Zezhong Qian

About noise difference in video & image generation

2

In image generation, different views use the same noise as validation initial. `torch.stack([latents]*6,dim=0)` But in the video generation, it seems not work to do the same way. The nosie between...

About Training

2

In video training,only unet's parameters are optimized. So why don't use `torch.no_grad()` in encoder_hidden_states progressing to enhance controlnet & vae_encode speed?

How to apply VADER on my own model

2

I trained a UNet diffusion model and I want to use VADER to tune a single unet. I followed the training script to assign a peft LoRA model on my...