Update app.py

app.py

@@ -3,28 +3,36 @@ import gradio as gr
 import torch
 import math
 
-orig_start_prompt = "A photograph of an adult Lion"
-orig_end_prompt = "A photograph of a Lion cub"
-model_list = ["kakaobrain/karlo-v1-alpha"]
+orig_start_prompt = "a photograph of an adult lion"
+orig_end_prompt = "a photograph of a lion cub"
+
+if torch.cuda.is_available():
+    device = "cuda"
+    dtype = torch.float16
+else:
+    device = "cpu"
+    dtype = torch.bfloat16
+
+pipe = DiffusionPipeline.from_pretrained("kakaobrain/karlo-v1-alpha", torch_dtype=dtype, custom_pipeline='unclip_text_interpolation')
+pipe.to(device)
 
 def unclip_text_interpolation(
-  model_path,
   start_prompt,
   end_prompt,
-  steps
+  steps,
+  seed
 ):
-    
-    pipe = DiffusionPipeline.from_pretrained(model_path, torch_dtype=torch.bfloat16, custom_pipeline='unclip_text_interpolation')
-
-    images = pipe(start_prompt, end_prompt, steps, enable_sequential_cpu_offload=False)
+    generator = torch.Generator()
+    generator.manual_seed(seed)
 
-    return images
+    output = pipe(start_prompt, end_prompt, steps, enable_sequential_cpu_offload=False, generator=generator)
+    return output.images
 
 inputs = [
-  gr.Dropdown(model_list, value=model_list[0], label="Model"),
-  gr.inputs.Textbox(lines=5, default=orig_start_prompt, label="Start Prompt"),
-  gr.inputs.Textbox(lines=1, default=orig_end_prompt, label="End Prompt"),
-  gr.inputs.Slider(minimum=2, maximum=12, default=5, step=1, label="Steps")
+  gr.Textbox(lines=2, default=orig_start_prompt, label="Start Prompt"),
+  gr.Textbox(lines=2, default=orig_end_prompt, label="End Prompt"),
+  gr.Slider(minimum=2, maximum=12, default=5, step=1, label="Steps"),
+  gr.Number(0, label="Seed", precision=0)
 ]
 
 output = gr.Gallery(
@@ -32,19 +40,15 @@ output = gr.Gallery(
         ).style(grid=[2], height="auto")
 
 examples = [
-  ["kakaobrain/karlo-v1-alpha", orig_start_prompt, orig_end_prompt, 6],
+  [orig_start_prompt, orig_end_prompt, 5, 42],
+  ["a photo of a landscape in winter","a photo of a landscape in fall", 5, 20],
+  ["a photo of a victorian house", "a photo of a modern house", 5, 20]
 ]
 
 title = "UnClip Text Interpolation Pipeline"
-description = """<p>For faster inference without waiting in queue, you may duplicate the space and upgrade to GPU in settings.
-<br/>
-<a href="https://huggingface.co/spaces/kadirnar/stable-diffusion-2-infinite-zoom-out?duplicate=true">
-<img style="margin-top: 0em; margin-bottom: 0em" src="https://bit.ly/3gLdBN6" alt="Duplicate Space"></a>
-<p/>"""
 
 demo_app = gr.Interface(
     fn=unclip_text_interpolation,
-    description=description,
     inputs=inputs,
     outputs=output,
     title=title,
@@ -52,4 +56,4 @@ demo_app = gr.Interface(
     examples=examples,
     cache_examples=True
 )
-demo_app.launch(debug=True, enable_queue=True)
+demo_app.launch(debug=True, enable_queue=True)