CRITIC: Large Language Models Can Self-Correct with Tool-Interactive Critiquing

TL;DR

CRITIC introduces a framework enabling large language models to self-validate and refine their outputs using external tools, significantly improving accuracy, correctness, and safety across various tasks.

Contribution

This work presents a novel tool-interactive critiquing framework allowing LLMs to self-correct by leveraging external tools, a significant step beyond traditional static models.

Findings

CRITIC improves factual accuracy in LLM outputs.

CRITIC reduces toxicity and harmful content.

Enhanced performance in question answering and code synthesis.

Abstract

Recent developments in large language models (LLMs) have been impressive. However, these models sometimes show inconsistencies and problematic behavior, such as hallucinating facts, generating flawed code, or creating offensive and toxic content. Unlike these models, humans typically utilize external tools to cross-check and refine their initial content, like using a search engine for fact-checking, or a code interpreter for debugging. Inspired by this observation, we introduce a framework called CRITIC that allows LLMs, which are essentially "black boxes" to validate and progressively amend their own outputs in a manner similar to human interaction with tools. More specifically, starting with an initial output, CRITIC interacts with appropriate tools to evaluate certain aspects of the text, and then revises the output based on the feedback obtained during this validation process. Comprehensive evaluations involving free-form question answering, mathematical program synthesis, and toxicity reduction demonstrate that CRITIC consistently enhances the performance of LLMs. Meanwhile, our research highlights the crucial importance of external feedback in promoting the ongoing self-improvement of LLMs.