Self-Improvement in Multimodal Large Language Models: A Survey
Shijian Deng
Abstract
A survey of self-improvement methods in Multimodal Large Language Models (MLLMs) from data collection, organization, and model optimization perspectives.
Recent advancements in self-improvement for Large Language Models (LLMs) have efficiently enhanced model capabilities without significantly increasing costs, particularly in terms of human effort. While this area is still relatively young, its extension to the multimodal domain holds immense potential for leveraging diverse data sources and developing more general self-improving models. This survey is the first to provide a comprehensive overview of self-improvement in Multimodal LLMs (MLLMs). We provide a structured overview of the current literature and discuss methods from three perspectives: 1) data collection, 2) data organization, and 3) model optimization, to facilitate the further development of self-improvement in MLLMs. We also include commonly used evaluations and downstream applications. Finally, we conclude by outlining open challenges and future research directions.
Community
Self-Improvement in Multimodal Large Language Models: A Survey
This is an automated message from the Librarian Bot. I found the following papers similar to this paper.
The following papers were recommended by the Semantic Scholar API
- Towards Alignment-Centric Paradigm: A Survey of Instruction Tuning in Large Language Models (2025)
- A Survey on Training-free Alignment of Large Language Models (2025)
- Empowering Multimodal LLMs with External Tools: A Comprehensive Survey (2025)
- M3PO: Multimodal-Model-Guided Preference Optimization for Visual Instruction Following (2025)
- SearchInstruct: Enhancing Domain Adaptation via Retrieval-Based Instruction Dataset Creation (2025)
- PromptCoT 2.0: Scaling Prompt Synthesis for Large Language Model Reasoning (2025)
- Implicit Reasoning in Large Language Models: A Comprehensive Survey (2025)
Please give a thumbs up to this comment if you found it helpful!
If you want recommendations for any Paper on Hugging Face checkout this Space
You can directly ask Librarian Bot for paper recommendations by tagging it in a comment:
@librarian-bot
recommend
It would be better to include a citation to the paper available at https://huggingface.co/papers/2505.22453 😀
Thank you for bringing this to our attention. We appreciate the recommendation and will review the paper for possible inclusion in the next iteration of our survey.
I believe this paper also focuses on the similar topic but is missed out in the survey.
Wang, X., Wu, J., Lin, Z., Zhang, F., Zhang, D., & Nie, L. (2025). Video DataFlywheel: Resolving the Impossible Data Trinity in Video-Language Understanding. IEEE Transactions on Pattern Analysis and Machine Intelligence.
"DataFlywheel" is what we call "self-improvment" in the industry. A fancy name, haha.
Thank you for flagging this for us. We appreciate the insight on the "DataFlywheel" terminology and will check out the paper for our next revision.
Models citing this paper 0
No model linking this paper
Datasets citing this paper 0
No dataset linking this paper
Spaces citing this paper 0
No Space linking this paper