Comments: On Distributional Assumptions and Whitened Cosine Similarities