Kaiwen Yang

I will pay for it.

+1

> For me, the problem was using v1.6 LLaVA source code with v1.5 base model. > > I still do not know how to fine-tune v1.6 - BUT this fork...

> > checkpoint folder name > > > > For me, the problem was using v1.6 LLaVA source code with v1.5 base model. > > > I still do not...

Now I could not find the imagenet pretrained model, but the implementation for imagenet is similar to cifar. The only difference is we use VQ-VAE for imagenet and I have...

+1