Catastrophic Forgetting

Follow this topic Ask about this topic

Catastrophic forgetting occurs when learning new tasks or adapting to shifting data degrades performance on earlier tasks, a central challenge for systems expected to learn over time. Common approaches include replaying examples, distilling prior outputs, regularizing important weights or gradients, and isolating updates with adapters or modular experts. Researchers study these methods in sequential vision and language benchmarks as well as foundation-model fine-tuning, multimodal systems, and robotics, measuring both new-task performance and retention. No single approach works best everywhere: stronger protection can limit adaptation, while memory, compute, and benchmark choices affect results; apparent forgetting may also reflect knowledge becoming inaccessible rather than erased.

Latest papers

351 papers59 papers in the last four weeks, up 119% on the four weeks before. 0.3% of all new papers.

Patch the Distribution Mismatch: RL Rewriting Agent for Stable Off-Policy SFT
Feb 11, 2026Jiacheng Wang, Zhijie Liu, Ping Jian +4Supervised Fine-Tuning Reinforcement Learning
Beijing Institute of Technology, Beijing, China · Tsinghua University, Beijing, China
Least but not Last: Fine-tuning Intermediate Principal Components for Better Performance-Forgetting Trade-Offs
Feb 3, 2026Alessio Quercia, Arya Bangun, Ira Assent +1Low-Rank Adaptation Parameter-Efficient Fine-Tuning
IAS-8, Forschungszentrum Juelich, Juelich, Germany · Department of Computer Science, RWTH Aachen University, Aachen, Germany · Aarhus University, Aarhus, Denmark
Spectral Imbalance Causes Forgetting in Low-Rank Continual Adaptation
Jan 31, 2026Hao Gu, Mao-Lin Luo, Zi-Hao Zhou +3VLM Adaptation Continual Learning
School of Computer Science and Engineering, Southeast University, Nanjing 210096, China · Key Laboratory of Computer Network and Information Integration (Southeast University), Ministry of Education, China
TTSA3R: Training-Free Temporal-Spatial Adaptive Persistent State for Streaming 3D Reconstruction
Jan 30, 2026Zhijie Zheng, Xinhao Xiang, Jiawei ZhangStreaming 3D Reconstruction 3D Reconstruction
University of California, Davis CA 95616, USA
Spatiotemporal Continual Federated Learning for Agentic Multi-UAV Edge Networks: Mitigating Catastrophic Forgetting
Jan 29, 2026Chuan-Chi LaiMulti-Objective Reinforcement Learning Federated RL
Department of Communications Engineering, National Chung Cheng University, Minxiong Township, Chiayi County 621301, Taiwan · Advanced Institute of Manufacturing with High-tech Innovations (AIM-HI), National Chung Cheng University, Minxiong Township, Chiayi County 621301, Taiwan
Beyond Transfer Accuracy: Mechanism-Guided Controlled Adaptation for Low-Resource Languages
Jan 13, 2026Khumaisa Nur'aini, Ayu Purwarianti, Alham Fikri Aji +1Selective Fine-Tuning Circuit Discovery
Monash University Indonesia · MBZUAI · Institute Teknologi Bandung +1
When Less is More: 8-bit Quantization Improves Continual Learning in Large Language Models
Dec 22, 2025Michael S. Zhang, Rishi A. Ruia, Arnav Kewalram +3LLM Quantization Continual Learning for LLMs
Algoverse
The Urysohn Ladder: Recursive Metric Contraction for Scalable Continual Learning
Dec 20, 2025Xin LiContinual Learning Hierarchical Representation Learning
Department of Computer Science University at Albany Albany, NY 12222
Hierarchical Semantic Tree Anchoring for CLIP-Based Class-Incremental Learning
Nov 19, 2025Tao Hu, Lan Li, Zhen-Hao Xie +1Class-Incremental Learning Continual Learning for VLMs
School of Artificial Intelligence, Nanjing University · State Key Laboratory for Novel Software Technology, Nanjing University
Are Current Continual Learning Methods Truly Agnostic? Introducing OPRE, a Step Toward Agnostic Continual Learning
Nov 11, 2025Raphaël Bayle, Martial Mermillod, Robert M. FrenchContinual Learning Coreset Selection
Univ. Grenoble Alpes, Univ. Savoie Mont Blanc, CNRS, LPNC, Grenoble, France · LEAD-CNRS UMR 5022, U. de Bourgogne Europe, Dijon, France
Mixtures of SubExperts for Large Language Continual Learning
Nov 9, 2025Haeyong Kang, Hee Suk Yoon, Dahua Feng +1Continual Learning for LLMs Mixture-of-Experts Language Models
Korea Advanced Institute of Science and Technology (KAIST)
Retaining by Doing: The Role of On-Policy Data in Mitigating Forgetting
Oct 21, 2025Howard Chen, Noam Razin, Karthik Narasimhan +1Continual Learning for LLMs RL Post-Training
Princeton Language and Intelligence, Princeton University.
How to Teach Large Multimodal Models New Skills
Oct 9, 2025Zhen Zhu, Yiming Gong, Yao Xiao +2Multimodal Large Language Models Continual Learning for VLMs
University of Illinois Urbana-Champaign · Google DeepMind · Carnegie Mellon University
Foundation vs. Specialized Models: Evaluating Catastrophic Forgetting in Continual Time Series Forecasting
Oct 1, 2025Nouha Karaouli, Denis Coquenet, Elisa Fromont +2Time Series Forecasting Time-Series Foundation Models
Univ. Rennes, CNRS, Inria, IRISA - UMR 6074, F-35000 Rennes, France · Univ. Grenoble Alpes, Univ. Savoie Mont Blanc, CNRS, LPNC, Grenoble, France · Univ. Grenoble Alpes, CEA, LIST, 38000 Grenoble, France
Reinforcement Fine-Tuning Naturally Mitigates Forgetting in Continual Post-Training
Jul 7, 2025Song Lai, Haohan Zhao, Rong Feng +9Continual Learning for LLMs LLM Post-Training
Centre for Artificial Intelligence and Robotics, Hong Kong Institute of Science & Innovation, Chinese Academy of Sciences, Hong Kong, China · City University of Hong Kong, Hong Kong, China · School of Artificial Intelligence, Jilin University, Changchun, China +3
Stochastic Engrams for Efficient Continual Learning
Mar 27, 2025Isabelle Aguilar, Luis Fernando Herbozo Contreras, Omid KaveheiClass-Incremental Learning Binary Neural Networks
School of Biomedical Engineering, The University of Sydney, NSW, 2006, Australia
Improving the adaptive and continuous learning capabilities of artificial neural networks: Lessons from multi-neuromodulatory dynamics
Jan 12, 2025Jie Mei, Alejandro Rodriguez-Garcia, Daigo Takeuchi +4Continual Learning Neural Network Training Dynamics
IT:U Interdisciplinary Transformation University Austria, Linz, Austria · International Research Center for Neurointelligence, The University of Tokyo, Tokyo, Japan · Department of Anatomy, University of Quebec in Trois-Rivieres, Trois-Rivieres, QC, Canada +8
Forget About It: Mitigating Forgetting Without Replay or Auxiliary Objectives
Jan 9, 2025Anat Kleiman, Gintare Karolina Dziugaite, Jonathan Frankle +2Catastrophic Forgetting Continual Learning for LLMs
Harvard University, Kempner Institute · Google DeepMind · Databricks
Continual Memorization of Factoids in Language Models
Nov 11, 2024Howard Chen, Jiayi Geng, Adithya Bhaskar +2Continual Learning for LLMs Memorization in Language Models
Princeton Language and Intelligence (PLI), Princeton University
Large Pretraining Datasets Don't Guarantee Robustness after Fine-Tuning in Image Classification
Oct 28, 2024Jaedong Hwang, Brian Cheung, Zhang-Wei Hong +3Neural Network Robustness OOD Generalization
Massachusetts Institute of Technology
Collaborative Knowledge Distillation via a Learning-by-Education Node Community
Sep 30, 2024Anestis Kaimakamidis, Ioannis Mademlis, Ioannis PitasOnline Knowledge Distillation Continual Learning
Aristotle University of Thessaloniki, Thessaloniki, 54124, Greece

Feb 11, 2026J/K move · Enter open · S save

Patch the Distribution Mismatch: RL Rewriting Agent for Stable Off-Policy SFT

Jiacheng Wang, Zhijie Liu, Ping Jian, Zirong Chen+3

Beijing Institute of Technology, Beijing, China · Tsinghua University, Beijing, China

Supervised Fine-Tuning Reinforcement Learning LLM Fine-Tuning Catastrophic Forgetting

Open paper AskPDF arXiv

Large language models are commonly adapted to downstream tasks through supervised fine-tuning (SFT), but substantial distribution mismatch between downstream supervision and a model's generation distribution can intensify catastrophic forgetting. Data rewriting offers a data-centric way to narrow this mismatch before SFT. Existing methods, however, typically sample rewrites from a prompt-induced conditional distribution, which need not align with the backbone's natural question-answering generation distribution, and fixed templates can reduce output diversity. We formulate data rewriting as a policy-learning problem and train a lightweight LoRA rewriting policy with reinforcement learning. The policy optimizes question-answering-style distributional alignment and semantic diversity under a hard task-consistency gate, producing verified supervision for downstream SFT. Across three instruction-tuned backbones, the resulting models attain downstream gains broadly comparable to standard SFT while reducing degradation on non-downstream benchmarks in every evaluated setting. Additional experiments on logical reasoning and medical question answering provide preliminary evidence that a rewriting policy can be reused across domains for the same backbone.