* Add mean_iou op.

* Add unitest for mean iou op.

* Add optional collections of confusion matrix and mean_iou.

* Fix cuda kernel.

* Refine code.
1. Merge computing in GPU to two kernel.
2. Use wrong array and correct array instead of confusion matrix.

* Add python api and fix cuda kernel.

* Fix comments.

* Small fix.

* Small fix.

* expose h0 in dynamic_lstm

* update by comment

* update by comment

* h0 to H0

* less_than
* cumsum
* multiplex
* open_recordio_file

* enable docstring checker and fix comments

* update

* update by comments

* fix build

* Add normalization operator.
1. Refine the raw norm_op and let it more general to support to normalize Tensor along any axis.
2. There is a bug in l2_normalize API, which lacks sqrt after `reduce_sum`.
3. Use norm_op to refine the l2_normalize API.
4. Fix bug in test_normalization_wrapper.py.

* row_conv
* uniform_random
* layer_norm
* create_parameter
* hard_shrink
* ssd_loss

* fix conflict

* code clean

* Add python wrapper for gather op.

* Add unitest for 'rank==1' and fix comments.

* Fix comments.

* Make bilinear_interp_op support attrs from input.

* Fix python api.

* unify UpsamplingBilinear2d interface specification

* unify UpsamplingBilinear2d interface specification

* fix name conventions

* small fix about computation order

* Expose bilinear operator into Python API

* delete unused lines

* delete unused lines

* fix typos

* fix name arg and typos

* Enhance reduce op for multi dims.

* Uncomment some unitest.

* Uncomment unitest.

* Remove unused code.

* Fix infershape and python wrapper.

* Add more examples.

* Fix l2_normalize.

* Fix normalization_wrapper.

* Polish code.
1. Rename unitest function.
2. Rename const variable.

* Add dice loss.

* Fix comments.

* Remove unused code.