Logical Reasoning on KandLogic 7obj
92.21AccuracyVLC
Evaluation results
| Method | Links | |
|---|---|---|
| VLCBackbone Scale=7B, Random Seeds=52026.03 | 92.21 | |
| ViperGPTBackbone Scale=7B, Random Seeds=52026.03 | 84.7 | |
| End2end RSBackbone Scale=7B, Random Seeds=52026.03 | 66.6 | |
| End2end FTBackbone Scale=7B, Random Seeds=52026.03 | 57.76 | |
| PrismBackbone Scale=7B, Random Seeds=52026.03 | 56.35 |