3 ms·
That amount of compute was used for training. For inference (applying the watermarks), hopefully no more than a few seconds per image. Llama 3 70B took 6.4M GP
by hnuser123456 2y ago
That amount of compute was used for training. For inference (applying the watermarks), hopefully no more than a few seconds per image.
Llama 3 70B took 6.4M GPU hours to train, emitting 1900 tons of CO2 equivalent.
- Jaxan 2y agoThanks! I was not at all aware of the scale of training! To me those are crazy amounts of gpu time and resources.
- pierrefdz 2y agoThe amounts of gpu time in the paper are for all experiments, not just training the last model that is OSS (which is usually reported). People don't just oneshot the final model.