4 ms·
No it isn't. Stable Diffusion is less than 200 grand to train.
by leesec 4y ago
No it isn't. Stable Diffusion is less than 200 grand to train.
- anononaut 4y agoI heard it was $4MM alone in AWS compute time.
- postalrat 4y agoSo about 40k on your own machines.
- inciampati 4y agoIt does make it seem like a box of H100s will easily be able to make an interesting open LLM.
- speedgoose 4y agoThis number seems to match the $200k, if you take into account the cloud margins of our favourite counterfeit products reseller.
- huijzer 4y agoAccording to Christopher Potts (Stanford Professor and Chair, Department of Linguistics, and Professor, by courtesy, Department of Computer Science), training a large language model costs about 50 million [1]. [1]: https://youtu.be/-lnHHWRCDGk?t=637 https://youtu.be/-lnHHWRCDGk?t=637
- nl 4y agoYeah this is way wrong, unless counting the salaries of everyone involved for a few years in the lead up while writing software that ended up being used https://bigscience.huggingface.co/blog/bloom https://bigscience.huggingface.co/blog/bloom was trained with a $7M grant, and that was the first time they'd done it (so there was a lot of failures).