Upgrade gradio

Fix pytz module not found

Add lightx2v deps to reduce startup time and laverage cached deps

Add deps required by fa4

Update dependencies

Set show_error=True

Fix app path

Fix torch version to torch==2.9.1

Add decord dep

Add back deps

Add debug log for app path

Add exp diffusers apps

Downgrade cuda to 12.8 for torchaudio and torchvision to match torch

move torch import after the correct version with cuda 12.9 installed

Log and trace roch initialization

Fix GRADIO_HOT_RELOAD error

Lauch app only when __name__ == __main__

Add deps instead of referencing them

Remove manuall deps installation

Disable transformers progress bar

Disable noisy logs

Support dynamically selecting app based on hf space repo name

For torch==2.9.1 built with cu129

Reference turbogen's requirements.txt

Downgrade torch to 2.9.1 with cuda 12.9

try to fix hf zero gpu config error

Put the index before the deps

Use https://download.pytorch.org/whl/cu130 as primary index

Upgrade python

fix torch version

Add torch dependency

Update readme

Switch to TurboGen

Remove sageattention installation command

Fix SageAttention build command

Fix SageAttention build

Specify target GPU architectures for building sage att

Use set FORCE_CUDA=1 env var to enable sageattention building from source

Upgrade gradio and remove cpu count logging

More accurate CPU count print for container envs

Print CPU count

Fix module name since the script expect __name__ == __main__

INstall pget

Also include inference/ in the sys path

Includ comfyui in the clone

Fix import issues

add inference dir to sys path for module import

Grant exec permission for deploy_gradio_inference.sh

Fix meta

Add readme with hf space meta