Goodfire’s Nick Cammarata: Latest analysis suggests most LLM neurons are legible
According to The Rundown AI citing Nick Cammarata, his 2023 OpenAI project explained about 1.7% of neurons, and he now reports at Goodfire that most neurons in models up to trillions of parameters appear understandable.
SourceAnalysis
Ex-OpenAI researcher Nick Cammarata now at Goodfire reports that large AI models are more interpretable than previously feared.
His 2023 OpenAI project explained roughly 1.7 percent of model neurons while current tools show the vast majority legible in models up to trillions of parameters with beautiful circuits according to his statement on X shared by The Rundown AI.
The Rundown AI
@TheRundownAIUpdating the world’s largest AI newsletter keeping 2,000,000+ daily readers ahead of the curve. Get the latest AI news and how to apply it in 5 minutes.