arXiv - CSCL: "Transformers Learn Higher-Order Optimization Meth…" - Qoto Mastodon

arXiv - CSCL @arxiv_cscl@qoto.org

Transformers Learn Higher-Order Optimization Methods for In-Context Learning: A Study with Linear Models. (arXiv:2310.17086v1 [cs.LG])

http://arxiv.org/abs/2310.17086 #arXiv #NLProc

Oct 29, 2023, 03:17 · · arxiv-cscl · · ·

Sign in to participate in the conversation