Logical Reasoning on KandLogic 3obj
99.97AccuracyEnd2end FT
Evaluation Results
| Method | Links | |
|---|---|---|
| End2end FTBackbone Scale=7B, Random Seeds=52026.03 | 99.97 | |
| ViperGPTBackbone Scale=7B, Random Seeds=52026.03 | 96.56 | |
| VLCBackbone Scale=7B, Random Seeds=52026.03 | 95.97 | |
| End2end RSBackbone Scale=7B, Random Seeds=52026.03 | 65.05 | |
| PrismBackbone Scale=7B, Random Seeds=52026.03 | 58.3 |