Generation Stuck at 0% with Emojich Model

[thomasperge]

When running the emoji generation script with the Emojich model, the process gets stuck at 0% and does not progress. The model initializes correctly but doesn’t generate any images.

Environment :
OS: Windows
Python Version: [v3.10.10]
pip Version: 24.2
rudalle Version: 1.1.3
torch Version: 2.4.0
tokenizers Version : 0.10.3

My script :

from rudalle.pipelines import generate_images, show
from rudalle import get_rudalle_model, get_tokenizer, get_vae
from rudalle.utils import seed_everything
import torch

device = 'cuda' if torch.cuda.is_available() else 'cpu'
fp16 = torch.cuda.is_available()

try:
    dalle = get_rudalle_model('Emojich', pretrained=True, fp16=fp16, device=device)
except RuntimeError as e:
    if 'Attempting to deserialize object on a CUDA device' in str(e):
        dalle = get_rudalle_model('Emojich', pretrained=True, fp16=False, device='cpu')
    else:
        raise

tokenizer = get_tokenizer()
vae = get_vae(dwt=True).to(device)

text = 'Дональд Трамп из лего'  # Donald Trump made of LEGO

seed_everything(42)

pil_images = []

try:
    for top_k, top_p, images_num in [
        (2048, 0.9, 4),
    ]:
        images, _ = generate_images(text, tokenizer, dalle, vae, top_k=top_k, images_num=images_num, top_p=top_p, bs=8)
        pil_images.extend(images)
except Exception as e:
    print(f"An error occurred: {e}")

if pil_images:
    show(pil_images, 4)
else:
    print("No images to display.")

Error Messages :

Using device: cpu
...
Generating images with top_k=256, top_p=0.9, images_num=4
0%| 

Feel free to adjust any details as needed!
Thanks