Circuit Recovery on ARC-C
1.11CPR Faithfulness ScoreNAP-IG-Activations
Evaluation Results
| Method | Links | |
|---|---|---|
| NAP-IG-ActivationsMethod Category=NAP-IG-Activations, Transfer Setting=Gold (Upper Bound)2026.04 | 1.11 | |
| NAPMethod Category=NAP, Transfer Setting=Gold (Upper Bound)2026.04 | 1.1 | |
| NAP-IG-InputsMethod Category=NAP-IG-Inputs, Transfer Setting=Gold (Upper Bound)2026.04 | 0.94 | |
| NAP-IG-InputsTransfer Setting=Gold (Upper Bound), Model Pair=Llama-3 (1B -> 3B)2026.04 | 0.48 | |
| NAP-IG-InputsTransfer Setting=DFA (In-Distribution), Model Pair=Llama-3 (1B -> 3B)2026.04 | 0.39 | |
| NAPMethod Category=NAP, Transfer Setting=DFA (Far/Zero-Shot)2026.04 | 0.38 | |
| NAP-IG-InputsTransfer Setting=DFA (Far/Zero-Shot), Model Pair=Llama-3 (1B -> 3B)2026.04 | 0.38 | |
| NAP-IG-InputsTransfer Setting=DFA (Near-Distribution), Model Pair=Llama-3 (1B -> 3B)2026.04 | 0.37 | |
| NAP-IG-ActivationsMethod Category=NAP-IG-Activations, Transfer Setting=DFA (Far/Zero-Shot)2026.04 | 0.34 | |
| NAP-IG-InputsMethod Category=NAP-IG-Inputs, Transfer Setting=DFA (In-Distribution)2026.04 | 0.31 | |
| NAP-IG-InputsMethod Category=NAP-IG-Inputs, Transfer Setting=DFA (Near-Distribution)2026.04 | 0.29 | |
| NAP-IG-InputsMethod Category=NAP-IG-Inputs, Transfer Setting=DFA (Far/Zero-Shot)2026.04 | 0.29 | |
| NAPMethod Category=NAP, Transfer Setting=DFA (In-Distribution)2026.04 | 0.29 | |
| NAP-IG-ActivationsMethod Category=NAP-IG-Activations, Transfer Setting=DFA (In-Distribution)2026.04 | 0.29 | |
| BaselineMethod Category=Baseline, Transfer Setting=Random Alignment2026.04 | 0.27 | |
| NAPMethod Category=NAP, Transfer Setting=DFA (Near-Distribution)2026.04 | 0.27 | |
| NAP-IG-ActivationsMethod Category=NAP-IG-Activations, Transfer Setting=DFA (Near-Distribution)2026.04 | 0.27 | |
| BaselineTransfer Setting=Random Alignment, Model Pair=Llama-3 (1B -> 3B)2026.04 | 0.25 |