Delete app.py

app.py

@@ -1,143 +0,0 @@
-import torch
-import torch.nn as nn
-import torch.nn.functional as F
-from tinydit import TinyDit
-from vae import Vae
-from sampler import ddim_sample, num_timesteps
-import gradio as gr
-import numpy as np
-from PIL import Image
-import os
-import subprocess
-
-device = 'cuda' if torch.cuda.is_available() else 'cpu'
-
-vae = Vae(latent_channels=16).to(device)
-tiny_dit = TinyDit(latent_channels=16, patch_size=2, dim=768, depth=12, num_classes=1).to(device)
-
-ckpt_url = "https://huggingface.co/nitesh501/tinydit/resolve/main/tinydit_.pth"
-
-if not os.path.exists("tinydit_.pth"):
-    subprocess.run(["wget", ckpt_url])
-
-ckpt = torch.load("tinydit_.pth", map_location=torch.device(device))
-
-tiny_dit.load_state_dict(ckpt['model'])
-vae.load_state_dict(ckpt['vae'])
-
-
-
-@torch.inference_mode()
-def generate_image(steps=50, seed=-1):
-    if seed == -1:
-        seed = torch.randint(0, 2**32, (1,)).item()
-
-    torch.manual_seed(int(seed))
-
-    x = torch.randn(1, 16, 8, 8).to(device)
-    x = x.clamp(-1, 1)
-
-    timesteps = torch.linspace(
-        num_timesteps - 1,
-        0,
-        steps,
-        device=device
-    ).long()
-
-    for i in range(len(timesteps) - 1):
-
-        t = timesteps[i]
-        t_prev = timesteps[i + 1]
-
-        t_tensor = torch.full(
-            (1,),
-            t,
-            device=device,
-            dtype=torch.long
-        )
-
-        t_prev_tensor = torch.full(
-            (1,),
-            t_prev,
-            device=device,
-            dtype=torch.long
-        )
-
-        label = torch.tensor([0], device=device)
-
-        x = ddim_sample(
-            tiny_dit,
-            x,
-            t_tensor,
-            t_prev_tensor,
-            label
-        )
-
-    img = vae.decoder(x / 0.18215)
-
-    img = (
-        img.squeeze(0)
-        .detach()
-        .cpu()
-        .numpy()
-        .transpose(1, 2, 0)
-    )
-
-    img = (img / 2 + 0.5).clip(0, 1)
-    img = (img * 255).astype(np.uint8)
-
-    pil_img = Image.fromarray(img)
-
-    return pil_img
-
-
-with gr.Blocks(theme=gr.themes.Soft()) as demo:
-
-    gr.Markdown(
-        """
-        # 🎨 TinyDiT AI Anime Face Generator
-        Generate AI anime faces using TinyDiT model
-        """
-    )
-
-    with gr.Row():
-
-        with gr.Column(scale=1):
-
-            steps = gr.Slider(
-                minimum=0,
-                maximum=100,
-                value=50,
-                step=1,
-                label="Sampling Steps"
-            )
-
-            seed = gr.Number(
-                value=-1,
-                label="Seed",
-                precision=0
-            )
-
-            generate_btn = gr.Button(
-                "🚀 Generate Image",
-                variant="primary",
-                size="lg"
-            )
-
-        with gr.Column(scale=2):
-
-            output_image = gr.Image(
-                label="",
-                type="pil",
-                height=256,
-                width=256,
-                interactive=False
-            )
-
-    generate_btn.click(
-        fn=generate_image,
-        inputs=[steps, seed],
-        outputs=output_image
-    )
-
-demo.launch()