On the Pitfalls of Analyzing Individual Neurons in Language Models

Open in new window