Code

update Paddle to newest version#1

Code

Merged

thisjiang merged 61 commits intothisjiang:developthisjiang/Paddle:developfrom PaddlePaddle:developPaddlePaddle/Paddle:developCopy head branch name to clipboardDec 18, 2020

+13,783-1,127

Commits

Commits on Dec 11, 2020

• Polish hash function of executor cache key (#29556)

  Show description for 2a42250
  Aurelius84

  authored
  2a42250

• Add the strategy of skipping cc/cu test compilation and execution in CI (#29499)

  Show description for b5d4a1f
  Avin0323

  authored
  b5d4a1f

• fix train eval set error in static mode (#29540)

  chenwhql

  authored
  c1a26e2

• add cast cuda kernel (#29352)

  zhangting2020

  authored
  30d9589

• improve dropout (#29465)

  Show description for 6702040
  zhangting2020

  authored
  6702040

• remove duplicated macro (#29563)

  zhiqiu

  authored
  1e72e03

• [Sharding] add hybrid-dp feature (#29518)

  Show description for d33d468
  JZ-LIANG

  authored
  d33d468

• update for xpu ci. (#29568)

  jiweibo

  authored
  740c0d5

Commits on Dec 14, 2020

• Fix precision problem (#29567)

  Show description for 08f24a3
  GeminiCarrie

  authored
  08f24a3

• [oneDNN] Making ThreadID info in caching key optional (#29272)

  jczaja

  authored
  f6cca62

• Fix Windows Unittest (#29543)

  Show description for 831e913
  zhhsplendid

  authored
  831e913

• [Dy2Stat] 1. Fix bug of for-range stmts. 2. Support that step value is negative in for-range stmts (#29519)

  Show description for 0cad115
  liym27

  authored
  0cad115

• support roi_align & affine_channel for kunlun (#29561)

  Show description for 79a41a9
  QingshuChen

  authored
  79a41a9

• Fix compile problem when cuda_arch < 6000 (#29576)

  Show description for c016383
  zhiqiu

  authored
  c016383

• add service (#29560)

  Show description for 0034273
  seiriosPlus

  authored
  0034273

• fix unittst unstable issue on ci machine (#29588)

  Show description for d72604c
  baiyfbupt

  authored
  d72604c

• gen nccl id use socket (#29431)

  wangxicoding

  authored
  467c716

• add some feature for paddle.flops (#29572)

  yukavio

  authored
  ee1a7d0

• add float16 into adaptive_avg_pool2d check list. (#29547)

  huangxu96

  authored
  2cb6f94

• update, test=develop (#29559)

  

  lilong12

  authored
  ff6a145

• Added verbose oneDNN lib version (#29378)

  arlesniak

  authored
  62d4483

• elementwise_add_grad Op optimization (#29575)

  wangchaochaohu

  authored
  ac4bae8

• disable test_parallel_executor_profiler in cuda 10.1 (#29581)

  Show description for 81acc32
  luotao1

  authored
  81acc32

Commits on Dec 15, 2020

• Add clip double grad (#29590)

  qingqing01

  authored
  8d549fc

• Simplify the prompt of const_cast check. (#29548)

  Show description for a908208
  wadefelix

  authored
  a908208

• Fix bug of matmul_v2 for broadcast case (#29599)

  Show description for 1efef8b
  ForFishes

  authored
  1efef8b

• Fix the dowanload bug in the case of multiple machines (#29551)

  Show description for fb6697b
  ForFishes

  authored
  fb6697b

• add alias for fluid.contrib.mixed_precision (#29562)

  Show description for c05170d
  huangxu96

  authored
  c05170d

• fix cache pip error (#29618)

  zhwesky2010

  authored
  18f9df0

• fix none-contiguous bug for python api. (#29615)

  jiweibo

  authored
  78dad78

• optimize for long width for elementwise (#29602)

  wangchaochaohu

  authored
  1b69e52

• add transpose double grad test=develop (#29600)

  Show description for 7779768
  lijianshe02

  authored
  7779768

• Add tf32 support for A100 tensor core acceleration for cuBLAS (#28732)

  AshburnLee

  authored
  efea540

• Add complex dtype op (add) test example (#29603)

  Show description for f02aece
  chenwhql

  authored
  f02aece

• New UT should not exceed 15s (#29492)

  Show description for 2926e74
  XieYunshen

  authored
  2926e74

• 1. add python version selection 2.add dynamic flags setting. (#29612)

  chen-zhiyu

  authored
  f5f8809

Commits on Dec 16, 2020

• fix gen_nccl_id_op_helper compile failed, test=develop (#29614)

  wangxicoding

  authored
  613c46b

• fix wmt14 doc, remove backward, add bidirect direction in rnn api (#29633)

  Show description for 84bae27
  joey12300

  authored
  84bae27

• disable ut test_cumsum_op (#29613)

  XieYunshen

  authored
  d0b789d

• refine (#29622)

  wangchaochaohu

  authored
  eab44e1

• improve dropout grad (#29605)

  Show description for 1e9127f
  zhangting2020

  authored
  1e9127f

• add static.amp into setup.pu.in (#29621)

  Show description for b96dada
  huangxu96

  authored
  b96dada

• 添加rocm平台支持代码 (#29342)

  Show description for 7673850
  XuanBaby

  authored
  7673850

• [Kunlun] PR1:Support one Kunlun card training in parallel executor (#29337)

  vslyu

  authored
  f13c3a9

• add pad and concat double grad (#29549)

  Show description for cc38715
  ceci3

  authored
  cc38715

• update activation op on kunlun (#29577)

  Show description for af8ded7
  tangzhiyi11

  authored
  af8ded7

Commits on Dec 17, 2020

• [GO] add two cgo api, test=develop (#29659)

  qili93

  authored
  7684b91

• Add approval monitor for unity_build_rule.cmake (#29701)

  Show description for bb5a785
  Avin0323

  authored
  bb5a785

• delete the code for fp16 optimization because it is not faster than common template code (#29715)

  wangchaochaohu

  authored
  2e0d1ed

• Update EarlyStopping sample code (#29723)

  Show description for 572810e
  LiuChiachi

  authored
  572810e

• Added missing format of oneDNN (#29670)

  jczaja

  authored
  9eff1a6

• [Complex] Add real & imag op and api for complex tensor (#29672)

  Show description for 6cfa59d
  chenwhql

  authored
  6cfa59d

• Modify CublasHandleHolder to Fix Random Unittest Failure. test=develop (#29617)

  Show description for 4c4d4ba
  zhhsplendid

  authored
  4c4d4ba

• Windows generate pdb and dump, for debug (#29628)

  Show description for 0c59ad2
  wanghuancoder

  authored
  0c59ad2

• fix ubuntu docker error (#29719)

  tianshuo78520a

  authored
  638ccaa

• fleet sync build strategy, test=develop (#29732)

  wangxicoding

  authored
  9cbcc6c

• [Inference] EnableUseGpu has higher priority than flags (#29697)

  Show description for b593d58
  jiweibo

  authored
  b593d58

• add conj op for complex types (#29527)

  Show description for 71063b8
  MingMingShangTian

  authored
  71063b8

• Update en docs of to_tensor (#29718)

  Show description for 10edfb6
  LielinJiang

  authored
  10edfb6

Commits on Dec 18, 2020

• update the operator registration for incompatible upgrade, test=develop (#29720)

  Shixiaowei02

  authored
  8bd2879

• Reduce batch size ot fix CPU memory, test=develop (#29736)

  Show description for 2e788bd
  zhhsplendid

  authored
  2e788bd