Fine-Grained Human Feedback Gives Better Rewards for Language Model
  Training

Zeqiu Wu; Yushi Hu; Weijia Shi; Nouha Dziri; Alane Suhr; Prithviraj Ammanabrolu; Noah A. Smith; Mari Ostendorf; Hannaneh Hajishirzi

Paper · arXiv 2306.01693

Fine-Grained Human Feedback Gives Better Rewards for Language Model Training

Zeqiu Wu, Yushi Hu, Weijia Shi, Nouha Dziri, Alane Suhr, Prithviraj Ammanabrolu, Noah A. Smith, Mari Ostendorf, Hannaneh Hajishirzi

3 upvotesJune 2, 2023arXiv 预印本

AI 摘要

Fine-Grained Reinforcement Learning from Human Feedback improves language model performance by using detailed human feedback to provide segment-specific rewards for diverse feedback types.

Reinforcement learning from human feedback (RLHF)fine-grained human feedbackFine-Grained RLHFdensity rewardsmultiple reward modelsfactual incorrectnessirrelevanceinformation incompleteness

Abstract

Language models (LMs) often exhibit undesirable text generation behaviors, including generating false, toxic, or irrelevant outputs. Reinforcement learning from human feedback (RLHF) - where human preference judgments on LM outputs are transformed into a learning signal - has recently shown promise in addressing these issues. However, such holistic feedback conveys limited information on long text outputs; it does not indicate which aspects of the outputs influenced user preference; e.g., which parts contain what type(s) of errors. In this paper, we use fine-grained human feedback (e.g., which sentence is false, which sub-sentence is irrelevant) as an explicit training signal. We introduce Fine-Grained RLHF, a framework that enables training and learning from reward functions that are fine-grained in two respects: (1) density, providing a reward after every segment (e.g., a sentence) is generated; and (2) incorporating multiple reward models associated with different feedback types (e.g., factual incorrectness, irrelevance, and information incompleteness). We conduct experiments on detoxification and long-form question answering to illustrate how learning with such reward functions leads to improved performance, supported by both automatic and human evaluation. Additionally, we show that LM behaviors can be customized using different combinations of fine-grained reward models. We release all data, collected human feedback, and codes at https://FineGrainedRLHF.github.io.