NVIDIA DGX Team is pleased to announce availability of DGX Optimized Frameworks Release 17.07

Answer ID 4527   |    Updated 07/18/2017 01:47 PM

**NEW**

This release includes the following framework updates. For complete details, please refer to the 17.07 Release Notes for each framework found here.

Caffe

• Fixed a bug that was introduced in 17.05 where the log output from the caffe time command was excessively large.

• Ubuntu 16.04 with June 2017 updates

Caffe2

• Updated version of the imagent script in the nvidia-examples directory.

• Ubuntu 16.04 with June 2017 updates

Cognitive Toolkit

• Ubuntu 16.04 with June 2017 updates

MXNet

• Support for multi-precision SGD

• cuBLAS® back-end for FullyConnected operation

• Ubuntu 16.04 with June 2017 updates

PyTorch

• Support for advanced tensor indexing

• Support multi-node or multi-process mode on the same node

• Support for double backward for most functions, including convolution

• Ubuntu 16.04 with June 2017 updates

TensorFlow

• Small bug-fixes in evaluation mode of nvidia-examples/cnn/nvcnn.py

• Ubuntu 16.04 with June 2017 updates

Theano

• Merged numerous upstream bug fixes and enhancements.

• Ubuntu 16.04 with June 2017 updates

Torch

• Ubuntu 16.04 with June 2017 updates

Was this answer helpful?
 
Your rating has been submitted, please tell us how we can make this answer more useful.

Print