Deep Learning doesn't Learn Deeply

The transformer expressiveness results are often a bit of a red herring as there tends to be a huge gap between what can be expressed in transformers, and what can be learned with gradient descent. Mind the Gap, a new paper lead by @SaldytLucas dives deeper into this issue 👇👇 https://t.co/Q3BTFR5v7T

— Subbarao Kambhampati (కంభంపాటి సుబ్బారావు) (@rao2z) May 30, 2025