TLDRocket
Sign in

Interpretability

12 summarised stories about Interpretability, each linking back to the original source. Browse all topics →

Wednesday, 6 March 2019

Introducing Activation Atlases

OpenAI Blog 7 years ago

Researchers have developed activation atlases, a visualization technique that shows how interactions between neurons in AI systems represent different concepts. The method maps neural activations across multiple input examples to reveal patterns in how the system processes information. Understanding these internal representations could help identify failures and weaknesses before deploying AI systems in sensitive applications.