Question Answering on CommonsenseQA (CSQA) (test)
40.85VWRCE
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| CEBackbone=Qwen-0.5B, Plug-in=None, Safety buffer gamma=0.252026.05 | 40.85 | 36.73 | 28.56 | 57.87 | |
| R3FBackbone=Qwen-0.5B, Plug-in=None, Safety buffer gamma=0.252026.05 | 40.84 | 36.95 | 28.67 | 57.53 | |
| SMARTBackbone=Qwen-0.5B, Plug-in=None, Safety buffer gamma=0.252026.05 | 39.22 | 34.8 | 30.22 | 59.33 | |
| CE+augBackbone=Qwen-0.5B, Plug-in=None, Safety buffer gamma=0.252026.05 | 38.32 | 34.02 | 40.33 | 59.27 | |
| SMARTBackbone=Mistral-7B, Plug-in=None, Safety buffer gamma=0.252026.05 | 37.73 | 31.78 | 34.56 | 69.33 | |
| R3FBackbone=Qwen-0.5B, Plug-in=FragileFlow, Safety buffer gamma=0.252026.05 | 32.03 | 28.59 | 28.22 | 57.73 | |
| SMARTBackbone=Qwen-0.5B, Plug-in=FragileFlow, Safety buffer gamma=0.252026.05 | 31.47 | 27.45 | 32.89 | 60.8 | |
| SMARTBackbone=Mistral-7B, Plug-in=FragileFlow, Safety buffer gamma=0.252026.05 | 30.37 | 26.21 | 38.89 | 70.67 | |
| CE+augBackbone=Qwen-0.5B, Plug-in=FragileFlow, Safety buffer gamma=0.252026.05 | 29.3 | 25.79 | 43.22 | 59.73 | |
| CEBackbone=Qwen-1.5B, Plug-in=None, Safety buffer gamma=0.252026.05 | 28.12 | 24.92 | 45.67 | 73.2 | |
| R3FBackbone=Qwen-1.5B, Plug-in=None, Safety buffer gamma=0.252026.05 | 27.82 | 24.71 | 44.67 | 73.53 | |
| SMARTBackbone=Qwen-1.5B, Plug-in=None, Safety buffer gamma=0.252026.05 | 27.57 | 24.25 | 49.78 | 74.8 | |
| CEBackbone=Mistral-7B, Plug-in=None, Safety buffer gamma=0.252026.05 | 26.56 | 22.13 | 51.89 | 76.4 | |
| R3FBackbone=Mistral-7B, Plug-in=None, Safety buffer gamma=0.252026.05 | 26.18 | 21.94 | 51.22 | 76.87 | |
| CE+augBackbone=Mistral-7B, Plug-in=None, Safety buffer gamma=0.252026.05 | 25.9 | 20.81 | 60.78 | 77.33 | |
| CE+augBackbone=Qwen-1.5B, Plug-in=None, Safety buffer gamma=0.252026.05 | 25.2 | 21.84 | 62.22 | 75.4 | |
| R3FBackbone=Qwen-1.5B, Plug-in=FragileFlow, Safety buffer gamma=0.252026.05 | 24.46 | 21.64 | 45.89 | 73.4 | |
| R3FBackbone=Mistral-7B, Plug-in=FragileFlow, Safety buffer gamma=0.252026.05 | 24.04 | 20.02 | 52 | 76.67 | |
| SMARTBackbone=Qwen-1.5B, Plug-in=FragileFlow, Safety buffer gamma=0.252026.05 | 23.41 | 20.26 | 50.67 | 74.87 | |
| CE+augBackbone=Mistral-7B, Plug-in=FragileFlow, Safety buffer gamma=0.252026.05 | 23.34 | 18.57 | 64.33 | 77.67 | |
| CE+augBackbone=Qwen-1.5B, Plug-in=FragileFlow, Safety buffer gamma=0.252026.05 | 21.87 | 18.86 | 65.67 | 75.13 |