News for hackers. Stuff that matters.

A comment on LLaMA-7B in Pure C++ with full Apple Silicon support. Parent comment

Re:LLaMA-7B in Pure C++ with full Apple Silicon support by schappim on Saturday March 11, 2023 @03:38PM (#35112447)
I've had difficulty obtaining useful results from the smaller (7B-sized) models. The issue lies in the content, not the speed. If you could stream the text-to-speech, the speed alone would be satisfactory.
[ Parent | on HN ]
Re:LLaMA-7B in Pure C++ with full Apple Silicon support by amusedcyclist on Saturday March 11, 2023 @03:49PM (#35112550)
You're right I overestimated how fast we talk!
[ Parent | on HN ]
Re:LLaMA-7B in Pure C++ with full Apple Silicon support by fwlr on Saturday March 11, 2023 @09:36PM (#35115820)
Some rules of thumb I use for estimating this kind of stuff

100wpm: Max typing speed

200wpm: Max speaking speed

300wpm: Max listening speed, max reading speed with subvocalisation

900wpm: Max reading speed without subvocalisation

[ Parent | on HN ]
Re:LLaMA-7B in Pure C++ with full Apple Silicon support by mromanuk on Sunday March 12, 2023 @08:57AM (#35119535)
Doin napkin math, this model should be hitting 900wpm
[ Parent | on HN ]