Neel Nanda comments on Transcoders enable fine-grained interpretable circuit analysis for language models

[Neel Nanda]

Nope to both of those, though I think both could be interesting directions!