Uncovering the Computational Ingredients of Human-Like Representations in LLMs

Open in new window