aljaz/DALLE2-pytorch
1
1
Fork 0
mirror of https://github.com/lucidrains/DALLE2-pytorch.git synced 2025-12-19 09:44:19 +01:00
Code Issues Packages Projects Releases Wiki Activity

revert restriction for classifier free guidance for diffusion prior, given @crowsonkb advice

Browse Source
This commit is contained in:
Phil Wang
2022-05-07 20:55:31 -07:00
parent 2eac7996fa
commit dde51fd362
2 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
dalle2_pytorch/dalle2_pytorch.py
View File

@@ -834,7 +834,7 @@ class DiffusionPrior(BaseGaussianDiffusion):
self.image_embed_dim = default(image_embed_dim, lambda: clip.dim_latent)
self.channels = default(image_channels, lambda: clip.image_channels)
self.cond_drop_prob = cond_drop_prob if not predict_x_start else 0.
self.cond_drop_prob = cond_drop_prob
self.condition_on_text_encodings = condition_on_text_encodings
# in paper, they do not predict the noise, but predict x0 directly for image embedding, claiming empirically better results. I'll just offer both.