2017 10 11

nickyfantasy

• Created a repo for Paddle/board and updated README.md to define purpose and proposal https://github.com/PaddlePaddle/board/blob/master/README.md
• Created a Wiki Doc for TensorFlow and Tensorboard overview https://github.com/PaddlePaddle/board/wiki/TensorFlow-basics-and-TensorBoard-Overview
• Initial commit for pdf slide, wiki assets and demo scripts for TensorFlow and TensorBoard https://github.com/PaddlePaddle/board/tree/master/wiki
• Created a Home Wiki Page for board and links to different docs https://github.com/PaddlePaddle/board/wiki
• Created a Project Info Page to define Project Goal, Structure and Schedule https://github.com/PaddlePaddle/board/wiki/Paddle-Board-Project-Info
• Created a Weekly Project Progress Report to keep track of status of the project and tasks https://github.com/PaddlePaddle/board/wiki/Weekly-Project-Progress-Report

wangkuiyi

• Demo development procedure for American developers
• Compile-time InferShape https://github.com/PaddlePaddle/Paddle/pull/4569#pullrequestreview-67743926
• Update IfElse operator design https://github.com/PaddlePaddle/Paddle/pull/4619
• Update ProgramDesc design https://github.com/PaddlePaddle/Paddle/pull/4618
• SelectedRows design: https://github.com/PaddlePaddle/Paddle/pull/4652#pullrequestreview-68159337
• The Python class Optimizer: https://github.com/PaddlePaddle/Paddle/pull/4656#pullrequestreview-68385726
• Executor and design: https://github.com/PaddlePaddle/Paddle/pull/4537#pullrequestreview-68480097

mkliegl

• Port conv shift operator https://github.com/PaddlePaddle/Paddle/pull/4591
• review and fix grammar in LoD tensor design doc: https://github.com/PaddlePaddle/Paddle/pull/4579
• review tensor array design doc: https://github.com/PaddlePaddle/Paddle/pull/4564

kavyasrinet

• Implement HardShrink operator https://github.com/PaddlePaddle/Paddle/pull/4653
• Implement RMSProp https://github.com/PaddlePaddle/Paddle/pull/4565
• Implement LeakyReLu activation https://github.com/PaddlePaddle/Paddle/pull/4604
• Implement Relu6 activation https://github.com/PaddlePaddle/Paddle/pull/4607
• Implement TanhShrink operator https://github.com/PaddlePaddle/Paddle/pull/4609
• Update Block.md design doc https://github.com/PaddlePaddle/Paddle/pull/4681
• Research on other frameworks about the activations/ operators that we are missing.

kexinzhao

• Farm:
  • Make unit tests fail on errors https://github.com/PaddlePaddle/farm/pull/4
  • Set up Travis CI for farm repo https://github.com/PaddlePaddle/farm/pull/5
• Paddle:
  • Implement adagrad operator https://github.com/PaddlePaddle/Paddle/pull/4558
  • Update new_op_cn/en.md design doc https://github.com/PaddlePaddle/Paddle/pull/4549
• Review:
  • https://github.com/PaddlePaddle/Paddle/pull/4538#pullrequestreview-67764487

helinwang

• Reviews: Paddle cloud web features design: https://github.com/PaddlePaddle/cloud/pull/378#pullrequestreview-68697101

product-team (Sunnyvale)

• Moved and created project related information to GitHub wiki for PaddlePaddle.org (https://github.com/PaddlePaddle/PaddlePaddle.org/wiki)
• Created Paddle Board project and all documents (https://github.com/PaddlePaddle/board/wiki)

cs2be (Thuan)

• Discussion on content structure https://github.com/PaddlePaddle/PaddlePaddle.org/issues/113
• Add code to generate full site map from config https://github.com/PaddlePaddle/PaddlePaddle.org/pull/90
• Add in command to update sitemap https://github.com/PaddlePaddle/PaddlePaddle.org/pull/94
• Change internationalization persistence to use session and cookie instead of path https://github.com/PaddlePaddle/PaddlePaddle.org/pull/96
• Fixed URL linking for home page, update to make header fixed https://github.com/PaddlePaddle/PaddlePaddle.org/pull/111
• Update app to support new sitemap format. Update URLs to allow linking on documentation, book, models, blog. Tie in chinese language and versioning feature https://github.com/PaddlePaddle/PaddlePaddle.org/pull/122
• Cache generated sitemap into a file. Consolidated a few methods. Cleaned up code https://github.com/PaddlePaddle/PaddlePaddle.org/pull/125

jetfuel (Jeff Wang)

• Version Selection Drop down menu: https://github.com/PaddlePaddle/PaddlePaddle.org/pull/95
• Support reading different languages from sitemap: https://github.com/PaddlePaddle/PaddlePaddle.org/pull/99
• Version Selection Drop down menu UI update: https://github.com/PaddlePaddle/PaddlePaddle.org/pull/108
• Helper script to pull the models and book content branch: https://github.com/PaddlePaddle/PaddlePaddle.org/pull/119
• Update script to pull documentations and blog: https://github.com/PaddlePaddle/PaddlePaddle.org/pull/126

varunarora

• Did initial draft of remapping URL paths for documentation pages based on navigational design (later scrapped) https://github.com/PaddlePaddle/PaddlePaddle.org/issues/70
• Create Django command for building documentation without running the webserver https://github.com/PaddlePaddle/PaddlePaddle.org/issues/102
• Wrote about unification of information architecture https://github.com/PaddlePaddle/PaddlePaddle.org/issues/58
• Did v0.1 of unifying styling between book, models, and doc, handed off to @cs2be https://github.com/PaddlePaddle/PaddlePaddle.org/issues/59, https://github.com/PaddlePaddle/PaddlePaddle.org/tree/transform-deploy-documentation
• Transformed book markdown, generated book HTML, and generated doc documentation into stripped contents that be "partially" embedded in app https://github.com/PaddlePaddle/PaddlePaddle.org/issues/101
• Came up with a clear plan on deploying different versions of documentation through Travis CI

gongweibao

• Review:
  • https://github.com/PaddlePaddle/Paddle/issues/3675
  • https://github.com/PaddlePaddle/Paddle/pull/4394
  • https://github.com/PaddlePaddle/Paddle/issues/4156
  • https://github.com/PaddlePaddle/cloud/issues/379
  • https://github.com/PaddlePaddle/Paddle/pull/4063
  • https://github.com/PaddlePaddle/cloud/pull/380
• PaddleCloud:
  • Test deploy paddlecloud on minicube.
• Doing
  • BlockExpand operator: https://github.com/PaddlePaddle/Paddle/compare/develop...gongweibao:blockexpand

tonyyang-svail

• Paddle:
  • Executor PR: https://github.com/PaddlePaddle/Paddle/pull/4537
  • Executor Design Doc: https://github.com/PaddlePaddle/Paddle/pull/4649
  • FeedOp and FetchOp Desgin Doc: https://github.com/PaddlePaddle/Paddle/pull/4599
  • Issue:
    • https://github.com/PaddlePaddle/Paddle/pull/4403
    • https://github.com/PaddlePaddle/Paddle/issues/4377
    • https://github.com/PaddlePaddle/Paddle/issues/4481

qiaolongfei

refactoring

• Add compile time infershape
  • https://github.com/PaddlePaddle/Paddle/pull/4569
• use EigenVector to get learning_rate for GPU device in SGD operator
  • https://github.com/PaddlePaddle/Paddle/pull/4598
• Merge infershapecontext and ExecutionContext
  • https://github.com/PaddlePaddle/Paddle/pull/4630
• Optimizer Design and Implementation
  • https://github.com/PaddlePaddle/Paddle/issues/4679
• Optimizer Design
  • https://github.com/PaddlePaddle/Paddle/pull/4656
• Impl optimizer
  • https://github.com/PaddlePaddle/Paddle/pull/4734

code review

• dynamic recurrent op forward c++ implentation
  • https://github.com/PaddlePaddle/Paddle/pull/4597
• Add persistable in framework.proto
  • https://github.com/PaddlePaddle/Paddle/pull/4600
• fix tensorarray unpack bug
  • https://github.com/PaddlePaddle/Paddle/pull/4614
• Feature/tensor array add python binding
  • https://github.com/PaddlePaddle/Paddle/pull/4616
• Design doc of SelectedRows
  • https://github.com/PaddlePaddle/Paddle/pull/4652
• Polish CompileTime InferShape
  • https://github.com/PaddlePaddle/Paddle/pull/4660
• Prune Design Doc
  • https://github.com/PaddlePaddle/Paddle/pull/4732

YuYang/yuyang

• Fix CI failing
• Sparse Design
  • https://github.com/PaddlePaddle/Paddle/pull/4652
• Python API design doc and implementaiton
  • Done:
    • Using Python API create compile-time Op
    • https://github.com/PaddlePaddle/Paddle/pull/4641
    • https://github.com/PaddlePaddle/Paddle/pull/4654
    • https://github.com/PaddlePaddle/Paddle/pull/4682
    • https://github.com/PaddlePaddle/Paddle/pull/4684
  • WIP:
    • Using Python API create layer
    • https://github.com/PaddlePaddle/Paddle/pull/4698
    • https://github.com/PaddlePaddle/Paddle/pull/4724
    • https://github.com/PaddlePaddle/Paddle/pull/4726
• Add persistable field in VarDesc
  • https://github.com/PaddlePaddle/Paddle/pull/4600
• BugFixes
  • https://github.com/PaddlePaddle/Paddle/pull/4644
  • https://github.com/PaddlePaddle/Paddle/pull/4646
  • https://github.com/PaddlePaddle/Paddle/pull/4667
• Enhancement
  • CompileTime InferShape
    • https://github.com/PaddlePaddle/Paddle/pull/4660
  • Use Protolite
    • https://github.com/PaddlePaddle/Paddle/pull/4648

nickyfantasy

• Created a repo for Paddle/board and updated README.md to define purpose and proposal https://github.com/PaddlePaddle/board/blob/master/README.md
• Created a Wiki Doc for TensorFlow and Tensorboard overview https://github.com/PaddlePaddle/board/wiki/TensorFlow-basics-and-TensorBoard-Overview
• Initial commit for pdf slide, wiki assets and demo scripts for TensorFlow and TensorBoard https://github.com/PaddlePaddle/board/tree/master/wiki
• Created a Home Wiki Page for board and links to different docs https://github.com/PaddlePaddle/board/wiki
• Created a Project Info Page to define Project Goal, Structure and Schedule https://github.com/PaddlePaddle/board/wiki/Paddle-Board-Project-Info
• Created a Weekly Project Progress Report to keep track of status of the project and tasks https://github.com/PaddlePaddle/board/wiki/Weekly-Project-Progress-Report

wangkuiyi

• Demo development procedure for American developers
• Compile-time InferShape https://github.com/PaddlePaddle/Paddle/pull/4569#pullrequestreview-67743926
• Update IfElse operator design https://github.com/PaddlePaddle/Paddle/pull/4619
• Update ProgramDesc design https://github.com/PaddlePaddle/Paddle/pull/4618
• SelectedRows design: https://github.com/PaddlePaddle/Paddle/pull/4652#pullrequestreview-68159337
• The Python class Optimizer: https://github.com/PaddlePaddle/Paddle/pull/4656#pullrequestreview-68385726
• Executor and design: https://github.com/PaddlePaddle/Paddle/pull/4537#pullrequestreview-68480097

mkliegl

• Port conv shift operator https://github.com/PaddlePaddle/Paddle/pull/4591
• review and fix grammar in LoD tensor design doc: https://github.com/PaddlePaddle/Paddle/pull/4579
• review tensor array design doc: https://github.com/PaddlePaddle/Paddle/pull/4564

srinetkavya

• Implement HardShrink operator https://github.com/PaddlePaddle/Paddle/pull/4653
• Implement RMSProp https://github.com/PaddlePaddle/Paddle/pull/4565
• Implement LeakyReLu activation https://github.com/PaddlePaddle/Paddle/pull/4604
• Implement Relu6 activation https://github.com/PaddlePaddle/Paddle/pull/4607
• Implement TanhShrink operator https://github.com/PaddlePaddle/Paddle/pull/4609
• Update Block.md design doc https://github.com/PaddlePaddle/Paddle/pull/4681
• Research on other frameworks about the activations/ operators that we are missing.

helinwang

• Reviews: Paddle cloud web features design: https://github.com/PaddlePaddle/cloud/pull/378#pullrequestreview-68697101

product-team (Sunnyvale)

• Moved and created project related information to GitHub wiki for PaddlePaddle.org (https://github.com/PaddlePaddle/PaddlePaddle.org/wiki)
• Created Paddle Board project and all documents (https://github.com/PaddlePaddle/board/wiki)

cs2be (Thuan)

• Discussion on content structure https://github.com/PaddlePaddle/PaddlePaddle.org/issues/113
• Add code to generate full site map from config https://github.com/PaddlePaddle/PaddlePaddle.org/pull/90
• Add in command to update sitemap https://github.com/PaddlePaddle/PaddlePaddle.org/pull/94
• Change internationalization persistence to use session and cookie instead of path https://github.com/PaddlePaddle/PaddlePaddle.org/pull/96
• Fixed URL linking for home page, update to make header fixed https://github.com/PaddlePaddle/PaddlePaddle.org/pull/111
• Update app to support new sitemap format. Update URLs to allow linking on documentation, book, models, blog. Tie in chinese language and versioning feature https://github.com/PaddlePaddle/PaddlePaddle.org/pull/122
• Cache generated sitemap into a file. Consolidated a few methods. Cleaned up code https://github.com/PaddlePaddle/PaddlePaddle.org/pull/125

jetfuel (Jeff Wang)

• Version Selection Drop down menu: https://github.com/PaddlePaddle/PaddlePaddle.org/pull/95
• Support reading different languages from sitemap: https://github.com/PaddlePaddle/PaddlePaddle.org/pull/99
• Version Selection Drop down menu UI update: https://github.com/PaddlePaddle/PaddlePaddle.org/pull/108
• Helper script to pull the models and book content branch: https://github.com/PaddlePaddle/PaddlePaddle.org/pull/119
• Update script to pull documentations and blog: https://github.com/PaddlePaddle/PaddlePaddle.org/pull/126

varunarora

• Did initial draft of remapping URL paths for documentation pages based on navigational design (later scrapped) https://github.com/PaddlePaddle/PaddlePaddle.org/issues/70
• Create Django command for building documentation without running the webserver https://github.com/PaddlePaddle/PaddlePaddle.org/issues/102
• Wrote about unification of information architecture https://github.com/PaddlePaddle/PaddlePaddle.org/issues/58
• Did v0.1 of unifying styling between book, models, and doc, handed off to @cs2be https://github.com/PaddlePaddle/PaddlePaddle.org/issues/59, https://github.com/PaddlePaddle/PaddlePaddle.org/tree/transform-deploy-documentation
• Transformed book markdown, generated book HTML, and generated doc documentation into stripped contents that be "partially" embedded in app https://github.com/PaddlePaddle/PaddlePaddle.org/issues/101
• Came up with a clear plan on deploying different versions of documentation through Travis CI

gongweibao

• Review:
  • https://github.com/PaddlePaddle/Paddle/issues/3675
  • https://github.com/PaddlePaddle/Paddle/pull/4394
  • https://github.com/PaddlePaddle/Paddle/issues/4156
  • https://github.com/PaddlePaddle/cloud/issues/379
  • https://github.com/PaddlePaddle/Paddle/pull/4063
  • https://github.com/PaddlePaddle/cloud/pull/380
• PaddleCloud:
  • Test deploy paddlecloud on minicube.
• Doing
  • BlockExpand operator: https://github.com/PaddlePaddle/Paddle/compare/develop...gongweibao:blockexpand

tonyyang-svail

• Paddle:
  • Executor PR: https://github.com/PaddlePaddle/Paddle/pull/4537
  • Executor Design Doc: https://github.com/PaddlePaddle/Paddle/pull/4649
  • FeedOp and FetchOp Desgin Doc: https://github.com/PaddlePaddle/Paddle/pull/4599
  • Issue:
    • https://github.com/PaddlePaddle/Paddle/pull/4403
    • https://github.com/PaddlePaddle/Paddle/issues/4377
    • https://github.com/PaddlePaddle/Paddle/issues/4481

Caoying

• PR review:
  • https://github.com/PaddlePaddle/Paddle/pull/4394#pullrequestreview-66048573
  • https://github.com/PaddlePaddle/Paddle/pull/4401#pullrequestreview-66304572
  • https://github.com/PaddlePaddle/Paddle/pull/4508#pullrequestreview-66305666
  • https://github.com/PaddlePaddle/Paddle/pull/4508#pullrequestreview-66306566
  • https://github.com/PaddlePaddle/Paddle/pull/4536#pullrequestreview-66305149
  • https://github.com/PaddlePaddle/Paddle/pull/4394#pullrequestreview-66048573
  • https://github.com/PaddlePaddle/Paddle/pull/4508#pullrequestreview-67887695
  • https://github.com/PaddlePaddle/Paddle/pull/4713#pullrequestreview-68601158
• PR:
  • bug fix of printing gradients: https://github.com/PaddlePaddle/models/pull/356
  • add configuration helper for resize layer: https://github.com/PaddlePaddle/Paddle/pull/4541
• WIP: the CRF opeator
  • https://github.com/PaddlePaddle/Paddle/compare/develop...lcy-seso:crf

Zhuoyuan Chen

• Check-in operators:
  • Gather, Scatter with cuda streams (building blocks for cond_op, while_op) https://github.com/PaddlePaddle/Paddle/pull/4483
• Working on operators:
  • Deconv https://github.com/PaddlePaddle/Paddle/pull/4739
• System design Doc: https://github.com/PaddlePaddle/Paddle/pull/4580 GAN design, provide solutions with two options: (1) dependency back-trace, like tf and torch; (2) block;
• Review op:
  • thresholded relu (with Arora Abhinav) https://github.com/PaddlePaddle/Paddle/pull/4685

Peng Li

• PR
  • Fixing random seeds in gradient checkers: https://github.com/PaddlePaddle/Paddle/pull/4498
  • Code review: https://github.com/PaddlePaddle/Paddle/pull/4479
• Model
  • Implement QA model: https://arxiv.org/abs/1607.06275

Yan Xu

• PR: Sequence Concat Op,https://github.com/PaddlePaddle/Paddle/pull/4508
• Deploy and verify PaddleCloud on the internal CPU cluster.
• Review: auto-scaling on cloud, https://github.com/PaddlePaddle/cloud/pull/380.
• Review: cloud web design doc, https://github.com/PaddlePaddle/cloud/pull/378#pullrequestreview-68806478
• Discuess about kubectl and paddlectl: https://github.com/PaddlePaddle/cloud/issues/383

qijun

Executor:

• Executor interface design and implementation
  • https://github.com/PaddlePaddle/Paddle/pull/4537

SelectedRows:

• [WIP] Implementing SelectedRows
  • https://github.com/PaddlePaddle/Paddle/pull/4746

Fix and Enhance:

• Unify CUDA stream in Tensor CopyFrom interface
  • https://github.com/PaddlePaddle/Paddle/pull/4692
• fix conflict of CopyFrom in vol2ol_test
  • https://github.com/PaddlePaddle/Paddle/pull/4728
• Debug executor test
  • https://github.com/PaddlePaddle/Paddle/pull/4721

hedaoyuan

• Mobile
  • Optimize the binary size of the Paddle program(From 4212KB to 1869KB).
  • https://github.com/PaddlePaddle/Paddle/pull/4509
  • https://github.com/PaddlePaddle/Paddle/issues/4678
  • https://github.com/PaddlePaddle/Mobile/pull/11
  • Review
    • https://github.com/PaddlePaddle/Paddle/pull/4638

luotao

• op porting:
  • remove rowwise_add_op: https://github.com/PaddlePaddle/Paddle/pull/4503
  • refine "Testing Backward Operators" in new_op_cn/en.md: https://github.com/PaddlePaddle/Paddle/pull/4505
  • Unify Reduce functions and simplify register code: https://github.com/PaddlePaddle/Paddle/pull/4632
  • add interpolation op: https://github.com/PaddlePaddle/Paddle/pull/4500
  • [WIP] add sqrt/last/first/max for seqpool op.
• fix compiler bug:
  • fix compiler error: ‘unique_ptr’ is not a member of ‘std’: https://github.com/PaddlePaddle/Paddle/pull/4658
  • fix Compile error [fatal error: boost/range/adaptor/reversed.hpp]: https://github.com/PaddlePaddle/Paddle/pull/4707
  • fix executor_test build failed: https://github.com/PaddlePaddle/Paddle/pull/4716
• code review:
  • MKLDNN:should reset env every time: https://github.com/PaddlePaddle/Paddle/pull/4441
• Disscuss the website:
  • Separate doc from Paddle repo and make it an independent compiling process: https://github.com/PaddlePaddle/PaddlePaddle.org/issues/18
  • Separate or remove doc_theme from Paddle repo: https://github.com/PaddlePaddle/PaddlePaddle.org/issues/83
  • shorten the doc url: https://github.com/PaddlePaddle/PaddlePaddle.org/issues/84
  • Install and demo in homepage: https://github.com/PaddlePaddle/PaddlePaddle.org/issues/123

wanghaoshuang

• op porting
  • Add sequence expand op:
    • https://github.com/PaddlePaddle/Paddle/pull/4740

zhouxiao

PR:

• Done: merged elu activation https://github.com/PaddlePaddle/Paddle/pull/4395
• Doing: layer normalization https://github.com/PaddlePaddle/Paddle/issues/4701 Review:
• https://github.com/PaddlePaddle/Paddle/pull/4674
• https://github.com/PaddlePaddle/Paddle/issues/4683

Liu Yiqun

• Porting operators
  • [Developing] warpctc operator
    • It is almost finished except the unit test
  • [Done] Simplify backward implementation when inserting a sum operator to accumulate all duplicated variables
    • https://github.com/PaddlePaddle/Paddle/pull/4703
  • [Review]
    • interpolation op: https://github.com/PaddlePaddle/Paddle/pull/4500
• Arm deployment
  • [Review]
    • Build mobile inference library for minimum size: https://github.com/PaddlePaddle/Paddle/pull/4509

Chen xi

• Paddle cloud master, pserver and trainer testing tool https://github.com/putcn/kubernetes-vagrant-coreos-cluster/tree/setup-paddle-testing/paddle-setup

ranqiu

• Add dot-product attention to networks.py

  https://github.com/PaddlePaddle/Paddle/pull/4674

• Update annotations of layers.py

  https://github.com/PaddlePaddle/Paddle/pull/4713

• convS2S (Doing)

  https://github.com/ranqiu92/models/tree/convs2s_param_initialization/conv_seq_to_seq

chengxiaohua1105

• fix a bug for adagrad optimizer algorithm
  • https://github.com/PaddlePaddle/Paddle/pull/4743
• communicate demands for paddle with other teams

yangyaming

• Tune decoder parameters for Aishell
  https://github.com/PaddlePaddle/models/issues/316
• Train iQiyi data
  • data preparation, parameter tuning
• Validate performance when enable rnn_use_batch
• Add document for Mandarin example and LM
  https://github.com/PaddlePaddle/models/pull/364
  https://github.com/PaddlePaddle/models/pull/368
• Bug fix for librispeech example
  https://github.com/PaddlePaddle/models/pull/360
• Code review:
  • https://github.com/PaddlePaddle/models/pull/363
  • https://github.com/PaddlePaddle/models/pull/359
  • https://github.com/PaddlePaddle/models/pull/355

fengjiayi

• Complete porting backward to OpDesc

  • https://github.com/PaddlePaddle/Paddle/pull/4573

• Complete low level Python API development (Variable, Op, Block and Program):

  • https://github.com/PaddlePaddle/Paddle/pull/4702
  • https://github.com/PaddlePaddle/Paddle/pull/4702

• bug fix：

  • https://github.com/PaddlePaddle/Paddle/pull/4582
  • https://github.com/PaddlePaddle/Paddle/pull/4640
  • https://github.com/PaddlePaddle/Paddle/pull/4651
  • https://github.com/PaddlePaddle/Paddle/pull/4664

• code review:

  • compile time gradient op registry:

    • https://github.com/PaddlePaddle/Paddle/pull/4551
    • https://github.com/PaddlePaddle/Paddle/pull/4566

  • Python layer design:

    • https://github.com/PaddlePaddle/Paddle/pull/4724
    • https://github.com/PaddlePaddle/Paddle/pull/4726

Yibing Liu

Deep Speech 2:

• Fix the core dump bug of DS2's training in docker
  • https://github.com/PaddlePaddle/models/pull/350
• Convert decoding results to unicode in DS2
  • https://github.com/PaddlePaddle/models/pull/347

Op Porting:

• [merged] Margin Rank Loss Operator
  • https://github.com/PaddlePaddle/Paddle/pull/4285
• [under review] LoD Reset Operator
  • https://github.com/PaddlePaddle/Paddle/pull/4747
• [WIP] CTC Edit Distance Evaluator

Dang Qingqing

• Op Porting:
  • https://github.com/PaddlePaddle/Paddle/projects/22
  • https://github.com/PaddlePaddle/Paddle/pulls?q=is%3Apr+is%3Aopen+label%3AOpPorting
• LSTM Operator:
  • WIP: https://github.com/PaddlePaddle/Paddle/compare/develop...qingqing01:lstm?expand=1
  • Issue:
    • https://github.com/PaddlePaddle/Paddle/issues/4629
    • https://github.com/PaddlePaddle/Paddle/issues/4675
• Code Review:
  • GRU Unit: https://github.com/PaddlePaddle/Paddle/pull/4443#pullrequestreview-67883978
  • CudnnConv: https://github.com/PaddlePaddle/Paddle/pull/4195#pullrequestreview-68261154
  • vol2col: https://github.com/PaddlePaddle/Paddle/pull/4462
  • Sequence Concat Op: https://github.com/PaddlePaddle/Paddle/pull/4508#pullrequestreview-68268306
  • Pool Op coc and class declarations: https://github.com/PaddlePaddle/Paddle/pull/4676#pullrequestreview-68499562

Yan Chunwei

• PR
  • open, feature/dynamic recurrent op forward test
  • merged, dynamic recurrent op forward c++ implentation
  • merged, fix tensorarray unpack bug
  • merged, add background for tensor array
  • merged, feature/tensor array add python binding
  • merged, tensorarray implementation
• review
  • https://github.com/PaddlePaddle/Paddle/pull/4725#issuecomment-335916698
  • https://github.com/PaddlePaddle/Paddle/pull/4553#pullrequestreview-66587384
  • https://github.com/PaddlePaddle/Paddle/pull/4580#pullrequestreview-66918376
  • https://github.com/PaddlePaddle/Paddle/pull/4569#pullrequestreview-67537544
  • https://github.com/PaddlePaddle/Paddle/pull/4580#pullrequestreview-67812582
  • https://github.com/PaddlePaddle/Paddle/pull/4537#pullrequestreview-68106782
  • https://github.com/PaddlePaddle/Paddle/pull/4602#pullrequestreview-68141951

ChengduoZhao

• WIP: the conv3d_gemm_op, pool_cudnn_op, finetune conv2d naive function
  • Add conv3d_gemm_op: https://github.com/PaddlePaddle/Paddle/pull/4709
  • Add pool cudnn op: https://github.com/PaddlePaddle/Paddle/pull/4636
    • Add cudnn pool2d maxpool and avgpool.
    • Add cudnn pool2d unit test.
  • finetune conv2d naive function: https://github.com/PaddlePaddle/Paddle/pull/4742
• PR merge:
  • Add pool op: https://github.com/PaddlePaddle/Paddle/pull/4146
    • Add pooling2d max and average (gpu and cpu)
    • Add pooling3d max and average (gpu and cpu)
    • Add unit test pooling2d and pooling3d
  • Add max pool op (with index): https://github.com/PaddlePaddle/Paddle/pull/4461
  • Add vol2col functor: https://github.com/PaddlePaddle/Paddle/pull/4462
  • Fix pool op doc and class declarations: https://github.com/PaddlePaddle/Paddle/pull/4676
• PR review:
  • Cudnn conv op: https://github.com/PaddlePaddle/Paddle/pull/4195

dongzhihong

• Refactorization:

  • Serialize LoDTensor, Save/Restore model
    • https://github.com/PaddlePaddle/Paddle/pull/4602
  • Benchmark of different serialization performance
    • https://github.com/PaddlePaddle/Paddle/pull/4610
  • change NewVar to Var
    • https://github.com/PaddlePaddle/Paddle/pull/4725

• Reviews

  • Implement Softplus Operator
    • https://github.com/PaddlePaddle/Paddle/pull/4690
  • Implement hardShrink Operator
    • https://github.com/PaddlePaddle/Paddle/pull/4653
  • Add logsigmoid (numerically stable) and softshrink
    • https://github.com/PaddlePaddle/Paddle/pull/4663
  • Adding relu6 activation functio
    • https://github.com/PaddlePaddle/Paddle/pull/4607
  • Added Leaky Relu activation
    • https://github.com/PaddlePaddle/Paddle/pull/4604
  • Conv Shift Operator
    • https://github.com/PaddlePaddle/Paddle/pull/4591
  • Add -D PADDLE_WITH_CUDA in cmake
    • https://github.com/PaddlePaddle/Paddle/pull/4590
  • Optimizer Design
    • https://github.com/PaddlePaddle/Paddle/pull/4656

Xinghai Sun

• DS2
  • Improve data pipeline for .seqbin format: 4x speedup (72hour --> 18 hour)
    • https://github.com/PaddlePaddle/models/pull/355
    • https://github.com/PaddlePaddle/models/pull/359
  • Others:
    • https://github.com/PaddlePaddle/models/pull/363
    • https://github.com/PaddlePaddle/models/pull/364#pullrequestreview-68813316
  • Debugging for the abrupt training divergence with English1W dataset.

xingzhaolong

• Pr

  • https://github.com/PaddlePaddle/Paddle/pull/4473 refine paddle_merge_model (in process)
  • use protobuf_lite during inference. The striped size of libpaddle_capi_shared.so now is 1.8M. https://github.com/PaddlePaddle/Paddle/pull/4638 (merged)

• Reviews:

  • https://github.com/PaddlePaddle/Paddle/pull/4509 Build mobile inference library for minimum size
  • https://github.com/PaddlePaddle/Paddle/pull/4707 fix compile bug
  • https://github.com/PaddlePaddle/Mobile/pull/11 refine paddlepaddle/Mobile repo

• Test performance of ENet of Paddle on mobile (in process)

Zhang Chao

• CAPI
  • add Subsequence example for capi
  • https://github.com/PaddlePaddle/models/pull/346
• Models
  • add nest text classification for paddle.v2
  • https://github.com/PaddlePaddle/models/pull/367

typhoonzero(wuyi)

• op porting: cudnn conv op refine
• paddle cloud web design doc
• paddle cloud autoscaling: kubernetes controller develop: https://github.com/PaddlePaddle/cloud/tree/controller
• paddle installation doc update
• pserver listen queue size bug fix
• reviews:
  • https://github.com/PaddlePaddle/Paddle/pull/4709
  • https://github.com/PaddlePaddle/Paddle/pull/4602
  • https://github.com/PaddlePaddle/Paddle/pull/4461

Hu Weiwei(wwhu)

Porting clip_by_norm operator (in progress) https://github.com/PaddlePaddle/Paddle/issues/4662

guosheng

• OP Port:
  • Add NormOp: https://github.com/PaddlePaddle/Paddle/pull/4496
  • Work on ChunkEvalOp with issue for help: https://github.com/PaddlePaddle/Paddle/issues/4700
  • Refine GRUUnitOp
• Review:
  • https://github.com/PaddlePaddle/Paddle/pull/4632#pullrequestreview-67938740
  • https://github.com/PaddlePaddle/Paddle/pull/4716#pullrequestreview-68595080
  • https://github.com/PaddlePaddle/Paddle/pull/4721#pullrequestreview-68802078

wangmeng

• Model:
  • Add preprocess and featurization scripts in globally normalized reader model (merged)
    • https://github.com/PaddlePaddle/models/pull/353
• New layer:
  • Add factorization machine layer (ongoing)
    • https://github.com/will-am/Paddle/tree/factorization_machine_layer

sidgoyal78 (Siddharth Goyal)

• Add momentum optimizer: https://github.com/PaddlePaddle/Paddle/pull/4571
• Add basic activations: https://github.com/PaddlePaddle/Paddle/pull/4663
  • log-sigmoid
  • softshrink

daming-lu

• Created a Wiki Doc for Version Branches of 'book' and 'model' https://github.com/PaddlePaddle/PaddlePaddle.org/wiki/Documentation-Versioning-Proposal
• Created a Slides for versioning.
• Add auto deployment for 'book' and model'
• Created version branches in 'book' and 'model' dev repos.

abhinavarora

Operator Porting

• Add Adamax Operator: https://github.com/PaddlePaddle/Paddle/pull/4538
• Add Adadelta Operator: https://github.com/PaddlePaddle/Paddle/pull/4576
• Add Adam Operator: https://github.com/PaddlePaddle/Paddle/pull/4685
• Add ThresholdedRELU Operator: https://github.com/PaddlePaddle/Paddle/pull/4685

Executor (with Tony and Qi)

• Copy Vector to Tensor
• Fill constant operators

Documentation Refactoring

• Refactorization Document: https://github.com/PaddlePaddle/Paddle/pull/4680
• Python_API Document: https://github.com/PaddlePaddle/Paddle/pull/4680

Issues

• https://github.com/PaddlePaddle/Paddle/issues/4585
• https://github.com/PaddlePaddle/Paddle/issues/4577

Reviews

• Optimizer Design: https://github.com/PaddlePaddle/Paddle/pull/4656
• https://github.com/PaddlePaddle/Paddle/pull/4618
• https://github.com/PaddlePaddle/Paddle/pull/4598