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

Home/Benchmarks

Multi-objective RLHF alignment on safeRLHF (test)

52Win Rate

MAVIS

Updated 1mo ago

Evaluation Results

Method	Links
MAVIS 2025.08		52

SOTA Paper

MAVIS

MAVIS: Multi-Objective Alignment via Inference-Time Value-Guided Selection

Dataset

SafeRLHF

Follow for update

@wizwand_team Discord

Related Benchmarks

Helpfulness on HH-RLHF (test)Harmlessness on HH-RLHF (test)Humor on HH-RLHF (test)Faithfulness on Summarization (test)

© 2026 wizwand

Blog Contact Changelog Swarm

Privacy Policy Terms of Service FAQs Swarm Docs