alt Hacker NewsI agree with you on every point but it is interesting to see real world benchmarks like this. Showing the standard benchmarks that all LLMs use is not only boring but at this point likely gamed or even has issues (according to OpenAI) by every LLM.
I agree with you on every point but it is interesting to see real world benchmarks like this. Showing the standard benchmarks that all LLMs use is not only boring but at this point likely gamed or even has issues (according to OpenAI) by every LLM.
>has issues (according to OpenAI)
What is this referring to?
I've been testing various big and small models for years and, about a year ago, switched my attitude from "biggest model is best!" to "best depends on task".
For example, I had a simple coding task which required making 3 trivial changes in 3 source files. Biggest Model completed task perfectly and took 90 seconds.
Smaller Brother also completed task perfectly, but took 30 seconds and cost 5x less.