patrickvonplaten commited on
Commit
3172812
1 Parent(s): e362959

Correct `sample_size` of Stable Diffusion 1's unet to have correct width and height default

Browse files

Since `diffusers==0.9.0` the width and height is automatically inferred from the `sample_size` attribute of your unet's config. It seems like your diffusion model has the same architecture as Stable Diffusion 1 which means that when using this model, by default an image size of 512x512 should be generated. This in turn means the unet's sample size should be **64**.

In order to suppress to update your configuration on the fly and to suppress the deprecation warning added in this PR: https://github.com/huggingface/diffusers/pull/1406/files#r1035703505 it is strongly recommended to merge this PR.

Files changed (1) hide show
  1. unet/config.json +1 -1
unet/config.json CHANGED
@@ -26,7 +26,7 @@
26
  "norm_eps": 1e-05,
27
  "norm_num_groups": 32,
28
  "out_channels": 4,
29
- "sample_size": 32,
30
  "up_block_types": [
31
  "UpBlock2D",
32
  "CrossAttnUpBlock2D",
 
26
  "norm_eps": 1e-05,
27
  "norm_num_groups": 32,
28
  "out_channels": 4,
29
+ "sample_size": 64,
30
  "up_block_types": [
31
  "UpBlock2D",
32
  "CrossAttnUpBlock2D",