...

• Enhanced the performance of `sparse.dot` operator
• Performance regression automation (vikram)

New Features - Sparse Tensor Support

• Added more sparse operators: `contrib.SparseEmbedding` and `sparse.mean`
• Added `asscipy()` for easier conversion to scipy
• Added support for custom sparse operators
• Added `check_format()` for sparse ndarrays to check if the array format is valid

New Features - Gluon

• Gluon - GA (vikram)
• Gluon IO performance (sukwon, Mu)

Other Planned Features

• Added `group2ctxs` argument to `Module` API to support model parallelism on CPU & GPU
• Advanced indexing Support: https://github.com/apache/incubator-mxnet/pull/8246 (Jun)
• slice operator supporting parameter step: https://github.com/apache/incubator-mxnet/pull/8558 (Jun)
• PR 8489 - https://github.com/apache/incubator-mxnet/pull/8489 (Eric)License Headers: https://github.com/apache/incubator-mxnet/issues/7749 (Meghna)
• Nvidia PR - https://github.com/apache/incubator-mxnet/pull/8294
• Gluon - GA (vikram)
• Gluon IO performance (sukwon, Mu)
• Performance regression automation (vikram)
• Caffe to MXNet translator (Indu)
• Gradient compression (sukwon)
• Advanced indexing fp-16 support (sukwon)

Bug Fixes

• Fixed a[-1] indexing doesn't work on NDArray
• Fixed `expand_dims` if axis < 0

Doc Updates

• License Headers: https://github.com/apache/incubator-mxnet/issues/7749 (Meghna)
• Documentation updates (markhama)fp-16 support (sukwon)

...