Create app.py

app.py

@@ -0,0 +1,51 @@
+
+import torch
+import random
+import gradio as gr
+from diffusers import StableDiffusionPipeline, DDIMScheduler
+
+# Fine-tuned model weights path
+model_path = "model"
+
+# Load fine-tuned pipeline
+pipe = StableDiffusionPipeline.from_pretrained(model_path, torch_dtype=torch.float32)
+
+# Change Scheduler to DDIMScheduler
+pipe.scheduler = DDIMScheduler.from_config(pipe.scheduler.config)
+
+# For memory optimization
+pipe.enable_xformers_memory_efficient_attention()
+
+# Generate the image from the given text prompt
+def generate_response(prompt):
+
+    # define the negative prompt
+    negative_prompt = "bad anatomy, ugly, deformed, desfigured, distorted face, poorly drawn, blurry, low quality, low definition, lowres, out of frame, out of image, cropped, cut off, signature, watermark"
+
+    # initialize the variables
+    num_samples = 5
+    guidance_scale = 7.5
+    num_inference_steps = 30
+    height = 512
+    width = 512
+    
+    seed = random.randint(0, 2147483647)
+    generator = torch.Generator(device='cuda').manual_seed(seed)
+    
+    with torch.inference_mode():
+        imgs = pipe(
+            prompt,
+            negative_prompt=negative_prompt,
+            height=height, 
+            width=width,
+            num_images_per_prompt=num_samples,
+            num_inference_steps=num_inference_steps,
+            guidance_scale=guidance_scale,
+            generator=generator
+        ).images
+
+    return imgs[0]
+
+# Create and launch the Gradio UI    
+gradio_ui = gr.Interface(fn=generate_response, inputs="text", outputs="image")
+gradio_ui.launch()