DAMO-NLP-SG / VCD

[CVPR 2024 Highlight] Mitigating Object Hallucinations in Large Vision-Language Models through Visual Contrastive Decoding
Apache License 2.0
196 stars 9 forks source link

Inference Sample #23

Open Shinyzenith opened 1 week ago

Shinyzenith commented 1 week ago

Hi,

Can a proper inference sample code be provided on which we can try VCD? I've been trying to set it up with LLaVA but I think I'm doing things wrong. If the authors can provide a simple script in which we can try and see the difference between a VCD and non VCD LVLM, that would be really helpful.

chenyangzhu1 commented 2 days ago

+1