Update app.py

app.py

@@ -487,27 +487,27 @@ scheduler = DDIMScheduler.from_pretrained(
 
 vae = AutoencoderKL.from_pretrained(base_model,
     subfolder="vae",
-    torch_dtype=torch.float16,
+    #torch_dtype=torch.float16,
 )
 if vae is None:
     vae = AutoencoderKL.from_pretrained(
         "stabilityai/sd-vae-ft-mse", 
-        torch_dtype=torch.float16,
+        #torch_dtype=torch.float16,
     )
 text_encoder = CLIPTextModel.from_pretrained(
     base_model,
     subfolder="text_encoder",
-    torch_dtype=torch.float16,
+    #torch_dtype=torch.float16,
 )
 tokenizer = CLIPTokenizer.from_pretrained(
     base_model,
     subfolder="tokenizer",
-    torch_dtype=torch.float16,
+    #torch_dtype=torch.float16,
 )
 unet = UNet2DConditionModel.from_pretrained(
     base_model,
     subfolder="unet",
-    torch_dtype=torch.float16,
+    #torch_dtype=torch.float16,
 )
 feature_extract = CLIPImageProcessor.from_pretrained(
     base_model,
@@ -604,24 +604,40 @@ def setup_model(name,clip_skip, lora_group=None,diffuser_pipeline = False ,contr
     if name not in unet_cache:
         if name not in models_single_file:
             try:
-                vae_model = AutoencoderKL.from_pretrained(model,subfolder="vae",torch_dtype=torch.float16)
+                vae_model = AutoencoderKL.from_pretrained(model,subfolder="vae"
+                                                          #,torch_dtype=torch.float16
+                                                         )
             except OSError:
-                vae_model = AutoencoderKL.from_pretrained("stabilityai/sd-vae-ft-mse", torch_dtype=torch.float16)
+                vae_model = AutoencoderKL.from_pretrained("stabilityai/sd-vae-ft-mse", 
+                                                          #torch_dtype=torch.float16
+                                                         )
 
             try:
-                unet = UNet2DConditionModel.from_pretrained(model, subfolder="unet", torch_dtype=torch.float16)
+                unet = UNet2DConditionModel.from_pretrained(model, subfolder="unet", 
+                                                            #torch_dtype=torch.float16
+                                                           )
             except OSError:
-                unet = UNet2DConditionModel.from_pretrained(base_model, subfolder="unet", torch_dtype=torch.float16)
+                unet = UNet2DConditionModel.from_pretrained(base_model, subfolder="unet", 
+                                                            #torch_dtype=torch.float16
+                                                           )
 
             try:
-                text_encoder = CLIPTextModel.from_pretrained(model, subfolder="text_encoder", torch_dtype=torch.float16)
+                text_encoder = CLIPTextModel.from_pretrained(model, subfolder="text_encoder", 
+                                                             #torch_dtype=torch.float16
+                                                            )
             except OSError:
-                text_encoder = CLIPTextModel.from_pretrained(base_model, subfolder="text_encoder", torch_dtype=torch.float16)
+                text_encoder = CLIPTextModel.from_pretrained(base_model, subfolder="text_encoder", 
+                                                             #torch_dtype=torch.float16
+                                                            )
 
             try:
-                tokenizer = CLIPTokenizer.from_pretrained(model,subfolder="tokenizer",torch_dtype=torch.float16)
+                tokenizer = CLIPTokenizer.from_pretrained(model,subfolder="tokenizer",
+                                                          #torch_dtype=torch.float16
+                                                         )
             except OSError:
-                tokenizer = CLIPTokenizer.from_pretrained(base_model,subfolder="tokenizer",torch_dtype=torch.float16)
+                tokenizer = CLIPTokenizer.from_pretrained(base_model,subfolder="tokenizer",
+                                                          #torch_dtype=torch.float16
+                                                         )
 
             try:
                 scheduler = DDIMScheduler.from_pretrained(model,subfolder="scheduler")
@@ -633,7 +649,9 @@ def setup_model(name,clip_skip, lora_group=None,diffuser_pipeline = False ,contr
             except OSError:
                 feature_extract = CLIPImageProcessor.from_pretrained(base_model,subfolder="feature_extractor")    
         else:
-            pipe_get = StableDiffusionPipeline_finetune.from_single_file(model,safety_checker= None,requires_safety_checker = False,torch_dtype=torch.float16).to(device)
+            pipe_get = StableDiffusionPipeline_finetune.from_single_file(model,safety_checker= None,requires_safety_checker = False,
+                                                                         #torch_dtype=torch.float16
+                                                                        ).to(device)
             vae_model = pipe_get.vae
             unet = pipe_get.unet
             text_encoder = pipe_get.text_encoder
@@ -2989,7 +3007,7 @@ with gr.Blocks(css=css) as demo:
 
                         prompt = gr.Textbox(
                             label="Prompt",
-                            value="loli cat girl, blue eyes, flat chest, solo, long messy silver hair, blue capelet, cat ears, cat tail, upper body",
+                            value="An adorable girl is sitting on the park",
                             show_label=True,
                             #max_lines=4,
                             placeholder="Enter prompt.",