Update app.py

app.py

@@ -86,17 +86,17 @@ class MagicTimeController:
         # config models
         self.inference_config      = OmegaConf.load(inference_config_path)[1]
 
-        # self.tokenizer             = CLIPTokenizer.from_pretrained(pretrained_model_path, subfolder="tokenizer")
-        # self.text_encoder          = CLIPTextModel.from_pretrained(pretrained_model_path, subfolder="text_encoder").cuda()
-        # self.vae                   = AutoencoderKL.from_pretrained(pretrained_model_path, subfolder="vae").cuda()
-        # self.unet                  = UNet3DConditionModel.from_pretrained_2d(pretrained_model_path, subfolder="unet", unet_additional_kwargs=OmegaConf.to_container(self.inference_config.unet_additional_kwargs)).cuda()
-        # self.text_model            = CLIPTextModel.from_pretrained("openai/clip-vit-large-patch14")
+        self.tokenizer             = CLIPTokenizer.from_pretrained(pretrained_model_path, subfolder="tokenizer")
+        self.text_encoder          = CLIPTextModel.from_pretrained(pretrained_model_path, subfolder="text_encoder").cuda()
+        self.vae                   = AutoencoderKL.from_pretrained(pretrained_model_path, subfolder="vae").cuda()
+        self.unet                  = UNet3DConditionModel.from_pretrained_2d(pretrained_model_path, subfolder="unet", unet_additional_kwargs=OmegaConf.to_container(self.inference_config.unet_additional_kwargs)).cuda()
+        self.text_model            = CLIPTextModel.from_pretrained("openai/clip-vit-large-patch14")
 
-        self.tokenizer    = tokenizer
-        self.text_encoder = text_encoder
-        self.vae          = vae
-        self.unet         = unet
-        self.text_model   = text_model
+        # self.tokenizer    = tokenizer
+        # self.text_encoder = text_encoder
+        # self.vae          = vae
+        # self.unet         = unet
+        # self.text_model   = text_model
 
         self.update_motion_module(self.motion_module_list[0])
         self.update_dreambooth(self.dreambooth_list[0])
@@ -198,13 +198,14 @@ class MagicTimeController:
         }
         return gr.Video(value=save_sample_path), gr.Json(value=json_config)
 
-inference_config = OmegaConf.load(inference_config_path)[1]
-tokenizer        =  CLIPTokenizer.from_pretrained(pretrained_model_path, subfolder="tokenizer")
-text_encoder     =  CLIPTextModel.from_pretrained(pretrained_model_path, subfolder="text_encoder").cuda()
-vae              =  AutoencoderKL.from_pretrained(pretrained_model_path, subfolder="vae").cuda()
-unet             =  UNet3DConditionModel.from_pretrained_2d(pretrained_model_path, subfolder="unet", unet_additional_kwargs=OmegaConf.to_container(inference_config.unet_additional_kwargs)).cuda()
-text_model       =  CLIPTextModel.from_pretrained("openai/clip-vit-large-patch14")
-controller       = MagicTimeController(tokenizer=tokenizer, text_encoder=text_encoder, vae=vae, unet=unet, text_model=text_model)
+# inference_config = OmegaConf.load(inference_config_path)[1]
+# tokenizer        =  CLIPTokenizer.from_pretrained(pretrained_model_path, subfolder="tokenizer")
+# text_encoder     =  CLIPTextModel.from_pretrained(pretrained_model_path, subfolder="text_encoder").cuda()
+# vae              =  AutoencoderKL.from_pretrained(pretrained_model_path, subfolder="vae").cuda()
+# unet             =  UNet3DConditionModel.from_pretrained_2d(pretrained_model_path, subfolder="unet", unet_additional_kwargs=OmegaConf.to_container(inference_config.unet_additional_kwargs)).cuda()
+# text_model       =  CLIPTextModel.from_pretrained("openai/clip-vit-large-patch14")
+# controller       = MagicTimeController(tokenizer=tokenizer, text_encoder=text_encoder, vae=vae, unet=unet, text_model=text_model)
+controller       = MagicTimeController()
 
 def ui():
     with gr.Blocks(css=css) as demo: