3x3x3 depthwise convolution with per channel quantization (#15775)
authorJongsoo Park <jongsoo@fb.com>
Sat, 12 Jan 2019 03:33:40 +0000 (19:33 -0800)
committerFacebook Github Bot <facebook-github-bot@users.noreply.github.com>
Sat, 12 Jan 2019 03:42:29 +0000 (19:42 -0800)
commite5266b4ba63182f9f86024965ca5330b80b8ad36
tree47c866fca81bc23bebf6679eb50ae25588811b66
parent264e16bffde7cf3a9491bc07fabe30c6d61d0490
3x3x3 depthwise convolution with per channel quantization (#15775)

Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/15775

Pull Request resolved: https://github.com/pytorch/FBGEMM/pull/55

fbgemm didn't have per-channel quantization for 3x3x3 depth-wise convolution

Reviewed By: jianyuh

Differential Revision: D13587438

fbshipit-source-id: 91c36fae7a0e8386e3bc49808e18918b01681dd1
caffe2/quantization/server/conv_depthwise_dnnlowp_op_test.py
caffe2/quantization/server/conv_dnnlowp_op.cc
caffe2/quantization/server/dnnlowp_test_utils.py
caffe2/quantization/server/fbgemm_pack_op.cc