CME295 Transformers and Large Language Models is open for enrollment until September 6. Learn more:
Every LLM you’ve used generates text one word at a time. But what if it didn’t?
Stanford Adjunct Professors Shervine and Afshine Amidi break down diffusion LLMs — models that generate text more like an image diffusion model, refining the whole output progressively instead of token-by-token.
#DiffusionLLM #LLM #Transformers #StanfordOnline #AI #MachineLearning #OnlineCourses
source
And how would it reflect on agents performance ?
arent we already using this for image?
Doesn't the term "auto-regressive" refer only to the training objective, not the inference method?
Incredible! Can’t wait 🙌🏻
Ami theonly one who thinks both people are same ?
Can diffusion LLMs be used as agents ?
What would be the main difference?
Nice