snapshot_download does not show the name of the files downloaded in parallel

[sgugger]

I was running a code sample sent by a user for a bug and realized that snapshot_download uses some parallelism (ncie!) but doesn't give the name of the file downloaded as a result, see screenshot:

This was by executing:

from huggingface_hub import snapshot_download

model_path = snapshot_download("EleutherAI/gpt-neox-20b")

cc @Wauplin

[Wauplin]

Yes, thanks for reporting @sgugger 👍
This happens when the file is downloaded from S3 (LFS files). I fixed it as part of #1267 (see #1267 (comment)). Fix will be available in next release :)

[sgugger]

Great, thanks and sorry for the dupe issue then.

[Wauplin]

No worries ! It was not documented anywhere :)