Hackernews
new
show
ask
jobs
NCMNext-Concept Modeling: Fast, Parallel Coarse-to-Fine Text Generation
2 points
posted 8 hours ago
by matt_d
(scalingintelligence.stanford.edu)
No comments yet