Saved in:
Bibliographic Details
Main Authors: Wang, Hanyin, Gao, Chufan, Xu, Qiping, Liu, Bolun, Hussein, Guleid, Korsapati, Hariprasad, Labban, Mohamad El, Iheasirim, Kingsley, Hassan, Mohamed, Anil, Gokhan, Bartlett, Brian, Sun, Jimeng
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2412.12583
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911141898223616
author Wang, Hanyin
Gao, Chufan
Xu, Qiping
Liu, Bolun
Hussein, Guleid
Korsapati, Hariprasad
Labban, Mohamad El
Iheasirim, Kingsley
Hassan, Mohamed
Anil, Gokhan
Bartlett, Brian
Sun, Jimeng
author_facet Wang, Hanyin
Gao, Chufan
Xu, Qiping
Liu, Bolun
Hussein, Guleid
Korsapati, Hariprasad
Labban, Mohamad El
Iheasirim, Kingsley
Hassan, Mohamed
Anil, Gokhan
Bartlett, Brian
Sun, Jimeng
contents Process-supervised reward models (PRMs) excel at providing step-by-step verification for large language model (LLM) outputs in domains like mathematics and coding. However, their application to fields lacking ground-truth answers, such as clinical note generation, poses significant challenges. We introduce a novel framework for training PRMs to deliver step-level reward signals for LLM-generated clinical notes. By precisely defining meaningful "steps," injecting realistic "errors" informed by domain expertise, and leveraging LLMs to generate process supervision data at scale, we overcome previous limitations. Our PRM, built on LLaMA-3.1 8B, consistently outperforms proprietary reasoning and non-reasoning models, achieving state-of-the-art performance on two key evaluations: (1) distinguishing gold-standard from error-containing samples with 98.8% accuracy, and (2) selecting physician-preferred clinical notes with 56.2% accuracy. We investigate critical components for effective PRM training, including optimal loss functions and data selection strategies, and present a comprehensive physician reader study identifying predictors of downstream Best-of-N performance. Our study sheds light on unlocking the potential of PRMs for diverse generative tasks across domains.
format Preprint
id arxiv_https___arxiv_org_abs_2412_12583
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Process-Supervised Reward Models for Verifying Clinical Note Generation: A Scalable Approach Guided by Domain Expertise
Wang, Hanyin
Gao, Chufan
Xu, Qiping
Liu, Bolun
Hussein, Guleid
Korsapati, Hariprasad
Labban, Mohamad El
Iheasirim, Kingsley
Hassan, Mohamed
Anil, Gokhan
Bartlett, Brian
Sun, Jimeng
Computation and Language
Process-supervised reward models (PRMs) excel at providing step-by-step verification for large language model (LLM) outputs in domains like mathematics and coding. However, their application to fields lacking ground-truth answers, such as clinical note generation, poses significant challenges. We introduce a novel framework for training PRMs to deliver step-level reward signals for LLM-generated clinical notes. By precisely defining meaningful "steps," injecting realistic "errors" informed by domain expertise, and leveraging LLMs to generate process supervision data at scale, we overcome previous limitations. Our PRM, built on LLaMA-3.1 8B, consistently outperforms proprietary reasoning and non-reasoning models, achieving state-of-the-art performance on two key evaluations: (1) distinguishing gold-standard from error-containing samples with 98.8% accuracy, and (2) selecting physician-preferred clinical notes with 56.2% accuracy. We investigate critical components for effective PRM training, including optimal loss functions and data selection strategies, and present a comprehensive physician reader study identifying predictors of downstream Best-of-N performance. Our study sheds light on unlocking the potential of PRMs for diverse generative tasks across domains.
title Process-Supervised Reward Models for Verifying Clinical Note Generation: A Scalable Approach Guided by Domain Expertise
topic Computation and Language
url https://arxiv.org/abs/2412.12583