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

[Neel Nanda]

That’s awesome, and insanely fast! Thanks so much, I really appreciate it