ConceptTBE
Table Batched Embedding
A core GPU operator in recommender system training that maps sparse ID features to dense vectors, supporting batched lookups across multiple GPUs and multiple embedding tables. Early implementations relied on hand-written CUDA kernels.
Timeline (last 90 days)
Oct 12
Table Batched Embedding (TBE) is a fundamental operator in recommendation model training, responsible for mapping sparse ID features to dense vectors and supporting batched queries across multiple GPUs and multiple embedding tables.
Unverified50%