# Self-improving MLLMs and synthetic data

Area: [Synthetic data and self-improvement for LLMs](https://mlsanity.com/synth/index.md). HTML: https://mlsanity.com/synth/self-improving-mllms-and-synthetic-data/

## Key facts

- Self-improving MLLMs and synthetic data: 178 papers on arXiv since 2021 (data through 14 Oct 2026).
- 40 of its 178 mapped papers (22%) cite “BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models” (2023, arXiv 2301.12597).
- 10 of its 178 mapped papers (6%) cite “Vision-Language Models Can Self-Improve Reasoning via Reflection” (2024, arXiv 2411.00855).
- 10 new papers 17 Sep–14 Oct.

## Most cited here

- [BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models](https://mlsanity.com/paper/2301.12597/) (2023), cited by 40 papers here
- [Vision-Language Models Can Self-Improve Reasoning via Reflection](https://mlsanity.com/paper/2411.00855/) (2024), cited by 10 papers here
- [Enhancing Large Vision Language Models with Self-Training on Image Comprehension](https://mlsanity.com/paper/2405.19716/) (2024), cited by 8 papers here
- [Vision-Zero: Scalable VLM Self-Improvement via Strategic Gamified Self-Play](https://mlsanity.com/paper/2509.25541/) (2025), cited by 7 papers here
- [Self-Rewarding Vision-Language Model via Reasoning Decomposition](https://mlsanity.com/paper/2508.19652/) (2025), cited by 6 papers here
- [VisPlay: Self-Evolving Vision-Language Models from Images](https://mlsanity.com/paper/2511.15661/) (2025), cited by 4 papers here
- [Calibrated Self-Rewarding Vision Language Models](https://mlsanity.com/paper/2405.14622/) (2024), cited by 3 papers here
- [Enhancing Visual-Language Modality Alignment in Large Vision Language Models via Self-Improvement](https://mlsanity.com/paper/2405.15973/) (2024), cited by 3 papers here
- [GPT4Tools: Teaching Large Language Model to Use Tools via Self-instruction](https://mlsanity.com/paper/2305.18752/) (2023), cited by 2 papers here
- [Video-STaR: Self-Training Enables Video Instruction Tuning with Any Supervision](https://mlsanity.com/paper/2407.06189/) (2024), cited by 2 papers here

## New (17 Sep–14 Oct)

- [MSP: Modality Self-Play for Training Multimodal LLMs Without Paired Cross-Modal Data](https://mlsanity.com/paper/2610.20965/) (2026-10-14)
- [NavAble: A Large-Scale Dataset and Synthetic Data Generation Pipeline for Blind Navigation](https://mlsanity.com/paper/2610.19330/) (2026-10-14)
- [SP-DocReader: Difference-Aware Self-Play for Precise Document OCR](https://mlsanity.com/paper/2610.11148/) (2026-10-08)
- [BabelFake: A Multilingual Audio-Visual DeepFake Benchmark](https://mlsanity.com/paper/2610.06339/) (2026-10-05)
- [Imagine to Act: High-Fidelity Data Synthesis via Image Editing World Model for Scalable GUI Agent Training](https://mlsanity.com/paper/2610.05861/) (2026-10-05)
- [Trinity: Self-Evolving Vision-Language Models with a Self-Verifier](https://mlsanity.com/paper/2610.04469/) (2026-10-03)
- [UniEvo-VL: An On-policy Self-Distillation Training Recipe for Multimodal Model Self-improvement](https://mlsanity.com/paper/2609.38721/) (2026-09-30)
- [Spatial-OPSD: Self-Improving Spatial Reasoning via Label-Free Self-Distillation](https://mlsanity.com/paper/2609.37055/) (2026-09-29)
- [SpatialSkill: Self-Evolving Skills for Cross-View Spatial Reasoning](https://mlsanity.com/paper/2609.34124/) (2026-09-28)
- [ReVision3D: Attribution-Guided Recursive Self-Improvement for 3D Medical Perception](https://mlsanity.com/paper/2609.32984/) (2026-09-26)