4 ms·
Look a cost per intelligence or cost per task instead of cost per token.
by energy123 6mo ago
Look a cost per intelligence or cost per task instead of cost per token.
- ulimn 6mo agoIsn't the outcome / solution for a given task non-deterministic? So can we reliably measure that?
- dns_snek 6mo agoRepetition and statistics, if you have $1000++ you didn't need anyway.
- genericresponse 6mo agoStatistically. Do many trials and measure how often it succeeds/fails.
- kridsdale1 6mo agoAka a benchmark.
- foota 6mo agoYes, sort of. Generally you can measure the pass rate on a benchmark given a fixed compute budget. A sufficiently smart model can hit a high pass rate with fewer tokens/compute. Check out the cost efficiency on https://artificialanalysis.ai/ https://artificialanalysis.ai/ (say this posted here the other day, pretty neat charts!)
- throwuxiytayq 6mo agoIt's much easier to measure a language model's intelligence than a human's because you can take as many samples as you want without affecting its knowledge. And we do measure human intelligence.
- torginus 6mo agoThis is the only correct take. The only metric that matters is cost per desired outcome.
- yokoprime 6mo agoHow do I reliably measure 1 unit of intelligence?
- wellthisisgreat 6mo agoIn pelicans, obviously