Can you retrain it yourself to accomodate an ever-changing world?
You can fine-tune it, but it turns out that’s mostly a waste of time. To accommodate an ever-changing world you can tell it what changed. You can get a model to write an entirely new programming language that’s not part of its training by dumping in a few KBs of examples of the new language and telling it to start writing more code like that.
You can fine-tune it, but it turns out that’s mostly a waste of time. To accommodate an ever-changing world you can tell it what changed. You can get a model to write an entirely new programming language that’s not part of its training by dumping in a few KBs of examples of the new language and telling it to start writing more code like that.
score 1 · canonical