569e5b056e
!53 [Auto parallel] Remove the redundant work in star elimination
...
Merge pull request !53 from Xiaoda/modify_star_and_triangle_elimination
5 years ago
fd96ebe3ea
fix typo in formula
5 years ago
849b84abb1
fix typo in formula
5 years ago
c46e267cfd
!21 update lenet , add alexnet in example
...
Merge pull request !21 from wukesong/add_lenet_alexnet
5 years ago
8f76c28017
!56 remove cce dependency
...
Merge pull request !56 from jojobugfree/dev-rm-cce
5 years ago
87040483ee
!58 fix two cast bug in auto parallel
...
Merge pull request !58 from lichen/fix_two_cast_bug_in_auto_parallel
5 years ago
60f7a95b1c
the size of tensor may be bigger than 2GB, should use memcpy instead of memcpy_s
5 years ago
110640e2ad
add parallel ops for neg and batchmatmul
5 years ago
c080ec7874
change star elimination: remove some redundant and checking works
5 years ago
0ae77bb0db
TFReaderOp fix, threads will exit after reading necessary amount of rows
...
changes from yanpanhui 524009: added set_dataset_size and changed get_dataest_size according to ME requirements
CI fixes
5 years ago
52166a85cf
!43 dump graph with type info when analysis failed
...
Merge pull request !43 from fary86/dump-typed-graph-when-analyze-fail
5 years ago
045a1c356c
!51 add op SpaceToBatch and BatchToSpace for ge
...
Merge pull request !51 from zhaozhenlong/op/space-to-batch
5 years ago
58cd173051
!63 dataset: repair parameter check problem in random_resize_crop and random_crop
...
Merge pull request !63 from ms_yan/repair_pytrans_check
5 years ago
1fc5a69d6f
!59 dataset: Repair parameter check problem in TFRecordDataset
...
Merge pull request !59 from ms_yan/repair_shard_tf
5 years ago
0727c2e76a
modify log level in DfGraphManager
5 years ago
816b60491d
Dump graph with type info when static analysis failed
5 years ago
e2df848597
!55 modify long time python ut
...
Merge pull request !55 from changzherui/master
5 years ago
2f031e1540
fixed: PrimitiveToInferImplMap map is global, and key of the map PrimitivePtr also a global variable. If key is initialized later than the map initialized during compilation, will cause the primitive map initialize failed. Variable initialization order is not guaranteed during compilation.
5 years ago
a8450dd761
!47 fix issue about the process don't exit when custom ops run failed
...
Merge pull request !47 from wchunjiang/master
5 years ago
cf40305bf0
add operator SpaceToBatch and BatchToSpace for ge
5 years ago
624ab97de6
repair parameter check problem in random_resize_crop and random_crop
5 years ago
da4c711dfb
!50 fix parallel related valuenode merging error
...
Merge pull request !50 from penn/fix_parallel_related_valuenode_merging_error
5 years ago
feb1c36811
fix parallel related valuenode merging error
5 years ago
7b5640da4e
Repair parameter check problem in TFRecordDataset
5 years ago
352c6faf85
!18 enable use float type learning rate in lars optimizer
...
Merge pull request !18 from gziyan/master
5 years ago
2da38ad401
fix two cast bug in auto parallel
5 years ago
da447b8d4d
!45 use std::vector instead of std::list to promote performance for parallel module
...
Merge pull request !45 from chentingting/permote_auto_prallel_performance
5 years ago
04c10bb93d
!42 add vgg scripts master
...
Merge pull request !42 from wangjun260/update_vgg_scripts
5 years ago
0173c40124
add lenet & alexnet in master branch
5 years ago
e6c15b82c8
remove cce dependency
5 years ago
eaf7146d46
modify longtime python ut
5 years ago
f946aea10d
fix grpah mode loop sink bug in auto parallel
5 years ago
31b165a57e
add vgg scripts
5 years ago
5a00d8cb58
This fixes an issue about mindspore process cannot exit when calling python api op_select_format failed in select kernel steps.
...
Previously function op_select_format and check_supported raise an exception directly on the tbe_process python side, but we don't deal with the exception, and raise an exeception on c++ side to frontend ME, that will cause some conflict when recycle resource on ME and tbe_process python interpreter.
This changes adding try...catch in function op_select_format and check_supported on the python side, and return the Exception string to c++ side, so that we can raise an exception to frontend ME and ME will deal with resouce clearning and exit.
5 years ago
3bb48ffee1
use std::vector instead of std::list to promote performance for parallel module
5 years ago
976226f9ac
!10 Add matmul biasadd fusion pass
...
Merge pull request !10 from YuJianfeng/master
5 years ago
6f03881b04
!39 add op Diag and DiagPart
...
Merge pull request !39 from zhaozhenlong/op/diag
5 years ago
2753aa8768
!15 Correct the comments for `RandomChoiceWithMask` op.
...
Merge pull request !15 from seatea/randomchoicewithmask-doc
5 years ago
7242c0fad5
!41 Revert 'Pull Request !17 : [AutoParallel]Fix bug in the case of two cast'
...
Merge pull request !41 from leonwanghui/revert-merge-17-master
5 years ago
e4b404e8ae
!32 auto-enable-dynamic-mem-pool
...
Merge pull request !32 from JoyLvliang/master
5 years ago
e535a5f50c
!40 fix bug of get a error dst format of transdata
...
Merge pull request !40 from lianliguang/master
5 years ago
84a61bd015
!27 add log for kernel runtime in order to trace performance
...
Merge pull request !27 from shibeiji/master
5 years ago
b3a306489d
auto enbale dynamic mem pool
5 years ago
976af212e9
回退 'Pull Request !17 : [AutoParallel]Fix bug in the case of two cast'
5 years ago
140a15924c
!17 [AutoParallel]Fix bug in the case of two cast
...
Merge pull request !17 from lichen/fix_cast_bug
5 years ago
02a25407c4
!30 use string::find instead of equal to distinguish training graph
...
Merge pull request !30 from chenhaozhe/enable-variable-acc-for-training-graph
5 years ago
9d5890d9b9
fix bug of got a error transdata's dest format
5 years ago
b12e6ff780
add operator diag and diag_part
5 years ago
c1c8fef9ca
!24 Change strategy for structure output
...
Merge pull request !24 from 步学/structure-output
5 years ago
4f5755003a
!29 Add some prompt information for ease of use
...
Merge pull request !29 from jonyguo/add_more_log_info_and_testcase
5 years ago
840280e784
Correct the comments for `RandomChoiceWithMask` op.
5 years ago
062b744b19
!12 Fix dtype bug for loss_scale and weight_decay
...
Merge pull request !12 from seatea/dynamic-loss-scale
5 years ago
3ab402e110
!7 adapt relu6grad
...
Merge pull request !7 from zhaozhenlong/adapte-relu6grad
5 years ago
b4d34973bc
fix_cast_bug
5 years ago
cab5503280
use find instead of equal to distinguish training graph
5 years ago
44cd0c1f90
!13 Check input shape for `NMSWithMask` op
...
Merge pull request !13 from seatea/NMSWithMask-check-shape
5 years ago
0da0bdcf40
Fix bug structure output when there is depend whose first input is constant in outputs
5 years ago
4cbcd8e907
enable use float type learning rate in lars optimizer
5 years ago
9862dea3cf
adapt relu6grad and graphengine modified
5 years ago
468e257a14
add log for kernel runtime in order to trace performance
5 years ago
34e42bd6f9
1. add more log info for dataset & mindrecord, 2. add two new testcase for MindDataset
5 years ago
6c03542eec
Fix dtype bug for loss_scale and weight_decay.
...
1.Change dtype of scale to dtype of grad in loss_scale.py;
2.Change dtype of weight_decay to dtype of weight in optimizer.py.
5 years ago
7b7a6a45a0
Check if the shape of the input of NMSWithMask is (N, 5).
5 years ago
86f5c69995
change parallel complie num: 32->16
5 years ago
468dbc3557
Add matmul biasadd fusion pass
5 years ago
930a1fb0a8
initial version
...
Signed-off-by: leonwanghui <leon.wanghui@huawei.com>
5 years ago
mindspore-ci-bot
万万没想到
WeibiaoYu
yangzhenzhang
Xiaoda Zhang
Peilin Wang
chenhaozhe
fary86
mxm
zhaozhenlong
ms_yan
panyifeng
ms_yan
lichenever
wukesong
jojobugfree
chang zherui
wangjun260
wenchunjiang
c00425699
lvliang
leonwanghui
lianliguang
seatea
buxue
Ziyan
shibeiji
jonyguo
jjfeing
GinFung
zhunaipan