top | item 37802645

(no title)

mmusson | 2 years ago

From what I have read, John had a genius for what you might call relentless optimization. Even if they just improve the performance or reduce the hardware requirements of executing an already trained LLM that would be a huge win. Current progress in that regard focuses on reducing the fidelity of the LLM to make it easier to execute, which isn’t ideal.

discuss

order

No comments yet.