News for hackers. Stuff that matters.

A comment on Learning to Reason with LLMs. Parent comment

Re:Learning to Reason with LLMs by brookst on Thursday September 12, 2024 @02:09PM (#41523760)
Token count was always an approximation of value. This may help break that silly idea.
[ Parent | on HN ]
Re:Learning to Reason with LLMs by regularfry on Thursday September 12, 2024 @02:41PM (#41524117)
I don't think it's much good as an approximation of value, but it seems ok as an approximation of cost.
[ Parent | on HN ]
Re:Learning to Reason with LLMs by brookst on Sunday September 15, 2024 @12:40AM (#41545256)
Fair, cost and value are only loosely related. Trying to price based on cost always turns into a mess.
[ Parent | on HN ]
Re:Learning to Reason with LLMs by regularfry on Monday September 16, 2024 @03:30AM (#41553634)
Its what you do when you're a commodity.
[ Parent | on HN ]