REVIEW 35 cited by
ProcessBench: Identifying Process Errors in Mathematical Reasoning
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
ProcessBench: Identifying Process Errors in Mathematical Reasoning
read the original abstract
As language models regularly make mistakes when solving math problems, automated identification of errors in the reasoning process becomes increasingly significant for their scalable oversight. In this paper, we introduce ProcessBench for measuring the ability to identify erroneous steps in mathematical reasoning. It consists of 3,400 test cases, primarily focused on competition- and Olympiad-level math problems. Each test case contains a step-by-step solution with error location annotated by human experts. Models are required to identify the earliest step that contains an error, or conclude that all steps are correct. We conduct extensive evaluation on ProcessBench, involving two types of models: process reward models (PRMs) and critic models, where for the latter we prompt general language models to critique each solution step by step. We draw two main observations: (1) Existing PRMs typically fail to generalize to more challenging math problems beyond GSM8K and MATH. They underperform both critic models (i.e., prompted general language models) and our own trained PRM that is straightforwardly fine-tuned on the PRM800K dataset. (2) The best open-source model, QwQ-32B-Preview, has demonstrated the critique capability competitive with the proprietary model GPT-4o, despite that it still lags behind the reasoning-specialized o1-mini. We hope ProcessBench can foster future research in reasoning process assessment, paving the way toward scalable oversight of language models.
Forward citations
Cited by 35 Pith papers
-
MedPRMBench: A Fine-grained Benchmark for Process Reward Models in Medical Reasoning
MedPRMBench is the first fine-grained benchmark for process reward models in medical reasoning, featuring 6500 questions, 13000 chains, 113910 step labels, and a baseline that improves downstream QA accuracy by 3.2-6....
-
Where Does Reasoning Break? Step-Level Hallucination Detection via Hidden-State Transport Geometry
Hallucination is detected as a transport-cost excursion in hidden-state trajectories, localized via contrastive PCA in a teacher model and distilled to a BiLSTM student.
-
OptiVerse: A Comprehensive Benchmark towards Optimization Problem Solving
OptiVerse is a new benchmark spanning neglected optimization domains that shows LLMs suffer sharp accuracy drops on hard problems due to modeling and logic errors, with a Dual-View Auditor Agent proposed to improve pe...
-
Semantic Step Prediction: Multi-Step Latent Forecasting in LLM Reasoning Trajectories via Step Sampling
Applying STP at consecutive semantic reasoning steps achieves 168x more accurate multi-step latent prediction on ProcessBench than frozen baselines, with trajectories forming smooth curves best captured by non-linear ...
-
Generate, Filter, Control, Replay: A Comprehensive Survey of Rollout Strategies for LLM Reinforcement Learning
This survey introduces the Generate-Filter-Control-Replay (GFCR) taxonomy to structure rollout pipelines for RL-based post-training of reasoning LLMs.
-
Fragile Thoughts: How Large Language Models Handle Chain-of-Thought Perturbations
LLMs show heterogeneous robustness to five types of chain-of-thought perturbations, with MathError causing 50-60% accuracy loss in small models but scaling benefits, UnitConversion remaining hard across sizes, and Ext...
-
Two Regimes of Chain-of-Thought Unfaithfulness: Behavioral Detection Fails Where Models Are Wrong
Answer correctness splits CoT unfaithfulness detection into two regimes: behavioral signals work only on correct answers, and fail at chance on incorrect answers where most unfaithfulness lives.
-
AdvancedMathBench: A Benchmark Suite for Advanced Mathematical Proof Generation and Verification
A 245-problem advanced proof benchmark plus 888 expert-labeled trajectories shows frontier LLMs remain far from reliable advanced proof generation and verification.
-
A Reference-Free Score for Detecting Silent Reasoning Failures in Large Language Models
RAFS is a reference-free composite score that flags 'silent reasoning failures'—correct answers produced by invalid chains—by combining step validity, counterfactual answer dependence, consensus, and reasoning stabili...
-
Multi-Turn On-Policy Distillation with Prefix Replay
ReOPD offline-distills multi-turn agentic LLMs via teacher-prefix replay plus step-decay sampling, matching online OPD accuracy at ≥4× speed with zero tool calls.
-
Test-Time Verification for Text-to-SQL via Outcome Reward Models
ORM-based test-time verification improves Text-to-SQL accuracy over heuristic selection by up to 4.33% on BIRD and 2.10% on Spider using automated labeling.
-
Curriculum Reinforcement Learning Can Incentivize Reasoning Capacity in LLMs Beyond the Base Model
Boundary-aware Curriculum RL raises average pass@256 by 9.8 points over base models and 10.3 points over vanilla RLVR on Qwen, Llama, and DeepSeek families.
-
Diagnosing Multi-step Reasoning Failures in Black-box LLMs via Stepwise Confidence Attribution
SCA framework applies Information Bottleneck to assign step-level confidence in black-box LLM reasoning traces, flagging errors and boosting self-correction success by up to 13.5% on math and QA tasks.
-
Diagnosing Multi-step Reasoning Failures in Black-box LLMs via Stepwise Confidence Attribution
SCA applies the Information Bottleneck principle via NIBS and GIBS methods to identify erroneous steps in black-box LLM reasoning and boosts self-correction success by up to 13.5%.
-
Process Supervision via Verbal Critique Improves Reasoning in Large Language Models
Verbal Process Supervision uses structured critiques from stronger models in an iterative loop to improve LLM reasoning, reaching 94.9% on GPQA Diamond and large gains on AIME 2025.
-
Dual-Cluster Memory Agent: Resolving Multi-Paradigm Ambiguity in Optimization Problem Solving
DCM-Agent improves LLM performance on multi-paradigm optimization problems by 11-21% via dual-cluster memory construction and dynamic inference guidance.
-
Intern-S1-MO: Long-horizon Reasoning Agent for Olympiad?Level Mathematical Problem Solving
A multi-round, lemma-memory reasoning agent with hierarchical RL reaches reported gold-medal-level scores on Olympiad math benchmarks, though the proof-based scores are self-graded.
-
E-valuator: Reliable Agent Verifiers with Sequential Hypothesis Testing
A density-ratio e-process wrapper converts black-box verifier scores into sequential decisions that control the false-alarm rate for agent trajectories, with empirical gains in early stopping.
-
Simultaneous Multi-objective Alignment Across Verifiable and Non-verifiable Rewards
MAHALO aligns LLMs to multiple objectives in one model via per-objective action heads and PRM-guided decoding, improving math, value, and tutoring metrics jointly.
-
Beyond Correctness: Harmonizing Process and Outcome Rewards through RL Training
PROF curates RL training data via PRM-ORM consistency to improve both final-answer accuracy and intermediate reasoning quality while reducing reliance on strong process reward models.
-
CoLD: Counterfactually-Guided Length Debiasing for Process Reward Models in Mathematical Reasoning
CoLD mitigates length bias in process reward models for mathematical reasoning via counterfactual guidance, length penalties, bias estimation, and joint training, improving step selection accuracy and conciseness on M...
-
Search-o1: Agentic Search-Enhanced Large Reasoning Models
Search-o1 integrates agentic retrieval-augmented generation and a Reason-in-Documents module into large reasoning models to dynamically supply missing knowledge and improve performance on complex science, math, coding...
-
Multi-Turn On-Policy Distillation with Prefix Replay
By replaying teacher prefixes with a step-decay schedule, multi-turn on-policy distillation can run without live environment interaction, matching or slightly beating online OPD accuracy.
-
Trust Region On-Policy Distillation
TrOPD stabilizes on-policy distillation for LLMs with trust-region learning, outlier estimation, and off-policy guidance, outperforming prior OPD methods on reasoning and code benchmarks.
-
Prefix-Safe Bayesian Belief Tracking for LLM Reasoning Reliability:Separating Calibration from Ranking
SBBT separates Brier-score calibration gains from AUROC ranking gains in prefix-conditioned success estimation for LLM math reasoning, with structure-aware signals yielding up to +0.110 AUROC over baselines.
-
ExComm: Exploration-Stage Communication for Error-Resilient Agentic Test-Time Scaling
ExComm adds cross-agent conflict detection and soft belief correction plus trajectory diversification to agentic test-time scaling, yielding 5-6% gains over baselines on AIME and GAIA benchmarks.
-
Dual-Cluster Memory Agent: Resolving Multi-Paradigm Ambiguity in Optimization Problem Solving
DCM-Agent improves LLM optimization solving by 11–21% on seven benchmarks via dual-cluster memory of Approaches, Checklists, and Pitfalls plus adaptive path switching.
-
Mechanistic Evidence for Faithfulness Decay in Chain-of-Thought Reasoning
Introduces NLDD, a step-level CoT faithfulness metric, and claims a consistent 'reasoning horizon' at 70-85% of chain length; the claim is unsupported by the paper's own reported data.
-
Attention Sink Forges Native MoE in Attention Layers: Sink-Aware Training to Address Head Collapse
Attention-sink weight is recast as an implicit MoE router per head, motivating a sink-aware head-balancing loss that yields small, consistent benchmark gains across three attention variants but rests on a definitional...
-
Attention Sink Forges Native MoE in Attention Layers: Sink-Aware Training to Address Head Collapse
Attention sinks forge native MoE mechanisms in attention layers that cause head collapse, addressed by sink-aware training with auxiliary load balancing.
-
Early Stopping Chain-of-thoughts in Large Language Models
ES-CoT shortens LLM chain-of-thought generation by tracking runs of identical step answers after linguistic markers, cutting tokens 16% on average while keeping accuracy comparable to full CoT across six datasets and ...
-
VRPRM: Process Reward Modeling via Visual Reasoning
VRPRM combines visual reasoning with a two-stage SFT-plus-RL strategy to deliver higher-quality process reward modeling using far less annotated data than prior non-thinking PRMs.
-
From LLM Reasoning to Autonomous AI Agents: A Comprehensive Review
A survey consolidating benchmarks, agent frameworks, real-world applications, and protocols for LLM-based autonomous agents into a proposed taxonomy with recommendations for future research.
-
Reinforcement Learning from Human Feedback
The book introduces the origins, mathematical setup, and optimization stages of RLHF including reward modeling, reinforcement learning, rejection sampling, and direct alignment algorithms.
-
Position: Multimodal Large Language Models Can Significantly Advance Scientific Reasoning
Position paper claims multimodal LLMs can significantly advance scientific reasoning and proposes a four-stage roadmap plus challenges and suggestions.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.