Windows: How to save models within RAVE installation directory instead of huggingface cache

[nitinmukesh]

Hello,

I managed to install successfully on Windows. The instalation steps are easy to follow, thanks for that.

During the launch it download a lot of models, config, etc... and put them in C:\Users<username>.cache\huggingface\hub

I have a lot of installed tools related to stable diffusion which uses the same models like
diffusion_pytorch_model.safetensors:1.45G
vae\diffusion_pytorch_model.safetensors not found
tokenizer/merges.txt:525k
tokenizer/vocab.json:1.06M
vae/diffusion_pytorch_model.bin:335M
text_encoder/pytorch_model.bin:492M
safety_checker/pytorch_model.bin:1.22G
unet/diffusion_pytorch_model.bin:3.44G

To save space I use windows hard link feature. Is there a way to put these inside RAVE installation directory instead of cache folder.

[ozgurkara99]

I think it's about the directory that Huggingface uses for default path when downloading something. You can change it by changing the 'HF_HOME' environment variable as follows:

import os
os.environ['HF_HOME'] = 'my_path'

Please find more details from 'https://stackoverflow.com/questions/63312859/how-to-change-huggingface-transformers-default-cache-directory'