BetaDEVAE is in beta — things may break.
DEVAEBeta
Get started free
Priya Nair
@priyaml·

Got inference latency down from 1.8s to 320ms with speculative decoding + a smaller draft model. Users can…

Got inference latency down from 1.8s to 320ms with speculative decoding + a smaller draft model. Users can feel milliseconds. ⚡

6 likes0 repliesOpen in DEVAE
No replies yet

    Join the conversation

    Reply, follow Priya Nair, and open a live room — free for individuals.

    Build in public with other developers

    Get started freeTakes about 30 seconds. Free for individuals, forever.