AI/diffusers
1
0
Fork 0
mirror of https://github.com/huggingface/diffusers.git synced 2026-01-29 07:22:12 +03:00
Code Activity

Fix more

Browse Source
This commit is contained in:
Patrick von Platen
2023-06-23 20:48:50 +00:00
parent 4309a2c63b
commit 51ab97a2f7
1 changed files with 3 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
src/diffusers/pipelines/stable_diffusion/convert_from_ckpt.py
View File

@@ -273,19 +273,18 @@ def create_unet_diffusers_config(original_config, image_size: int, controlnet=Fa
head_dim = [5 * c for c in list(unet_params.channel_mult)]
class_embed_type = None
addition_embed_type = None
addition_time_embed_dim = None
projection_class_embeddings_input_dim = None
if "num_classes" in unet_params:
if unet_params.num_classes == "sequential":
if unet_params.context_dim == 2048:
# SDXL
class_embed_type = None
addition_embed_type = "text_time"
addition_time_embed_dim = 256
else:
class_embed_type = "projection"
addition_embed_type = None
addition_time_embed_dim = None
assert "adm_in_channels" in unet_params
projection_class_embeddings_input_dim = unet_params.adm_in_channels
else:
@@ -1244,8 +1243,6 @@ def download_from_original_stable_diffusion_ckpt(
checkpoint, unet_config, path=checkpoint_path, extract_ema=extract_ema
)
unet.load_state_dict(converted_unet_checkpoint)
# Works!
import ipdb; ipdb.set_trace()
# Convert the VAE model.
if vae_path is None:
@@ -1396,7 +1393,7 @@ def download_from_original_stable_diffusion_ckpt(
tokenizer = CLIPTokenizer.from_pretrained("openai/clip-vit-large-patch14")
text_encoder = CLIPTextModel.from_pretrained("openai/clip-vit-large-patch14")
tokenizer_2 = CLIPTokenizer.from_pretrained("laion/CLIP-ViT-bigG-14-laion2B-39B-b160k")
text_encoder_2 = CLIPTextModelWithProjection.from_pretrained("laion/CLIP-ViT-bigG-14-laion2B-39B-b160k")
text_encoder_2 = CLIPTextModelWithProjection.from_pretrained("laion/CLIP-ViT-bigG-14-laion2B-39B-b160k", projection_dim=1280)
pipe = StableDiffusionXLPipeline(
vae=vae,
text_encoder=text_encoder,