Create app.py

app.py

@@ -0,0 +1,60 @@
+import torch
+import torch.nn as nn
+import torch.optim as optim
+from torch.utils.data import DataLoader, Dataset
+import numpy as np
+import gradio as gr
+
+class VideoDataset(Dataset):
+    def __init__(self, videos):
+        self.videos = videos
+
+    def __len__(self):
+        return len(self.videos)
+
+    def __getitem__(self, idx):
+        return torch.tensor(self.videos[idx], dtype=torch.float32)
+
+class Generator(nn.Module):
+    def __init__(self, latent_dim, video_shape):
+        super(Generator, self).__init__()
+        self.latent_dim = latent_dim
+        self.video_shape = video_shape
+        self.fc = nn.Sequential(
+            nn.Linear(latent_dim, 128),
+            nn.ReLU(),
+            nn.Linear(128, np.prod(video_shape)),
+            nn.Tanh()
+        )
+
+    def forward(self, z):
+        x = self.fc(z)
+        x = x.view(x.size(0), *self.video_shape)
+        return x
+
+# Hyperparameters
+latent_dim = 100
+video_shape = (10, 1, 32, 32)  # 10 frames of 32x32 grayscale images
+
+# Instantiate the Generator
+generator = Generator(latent_dim, video_shape)
+
+def generate_video(noise_seed):
+    torch.manual_seed(noise_seed)
+    z = torch.randn(1, latent_dim)
+    video = generator(z).detach().numpy()[0]
+
+    # Rescale the pixel values from [-1, 1] to [0, 255]
+    video = (video + 1) * 127.5
+    video = video.astype(np.uint8)
+    return [frame[0] for frame in video]  # Return list of frames
+
+demo = gr.Interface(
+    fn=generate_video,
+    inputs=gr.Slider(0, 1000, step=1, label="Noise Seed"),
+    outputs=gr.Video(label="Generated Video"),
+    title="Simple GAN Video Generator",
+    description="Generate random grayscale videos using a simple GAN generator model."
+)
+
+demo.launch()