Error Downloading Mdodel

#1
by themessier - opened

While using git commands to download the model I am facing the following error

Cloning into 'gpt-j-6B'...
remote: Enumerating objects: 100, done.
remote: Counting objects: 100% (100/100), done.
remote: Compressing objects: 100% (99/99), done.
remote: Total 100 (delta 46), reused 0 (delta 0), pack-reused 0
Receiving objects: 100% (100/100), 1.40 MiB | 1.22 MiB/s, done.
Resolving deltas: 100% (46/46), done.
Filtering content: 100% (3/3), 7.62 GiB | 5.15 MiB/s, done.
Encountered %!d(MISSING) file(s) that may not have been copied correctly on Windows:
    flax_model.msgpack
    tf_model.h5
    pytorch_model.bin

See: `git lfs help smudge` for more details.
EleutherAI org

What happens when you run git lfs help smudge

"Sorry, no usage text found for "git-lfs"

I also tried

set GIT_LFS_SKIP_SMUDGE=1
git clone https://huggingface.co/EleutherAI/gpt-j-6B

but still get the same error.

Are you on Windows?

If you're not, Encountered %!d(MISSING) file(s) that may not have been copied correctly on Windows is just a warning.

If you are on Windows, I'm not 100% sure git-lfs works with >4GB files.... 😱

No, I am not on windows, I am using Ubuntu 18.
Does that mean that the required file is being downloaded and can be used? I used the same git-lfs command for GTP-neo-125M and it cloned without any errors and warnings.

yes, that's just a warning for Windows, you should be fine

EleutherAI org

@themessier Why don’t you try using the model and see if it works.

stellaathena changed discussion status to closed

How do i use this module for a discord bot?

Sign up or log in to comment