Commonsense Reasoning on CSQA (ACC and Length Metrics)
66.4Accuracy (CSQA)JET
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| JETBackbone=DS-Qwen-7B2025.09 | 66.4 | 531 | |
| DPOBackbone=DS-Qwen-7B2025.09 | 65.9 | 586 | |
| AdaptThinkBackbone=DS-Qwen-7B2025.09 | 65.9 | 536 | |
| Laser-DBackbone=DS-Qwen-7B2025.09 | 65.8 | 600 | |
| DAPOBackbone=DS-Qwen-7B2025.09 | 65.7 | 519 | |
| SFTBackbone=DS-Qwen-7B2025.09 | 65.4 | 593 | |
| LCR1Backbone=DS-Qwen-7B2025.09 | 65 | 404 | |
| Laser-DEBackbone=DS-Qwen-7B2025.09 | 64.8 | 553 | |
| BaseBackbone=DS-Qwen-7B2025.09 | 63.7 | 631 | |
| AdaptThinkBackbone=DS-Qwen-1.5B2025.09 | 48 | 887 | |
| Laser-DEBackbone=DS-Qwen-1.5B2025.09 | 47.8 | 685 | |
| Laser-DBackbone=DS-Qwen-1.5B2025.09 | 47.1 | 685 | |
| SFTBackbone=DS-Qwen-1.5B2025.09 | 47 | 740 | |
| DAPOBackbone=DS-Qwen-1.5B2025.09 | 45.6 | 479 | |
| JETBackbone=DS-Qwen-1.5B2025.09 | 45.6 | 407 | |
| DPOBackbone=DS-Qwen-1.5B2025.09 | 44.6 | 690 | |
| LCR1Backbone=DS-Qwen-1.5B2025.09 | 44.3 | 431 | |
| BaseBackbone=DS-Qwen-1.5B2025.09 | 44.2 | 787 |