Caffe2 Release 17.09
The NVIDIA container image of Caffe2, release 17.09, is available.
Caffe2 container image version 17.09 is based on Caffe2 0.8.1.
Contents of Caffe2
This container image contains the complete source of the version of Caffe2 in /opt/caffe2
. It is pre-built and installed into the /opt/caffe2/[binaries,lib]
directories in the container image.
The container also includes the following:
- Ubuntu 16.04
- NVIDIA CUDA 9.0
- NVIDIA CUDA® Deep Neural Network library™ (cuDNN) 7.0.2
- NCCL 2.0.5 (optimized for NVLink™ )
- OpenMPI™ 1.10.3
Driver Requirements
Release 17.09 is based on CUDA 9, which requires NVIDIA Driver release 384.xx.
Key Features and Enhancements
This Caffe2 release includes the following key features and enhancements.
- Support Tensor Core Operations for Volta
- Improved the
ImageNet
training script innvidia-examples
- Data input pipeline improvements - use of
libturbojpeg
directly for JPEG decode, and resize work-saving improvements. - Improved the sequence-to-sequence (seq2seq) example in
caffe2/python/models/seq2seq
- Added FP16 support for RNNs via the updated seq2seq example
- Latest version of CUDA
- Latest version of cuDNN
- Latest version of NCCL
- Ubuntu 16.04 with August 2017 updates
Known Issues
There are no known issues in this release.