Skywork-Reward-V2 models trained on 26 million human-AI curated preference pairs set new state-of-the-art results on seven major reward model benchmarks.
Our results suggest that roughly O(105) carefully-selected gold pairs suffice to train strong gold and Stage-1 RMs
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Skywork-Reward-V2: Scaling Preference Data Curation via Human-AI Synergy
Skywork-Reward-V2 models trained on 26 million human-AI curated preference pairs set new state-of-the-art results on seven major reward model benchmarks.