Hacker News new | ask | show | jobs
by chrisMyzel 1189 days ago
Petals is an impressive feat but be aware it is very slow with 1-4 sec / token (depending on the hardware you have). I find it too slow for even experimenting, as a developer I want faster feedback cycles. Super interesting to see the project evolve over time, onboarding could not be easier.
1 comments

What is the rate of tokens per second when you are talking to ChatGPT on GPT-4?