EMNLP Findings2024
LLM Self-Correction with DeCRIM: Decompose, Critique, and Refine for Enhanced Following of Instructions with Multiple Constraints
Thomas Palmeira Ferraz, Kartik Mehta, Yu-Hsiang Lin, Haw-Shiuan Chang, Shereen Oraby, Sijia Liu, Vivek Subramanian, Tagyoung Chung, Mohit Bansal, Nanyun Peng
Abstract
Instruction following is a key capability for LLMs. However, recent studies have shown that LLMs often struggle with instructions containing multiple constraints (e.g. a request to create a social media post "in a funny tone" with "no hashtag"). Despite this, most evaluations focus solely on synthetic data. To address this, we introduce REALINSTRUCT, the first benchmark designed to evaluate LLMs' ability to follow real-world multi-constrained instructions by leveraging queries real users asked AI assistants. We also investigate modelbased evaluation as a cost-effective alternative to human annotation for this task. Our findings reveal that even the proprietary GPT-4 model fails to meet at least one constraint on over 21% of instructions, highlighting the limitations of state-of-the-art models. To address the performance gap between open-source and proprietary models, we propose the DECOM-POSE, CRITIQUE, AND REFINE (DECRIM) self-correction pipeline, which enhances LLMs' ability to follow constraints. DECRIM works by decomposing the original instruction into a list of constraints and using a Critic model to decide when and where the LLM's response needs refinement. Our results show that DECRIM improves Mistral's performance by 7.3% on RE-ALINSTRUCT and 8.0% on IFEval even with weak feedback. Moreover, we demonstrate that with strong feedback, open-source LLMs with DECRIM can outperform GPT-4 on both benchmarks.
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
8 top-tier papers have cited it since. Know what they found before your reviewers do.
Lune can draw on
the full paper · the top-tier papers that cite it · related work
Free to start. No credit card required.
Lineage
- LoRA: Low-Rank Adaptation of Large Language Models cites Language Models are Few-Shot Learners
- Incentivizing Reasoning for Advanced Instruction-Following of Large Language Models cites Chain-of-Thought Prompting Elicits Reasoning in Large Language Models
- Measuring Massive Multitask Language Understanding cites Language Models are Few-Shot Learners
- Revisiting the Reliability of Language Models in Instruction-Following cites LoRA: Low-Rank Adaptation of Large Language Models
- Safe4U: Identifying Unsound Safe Encapsulations of Unsafe Calls in Rust using LLMs cites Chain-of-Thought Prompting Elicits Reasoning in Large Language Models
- Are Checklists Really Useful for Automatic Evaluation of Generative Tasks? cites Chain-of-Thought Prompting Elicits Reasoning in Large Language Models
Cited by top-tier papers8
- Incentivizing Reasoning for Advanced Instruction-Following of Large Language ModelsYulei Qin, Gang Li, Zongyi Li, Zihan Xu et al.NeurIPS 2025 · 17 citations
- RECAST: Expanding the Boundaries of LLMs' Complex Instruction Following with Multi-Constraint DataZhengkang Guo, Wenhao Liu, Mingchen Xie, Jingwen Xu et al.ICLR 2026 · 13 citations
- Revisiting the Reliability of Language Models in Instruction-FollowingJianshuo Dong, Yutong Zhang, Liu Yan, Zhenyu Zhong et al.ACL 2026 · 3 citations
- Safe4U: Identifying Unsound Safe Encapsulations of Unsafe Calls in Rust using LLMsHuan Li, Bei Wang, Xing Hu, Xin XiaISSTA 2025 · 1 citation
- Are Checklists Really Useful for Automatic Evaluation of Generative Tasks?Momoka Furuhashi, Kouta Nakayama, Takashi Kodama, Saku SugawaraEMNLP 2025 · 1 citation
Builds on46
- Language Models are Few-Shot LearnersTom B. Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah et al.NeurIPS 2020 · 64,255 citations
- Chain-of-Thought Prompting Elicits Reasoning in Large Language ModelsJason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma et al.NeurIPS 2022 · 22,562 citations
- LoRA: Low-Rank Adaptation of Large Language ModelsEdward J. Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu et al.ICLR 2022 · 18,833 citations
- BERTScore: Evaluating Text Generation with BERTTianyi Zhang, Varsha Kishore, Felix Wu, Kilian Q. Weinberger et al.ICLR 2020 · 8,443 citations
- Measuring Massive Multitask Language UnderstandingDan Hendrycks, Collin Burns, Steven Basart, Andy Zou et al.ICLR 2021 · 7,905 citations
Related papers
- FollowBench: A Multi-level Fine-grained Constraints Following Benchmark for Large Language ModelsYuxin Jiang, Yufei Wang, Xingshan Zeng, Wanjun Zhong et al.ACL 2024 · 10 citations
- ConInstruct: Evaluating Large Language Models on Conflict Detection and Resolution in InstructionsXingwei He, Qianru Zhang, Pengfei Chen, Guanhua Chen et al.AAAI 2026 · 2 citations
- UltraIF: Advancing Instruction Following from the WildKaikai An, Li Sheng, Ganqu Cui, Shuzheng Si et al.EMNLP 2025 · 1 citation
- CFBench: A Comprehensive Constraints-Following Benchmark for LLMsTao Zhang, Chenglin Zhu, Yanjun Shen, Wenjing Luo et al.ACL 2025 · 53 citations
- MM-IFEngine: Towards Multimodal Instruction FollowingShengyuan Ding, Shenxi Wu, Xiangyu Zhao, Yuhang Zang et al.ICCV 2025 · 3 citations
