-
There seems to be a problem with the recently updated code about EMA with
**RuntimeError: Subtraction, the `-` operator, with two bool tensors is not supported. Use the `^` or `logical_xor()` operato…
-
Hi @lucidrains
In your implementation [here](https://github.com/lucidrains/DALLE2-pytorch/blob/0f319803620c204e61c6b580f65e8689ea71d6e8/dalle2_pytorch/dalle2_pytorch.py#L1566), the image embedding…
-
how are people's experience with pytorch lightning for distributed training? i tried it a while ago for GANs and it was not flexible enough, but perhaps it can do well here (if it can handle only upda…
-
Hi Phil,
I noticed that Imagen applies `noise conditional augmentation` for both two super-res models.
`we corrupt the low-resolution image with the augmentation (corresponding to aug_level), and…
-
Hi Phil,
great to see your big contributions.
Here I find might be a typo. I thought it should be like `last_dec_layer = self.enc_dec.decoders[-1].weight`
https://github.com/lucidrains/DALLE2-pyto…
-
Do you have any suggestions for generating images by not only relying on text, but on a masked image, as open describes in their blog https://openai.com/dall-e-2/?
-
so basically i took example code and modified it to use CPU instead of cuda (m1 mac)
```import torch
from dalle2_pytorch import DALLE2, DiffusionPriorNetwork, DiffusionPrior, Unet, Decoder, CLIP
…
-
I have trained the dalle 2 model, but the loss is still stuck on 0.1 - 0.2. I use AdamW optimizer with ExponentialLR scheduler. Is there any solution for this problem? I have attached my training code…
-
Hi, I meet a problem when using clip-retrieval. My dataset path is like this: /data1/train-{00000..00099}.tar. Each tar file contains .jpg and .cls which match each other. I want to use clip-retriecal…
-
It seems that unet1 progresses quickly. On my dataset, within 10-20 epochs I can get pretty good results at 64px:
![image](https://user-images.githubusercontent.com/70453896/183140800-ad429d5b-bde…