Mechanistic interpretability, are we any closer than we were 5 years ago?
Mechanistic interpretability, are we any closer than we were 5 years ago?