Delete organizations / models from the Hub

Hi @julien-c, I made the same mistake as others. Could you delete the user docketanalyzer and this organization. Thanks, I appreciate any help!

done – feel free to start over :slight_smile:

Hello @julien-c,
I also made a mistake in naming until I figured out what I need to do. Can you please delete this organization and the user named “bolbolzaban” so that I can start over. I send another message with the “bolbolzaban” user name to confirm my identity.

Many Thanks!

deleted both accounts @khashei

@Simran you’ll need a user to create an org – do you still want me to delete that particular user account? Or will you create the org from that account?

@Simran got it – done

Hi @julien-c,
Thanks for the quick turnaround. Seems like my organization named “Bolbol-zaban” which was under user @khashei is also deleted and now I lost access to my model which was in this repo: “Bolbol-zaban/gpt2-persian”

I uploaded the repo again in its final and correct name which is “bolbolzaban/gpt2-persian”.

Can you please delete the “Bolbol-zaban/gpt2-persian” (with dash in org name)? This repo is now orphan and I don’t have access from the UI to delete that anymore.

Many Thanks

done @khashei

1 Like

Hi @julien-c,

I’ve got a little request. We have got duplicated herbert-base-cased model it’s on the old s3 storage and the new one: herbert-base-cased. This old version is loaded by default. Tokenizer there doesn’t work as expected. Unfortunately, I’m not able to delete this version (tried transformers-cli in several versions). I wonder, if you could remove the s3 version from the storage ?

Which version of transformers are you using? If I’m not mistaken transformers either load from the hub, or from S3, so the old version shouldn’t be an issue

(I can still delete the old version, but just wanted to understand the root cause)

Of course, I use transformers version 4.3.2, in this line one first check the s3 storage for tokenizer files and only if it’s missing it goes to the HUB

Arf you are right. This code should be removed soon :slight_smile: (cc @lysandre @sgugger :scream:)

I’ve deleted the allegro/herbert-base-cased files on S3

It has already been removed and the change will be reflected in the upcoming v4.4.0 release.

Thanks, @sgugger, @julien-c with the newest v4.4.0 release it works like a charm :slight_smile:

@julien-c Herbert model is used in Poland by a large NLP community is there any possibility to ensure backward compatibility and reupload allegro/herbert-base-cased to s3 storage ? I’m not able to check what is wrong with those files

Ok… do you want me to un-delete the files that were on S3? or to upload the files that are currently in the model on the model hub?

ok i’ve readded the files to S3

@julien-c I don’t even know how to thank you for your help. Now Herbert also works with older versions!

1 Like

Thank YOU for contributing to the community!!

@julien-c Could I ask you to rename the CLASSLA organization so that its name becomes lowercased (classla)? The uppercase simply hurts my eyes (and eyes of others). I tried adapting, but it did not work. Thanks!