Enregistré dans:
Détails bibliographiques
Auteurs principaux: Ning, Meiling, Zhang, Zhongbao, Ye, Junda, Guo, Jiabao, Guan, Qingyuan
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:https://arxiv.org/abs/2508.18212
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866915619684745216
author Ning, Meiling
Zhang, Zhongbao
Ye, Junda
Guo, Jiabao
Guan, Qingyuan
author_facet Ning, Meiling
Zhang, Zhongbao
Ye, Junda
Guo, Jiabao
Guan, Qingyuan
contents The emergence of LM-based judging reward modeling, represented by generative reward models, has successfully made reinforcement learning from AI feedback (RLAIF) efficient and scalable. To further advance this paradigm, we propose a core insight: this form of reward modeling shares fundamental formal consistency with natural language inference (NLI), a core task in natural language understanding. This reframed perspective points to a key path for building superior reward models: scaling the model's comprehension boundaries. Pursuing this path, exploratory experiments on NLI tasks demonstrate that the slot prediction masked language models (MLMs) incorporating contextual explanations achieve significantly better performance compared to mainstream autoregressive models. Based on this key finding, we propose ESFP-RM, a two-stage LM-based judging reward model that utilizes an explanation based slot framework for prediction to fully leverage the advantages of MLMs. Extensive experiments demonstrate that in both reinforcement learning from human feedback (RLHF) and out-of-distribution (OOD) scenarios, the ESFP-RM framework delivers more stable and generalizable reward signals compared to generative reward models.
format Preprint
id arxiv_https___arxiv_org_abs_2508_18212
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Better Language Model-Based Judging Reward Modeling through Scaling Comprehension Boundaries
Ning, Meiling
Zhang, Zhongbao
Ye, Junda
Guo, Jiabao
Guan, Qingyuan
Computation and Language
The emergence of LM-based judging reward modeling, represented by generative reward models, has successfully made reinforcement learning from AI feedback (RLAIF) efficient and scalable. To further advance this paradigm, we propose a core insight: this form of reward modeling shares fundamental formal consistency with natural language inference (NLI), a core task in natural language understanding. This reframed perspective points to a key path for building superior reward models: scaling the model's comprehension boundaries. Pursuing this path, exploratory experiments on NLI tasks demonstrate that the slot prediction masked language models (MLMs) incorporating contextual explanations achieve significantly better performance compared to mainstream autoregressive models. Based on this key finding, we propose ESFP-RM, a two-stage LM-based judging reward model that utilizes an explanation based slot framework for prediction to fully leverage the advantages of MLMs. Extensive experiments demonstrate that in both reinforcement learning from human feedback (RLHF) and out-of-distribution (OOD) scenarios, the ESFP-RM framework delivers more stable and generalizable reward signals compared to generative reward models.
title Better Language Model-Based Judging Reward Modeling through Scaling Comprehension Boundaries
topic Computation and Language
url https://arxiv.org/abs/2508.18212