AdamW and Super-convergence is now the fastest way to train neural nets

Comments

from Hacker News https://ift.tt/2z6nqP4
via

Comments

Popular posts from this blog