合并请求 · PaddlePaddle / Paddle

Support grouped convolution layer with mkldnn.

!13378 · opened 9月 13, 2018 by saxon_zh Intel
- 已合并
- 5
更新于9月 14, 2018
created conv relu pass

!13258 · opened 9月 05, 2018 by saxon_zh Intel
- 已合并
- 38
更新于9月 13, 2018
[MKLDNN] Fuse Conv+BatchNorm + ReLU

!13209 · opened 9月 04, 2018 by saxon_zh Intel
- 已合并
- 3
更新于9月 11, 2018
[MKLDNN] Reusing once reordered convolution weights in test mode

!13327 · opened 9月 10, 2018 by saxon_zh Intel
- 已合并
- 2
更新于9月 11, 2018
Enable conv and batch norm by default

!13108 · opened 8月 31, 2018 by saxon_zh Intel
- 已合并
- 2
更新于9月 10, 2018
Enable lstm peephole

!13160 · opened 9月 03, 2018 by saxon_zh Intel
- 已合并
- 24
更新于9月 05, 2018
Enforce requested size of tensor to be sufficiently large

!12982 · opened 8月 27, 2018 by saxon_zh Intel enhancement
- 已合并
- 1
更新于8月 28, 2018
Reusing primitives for forward Batch Norm operator

!12515 · opened 8月 02, 2018 by saxon_zh Intel
- 已合并
- 8
更新于8月 28, 2018
Update MKLDNN to 0.15, fix convolution integration

!12675 · opened 8月 13, 2018 by saxon_zh Intel
- 已合并
- 12
更新于8月 27, 2018
Fuse Convolution and Eltwise Add into MKLDNN's Conv+Bias

!12669 · opened 8月 13, 2018 by saxon_zh Intel
- 已合并
- 16
更新于8月 21, 2018
Optimize CRF Decoding with AVX/AVX2/AVX512F instruction

!12767 · opened 8月 17, 2018 by saxon_zh Intel
- 已合并
- 10
更新于8月 20, 2018
fix UT for mkldnn 0.15

!12618 · opened 8月 09, 2018 by saxon_zh Intel enhancement
- 已合并
- 9
更新于8月 10, 2018
MKLDNN sum fix: remove in_place condition in loop with memory primitives

!12180 · opened 7月 17, 2018 by saxon_zh Intel
- 已合并
- 0
更新于7月 17, 2018
MKLDNN: Extending Conv grad MKLDNN op with reusing MKLDNN primitives

!12099 · opened 7月 11, 2018 by saxon_zh Intel
- 已合并
- 6
更新于7月 17, 2018
reserve vector space before loop in top-k

!12100 · opened 7月 11, 2018 by saxon_zh Intel
- 已合并
- 5
更新于7月 13, 2018
MKLDNN: Extending Conv MKLDNN op to reuse MKLDNN primitives

!11750 · opened 6月 26, 2018 by saxon_zh Intel
- 已合并
- 3
更新于7月 11, 2018
added cycling the dataset for cifar and flowers

!12077 · opened 7月 10, 2018 by saxon_zh Intel
- 已合并
- 0
更新于7月 11, 2018
workaround for no MKLDNN kernel

!11699 · opened 6月 25, 2018 by saxon_zh Intel
- 已合并
- 8
更新于7月 09, 2018
The mkldnn batch norm supports other data format

!11666 · opened 6月 22, 2018 by saxon_zh Intel
- 已合并
- 6
更新于6月 29, 2018
bnorm+relu fuse for mkldnn (inference)

!11434 · opened 6月 13, 2018 by saxon_zh Intel
- 已合并
- 23
更新于6月 27, 2018

PaddlePaddle / Paddle 大约 1 年 前同步成功

PaddlePaddle / Paddle
大约 1 年前同步成功