Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Distill and Align Decomposition for Enhanced Claim Verification

Jabez Magomere, Elena Kochkina, Samuel Mensah, Simerjot Kaur, Fernando Acero, Arturo Oncevay, Charese H. Smiley, Xiaomo Liu, Manuela Veloso

TL;DR

This framework enables smaller language models to achieve state-of-the-art claim verification by jointly optimising for verification accuracy and decomposition quality.

Abstract

Complex claim verification requires decomposing sentences into verifiable subclaims, yet existing methods struggle to align decomposition quality with verification performance. We propose a reinforcement learning (RL) approach that jointly optimizes decomposition quality and verifier alignment using Group Relative Policy Optimization (GRPO). Our method integrates: (i) structured sequential reasoning; (ii) supervised finetuning on teacher-distilled exemplars; and (iii) a multi-objective reward balancing format compliance, verifier alignment, and decomposition quality. Across six evaluation settings, our trained 8B decomposer improves downstream verification performance to (71.75%) macro-F1, outperforming prompt-based approaches ((+1.99), (+6.24)) and existing RL methods ((+5.84)). Human evaluation confirms the high quality of the generated subclaims. Our framework enables smaller language models to achieve state-of-the-art claim verification by jointly optimising for verification accuracy and decomposition quality.

Distill and Align Decomposition for Enhanced Claim Verification

TL;DR

This framework enables smaller language models to achieve state-of-the-art claim verification by jointly optimising for verification accuracy and decomposition quality.

Abstract

Complex claim verification requires decomposing sentences into verifiable subclaims, yet existing methods struggle to align decomposition quality with verification performance. We propose a reinforcement learning (RL) approach that jointly optimizes decomposition quality and verifier alignment using Group Relative Policy Optimization (GRPO). Our method integrates: (i) structured sequential reasoning; (ii) supervised finetuning on teacher-distilled exemplars; and (iii) a multi-objective reward balancing format compliance, verifier alignment, and decomposition quality. Across six evaluation settings, our trained 8B decomposer improves downstream verification performance to (71.75%) macro-F1, outperforming prompt-based approaches ((+1.99), (+6.24)) and existing RL methods ((+5.84)). Human evaluation confirms the high quality of the generated subclaims. Our framework enables smaller language models to achieve state-of-the-art claim verification by jointly optimising for verification accuracy and decomposition quality.
Paper Structure (53 sections, 3 equations, 3 figures, 11 tables)

This paper contains 53 sections, 3 equations, 3 figures, 11 tables.

Table of Contents

  1. Introduction
  2. Problem Formulation
  3. Our Approach
  4. Reframing Decomposition as a Sequential Reasoning Task
  5. Distilling High-Quality Decomposition Exemplars from a Teacher Model
  6. Aligning the Student Policy via Reinforcement Learning with a Multi-Objective Reward
  7. Reinforcement Learning Formulation
  8. Group Relative Policy Optimization
  9. Multi-Objective Reward Function
  10. Experimental Setup
  11. Datasets
  12. RL.
  13. Evaluation Datasets.
  14. Models
  15. Baselines.
  16. ...and 38 more sections

Figures (3)

  • Figure 1: Distill-Align-Decompose (DAD). We (1) distill examples from a teacher model, (2) frame decomposition as a sequential reasoning task, and (3) align the policy via GRPO using multi-objective rewards.
  • Figure 2: Verifier training rewards (dense vs. sparse). Training the base policy with the verifier Brier reward ($\bullet$ dense) yields smoother, more sample-efficient learning than the sparse accuracy reward ($\bullet$ sparse).
  • Figure 3: Human-evaluated subclaim decomposition quality across the five desiderata. Our method achieves comparable completeness to baselines while maintaining high scores across other desiderata. Scores represent sentence-level averages of subclaim quality with higher values indicating better quality for a given desideratum.