List your company

Sparse Probing on SAEBench

81.9Average F1 Scorefmxcoder (p = 0)
45.60455.02764.4573.873May 10, 2026

Evaluation results

MethodLinks
fmxcoder (p = 0)LLM=Gemma2-2b2026.0581.9198.7
fmxcoder (p = 0)LLM=Pythia-1.4b2026.0578.436.1
fmxcoder (p = 0)LLM=GPT2-Small2026.0578.3120.2
fmxcoder (p = 0)LLM=Pythia-410m2026.0574.310.2
fmxcoder (p = 0.05)LLM=GPT2-Small2026.0573.568.4
fmxcoder (p = 0.05)LLM=Gemma2-2b2026.057339.1
fmxcoder (p = 0.05)LLM=Pythia-1.4b2026.0571.213.1
fmxcoder (p = 0.05)LLM=Pythia-410m2026.0568.14.5
CrosscoderLLM=GPT2-Small2026.0567.611.6
SAELLM=GPT2-Small2026.0565.78.4
SAELLM=Pythia-410m2026.05650.5
SAELLM=Gemma2-2b2026.0564.82.7
SAELLM=Pythia-1.4b2026.0564.60.8
CrosscoderLLM=Gemma2-2b2026.0551.20.2
CrosscoderLLM=Pythia-410m2026.0547.40.1
CrosscoderLLM=Pythia-1.4b2026.05470.2