EleutherAI/math-lm

Optimizer states?

Opened this issue · 1 comments

jenkspt commented

Hi I'm interested in getting access to the optimizer states for the final llemma checkpoints. Are these available? - and if not, can they be made public?

Thanks for your amazing work!

jenkspt commented

Also interested in older checkpoints of 7b model -- specifically interested in estimating the performance of the 34b model with more tokens.