Earlier quoted context omitted.
This deepmind paper showed big limitations on nested structures with transformers, LSTMs actually did better: Neural Networks and the Chomsky Hierarchy https://arxiv.org/abs/2207.02098
Interesting. Can someone quickly summarize their results? Are they saying that using languages with simpler syntax Lisp lead to LLM being able to generalize their knowledge better?
It was a good bit older of a paper though, if I remember it's somewhat expected from the pure feed forward nature of them and limited circuit depth, where LSTMs have some recurrence.