Post #1368603

@lefticus it would be interesting (and fun) to see the results of an experiment where you task the LLM (I refrain from calling them AI, yet) to do a port which you (or preferably any one else) hasn't already done or isn't in public domain. Then once it has completed its task, you try to do it later without looking at the LLM's process or results. Yet another interesting version would be that once the LLM has completed its task you *"get to train"* on it, meaning, you get to look at how it has done its job and what is the result, and then create your own version. I think that would be a fair apples to apples comparison. What do you think? 🙂😊