lirongzhen1
|
48c734b669
|
auto parallel for sparse gradient
|
5 years ago |
Wei Luning
|
ef13a4b6fb
|
adjust cse code when op has side effect.
|
5 years ago |
Wei Luning
|
8f56528f8c
|
add flags on function
|
5 years ago |
mindspore-ci-bot
|
5b0472683c
|
!1737 sparse feature backpropagation
Merge pull request !1737 from lirongzhen1/sparse
|
5 years ago |
lirongzhen1
|
516b56cb64
|
sparse feature bp
|
5 years ago |
Kang
|
041f628cce
|
Optimize MixedPrecisionCast function
|
5 years ago |
rick_sanchez
|
e2a322b6b7
|
Modify code to support dynamic graph.
|
5 years ago |
mindspore-ci-bot
|
1234439661
|
!1174 Make optimizer parameter same as gradient
Merge pull request !1174 from ghzl/fix-beg-group-parameters
|
6 years ago |
guohongzilong
|
e0c0c8bc86
|
make optimizer parameter same as gradient
|
6 years ago |
Wei Luning
|
3e89f7baaa
|
fix sens shape error in loss_scale one_step_wrap
|
6 years ago |
mindspore-ci-bot
|
57085bb18d
|
!996 [Auto parallel] Supporting VirtualDataset having three inputs
Merge pull request !996 from Xiaoda/supporting-virtualdataset-has-three-inputs
|
6 years ago |
zhangz0911gm
|
b27e0740ba
|
Getting a simple issues fixed
|
6 years ago |
Xiaoda Zhang
|
5e041966f1
|
add a new vritualdataset cell for three inputs
|
6 years ago |
guohongzilong
|
824bc30a94
|
learning rate and weight decay support group mode
|
6 years ago |
Wei Luning
|
8f1984a8d1
|
only cast when level is O2
|
6 years ago |
zjun
|
c538b83712
|
remove enbale hccl
|
6 years ago |
mindspore-ci-bot
|
507b63ea20
|
!461 Add interface to get attributes of network.
Merge pull request !461 from wsc/master
|
6 years ago |
wangnan39@huawei.com
|
f38d18c665
|
fix bug in checkpoint when save scaler
|
6 years ago |
wangnan39@huawei.com
|
b812b18c02
|
support update parameter for vm
|
6 years ago |
VectorSL
|
4740c70fc3
|
gpu add testcases
|
6 years ago |
wsc
|
f51d5fc411
|
Add interface to get the attributes of network
|
6 years ago |
mindspore-ci-bot
|
e09364df48
|
!298 bugfix(side effect): fix adding wrong control depend between AllReduce and GetStatus
Merge pull request !298 from gongchen/fix_allreduce_control_depend
|
6 years ago |
Wei Luning
|
e1c8f248e0
|
Fix the output is not tuple, when eval
|
6 years ago |
gong chen
|
5d4144de11
|
bugfix(side effect): fix adding wrong control depend between AllReduce and GetStatus.
|
6 years ago |
jinyaohui
|
0790ef8a85
|
modify comment
|
6 years ago |
mindspore-ci-bot
|
40d4a4baa3
|
!20 modify _set_dataset_mode_config api param
Merge pull request !20 from jinyaohui/master
|
6 years ago |
jinyaohui
|
ac62faa388
|
modify set_dataset_mode_config api param
|
6 years ago |
VectorSL
|
d0c24fb706
|
update lossscale for gpu
|
6 years ago |
chengang
|
cc9a0e1310
|
回退 'Pull Request !133 : Edit loss_scale to fit GPU'
|
6 years ago |
mindspore-ci-bot
|
315036b1a5
|
!133 Edit loss_scale to fit GPU
Merge pull request !133 from VectorSL/master
|
6 years ago |
VectorSL
|
2ff6f0de46
|
edit loss_scale for gpu
|
6 years ago |
zhongligeng
|
144a636b51
|
resolve some issue in nn comments
|
6 years ago |
seatea
|
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.
|
6 years ago |
zhunaipan
|
930a1fb0a8
|
initial version
Signed-off-by: leonwanghui <leon.wanghui@huawei.com>
|
6 years ago |