- 09 10月, 2019 1 次提交
-
-
由 liuwei1031 提交于
-
- 03 9月, 2019 1 次提交
-
-
由 gongweibao 提交于
Change backward_guard to optimize_guard to maximize the allreduce overlap
-
- 23 7月, 2019 1 次提交
-
-
由 wangchaochaohu 提交于
* rewrite the conv_op using cudnn_conv_helper * add workspace limit for v7 test=develop * fix test=develop * add half float test=develop * fix test=develop * fix test=develop * revise code style test=develop * fix test=develop
-
- 10 5月, 2019 1 次提交
-
-
由 qingqing01 提交于
* Add conv2d_grad_grad_op * Extracte the cuDNN conv algo searching code in conv_cudnn_helper.h. - Now use it in conv2d_grad_grad. - Will simply the searching code in conv2d and conv2d_grad in next PR. * Enhance and fix bug in unit testing of gradient_checker. * Support to fetch empty variables，return None in Python.
-