Commit Graph

564 Commits

Author SHA1 Message Date
i-robot 1da59bf782
!61455 Support starred_expression in graph mode.
Merge pull request !61455 from Margaret_wangrui/starred_expression_unpack_r2.3
2023-11-11 02:16:15 +00:00
liangzhibo 1484a1a608 Fix range with empty output 2023-11-10 23:10:17 +08:00
Margaret_wangrui cacd854fed Support starred_expression in graph mode. 2023-11-09 11:28:21 +08:00
ligan 2d7808f8ac fix ut bug. 2023-11-08 22:35:08 +08:00
chenfei cd4d7779af fix cum defination 2023-11-08 22:35:06 +08:00
lilinjie aee4ae3afc fix kernelTensor adaptation pooling op gpu bugs
pooling op cpu bugs
2023-11-08 22:34:56 +08:00
lilinjie a7d156662b dyn shape dev: CumSum, CumProd, CumMax, CumMin 2023-11-08 22:34:49 +08:00
zengxianglong 73c8843155 add scalar arithmetic 2023-11-08 22:34:46 +08:00
lilinjie b2c06c093e dyn shape dev: Div 2023-11-08 22:34:46 +08:00
chenfei d5281c5ad2 support kernel tensor for frontend infer 2023-11-08 22:34:43 +08:00
hanhuifeng2020 da3fd7fbde add auto_generate ops to import 2023-11-08 22:34:43 +08:00
chenfei 7cc7e890f6 fix recursive import 2023-11-08 22:34:42 +08:00
lilinjie 1ab733c60e KernelTensor adaptation 2023-11-08 22:34:32 +08:00
shen_jingxing 4d03585b40 st support on ascend platform 2023-11-08 22:34:28 +08:00
huangbingjian 5bd105613d Use prim_arg_register instead of prim_attr_register 2023-11-08 22:34:27 +08:00
yangruoqi713 9311a05595 [OPS][Dynamic] add type cast process for yaml args 2023-11-08 22:33:59 +08:00
yangruoqi713 9b112a1eb7 [OPS][Dynamic] enum c++ generate 2023-11-08 22:33:59 +08:00
yangruoqi713 e3695be45f [OPS][Dynamic] move ops generate directory 2023-11-08 22:33:56 +08:00
ckey_Dou 0ee0f11780 support register backend op into op_table 2023-11-08 22:33:41 +08:00
liangcanli 68c768f68d support user made metafg 2023-11-02 10:20:08 +08:00
panzhihui 46ed823605 Clean code 2023-10-25 15:38:28 +08:00
yiyanzhi_akane 098caf9b3e clean code 2023-10-23 16:00:20 +08:00
huan 2803a1120d update the release notes in 2.2 2023-10-18 18:35:29 +08:00
zengxianglong 4f60f8af93 fix lpnorm cpu precision problem 2023-10-09 15:32:38 +08:00
hedongdong 39654bf5d3 fix bug of inplaceupdate 2023-10-07 15:36:50 +08:00
greatpanc 7fe57b5b04 transform bugfix 2023-09-25 09:25:40 +08:00
TingWang d635bacbf6
!59221 add 2.1.1 release notes
Merge pull request !59221 from yanghaoran/master
2023-09-18 07:09:45 +00:00
yanghaoran a4c94933f0 add 2.1.1 release notes 2023-09-15 09:45:35 +08:00
panzhihui be2abc0952 Add regop and adapters for custom aicpu 2023-09-14 22:24:15 +08:00
i-robot 59f08851ae
!58809 [MS]remove useless file
Merge pull request !58809 from XianglongZeng/myms_2
2023-09-06 09:34:48 +00:00
panzhihui a2adaa3917 Add regop and adapters for custom aicpu 2023-09-06 14:17:57 +08:00
zengxianglong 43f7391efe remove useless file 2023-09-06 14:14:23 +08:00
i-robot f6b9c60347
!58502 merge global dict to local dict
Merge pull request !58502 from lianliguang/move-self-to-local-dict
2023-09-05 12:36:56 +00:00
lianliguang da6a478b78 merge global dict to local dict 2023-09-04 14:59:39 +08:00
zhoufeng 97118bdac9 owners
Signed-off-by: zhoufeng <zhoufeng54@huawei.com>
2023-08-31 11:05:33 +08:00
i-robot 5dc7431fcb
!58302 [MS][LITE][ASCEND] device data for ascend (python api and C++ api)
Merge pull request !58302 from yefeng/625-device_data_for_master
2023-08-30 17:01:49 +00:00
lizhenyu 8d359adcec Optimize embedding cache module performance:
1.embedding cache support multi-level pipeline
2.add big batch for embedding cache
3.local cache support lru strategy for embedding cache mode
2023-08-27 12:41:24 +08:00
yefeng 947cb5370b device data 2023-08-24 22:27:26 +08:00
i-robot eb51dea967
!58010 fix opencl resize failed
Merge pull request !58010 from liyan2022/master
2023-08-24 06:52:57 +00:00
albert-yan 6ea6e349ab fix opencl resize bug 2023-08-23 16:24:53 +08:00
i-robot 407947397e
!57923 框架新增支持bfloat16数据类型
Merge pull request !57923 from GuoZhibin/add_support_bfloat16
2023-08-21 07:24:35 +00:00
i-robot 54805623ce
!57914 add input and output variable pass
Merge pull request !57914 from 周超/master3
2023-08-19 09:32:11 +00:00
zhou_chao1993 ac8e413b52 add input and output variable pass 2023-08-18 16:31:27 +08:00
guozhibin b1b78562e5 add support to bfloat16 2023-08-18 12:41:34 +08:00
mengyuanli 9c7149c408 1.add ascendc folder
2.change vendor name form mslite to mslite_tbe_and_aicpu
2023-08-18 09:54:29 +08:00
i-robot 769d99dfe8
!57812 Refactor ExpanderFallback, base on Emitter class
Merge pull request !57812 from DeshiChen/0808_expander_fallback
2023-08-16 08:31:43 +00:00
dayschan 3d3fa61df5 ExpanderFallback: use the same framework as bpropexpander
1. change the virtual function of Emitter to non-const function
2. refactor the ExpanderFallback, base on the 'Emitter'
3. rewrite SiLU/SiLUGrad/VmapStackAssign/VmapUnstackAssign/Dense with new framework
4. support backoff on cpu for expanded nodes.
2023-08-15 21:06:44 +08:00
lianliguang 6650618d92 fix error for graph 2023-08-15 09:57:16 +08:00
i-robot e904004af2
!56274 Eps
Merge pull request !56274 from shenzhangyi/master
2023-08-10 12:42:00 +00:00
i-robot 1709eba9e2
!57689 move optim_ex to experimental dir
Merge pull request !57689 from liutongtong9/change_optim
2023-08-10 07:31:27 +00:00
shenzhangyi 513dcf1b27 Add Eps 2023-08-10 15:27:41 +08:00
testtong 4701cd9ad0 move optim to experimental dir 2023-08-09 20:46:57 +08:00
hangangqiang fa0988ef37 fix crop transpose axis function and format_pass bugs 2023-08-09 16:11:11 +08:00
zuochuanyong 0583b486f4 switch Pooling3D from onednn to self-developed. 2023-08-06 20:03:08 +08:00
gaoshuanglong 890683caa7 multinomial remove UniformReal cache, rrelu uniformsupport dynamic shape. 2023-08-04 11:31:41 +08:00
xuanyue fd2f2c9a1e [lite][lstm][fp16]unify the packed logic to right matrix 2023-08-03 15:44:14 +08:00
liangzhibo 6cabd733b6 Enable list getitem with bool index 2023-08-01 16:10:25 +08:00
xuanyue 1176e24ec2 [lite][lstm]prepare to support project-matmul part2 2023-07-28 14:19:14 +08:00
wang_shaocong b5c19ead4c [MSLITE] codex clean 2023-07-24 09:14:53 +08:00
kswang 3b1c8383a8 update owner 2023-07-20 19:43:56 +08:00
张清华 aca8dc2a3e Move the directory "jit" to "jit/ps". 2023-07-20 11:33:11 +08:00
i-robot 7e583f5575
!56848 utils for testing dynamic shape rapidly
Merge pull request !56848 from chengbin/test_util
2023-07-19 09:28:45 +00:00
ckey_Dou 4623398ccf utils for testing dyanmic shape rapidly 2023-07-19 15:32:08 +08:00
yanghaoran 255905661f update docker 2.1.0 2023-07-19 14:45:42 +08:00
i-robot a4e0b473a9
!56531 增加lite onnx parser 2(new)
Merge pull request !56531 from zhangdanyang/0710_master
2023-07-13 06:09:19 +00:00
zhangdanyang f7056eaee9 mvn multiple parser mapper 2023-07-12 10:48:30 +08:00
i-robot 673b4669b4
!55514 aicpu 910b support
Merge pull request !55514 from panzhihui/aicpu910b
2023-07-12 01:31:36 +00:00
panzhihui 1e3039cf16 add custom aicpu proto and cfg 2023-07-11 15:40:57 +08:00
lizhenyu 6fd8997e87 bugfix for failover to handle empty output value 2023-07-10 21:01:13 +08:00
liangcanli 1fac2e4d75 raise error when using dict as input of some multitypefuncgraph 2023-07-07 11:44:02 +08:00
i-robot 135d209c93
!56027 fix docs
Merge pull request !56027 from liutongtong9/docsfx
2023-07-04 11:30:28 +00:00
liutongtong9 41b7b121fc fix doc 2023-07-04 15:15:36 +08:00
LLLRT a534a290ad
!56073 更新CI门禁任务类型
Merge pull request !56073 from LLLRT/master
2023-07-03 11:28:33 +00:00
liuruotao c73040bae3 add CI task type 2023-07-03 15:29:53 +08:00
LuoJianing 4852a47ff6 check import for mindspore & amp 2023-06-30 10:27:21 +08:00
guozhibin 0dabd014b2 rewrite教程放到隐藏页;不展示非转测的接口" 2023-06-28 15:31:41 +08:00
i-robot 503df5d646
!54482 新增API ops.vector_norm 和 ops.matrix_norm
Merge pull request !54482 from tangdezhi_123/br_tdz_11
2023-06-26 03:09:33 +00:00
lizhenyu c7e3eafa1d Refine Infer function for EmbededdingLookup operator 2023-06-25 19:26:15 +08:00
tangdezhi_123 a515a0e842 新增API ops.vector_norm 和 ops.matrix_norm 2023-06-25 18:15:37 +08:00
i-robot 995c65d348
!55260 [lite]conv-arm64 support slide window
Merge pull request !55260 from 徐安越/master5
2023-06-21 02:22:03 +00:00
i-robot 8f13ed1829
!55448 Fix PyNative tasksink bug.
Merge pull request !55448 from caifubi/master-pynative-tasksink-bugfix
2023-06-21 01:48:24 +00:00
xuanyue a44bff4c4a conv-arm64 support slide window 2023-06-20 21:32:19 +08:00
caifubi 01a33913b0 Bugfix for PyNative TaskSink 2023-06-20 14:49:57 +08:00
ckey_Dou fc3580126d fix api dynamic shape 2023-06-14 14:25:25 +08:00
i-robot 86a75c9b46
!54013 regisgter ValudeDependedArgIndices in Infer function
Merge pull request !54013 from chengbin/remove_depend_map
2023-06-13 04:16:15 +00:00
i-robot ea8e09e654
!55024 dataset gallery zh-cn
Merge pull request !55024 from luoyang/code_docs_gallery
2023-06-12 07:18:43 +00:00
YangLuo 387c71162d dataset gallery zh-cn 2023-06-12 11:42:14 +08:00
i-robot d2360a364d
!54770 supplement dtypes for some meta ops master
Merge pull request !54770 from 李林杰/0530_add_dtypes_for_some_meta_ops_master
2023-06-09 06:33:59 +00:00
i-robot 10a0663d00
!54992 infer shape inside CustomAkgOpInferFunc
Merge pull request !54992 from looop5/ascend_dyn_commit
2023-06-09 02:15:21 +00:00
i-robot 4cb2af2e8d
!55007 Opt symbol resolve performance, by parse jit function in parser. Support condition checking in parser, and eliminate dead branch during parser.
Merge pull request !55007 from zhangqinghua/master
2023-06-09 01:41:36 +00:00
i-robot 87df235f06
!54593 adapt Cast on aicpu with complex master
Merge pull request !54593 from 李林杰/0524_adapt_Cast_on_ascend_with_complex_master
2023-06-08 16:55:17 +00:00
张清华 68a3b8bc31 Opt symbol resolve performance, by parse jit function in parser.
Support condition checking in parser, and eliminate dead branch during parser.
2023-06-08 11:28:34 +08:00
looop5 0567a3a929 infer shape inside CustomAkgOpInferFunc
delete custom_op_infer.h

refine CustomAkgOpInferFunc

refine Infer

refine InputsInfoNotChanged

refine InferOnline
2023-06-07 17:59:17 +08:00
ckey_Dou 926294d3a0 regisgter ValudeDependedArgIndices in Infer function 2023-06-06 16:43:43 +08:00
LuoJianing 8167a84e1f gallery docs change 2023-06-06 15:58:34 +08:00
YangLuo 17353f611a fix gallery links 2023-06-05 19:11:39 +08:00
LuoJianing a76395ba65 add communicate_ops and pynative api 2023-06-05 12:49:53 +08:00
YangLuo 43a790f059 minddata gallery en 2023-06-02 17:20:01 +08:00
lilinjie 663d578c05 supplement dtypes for some meta ops 2023-06-01 11:54:56 +08:00
mengyuanli 28dec3fc1f fix bug of xdivy 2023-05-27 11:48:26 +08:00
lilinjie e97ba2e4d6 adapt Cast on aicpu 2023-05-25 14:48:51 +08:00
i-robot 651f8ae568
!54434 broadcast 优化
Merge pull request !54434 from zong_shuai/broadcast
2023-05-23 09:53:51 +00:00
zong-shuai c5d81658b2 binary ops 的gpu性能优化 2023-05-23 14:49:13 +08:00
i-robot 5dbab6e61e
!54349 bump up version number to 2.1.0
Merge pull request !54349 from yanghaoran/master
2023-05-20 07:04:32 +00:00
yanghaoran 50b2d20dcd
回退 'Pull Request !54166 : binary broadcast类型gpu算子性能优化' 2023-05-19 10:28:22 +00:00
yanghaoran 3d8af70ae8 bump up version number to 2.1.0 2023-05-19 17:31:02 +08:00
ZPaC 407343bd86 Update versions of nccl and ompi in MindSpore. 2023-05-19 15:29:24 +08:00
zong-shuai 0a08686ef7 binary ops 的gpu性能优化 2023-05-18 22:02:08 +08:00
i-robot 3da97434c0
!53551 [MS]{Lite][Task] optimize ascend kernel build
Merge pull request !53551 from 刘力力/ascend_custom_build_master
2023-05-09 12:09:31 +00:00
liu lili 0737256dad t # This is a combination of 10 commits.
mindspore lite: add ascend custom op into packages

mindspore lite: move dir from ascend_new to ascend

mindspore lite: add check block

mindspore lite: add check block

mindspore lite: add check block

mindspore lite: add check block

mindspore lite: add check block

mindspore lite: add check block

mindspore lite: add check block

mindspore lite: add cloud macro condition

lll: update code
2023-05-08 12:00:42 +08:00
i-robot 1fb62a928c
!53597 add new aicpu operator supported and code clean
Merge pull request !53597 from 沈竞兴/MigrateOperator
2023-04-28 13:42:32 +00:00
张清华 130dee335c [JIT Fallback] Optimize comment annotation usage, and so on. 2023-04-28 10:40:36 +08:00
shen_jingxing eda49dffeb add new aicpu operator supported and code clean 2023-04-28 09:24:24 +08:00
i-robot 5bf06a5ad7
!53354 [MS][LITE]add menjin import ms and mslite
Merge pull request !53354 from luoyuan/add-menjin-import-yun-lite
2023-04-27 06:23:50 +00:00
i-robot 6173398cc8
!53519 migrate canndev AICPU operator to MindSpore
Merge pull request !53519 from 沈竞兴/MigrateOperator
2023-04-27 02:57:54 +00:00
luoyuan 206c37d803 [MS][LITE]add menjin import ms and mslite 2023-04-27 09:06:04 +08:00
shen_jingxing 9cff2b9b31 migrate canndev AICPU operator to MindSpore 2023-04-26 16:30:47 +08:00
tangdezhi_123 ed2a1f6b28 新增Tensor下的API,新增内容包括,实现,英文注释,中文注释, ST用例;
新增API如下,
Tensor.aminmax
Tensor.fmax
Tensor.fmin
Tensor.nanquantile
Tensor.orgqr
Tensor.outer
Tensor.quantile
2023-04-26 14:43:12 +08:00
i-robot 6f8072fb44
!53281 canndev operator LeftShift merge to mindspore
Merge pull request !53281 from 沈竞兴/MigrateOperator
2023-04-24 02:09:41 +00:00
shen_jingxing 3975462b80 canndev code merge to mindspore 2023-04-23 14:24:26 +08:00
VectorSL 919bd40093 add mux send/recv for ascend 2023-04-21 16:30:31 +08:00
i-robot dc76a17da8
!52757 support ascend infer process
Merge pull request !52757 from looop5/ascend_infer_commit
2023-04-14 08:56:08 +00:00
looop5 ec1570cac8 support ascend infer process
fix compile

add FuseLayerNorm pattern
2023-04-13 20:28:37 +08:00
i-robot c7271ac52d
!52661 merge canndev code to mindspore
Merge pull request !52661 from 沈竞兴/codeclean320
2023-04-13 11:13:02 +00:00
i-robot aa80a9644b
!52664 move grad to grad_experiment
Merge pull request !52664 from r1chardf1d0/rmpb
2023-04-13 00:18:22 +00:00
r1chardf1d0 9a45e88e9b move grad to grad_ex 2023-04-12 15:17:54 +08:00
i-robot 645ed6a23d
!50828 [lite]GRU fusion for train
Merge pull request !50828 from 徐安越/master4
2023-04-12 06:10:48 +00:00
shen_jingxing 380ee5aac8 merge canndev aicpu code to mindspore 2023-04-12 09:45:43 +08:00
xuanyue efaeeebb95 GRU-fusion for train-export 2023-04-11 14:11:55 +08:00
r1chardf1d0 1f2f02a378 expander 2023-04-11 11:06:57 +08:00
i-robot 191606cf26
!52287 Fix bugs
Merge pull request !52287 from panzhihui/bugfix
2023-04-06 10:39:10 +00:00
yanghaoran d6af23d40b docker for 2.0rc1 2023-04-04 21:08:53 +08:00
panzhihui d3a9894beb Fix bugs 2023-04-04 15:48:42 +08:00
i-robot 5586e7afed
!52162 fix check in api
Merge pull request !52162 from chengbin/fix_check
2023-04-04 06:06:06 +00:00
ckey_Dou 4c5d245a78 add check in api
remove useless return None
2023-04-03 20:48:26 +08:00
i-robot ea05eba9f8
!51279 fix mindir pb compiler error
Merge pull request !51279 from lianliguang/fix-pb
2023-04-03 11:30:26 +00:00
i-robot 0fb84bb2c0
!51823 add int32 and seed for uniformcandidatesampler on Ascend
Merge pull request !51823 from yanzhenxiang2020/br_candidate_sampler
2023-04-03 09:15:48 +00:00
lianliguang 39377a3c92 fix mindir pd compile error 2023-04-03 10:08:49 +08:00
yanzhenxiang2020 6f415843a9 add int32 and seed for candidate_sampler aicpu 2023-04-01 14:49:00 +08:00
i-robot c0b59a690f
!49303 revert check in constexpr
Merge pull request !49303 from chengbin/fix_check
2023-04-01 02:21:22 +00:00
ckey_Dou 4988e94dfd fix check in api 2023-03-31 16:38:39 +08:00
lizhenyu e25ae03d37 clean code warnings 2023-03-31 11:56:18 +08:00
i-robot 8b6a7afbdb
!51163 clean code master
Merge pull request !51163 from jiangzhenguang/clean_code_master
2023-03-30 11:58:24 +00:00
moran abacfb2e8b fix CodeCheck in 3.28 2023-03-30 11:06:35 +08:00
jiangzhenguang 676edf7950 code clean 2023-03-30 10:28:57 +08:00
i-robot 2382b8c8d4
!51716 Grad of make_range to C++ and fix some bugs
Merge pull request !51716 from NaCN/fix_make_range_grad
2023-03-29 02:40:04 +00:00
huangchengnuo bbe31a16e7 Grad of make_range to C++ and fix some bugs 2023-03-27 15:50:44 +08:00
i-robot e4f2b77af5
!50119 [API] Add bias to ops.conv2d
Merge pull request !50119 from shaojunsong/conv_bias
2023-03-25 03:46:10 +00:00
i-robot e7d3e467fe
!51187 Clean code for compiler
Merge pull request !51187 from YuJianfeng/master
2023-03-25 01:56:16 +00:00
shaojunsong f6d795bf40 add bias to conv2d 2023-03-24 23:45:25 +08:00