Report
How to scale training to 1M-token context
A team member points to low-level algorithmic and systems details; Lila Sciences describes the linked results as a collaboration.
TLDR
A team member shares a post they say covers the algorithmic and systems details of scaling training to 1M-token context. Lila Sciences links to the same post and calls its results a collaboration.
Combined views
6.2K
3 Sources, first seen ago
