# General Purpose Verification for Chain of Thought Prompting

arXiv: https://arxiv.org/abs/2405.00204 · HTML: https://mlsanity.com/paper/2405.00204/

## Key facts

- Cited by 17 papers in the map (data through 14 Oct 2026).
- Placed in Reasoning benchmarks and evaluation, one of 17 directions in LLM reasoning.
- Builds on 3 of its 42 references that are in the map.
- First posted on arXiv 30 Apr 2024.

## Placed in

- LLM reasoning / [Reasoning benchmarks and evaluation](https://mlsanity.com/lr/reasoning-benchmarks-and-evaluation/)

## Builds on (most cited first, 3 of 3)

- [Constitutional AI: Harmlessness from AI Feedback](https://mlsanity.com/paper/2212.08073/) (2022)
- [Graph of Thoughts: Solving Elaborate Problems with Large Language Models](https://mlsanity.com/paper/2308.09687/) (2023)
- [Towards a Human-like Open-Domain Chatbot](https://mlsanity.com/paper/2001.09977/) (2020)

## Built on by (most cited first, 17 of 17)

- [Towards Reasoning Era: A Survey of Long Chain-of-Thought for Reasoning Large Language Models](https://mlsanity.com/paper/2503.09567/) (2025)
- [Efficient Inference for Large Reasoning Models: A Survey](https://mlsanity.com/paper/2503.23077/) (2025)
- [What's Wrong with Your Code Generated by Large Language Models? An Extensive Study](https://mlsanity.com/paper/2407.06153/) (2024)
- [The Limits of Inference Scaling Through Resampling](https://mlsanity.com/paper/2411.17501/) (2024)
- [A Survey on Feedback-based Multi-step Reasoning for Large Language Models on Mathematics](https://mlsanity.com/paper/2502.14333/) (2025)
- [VeriCoT: Neuro-symbolic Chain-of-Thought Validation via Logical Consistency Checks](https://mlsanity.com/paper/2511.04662/) (2025)
- [REFLECT: Intervention-Supported Error Attribution for Silent Failures in LLM Agent Traces](https://mlsanity.com/paper/2606.09071/) (2026)
- [DAC: Decomposed Automation Correction for Text-to-SQL](https://mlsanity.com/paper/2408.08779/) (2024)
- [RATIONALYST: Mining Implicit Rationales for Process Supervision of Reasoning](https://mlsanity.com/paper/2410.01044/) (2024)
- [Nunchi-Bench: Benchmarking Language Models on Cultural Reasoning with a Focus on Korean Superstition](https://mlsanity.com/paper/2507.04014/) (2025)
- [From Chatbot to Digital Colleague: The Paradigm Shift Toward Persistent Autonomous AI](https://mlsanity.com/paper/2606.14502/) (2026)
- [Characterizing, Evaluating, and Optimizing Complex Reasoning](https://mlsanity.com/paper/2602.08498/) (2026)
- [Flying Pigs, FaR and Beyond: Evaluating LLM Reasoning in Counterfactual Worlds](https://mlsanity.com/paper/2505.22318/) (2025)
- [Correlation or Causation: Analyzing the Causal Structures of LLM and LRM Reasoning Process](https://mlsanity.com/paper/2509.17380/) (2025)
- [Native Reasoning Models: Training Language Models to Reason on Unverifiable Data](https://mlsanity.com/paper/2602.11549/) (2026)
- [Systematic Optimization of Open Source Large Language Models for Mathematical Reasoning](https://mlsanity.com/paper/2509.07238/) (2025)
- [LCoT-GV: Graph Attention Networks for Verifying Long Reasoning Chains in Large Language Models](https://mlsanity.com/paper/2608.30679/) (2026)

## Abstract (arXiv)

Many of the recent capabilities demonstrated by Large Language Models (LLMs) arise primarily from their ability to exploit contextual information. In this paper, we explore ways to improve reasoning capabilities of LLMs through (1) exploration of different chains of thought and (2) validation of the individual steps of the reasoning process. We propose three general principles that a model should adhere to while reasoning: (i) Relevance, (ii) Mathematical Accuracy, and (iii) Logical Consistency. We apply these constraints to the reasoning steps generated by the LLM to improve the accuracy of the final generation. The constraints are applied in the form of verifiers: the model itself is asked to verify if the generated steps satisfy each constraint. To further steer the generations towards high-quality solutions, we use the perplexity of the reasoning steps as an additional verifier. We evaluate our method on 4 distinct types of reasoning tasks, spanning a total of 9 different datasets. Experiments show that our method is always better than vanilla generation, and, in 6 out of the 9 datasets, it is better than best-of N sampling which samples N reasoning chains and picks the lowest perplexity generation.