Update app.py

app.py

@@ -66,10 +66,9 @@ device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
 
 request_log = []  
 
-clip_model = CLIPModel.from_pretrained("openai/clip-vit-base-patch32", cache_dir=model_path).to(device)
+clip_model = CLIPModel.from_pretrained("openai/clip-vit-base-patch32", cache_dir=model_path).to(torch.device("cuda:0"))
 clip_processor = CLIPProcessor.from_pretrained("openai/clip-vit-base-patch32", cache_dir=model_path)
 
-
 def compute_clip_embedding(text=None, image=None):
     """
     Compute CLIP embedding for a given text or image.
@@ -218,7 +217,7 @@ vae = load_vae(vae_dir)
 unet = load_unet(unet_dir)
 scheduler = load_scheduler(scheduler_dir)
 patchifier = SymmetricPatchifier(patch_size=1)
-text_encoder = T5EncoderModel.from_pretrained("PixArt-alpha/PixArt-XL-2-1024-MS", subfolder="text_encoder").to(device)
+text_encoder = T5EncoderModel.from_pretrained("PixArt-alpha/PixArt-XL-2-1024-MS", subfolder="text_encoder").to(torch.device("cuda:0"))
 tokenizer = T5Tokenizer.from_pretrained("PixArt-alpha/PixArt-XL-2-1024-MS", subfolder="tokenizer")
 
 pipeline = XoraVideoPipeline(
@@ -228,7 +227,7 @@ pipeline = XoraVideoPipeline(
     tokenizer=tokenizer,
     scheduler=scheduler,
     vae=vae,
-).to(device)
+).to(torch.device("cuda:0"))
 
 @spaces.GPU(duration=90)  # Dynamic duration
 def generate_video_from_text_90(