Twitxy Saved

Petr

@pmolodyk

Machine Learning researcher, PhD student at Georgia Tech.
Atlanta, GA Joined December 2018
60 Following 36 Followers 17 Posts

Latest posts 1

Petr @pmolodyk
[1/8] Can we improve parallel drafters with on-policy distillation?
Introducing MDraft: on-policy distillation for diffusion drafters.
At T=1, mean acceptance length vs. DFlash:
• Qwen3-4B: +11.9%
• Qwen3-8B: +13.4%
Same inference, just swap the drafter checkpoint! 🧵