Volodymyr Kuleshov 🇺🇦
@volokuleshov
New blog post: How to Build a Diffusion Language Model.
Diffusion LLMs went from open problem to reality in 2 years (Mercury, Gemma Diffusion, Nemotron Diffusion). With my Cornell group, we wrote up the research advances that make them work.
We use today's OSS large diffusion
Diffusion LLMs went from open problem to reality in 2 years (Mercury, Gemma Diffusion, Nemotron Diffusion). With my Cornell group, we wrote up the research advances that make them work.
We use today's OSS large diffusion
· Edited
17 1.5K