Matrix Factorization w/ Sparse Embedding

The example demonstrates the basic usage of the SparseEmbedding operator in MXNet, adapted based on @leopd's recommender examples. The operator is available on both CPU and GPU. This is for demonstration purpose only.

• python train.py
• To compare the training speed with (dense) Embedding, run python train.py --use-dense
• To run the example on the GPU, run python train.py --use-gpu