BigDL v0.1.1 Release Notes
Release Date: 2017-06-16 // almost 7 years ago-
Release Notes
- API Change
๐ฆ 1. Use bigdl as the top level package name for all bigdl python module โก๏ธ 2. Allow user to change the model in the optimizer
- Allow user to define a model in python API
- Allow user to invoke BigDL scala code from python in 3rd prject
- Allow user to use BigDL random generator in python
- Allow user to use forward/backward method in python
- Add BiRnn layer to python ๐ 8. Remove useless CriterionTable layer
- โจ Enhancement
- Load libjmkl.so in the class load phase ๐ 2. Support python 3.5
- Initialize gradient buffer at the start of backward to reduce the memory usage
- Auto pack python dependency in yarn cluster mode
- ๐ Bug Fix
โก๏ธ 1. Fix optimizer continue without failure after retry maximum number
- Fix LookupTable python API throw noSuchMethod error
- Fix an addmv bug for 1x1 matrix
- Fix lenet python example error
- Fix python load text file encoding issue ๐ 6. Fix HardTanh performance issue
- Fix data may distribute unevenly in vgg example when input partition is too large
- Fix a bug in SpatialDilatedConvolution
- Fix a bug in BCECriterion loss function
- Fix a bug in Add layer
- Fix runtime error when run BigDL on Pyspark 1.5