Share your thoughts, 1 month free Claude Pro on usSee more

Home/Benchmarks

Safety Alignment on AdvBench (HarmRate)

0Harm Rate

BoN64

Updated 1mo ago

Evaluation Results

Method	Links
BoN64 2025.05		0
SEA 2025.05		0
SFT 2025.05		38

SOTA Paper

BoN64

WebGPT: Browser-assisted question-answering with human feedback

Dataset

AdvBench

Follow for update

@wizwand_team Discord

Related Benchmarks

Truthfulness on TruthfulQA (Reward, TR, IR, Div)Mathematical Reasoning on GSM8K (test) (Accuracy, Reward)Mathematical Reasoning on MATH (test) (Reward, Acc (%))Truthful Question Answering on TruthfulQA (TruthRate and InfoRate)Mathematical Reasoning on GSM8K (Acc)

© 2026 wizwand

Blog Contact Changelog Swarm

Privacy Policy Terms of Service FAQs Swarm Docs