BEAR Applications

TensorFlow-Large-Model-Support 0.1.0-fosscuda-2019a-Python-3.7.2

This library provides an approach to training large models that cannot be fit into GPU memory. It takes a computational graph defined by users, and automatically adds swap-in and swap-out nodes for transferring tensors from GPUs to the host and vice versa. The computational graph is statically modified. Hence, it needs to be done before a session actually starts.

Accessing TensorFlow-Large-Model-Support 0.1.0-fosscuda-2019a-Python-3.7.2

To load the module for TensorFlow-Large-Model-Support 0.1.0-fosscuda-2019a-Python-3.7.2 please use this command on the BEAR systems (BlueBEAR, BEAR Cloud VMs, and CaStLeS VMs):

module load TensorFlow-Large-Model-Support/0.1.0-fosscuda-2019a-Python-3.7.2

There is a CPU version of this module: TensorFlow-Large-Model-Support 0.1.0-foss-2019a-Python-3.7.2

BEAR Apps Version

2020a

Extensions

• TensorFlow-Large-Model-Support-0.1.0
• toposort 1.5

More Information

For more information visit the TensorFlow-Large-Model-Support website.

Dependencies

This version of TensorFlow-Large-Model-Support has a direct dependency on: fosscuda/2019a Python/3.7.2-GCCcore-8.2.0 TensorFlow/1.13.1-fosscuda-2019a-Python-3.7.2

Other Versions

These versions of TensorFlow-Large-Model-Support are available on the BEAR systems (BlueBEAR, BEAR Cloud VMs, and CaStLeS VMs). These will be retained in accordance with our Applications Support and Retention Policy.

Last modified on 31st July 2019