Based on training my own flux model

[KoDelioDa]

when I use run.py to train the model, the model is downloaded automatically. I want to know where this model is downloaded to, and I want to replace it with my own base model for training

[prp-e]

The better way would be tweaking the code to read from your own huggingface repo or local machine.

[0x1355]

'name_or_path' in config.

Just point that to the path of your own model.

By default huggingface-hub downloads to cache directory - typically ~/.cache on Linux. Not sure about Windows. You can use huggingface-hub to specify local_dir and put that in 'name_or_path'.

No need to tweak the code base.

[jsemrau]

You can set the HF_HOME variable to change local dir. Environment variables https://huggingface.co/docs/huggingface_hub/main/en/package_reference/environment_variables

…

On 01/22/2025 11:00 AM CET 0x1355 ***@***.***> wrote: 'name_or_path' in config. Just point that to the path of your own model. By default huggingface-hub downloads to cache directory - typically ~/.cache on Linux. Not sure about Windows. You can use huggingface-hub to specify local_dir and put that in 'name_or_path'. No need to tweak the code base. — Reply to this email directly, view it on GitHub #236 (comment), or unsubscribe https://github.com/notifications/unsubscribe-auth/AADFU66YVQZWODFHE2W2JS32L5T27AVCNFSM6AAAAABTTJUYQSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDMMBWG44DQMRZHA. You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>