Alex Li

icon indicating a website link http://alexanderli.com icon indicating an email [email protected]

icon location Pittsburgh, PA icon location PhD student in machine learning at Carnegie Mellon University. Prev: undergrad at UC Berkeley.

Results 2 issues of Alex Li

Code for training Transformer++

Thanks for the great work, especially the scaling curves! I'm particularly interested in how you trained the Transformer++ models. Is there any public codebase that you used to train them?...

Pretrained checkpoints for other model sizes

Great work! Is it possible to released pretrained checkpoints of the other model sizes (e.g., S, B, L)? I'm aware that their generation performance is not that great. However, they'd...