05 | 42.5 | 943.5 | |
| LC-R1Backbone=DeepSeek-R1-Distill-Qwen-1.5B2026.05 | 41 | 927 | |
| AutoThinkBackbone=DeepSeek-R1-Distill-Qwen-1.5B2026.05 | 40.5 | 1,068.1 | |
| VanillaBackbone=DeepSeek-R1-Distill-Qwen-1.5B2026.05 | 38.8 | 1,680.3 |