Yanhao Li

Results 2 comments of Yanhao Li

Same issue. I wrote a function with taichi for data preprocessing in a script, and call this function in my custom dataset class derived from torch.utils.data.Dataset. When the main training...

I played a little bit with the unconditional generation at 256x256 resolution and trained a model from scratch. Some generated samples during the training: step 50 ![sample-step-00050](https://github.com/lucidrains/imagen-pytorch/assets/45092998/8cc59a77-e02a-4f8f-9267-f4b4b71b5167) step 10000 ![sample-step-10001](https://github.com/lucidrains/imagen-pytorch/assets/45092998/b19dd468-cfe1-4ac2-bf5c-8b1db0ae10bf)...