wb-droid's picture
initial commit.
b5c4f7a
import gradio as gr
import torch
from torch import nn
import torchvision
from diffusers import UNet2DModel, UNet2DConditionModel, DDPMScheduler, DDPMPipeline, DDIMScheduler
from fastprogress.fastprogress import progress_bar
labels_map = {
0: "T-Shirt",
1: "Trouser",
2: "Pullover",
3: "Dress",
4: "Coat",
5: "Sandal",
6: "Shirt",
7: "Sneaker",
8: "Bag",
9: "Ankle Boot",
}
l2i = {l:i for i,l in labels_map.items()}
def label2idx(l):
return l2i[l]
unet = torch.load("unconditional01.pt", map_location=torch.device('cpu')).to("cpu")
Emb = torch.load("unconditional_emb_01.pt", map_location=torch.device('cpu')).to("cpu")
unet.eval()
sched = DDIMScheduler(beta_end=0.01)
sched.set_timesteps(20)
@torch.no_grad
def diff_sample(model, sz, sched, hidden, **kwargs):
x_t = torch.randn(sz)
preds = []
for t in progress_bar(sched.timesteps):
with torch.no_grad(): noise = model(x_t, t, hidden).sample
x_t = sched.step(noise, t, x_t, **kwargs).prev_sample
preds.append(x_t.float().cpu())
return preds
@torch.no_grad()
def generate(classChoice):
sz = (1,1,32,32)
print(classChoice)
hidden = Emb(torch.tensor([label2idx(classChoice)]*1)[:,None]).detach().to("cpu")
preds = diff_sample(unet, sz, sched, hidden, eta=1.)
return((preds[-1][0] + 0.5).squeeze().clamp(-1,1).detach().numpy())
with gr.Blocks() as demo:
gr.HTML("""<h1 align="center">Conditional Diffusion with DDIM</h1>""")
gr.HTML("""<h1 align="center">trained with FashionMNIST</h1>""")
session_data = gr.State([])
classChoice = gr.Radio(list(labels_map.values()), value="T-Shirt", label="Select the type of image to generate", info="")
sampling_button = gr.Button("Conditional image generation")
final_image = gr.Image(height=250,width=200)
sampling_button.click(
generate,
[classChoice],
[final_image],
)
demo.queue().launch(share=False, inbrowser=True)