Argo is a library for deep learning algorithms based on TensorFlow and Sonnet. The library allows you to train different models (feed-forwards neural networks for regression and classification problems, autoencoders and variational autoencoders, Bayesian neural networks, Helmholtz machines, etc) by specifying their parameters as well as the network topologies in a configuration file. The models can then be trained in parallel in presence of multiple GPUs. The library is easy to expand for alternative models and training algorithms, as well as for different network topologies.
Requirements (stable):
- tensorflow-datasets 1.2.0
- tensorflow-estimator 1.14.0
- tensorflow-gpu 1.14.0
- tensorflow-metadata 0.14.0
- tensorflow-probability 0.7.0
- sonnet 1.32
- torchfile
- seaborn
- matplotlib
- numpy
Or:
pip install -r requirements.txt
To run the examples provided in the framework (or new ones) one can choose between three separate modes of running:
- single:
Runs a single instance of the configuration file
python argo/runTraining.py configFile.conf single
- pool:
Runs a muliple experiments (if defined) from the configuration file
python argo/runTraining.py configFile.conf pool
python argo/runTraining.py examples/MNISTcontinuous.conf single
python argo/runTraining.py examples/ThreeByThree.conf single
python argo/runTraining.py examples/GTSRB.conf single
How to run the code:
python3 argo/runTrainingVAE.py configFile.conf single/pool/stats
See ConfOptions.conf in examples/ for details regarding meaning of parameters and logging options.
In alphabetical order.
- Luigi Malagò
- Csongor Varady
- Riccardo Volpi
- Alexandra Albu
- Cristian Alecsa
- Norbert Cristian Bereczki
- Robert Colt
- Delia Dumitru
- Alina Enescu
- Petru Hlihor
- Hector Javier Hortua
- Uddhipan Thakur
- Ria Arora
- Dimitri Marinelli
- Titus Nicolae
- Alexandra Peste
- Marginean Radu
- Septimia Sarbu
The library has been developed in the context of the DeepRiemann project, co-funded by the European Regional Development Fund and the Romanian Government through the Competitiveness Operational Programme 2014-2020, Action 1.1.4, project ID P_37_714, SMIS code 103321, contract no. 136/27.09.2016.