Requirements? about tokenflow HOT 5 OPEN

like regular stable diffusion

from tokenflow.

DL'd miniconda, created env, installed reqs, got some errors that some packages are missing, installed each missing item, then got stuck with:
\.conda\envs\tokenflow\lib\site-packages\torchvision\io\video.py:161: UserWarning: The pts_unit 'pts' gives wrong results. Please use pts_unit 'sec'. warnings.warn("The pts_unit 'pts' gives wrong results. Please use pts_unit 'sec'.") [INFO] loading stable diffusion... Traceback (most recent call last): File "F:\Progz\TokenFlow\preprocess.py", line 354, in <module> prep(opt) File "F:\Progz\TokenFlow\preprocess.py", line 315, in prep model = Preprocess(device, opt) File "F:\Progz\TokenFlow\preprocess.py", line 51, in __init__ self.vae = AutoencoderKL.from_pretrained(model_key, subfolder="vae", revision="fp16", File "\.conda\envs\tokenflow\lib\site-packages\torch\nn\modules\module.py", line 1145, in to return self._apply(convert) File "\.conda\envs\tokenflow\lib\site-packages\torch\nn\modules\module.py", line 797, in _apply module._apply(fn) File "\.conda\envs\tokenflow\lib\site-packages\torch\nn\modules\module.py", line 797, in _apply module._apply(fn) File "\.conda\envs\tokenflow\lib\site-packages\torch\nn\modules\module.py", line 820, in _apply param_applied = fn(param) File "\.conda\envs\tokenflow\lib\site-packages\torch\nn\modules\module.py", line 1143, in convert return t.to(device, dtype if t.is_floating_point() or t.is_complex() else None, non_blocking) File "\.conda\envs\tokenflow\lib\site-packages\torch\cuda\__init__.py", line 239, in _lazy_init raise AssertionError("Torch not compiled with CUDA enabled") AssertionError: Torch not compiled with CUDA enabled

Though I have CUDA and PyTorch installed

from tokenflow.

None of PyTorch, TensorFlow >= 2.0, or Flax have been found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. Traceback (most recent call last): File "F:\Progz\TokenFlow\preprocess.py", line 2, in <module> from diffusers import AutoencoderKL, UNet2DConditionModel, DDIMScheduler File "\.conda\envs\tokenflow\lib\site-packages\diffusers\__init__.py", line 3, in <module> from .configuration_utils import ConfigMixin File "\.conda\envs\tokenflow\lib\site-packages\diffusers\configuration_utils.py", line 34, in <module> from .utils import ( File "\.conda\envs\tokenflow\lib\site-packages\diffusers\utils\__init__.py", line 21, in <module> from .accelerate_utils import apply_forward_hook File "\.conda\envs\tokenflow\lib\site-packages\diffusers\utils\accelerate_utils.py", line 24, in <module> import accelerate File "\.conda\envs\tokenflow\lib\site-packages\accelerate\__init__.py", line 3, in <module> from .accelerator import Accelerator File "\.conda\envs\tokenflow\lib\site-packages\accelerate\accelerator.py", line 32, in <module> import torch ModuleNotFoundError: No module named 'torch'

although:
pip show torch

Name: torch
Version: 2.0.1
Summary: Tensors and Dynamic neural networks in Python with strong GPU acceleration
Home-page: https://pytorch.org/
Author: PyTorch Team
Author-email: [email protected]
License: BSD-3
Location: .conda\envs\tokenflow\lib\site-packages
Requires: filelock, jinja2, networkx, sympy, typing-extensions
Required-by: accelerate, kornia, torchaudio, torchvision, xformers

from tokenflow.

after creating conda env, also had to

pip install opencv-python diffusers transformers
conda install av -c conda-forge

and possibly pip install accelerate as well

from tokenflow.

nah, I think this needs freaking big VRAM.
I tried a video which is 1024x576, 30fps, 100frames, and it said it needs 50GB VRAM lmao

torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 13.73 GiB (GPU 0; 23.99 GiB total capacity; 22.73 GiB already allocated; 0 bytes free; 27.96 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF

from tokenflow.