Large language models implicitly learn to straighten neural sentence trajectories to construct a predictive representation of natural language

Add code
Nov 05, 2023

Share this with someone who'll enjoy it:

View paper onarxiv icon

Share this with someone who'll enjoy it: