Hacker News new | past | comments | ask | show | jobs | submit | from login

It costs you $9M to $17M to train PaLM (Google's 530B param. language model) (twitter.com/ohlennart) 22 points by ohlennart on April 8, 2022 | past | 2 comments ML training compute has grown by a factor of 10B over the last 12 years (twitter.com/ohlennart) 4 points by innerfestival on March 14, 2022 | past ML training compute has been doubling every 6 months since 2010 (twitter.com/ohlennart) 129 points by innerfestival on Feb 15, 2022 | past | 57 comments

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact Search: