5 ms·
No Train No Gain:Revisiting Efficient Training Algrthm for Transformer-BasedLM
- froster 3y agoRecent paper highlights the difficulty of creating a new optimizer as drop-in replacement. Sophia and Lion were recently proposed as superior alternatives to Adam, but appeared worse in an independent eval