openai / automated-interpretability

940 stars 110 forks source link

Is there a demo that shows this great project? #35

Closed guotong1988 closed 8 months ago

guotong1988 commented 8 months ago

Thank you very much!

hijohnnylin commented 8 months ago

[disclaimer: i am the creator of neuronpedia]

check out neuronpedia.org. it uses automated-interpretability for scoring gpt2-small neuron explanations - and it lets anyone contribute their own explanations too.