logoalt Hacker News

aesthesia • today at 8:29 PM • 1 reply • view on HN

There are good theoretical reasons to think superposition exists in neural networks, particularly in the residual stream rather than specifically in MLP neuron activations. It wasn't just a single example that got people thinking about superposition.


Replies

mrguyorama • today at 8:58 PM

Yeah that makes sense too. Again, downsides of having zero fundamental frameworks. I'm not equipped to really take it apart.