Less Data, Faster Convergence: Goal-Driven Data Optimization for Multimodal Instruction Tuning

Open paper

Benchmark	Model	Rank	Results
video-question-answering-on-lvbench	GDO (Temp+)	#49	Accuracy: 41.06
video-question-answering-on-mlvu	GDO (Temp+)	#55	Accuracy: 46.89
video-question-answering-on-mvbench	GDO (Temp+)	#55	Avg.: 63.65