3 ms·
> Big enough to train llama 65B in ~10 days Y'all could totally eat Meta's lunch and train an open LLM with all the innovations that have come since LLaMA's re
by brucethemoose2 3y ago
> Big enough to train llama 65B in ~10 days
Y'all could totally eat Meta's lunch and train an open LLM with all the innovations that have come since LLaMA's release. Other startups are trying, but they all seem bottlenecked by training time/resources.
This could be where the next Stable Diffusion 1.5 comes from.