mindspore-ci-bot
30005c9c64
!12301 [GraphKernel] Eliminate redundant split nodes.
From: @tronzhang
Reviewed-by: @gaoxiong1,@anyrenwei
Signed-off-by: @anyrenwei
5 years ago
mindspore-ci-bot
5394ff8ac3
!12316 [GraphKernel] Exclude special node in expander and basic fusion.
From: @tronzhang
Reviewed-by: @gaoxiong1,@anyrenwei
Signed-off-by: @anyrenwei
5 years ago
tronzhang
be2b9978be
exclude special node when expand or basic fusion
5 years ago
yuchaojie
ad93497838
fix SliceGradUnifyMindIR
5 years ago
tronzhang
e953705521
eliminate redundant split ops
5 years ago
luopengting
b4d56f6358
add functions to check for directory and filename
1. add validation for directory and filename
2. refactor the file operations about path
3. add in CMakeLists
4. add logger in nn_ops
5 years ago
mindspore-ci-bot
a063d7633d
!12241 [auto-monad] Support side-effects by auto-monad
From: @hwhewei
Reviewed-by: @zhunaipan,@zh_qh
Signed-off-by: @zh_qh
5 years ago
mindspore-ci-bot
61ab50915f
!11959 Add BCEWithLogitsLoss op for Ascend.
From: @liu_xiao_93
Reviewed-by:
Signed-off-by:
5 years ago
He Wei
7d9a783993
[auto-monad] Support side-effects by auto-monad
The basic idea is: exploits data dependency to control the execution order
of side-effect operations, and keep the semantics of ANF unchanged.
The ControlDepend primitive is removed and there are two primitives added:
1. UpdateState:
```
a = Assign(para, value)
```
became:
```
a = Assign(para, value, u)
u = UpdateState(u, a)
```
2. Load:
```
x = Add(para, value)
```
became:
```
p = Load(para, u)
x = Add(p, value)
u = UpdateState(u, p)
```
5 years ago
mindspore-ci-bot
228a64de0f
!11390 convert part of prim attr from str to int
From: @wangnan39
Reviewed-by: @kingxian
Signed-off-by:
5 years ago
mindspore-ci-bot
7f1a9508f8
!12045 [MS][RDR] change saving filename in RDR module
From: @louie5
Reviewed-by:
Signed-off-by:
5 years ago
wangnan39@huawei.com
2f79905f60
convert attr from str to enum
5 years ago
louei5
72e567d06b
change saving filename in RDR module
5 years ago
dayschan
e0e6c39eae
Refactor GraphKernelExpander (1st submission)
Decoupled from the front-end interfaces.
1. Removed the call to "Renormalize".
Completed the infer-format in model_builder.
Only used the device shape and device format to
infer an abstract shape without considering padding.
2. Removed the call to python's Primitive interfaces.
The "Renormalize" relies on the PrimitivePy, so they can be
removed together. After that, the functions "ConstAttrToInput",
"DeleteAttrInInput" and related can be removed.
3. Reuse the AkgKernelJsonGenerator in GraphKernelExpander.
1) set the attribute "extract_opinfo_from_anf" to true, so that
the generator can handle the basic operator with anf info.
2) added a function "extract_expand_info" in expander.py
to convert the json into a more friendly format. The attrs
was converted to a dict instead of a list.
4. Scalars only support DefaultFormat.
Removed the argument "format" from graph_builder.value
5. Moved the expander op list from graph_kernel_helper.cc to graph_kernel_expander.cc
5 years ago
mindspore-ci-bot
5bbf009829
!12108 [GraphKernel] Set attribute to node safely.
From: @tronzhang
Reviewed-by: @ckey_dou
Signed-off-by:
5 years ago
liu_xiao_93
fabc25538e
Add BCEWithLogitsLoss
5 years ago
jinyaohui
30a27b2adb
modify Gelu、FastGelu to GeLU and FastGeLU
5 years ago
mindspore-ci-bot
aebe263dce
!11895 unify mindir for different backend: the output num of optimizer ops, the backward of concat
From: @wangnan39
Reviewed-by:
Signed-off-by:
5 years ago
mindspore-ci-bot
c9aaa70b39
!12092 Change L2Norm, merge from r1.1 to master
From: @liangzhibo
Reviewed-by: @ginfung
Signed-off-by:
5 years ago
mindspore-ci-bot
ad5b033cc5
Change L2Norm, r1.1 to master
5 years ago
tronzhang
c1e63d4824
set attr safetly
5 years ago
mindspore-ci-bot
2847a1f3e3
!12052 somas solver optimization
From: @laiyongqiang
Reviewed-by: @zhoufeng54,@chujinjin
Signed-off-by: @chujinjin
5 years ago
mindspore-ci-bot
74652eb942
!12044 modify pack to stack
From: @jinyaohui
Reviewed-by:
Signed-off-by:
5 years ago
mindspore-ci-bot
c6e8204915
!12084 fix Conv2DBackpropInputUnifyMindIR when it is a forward op in pynative
From: @yuchaojie
Reviewed-by: @kisnwang,@zhoufeng54
Signed-off-by: @zhoufeng54
5 years ago
mindspore-ci-bot
39d0f2e95a
!12125 Dump complete trace info for hwop_d_end_xxx.ir
From: @irmo
Reviewed-by: @zhoufeng54,@chujinjin
Signed-off-by: @chujinjin
5 years ago
jinyaohui
8022f9a6ed
modify pack to stack
5 years ago
wangnan39@huawei.com
cd9173fdfd
unify the output num of optimizer ops
5 years ago
huanghui
e10b9cc60b
dump complete trace info for hwop_d_end_xxx.ir
5 years ago
mindspore-ci-bot
8e4ec7c2ca
!12047 move somas model info into somas_meta directory
From: @laiyongqiang
Reviewed-by: @stsuteng,@zhoufeng54
Signed-off-by: @stsuteng
5 years ago
yuchaojie
90a83db5f1
fix Conv2DBackpropInputUnifyMindIR when it is a forward op in pynative
5 years ago
jinyaohui
d9be0c102d
add some ops
5 years ago
mindspore-ci-bot
f9d9bba927
!12006 [GraphKernel][Gpu]enable GraphKernel for layernorm and layernormGrad (sync from r1.1)
From: @hanhuifeng2020
Reviewed-by: @gaoxiong1,@anyrenwei
Signed-off-by: @anyrenwei
5 years ago
mindspore-ci-bot
f8fc047f2b
!11997 update io_format to format
From: @liubuyu
Reviewed-by: @kisnwang,@zhoufeng54
Signed-off-by: @zhoufeng54
5 years ago
laiyongqiang
75fb5487b5
somas solver optimization
5 years ago
mindspore-ci-bot
097431af6d
!11961 fix solver log bug
From: @laiyongqiang
Reviewed-by: @zhoufeng54,@kisnwang
Signed-off-by: @kisnwang
5 years ago
laiyongqiang
b625073d4f
move somas model info into somas_meta directory
5 years ago
mindspore-ci-bot
0ff27ef3b4
!11930 【GraphKernel】Replace Assign with InplaceAssign
From: @dayschan
Reviewed-by: @gaoxiong1,@dylangeng
Signed-off-by: @gaoxiong1
5 years ago
laiyongqiang
15240c3614
use the correct filename for solver output log
5 years ago
mindspore-ci-bot
b04e2aabff
!11996 fix bug of split optimizer.
From: @liu_xiao_93
Reviewed-by: @liangchenghui,@wuxuejian
Signed-off-by: @liangchenghui
5 years ago
hanhuifeng2020
c5f261d894
enable GraphKernel for layernorm and layernormGrad (sync from r1.1)
5 years ago
mindspore-ci-bot
a1f47b5ebf
!11960 somas cache with model hash id
From: @laiyongqiang
Reviewed-by:
Signed-off-by:
5 years ago
mindspore-ci-bot
24ef3efcb8
!11545 unify SpaceToBatchND's attr
From: @yuchaojie
Reviewed-by:
Signed-off-by:
5 years ago
liubuyu
91f5b1b68e
update io_format to format
5 years ago
liu_xiao_93
2efd12dec6
fix bug of split optimizer.
5 years ago
laiyongqiang
3a2b9564c3
somas cache with model hash id
5 years ago
mindspore-ci-bot
a24ff36d9c
!11777 stitch fusion
From: @r1chardf1d0
Reviewed-by:
Signed-off-by:
5 years ago
mindspore-ci-bot
9efbef72fc
!11622 【GraphKernel】Moved ShapeOpsSplitter before GraphKernelSplitter
From: @dayschan
Reviewed-by: @gaoxiong1,@dylangeng
Signed-off-by: @gaoxiong1
5 years ago
yuchaojie
7a7f8b5e13
unify SpaceToBatchND's attr
5 years ago
mindspore-ci-bot
e897eb4c41
!11915 Change TensorAdd to Add, merge from r1.1 to master
From: @liangzhibo
Reviewed-by: @ginfung,@zh_qh
Signed-off-by: @zh_qh
5 years ago
l00591931
9ec100d069
Change TensorAdd to Add, from r1.1 to master
5 years ago