Open-Assistant icon indicating copy to clipboard operation
Open-Assistant copied to clipboard

Published 20 hours ago • verified publisher icon LAION-AI

Resolve issue with ShareGPT_vicuna_unfiltered

Open olliestanley opened this issue 2 years ago • 3 comments

In SFT-8 branch we use HF dataset gozfarb/ShareGPT_Vicuna_unfiltered. However gozfarb's account has been nuked from HF including all datasets, so this dataset can no longer be used by anyone who doesn't have it cached

olliestanley avatar May 14 '23 21:05 olliestanley

need to reach out to gozfarb to get a copy of that dataset and republish it

ehartford avatar May 15 '23 01:05 ehartford

why was gozfarb's account deleted?

ehartford avatar May 15 '23 01:05 ehartford

should we use https://huggingface.co/datasets/anon8231489123/ShareGPT_Vicuna_unfiltered instead?

ehartford avatar May 15 '23 02:05 ehartford

why was gozfarb's account deleted?

Not sure

should we use https://huggingface.co/datasets/anon8231489123/ShareGPT_Vicuna_unfiltered instead?

This seems like the best option, although I think it's slightly less cleaned than gozfarb's latest version

olliestanley avatar May 15 '23 20:05 olliestanley

There is now also: Aeala/ShareGPT_Vicuna_unfiltered

andreaskoepf avatar Jun 02 '23 19:06 andreaskoepf