-
Notifications
You must be signed in to change notification settings - Fork 5.6k
2017 11 08
Tao Luo edited this page Dec 9, 2019
·
1 revision
- Fix warnings https://github.com/PaddlePaddle/Paddle/pull/5359#pullrequestreview-74220912
- Too slow unit test https://github.com/paddlepaddle/paddle/issues/5342
- Compilation warnings https://github.com/paddlepaddle/paddle/issues/5341
- Compilation warnings https://github.com/paddlepaddle/paddle/issues/5340
- Compilation warnings https://github.com/paddlepaddle/paddle/issues/5339
- Compilation warnings https://github.com/paddlepaddle/paddle/issues/5338
- test_CustomStackTrace takes too much time https://github.com/paddlepaddle/paddle/issues/5330
- test_LayerGrad takes too much time to run https://github.com/paddlepaddle/paddle/issues/5329
- C/C++ compilation warnings about comparing signed and unsigned integers https://github.com/paddlepaddle/paddle/issues/5322
- {init_program,program} => {startup_program,main_program} https://github.com/paddlepaddle/paddle/issues/5319
- Pip install support https://github.com/paddlepaddle/paddle/issues/5318
- Unused parameters in framework.layers.py https://github.com/paddlepaddle/paddle/issues/5316
- Rename Python package layers into layer https://github.com/paddlepaddle/paddle/issues/5315
- Polish C++ operators documentation https://github.com/paddlepaddle/paddle/issues/5265
- Issues related with buildtools:
-
Edit documentation for all operators
-
Write documentation for layers
-
Design documentation:
- Background for multi-device: https://github.com/PaddlePaddle/Paddle/pull/5284
- Explaining how multi-device works on Tensorflow: https://github.com/PaddlePaddle/Paddle/pull/5412
-
Edit documentation for C++ operators
- https://github.com/PaddlePaddle/Paddle/pull/5334
- https://github.com/PaddlePaddle/Paddle/pull/5347
- https://github.com/PaddlePaddle/Paddle/pull/5354
- https://github.com/PaddlePaddle/Paddle/pull/5357
- https://github.com/PaddlePaddle/Paddle/pull/5367
- https://github.com/PaddlePaddle/Paddle/pull/5375
- https://github.com/PaddlePaddle/Paddle/pull/5376
- https://github.com/PaddlePaddle/Paddle/pull/5377
-
Check that operators follow naming conventions
-
Review
- Mobile
- Build for minimum size https://github.com/PaddlePaddle/Mobile/pull/21
- Inference Acceleration with Mobile GPU https://github.com/PaddlePaddle/Paddle/issues/5469
- Review
- Define a customized simple protobuf format for PaddleBoard https://github.com/PaddlePaddle/board/pull/13
- Discussed with YongFeng and YiQun for a AICamera mobile demo (SSD Model) that will be shown in Baidu World https://github.com/PaddlePaddle/Mobile/issues/25
- Successfully compiled iOS and Android Paddle library https://github.com/PaddlePaddle/Paddle/blob/develop/doc/mobile/cross_compiling_for_ios_cn.md https://github.com/PaddlePaddle/Paddle/blob/develop/doc/mobile/cross_compiling_for_android_cn.md
- Initial check in iOS code with Paddle C++ code integration https://github.com/PaddlePaddle/Mobile/pull/26
- Contributed to building a development guide: https://github.com/PaddlePaddle/PaddlePaddle.org/pull/259
- Fixed quick start guide to make instructions clearer and for the code to reliably work with the current API: https://github.com/PaddlePaddle/PaddlePaddle.org/pull/260
- Built operators documentation and injected into documentation from JSON from
print_operators_doc: https://github.com/PaddlePaddle/PaddlePaddle.org/pull/257 - Helped with fixing formula rendering on Chinese pages: https://github.com/PaddlePaddle/PaddlePaddle.org/pull/263
- Completed pending UI/UX proposal document: https://github.com/PaddlePaddle/PaddlePaddle.org/wiki/User-Research-(UI-UX)
- Autoscaling experiment and report: https://github.com/PaddlePaddle/cloud/blob/develop/doc/autoscale/experiment/README.md
- Reviews:
- Fixes:
- Add Mobile documentation to "DOC_MODE" https://github.com/PaddlePaddle/PaddlePaddle.org/pull/250
- Update sitemap generator for Mobile to pull in li>a and add them to side menu https://github.com/PaddlePaddle/PaddlePaddle.org/pull/254
- Add development guide https://github.com/PaddlePaddle/PaddlePaddle.org/pull/259
- Add manage.py method to update operators documentation. https://github.com/PaddlePaddle/PaddlePaddle.org/pull/261
- Fix line height issues on side navigation when text is too long https://github.com/PaddlePaddle/PaddlePaddle.org/pull/262
- Add document view only mode https://github.com/PaddlePaddle/PaddlePaddle.org/pull/269
- Fixed mal-format math formulas in book :
- Added one-line deployment to documentation:
- Add cron job for PaddlePaddle.org's daily release (WIP)
Working on Move core job management codes to controller in go
- https://github.com/PaddlePaddle/cloud/issues/462
- code branch is here: https://github.com/gongweibao/cloud/commits/movetogo
-
Add convs2s to models
-
Update doc of layers.py
- discuss with Abhinav and longfei about beam search support
- update design of sequence decoder (beamsearch)
- beam search op
- PR
- fix bugs of softmax with cross entropy:
- CRF decoding:
- fix NCE:
- PR review
-
Refine bilinear tensor product op :
-
Help to transform resnet from tf to paddle.
-
[WIP] Refine doc for capi in models:
- PR
- https://github.com/PaddlePaddle/models/pull/415 : fix bug in document
- code review:
- Refine sequence max-pooling and add unit testing of gradient check.
- Check errors for the CUDA kernel calls.
- Enable hidden/cell state initialization and enhance unit testing in LSTM operator.
- Fix cmake error when building with WITH_AVX=OFF.
- Fix the doc for Momentum and Adam optimizer.
- [Speed up compiling]: reduce the NVCC compiling (some .cu operators can be compiled by G++)
- Code Review:
-
Op porting
- Add NCE op(WIP):
-
code review
- reduce elasped time of test_LayerGrad: https://github.com/PaddlePaddle/Paddle/pull/5478
- add doc for image.py: https://github.com/PaddlePaddle/Paddle/pull/5485
- code review:
- MKLDNN:
- add MKLDNNAddtoLayer: https://github.com/PaddlePaddle/Paddle/pull/5309
- enable benchmark resnet with MKLDNN：https://github.com/PaddlePaddle/Paddle/pull/5422
- enable bias for mkldnn_addto: https://github.com/PaddlePaddle/Paddle/pull/5440
- Op/compiler/doc/models/website:
- Enable hidden/cell state initialization and enhance unit testing in LSTM operator. https://github.com/PaddlePaddle/Paddle/pull/5429#pullrequestreview-75032423
- Put files to platlib：https://github.com/PaddlePaddle/Paddle/pull/5396
- Fix ci not exit 1 when error：https://github.com/PaddlePaddle/Paddle/pull/5431
- Fix the doc for Momentum and Adam optimizer：https://github.com/PaddlePaddle/Paddle/pull/5391
- Rename doc/howto/cross_compiling into doc/mobile：https://github.com/PaddlePaddle/Paddle/pull/5399
- rename the directory of neural sequence QA：https://github.com/PaddlePaddle/models/pull/425
- update the README for the Models repo：https://github.com/PaddlePaddle/models/pull/426
- uci_housing.py can download the trained model automatically: https://github.com/PaddlePaddle/Paddle/pull/5463
- Add quick start code: https://github.com/PaddlePaddle/book/pull/443
- MKLDNN:
Deep Speech 2:
- Add the document about docker running for DS2
- Fix a bug in CTC decoder
- Re-tune the decoder and update benchmark results for different models
Intel MKL_DNN:
- Carry out benchmark test on laptop and desktop computer
- DS2
- Remove padding
https://github.com/PaddlePaddle/models/issues/414
https://github.com/PaddlePaddle/Paddle/pull/5388
https://github.com/PaddlePaddle/Paddle/pull/5488 - Refine PR
https://github.com/PaddlePaddle/models/pull/396 - Docs
https://github.com/PaddlePaddle/models/pull/368
https://github.com/PaddlePaddle/models/pull/364
- Remove padding
- Paddle
- More robust memory handler
https://github.com/PaddlePaddle/Paddle/pull/5418 - Refine expand_op
https://github.com/PaddlePaddle/Paddle/pull/4061
- More robust memory handler
- autoscaler scheduler design:
- manylinux compile env and CI:
- Some op refines
- Send Recv op WIP.
- Auto-scaling experiment
- Op porting
- Fix seq concat op with refactoring LoD:https://github.com/PaddlePaddle/Padd
- OP Porting
- Refine GRU Operator(Merged):
- Refine ChunkEval Operator:
- Others
- Convert Model from tensorflow to PaddlePaddle
- porting clip_by_norm op (merged)
- https://github.com/PaddlePaddle/Paddle/pull/5296
- Op Porting
- Add conv3d_gemm_op. (merged)
- conv2d kernel and conv3d kernel are wrote together
- https://github.com/PaddlePaddle/Paddle/pull/4709
- Add 3D convolution transposed operator. (merged)
- deconv2d kernel and deconv3d kernel write together
- https://github.com/PaddlePaddle/Paddle/pull/5118
- Add conv3d_gemm_op. (merged)
- enhancement
- Add dilations for conv2d and optimize conv2d code
- Add 3D data type for cudnn (conv3d/deconv3d/pool3d)
- fix_pool_op_attr_name
- Mobile deployment
- [Merged] Rename doc/howto/cross_compiling into doc/mobile
- [Merged] Fix the lack of linking libraries to libpaddle_capi_engine
- [Merged] Enable the build for iOS simulator
- SSD Demo on Android and iOS
- issue: https://github.com/PaddlePaddle/Mobile/issues/25
- capi example on linux: https://github.com/Xreki/Mobile/tree/add_ssd_linux_demo/demo/linux/ssd
- Review
- Init commit for iOS AICamera demo: https://github.com/PaddlePaddle/Mobile/pull/26
PR:
- Add attribute check in operator constructor
- Forward: feature/while_op
- Refactor Static RNN unittest
- Update lod_tensor.md
- Static lstm sanity check
- Add fill_constant_batch_size_like_op to Static RNN's h_boot
review:
- Rewrite StaticRNN with Executor
- Remove fill_constant_batch_size_like_op.h
- Feature/shrink memory op
- Feature/rnn to array to lod tensor
- Add unittest, backward of array read/write op
- Add To Array operator for dynamic RNN
- ReadFromArray/WriteToArray op
- Compare Operator
- Feature/rank sort lod op
- Remove warning in executor
- Add net drawer for visualizing the graph
- Update the Chinese navigation bar text.
- Update translated Chinese strings.
- Consolidate the URL pattern for documentation, book, models, and mob…
- uci_housing.py can download the trained model automatically.
- Add quickstart code
- SSD iOS Demo
- PR for fixing attribute naming: https://github.com/PaddlePaddle/Paddle/pull/5453