Meta FAIR Releases e-means: An Algorithm Beating k-means in 10x Fewer Epochs

Meta FAIR has introduced e-means, a mini-batch k-means clustering algorithm that achieves similar or better loss than Lloyd's k-means in about one-tenth the epochs, significantly speeding up clustering tasks. It is available in FAISS 1.16.0 and supports CPU and GPU with PyTorch or NumPy, showing 10-45x speedups over Flash-KMeans on GPUs.