ScalSelect: Scalable Training-Free Multimodal Data Selection for Efficient Visual Instruction Tuning
Open Access
- 9 November 2026
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 11478-11487
Abstract
No abstract availableFunding Information
- Beijing Zhongguancun Academy (c20250510)
- National Natural Science Foundation of China (62372177)
- High Innovation Plan (202504841022)
This publication has 22 references indexed in Scilit:
- A Survey on Efficient Vision‐Language ModelsWIREs Data Mining and Knowledge Discovery, 2025
- A Survey of State of the Art Large Vision Language Models: Alignment, Benchmark, Evaluations and Challenges2025
- Cross-modal Information Flow in Multimodal Large Language Models2025
- Align-KD: Distilling Cross-Modal Alignment Knowledge for Mobile Vision-Language Large Model Enhancement2025
- OCRBench: on the hidden mystery of OCR in large multimodal modelsScience China Information Sciences, 2024
- Improved Baselines with Visual Instruction Tuning2024
- A Diagram is Worth a Dozen Images2016
- CUR matrix decompositions for improved data analysisProceedings of the National Academy of Sciences, 2009
- Singular Value Decomposition and Principal Component Analysis2005
- The Approximation of One Matrix by Another of Lower RankPsychometrika, 1936