How to Modify the Designated Location of the Model

SWivid / F5-TTS

Official code for "F5-TTS: A Fairytaler that Fakes Fluent and Faithful Speech with Flow Matching"

https://arxiv.org/abs/2410.06885

MIT License

7.23k stars 862 forks source link

How to Modify the Designated Location of the Model #468

Open BecarefulW opened 4 days ago

BecarefulW commented 4 days ago

Checks

[X] This template is only for question, not feature requests or bug reports.
[X] I have thoroughly reviewed the project documentation and read the related paper(s).
[X] I have searched for existing issues, including closed ones, no similar questions.
[X] I confirm that I am using English to submit this report in order to facilitate communication.

Question details

The automatically downloaded models are in the cache folder, which is not conducive to file management. May I ask how to modify the loading location of the model?

JarodMica commented 4 days ago

You could probably do something similar like my modifications here: https://github.com/JarodMica/F5-TTS/commit/760a0ee6662ab0f8c94b7e44ba839c4681fd72ee#diff-e04b7ac8329048ac5038a06eb8d8db889aebb858e399eaa1eb3690739064d1afR66

I haven't had time to formalize a PR to the main repo

SWivid commented 4 days ago

https://github.com/SWivid/F5-TTS/blob/8f65f9f3e4d47611f2c0628cb3732b1b1ba3fd0b/src/f5_tts/infer/utils_infer.py#L185 so you could just pass in you local path and fixed here https://github.com/SWivid/F5-TTS/blob/8f65f9f3e4d47611f2c0628cb3732b1b1ba3fd0b/src/f5_tts/infer/utils_infer.py#L218

BecarefulW commented 3 days ago

https://github.com/SWivid/F5-TTS/blob/8f65f9f3e4d47611f2c0628cb3732b1b1ba3fd0b/src/f5_tts/infer/utils_infer.py#L185

so you could just pass in you local path and fixed here https://github.com/SWivid/F5-TTS/blob/8f65f9f3e4d47611f2c0628cb3732b1b1ba3fd0b/src/f5_tts/infer/utils_infer.py#L218

I tried to set it as os.environ["TRANSFORMERS_CACHE"] = ".model" in the constants.py of huggingface hub, but it seems that the main model is still downloaded to the original cache directory. Since there are four models, it seems easy to make mistakes when modifying them correspondingly. Is there a way to set the cache location to the project directory or specify a folder for the project to read models?