作者位次 #2/3ACL 2026Poster· Natural Language Processing
Debiasing Reward Models via Causally Motivated Inference-Time Intervention
Kazutoshi Shinoda · Kosuke Nishida · Kyosuke Nishida
摘要
Reward models (RMs) play a central role in aligning large language models (LLMs) with human preferences. However, RMs are often sensitive to spurious features such as response length. Existing inference-time approaches for mitigating these biases typically focus exclusively on response length, re...