add the softsign activation.

fix bug (conv2d  groups: 'num_channels % groups is not 0:' )

enable mkldnn benchmark with googlenet

Fix pool max with index.(Mask type should be int, not float)

Add cudnn pool3d

Add sequence slice operator

remove redundant tests in layer helper's unittest.

Add the configuration helper for multi-head attention.

update the dot_product_attention.

Add the L2 distance layer.

* enforce shape of backward target to be {1}

* fix test_regularizer.py

* rm unused code

* fix backward_test

* fix a type bug

* fix test_program

fix compiler error in "WITH_MKL"

change message level from warning to status, and fix hard number in paddle version

add conv3d_trans_cudnn_op

[Speed compiling]: Refine cmake about CUDA to automatically detect GPU arch by default.

Fix data order of H0 in GRU Operator

add dot_prod_layer