Paddle

Author	SHA1	Message	Date
tensor-tang	c961fbf09a	change the condition to reset the forward in MKLDNNLayer	8 years ago
tensor-tang	a8eeef86ac	make MKLDNNLayer input grad as a vector	8 years ago
tensor-tang	bc0d255796	make MKLDNNLayer input value as a vector	8 years ago
tensor-tang	c397599dfd	remove weight and bias in MKLDNN reset function, since not all layers have weight and bias. and remove some comments.	8 years ago
tensor-tang	a9490a1053	make output channels changeable in reshape function	8 years ago
tensor-tang	2826ccbbd3	remove unused code and fix typo	8 years ago
tensor-tang	c1914543b0	refine mkldnn logic, move reset buffers into MKLDNNLayer	8 years ago
tensor-tang	1199aa6876	fix bug: clear grad and always share data when output has cpu and add activation in unit tests	8 years ago
tensor-tang	698071cc33	share mkldnn output value data if next layer is cpu device	8 years ago
tensor-tang	6715beaace	enable merge output grad of mkldnn	8 years ago
tensor-tang	49d4b39f28	fix typo and remove some unused code	8 years ago
tensor-tang	37268dbd33	update using exclusive padding for avg-pool	8 years ago
tensor-tang	9d108a21a7	add unit test for mkldnn_pool and pass them	8 years ago
tensor-tang	2a98cba2d2	enable mkldnn_pool forward and backward	8 years ago
tensor-tang	d3a86c673b	init mkldnn_pool files	8 years ago