Improving Multimodal Large Language Models Using Continual Learning

Shikhar Srivastava, Md Yousuf Harun, Robik Singh Shrestha, Christopher Kanan
Proceedings of The 4th Conference on Lifelong Learning Agents, PMLR 330:736-755, 2026.

Abstract

Generative large language models (LLMs) exhibit impressive capabilities, which can be further augmented by integrating a pre-trained vision model into the original LLM to create a multimodal LLM (MLLM). However, this integration often significantly decreases performance on natural language understanding and generation tasks, compared to the original LLM. This study investigates this issue using the LLaVA MLLM, treating the integration as a continual learning problem. We evaluate five continual learning methods to mitigate forgetting and identify a technique that enhances visual understanding while minimizing linguistic performance loss. Our approach reduces linguistic performance degradation by up to 15% over the LLaVA recipe, while maintaining high multimodal accuracy. We also demonstrate the robustness of our method through continual learning on a sequence of vision-language tasks, effectively preserving linguistic skills while acquiring new multimodal capabilities. Project webpage: https://shikhar-srivastava.github.io/cl-for-improving-mllms

Cite this Paper

BibTeX
@InProceedings{pmlr-v330-srivastava26a, title = {Improving Multimodal Large Language Models Using Continual Learning}, author = {Srivastava, Shikhar and Harun, Md Yousuf and Shrestha, Robik Singh and Kanan, Christopher}, booktitle = {Proceedings of The 4th Conference on Lifelong Learning Agents}, pages = {736--755}, year = {2026}, editor = {Chandar, Sarath and Pascanu, Razvan and Eaton, Eric and Liu, Bing and Mahmood, Rupam and Rannen-Triki, Amal}, volume = {330}, series = {Proceedings of Machine Learning Research}, month = {11--14 Aug}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v330/main/assets/srivastava26a/srivastava26a.pdf}, url = {https://proceedings.mlr.press/v330/srivastava26a.html}, abstract = {Generative large language models (LLMs) exhibit impressive capabilities, which can be further augmented by integrating a pre-trained vision model into the original LLM to create a multimodal LLM (MLLM). However, this integration often significantly decreases performance on natural language understanding and generation tasks, compared to the original LLM. This study investigates this issue using the LLaVA MLLM, treating the integration as a continual learning problem. We evaluate five continual learning methods to mitigate forgetting and identify a technique that enhances visual understanding while minimizing linguistic performance loss. Our approach reduces linguistic performance degradation by up to 15% over the LLaVA recipe, while maintaining high multimodal accuracy. We also demonstrate the robustness of our method through continual learning on a sequence of vision-language tasks, effectively preserving linguistic skills while acquiring new multimodal capabilities. Project webpage: https://shikhar-srivastava.github.io/cl-for-improving-mllms} }
Endnote
%0 Conference Paper %T Improving Multimodal Large Language Models Using Continual Learning %A Shikhar Srivastava %A Md Yousuf Harun %A Robik Singh Shrestha %A Christopher Kanan %B Proceedings of The 4th Conference on Lifelong Learning Agents %C Proceedings of Machine Learning Research %D 2026 %E Sarath Chandar %E Razvan Pascanu %E Eric Eaton %E Bing Liu %E Rupam Mahmood %E Amal Rannen-Triki %F pmlr-v330-srivastava26a %I PMLR %P 736--755 %U https://proceedings.mlr.press/v330/srivastava26a.html %V 330 %X Generative large language models (LLMs) exhibit impressive capabilities, which can be further augmented by integrating a pre-trained vision model into the original LLM to create a multimodal LLM (MLLM). However, this integration often significantly decreases performance on natural language understanding and generation tasks, compared to the original LLM. This study investigates this issue using the LLaVA MLLM, treating the integration as a continual learning problem. We evaluate five continual learning methods to mitigate forgetting and identify a technique that enhances visual understanding while minimizing linguistic performance loss. Our approach reduces linguistic performance degradation by up to 15% over the LLaVA recipe, while maintaining high multimodal accuracy. We also demonstrate the robustness of our method through continual learning on a sequence of vision-language tasks, effectively preserving linguistic skills while acquiring new multimodal capabilities. Project webpage: https://shikhar-srivastava.github.io/cl-for-improving-mllms
APA
Srivastava, S., Harun, M.Y., Shrestha, R.S. & Kanan, C.. (2026). Improving Multimodal Large Language Models Using Continual Learning. Proceedings of The 4th Conference on Lifelong Learning Agents, in Proceedings of Machine Learning Research 330:736-755 Available from https://proceedings.mlr.press/v330/srivastava26a.html.

Related Material