Wide and Deep Recommender Inference on GPU

Discussing an accelerated inference deployment of a common recommendation architecture using a custom embedding kernel with TensorFlow APIs.