and it kind of saddens me that genuinely excellent thinkers like Neel Nanda are putting their writeups on #interpretability on LW, which ranges from noobish to terrible.

#machinelearning #ai #mechanisticinterpretability