Version 1.0.12 Release

• Add CUDNN RNN/LSTM/GRU
• [function] atan2

Install the latest nnabla by:

pip install nnabla
pip install nnabla_ext_cuda  # For CUDA users

Users with python <= 3.4 may experience errors with pip install nnabla and pip install nnabla-ext-cuda.
■ Workaround
Please install matplotlib == 2.2.3 and re-install nnabla, nnabla_ext_cuda.

pip install matplotlib==2.2.3
pip install nnabla
pip install nnabla_ext_cuda

Note that CUDA 9.2 and cuDNN 7.3 are set as default if versions are not specified. You can also install the cuda extension with specific versions from one of the following. See also FAQ

• nnabla-ext-cuda80 (CUDA 8.0 x cuDNN 7.1)
• nnabla-ext-cuda90 (CUDA 9.0 x cuDNN 7.3(win), 7.4(linux))
• nnabla-ext-cuda92 (CUDA 9.2 x cuDNN 7.3(win), 7.4(linux))
• nnabla-ext-cuda100 (CUDA 10.0 x cuDNN 7.3(win), 7.4(linux))

pip install nnabla
pip install nnabla_ext_cuda92  # For CUDA 9.2 x cuDNN 7.3 users

Additional setup may be required depending on your OS or environment. Please check Python Package Installation Guide for details.

Install Guide

To use C++ inference feature, follow the demonstration on MNIST inference in C++.

Demonstration

For distributed training, you need to build a binary from source. See the guide for building multi-GPU training package.

Build Guide

Version v1.0.11 Release

• Fix pointer arithmetic and handling of half type
• Add functions: IsInf, IsNaN, ResetNaN, ResetInf, and Where
• Feature/20190121 build with python35
• Print error for cuda ver less than 7
• Use dedicated function to determine workspace size for alogorithm.
• Add CuDNN max and average pooling for 3D case.
• Serialization of SolverState
• Fix CuDNN reduction for output shape equal input shape.
• Fix binary functions

Install the latest nnabla by:

pip install nnabla
pip install nnabla_ext_cuda  # For CUDA users

Users with python <= 3.4 may experience errors with pip install nnabla and pip install nnabla-ext-cuda.
■ Workaround
Please install matplotlib == 2.2.3 and re-install nnabla, nnabla_ext_cuda.

pip install matplotlib==2.2.3
pip install nnabla
pip install nnabla_ext_cuda

Note that CUDA 9.2 and cuDNN 7.3 are set as default if versions are not specified. You can also install the cuda extension with specific versions from one of the following. See also FAQ

• nnabla-ext-cuda80 (CUDA 8.0 x cuDNN 7.1)
• nnabla-ext-cuda90 (CUDA 9.0 x cuDNN 7.3)
• nnabla-ext-cuda92 (CUDA 9.2 x cuDNN 7.3)
• nnabla-ext-cuda100 (CUDA 10.0 x cuDNN 7.3)

pip install nnabla
pip install nnabla_ext_cuda92  # For CUDA 9.2 x cuDNN 7.3 users

Additional setup may be required depending on your OS or environment. Please check Python Package Installation Guide for details.

Install Guide

To use C++ inference feature, follow the demonstration on MNIST inference in C++.

Demonstration

For distributed training, you need to build a binary from source. See the guide for building multi-GPU training package.

Build Guide

Version 1.0.10 Release

• Add openmpi-bin to Multi-GPU docker.
• Build with Python3.7
• Get version info from VERSION.txt

Version 1.0.8 Release

Merge pull request #113 from sony/feature/20181114-fix-multi-gpu-docker

Add openmpi-bin to Multi-GPU docker.

Version 1.0.9 Release

• Add reflection padding and generalize for N-D input.
• Add CuDNN versions of sum/mean/prod reduction
• Add CUDA10 support
• Add implementation of arange function.
• Add option for choosing deterministic algorithms in cuDNN convolution
• [fix] Add include in sort.cu
• Add options for F.min() and F.max() to return indices
• [doc] Avoid unnecessary upgrade of dependent packages
• [function] Prune function
• Add the CUDA implementation for sort function.
• [format] Add auto-format make commands

Install the latest nnabla by:

pip install nnabla
pip install nnabla_ext_cuda  # For CUDA users

Users with python <= 3.4 may experience errors with pip install nnabla and pip install nnabla-ext-cuda.
■ Workaround
Please install matplotlib == 2.2.3 and re-install nnabla, nnabla_ext_cuda.

pip install matplotlib==2.2.3
pip install nnabla
pip install nnabla_ext_cuda

Note that CUDA 9.2 and cuDNN 7.3 are set as default if versions are not specified. You can also install the cuda extension with specific versions from one of the following. See also FAQ

• nnabla-ext-cuda80 (CUDA 8.0 x cuDNN 7.1)
• nnabla-ext-cuda90 (CUDA 9.0 x cuDNN 7.3)
• nnabla-ext-cuda92 (CUDA 9.2 x cuDNN 7.3)
• nnabla-ext-cuda100 (CUDA 10.0 x cuDNN 7.3)

pip install nnabla
pip install nnabla_ext_cuda92  # For CUDA 9.2 x cuDNN 7.3 users

Additional setup may be required depending on your OS or environment. Please check Python Package Installation Guide for details.

Install Guide

To use C++ inference feature, follow the demonstration on MNIST inference in C++.

Demonstration

For distributed training, you need to build a binary from source. See the guide for building multi-GPU training package.

Build Guide

Version 1.0.7 Release

• Fix link.
• Numpy-like basic indexing (negative indexing)
• Create ~/.ccache before prepare docker build.
• Delete try-catch-all logic.

Install the latest nnabla by:

pip install nnabla
pip install nnabla_ext_cuda  # For CUDA users

Users with python <= 3.4 may experience errors with pip install nnabla and pip install nnabla-ext-cuda.
■ Workaround
Please install matplotlib == 2.2.3 and re-install nnabla, nnabla_ext_cuda.

pip install matplotlib==2.2.3
pip install nnabla
pip install nnabla_ext_cuda

Note that CUDA 9.2 and cuDNN 7.2 are set as default if versions are not specified. You can also install the cuda extension with specific versions from one of the following. See also FAQ

• nnabla-ext-cuda80 (CUDA 8.0 x cuDNN 7.2)
• nnabla-ext-cuda90 (CUDA 9.0 x cuDNN 7.2)
• nnabla-ext-cuda91 (CUDA 9.1 x cuDNN 7.1)
• nnabla-ext-cuda92 (CUDA 9.2 x cuDNN 7.2)

pip install nnabla
pip install nnabla_ext_cuda92  # For CUDA 9.2 x cuDNN 7.2 users

Additional setup may be required depending on your OS or environment. Please check Python Package Installation Guide for details.

Install Guide

To use C++ inference feature, follow the demonstration on MNIST inference in C++.

Demonstration

For distributed training, you need to build a binary from source. See the guide for building multi-GPU training package.

Build Guide

Version 1.0.6 Release

• Fix misspell

Install the latest nnabla by:

pip install nnabla
pip install nnabla_ext_cuda  # For CUDA users

Note that CUDA 9.2 and cuDNN 7.2 are set as default if versions are not specified. You can also install the cuda extension with specific versions from one of the following. See also FAQ

• nnabla-ext-cuda80 (CUDA 8.0 x cuDNN 7.2)
• nnabla-ext-cuda90 (CUDA 9.0 x cuDNN 7.2)
• nnabla-ext-cuda91 (CUDA 9.1 x cuDNN 7.1)
• nnabla-ext-cuda92 (CUDA 9.2 x cuDNN 7.2)

pip install nnabla
pip install nnabla_ext_cuda92  # For CUDA 9.2 x cuDNN 7.2 users

Additional setup may be required depending on your OS or environment. Please check Python Package Installation Guide for details.

Install Guide

To use C++ inference feature, follow the demonstration on MNIST inference in C++.

Demonstration

For distributed training, you need to build a binary from source. See the guide for building multi-GPU training package.

Build Guide

Version 1.0.5 Release

• Get cmake 3.x from EPEL repository.
• Discard MULTI_GPU_NAME and MULTI_GPU, and enable MULTI_GPU_SUFFIX and…

Install the latest nnabla by:

pip install nnabla
pip install nnabla_ext_cuda  # For CUDA users

Note that CUDA 9.2 and cuDNN 7.2 are set as default if versions are not specified. You can also install the cuda extension with specific versions from one of the following. See also FAQ

• nnabla-ext-cuda80 (CUDA 8.0 x cuDNN 7.2)
• nnabla-ext-cuda90 (CUDA 9.0 x cuDNN 7.2)
• nnabla-ext-cuda91 (CUDA 9.1 x cuDNN 7.1)
• nnabla-ext-cuda92 (CUDA 9.2 x cuDNN 7.2)

pip install nnabla
pip install nnabla_ext_cuda92  # For CUDA 9.2 x cuDNN 7.2 users

Additional setup may be required depending on your OS or environment. Please check Python Package Installation Guide for details.

Install Guide

To use C++ inference feature, follow the demonstration on MNIST inference in C++.

Demonstration

For distributed training, you need to build a binary from source. See the guide for building multi-GPU training package.

Build Guide

Version 1.0.4 Release

• Feature/20180907 multi gpu package
• Resize function with interpolation (add only 2D bilinear)
• Fix -u option.
• Multi-GPU docker build
• Remove unnecessary print
• LeakyReLU with inplace
• Update numpy version to 1.12 and install scikit-image by PIP instead of conda
• Remove MAKE_MANYLINUX_WHEEL from build-with-docker
• [c++] Enable Group in (De)ConvolutionCudaCudnn
• Fix missing SCL package issue
• Fix bug for division of collectives

Install the latest nnabla by:

pip install nnabla
pip install nnabla_ext_cuda  # For CUDA users

Note that CUDA 9.2 and cuDNN 7.2 are set as default if versions are not specified. You can also install the cuda extension with specific versions from one of the following. See also FAQ

• nnabla-ext-cuda80 (CUDA 8.0 x cuDNN 7.2)
• nnabla-ext-cuda90 (CUDA 9.0 x cuDNN 7.2)
• nnabla-ext-cuda91 (CUDA 9.1 x cuDNN 7.1)
• nnabla-ext-cuda92 (CUDA 9.2 x cuDNN 7.2)

pip install nnabla
pip install nnabla_ext_cuda92  # For CUDA 9.2 x cuDNN 7.2 users

Additional setup may be required depending on your OS or environment. Please check Python Package Installation Guide for details.

Install Guide

To use C++ inference feature, follow the demonstration on MNIST inference in C++.

Demonstration

For distributed training, you need to build a binary from source. See the guide for building multi-GPU training package.

Build Guide

Version 1.0.3 Release

• Fix/20180802 eigen download source
• Properly check architecture in cublas gemm and batched gemm