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