Update app.py

app.py

@@ -25,7 +25,7 @@ from transformers import CLIPTextModel, CLIPTokenizer
 from diffusers_vdm.pipeline import LatentVideoDiffusionPipeline
 from diffusers_vdm.utils import resize_and_center_crop, save_bcthw_as_mp4
 
-
+@spaces.GPU()
 class ModifiedUNet(UNet2DConditionModel):
     @classmethod
     def from_config(cls, *args, **kwargs):
@@ -73,7 +73,7 @@ def find_best_bucket(h, w, options):
             best_bucket = (bucket_h, bucket_w)
     return best_bucket
 
-
+@spaces.GPU()
 @torch.inference_mode()
 def encode_cropped_prompt_77tokens(txt: str):
     memory_management.load_models_to_gpu(text_encoder)
@@ -85,7 +85,7 @@ def encode_cropped_prompt_77tokens(txt: str):
     text_cond = text_encoder(cond_ids, attention_mask=None).last_hidden_state
     return text_cond
 
-
+@spaces.GPU()
 @torch.inference_mode()
 def pytorch2numpy(imgs):
     results = []
@@ -96,7 +96,7 @@ def pytorch2numpy(imgs):
         results.append(y)
     return results
 
-
+@spaces.GPU()
 @torch.inference_mode()
 def numpy2pytorch(imgs):
     h = torch.from_numpy(np.stack(imgs, axis=0)).float() / 127.5 - 1.0
@@ -109,12 +109,12 @@ def resize_without_crop(image, target_width, target_height):
     resized_image = pil_image.resize((target_width, target_height), Image.LANCZOS)
     return np.array(resized_image)
 
-
+@spaces.GPU()
 @torch.inference_mode()
 def interrogator_process(x):
     return wd14tagger.default_interrogator(x)
 
-
+@spaces.GPU()
 @torch.inference_mode()
 def process(input_fg, prompt, input_undo_steps, image_width, image_height, seed, steps, n_prompt, cfg,
             progress=gr.Progress()):
@@ -154,7 +154,7 @@ def process(input_fg, prompt, input_undo_steps, image_width, image_height, seed,
 
     return pixels
 
-
+@spaces.GPU()
 @torch.inference_mode()
 def process_video_inner(image_1, image_2, prompt, seed=123, steps=25, cfg_scale=7.5, fs=3, progress_tqdm=None):
     random.seed(seed)