mlsanityData to 7 Oct 2026

Investigating the Catastrophic Forgetting in Multimodal Large Language Models

arXiv 2309.10313 ↗First posted 19 Sep 2023cs.CL cs.AI cs.LGcited by 119

On the map: Multimodal LLMs / vision-language models → MLLM training and visual instruction tuning

Abstract (arXiv)

Following the success of GPT4, there has been a surge in interest in multimodal large language model (MLLM) research. This line of research focuses on developing general-purpose LLMs through fine-tuning pre-trained LLMs and vision models. However, catastrophic forgetting, a notorious phenomenon where the fine-tuned model fails to retain similar performance compared to the pre-trained model, still remains an inherent problem in multimodal LLMs (MLLM). In this paper, we introduce EMT: Evaluating MulTimodality for evaluating the catastrophic forgetting in MLLMs, by treating each MLLM as an image classifier. We first apply EMT to evaluate several open-source fine-tuned MLLMs and we discover that almost all evaluated MLLMs fail to retain the same performance levels as their vision encoders on standard image classification tasks. Moreover, we continue fine-tuning LLaVA, an MLLM and utilize EMT to assess performance throughout the fine-tuning. Interestingly, our results suggest that early-stage fine-tuning on an image dataset improves performance across other image datasets, by enhancing the alignment of text and visual features. However, as fine-tuning proceeds, the MLLMs begin to hallucinate, resulting in a significant loss of generalizability, even when the image encoder remains frozen. Our results suggest that MLLMs have yet to demonstrate performance on par with their vision models on standard image classification tasks and the current MLLM fine-tuning procedure still has room for improvement.

Key facts. Cited by 119 papers in the map (data through 7 Oct 2026). Placed in MLLM training and visual instruction tuning, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 48 of its 62 references that are in the map. First posted on arXiv 19 Sep 2023.

Builds on

48 of 62 references.

Closest ancestors · 4

Same direction, newest first

  1. 2023A Survey on Multimodal Large Language Models23 Jun
  2. 2023MME: A Comprehensive Evaluation Benchmark for Multimodal Large Language Models23 Jun
  3. 2023InstructBLIP: Towards General-purpose Vision-Language Models with Instruction Tuning11 May
  4. 2023Otter: A Multi-Modal Model with In-Context Instruction Tuning5 May

Other references · 44

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,204
  3. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,456
  4. 2020Language Models are Few-Shot Learnerscited by 32,199
  5. 2023GPT-4 Technical Reportcited by 18,132
  6. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,084
  7. 2023LLaMA: Open and Efficient Foundation Language Modelscited by 13,958
  8. 2023Llama 2: Open Foundation and Fine-Tuned Chat Modelscited by 12,865
  9. 2020A Simple Framework for Contrastive Learning of Visual Representationscited by 11,979
  10. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 7,916
  11. 2023Visual Instruction Tuningcited by 7,792
  12. 2023BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Modelscited by 6,069
32 more
  1. 2022Flamingo: a Visual Language Model for Few-Shot Learningcited by 4,337
  2. 2022LAION-5B: An open large-scale dataset for training next generation image-text modelscited by 4,131
  3. 2022Scaling Instruction-Finetuned Language Modelscited by 2,997
  4. 2023MiniGPT-4: Enhancing Vision-Language Understanding with Advanced Large Language Modelscited by 2,689
  5. 2015VQA: Visual Question Answeringcited by 2,489
  6. 2023Voyager: An Open-Ended Embodied Agent with Large Language Modelscited by 2,081
  7. 2022Survey of Hallucination in Natural Language Generationcited by 2,056
  8. 2022Learn to Explain: Multimodal Reasoning via Thought Chains for Science Question Answeringcited by 1,881
  9. 2023Evaluating Object Hallucination in Large Vision-Language Modelscited by 1,696
  10. 2023Video-LLaMA: An Instruction-tuned Audio-Visual Language Model for Video Understandingcited by 1,461
  11. 2023LLaMA-Adapter: Efficient Fine-tuning of Language Models with Zero-init Attentioncited by 776
  12. 2020Prevalence of Neural Collapse during the terminal phase of deep learning trainingcited by 713
  13. 2013An Empirical Investigation of Catastrophic Forgetting in Gradient-Based Neural Networkscited by 711
  14. 2023Language Is Not All You Need: Aligning Perception with Language Modelscited by 543
  15. 2020On the Stability of Fine-tuning BERT: Misconceptions, Explanations, and Strong Baselinescited by 254
  16. 2023The False Promise of Imitating Proprietary LLMscited by 254
  17. 2020Revisiting Few-sample BERT Fine-tuningcited by 237
  18. 2021A Geometric Analysis of Neural Collapse with Unconstrained Featurescited by 226
  19. 2023Mitigating the Alignment Tax of RLHFcited by 181
  20. 2020Recall and Learn: Fine-tuning Deep Pretrained Language Models with Less Forgettingcited by 169
  21. 2021Exploring Deep Neural Networks via Layer-Peeled Model: Minority Collapse in Imbalanced Trainingcited by 166
  22. 2019Mixout: Effective Regularization to Finetune Large-scale Pretrained Language Modelscited by 141
  23. 2020Neural collapse with unconstrained featurescited by 140
  24. 2023Neural Collapse Inspired Feature-Classifier Alignment for Few-Shot Class Incremental Learningcited by 112
  25. 2022Inducing Neural Collapse in Imbalanced Learning: Do We Really Need a Learnable Classifier at the End of Deep Neural Network?cited by 100
  26. 2022Imbalance Trouble: Revisiting Neural-Collapse Geometrycited by 80
  27. 2023Towards Language Models That Can See: Computer Vision Through the LENS of Natural Languagecited by 53
  28. 2021How Should Pre-Trained Language Models Be Fine-Tuned Towards Adversarial Robustness?cited by 49
  29. 2023Understanding Imbalanced Semantic Segmentation Through Neural Collapsecited by 46
  30. 2022Neural Collapse Inspired Attraction-Repulsion-Balanced Loss for Imbalanced Learningcited by 43
  31. 2021Controlling Conditional Language Models without Catastrophic Forgettingcited by 32
  32. 2023On the Implicit Geometry of Cross-Entropy Parameterizations for Label-Imbalanced Datacited by 23

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 15; 2024: 35; 2025: 41; 2026: 28.

119 papers.

Sort:

By area:

  1. 2024Cambrian-1: A Fully Open, Vision-Centric Exploration of Multimodal LLMscited by 822
  2. 2024Eyes Wide Shut? Exploring the Visual Shortcomings of Multimodal LLMscited by 719
  3. 2023HallusionBench: An Advanced Diagnostic Suite for Entangled Language Hallucination and Visual Illusion in Large Vision-Language Modelscited by 572
  4. 2025LMM-R1: Empowering 3B LMMs with Strong Reasoning Abilities Through Two-Stage Rule-Based RLcited by 244
  5. 2024Continual Learning of Large Language Models: A Comprehensive Surveycited by 228
  6. 2023Survey on Factuality in Large Language Models: Knowledge, Retrieval and Domain-Specificitycited by 204
  7. 2025RoboBrain: A Unified Brain Model for Robotic Manipulation from Abstract to Concretecited by 134
  8. 2024The Emerged Security and Privacy of LLM Agent: A Survey with Case Studiescited by 127
  9. 2024Vision-Language Models for Medical Report Generation and Visual Question Answering: A Reviewcited by 121
  10. 2025ChatVLA: Unified Multimodal Understanding and Robot Control with Vision-Language-Action Modelcited by 112
  11. 2025Words or Vision: Do Vision-Language Models Have Blind Faith in Text?cited by 86
  12. 2024Mono-InternVL: Pushing the Boundaries of Monolithic Multimodal Large Language Models with Endogenous Visual Pre-trainingcited by 82
38 more
  1. 2023LION : Empowering Multimodal Large Language Model with Dual-Level Visual Knowledgecited by 80
  2. 2024Look Twice Before You Answer: Memory-Space Visual Retracing for Hallucination Mitigation in Multimodal Large Language Modelscited by 72
  3. 2024Towards Synergistic, Generalized, and Efficient Dual-System for Robotic Manipulationcited by 71
  4. 2024Embodied Understanding of Driving Scenarioscited by 63
  5. 2024Yo'LLaVA: Your Personalized Language and Vision Assistantcited by 61
  6. 2024CoIN: A Benchmark of Continual Instruction tuNing for Multimodel Large Language Modelcited by 60
  7. 2023Let's Think Outside the Box: Exploring Leap-of-Thought in Large Language Models with Creative Humor Generationcited by 56
  8. 2024Model Tailor: Mitigating Catastrophic Forgetting in Multi-modal Large Language Modelscited by 54
  9. 2024LOKI: A Comprehensive Synthetic Data Detection Benchmark using Large Multimodal Modelscited by 51
  10. 2023Continual Instruction Tuning for Large Multimodal Modelscited by 49
  11. 2023Can We Edit Multimodal Large Language Models?cited by 44
  12. 2024Personalized Large Language Modelscited by 44
  13. 2024Text-space Graph Foundation Models: Comprehensive Benchmarks and New Insightscited by 44
  14. 2024Recent Advances of Multimodal Continual Learning: A Comprehensive Surveycited by 39
  15. 2025SEFE: Superficial and Essential Forgetting Eliminator for Multimodal Continual Instruction Tuningcited by 32
  16. 2023What Makes for Good Visual Instructions? Synthesizing Complex Visual Reasoning Instructions for Visual Instruction Tuningcited by 28
  17. 2023Machine Vision Therapy: Multimodal Large Language Models Can Enhance Visual Robustness via Denoising In-Context Learningcited by 27
  18. 2025Learn Globally, Speak Locally: Bridging the Gaps in Multilingual Reasoningcited by 23
  19. 2025MLLM-CL: Continual Learning for Multimodal Large Language Modelscited by 21
  20. 2024WOMD-Reasoning: A Large-Scale Dataset for Interaction Reasoning in Drivingcited by 19
  21. 2024SMoLoRA: Exploring and Defying Dual Catastrophic Forgetting in Continual Visual Instruction Tuningcited by 19
  22. 2025A Survey on Video Temporal Grounding with Multimodal Large Language Modelcited by 19
  23. 2024MallowsPO: Fine-Tune Your LLM with Preference Dispersionscited by 18
  24. 2023Vision-Language Instruction Tuning: A Review and Analysiscited by 17
  25. 2024Parameter Efficient Fine-tuning of Self-supervised ViTs without Catastrophic Forgettingcited by 17
  26. 2024AI Risk Management Should Incorporate Both Safety and Securitycited by 17
  27. 2024MUNBa: Machine Unlearning via Nash Bargainingcited by 17
  28. 2025Multimodal Large Language Models for Medicine: A Comprehensive Surveycited by 16
  29. 2024Exploring the Transferability of Visual Prompting for Multimodal Large Language Modelscited by 15
  30. 2023Neural Collapse in Multi-label Learning with Pick-all-label Losscited by 14
  31. 2025Mitigating Visual Knowledge Forgetting in MLLM Instruction-tuning via Modality-decoupled Gradient Descentcited by 13
  32. 2025Mono-InternVL-1.5: Towards Cheaper and Faster Monolithic Multimodal Large Language Modelscited by 13
  33. 2024On the Limitations and Prospects of Machine Unlearning for Generative AIcited by 12
  34. 2024LLMs Can Evolve Continually on Modality for X-Modal Reasoningcited by 12
  35. 2025RationalVLA: A Rational Vision-Language-Action Model with Dual Systemcited by 12
  36. 2026ThinkJEPA: Empowering Latent World Models with Large Vision-Language Reasoning Modelcited by 12
  37. 2024Evaluating the Factuality of Large Language Models using Large-Scale Knowledge Graphscited by 10
  38. 2025A Causality-aware Paradigm for Evaluating Creativity of Multimodal Large Language Modelscited by 10

50 most cited of 119 All 119 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2026Fine-tuning MLLMs Without Forgetting Is Easier Than You Think
  2. 2024MM1: Methods, Analysis & Insights from Multimodal LLM Pre-training
  3. 2024Improving Multimodal Large Language Models Using Continual Learning
  4. 2024Learn from Downstream and Be Yourself in Multimodal Large Language Model Fine-Tuning
  5. 2026Multimodal Large Language Models as Image Classifiers
  6. 2026Predictive Regularization Against Visual Representation Degradation in Multimodal Large Language Models
  7. 2025Multi-Modality Expansion and Retention for LLMs through Parameter Merging and Decoupling
  8. 2026Model-Dowser: Data-Free Importance Probing to Mitigate Catastrophic Forgetting in Multimodal Large Language Models