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

Anthropic-SafeRLHF

Benchmarks

Task NameDataset NameSOTA ResultTrend
Preference evaluationAnthropic-SafeRLHF (target)
Win Rate41.7
2
Preference evaluationAnthropic-SafeRLHF benchmark
Win Rate33.7
2
Showing 2 of 2 rows