Ziyan
2a752f24bf
enable not fully use opt shard
5 years ago
ZPaC
12f95b51f4
Add server code part2
4 years ago
mindspore-ci-bot
a2a24f7833
!15810 fix gather_p_info judgement
From: @yao_yf
Reviewed-by: @stsuteng,@yangzhenzhang
Signed-off-by: @stsuteng
4 years ago
Ziyan
3a11b8b39c
fix accu grads shape when enable opt shard
4 years ago
yao_yf
61ef56a26c
fix gather_p_info judgements
4 years ago
Xiaoda Zhang
5fecfe92a6
code style warnings fixing
4 years ago
yao_yf
093ef784de
dont insert virtualoutput for scalar
4 years ago
mindspore-ci-bot
3cfd58e8e0
!15643 insert virtual div only for first input of dropout do mask
From: @yangzhenzhang
Reviewed-by: @stsuteng,@kisnwang
Signed-off-by: @stsuteng
4 years ago
yangzhenzhang
5828973978
fix bug for dropout do mask
4 years ago
yao_yf
21276408b8
parallel virtual_out_ops
5 years ago
yao_yf
17354e3c4e
fix find nodes with param
4 years ago
yangzhenzhang
bcd2ecc403
check layouts for shared parameter
4 years ago
yao_yf
d7641123bb
strategy_ckpt_file_adapt_optimizer_shard
4 years ago
yangzhenzhang
689e50a3d0
fix grad accu bug for no used parameter
5 years ago
mindspore-ci-bot
29bf2909b2
!13105 insert mirror before load
From: @yangzhenzhang
Reviewed-by:
Signed-off-by:
5 years ago
yangzhenzhang
6eadd241a0
insert mirror before load
5 years ago
Ziyan
4109308e34
insert parallel optimizer once
5 years ago
Ziyan
ec9793861f
fix grad accu
5 years ago
mindspore-ci-bot
7fcce73c51
!12700 add grad accumulation combined with optimizer parallel
From: @yangzhenzhang
Reviewed-by:
Signed-off-by:
5 years ago
chendongsheng
db0a6f1e19
replace ps-lite
5 years ago
yangzhenzhang
a70d616841
mini step grad accumulation
5 years ago
huangbingjian
0bbd95d7a0
modify CheckpointStrategy to adapt load operator
5 years ago
mindspore-ci-bot
2c5d19260e
!12434 fix operator_info is null
From: @Margaret_wangrui
Reviewed-by:
Signed-off-by:
5 years ago
mindspore-ci-bot
da6e6728b1
!12515 fix shape_ptr is nullptr:Umonad should not get shape
From: @Margaret_wangrui
Reviewed-by:
Signed-off-by:
5 years ago
Margaret_wangrui
0aaa31764e
Do not get shape for monad type
5 years ago
Margaret_wangrui
a5fa4918f5
fix operator_info is null
5 years ago
yangzhenzhang
70aa0dc5e2
modify get output layout
5 years ago
huangbingjian
b56fc0c2af
[auto-monad] Do not insert VirtualDiv after UpdateState
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
b189f177bb
Change tuple_getitem to TupleGetItem and some other ops, merge from r1.1 to master
5 years ago
mindspore-ci-bot
9fa0499fa0
Change GatherV2 to Gather r1.1 to master
5 years ago
yangzhenzhang
7303c3d3b8
add group ckpt
5 years ago
yangzhenzhang
9da3f9bec9
mini step grad accumulation
5 years ago
Ziyan
98566ddc07
enable gradients mean in opt shard
5 years ago
lichenever
39306d64fb
opt_pipeline_split
5 years ago
mindspore-ci-bot
b67aaf6773
!9832 expose_allgather_fusion_to_users
From: @gong_zi_yan
Reviewed-by:
Signed-off-by:
5 years ago
Ziyan
bbf8ec82b9
expose allgather fusion interface to users
5 years ago
lichenever
9595502278
optimizer_pipline_split
5 years ago
yao_yf
19fe28cb9b
hange strategys of last nodes in eval/predict at auto parallel mode
5 years ago
Xiaoda Zhang
e78228603b
move parallel-related black-list to core/ir, and fix the cloneCNode bug
5 years ago
lizhenyu
e3f7ae61db
add ps cache manager
5 years ago
Xiaoda Zhang
14d4926cf0
simplifying step-auto-parallel
5 years ago
mindspore-ci-bot
4bb7303463
!9424 [PipelineSplit]Fix Shared Parameter bug
From: @lichen666
Reviewed-by: @zh_qh
Signed-off-by:
5 years ago
mindspore-ci-bot
6b9e402790
!9396 enable allgather fusion
From: @gong_zi_yan
Reviewed-by: @stsuteng,@yangzhenzhang,@kisnwang
Signed-off-by: @stsuteng,@kisnwang
5 years ago
lichenever
818e920f02
fix_pipeline_split_param_shared_bug
5 years ago
Ziyan
e29f5c96cb
enable_allgather_fusion
5 years ago
lichenever
78e131cf15
pipeline_split adapt parallel
5 years ago
huangxinjing
f2d5f14e37
Fix review bot
5 years ago
yangzhenzhang
278e82a849
update pipeline parallel
5 years ago
yangzhenzhang
d4d6c4beae
update get device list in parallel ops
5 years ago