BLIP
BLIP copied to clipboard
How does the BLIP model use half-precision (FP16) inference?
How to use the demo code provided on the official website to do half-precision (FP16) inference?