mlsanityData to 7 Oct 2026

ChatGLM-Math: Improving Math Problem-Solving in Large Language Models with a Self-Critique Pipeline

arXiv 2404.02893 ↗First posted 3 Apr 2024cs.CLcited by 47

Not on the map. Why?

Cited by 24 in LLM reasoning, but placed by its own abstract, which didn’t fit.

Abstract (arXiv)

Large language models (LLMs) have shown excellent mastering of human language, but still struggle in real-world applications that require mathematical problem-solving. While many strategies and datasets to enhance LLMs' mathematics are developed, it remains a challenge to simultaneously maintain and improve both language and mathematical capabilities in deployed LLM systems.In this work, we tailor the Self-Critique pipeline, which addresses the challenge in the feedback learning stage of LLM alignment. We first train a general Math-Critique model from the LLM itself to provide feedback signals. Then, we sequentially employ rejective fine-tuning and direct preference optimization over the LLM's own generations for data collection. Based on ChatGLM3-32B, we conduct a series of experiments on both academic and our newly created challenging dataset, MathUserEval. Results show that our pipeline significantly enhances the LLM's mathematical problem-solving while still improving its language ability, outperforming LLMs that could be two times larger. Related techniques have been deployed to ChatGLMfootnoteurlhttps://chatglm.cn, an online serving LLM. Related evaluation dataset and scripts are released at urlhttps://github.com/THUDM/ChatGLM-Math.

Key facts. Cited by 47 papers in the map (data through 7 Oct 2026). Builds on 50 of its 69 references that are in the map. First posted on arXiv 3 Apr 2024.

Builds on

50 of 69 references.

References · 50

  1. 2020Language Models are Few-Shot Learnerscited by 32,199
  2. 2022Training language models to follow instructions with human feedbackcited by 15,712
  3. 2023LLaMA: Open and Efficient Foundation Language Modelscited by 13,958
  4. 2021Training Verifiers to Solve Math Word Problemscited by 9,743
  5. 2023Judging LLM-as-a-Judge with MT-Bench and Chatbot Arenacited by 8,393
  6. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 7,916
  7. 2020Measuring Massive Multitask Language Understandingcited by 7,768
  8. 2020Scaling Laws for Neural Language Modelscited by 6,411
  9. 2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 5,601
  10. 2022PaLM: Scaling Language Modeling with Pathwayscited by 5,304
  11. 2021Measuring Mathematical Problem Solving With the MATH Datasetcited by 5,203
  12. 2023Tree of Thoughts: Deliberate Problem Solving with Large Language Modelscited by 3,155
38 more
  1. 2022Constitutional AI: Harmlessness from AI Feedbackcited by 2,950
  2. 2019PIQA: Reasoning about Physical Commonsense in Natural Languagecited by 2,818
  3. 2022BLOOM: A 176B-Parameter Open-Access Multilingual Language Modelcited by 1,968
  4. 2022Challenging BIG-Bench Tasks and Whether Chain-of-Thought Can Solve Themcited by 1,767
  5. 2023LongBench: A Bilingual, Multitask Benchmark for Long Context Understandingcited by 1,378
  6. 2018Don't Give Me the Details, Just the Summary! Topic-Aware Convolutional Neural Networks for Extreme Summarizationcited by 1,308
  7. 2023AgentBench: Evaluating LLMs as Agentscited by 1,211
  8. 2015Teaching Machines to Read and Comprehendcited by 1,039
  9. 2021GLM: General Language Model Pretraining with Autoregressive Blank Infillingcited by 1,026
  10. 2021A General Language Assistant as a Laboratory for Alignmentcited by 1,023
  11. 2023Math-Shepherd: Verify and Reinforce LLMs Step-by-step without Human Annotationscited by 895
  12. 2023Graph of Thoughts: Solving Elaborate Problems with Large Language Modelscited by 873
  13. 2022GLM-130B: An Open Bilingual Pre-trained Modelcited by 837
  14. 2023Large Language Models as Optimizerscited by 830
  15. 2023MetaMath: Bootstrap Your Own Mathematical Questions for Large Language Modelscited by 724
  16. 2023AGIEval: A Human-Centric Benchmark for Evaluating Foundation Modelscited by 701
  17. 2017Program Induction by Rationale Generation : Learning to Solve and Explain Algebraic Word Problemscited by 642
  18. 2024DeepSeek LLM: Scaling Open-Source Language Models with Longtermismcited by 621
  19. 2023WizardMath: Empowering Mathematical Reasoning for Large Language Models via Reinforced Evol-Instructcited by 618
  20. 2023MAmmoTH: Building Math Generalist Models through Hybrid Instruction Tuningcited by 483
  21. 2023Scaling Relationship on Learning Mathematical Reasoning with Large Language Modelscited by 417
  22. 2020Generative Language Modeling for Automated Theorem Provingcited by 379
  23. 2023Character-LLM: A Trainable Agent for Role-Playingcited by 335
  24. 2019Analysing Mathematical Reasoning Abilities of Neural Modelscited by 331
  25. 2023Evaluating the Performance of Large Language Models on GAOKAO Benchmarkcited by 182
  26. 2024CharacterEval: A Chinese Benchmark for Role-Playing Conversational Agent Evaluationcited by 139
  27. 2023How well do Large Language Models perform in Arithmetic tasks?cited by 130
  28. 2023Black-Box Prompt Optimization: Aligning Large Language Models without Model Trainingcited by 122
  29. 2022NumGLUE: A Suite of Fundamental yet Challenging Mathematical Reasoning Taskscited by 110
  30. 2023Chain-of-Thought Hub: A Continuous Effort to Measure Large Language Models' Reasoning Performancecited by 102
  31. 2023AlignBench: Benchmarking Chinese Alignment of Large Language Modelscited by 68
  32. 2020Ape210K: A Large-Scale and Template-Rich Dataset of Math Word Problemscited by 60
  33. 2023Evaluating and Improving Tool-Augmented Computation-Intensive Math Reasoningcited by 49
  34. 2022CSL: A Large-scale Chinese Scientific Literature Datasetcited by 45
  35. 2021Neural-Symbolic Solver for Math Word Problems with Auxiliary Taskscited by 38
  36. 2023CharacterGLM: Customizing Chinese Conversational AI Characters with Large Language Modelscited by 34
  37. 2023MathAttack: Attacking Large Language Models Towards Math Solving Abilitycited by 31
  38. 2021Adversarial Examples for Evaluating Math Word Problem Solverscited by 20

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 24; 2025: 16; 2026: 7.

47 papers.

Sort:

By area:

  1. 2024RLHF Workflow: From Reward Modeling to Online RLHFcited by 265
  2. 2024Step-DPO: Step-wise Preference Optimization for Long-chain Reasoning of LLMscited by 265
  3. 2025VisuLogic: A Benchmark for Evaluating Visual Reasoning in Multi-modal Large Language Modelscited by 102
  4. 2024DART-Math: Difficulty-Aware Rejection Tuning for Mathematical Problem-Solvingcited by 92
  5. 2025Lifelong Learning of Large Language Model based Agents: A Roadmapcited by 78
  6. 2024Self-Evolving Multi-Agent Collaboration Networks for Software Developmentcited by 73
  7. 2024A Survey of Mathematical Reasoning in the Era of Multimodal Large Language Model: Benchmark, Method & Challengescited by 55
  8. 2024Internal Consistency and Self-Feedback in Large Language Models: A Surveycited by 49
  9. 2024JiuZhang3.0: Efficiently Improving Mathematical Reasoning by Training Small Data Synthesis Modelscited by 44
  10. 2024MathCoder2: Better Math Reasoning from Continued Pretraining on Model-translated Mathematical Codecited by 39
  11. 2024Step-Controlled DPO: Leveraging Stepwise Error for Enhanced Mathematical Reasoningcited by 35
  12. 2024InternLM2.5-StepProver: Advancing Automated Theorem Proving via Critic-Guided Searchcited by 34
35 more
  1. 2025Position: Multimodal Large Language Models Can Significantly Advance Scientific Reasoningcited by 33
  2. 2024Weak-to-Strong Reasoningcited by 32
  3. 2024Diving into Self-Evolving Training for Multimodal Reasoningcited by 31
  4. 20243D-Properties: Identifying Challenges in DPO and Charting a Path Forwardcited by 24
  5. 2024Skywork-Math: Data Scaling Laws for Mathematical Reasoning in Large Language Models -- The Story Goes Oncited by 24
  6. 2025LEMMA: Learning from Errors for MatheMatical Advancement in LLMscited by 20
  7. 2025BoostStep: Boosting mathematical capability of Large Language Models via improved single-step reasoningcited by 19
  8. 2024Multilingual Large Language Models: A Systematic Surveycited by 18
  9. 2024Self-Tuning: Instructing LLMs to Effectively Acquire New Knowledge through Self-Teachingcited by 17
  10. 2024Training Language Models to Critique With Multi-agent Feedbackcited by 16
  11. 2025Full-Step-DPO: Self-Supervised Preference Optimization with Step-wise Rewards for Mathematical Reasoningcited by 14
  12. 2025APT: Improving Specialist LLM Performance with Weakness Case Acquisition and Iterative Preference Trainingcited by 12
  13. 2025HiPlan: Hierarchical Planning for LLM-Based Agents with Adaptive Global-Local Guidancecited by 10
  14. 2024System-2 Mathematical Reasoning via Enriched Instruction Tuningcited by 7
  15. 2024SIaM: Self-Improving Code-Assisted Mathematical Reasoning of Large Language Modelscited by 6
  16. 2026GanitLLM: Difficulty-Aware Bengali Mathematical Reasoning through Curriculum-GRPOcited by 6
  17. 2024SED: Self-Evaluation Decoding Enhances Large Language Models for Better Generationcited by 5
  18. 2024Leveraging Web-Crawled Data for High-Quality Fine-Tuningcited by 5
  19. 2024Can Large Language Models Invent Algorithms to Improve Themselves?: Algorithm Discovery for Recursive Self-Improvement through Reinforcement Learningcited by 4
  20. 2024Mars-PO: Multi-Agent Reasoning System Preference Optimizationcited by 3
  21. 2025DIVE: Diversified Iterative Self-Improvementcited by 3
  22. 2025Large Language Models for Predictive Analysis: How Far Are They?cited by 3
  23. 2026Supervised Fine-Tuning versus Reinforcement Learning: A Study of Post-Training Methods for Large Language Modelscited by 3
  24. 2024When Not to Answer: Evaluating Prompts on GPT Models for Effective Abstention in Unanswerable Math Word Problemscited by 2
  25. 2026Beyond Rejection Sampling: Trajectory Fusion for Scaling Mathematical Reasoningcited by 2
  26. 2025CuDIP: Enhancing Theorem Proving in LLMs via Curriculum Learning-based Direct Preference Optimizationcited by 1
  27. 2025Human-Aligned Bench: Fine-Grained Assessment of Reasoning Ability in MLLMs vs. Humanscited by 1
  28. 2025p2-TQA: A Process-based Preference Learning Framework for Self-Improving Table Question Answering Modelscited by 1
  29. 2025End-to-End Bangla AI for Solving Math Olympiad Problem Benchmark: Leveraging Large Language Model Using Integrated Approachcited by 0
  30. 2025Is the Top Still Spinning? Evaluating Subjectivity in Narrative Understandingcited by 0
  31. 2025ICH-Qwen: A Large Language Model Towards Chinese Intangible Cultural Heritagecited by 0
  32. 2026Seeking Information with RAG-Assistants: Does Model Size Matter in Human-AI Collaborations?cited by 0
  33. 2026From Prompts to Context: An Ontology-Driven Framework for Human-Generative AI Collaborationcited by 0
  34. 2026INSPIRE: An Internalize-Then-Improve Approach for Example-Driven Mathematical Reasoningcited by 0
  35. 2026When Tokenization is Secretly Output Supervisioncited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025DeepCritic: Deliberate Critique with Large Language Models
  2. 2024MathChat: Benchmarking Mathematical Reasoning and Instruction Following in Multi-Turn Interactions
  3. 2024Benchmarking Large Language Models for Math Reasoning Tasks
  4. 2024Self-Explore: Enhancing Mathematical Reasoning in Language Models with Fine-grained Rewards
  5. 2025Do LLMs Overthink Basic Math Reasoning? Benchmarking the Accuracy-Efficiency Tradeoff in Language Models
  6. 2024Enhancing LLM Reasoning via Critique Models with Test-Time and Training-Time Supervision
  7. 2024Large Language Models for Mathematical Reasoning: Progresses and Challenges
  8. 2024Large Language Models Struggle with Unreasonability in Math Problems