Exploring Llama-3-8B MLP Neurons — LessWrong