I might have created this redundant change while trying to make the inference work. This is unnecessary, as we have LoRA updates happening below in the code, which takes care of the two stage idea while loading trainable sam. This is GTG from my side. Thanks!
@constantinpape,
I might have created this redundant change while trying to make the inference work. This is unnecessary, as we have LoRA updates happening below in the code, which takes care of the two stage idea while loading trainable sam. This is GTG from my side. Thanks!