added changes

app.py

@@ -0,0 +1,23 @@
+import gradio as gr
+from torchvision.transforms import Compose, Resize, ToTensor, Normalize
+from PIL import Image
+from torchvision.utils import save_image
+
+from huggan.pytorch.pix2pix.modeling_pix2pix import GeneratorUNet
+
+transform = Compose(
+    [
+        Resize((512, 512), Image.BICUBIC),
+        ToTensor(),
+        Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5)),
+    ]
+)
+model = GeneratorUNet.from_pretrained('huggan/pix2pix-uavid-15')
+
+def predict_fn(img):
+    inp = transform(img).unsqueeze(0)
+    out = model(inp)
+    save_image(out, 'out.png', normalize=True)
+    return 'out.png'
+
+gr.Interface(predict_fn, inputs=gr.inputs.Image(type='pil'), outputs='image', examples=[['sample.jpg'], ['sample2.jpg'], ['sample3.jpg']]).launch()

requirements.txt

@@ -0,0 +1 @@
+git+https://github.com/huggingface/community-events@main