Closed tonychenxyz closed 1 month ago
Hi! We haven't planned to release weights for the model as they are pretty undertrained with modern standards. The paper is a proof of concept and not a generally useful model for downstream tasks. That said, we will try to release the weights if you find that useful with the training code release by the end of this month.
I uploaded the checkpoints to Hugginface, and added notes to the training code on how to load it. Unfortunately it is not for the old, experimental code, not this simple, cleaned up one.
Hi! I was wondering if the model weight will be released any time soon?