Nigel
/
mindspore
forked from mindspore-Ecosystem/mindspore
0
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
35,316 Commits 24 Branches 0 Tags 3.9 GiB
Commit Graph

450 Commits

Author SHA1 Message Date
lichenever a7f8024c29 add_replace_graph_for_conv2d 2021-07-28 11:02:21 +08:00
yao_yf dc7dc7d3fa dataset strategy set 2021-07-27 18:44:10 +08:00
yangzhenzhang 80e5cc0e52 add parallel op for gatherd 2021-07-26 11:14:25 +08:00
Xiaoda Zhang bb5d4212f7 enable All2All in infering redistribution ops 2021-07-24 09:09:01 +08:00
lichenever 3c7cfb7c08 auto_parallel_support_control_flow 2021-07-22 17:31:31 +08:00
i-robot a7d40fc220 !20520 [AutoParallel]Add op AllToAllv 
Merge pull request !20520 from lichen/add_op_AllToAllv
 2021-07-20 07:57:03 +00:00
lichenever 8c1998fd6b add_op_AllToAllv 2021-07-20 10:49:28 +08:00
i-robot c9d3c1d346 !20411 enable optimizer parallel for inference 
Merge pull request !20411 from gziyan/enable_opt_shard_predict
 2021-07-20 01:37:06 +00:00
yangzhenzhang b31cd27a08 update check strategy for conv2d 2021-07-19 15:03:26 +08:00
Ziyan 1c9166e0a6 remove restriction for opt shard in inference 2021-07-16 17:16:12 +08:00
Xiaoda Zhang 04381273b3 Add the sharding propagation function: 
1) users configure sharding strategies for operators;
2) framework will propagate the strategies from configured-ops to
non-configured ops using BFS;
3) the propagation goal is to minimize redistribution communication
cost;
 2021-07-14 10:54:18 +08:00
chenhaozhe 086a871975 Change Loss to LossBase 2021-07-02 16:41:02 +08:00
lichenever db8850a4a3 pipeline_support_predict_master 2021-06-26 09:25:59 +08:00
Ziyan be1f5a43d7 opt shard fit micro batch 2021-06-24 14:52:19 +08:00
yangzhenzhang 69acf757d0 add parallel op for conv2d backprop input 2021-06-24 10:01:49 +08:00
yangzhenzhang 24370b5613 add parallel op for maxpool 2021-06-21 09:00:56 +08:00
yangzhenzhang af0d28de48 add parallel op for batchnorm 2021-06-17 15:43:24 +08:00
i-robot 85d860e6a2 !16457 [AutoParallel]pipeline_split_adapt_master 
Merge pull request !16457 from lichen/pipeline_split_adapt_master
 2021-06-11 11:37:40 +08:00
lichenever db5d508356 pipeline_split_adapt_master 2021-06-10 20:17:33 +08:00
yangzhenzhang 7a40741048 add parallel operator for conv2d 2021-06-10 14:58:51 +08:00
Ziyan 95ac0f6d58 fix optimizer weight shard config 2021-06-08 10:44:17 +08:00
chenhaozhe 9da8534396 change _Loss to Loss 2021-06-03 15:26:59 +08:00
mindspore-ci-bot 1c8fda25ef !16478 handle load op in step parallel 
From: @gong_zi_yan
Reviewed-by: @yangzhenzhang,@stsuteng
Signed-off-by: @stsuteng
 2021-05-29 09:32:35 +08:00
mindspore-ci-bot b45b63fc58 !17239 add parallel gathernd test case 
From: @hanyang001
Reviewed-by: @stsuteng,@yangzhenzhang
Signed-off-by: @stsuteng
 2021-05-29 09:11:25 +08:00
Wan Hanyang c51dff2634 add parallel gathernd test case 2021-05-28 17:21:11 +08:00
Wan Hanyang 3ce521d78f add parallel layernorm test case 2021-05-28 17:20:04 +08:00
Ziyan 4b17493e52 handle load in step parallel 2021-05-28 09:16:40 +08:00
yangzhenzhang d711d98f07 clean duplicate code 2021-05-25 17:25:34 +08:00
yao_yf 732d13ccff parallel dropout support repeated compute 2021-05-20 19:42:13 +08:00
yangzhenzhang 6aa3859131 modify check strategy for scatter update 2021-05-10 11:22:00 +08:00
Ziyan 2a752f24bf enable not fully use opt shard 2021-05-07 15:32:48 +08:00
yao_yf e967f1939b parallel envs variable check 2021-05-07 09:12:53 +08:00
mindspore-ci-bot 78fcdbc7c9 !15790 modify scatter update op 
From: @yangzhenzhang
Reviewed-by: @kisnwang,@stsuteng
Signed-off-by: @stsuteng
 2021-04-28 10:48:19 +08:00
yangzhenzhang 075f680a42 modify scatter update op 2021-04-27 20:23:14 +08:00
Xiaoda Zhang aa52399200 Making the Tile operator to have more parallel strategies 2021-04-27 11:36:40 +08:00
yao_yf 093ef784de dont insert virtualoutput for scalar 2021-04-26 19:55:16 +08:00
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
 2021-04-26 09:21:00 +08:00
mindspore-ci-bot 49d6c029a6 !15542 split axis and batch for gather 
From: @yangzhenzhang
Reviewed-by: @kisnwang,@stsuteng,@stsuteng
Signed-off-by: @stsuteng,@stsuteng
 2021-04-25 19:33:09 +08:00
yangzhenzhang 5828973978 fix bug for dropout do mask 2021-04-25 16:47:44 +08:00
yao_yf 21276408b8 parallel virtual_out_ops 2021-04-25 11:18:54 +08:00
yangzhenzhang 213922574e split axis and batch for gatherv2 2021-04-23 16:59:35 +08:00
yangzhenzhang c2ca2232c5 add select op 2021-04-20 09:23:22 +08:00
mindspore-ci-bot 1c9d3c0aa0 !15353 add parallel operator for scatter update 
From: @yangzhenzhang
Reviewed-by: @kisnwang,@stsuteng
Signed-off-by: @stsuteng
 2021-04-20 09:03:08 +08:00
mindspore-ci-bot 0fd1726e79 !15172 Clean GraphKernel's codes from frontend 
From: @dayschan
Reviewed-by: @gaoxiong1,@dylangeng,@gaoxiong1
Signed-off-by: @dylangeng,@dylangeng
 2021-04-19 09:34:35 +08:00
yangzhenzhang 9cdd70433f add scatterupdate op 2021-04-17 17:32:48 +08:00
yangzhenzhang d070af122f add topk op 2021-04-17 14:12:49 +08:00
dayschan 771e3f61f3 Clean GraphKernel's codes from frontend 
1. set class GraphKernel as deprecated, and treat it as Cell
2. set class InplaceAssign as deprecated, suggested using Assign instead.
3. set op_selector as deprecated, removed the _selected_ops and _selected_grad_ops, replaced with real operations
4. removed the two passes of GraphKernel from frontend
5. removed the GraphKernel's codes from other modules
 2021-04-17 11:03:34 +08:00
yangzhenzhang f9f5df368e add gathernd op 2021-04-16 14:29:03 +08:00
yangzhenzhang bcd2ecc403 check layouts for shared parameter 2021-04-15 10:39:05 +08:00
yao_yf a83fb3316b fix parallel timeout 2021-04-02 20:19:26 +08:00
yao_yf 4d0635eabe set parallel communication init flag in parallel ut 2021-03-30 11:21:57 +08:00
dingpeifei 87e41aaeee IR operators of GPU and CPU are unified as batchnorm 2021-03-18 19:02:28 +08:00
mindspore-ci-bot 7454ac8ecd !13382 [PipelineSplit]change pipeline key word 
From: @lichen666
Reviewed-by: @kisnwang,@zhunaipan
Signed-off-by: @zhunaipan
 2021-03-16 20:20:32 +08:00
lichenever a2b2727ba8 change_pipeline_key_word 2021-03-16 14:02:35 +08:00
LianLiguang 17b9758543 unify range ops 2021-03-16 10:47:42 +08:00
mindspore-ci-bot 7ba21f8d8c !12900 Add communication parallel mode. 
From: @liujunzhu
Reviewed-by: @zhoufeng54,@guoqi1024
Signed-off-by: @guoqi1024
 2021-03-06 15:55:49 +08:00
liujunzhu 6541b96c40 Add communication parallel mode. 2021-03-05 21:36:03 +08:00
Ziyan ec9793861f fix grad accu 2021-03-05 10:28:10 +08:00
mindspore-ci-bot 7ff2b3b499 !12781 fix bug of amp bn cast 
From: @jojobugfree
Reviewed-by: 
Signed-off-by:
 2021-03-04 10:14:51 +08:00
caifubi a6959c2a13 fix bn cast bug 2021-03-02 17:24:47 +08:00
yangzhenzhang a70d616841 mini step grad accumulation 2021-03-01 10:20:12 +08:00
wangshuide2020 72e938eb06 change dimension of input for FusedBatchNormEx from 2D to 4D in test_two_matmul_batchnorm_ex. 2021-02-25 09:37:59 +08:00
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)
```
 2021-02-08 09:01:15 +08:00
jinyaohui 30a27b2adb modify Gelu、FastGelu to GeLU and FastGeLU 2021-02-05 17:19:52 +08:00
mindspore-ci-bot 74652eb942 !12044 modify pack to stack 
From: @jinyaohui
Reviewed-by: 
Signed-off-by:
 2021-02-04 21:02:09 +08:00
jinyaohui 8022f9a6ed modify pack to stack 2021-02-04 18:54:31 +08:00
yangzhenzhang 726ea32778 merge parameter slice compile graph only once 2021-02-03 19:39:32 +08:00
l00591931 9ec100d069 Change TensorAdd to Add, from r1.1 to master 2021-02-01 17:53:52 +08:00
mindspore-ci-bot 9fa0499fa0 Change GatherV2 to Gather r1.1 to master 2021-01-30 14:13:47 +08:00
yangzhenzhang cbca482e59 delete useless parameter in pipeline parallel 2021-01-21 19:05:40 +08:00
yangzhenzhang 7303c3d3b8 add group ckpt 2021-01-21 14:41:05 +08:00
lilei 9a45c4419c modify batch_normal 2021-01-13 20:59:46 +08:00
yangzhenzhang 9da3f9bec9 mini step grad accumulation 2021-01-06 09:25:06 +08:00
mindspore-ci-bot 2e684df5b1 !10686 fix infer rank list typo and add testcase 
From: @gong_zi_yan
Reviewed-by: @stsuteng,@zhunaipan
Signed-off-by: @stsuteng
 2020-12-28 21:20:19 +08:00
Ziyan 2c3b99ce91 fix infer rank list typo 2020-12-28 18:46:01 +08:00
mindspore-ci-bot b67aaf6773 !9832 expose_allgather_fusion_to_users 
From: @gong_zi_yan
Reviewed-by: 
Signed-off-by:
 2020-12-28 09:54:57 +08:00
Ziyan bbf8ec82b9 expose allgather fusion interface to users 2020-12-27 17:44:45 +08:00
ms_yan deb1e6e965 use from_numpy and add do_copy option 2020-12-23 21:38:05 +08:00
Ziyan c5c905fdf5 add restriction for opt shard 2020-12-19 17:43:55 +08:00
huangxinjing a8446af1ab Fix condition check 2020-12-18 16:43:43 +08:00
jjfeing 1984cf8e20 unify mindir 2020-12-12 17:55:16 +08:00
mindspore-ci-bot d5db8872fd !9834 Fix wrong input argument of Reshape for multi field embedding 
From: @huangxinjing
Reviewed-by: @zhunaipan,@stsuteng
Signed-off-by: @stsuteng
 2020-12-11 21:31:48 +08:00
huangxinjing 996ee72c50 Fix embedding layer 
Fix input check
 2020-12-11 19:13:09 +08:00
yao_yf 19fe28cb9b hange strategys of last nodes in eval/predict at auto parallel mode 2020-12-11 11:11:28 +08:00
Xiaoda Zhang e78228603b move parallel-related black-list to core/ir, and fix the cloneCNode bug 2020-12-10 15:57:01 +08:00
mindspore-ci-bot ec3983b77d !9577 support distributed predict 
From: @gong_zi_yan
Reviewed-by: @caozhou_huawei,@yao_yf,@stsuteng,@zh_qh
Signed-off-by: @stsuteng
 2020-12-08 10:28:30 +08:00
Ziyan e7e9dae54d support distributed predict 2020-12-07 16:12:45 +08:00
Xiaoda Zhang 9a9e3a751e set cnode's fullname when cloning 2020-12-04 15:59:44 +08:00
lichenever 818e920f02 fix_pipeline_split_param_shared_bug 2020-12-03 19:30:29 +08:00
lichenever 78e131cf15 pipeline_split adapt parallel 2020-12-02 18:52:40 +08:00
yangzhenzhang 7b33f3e2ac gatherv2 axis split repeated calculation 2020-11-30 17:03:29 +08:00
yangzhenzhang 7278f5c109 update_gatherv2_op 2020-11-28 16:56:39 +08:00
Xiaoda Zhang c79e988b0d set fullname for reshape after reshape-elimination 2020-11-28 12:16:51 +08:00
yao_yf 9cda064716 auto parallel predict 2020-11-26 20:13:15 +08:00
mindspore-ci-bot 3d6d820612 !8154 Add nn.MultiFieldEmbedding for the embedding lookup opearations 
From: @huangxinjing
Reviewed-by: 
Signed-off-by:
 2020-11-24 09:49:39 +08:00
mindspore-ci-bot d915fd9b96 !8592 change repeat_elements to a composite op 
From: @tom__chen
Reviewed-by: 
Signed-off-by:
 2020-11-23 23:15:23 +08:00
huangxinjing b0deb7a289 Add dense embedding 2020-11-23 20:21:58 +08:00
tom__chen a52cd685fc change repeat_element op to a composite op 2020-11-19 15:42:22 -05:00
huangxinjing 89e7778497 Add UnsortedSegmentMax Operation 2020-11-19 17:41:54 +08:00
huangxinjing 8129806475 Add slice parallel op 2020-11-18 20:20:33 +08:00
mindspore-ci-bot ee72de1db2 !8441 Add Parallel Implements of UniformCandidateSampler 
From: @huangxinjing
Reviewed-by: 
Signed-off-by:
 2020-11-18 18:59:31 +08:00
mindspore-ci-bot cd6236c0a0 !8712 update pipeline parallel interface 
From: @yangzhenzhang
Reviewed-by: @kisnwang
Signed-off-by:
 2020-11-18 18:59:04 +08:00
mindspore-ci-bot fb0e866ad1 !8269 forward unique dynamic shape 
From: @yao_yf
Reviewed-by: 
Signed-off-by:
 2020-11-18 18:53:46 +08:00
yao_yf 31819bb4a7 support forward unique 2020-11-18 15:22:14 +08:00
yangzhenzhang 278e82a849 update pipeline parallel 2020-11-18 13:55:34 +08:00
huangxinjing 2730cef047 Uniform Sampler Base Update 2020-11-18 09:39:32 +08:00
Ziyan 0ddb754edb run cast before parallel optimizer 2020-11-13 14:04:04 +08:00
huangxinjing 12e9107162 Fix VirtualDiv Int32 error 2020-11-11 09:32:57 +08:00
mindspore-ci-bot 4e07f43dff !8323 [Auto parallel] Supporting for-loop in strategy-searching 
From: @xiaoda_zh
Reviewed-by: 
Signed-off-by:
 2020-11-10 10:33:25 +08:00
Xiaoda Zhang aa13d6b1cd support for-loop in auto-parallel 2020-11-09 17:18:30 +08:00
yangzhenzhang 9747bde861 add range op 2020-11-09 15:14:07 +08:00
yangzhenzhang 0a79ab82ae add parallel ops 2020-11-06 12:09:19 +08:00
Xiaoda Zhang aa84484049 enabling approximation in DP algorithms 2020-11-05 16:07:26 +08:00
Ziyan c33f2cd796 fix auto optimizer weight shard 2020-11-02 16:25:48 +08:00
Yi Huaijie 148d645915 fix ReLUV2 error 2020-10-29 20:29:01 +08:00
huangxinjing bf5d21770a Add UnsortedSegmentSum and UnosrtedSemenMin For Oparallel Implements 2020-10-28 08:57:02 +08:00
mindspore-ci-bot d4142d682d !7437 auto parallel support dynamic 
Merge pull request !7437 from yao_yf/auto_parallel_support_dynamic_shape
 2020-10-27 21:02:28 +08:00
mindspore-ci-bot 08e6ac0b09 !7805 support split ValueList 
Merge pull request !7805 from yihuaijie/master
 2020-10-27 19:22:52 +08:00
yao_yf 0cfe72cd22 auto parallel dynamic 2020-10-27 19:17:27 +08:00
Yi Huaijie 3102c4ff8d support split ValueList 2020-10-27 17:58:36 +08:00
yao_yf 65d8e63580 set last node data parallel or repeat calculate in eval/predict 2020-10-26 20:05:00 +08:00
mindspore-ci-bot f711e01f61 !7719 refactor cnode strategy 
Merge pull request !7719 from gziyan/refactor_cnode
 2020-10-24 16:16:29 +08:00
Ziyan 069318899a refactor get cnode strategy 2020-10-24 15:06:26 +08:00
mindspore-ci-bot d50736df2c !7607 modify_normal_seed 
Merge pull request !7607 from lilei/modify_normal
 2020-10-23 10:40:58 +08:00
lilei 0c043d18ef modify_normal 2020-10-22 21:35:44 +08:00
mindspore-ci-bot 8bfe141680 !7571 fix bug in reshape strategy search when reshape as the first operator 
Merge pull request !7571 from yao_yf/reshape_strategy_search_fix_when_reshape_is_first_op
 2020-10-22 10:16:33 +08:00
yao_yf f7189adb91 fix bug in reshape strategy search when reshape is first operator 2020-10-22 09:15:18 +08:00
mindspore-ci-bot f487797694 !7495 [Auto-parallel] Overwrite strategies for star graph structures 
Merge pull request !7495 from Xiaoda/27-star-elimination-overload
 2020-10-22 09:10:52 +08:00
mindspore-ci-bot b2b9016ddf !7429 modify normal seed 
Merge pull request !7429 from lilei/modify_bug
 2020-10-21 09:37:47 +08:00
lilei 04075671cf modify_normal_seed 2020-10-20 22:06:17 +08:00
Xiaoda Zhang fba2bfeb54 overwrite strategies for star graph structure 2020-10-20 13:40:24 +08:00
yangzhenzhang fc4ed975c4 handle repeated calculation 2020-10-19 10:47:01 +08:00
yao_yf 022005b94a fix a bug cases in reshape redistribution 2020-10-15 16:58:09 +08:00
yao_yf f60d81a15f support reshape redistribution in all scenes 2020-10-13 21:21:17 +08:00
mindspore-ci-bot fd0c03c493 !7090 implement parallel BroadcastTo 
Merge pull request !7090 from yihuaijie/master
 2020-10-09 19:10:20 +08:00
Yi Huaijie 45d373d40e implement parallel BroadcastTo 2020-10-09 15:17:37 +08:00
Ziyan ddc0113058 enable parallel optimizer in auto parallel 2020-10-08 14:51:15 +08:00
mindspore-ci-bot 58610443cb !7023 modify endofsequence for multi-machine 
Merge pull request !7023 from HW_KK/master
 2020-09-30 08:40:57 +08:00
wuweikang a32811e160 modify endofsequence for multi-machine 2020-09-29 16:17:38 +08:00
mindspore-ci-bot 2a799fe90e !6018 Set the number of epochs in mode.train() non-sink Mode 
Merge pull request !6018 from h.farahat/ms_nonsink_epochs
 2020-09-28 22:15:54 +08:00
hesham 9cee0d2143 Add num_epochs to non-sink training 2020-09-24 23:24:43 -04:00
huangxinjing 4ef439e27b Add stage information for ops and strategy 2020-09-25 10:42:43 +08:00
Yi Huaijie 6066b16838 implement parallel Pack 2020-09-23 09:22:21 +08:00
mindspore-ci-bot 9475f9a19a !6548 Implement parallel Split 
Merge pull request !6548 from yihuaijie/dev
 2020-09-22 19:41:04 +08:00
mindspore-ci-bot dfe77372f5 !6505 Set top graph parameters' name the same as original graph parameters. 
Merge pull request !6505 from 张清华/master2
 2020-09-21 20:06:59 +08:00
Zhang Qinghua 6c72d88ba1 Set top graph parameters' name as original graph parameters. 2020-09-21 17:44:02 +08:00
guohongzilong a754dea90c delete SoftmaxCrossEntropyExpand 2020-09-20 21:17:24 +08:00
mindspore-ci-bot 5a20b11012 !6502 [AutoParallel]Fix auto parallel find loss bug 
Merge pull request !6502 from lichen/fix_auto_parallel_find_loss_bug
 2020-09-19 15:56:56 +08:00
Yi Huaijie 18ed2bec53 implement parallel Split 2020-09-19 12:35:56 +08:00
lichenever d4bba3f1d2 fix_auto_parallel_find_loss_bug 2020-09-18 20:27:33 +08:00