Version 1.31.0 Release

release-note-bugfix

• Support compile nnabla-ext-cuda with cuTENSOR in non-docker environment

release-note-core

• Add narrow function for NdArray

release-note-examples

• add example of infer semantic segmentation by cpp

release-note-op-layer

• Add TopK support to ONNX importer
• Error function (erf) implementation (CUDA)

Install the latest nnabla by:

pip install nnabla
pip install nnabla-ext-cuda110 # For CUDA version 11.0.3 users

You can also install the cuda extension with specific versions from one of the following. See also FAQ

• nnabla-ext-cuda110 (CUDA 11.0.3 x cuDNN 8.0)
• nnabla-ext-cuda114 (CUDA 11.4.3 x cuDNN 8.0)

pip install nnabla
pip install nnabla_ext_cuda110  # For CUDA 11.0.3 x cuDNN 8.0 users

The nnabla cuda extension supports both centralized and distributed training.
Below packages detect MPI and NCCL2 installed on your system. If either one
of then is unavailable, distributed training is disabled.

• nnabla-ext-cuda110 (Ubuntu18.04 default)
• nnabla-ext-cuda114 (Ubuntu18.04 default)

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

Build Guide
The "nnabla-ext-cuda" package is temporarily unavailable. Use of this package is not recommended. Please install nnabla-ext-cuda110, nnabla-ext-cuda114 instead.
The following nnabla CUDA extension packages have been deprecated and the PyPi repository has been (or going to be) closed.

• nnabla-ubuntu16
• nnabla-ubuntu18
• nnabla-ext-cuda
• nnabla-ext-cuda80
• nnabla-ext-cuda91
• nnabla-ext-cuda92
• nnabla-ext-cuda100
• nnabla-ext-cuda101
• nnabla-ext-cuda90-nccl2-ubuntu16
• nnabla-ext-cuda100-nccl2-ubuntu16
• nnabla-ext-cuda100-nccl2-ubuntu18
• nnabla-ext-cuda101-nccl2-ubuntu16
• nnabla-ext-cuda101-nccl2-ubuntu18
• nnabla-ext-cuda102-nccl2-mpi2-1-1
• nnabla-ext-cuda102-nccl2-mpi3-1-6
• nnabla-ext-cuda110-nccl2-mpi2-1-1
• nnabla-ext-cuda110-nccl2-mpi3-1-6

The following "nnabla-ext-cuda" docker images have been deprecated.

• py27-cuda92
• py36-cuda92
• py37-cuda92
• py27-cuda92-v1.0.xx
• py36-cuda92-v1.0.xx
• py37-cuda92-v1.0.xx
• py36-cuda100-v1.xx.x
• py37-cuda100-v1.xx.x
• py38-cuda100-v1.xx.x
• py39-cuda100-v1.xx.x
• py36-cuda102-v1.xx.x
• py37-cuda102-v1.xx.x
• py38-cuda102-v1.xx.x
• py39-cuda102-v1.xx.x

We've decided to change nnabla's versioning policy to semantic versioning.
This change has been applied from version 1.1.0.

Python 2 is no longer be supported from v1.5.0.
CUDA8.0 is no longer be supported from v1.6.0.
Python3.5 is no longer be supported from v1.14.0.
CUDA9.0 is no longer be supported from v1.14.0.
Python3.6 is no longer be supported from v1.26.0.
CUDA10.0 is no longer be supported from v1.26.0.
CUDA10.2 is no longer be supported from v1.30.0.

Known Issue:

• cufftXtMakePlanMany returns CUFFT_INVALID_PLAN on Windows + CUDA11.4.3 environment. See details here

Version 1.30.0 Release

release-note-bugfix

• Set to use openib by default
• Update nvidia/cuda base image
• disable dl_nccl_generate when nccl_disabled

release-note-build

• remove flatc build from Dockerfile
• include cuda/cudnn libraries in wheel
• Fix CUDA_VERSION_MINOR to various condition
• add empty targz and merge cuda wheel
• Prepare to build CUDA11.4 support

release-note-core

• Optimize TransposeCuda with cuTENSOR

release-note-utility

• Better error message in CUDNN.

Install the latest nnabla by:

pip install nnabla
pip install nnabla-ext-cuda110 # For CUDA version 11.0.3 users

You can also install the cuda extension with specific versions from one of the following. See also FAQ

• nnabla-ext-cuda110 (CUDA 11.0.3 x cuDNN 8.0)
• nnabla-ext-cuda114 (CUDA 11.4.3 x cuDNN 8.0)

pip install nnabla
pip install nnabla_ext_cuda110  # For CUDA 11.0.3 x cuDNN 8.0 users

The nnabla cuda extension supports both centralized and distributed training.
Below packages detect MPI and NCCL2 installed on your system. If either one
of then is unavailable, distributed training is disabled.

• nnabla-ext-cuda110 (Ubuntu18.04 default)
• nnabla-ext-cuda114 (Ubuntu18.04 default)

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

Build Guide
The "nnabla-ext-cuda" package is temporarily unavailable. Use of this package is not recommended. Please install nnabla-ext-cuda110, nnabla-ext-cuda114 instead.
The following nnabla CUDA extension packages have been deprecated and the PyPi repository has been (or going to be) closed.

• nnabla-ubuntu16
• nnabla-ubuntu18
• nnabla-ext-cuda
• nnabla-ext-cuda80
• nnabla-ext-cuda91
• nnabla-ext-cuda92
• nnabla-ext-cuda100
• nnabla-ext-cuda101
• nnabla-ext-cuda90-nccl2-ubuntu16
• nnabla-ext-cuda100-nccl2-ubuntu16
• nnabla-ext-cuda100-nccl2-ubuntu18
• nnabla-ext-cuda101-nccl2-ubuntu16
• nnabla-ext-cuda101-nccl2-ubuntu18
• nnabla-ext-cuda102-nccl2-mpi2-1-1
• nnabla-ext-cuda102-nccl2-mpi3-1-6
• nnabla-ext-cuda110-nccl2-mpi2-1-1
• nnabla-ext-cuda110-nccl2-mpi3-1-6

The following "nnabla-ext-cuda" docker images have been deprecated.

• py27-cuda92
• py36-cuda92
• py37-cuda92
• py27-cuda92-v1.0.xx
• py36-cuda92-v1.0.xx
• py37-cuda92-v1.0.xx
• py36-cuda100-v1.xx.x
• py37-cuda100-v1.xx.x
• py38-cuda100-v1.xx.x
• py39-cuda100-v1.xx.x
• py36-cuda102-v1.xx.x
• py37-cuda102-v1.xx.x
• py38-cuda102-v1.xx.x
• py39-cuda102-v1.xx.x

We've decided to change nnabla's versioning policy to semantic versioning.
This change has been applied from version 1.1.0.

Python 2 is no longer be supported from v1.5.0.
CUDA8.0 is no longer be supported from v1.6.0.
Python3.5 is no longer be supported from v1.14.0.
CUDA9.0 is no longer be supported from v1.14.0.
Python3.6 is no longer be supported from v1.26.0.
CUDA10.0 is no longer be supported from v1.26.0.
CUDA10.2 is no longer be supported from v1.30.0.

Known Issue:

• cufftXtMakePlanMany returns CUFFT_INVALID_PLAN on Windows + CUDA11.4.3 environment. See details here

Version 1.29.0 Release

release-note-bugfix

• support libnccl.so.2
• improve libmpi.so path searching method

release-note-build

• Update protobuf to 3.19.4
• use numpy 1.20.0 and later
• Update gpg keys for centos based nvidia docker
• update nvidia gpg key
• Build flatc from latest upstream release
• enable dynamic load nccl library

Install the latest nnabla by:

pip install nnabla
pip install nnabla-ext-cuda102 # For CUDA version 10.2 users

You can also install the cuda extension with specific versions from one of the following. See also FAQ

• nnabla-ext-cuda102 (CUDA 10.2 x cuDNN 8.0)
• nnabla-ext-cuda110 (CUDA 11.0 x cuDNN 8.0)

pip install nnabla
pip install nnabla_ext_cuda102  # For CUDA 10.2 x cuDNN 8.0 users

For distributed training, You need to install the correct package that
matches the version of MPI installed on your system.

We prepared following packages.

• nnabla-ext-cuda102-nccl2-mpi2-1-1 (Ubuntu18.04 default)
• nnabla-ext-cuda102-nccl2-mpi3-1-6
• nnabla-ext-cuda110-nccl2-mpi2-1-1 (Ubuntu18.04 default)
• nnabla-ext-cuda110-nccl2-mpi3-1-6

If you want to use a version of MPI not listed above, you need to build it from the source.

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

Build Guide
The "nnabla-ext-cuda" package is temporarily unavailable. Use of this package is not recommended. Please install nnabla-ext-cuda110, nnabla-ext-cuda102 instead.
The following nnabla CUDA extension packages have been deprecated and the PyPi repository has been (or going to be) closed.

• nnabla-ubuntu16
• nnabla-ubuntu18
• nnabla-ext-cuda
• nnabla-ext-cuda80
• nnabla-ext-cuda91
• nnabla-ext-cuda92
• nnabla-ext-cuda100
• nnabla-ext-cuda101
• nnabla-ext-cuda90-nccl2-ubuntu16
• nnabla-ext-cuda100-nccl2-ubuntu16
• nnabla-ext-cuda100-nccl2-ubuntu18
• nnabla-ext-cuda101-nccl2-ubuntu16
• nnabla-ext-cuda101-nccl2-ubuntu18

The following "nnabla-ext-cuda" docker images have been deprecated.

• py27-cuda92
• py36-cuda92
• py37-cuda92
• py27-cuda92-v1.0.xx
• py36-cuda92-v1.0.xx
• py37-cuda92-v1.0.xx
• py36-cuda100

We've decided to change nnabla's versioning policy to semantic versioning.
This change has been applied from version 1.1.0.

Python 2 is no longer be supported from v1.5.0.
CUDA8.0 is no longer be supported from v1.6.0.
Python3.5 is no longer be supported from v1.14.0.
CUDA9.0 is no longer be supported from v1.14.0.
Python3.6 is no longer be supported from v1.26.0.
CUDA10.0 is no longer be supported from v1.26.0.

Version 1.28.0 Release

release-note-bugfix

• Fix: Slow Softmax backward when H and W are 1

release-note-build

• Fix flatc build issue

Install the latest nnabla by:

pip install nnabla
pip install nnabla-ext-cuda102 # For CUDA version 10.2 users

You can also install the cuda extension with specific versions from one of the following. See also FAQ

• nnabla-ext-cuda102 (CUDA 10.2 x cuDNN 8.0)
• nnabla-ext-cuda110 (CUDA 11.0 x cuDNN 8.0)

pip install nnabla
pip install nnabla_ext_cuda102  # For CUDA 10.2 x cuDNN 8.0 users

For distributed training, You need to install the correct package that
matches the version of MPI installed on your system.

We prepared following packages.

• nnabla-ext-cuda102-nccl2-mpi2-1-1 (Ubuntu18.04 default)
• nnabla-ext-cuda102-nccl2-mpi3-1-6
• nnabla-ext-cuda110-nccl2-mpi2-1-1 (Ubuntu18.04 default)
• nnabla-ext-cuda110-nccl2-mpi3-1-6

If you want to use a version of MPI not listed above, you need to build it from the source.

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

Build Guide
The "nnabla-ext-cuda" package is temporarily unavailable. Use of this package is not recommended. Please install nnabla-ext-cuda110, nnabla-ext-cuda102 instead.
The following nnabla CUDA extension packages have been deprecated and the PyPi repository has been (or going to be) closed.

• nnabla-ubuntu16
• nnabla-ubuntu18
• nnabla-ext-cuda
• nnabla-ext-cuda80
• nnabla-ext-cuda91
• nnabla-ext-cuda92
• nnabla-ext-cuda100
• nnabla-ext-cuda101
• nnabla-ext-cuda90-nccl2-ubuntu16
• nnabla-ext-cuda100-nccl2-ubuntu16
• nnabla-ext-cuda100-nccl2-ubuntu18
• nnabla-ext-cuda101-nccl2-ubuntu16
• nnabla-ext-cuda101-nccl2-ubuntu18

The following "nnabla-ext-cuda" docker images have been deprecated.

• py27-cuda92
• py36-cuda92
• py37-cuda92
• py27-cuda92-v1.0.xx
• py36-cuda92-v1.0.xx
• py37-cuda92-v1.0.xx
• py36-cuda100

We've decided to change nnabla's versioning policy to semantic versioning.
This change has been applied from version 1.1.0.

Python 2 is no longer be supported from v1.5.0.
CUDA8.0 is no longer be supported from v1.6.0.
Python3.5 is no longer be supported from v1.14.0.
CUDA9.0 is no longer be supported from v1.14.0.
Python3.6 is no longer be supported from v1.26.0.
CUDA10.0 is no longer be supported from v1.26.0.

Version 1.27.0 Release

release-note-bugfix

• Fix: copy function was missing in MaxPoolingCudnn

release-note-core

• Fused weight decay into Solver update / Fix SgdW / Add Lamb solver
• Enable customizing std types and functions with memory allocation

release-note-op-layer

• Feature/20210608 cholesky decomposition
• Speedup SortCuda with cub::DeviceSegmentedRadixSort

release-note-utility

• Auto restart DALI iterator at StopIteration
• Support NVTX range events

Install the latest nnabla by:

pip install nnabla
pip install nnabla-ext-cuda102 # For CUDA version 10.2 users

You can also install the cuda extension with specific versions from one of the following. See also FAQ

• nnabla-ext-cuda102 (CUDA 10.2 x cuDNN 8.0)
• nnabla-ext-cuda110 (CUDA 11.0 x cuDNN 8.0)

pip install nnabla
pip install nnabla_ext_cuda102  # For CUDA 10.2 x cuDNN 8.0 users

For distributed training, You need to install the correct package that
matches the version of MPI installed on your system.

We prepared following packages.

• nnabla-ext-cuda102-nccl2-mpi2-1-1 (Ubuntu18.04 default)
• nnabla-ext-cuda102-nccl2-mpi3-1-6
• nnabla-ext-cuda110-nccl2-mpi2-1-1 (Ubuntu18.04 default)
• nnabla-ext-cuda110-nccl2-mpi3-1-6

If you want to use a version of MPI not listed above, you need to build it from the source.

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

Build Guide
The "nnabla-ext-cuda" package is temporarily unavailable. Use of this package is not recommended. Please install nnabla-ext-cuda110, nnabla-ext-cuda102 instead.
The following nnabla CUDA extension packages have been deprecated and the PyPi repository has been (or going to be) closed.

• nnabla-ubuntu16
• nnabla-ubuntu18
• nnabla-ext-cuda
• nnabla-ext-cuda80
• nnabla-ext-cuda91
• nnabla-ext-cuda92
• nnabla-ext-cuda100
• nnabla-ext-cuda101
• nnabla-ext-cuda90-nccl2-ubuntu16
• nnabla-ext-cuda100-nccl2-ubuntu16
• nnabla-ext-cuda100-nccl2-ubuntu18
• nnabla-ext-cuda101-nccl2-ubuntu16
• nnabla-ext-cuda101-nccl2-ubuntu18

The following "nnabla-ext-cuda" docker images have been deprecated.

• py27-cuda92
• py36-cuda92
• py37-cuda92
• py27-cuda92-v1.0.xx
• py36-cuda92-v1.0.xx
• py37-cuda92-v1.0.xx
• py36-cuda100

We've decided to change nnabla's versioning policy to semantic versioning.
This change has been applied from version 1.1.0.

Python 2 is no longer be supported from v1.5.0.
CUDA8.0 is no longer be supported from v1.6.0.
Python3.5 is no longer be supported from v1.14.0.
CUDA9.0 is no longer be supported from v1.14.0.
Python3.6 is no longer be supported from v1.26.0.
CUDA10.0 is no longer be supported from v1.26.0.

Version 1.26.0 Release

release-note-build

• Replace pow to products
• drop support python3.6 and CUDA-cuDNN 10.0/7
• Support dynamic load mpi library
• Feature/20211203 unify dockerfiles

release-note-core

• restrict condition with dst for communicator reduce
• Add computation graph active inputs handling.

release-note-op-layer

• Change random_choice to use faster cumulative summing

release-note-utility

• caching device properties for avoiding slow recall

Install the latest nnabla by:

pip install nnabla
pip install nnabla-ext-cuda102 # For CUDA version 10.2 users

You can also install the cuda extension with specific versions from one of the following. See also FAQ

• nnabla-ext-cuda102 (CUDA 10.2 x cuDNN 8.0)
• nnabla-ext-cuda110 (CUDA 11.0 x cuDNN 8.0)

pip install nnabla
pip install nnabla_ext_cuda102  # For CUDA 10.2 x cuDNN 8.0 users

For distributed training, You need to install the correct package that
matches the version of MPI installed on your system.

We prepared following packages.

• nnabla-ext-cuda102-nccl2-mpi2-1-1 (Ubuntu18.04 default)
• nnabla-ext-cuda102-nccl2-mpi3-1-6
• nnabla-ext-cuda110-nccl2-mpi2-1-1 (Ubuntu18.04 default)
• nnabla-ext-cuda110-nccl2-mpi3-1-6

If you want to use a version of MPI not listed above, you need to build it from the source.

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

Build Guide
The "nnabla-ext-cuda" package is temporarily unavailable. Use of this package is not recommended. Please install nnabla-ext-cuda110, nnabla-ext-cuda102 instead.
The following nnabla CUDA extension packages have been deprecated and the PyPi repository has been (or going to be) closed.

• nnabla-ubuntu16
• nnabla-ubuntu18
• nnabla-ext-cuda
• nnabla-ext-cuda80
• nnabla-ext-cuda91
• nnabla-ext-cuda92
• nnabla-ext-cuda100
• nnabla-ext-cuda101
• nnabla-ext-cuda90-nccl2-ubuntu16
• nnabla-ext-cuda100-nccl2-ubuntu16
• nnabla-ext-cuda100-nccl2-ubuntu18
• nnabla-ext-cuda101-nccl2-ubuntu16
• nnabla-ext-cuda101-nccl2-ubuntu18

The following "nnabla-ext-cuda" docker images have been deprecated.

• py27-cuda92
• py36-cuda92
• py37-cuda92
• py27-cuda92-v1.0.xx
• py36-cuda92-v1.0.xx
• py37-cuda92-v1.0.xx
• py36-cuda100

We've decided to change nnabla's versioning policy to semantic versioning.
This change has been applied from version 1.1.0.

Python 2 is no longer be supported from v1.5.0.
CUDA8.0 is no longer be supported from v1.6.0.
Python3.5 is no longer be supported from v1.14.0.
CUDA9.0 is no longer be supported from v1.14.0.
Python3.6 is no longer be supported from v1.26.0.
CUDA10.0 is no longer be supported from v1.26.0.

Version 1.25.0 Release

release-note-bugfix

• Fix FusedBN not to fall back to cpu implementation always
• Update deformable_convolution.cu
• Add AssignCuda not to perform host-device copy
• Use get_current_device_attribute rather than properties

release-note-build

• upgrade g++ from 4.9.2 to 6.5.0 and enable stdc++14

release-note-doc

• Fix spelling "Currently"

release-note-examples

• make corresponding modification for the files in reconstruction directories

Install the latest nnabla by:

pip install nnabla
pip install nnabla-ext-cuda102 # For CUDA version 10.2 users

You can also install the cuda extension with specific versions from one of the following. See also FAQ

• nnabla-ext-cuda100 (CUDA 10.0 x cuDNN 7.6)
• nnabla-ext-cuda102 (CUDA 10.2 x cuDNN 8.0)
• nnabla-ext-cuda110 (CUDA 11.0 x cuDNN 8.0)

pip install nnabla
pip install nnabla_ext_cuda102  # For CUDA 10.2 x cuDNN 8.0 users

For distributed training, You need to install the correct package that
matches the version of MPI installed on your system.

We prepared following packages.

• nnabla-ext-cuda100-nccl2-mpi2-1-1 (Ubuntu18.04 default)
• nnabla-ext-cuda100-nccl2-mpi3-1-6
• nnabla-ext-cuda102-nccl2-mpi2-1-1 (Ubuntu18.04 default)
• nnabla-ext-cuda102-nccl2-mpi3-1-6
• nnabla-ext-cuda110-nccl2-mpi2-1-1 (Ubuntu18.04 default)
• nnabla-ext-cuda110-nccl2-mpi3-1-6

If you want to use a version of MPI not listed above, you need to build it from the source.

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

Build Guide
The "nnabla-ext-cuda" package is temporarily unavailable. Use of this package is not recommended. Please install nnabla-ext-cuda110, nnabla-ext-cuda102, nnabla-ext-cuda100 instead.
The following nnabla CUDA extension packages have been deprecated and the PyPi repository has been (or going to be) closed.

• nnabla-ubuntu16
• nnabla-ubuntu18
• nnabla-ext-cuda
• nnabla-ext-cuda80
• nnabla-ext-cuda91
• nnabla-ext-cuda92
• nnabla-ext-cuda101
• nnabla-ext-cuda90-nccl2-ubuntu16
• nnabla-ext-cuda100-nccl2-ubuntu16
• nnabla-ext-cuda100-nccl2-ubuntu18
• nnabla-ext-cuda101-nccl2-ubuntu16
• nnabla-ext-cuda101-nccl2-ubuntu18

The following "nnabla-ext-cuda" docker images have been deprecated.

• py27-cuda92
• py36-cuda92
• py37-cuda92
• py27-cuda92-v1.0.xx
• py36-cuda92-v1.0.xx
• py37-cuda92-v1.0.xx

We've decided to change nnabla's versioning policy to semantic versioning.
This change has been applied from version 1.1.0.

Python 2 is no longer be supported from v1.5.0.
CUDA8.0 is no longer be supported from v1.6.0.
Python3.5 is no longer be supported from v1.14.0.
CUDA9.0 is no longer be supported from v1.14.0.

Version 1.24.0 Release

release-note-bugfix

• Remove the argument "output_mask" from Dropout
• fix multithread issues

release-note-build

• Add PYTEST_OPTS for pytest parallel execution
• support python3.9

release-note-op-layer

• Support broadcast in instance norm kernel to improve performance
• Make ISTFT consistent with PyTorch implementation (NOLA condition)
• Add linspace function
• Optimize CumProd/CumSum

release-note-utility

• Default disable watchdog and enable it with environment variable

Install the latest nnabla by:

pip install nnabla
pip install nnabla-ext-cuda102 # For CUDA version 10.2 users

You can also install the cuda extension with specific versions from one of the following. See also FAQ

• nnabla-ext-cuda100 (CUDA 10.0 x cuDNN 7.6)
• nnabla-ext-cuda102 (CUDA 10.2 x cuDNN 8.0)
• nnabla-ext-cuda110 (CUDA 11.0 x cuDNN 8.0)

pip install nnabla
pip install nnabla_ext_cuda102  # For CUDA 10.2 x cuDNN 8.0 users

For distributed training, You need to install the correct package that
matches the version of MPI installed on your system.

We prepared following packages.

• nnabla-ext-cuda100-nccl2-mpi2-1-1 (Ubuntu18.04 default)
• nnabla-ext-cuda100-nccl2-mpi3-1-6
• nnabla-ext-cuda102-nccl2-mpi2-1-1 (Ubuntu18.04 default)
• nnabla-ext-cuda102-nccl2-mpi3-1-6
• nnabla-ext-cuda110-nccl2-mpi2-1-1 (Ubuntu18.04 default)
• nnabla-ext-cuda110-nccl2-mpi3-1-6

If you want to use a version of MPI not listed above, you need to build it from the source.

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

Build Guide
The "nnabla-ext-cuda" package is temporarily unavailable. Use of this package is not recommended. Please install nnabla-ext-cuda110, nnabla-ext-cuda102, nnabla-ext-cuda100 instead.
The following nnabla CUDA extension packages have been deprecated and the PyPi repository has been (or going to be) closed.

• nnabla-ubuntu16
• nnabla-ubuntu18
• nnabla-ext-cuda
• nnabla-ext-cuda80
• nnabla-ext-cuda91
• nnabla-ext-cuda92
• nnabla-ext-cuda101
• nnabla-ext-cuda90-nccl2-ubuntu16
• nnabla-ext-cuda100-nccl2-ubuntu16
• nnabla-ext-cuda100-nccl2-ubuntu18
• nnabla-ext-cuda101-nccl2-ubuntu16
• nnabla-ext-cuda101-nccl2-ubuntu18

The following "nnabla-ext-cuda" docker images have been deprecated.

• py27-cuda92
• py36-cuda92
• py37-cuda92
• py27-cuda92-v1.0.xx
• py36-cuda92-v1.0.xx
• py37-cuda92-v1.0.xx

We've decided to change nnabla's versioning policy to semantic versioning.
This change has been applied from version 1.1.0.

Python 2 is no longer be supported from v1.5.0.
CUDA8.0 is no longer be supported from v1.6.0.
Python3.5 is no longer be supported from v1.14.0.
CUDA9.0 is no longer be supported from v1.14.0.

Version 1.23.0 Release

release-note-bugfix

• Need explicit Shape_t initialization since CUDA 11.1

release-note-build

• Revert "replace centos with ubuntu"
• Add pytest-xdist options
• Apply new pyyaml loader()

Install the latest nnabla by:

pip install nnabla
pip install nnabla-ext-cuda102 # For CUDA version 10.2 users

You can also install the cuda extension with specific versions from one of the following. See also FAQ

• nnabla-ext-cuda100 (CUDA 10.0 x cuDNN 7.6)
• nnabla-ext-cuda102 (CUDA 10.2 x cuDNN 8.0)
• nnabla-ext-cuda110 (CUDA 11.0 x cuDNN 8.0)

pip install nnabla
pip install nnabla_ext_cuda102  # For CUDA 10.2 x cuDNN 8.0 users

For distributed training, You need to install the correct package that
matches the version of MPI installed on your system.

We prepared following packages.

• nnabla-ext-cuda100-nccl2-mpi2-1-1 (Ubuntu18.04 default)
• nnabla-ext-cuda100-nccl2-mpi3-1-6
• nnabla-ext-cuda102-nccl2-mpi2-1-1 (Ubuntu18.04 default)
• nnabla-ext-cuda102-nccl2-mpi3-1-6
• nnabla-ext-cuda110-nccl2-mpi2-1-1 (Ubuntu18.04 default)
• nnabla-ext-cuda110-nccl2-mpi3-1-6

If you want to use a version of MPI not listed above, you need to build it from the source.

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

Build Guide
The "nnabla-ext-cuda" package is temporarily unavailable. Use of this package is not recommended. Please install nnabla-ext-cuda110, nnabla-ext-cuda102, nnabla-ext-cuda100 instead.
The following nnabla CUDA extension packages have been deprecated and the PyPi repository has been (or going to be) closed.

• nnabla-ubuntu16
• nnabla-ubuntu18
• nnabla-ext-cuda
• nnabla-ext-cuda80
• nnabla-ext-cuda91
• nnabla-ext-cuda92
• nnabla-ext-cuda101
• nnabla-ext-cuda90-nccl2-ubuntu16
• nnabla-ext-cuda100-nccl2-ubuntu16
• nnabla-ext-cuda100-nccl2-ubuntu18
• nnabla-ext-cuda101-nccl2-ubuntu16
• nnabla-ext-cuda101-nccl2-ubuntu18

The following "nnabla-ext-cuda" docker images have been deprecated.

• py27-cuda92
• py36-cuda92
• py37-cuda92
• py27-cuda92-v1.0.xx
• py36-cuda92-v1.0.xx
• py37-cuda92-v1.0.xx

We've decided to change nnabla's versioning policy to semantic versioning.
This change has been applied from version 1.1.0.

Python 2 is no longer be supported from v1.5.0.
CUDA8.0 is no longer be supported from v1.6.0.
Python3.5 is no longer be supported from v1.14.0.
CUDA9.0 is no longer be supported from v1.14.0.

Version 1.22.0 Release

release-note-break-compat

• Disable TF32 by default

release-note-bugfix

• Use generic transpose kernel if the grid y-dimension exceeds the CUDA limit.
• Bugfix the higher-order gradient of gather_nd

release-note-build

• replace centos with ubuntu and use PEP600 tag

release-note-core

• Improve ioctl time on A100 by cudaDeviceGetAttribute

release-note-op-layer

• Fast reduction
• Optimize normalization functions

Install the latest nnabla by:

pip install nnabla
pip install nnabla-ext-cuda102 # For CUDA version 10.2 users

You can also install the cuda extension with specific versions from one of the following. See also FAQ

• nnabla-ext-cuda100 (CUDA 10.0 x cuDNN 7.6)
• nnabla-ext-cuda102 (CUDA 10.2 x cuDNN 8.0)
• nnabla-ext-cuda110 (CUDA 11.0 x cuDNN 8.0)

pip install nnabla
pip install nnabla_ext_cuda102  # For CUDA 10.2 x cuDNN 8.0 users

For distributed training, You need to install the correct package that
matches the version of MPI installed on your system.

We prepared following packages.

• nnabla-ext-cuda100-nccl2-mpi2-1-1 (Ubuntu18.04 default)
• nnabla-ext-cuda100-nccl2-mpi3-1-6
• nnabla-ext-cuda102-nccl2-mpi2-1-1 (Ubuntu18.04 default)
• nnabla-ext-cuda102-nccl2-mpi3-1-6
• nnabla-ext-cuda110-nccl2-mpi2-1-1 (Ubuntu18.04 default)
• nnabla-ext-cuda110-nccl2-mpi3-1-6

If you want to use a version of MPI not listed above, you need to build it from the source.

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

Build Guide
The "nnabla-ext-cuda" package is temporarily unavailable. Use of this package is not recommended. Please install nnabla-ext-cuda110, nnabla-ext-cuda102, nnabla-ext-cuda100 instead.
The following nnabla CUDA extension packages have been deprecated and the PyPi repository has been (or going to be) closed.

• nnabla-ubuntu16
• nnabla-ubuntu18
• nnabla-ext-cuda
• nnabla-ext-cuda80
• nnabla-ext-cuda91
• nnabla-ext-cuda92
• nnabla-ext-cuda101
• nnabla-ext-cuda90-nccl2-ubuntu16
• nnabla-ext-cuda100-nccl2-ubuntu16
• nnabla-ext-cuda100-nccl2-ubuntu18
• nnabla-ext-cuda101-nccl2-ubuntu16
• nnabla-ext-cuda101-nccl2-ubuntu18

The following "nnabla-ext-cuda" docker images have been deprecated.

• py27-cuda92
• py36-cuda92
• py37-cuda92
• py27-cuda92-v1.0.xx
• py36-cuda92-v1.0.xx
• py37-cuda92-v1.0.xx

We've decided to change nnabla's versioning policy to semantic versioning.
This change has been applied from version 1.1.0.

Python 2 is no longer be supported from v1.5.0.
CUDA8.0 is no longer be supported from v1.6.0.
Python3.5 is no longer be supported from v1.14.0.
CUDA9.0 is no longer be supported from v1.14.0.