DarrenZhaoFR

Results 3 comments of DarrenZhaoFR

RuntimeError when using batch size > 1

After some digging, I think the reason behind this is because, during training, they initially get a batch of image embedding with shape: (N, 256, 64, 64) , and they...

Fixes #4016 by using websockets with callback so user can manage freeing up of state objects

Hi, I was wondering if there have been any updates on this issue?

Inquiry about the Role of use_dynamic_rescale in Training vs. Inference

Same here, hope @Doubiiu can share some insights