TLCR: Token-Level Continuous Reward for Fine-grained Reinforcement Learning from Human Feedback

Eunseop Yoon, Hee Suk Yoon, SooHwan Eom, Gunsoo Han, Daniel Nam, Daejin Jo, Kyoung-Woon On, Mark Hasegawa-Johnson, Sungwoong Kim, Chang Yoo


Abstract
Reinforcement Learning from Human Feedback (RLHF) leverages human preference data to train language models to align more closely with human essence. These human preference data, however, are labeled at the sequence level, creating a mismatch between sequence-level preference labels and tokens, which are autoregressively generated from the language model. Although several recent approaches have tried to provide token-level (i.e., dense) rewards for each individual token, these typically rely on predefined discrete reward values (e.g., positive: +1, negative: -1, neutral: 0), failing to account for varying degrees of preference inherent to each token. To address this limitation, we introduce TLCR (Token-Level Continuous Reward) for RLHF, which incorporates a discriminator trained to distinguish positive and negative tokens, and the confidence of the discriminator is used to assign continuous rewards to each token considering the context. Extensive experiments show that our proposed TLCR leads to consistent performance improvements over previous sequence-level or token-level discrete rewards on open-ended generation benchmarks.
Anthology ID:
2024.findings-acl.889
Volume:
Findings of the Association for Computational Linguistics: ACL 2024
Month:
August
Year:
2024
Address:
Bangkok, Thailand
Editors:
Lun-Wei Ku, Andre Martins, Vivek Srikumar
Venue:
Findings
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
14969–14981
Language:
URL:
https://aclanthology.org/2024.findings-acl.889
DOI:
10.18653/v1/2024.findings-acl.889
Bibkey:
Cite (ACL):
Eunseop Yoon, Hee Suk Yoon, SooHwan Eom, Gunsoo Han, Daniel Nam, Daejin Jo, Kyoung-Woon On, Mark Hasegawa-Johnson, Sungwoong Kim, and Chang Yoo. 2024. TLCR: Token-Level Continuous Reward for Fine-grained Reinforcement Learning from Human Feedback. In Findings of the Association for Computational Linguistics: ACL 2024, pages 14969–14981, Bangkok, Thailand. Association for Computational Linguistics.
Cite (Informal):
TLCR: Token-Level Continuous Reward for Fine-grained Reinforcement Learning from Human Feedback (Yoon et al., Findings 2024)
Copy Citation:
PDF:
https://aclanthology.org/2024.findings-acl.889.pdf