Evaluating Language Models
The Data Exchange with Ben Lorica
00:00
How to Score Language Models in a Benchmark
The tests tend to be very English centric. So all of the scenarios are English only for now. How long does it take to score one model comprehensively? I guess depends on the model size. There's models that we're also missing. We don't have deep minds models. Like colactica from meta that just came out, which we're going to evaluate but haven't had a chance to do that yet.
Transcript
Play full episode
