提交 2091a734 编写于 作者: H hedaoyuan

Merge branch 'release/0.10.0' of https://github.com/baidu/Paddle into release/0.10.0

...@@ -2,10 +2,18 @@ ...@@ -2,10 +2,18 @@
## New Features ## New Features
* Support rectangle input for CNN.
* Support stride pooling for seqlastin and seqfirstin.
* Expose seq_concat_layer/seq_reshape_layer in `trainer_config_helpers`.
* Add the cpu implementation of cmrnorm-projection. * Add the cpu implementation of cmrnorm-projection.
## Improvements ## Improvements
* Speedup the backward of average layer on GPU.
* Reorganize the catalog of doc/ and refine several docs.
* Add Travis-CI for checking dead links.
* Add a example for explaining sparse_vector.
* Add Relu in layer_math.py
* Add a new `TensorExpression` implementation for matrix-related expression evaluations. * Add a new `TensorExpression` implementation for matrix-related expression evaluations.
* Add Lazy Assignment for optimize the calculation of multiple expressions. * Add Lazy Assignment for optimize the calculation of multiple expressions.
* Add `Function` to reconstruct the computation function. * Add `Function` to reconstruct the computation function.
...@@ -19,6 +27,11 @@ ...@@ -19,6 +27,11 @@
## Bug Fixes ## Bug Fixes
* Add layer check for recurrent_group.
* Clang-format off on some cuda .cc files.
* Fix LogActivation which is not defined.
* Fix bug when run test_layerHelpers multiple times.
* Fix protobuf size limit on seq2seq demo.
* Fix unit test of paramRelu. * Fix unit test of paramRelu.
* Fix some warning about CpuSparseMatrix. * Fix some warning about CpuSparseMatrix.
* Fix MultiGradientMachine error if trainer_count > batch_size. * Fix MultiGradientMachine error if trainer_count > batch_size.
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册