Generalization Reasoning on CSQA, GPQA-Diamond, MMLU AVG
60.9AccuracyJET
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| JETBackbone=DS-Qwen-7B2025.09 | 60.9 | 1,013 | |
| Laser-DBackbone=DS-Qwen-7B2025.09 | 59.7 | 1,090 | |
| DAPOBackbone=DS-Qwen-7B2025.09 | 58.5 | 1,034 | |
| DPOBackbone=DS-Qwen-7B2025.09 | 58.3 | 1,083 | |
| Laser-DEBackbone=DS-Qwen-7B2025.09 | 58.2 | 1,023 | |
| LCR1Backbone=DS-Qwen-7B2025.09 | 57.2 | 700 | |
| BaseBackbone=DS-Qwen-7B2025.09 | 57.1 | 1,191 | |
| SFTBackbone=DS-Qwen-7B2025.09 | 57 | 1,136 | |
| AdaptThinkBackbone=DS-Qwen-7B2025.09 | 55.9 | 962 | |
| JETBackbone=DS-Qwen-1.5B2025.09 | 44.5 | 806 | |
| DAPOBackbone=DS-Qwen-1.5B2025.09 | 43.3 | 938 | |
| Laser-DBackbone=DS-Qwen-1.5B2025.09 | 42.9 | 1,223 | |
| Laser-DEBackbone=DS-Qwen-1.5B2025.09 | 42.6 | 1,279 | |
| SFTBackbone=DS-Qwen-1.5B2025.09 | 42.2 | 1,227 | |
| AdaptThinkBackbone=DS-Qwen-1.5B2025.09 | 41.6 | 1,299 | |
| LCR1Backbone=DS-Qwen-1.5B2025.09 | 40.3 | 802 | |
| BaseBackbone=DS-Qwen-1.5B2025.09 | 40.1 | 1,370 | |
| DPOBackbone=DS-Qwen-1.5B2025.09 | 39.8 | 1,221 |