List your company

Safety Evaluation on Opposite-day

4.5ASRClaude 3.5 Sonnet
1.5421.5241.561.48Feb 28, 2025

Evaluation results

MethodLinks
Claude 3.5 SonnetNBF steering=true2025.024.5
Claude 3.5 SonnetNBF steering=false2025.029.5
o1NBF steering=true2025.0221
GPT-4oNBF steering=true2025.0232.5
GPT-3.5-turboNBF steering=true2025.0237.5
o1NBF steering=false2025.0253
GPT-4oNBF steering=false2025.0272.5
GPT-3.5-turboNBF steering=false2025.0278.5