Stuck on 0% when generating

[haddad-github]

In the first line, you can see the command that I'm using.

Loading the checkpoint shards works.

Then you see at the end it's at: 0% 0/40 [00:00<?, ?it/s]

It's been stuck like that with no progress for the past 20 minutes.

The Beautiful Soup warning shouldn't be an issue and the torch.meshgrid UserWarning is just a warning for a future release, so I don't see what is causing it to hang up.

Any ideas?

(env) C:\Users\XXXXX\Videos\LTX\LTX-Video>python inference.py --ckpt_path C:\Users\XXXXX\Videos\LTX\LTX-Video\model\ltx-video-2b-v0.9.safetensors --prompt "A beautiful sunset over the ocean" --height 512 --width 512 --num_frames 30 --seed 123

Running generation with arguments: Namespace(ckpt_path='C:\\Users\\XXXXX\\Videos\\LTX\\LTX-Video\\model\\ltx-video-2b-v0.9.safetensors', input_video_path=None, input_image_path=None, output_path=None, seed=123, num_inference_steps=40, num_images_per_prompt=1, guidance_scale=3, image_cond_noise_scale=0.15, height=512, width=512, num_frames=30, frame_rate=25, bfloat16=False, prompt='A beautiful sunset over the ocean', negative_prompt='worst quality, inconsistent motion, blurry, jittery, distorted')
Padded dimensions: 512x512x33
Downloading shards: 100%|███████████████████████████████████████████████████████████████| 2/2 [00:00<00:00, 499.95it/s]
Loading checkpoint shards: 100%|█████████████████████████████████████████████████████████| 2/2 [00:00<00:00,  3.87it/s]

Setting `clean_caption=True` requires the Beautiful Soup library but it was not found in your environment. You can install it with pip:
`pip install beautifulsoup4`. Please note that you may need to restart your runtime after installation.

Setting `clean_caption` to False...

Setting `clean_caption=True` requires the Beautiful Soup library but it was not found in your environment. You can install it with pip:
`pip install beautifulsoup4`. Please note that you may need to restart your runtime after installation.

Setting `clean_caption` to False...
  0%|                                                                                           | 0/40 [00:00<?, ?it/s]C:\Users\XXXXX\Videos\LTX\LTX-Video\env\Lib\site-packages\torch\functional.py:534: UserWarning: torch.meshgrid: in an upcoming release, it will be required to pass the indexing argument. (Triggered internally at C:\actions-runner\_work\pytorch\pytorch\builder\windows\pytorch\aten\src\ATen\native\TensorShape.cpp:3596.)
  return _VF.meshgrid(tensors, **kwargs)  # type: ignore[attr-defined]

Here are the versions I'm using:

CUDA version:

+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 566.14                 Driver Version: 566.14         CUDA Version: 12.7     |
|-----------------------------------------+------------------------+----------------------+
| GPU  Name                  Driver-Model | Bus-Id          Disp.A | Volatile Uncorr. ECC |
| Fan  Temp   Perf          Pwr:Usage/Cap |           Memory-Usage | GPU-Util  Compute M. |
|                                         |                        |               MIG M. |
|=========================================+========================+======================|
|   0  NVIDIA GeForce RTX 3060 Ti   WDDM  |   00000000:01:00.0  On |                  N/A |
|  0%   45C    P8             25W /  200W |    2045MiB /   8192MiB |     11%      Default |
|                                         |                        |                  N/A |
+-----------------------------------------+------------------------+----------------------+

Python version:

Python 3.12.7

PyTorch version:

2.5.1+cpu

[haddad-github]

UPDATE: I bypassed that particular issue by uninstalling my PyTorch installation and re-installing a particular version to use CUDA (instead of the 2.5.1 +cpu version I had) by following this:

pip uninstall torch

pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118

However now I'm getting this issue, which I don't think is related, but I'll see if I can fix it:

torch.OutOfMemoryError: CUDA out of memory. Tried to allocate 64.00 MiB. GPU 0 has a total capacity of 8.00 GiB of which 0 bytes is free. Of the allocated memory 22.17 GiB is allocated by PyTorch, and 355.16 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation.  See documentation for Memory Management  (https://pytorch.org/docs/stable/notes/cuda.html#environment-variables)

[yunming181920]

Hi haddad, I think you should try to adjust the width,height or the frames so that there won't be an OOM. I successfully run the three jsons by this way. Also there is a chance that the vram of your 3060ti is too small to load the model.

[haddad-github]

Hi haddad, I think you should try to adjust the width,height or the frames so that there won't be an OOM. I successfully run the three jsons by this way. Also there is a chance that the vram of your 3060ti is too small to load the model.

Hi Yun,

I ended up using ComfyUI with quantized models, which works.