Program Induction on Custom Suite Shift-P OOD
100AccuracyNLI Gradient Search
Evaluation results
| Method | Links | |
|---|---|---|
| NLI Gradient SearchInference Strategy=Gradient Search, Decoder Layers=22026.04 | 100 | |
| In-ContextInference Strategy=In-Context Learning, Decoder Layers=22026.04 | 0 | |
| TTTInference Strategy=Test-Time Training, Decoder Layers=22026.04 | 0 | |
| LPNInference Strategy=Base Inference, Decoder Layers=22026.04 | 0 | |
| LPN Gradient SearchInference Strategy=Gradient Search, Decoder Layers=22026.04 | 0 | |
| D-LPNInference Strategy=Base Inference, Decoder Layers=22026.04 | 0 | |
| D-LPN Gradient SearchInference Strategy=Gradient Search, Decoder Layers=22026.04 | 0 | |
| NLIInference Strategy=Base Inference, Decoder Layers=22026.04 | 0 | |
| NLI Prior SearchInference Strategy=Prior Search, Decoder Layers=22026.04 | 0 |