galai icon indicating copy to clipboard operation
galai copied to clipboard

Galai not cleaning the downloaded zip files

Open Kaelorn opened this issue 2 years ago • 2 comments

I just installed galai and started downloading the standard version on notebook.

Since each update of the progress bar creates a new line of log instead of refreshing the progress bar I stopped the download to start again on terminal.

The problem is when I try to download the model again it begins to download the second file... And if you interrupt again and try again to download the model it will crash with this error message: PytorchStreamReader failed reading zip archive: failed finding central directory

I am pretty sure the zip file it downloads are not cleared so it tries to unpack incomplete files and crashes because of that

Is there any way to delete the model files to download again?

Kaelorn avatar Nov 17 '22 15:11 Kaelorn

I also stopped downloading one model file and got the same error when trying to re-download it. A manual way to delete the zip file would seem reasonable. How can I find the saving path ?

sangfrois avatar Nov 17 '22 16:11 sangfrois

I've been exploring this as well. My local files are in ~/.cache/galactica/. The comment in the code seems to be off. It indicates ~/.cache/galai/ but the code shows ~/.cache/galactica/ - https://github.com/paperswithcode/galai/blob/bef775fc8bccdd172e0bef69e77023de6cb35c5f/galai/utils.py#L121

brandonbiggs avatar Nov 18 '22 19:11 brandonbiggs

Hi all, in galai 1.1.0 we switched to transformers for checkpoints management. See the details at https://huggingface.co/docs/transformers/installation#cache-setup for information about where the cache is located and how to change it. Please reopen if you still have any issues.

mkardas avatar Dec 09 '22 11:12 mkardas