List your company

SAE Activation on Neuronpedia SAE features entropy 3-9

81Win PercentageGCG
15.4832.4949.566.51Jun 15, 2025

Evaluation results

MethodLinks
GCGOptimisation target=mean activation, Comparison baseline=GPT-4o2025.0681
EPOOptimisation target=mean activation, Comparison baseline=GPT-4o2025.0675.5
GCGOptimisation target=mean activation, Comparison baseline=Max Act Examples2025.0675.2
Max Act ExamplesOptimisation target=mean activation, Comparison baseline=GPT-4o2025.0673.7
GCGOptimisation target=mean activation, Comparison baseline=EPO2025.0668.4
EPO-AssistOptimisation target=mean activation, Comparison baseline=GPT-4o2025.0668.3
EPOOptimisation target=mean activation, Comparison baseline=Max Act Examples2025.0667.3
EPO-AssistOptimisation target=mean activation, Comparison baseline=EPO-Inpaint2025.0664.4
GCGOptimisation target=mean activation, Comparison baseline=EPO-Assist2025.0662.7
EPO-InpaintOptimisation target=mean activation, Comparison baseline=GPT-4o2025.0661.8
GCGOptimisation target=mean activation, Comparison baseline=EPO-Inpaint2025.0659.8
EPO-AssistOptimisation target=mean activation, Comparison baseline=Max Act Examples2025.0657.8
EPO-InpaintOptimisation target=mean activation, Comparison baseline=EPO2025.0653.5
EPO-InpaintOptimisation target=mean activation, Comparison baseline=Max Act Examples2025.0650
Max Act ExamplesOptimisation target=mean activation, Comparison baseline=EPO-Inpaint2025.0650
EPO-AssistOptimisation target=mean activation, Comparison baseline=EPO2025.0648.5
EPOOptimisation target=mean activation, Comparison baseline=EPO-Assist2025.0647.5
EPOOptimisation target=mean activation, Comparison baseline=EPO-Inpaint2025.0646.5
Max Act ExamplesOptimisation target=mean activation, Comparison baseline=EPO-Assist2025.0641.2
EPO-InpaintOptimisation target=mean activation, Comparison baseline=GCG2025.0639.2
EPO-AssistOptimisation target=mean activation, Comparison baseline=GCG2025.0637.3
GPT-4oOptimisation target=mean activation, Comparison baseline=EPO-Inpaint2025.0637.3
EPO-InpaintOptimisation target=mean activation, Comparison baseline=EPO-Assist2025.0634.7
Max Act ExamplesOptimisation target=mean activation, Comparison baseline=EPO2025.0632.7
GPT-4oOptimisation target=mean activation, Comparison baseline=EPO-Assist2025.0631.7
EPOOptimisation target=mean activation, Comparison baseline=GCG2025.0629.6
GPT-4oOptimisation target=mean activation, Comparison baseline=Max Act Examples2025.0626.3
Max Act ExamplesOptimisation target=mean activation, Comparison baseline=GCG2025.0624.8
GPT-4oOptimisation target=mean activation, Comparison baseline=EPO2025.0624.5
GPT-4oOptimisation target=mean activation, Comparison baseline=GCG2025.0618