HN
login
Transformer Can Hold Two Thoughts at Once: Evidence of Lineararxiv.org

While Large Language Models (LLMs) rely on highly non-linear components, in this work we demonstrate that they exhibit fundamental linearity: when inputs from distinct text streams are linearly combined, the model…

3 points•sbulaev•5 days ago•0 comments•
#linear#evidence#thoughts#transformer#two thoughts#hold two#hold two thoughts#llms

0 comments

No comments yet.

Related stories