Topological Interpretations of GPT-3