Gradio demos do not use venv

[mykeehu]

I installed the program on Windows using the included requirements-Windows.txt. When I try to start the gradio demo, I get this error, which is also in the title in my venv (created by python and pip, not conda):

Traceback (most recent call last):
  File "I:\img2img-turbo\gradio_sketch2image.py", line 13, in <module>
    model = Pix2Pix_Turbo("sketch_to_image_stochastic")
  File "I:\img2img-turbo\src\pix2pix_turbo.py", line 105, in __init__
    vae.add_adapter(vae_lora_config, adapter_name="vae_skip")
  File "C:\Users\Mykee\AppData\Local\Programs\Python\Python310\lib\site-packages\diffusers\models\modeling_utils.py", line 218, in __getattr__
    return super().__getattr__(name)
  File "C:\Users\Mykee\AppData\Local\Programs\Python\Python310\lib\site-packages\torch\nn\modules\module.py", line 1695, in __getattr__
    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
AttributeError: 'AutoencoderKL' object has no attribute 'add_adapter'

If I install diffusers not in venv but in global python, it starts, but why doesn't the gradio demo work from venv?

Here is my modified files for Windows:

environment-Windows.yaml.txt
requirements-Windows.txt