Unverified50% confidenceOpinionExact time
Camerata points out that the car features hidden in the dog neuron are not due to compositional readout but because cars and dogs share poses; there are hundreds of such pose neurons in Inception V1.
1
Sources
50%
Confidence
Long-term
Relevance
10/22/2026
First Seen
Sources
Was the Neural Network Always "Clean"? Interpreting Goodfire's Long-Form Explainability Essay
youtubeXiaoyu Li10/16/2026
Related Entities
Related Claims
UnverifiedIn 2020, Cammarata and Chris Olah provided evidence in the paper "Zoom In" showing that the network hides car feature signals inside a group of dog neurons, reinforcing the superposition hypothesis.70% similarUnverifiedThe robot dog features 12 biomimetic movements and an LCD face capable of expressing emotions.59% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1085039API
curl https://kongchang.com/api/v1/knowledge/claims/1085039MCP
get_claim(id=1085039)