Tensorflow¶

pagelastupdated:
supportlevel:	A
	2020-05-15
maintainer:

Tensorflow is a commonly used Python package for deep learning.

Basic usage¶

First, check the tutorials up to and including GPU computing.

If you plan on using NVIDIA’s containers to run your model, please check the page about NVIDIA’s singularity containers.

The basic way to use is via the Python in the anaconda module. The versions with -tf2 (the default ones) have Tensorflow 2 installed. If you use module spider anaconda, you can see a -tf1 version available.

Warning

Older versions of Tensorflow were CPU-only or GPU-only

With older versions of tensorflow (<1.15.0), you have to decide at install time if you want a version that runs on CPUs or GPUs. This means that we can’t install it for everyone and expect it to work everywhere - you have to load something different if you want it to run on login node/regular nodes (probably for testing) or GPU nodes. The old -cpu and -gpu versions in the anaconda2- and anaconda3-modules denoted this.

From tensorflow versions >= 1.15.0, they solved this problem (thankfully)

Don’t load any additional CUDA modules, anaconda includes everything.

If you use GPUs, you need --constraint='kepler|pascal|volta' in order to select a GPU new enough to run tensorflow. (Note that as we get newer cards, this will need further updating).

Simple Tensorflow/Keras model¶

Let’s run the MNIST example from Tensorflow’s tutorials:

model = tf.keras.models.Sequential([
  tf.keras.layers.Flatten(input_shape=(28, 28)),
  tf.keras.layers.Dense(512, activation=tf.nn.relu),
  tf.keras.layers.Dropout(0.2),
  tf.keras.layers.Dense(10, activation=tf.nn.softmax)
])

The full code for the example is in tensorflow_mnist.py. One can run this example with srun:

wget https://raw.githubusercontent.com/AaltoSciComp/scicomp-docs/master/triton/examples/tensorflow/tensorflow_mnist.py
module load anaconda
srun -t 00:15:00 --gres=gpu:1 python tensorflow_mnist.py

or with sbatch by submitting tensorflow_mnist.sh:

#!/bin/bash
#SBATCH --gres=gpu:1
#SBATCH --time=00:15:00

module load anaconda

python tensorflow_mnist.py

Do note that by default Keras downloads datasets to $HOME/.keras/datasets.

Running simple Tensorflow/Keras model with NVIDIA’s containers¶

Let’s run the MNIST example from Tensorflow’s tutorials:

model = tf.keras.models.Sequential([
  tf.keras.layers.Flatten(input_shape=(28, 28)),
  tf.keras.layers.Dense(512, activation=tf.nn.relu),
  tf.keras.layers.Dropout(0.2),
  tf.keras.layers.Dense(10, activation=tf.nn.softmax)
])

The full code for the example is in tensorflow_mnist.py. One can run this example with srun:

wget https://raw.githubusercontent.com/AaltoSciComp/scicomp-docs/master/triton/examples/tensorflow/tensorflow_mnist.py
module load nvidia-tensorflow/20.02-tf1-py3
srun -t 00:15:00 --gres=gpu:1 singularity_wrapper exec python tensorflow_mnist.py

or with sbatch by submitting tensorflow_singularity_mnist.sh:

#!/bin/bash
#SBATCH --gres=gpu:1
#SBATCH --time=00:15:00

module load nvidia-tensorflow/20.02-tf1-py3

singularity_wrapper exec python tensorflow_mnist.py

Do note that by default Keras downloads datasets to $HOME/.keras/datasets.

Common problems¶

ImportError: libcuda.so.1: cannot open shared object file: No such file or directory. Older versions of GPU tensorflow can only be imported on GPU nodes (even though you’d think that you can import it and just not use the GPUs). So you can only run this code in the GPU queue. Solution for this is to use the newer anaconda-modules.
Random CUDA errors: don’t load any other CUDA modules, only anaconda. Anaconda includes the necessary libraries in compatible versions.