Hacker News new | ask | show | jobs
by flaviojuvenal 2164 days ago
"Reading the OpenAI GPT-3 paper. Impressive performance on many few-shot language tasks. The cost to train this 175 billion parameter language model appears to be staggering: Nearly $12 million dollars in compute based on public cloud GPU/TPU cost models (200x the price of GPT-2)"

https://twitter.com/eturner303/status/1266264358771757057

1 comments

Nearly $12 million dollars in compute based on public cloud GPU/TPU cost models (200x the price of GPT-2)"

Oh! So only about $1 million on bare metal.