News for hackers. Stuff that matters.
Re:WebGPU GPT Model Demo by FL33TW00D on Friday April 21, 2023 @03:29AM (#35650757)
My 250M parameter model runs in 50ms/token ;)

Releasing April 26th when Chrome 113 hits stable. Open source NPM library you can add to any project.

Preview here: https://twitter.com/fleetwood___/status/1646608499126816799?...

Re:WebGPU GPT Model Demo by lxe on Friday April 21, 2023 @03:34AM (#35650796)
That's pretty impressive. What model are you using?
Re:WebGPU GPT Model Demo by FL33TW00D on Friday April 21, 2023 @05:00AM (#35651321)
FLAN-T5 Base currently, 780M parameter variant coming shortly!
Re:WebGPU GPT Model Demo by VierScar on Friday April 21, 2023 @09:43AM (#35653616)
What kind of model? Question-Answering? I imagine it must be quite specialised at <1b Params when many are 7b, 13b, or more?