# Circuit Recovery on MCQA Source: [Circuit Recovery on MCQA]() Collection index: [llms.txt]() Data updated: 2026-04-28T10:11:43.480Z Evaluation of model performance on the Circuit Recovery task using the MCQA dataset, measured by CPR (faithfulness score). A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Circuit Recovery | | Dataset (including any stated split) | MCQA | | Metric | CPR (Faithfulness Score) | | Unit | SCALAR | | Direction | Higher is better | | Results loaded | 14 of 14 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [CPR (Faithfulness Score)](): Higher is better; unit: SCALAR. ## Recorded results | Method | CPR (Faithfulness Score) | Result date | Report paper | | --- | --- | --- | --- | | NAP-IG-Inputs | 1.13 | Not recorded | 69f03f7555a3c32af89fa4df | | NAP-IG-Activations | 0.9 | Not recorded | 69f03f7555a3c32af89fa4df | | NAP | 0.82 | Not recorded | 69f03f7555a3c32af89fa4df | | NAP-IG-Inputs | 0.48 | Not recorded | 69f03f7555a3c32af89fa4df | | NAP-IG-Inputs | 0.47 | Not recorded | 69f03f7555a3c32af89fa4df | | NAP-IG-Inputs | 0.41 | Not recorded | 69f03f7555a3c32af89fa4df | | NAP-IG-Inputs | 0.33 | Not recorded | 69f03f7555a3c32af89fa4df | | NAP-IG-Inputs | 0.32 | Not recorded | 69f03f7555a3c32af89fa4df | | NAP-IG-Activations | 0.3 | Not recorded | 69f03f7555a3c32af89fa4df | | Baseline | 0.28 | Not recorded | 69f03f7555a3c32af89fa4df | | NAP | 0.28 | Not recorded | 69f03f7555a3c32af89fa4df | | NAP | 0.26 | Not recorded | 69f03f7555a3c32af89fa4df | | NAP-IG-Activations | 0.25 | Not recorded | 69f03f7555a3c32af89fa4df | | Baseline | 0.25 | Not recorded | 69f03f7555a3c32af89fa4df | ### NAP-IG-Inputs CPR (Faithfulness Score): 1.13. Result ID: 69f087cc47f34ce924d70ea6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Method Category":"NAP-IG-Inputs","Transfer Setting":"Gold (Upper Bound)"}. This record does not assert independent reproduction. ### NAP-IG-Activations CPR (Faithfulness Score): 0.9. Result ID: 69f087cc47f34ce924d70eac. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Method Category":"NAP-IG-Activations","Transfer Setting":"Gold (Upper Bound)"}. This record does not assert independent reproduction. ### NAP CPR (Faithfulness Score): 0.82. Result ID: 69f087cc47f34ce924d70ea9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Method Category":"NAP","Transfer Setting":"Gold (Upper Bound)"}. This record does not assert independent reproduction. ### NAP-IG-Inputs CPR (Faithfulness Score): 0.48. Result ID: 69f087cc47f34ce924d70f11. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Transfer Setting":"DFA (Far/Zero-Shot)","Model Pair":"Llama-3 (1B -\> 3B)"}. This record does not assert independent reproduction. ### NAP-IG-Inputs CPR (Faithfulness Score): 0.47. Result ID: 69f087cc47f34ce924d70f0f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Transfer Setting":"Gold (Upper Bound)","Model Pair":"Llama-3 (1B -\> 3B)"}. This record does not assert independent reproduction. ### NAP-IG-Inputs CPR (Faithfulness Score): 0.41. Result ID: 69f087cc47f34ce924d70f10. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Transfer Setting":"DFA (In-Distribution)","Model Pair":"Llama-3 (1B -\> 3B)"}. This record does not assert independent reproduction. ### NAP-IG-Inputs CPR (Faithfulness Score): 0.33. Result ID: 69f087cc47f34ce924d70ea8. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Method Category":"NAP-IG-Inputs","Transfer Setting":"DFA (Far/Zero-Shot)"}. This record does not assert independent reproduction. ### NAP-IG-Inputs CPR (Faithfulness Score): 0.32. Result ID: 69f087cc47f34ce924d70ea7. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Method Category":"NAP-IG-Inputs","Transfer Setting":"DFA (In-Distribution)"}. This record does not assert independent reproduction. ### NAP-IG-Activations CPR (Faithfulness Score): 0.3. Result ID: 69f087cc47f34ce924d70eae. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Method Category":"NAP-IG-Activations","Transfer Setting":"DFA (Far/Zero-Shot)"}. This record does not assert independent reproduction. ### Baseline CPR (Faithfulness Score): 0.28. Result ID: 69f087cc47f34ce924d70ea5. [Reporting paper]() Recorded evaluation configuration: {"Method Category":"Baseline","Transfer Setting":"Random Alignment"}. This record does not assert independent reproduction. ### NAP CPR (Faithfulness Score): 0.28. Result ID: 69f087cc47f34ce924d70eab. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Method Category":"NAP","Transfer Setting":"DFA (Far/Zero-Shot)"}. This record does not assert independent reproduction. ### NAP CPR (Faithfulness Score): 0.26. Result ID: 69f087cc47f34ce924d70eaa. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Method Category":"NAP","Transfer Setting":"DFA (In-Distribution)"}. This record does not assert independent reproduction. ### NAP-IG-Activations CPR (Faithfulness Score): 0.25. Result ID: 69f087cc47f34ce924d70ead. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Method Category":"NAP-IG-Activations","Transfer Setting":"DFA (In-Distribution)"}. This record does not assert independent reproduction. ### Baseline CPR (Faithfulness Score): 0.25. Result ID: 69f087cc47f34ce924d70f0e. [Reporting paper]() Recorded evaluation configuration: {"Transfer Setting":"Random Alignment","Model Pair":"Llama-3 (1B -\> 3B)"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.