List your company

Circuit Discovery on Induction

99ROC-AUCCo-activation
23.0842.7962.582.21Jul 2, 2026

Evaluation results

MethodLinks
Co-activationModel=GPT-2-lg, Signal=co-activation, Calibration=repeated-token2026.0799
Co-activationModel=GPT-2-md, Signal=co-activation, Calibration=repeated-token2026.0798
Co-activationModel=GPT-Neo-1.3B, Signal=co-activation, Calibration=repeated-token2026.0797
Co-activationModel=Pythia-410M, Signal=co-activation, Calibration=repeated-token2026.0795
Co-activationModel=Pythia-160M, Signal=co-activation, Calibration=repeated-token2026.0794
Co-activationModel=GPT-2-sm, Signal=co-activation, Calibration=repeated-token2026.0794
Co-activationModel=Qwen2.5-7B, Signal=co-activation, Calibration=repeated-token2026.0793
Co-activationModel=Llama-3.1-8B, Signal=co-activation, Calibration=repeated-token2026.0792
Co-activationModel=Gemma-2-9B, Signal=co-activation, Calibration=repeated-token2026.0789
Co-activationModel=Gemma-2-2B, Signal=co-activation, Calibration=repeated-token2026.0787
Co-activationModel=OLMo-2-7B, Signal=co-activation, Calibration=repeated-token2026.0786
Co-activationModel=Pythia-1.4B, Signal=co-activation, Calibration=repeated-token2026.0783
Conditional Co-ablationModel=GPT-2-lg, Signal=co-ablation, Calibration=repeated-token2026.0779
Conditional Co-ablationModel=Qwen2.5-7B, Signal=co-ablation, Calibration=repeated-token2026.0771
Conditional Co-ablationModel=Llama-3.1-8B, Signal=co-ablation, Calibration=repeated-token2026.0758
Conditional Co-ablationModel=Pythia-410M, Signal=co-ablation, Calibration=repeated-token2026.0753
Conditional Co-ablationModel=GPT-2-md, Signal=co-ablation, Calibration=repeated-token2026.0752
Conditional Co-ablationModel=GPT-2-sm, Signal=co-ablation, Calibration=repeated-token2026.0748
Conditional Co-ablationModel=Pythia-160M, Signal=co-ablation, Calibration=repeated-token2026.0743
Conditional Co-ablationModel=GPT-Neo-1.3B, Signal=co-ablation, Calibration=repeated-token2026.0740
Conditional Co-ablationModel=OLMo-2-7B, Signal=co-ablation, Calibration=repeated-token2026.0740
Conditional Co-ablationModel=Pythia-1.4B, Signal=co-ablation, Calibration=repeated-token2026.0739
Conditional Co-ablationModel=Gemma-2-2B, Signal=co-ablation, Calibration=repeated-token2026.0727
Conditional Co-ablationModel=Gemma-2-9B, Signal=co-ablation, Calibration=repeated-token2026.0726