{"url":"/dataset/open-compass-criticbench","name":"open-compass/CriticBench","full_name":"open-compass/CriticBench","description_markdown":"[![arXiv](https://img.shields.io/badge/arXiv-2307.04725-b31b1b.svg)](https://arxiv.org/abs/2402.13764)\r\n[![license](https://img.shields.io/github/license/InternLM/opencompass.svg)](./LICENSE)\r\n\r\n[[Dataset on HF](https://huggingface.co/datasets/opencompass/CriticBench)]\r\n[[Project Page](https://open-compass.github.io/CriticBench/)]\r\n[[Subjective LeaderBoard](https://open-compass.github.io/CriticBench/leaderboard_subjective.html)]\r\n[[Objective LeaderBoard](https://open-compass.github.io/CriticBench/leaderboard_objective.html)]\r\n\r\n**CriticBench** is a novel benchmark designed to comprehensively and reliably evaluate the critique abilities of **Large Language Models (LLMs)**. These critique abilities are crucial for scalable oversight and self-improvement of LLMs. While many recent studies explore how LLMs can judge and refine flaws in their generated outputs, the measurement of critique abilities remains under-explored.\r\n\r\nHere are the key aspects of **CriticBench**:\r\n\r\n1. **Purpose**: To assess LLMs' critique abilities across four dimensions:\r\n    - **Feedback**: How well an LLM provides constructive feedback.\r\n    - **Comparison**: The ability to compare and contrast different responses.\r\n    - **Refinement**: How effectively an LLM can refine flawed or suboptimal outputs.\r\n    - **Meta-feedback**: The LLM's ability to reflect on its own performance.\r\n\r\n2. **Tasks**: CriticBench encompasses **nine diverse tasks**, each evaluating LLMs' critique abilities at varying levels of quality granularity.\r\n\r\n3. **Evaluation**: The benchmark evaluates both open-source and closed-source LLMs, revealing intriguing relationships between critique abilities, response qualities, and model scales.\r\n\r\n4. **Resources**: Datasets, resources, and an evaluation toolkit for CriticBench will be publicly released.\r\n\r\nIn summary, CriticBench aims to provide a comprehensive framework for assessing LLMs' critique and self-improvement capabilities, contributing to the advancement of large-scale language models in various applications.","description_withheld":null,"homepage":"https://github.com/open-compass/CriticBench","introduced_date":"2024-02-21","introduced_date_note":null,"introduced_by":{"paper":"/paper/criticbench-evaluating-large-language-models","title":"CriticEval: Evaluating Large Language Model as Critic","first_author":"Tian Lan","url":null},"license":{"name":"Apache License, Version 2.0","url":"https://opensource.org/license/apache-2-0"},"modalities":[{"name":"Texts","url":"/datasets/modality/texts"}],"tasks":[],"languages":[{"name":"English","url":"/datasets/language/english"}],"variants":["open-compass/CriticBench"],"data_loaders":[],"num_papers_in_archive":1,"source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28"},"benchmarks":[],"papers_with_a_benchmark_row":[],"syntology_totals":{"read_at":"2026-09-24T18:15:14+00:00","papers_with_samples":0,"samples_harvested":0,"samples_ran":0,"samples_unverified":0,"pointer_only_for_licence":0,"papers_with_no_sample_that_ran":0,"note":"the per-paper counts above, summed; not a rate"},"papers_note":"The archive never published its papers-using-dataset list; these are papers with a leaderboard row on this dataset's benchmarks."}