Comment by dmarinus
6 hours ago
my experience with LLMs is that when you get answers to questions you didn't ask that it's repeating something that already exists. In other words, it probably regenerated a tool from which it was trained on.
6 hours ago
my experience with LLMs is that when you get answers to questions you didn't ask that it's repeating something that already exists. In other words, it probably regenerated a tool from which it was trained on.
Sounds like your experience from LLMs dates back to 2022.
How do you know this? How would you falsify this hypothesis?
You could train an LLM on a dataset that intentionally excludes certain data, and see if it is able to extrapolate outside of its dataset and come up with those excluded items independently.
For example, remove all code from your training data, and then see if the model can code regardless.
Thanks for the elaboration. (Though I'm not sure dmarinus would agree with your criterion?)
> For example, remove all code from your training data, and then see if the model can code regardless.
Would you accept this weaker version: make up a new programming language (that's guaranteed not to be in the training set), and see if the model (which has trained on programming in existing languages) can cope?
[dead]