Versions Compared

Version Old Version 1 New Version Current
Changes made by

Shawn Doughty (Deactivated)

Shawn Doughty (Deactivated)

Saved on

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

Excerpt

...

TensorFlow is an open source

...

Users can install their own individual version of Jupyter using the following steps.

  1. Install your own version of python to run jupyter. Instructions go on another page.
  2. Code Block
    titleInstall jupyter within anaconda
    login001: conda install jupyter
  3. Start up jupyter
    1. Code Block
      languagebash
      titleGet an allocation on a compute node. Write down the name of the assigned compute node
      login001: srun --pty --x11=first -p interactive bash
alpha001:
    2. Code Block
      titleStart jupyter on the compute node. It will display the port # the software is running on, usually 8888
      alpha001: jupyter notebook --no-browser
[NotebookApp] The Jupyter Notebook is running at: http://localhost:8888/
[NotebookApp] Use Control-C to stop this server and shut down all kernels (twice to skip confirmation).
  4. Access jupyter
    1. Code Block
      languagebash
      titleNow setup forwarding from your workstation thru the headnode to the assigned compute node, in this example it is alpha001 
      Your Workstation: ssh sdough01@login.cluster.tufts.edu -L 8888:localhost:8888 ssh alpha001 -L 8888:localhost:8888

    2. Point the browser on your workstation to http://localhost:8888/    and the jupyter web interface should come up.

    3. Do computation, do science!
  5. Exit jupyter
    1. Code Block
      titleDon't forget to exit jupyter so it isn't taking up resources
      The Jupyter Notebook is running at: http://localhost:8888/
Shutdown this notebook server (y/[n])? y
[C 14:19:54.199 NotebookApp] Shutdown confirmed
[I 14:19:54.199 NotebookApp] Shutting down kernels

Project Home Page

...

software library for numerical computation using data flow graphs. Nodes in the graph represent mathematical operations, while the graph edges represent the multidimensional data arrays (tensors) communicated between them. The flexible architecture allows you to deploy computation to one or more CPUs or GPUs in a desktop, server, or mobile device with a single API.

TensorFlow was originally developed by researchers and engineers working on the Google Brain Team within Google's Machine Intelligence research organization for the purposes of conducting machine learning and deep neural networks research, but the system is general enough to be applicable in a wide variety of other domains as well." 

TensorFlow site has tutorials and other resources for using this application.  TensorFlow has an API for python and C++.
There are two modules for tensorflow on the cluster for python and both  are gpu aware. Use the gpu partition to take advantage of the speedup that the GPUs provide.
tensorflow/11-python2.7
tensorflow/11-python3.5

You must use python3 instead of python if using the tensorflow/11-python3.5 module.

Code Block
$module load tensorflow/11-python3.5
$python3
Python 3.5.0 (default, Nov  4 2015, 11:43:11)  
[GCC 4.4.7 20120313 (Red Hat 4.4.7-4)] on linux 
Type "help", "copyright", "credits" or "license" for more information. 
>>>import tensorflow as tf
>>> ( more code)