arXiv - CSCL: "Effects of Parameter Norm Growth During Transform…" - Qoto Mastodon

arXiv - CSCL @arxiv_cscl@qoto.org

Effects of Parameter Norm Growth During Transformer Training: Inductive Bias from Gradient Descent. (arXiv:2010.09697v5 [cs.LG] UPDATED)

http://arxiv.org/abs/2010.09697 #arXiv #NLProc

Mar 09, 2023, 03:13 · · arxiv-cscl · · ·

Sign in to participate in the conversation