Comment by ACCount37
21 hours ago
Model inference compute outweights training compute by 10:1 and more for frontier LLMs. "LLM depreciation" is an expense, but not a dealbreaker.
21 hours ago
Model inference compute outweights training compute by 10:1 and more for frontier LLMs. "LLM depreciation" is an expense, but not a dealbreaker.
No comments yet
Contribute on Hacker News ↗