Multi-step Problem Solving Through a Verifier: An Empirical Analysis on Model-induced Process Supervision

• URL: http://arxiv.org/abs/2402.02658v2
• Date: Mon, 14 Oct 2024 19:33:00 GMT
• Title: Multi-step Problem Solving Through a Verifier: An Empirical Analysis on Model-induced Process Supervision
• Authors: Zihan Wang, Yunxuan Li, Yuexin Wu, Liangchen Luo, Le Hou, Hongkun Yu, Jingbo Shang,
• Abstract summary: We introduce Model-induced Process Supervision (MiPS), a novel method for automating data curation. MiPS annotates an intermediate step by sampling completions of this solution through the reasoning model, and obtaining an accuracy defined as the proportion of correct completions. Our approach significantly improves the performance of PaLM 2 on math and coding tasks.
• Score: 40.984680166762345
• License:
• Abstract: Process supervision, using a trained verifier to evaluate the intermediate steps generated by a reasoner, has demonstrated significant improvements in multi-step problem solving. In this paper, to avoid the expensive effort of human annotation on the verifier training data, we introduce Model-induced Process Supervision (MiPS), a novel method for automating data curation. MiPS annotates an intermediate step by sampling completions of this solution through the reasoning model, and obtaining an accuracy defined as the proportion of correct completions. Inaccuracies of the reasoner would cause MiPS underestimating the accuracy of intermediate steps, therefore, we suggest and empirically show that verification focusing on high predicted scores of the verifier shall be preferred over that of low predicted scores, contrary to prior observations on human curated data. Our approach significantly improves the performance of PaLM 2 on math and coding tasks (accuracy +0.67% on GSM8K, +4.16% on MATH, +0.92% on MBPP compared with an output supervision trained verifier). Additionally, our study demonstrates that the verifier exhibits strong generalization ability across different reasoning models.

Related papers

• The Lessons of Developing Process Reward Models in Mathematical Reasoning [62.165534879284735]
  Process Reward Models (PRMs) aim to identify and mitigate intermediate errors in the reasoning processes. We develop a consensus filtering mechanism that effectively integrates Monte Carlo (MC) estimation with Large Language Models (LLMs) We release a new state-of-the-art PRM that outperforms existing open-source alternatives.
  arXiv  Detail & Related papers  (2025-01-13T13:10:16Z)
• ToolComp: A Multi-Tool Reasoning & Process Supervision Benchmark [0.0]
  We introduce ToolComp, a benchmark designed to evaluate multi-step tool-use reasoning. ToolComp is developed through a collaboration between models and human annotators. We generate synthetic training data to compare the performance of outcome-supervised reward models with process-supervised reward models.
  arXiv  Detail & Related papers  (2025-01-02T15:10:52Z)
• What Do Learning Dynamics Reveal About Generalization in LLM Reasoning? [83.83230167222852]
  We find that a model's generalization behavior can be effectively characterized by a training metric we call pre-memorization train accuracy. By connecting a model's learning behavior to its generalization, pre-memorization train accuracy can guide targeted improvements to training strategies.
  arXiv  Detail & Related papers  (2024-11-12T09:52:40Z)
• Step-by-Step Reasoning for Math Problems via Twisted Sequential Monte Carlo [55.452453947359736]
  We introduce a novel verification method based on Twisted Sequential Monte Carlo (TSMC) We apply TSMC to Large Language Models by estimating the expected future rewards at partial solutions. This approach results in a more straightforward training target that eliminates the need for step-wise human annotations.
  arXiv  Detail & Related papers  (2024-10-02T18:17:54Z)
• Semi-supervised 2D Human Pose Estimation via Adaptive Keypoint Masking [2.297586471170049]
  This paper proposes an adaptive keypoint masking method, which can fully mine the information in the samples and obtain better estimation performance. The effectiveness of the proposed method is verified on COCO and MPII, outperforming the state-of-the-art semi-supervised pose estimation by 5.2% and 0.3%, respectively.
  arXiv  Detail & Related papers  (2024-04-23T08:41:50Z)
• Let's Verify Step by Step [73.58107073356732]
  We show that process supervision significantly outperforms outcome supervision for training models to solve problems. Our model solves 78% of problems from a representative subset of the MATH test set. We also release PRM800K, the complete dataset of 800,000 step-level human feedback labels used to train our best reward model.
  arXiv  Detail & Related papers  (2023-05-31T17:24:00Z)
• Boosting Out-of-Distribution Detection with Multiple Pre-trained Models [41.66566916581451]
  Post hoc detection utilizing pre-trained models has shown promising performance and can be scaled to large-scale problems. We propose a detection enhancement method by ensembling multiple detection decisions derived from a zoo of pre-trained models. Our method substantially improves the relative performance by 65.40% and 26.96% on the CIFAR10 and ImageNet benchmarks.
  arXiv  Detail & Related papers  (2022-12-24T12:11:38Z)
• Leveraging Unlabeled Data to Predict Out-of-Distribution Performance [63.740181251997306]
  Real-world machine learning deployments are characterized by mismatches between the source (training) and target (test) distributions. In this work, we investigate methods for predicting the target domain accuracy using only labeled source data and unlabeled target data. We propose Average Thresholded Confidence (ATC), a practical method that learns a threshold on the model's confidence, predicting accuracy as the fraction of unlabeled examples.
  arXiv  Detail & Related papers  (2022-01-11T23:01:12Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.