Vladimir Mandic
Vladimir Mandic
please edit the pr with the description what its trying to do. leaving the templated as-is without any explanation is not going to help.
there are quite a few available on youtube. and a good series of tutorial videos does exist and is frequently mentioned in discussions.
> Last time runing, has downloading pytorch_model for upgrating, but failed download. like you said - so now you have a partial model on disk and it fails to load....
it tells you in the console log itself > Error verifying pickled file from C:\Users\Admin/.cache\huggingface\hub\models--openai--clip-vit-large-patch14\snapshots\8d052a0f05efbaefbc9e8786ba291cfdf93e5bff\pytorch_model.bin:
```shell pip install -U GitPython==3.1.27 ``` newer version of the lib is not compatible (PRs are ready and pending approval, but in the meantime use older supported version)
i don't track model load time, but looking at different issues reported around that, any extreme times are almost always related to initial model hash calculations when model resides on...
fair point. i'll take a look at that as well, but lets get this comited first.
> The intended approach is to throw rather than return something. Do this with `assert`. done.
> 1.Add a png/jpg of the same name as one of the embeddings TI preview files need to be in the format of `.preview.png`, NOT "same name as embedding" Reason...
btw, link to youtube video is ok as intro, but at the end not much will happen unless you have a link to a doc or a repo.