4 ms·

So do most LLMs now days, no?

• viraptor 2y ago
  Kind of. Those that are explicitly trained to do that with consistent formats will do it better. They'll also save you the extra tokens needed to explain the format/method of interacting with functions. But yeah, you can simulate this with any recent model and enough explanation.