Hi Pierz. No I didn’t find a solution. I just stopped trying to use that model on spaces. It is a big model so maybe that was the issue. I haven’t had any problems with the 13b or 7b model. Still open to suggestions if anyone has thoughts.
Thanks Osanseviero, this is useful. I’m new to using hugging face, I’m running the models on spaces currently and have checked out the code for the space but can’t see where the code for the endpoint sits to update it. Do you have any suggestions on how to do this? I do have a pro account.
Hi Stradegio. I see you have a Space (7b chat) which is working fine. Sorry for the confusion, the 7B and 13B Spaces actually run the model directly on the space, so duplicating the base ones and assigning a GPU + adding your token should be enough.