Hacker News new | ask | show | jobs
Processing 24T tokens for LLM training with 0 crashes (what made it possible) (daft.ai)
1 points by DISCURSIVE 299 days ago