יום שני, 16 בנובמבר 2026 LIVE
AI־INFO

כתבה arXiv cs.AI ·

EnGRICH: Enhancing Generative Reward Modeling with Critiques from Humans

תקציר מקורי באנגליתarXiv:2610.05370v2 Announce Type: replace Abstract: Generative reward models (GRMs) are important for LLM optimization. Unlike scalar reward models, GRMs generate natural-language critiques alongside preference judgments, providing finer-grained evaluation signals. Their effectiveness depends heavily on critique reliability. However, existing GRM training typically uses final preference correctness as outcome supervision. Because the preference outcome space is highly constrained, unreliable critiques can still yield correct outcomes and thus be reinforced. Recent work leverages human critiques for process supervision, but such critiques are scarce and are often reduced to scalar rewards, leaving their fine-grained evaluative information underutilized. We argue that evaluative criteria lea
קרא במקור המקורי