The official PyTorch blog published a technical article titled "Modernizing Table Batched Embeddings with FBTriton," introducing the redesign of the TBE operator with the FBTriton kernel written in Triton.