List your company

Inference Efficiency on POPE (test)

756,000Total Time (ms)VisionZip
694,5201,109,5101,524,5001,939,490Dec 5, 2024

Evaluation results

MethodLinks
VisionZipToken=160, Backbone=LLaVA-NeXT 7B, Hardware=single NVIDIA A800-80GB2024.12756,000—27.8—37.8
FastVToken=160, Backbone=LLaVA-NeXT 7B, Hardware=single NVIDIA A800-80GB2024.121,792,000—119—1.31.8
SparseVLMToken=160, Backbone=LLaVA-NeXT 7B, Hardware=single NVIDIA A800-80GB2024.121,895,000—128—1.21.7
BaselineToken=2880, Backbone=LLaVA-NeXT 7B, Hardware=single NVIDIA A800-80GB2024.122,293,000—218———