Yinghai Lu 346de2535d Workaround lack of 0-dim support in ideep (#8959) ... Summary: Closes https://github.com/pytorch/pytorch/pull/8959 MKL-DNN doesn't have support to 0-dim tensor. As a workaround, we produce CPUTensor instead of Ideep tensor in the fallback ops. And for those tensors, we don't need Ideep copy op anymore. Reviewed By: viswanathgs Differential Revision: D8665168 fbshipit-source-id: 59678de2c5aed8c691ab5caaadede6d6c000dd7b		2018-06-27 20:24:28 -07:00
..
mkl_concat_op_test.py
mkl_conv_op_test.py
mkl_copy_op_test.py
mkl_elementwise_add_op_test.py
mkl_elementwise_sum_op_test.py
mkl_fc_op_test.py
mkl_fc_speed_test.py
mkl_fill_op_test.py
mkl_LRN_op_test.py
mkl_LRN_speed_test.py
mkl_pool_op_test.py
mkl_pool_speed_test.py
mkl_relu_op_test.py
mkl_sbn_op_test.py
mkl_sbn_speed_test.py
mkl_sigmoid_op_test.py
mkl_speed_test.py
mkl_squeeze_op_test.py
rewrite_graph.py	Workaround lack of 0-dim support in ideep (#8959)	2018-06-27 20:24:28 -07:00
rewrite_graph_test.py	Update MKL exporter to IDEEP ops (#8228)	2018-06-06 21:43:43 -07:00