mlsanityData to 7 Oct 2026

Revisiting Catastrophic Forgetting in Large Language Model Tuning

arXiv 2406.04836 ↗First posted 7 Jun 2024cs.CL cs.AIcited by 93

Not on the map. Why?

Abstract (arXiv)

Catastrophic Forgetting (CF) means models forgetting previously acquired knowledge when learning new data. It compromises the effectiveness of large language models (LLMs) during fine-tuning, yet the underlying causes have not been thoroughly investigated. This paper takes the first step to reveal the direct link between the flatness of the model loss landscape and the extent of CF in the field of LLMs. Based on this, we introduce the sharpness-aware minimization to mitigate CF by flattening the loss landscape. Experiments on three widely-used fine-tuning datasets, spanning different model scales, demonstrate the effectiveness of our method in alleviating CF. Analyses show that we nicely complement the existing anti-forgetting strategies, further enhancing the resistance of LLMs to CF.

Key facts. Cited by 93 papers in the map (data through 7 Oct 2026). Builds on 40 of its 59 references that are in the map. First posted on arXiv 7 Jun 2024.

Builds on

40 of 59 references.

References · 40

  1. 2020Language Models are Few-Shot Learnerscited by 32,199
  2. 2021Training Verifiers to Solve Math Word Problemscited by 9,743
  3. 2020Measuring Massive Multitask Language Understandingcited by 7,768
  4. 2021Measuring Mathematical Problem Solving With the MATH Datasetcited by 5,203
  5. 2016Overcoming catastrophic forgetting in neural networkscited by 4,718
  6. 2018Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challengecited by 4,710
  7. 2019HellaSwag: Can a Machine Really Finish Your Sentence?cited by 4,018
  8. 2021Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsitycited by 3,014
  9. 2021TruthfulQA: Measuring How Models Mimic Human Falsehoodscited by 2,919
  10. 2019BoolQ: Exploring the Surprising Difficulty of Natural Yes/No Questionscited by 2,140
  11. 2018Can a Suit of Armor Conduct Electricity? A New Dataset for Open Book Question Answeringcited by 2,114
  12. 2017Gradient Episodic Memory for Continual Learningcited by 1,871
28 more
  1. 2019SuperGLUE: A Stickier Benchmark for General-Purpose Language Understanding Systemscited by 1,718
  2. 2020Sharpness-Aware Minimization for Efficiently Improving Generalizationcited by 1,404
  3. 2016On Large-Batch Training for Deep Learning: Generalization Gap and Sharp Minimacited by 1,337
  4. 2017Continual Learning with Deep Generative Replaycited by 1,152
  5. 2018Efficient Lifelong Learning with A-GEMcited by 1,044
  6. 2017RACE: Large-scale ReAding Comprehension Dataset From Examinationscited by 851
  7. 2024TinyLlama: An Open-Source Small Language Modelcited by 604
  8. 2017Computing Nonvacuous Generalization Bounds for Deep (Stochastic) Neural Networks with Many More Parameters than Training Datacited by 527
  9. 2020BatchEnsemble: An Alternative Approach to Efficient Ensemble and Lifelong Learningcited by 368
  10. 2023AgentTuning: Enabling Generalized Agent Abilities for LLMscited by 319
  11. 2022Predictability and Surprise in Large Generative Modelscited by 209
  12. 2019LAMOL: LAnguage MOdeling for Lifelong Language Learningcited by 156
  13. 2023Can ChatGPT Understand Too? A Comparative Study on ChatGPT and Fine-tuned BERTcited by 155
  14. 2022Fine-tuned Language Models are Continual Learnerscited by 152
  15. 2023Platypus: Quick, Cheap, and Powerful Refinement of LLMscited by 150
  16. 2023Towards Making the Most of ChatGPT for Machine Translationcited by 138
  17. 2020Neural CRF Model for Sentence Alignment in Text Simplificationcited by 106
  18. 2020ASSET: A Dataset for Tuning and Evaluation of Sentence Simplification Models with Multiple Rewriting Transformationscited by 101
  19. 2023Error Analysis Prompting Enables Human-Like Translation Evaluation in Large Language Modelscited by 87
  20. 2021Flattening Sharpness for Dynamic Gradient Projection Memory Benefits Continual Learningcited by 71
  21. 2019Visualizing and Understanding the Effectiveness of BERTcited by 67
  22. 2023LM-Cocktail: Resilient Tuning of Language Models via Model Mergingcited by 44
  23. 2022Improving Sharpness-Aware Minimization with Fisher Mask for Better Generalization on Language Modelscited by 42
  24. 2023Merging Experts into One: Improving Computational Efficiency of Mixture of Expertscited by 36
  25. 2021Progressive Multi-Granularity Training for Non-Autoregressive Translationcited by 25
  26. 2021Self-Guided Curriculum Learning for Neural Machine Translationcited by 9
  27. 2023From MNIST to ImageNet and Back: Benchmarking Continual Curriculum Learningcited by 6
  28. 2023Create and Find Flatness: Building Flat Training Spaces in Advance for Continual Learningcited by 3

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 6; 2025: 32; 2026: 55.

93 papers.

Sort:

By area:

  1. 2025LLM Post-Training: A Deep Dive into Reasoning Large Language Modelscited by 129
  2. 2025Memento: Fine-tuning LLM Agents without Fine-tuning LLMscited by 119
  3. 2026MemRL: Self-Evolving Agents via Runtime Reinforcement Learning on Episodic Memorycited by 113
  4. 2024Towards Lifelong Learning of Large Language Models: A Surveycited by 60
  5. 2025SEFE: Superficial and Essential Forgetting Eliminator for Multimodal Continual Instruction Tuningcited by 32
  6. 2025RL Fine-Tuning Heals OOD Forgetting in SFTcited by 27
  7. 2025R1-T1: Fully Incentivizing Translation Capability in LLMs via Reasoning Learningcited by 21
  8. 2026Just-In-Time Reinforcement Learning: Continual Learning in LLM Agents Without Gradient Updatescited by 17
  9. 2025Agent Data Protocol: Unifying Datasets for Diverse, Effective Fine-tuning of LLM Agentscited by 16
  10. 2025PixelVLA: Advancing Pixel-level Understanding in Vision-Language-Action Modelcited by 16
  11. 2025Recurrent Knowledge Identification and Fusion for Language Model Continual Learningcited by 14
  12. 2026PivotRL: High Accuracy Agentic Post-Training at Low Compute Costcited by 12
38 more
  1. 2025Online Reasoning Video Segmentation with Just-in-Time Digital Twinscited by 11
  2. 2025Midtraining Bridges Pretraining and Posttraining Distributionscited by 10
  3. 2025SmartSight: Mitigating Hallucination in Video-LLMs Without Compromising Video Understanding via Temporal Attention Collapsecited by 9
  4. 2024Demystifying Language Model Forgetting with Low-rank Example Associationscited by 8
  5. 2026From Weights to Activations: Is Steering the Next Frontier of Adaptation?cited by 8
  6. 2025LookAhead Tuning: Safer Language Models via Partial Answer Previewscited by 7
  7. 2026FOREVER: Forgetting Curve-Inspired Memory Replay for Language Model Continual Learningcited by 7
  8. 2025Keeping Medical AI Healthy and Trustworthy: A Review of Detection and Correction Methods for System Degradationcited by 6
  9. 2026Can Muon Fine-tune Adam-Pretrained Models?cited by 6
  10. 2026Not Just What's There: Enabling CLIP to Comprehend Negated Visual Descriptions Without Fine-tuningcited by 5
  11. 2026Decomposing the Basic Abilities of Large Language Models: Mitigating Cross-Task Interference in Multi-Task Instruct-Tuningcited by 5
  12. 2025Being Strong Progressively! Enhancing Knowledge Distillation of Large Language Models through a Curriculum Learning Frameworkcited by 4
  13. 2025AIMMerging: Adaptive Iterative Model Merging Using Training Trajectories for Language Model Continual Learningcited by 4
  14. 2025Donors and Recipients: On Asymmetric Transfer Across Tasks and Languages with Parameter-Efficient Fine-Tuningcited by 4
  15. 2025The Devil in the Details: Emergent Misalignment, Format and Coherence in Open-Weights LLMscited by 4
  16. 2026Agentic Design Patterns: A System-Theoretic Frameworkcited by 4
  17. 2026RAG or Learning? Understanding the Limits of LLM Adaptation under Continuous Knowledge Drift in the Real Worldcited by 4
  18. 2026(How) Learning Rates Regulate Catastrophic Overtrainingcited by 4
  19. 2026End-to-End Context Compression at Scalecited by 4
  20. 2024CogSteer: Cognition-Inspired Selective Layer Intervention for Efficiently Steering Large Language Modelscited by 3
  21. 2024A Primer on Large Language Models and their Limitationscited by 3
  22. 2025Understanding the Dilemma of Unlearning for Large Language Modelscited by 3
  23. 2026M3Searcher: Modular Multimodal Information Seeking Agency with Retrieval-Oriented Reasoningcited by 3
  24. 2026Understanding and Enforcing Weight Disentanglement in Task Arithmeticcited by 3
  25. 2026Your "Pro" LLM Subscription May Actually Be "Free": Exposing Fingerprint Spoofing Risks in LLM Inference Servicescited by 3
  26. 2025LoKI: Low-damage Knowledge Implanting of Large Language Modelscited by 2
  27. 2025Review of Hallucination Understanding in Large Language and Vision Modelscited by 2
  28. 2025Pragmatic Theories Enhance Understanding of Implied Meanings in LLMscited by 2
  29. 2025The Effectiveness of Approximate Regularized Replay for Efficient Supervised Fine-Tuning of Large Language Modelscited by 2
  30. 2026CapTrack: Multifaceted Evaluation of Forgetting in LLM Post-Trainingcited by 2
  31. 2026Learning from Execution: Self-Evolving Memory for Private-Library Code Generationcited by 2
  32. 2026Scaling Laws for Hypernetwork-Based Knowledge Injection in Large Language Modelscited by 2
  33. 2026MemSFT: Mitigating Alignment Tax with an External Parametric Memorycited by 2
  34. 2024ORBIT: Cost-Effective Dataset Curation for Large Language Model Domain Adaptation with an Astronomy Case Studycited by 1
  35. 2025Arti-"fickle" Intelligence: Using LLMs as a Tool for Inference in the Political and Social Sciencescited by 1
  36. 2025Logits Replay + MoClip: Stabilized, Low-Cost Post-Training with Minimal Forgettingcited by 1
  37. 2026Reinforcement Learning of Large Language Models for Interpretable Credit Card Fraud Detectioncited by 1
  38. 2026TMS: Trajectory-Mixed Supervision for Reward-Free, On-Policy SFTcited by 1

50 most cited of 93 All 93 (JSON)