Comment by Marchant_hq

6 hours ago

CDLMs sound promising for smoother, more coherent text generation. Excited to see how they tackle the token-level discontinuities.

0 comments