https://github.com/sparticlesteve/cosmoflow-benchmark

Benchmark implementation of CosmoFlow in TensorFlow Keras
https://github.com/sparticlesteve/cosmoflow-benchmark

Last synced: 11 days ago
JSON representation

Benchmark implementation of CosmoFlow in TensorFlow Keras

• Host: GitHub
• URL: https://github.com/sparticlesteve/cosmoflow-benchmark
• Owner: sparticlesteve
• License: apache-2.0
• Created: 2019-06-11T14:46:53.000Z (over 5 years ago)
• Default Branch: master
• Last Pushed: 2024-02-07T22:43:41.000Z (9 months ago)
• Last Synced: 2024-08-01T16:46:43.217Z (3 months ago)
• Language: Jupyter Notebook
• Size: 3.26 MB
• Stars: 19
• Watchers: 7
• Forks: 10
• Open Issues: 4
• Metadata Files:
  • Readme: README.md
  • License: LICENSE.md

Awesome Lists containing this project

README

        
# CosmoFlow TensorFlow Keras benchmark implementation

**WARNING: this repo is old. For the latest MLPerf HPC reference implementation of cosmoflow, see https://github.com/mlcommons/hpc/tree/main/cosmoflow**

This is a an implementation of the

[CosmoFlow](https://arxiv.org/abs/1808.04728) 3D convolutional neural network

for benchmarking. It is written in TensorFlow with the Keras API and uses

[Horovod](https://github.com/horovod/horovod) for distributed training.

You can find the previous TensorFlow implementation which accompanied the CosmoFlow paper at

https://github.com/NERSC/CosmoFlow

## Datasets

The dataset we use for this benchmark comes from simulations run by the

ExaLearn group and hosted at NERSC. The following web portal describes the

technical content of the dataset and provides links to the raw data.

https://portal.nersc.gov/project/m3363/

For this benchmark we currently use a preprocessed version of the dataset which

generates crops of size (128, 128, 128, 4) and stores in TFRecord format.

This preprocessing is done using the [prepare.py](prepare.py) script included

in this package. We describe here how to get access to this processed dataset,

but please refer to the ExaLearn web portal for additional technical details.

Globus is the current recommended way to transfer the dataset locally.

There is a globus endpoint at:

https://app.globus.org/file-manager?origin_id=d0b1b73a-efd3-11e9-993f-0a8c187e8c12&origin_path=%2F

The contents are also available via HTTPS at:

https://portal.nersc.gov/project/dasrepo/cosmoflow-benchmark/

### MLPerf HPC v1.0 preliminary dataset

Preprocessed TFRecord files are available in a 1.7TB tarball named

`cosmoUniverse_2019_05_4parE_tf_v2.tar`. It contains subfolders for

train/val/test file splits.

In this preparation, there are 524288 samples for training and 65536 samples for

validation. The TFRecord files are written with gzip compression to reduce total

storage size.

### MLPerf HPC v0.7 dataset

The pre-processed dataset in TFRecord format is in the

`cosmoUniverse_2019_05_4parE_tf` folder, which contains training and validation

subfolders. There are 262144 samples for training and 65536 samples

for validation/testing. The combined size of the dataset is 5.1 TB.

For getting started, there is also a small tarball (179MB) with 32 training

samples and 32 validation samples, called `cosmoUniverse_2019_05_4parE_tf_small.tgz`.

## Running the benchmark

Submission scripts are in `scripts`. YAML configuration files go in `configs`.

### Running at NERSC

`sbatch -N 64 scripts/train_cori.sh`