This is a tutorial repository on how to get started with the MLE-Infrastructure:
- Google Colab walkthrough of the sub-packages:
- Google Colab walkthrough of the
mle-toolbox
: - Example template project repository:
mle-project
.
- 12/2021: Sprekeler lab group presentation slides
mle-logging
: A Lightweight Logger for ML Experiments πmle-scheduler
: A Lightweight Cluster/Cloud VM Job Management Tool πmle-hyperopt
: A Lightweight Hyperparameter Optimization Tool πmle-monitor
: A Lightweight Experiment & Resource Monitoring Tool πΊmle-toolbox
: A Lightweight Tool to Manage Distributed ML Experiments π
Job Types | Description | |
---|---|---|
π Single-Objective | multi-configs , hyperparameter-search |
Core experiment types. |
π Multi-Objective | hyperparameter-search |
Multi-objective tuning. |
π Multi Bash | multi-configs |
Bash-based jobs. |
π Quadratic PBT | hyperparameter-search |
PBT on toy quadratic surrogate. |
π Hyperband | hyperparameter-search |
Hyperband on toy polynomial problem. |
Description | Colab | |
---|---|---|
π Getting Started | Get started with the toolbox. | |
π Subpackages | Get started with the toolbox subpackages. | |
π MLExperiment |
Introduction to MLExperiment wrapper. |
|
π Evaluation | Evaluation of gridsearch results. | |
π GIF Animations | Walk through a set of animation helpers. | |
π Testing | Perform hypothesis tests on logs. |
Here is a set of published papers using parts of the mle-toolbox
experiment infrastructure to generate their results:
- Vischer*, M. A., Lange*, R. T., & Sprekeler, H. (2021). On Lottery Tickets and Minimal Task Representations in Deep Reinforcement Learning. arXiv preprint arXiv:2105.01648.
- Lange, R. T., & Sprekeler, H. (2020). Learning not to learn: Nature versus nurture in silico. Published as a conference paper at AAAI 2022.
And here is a set of blogs and other projects using parts of the mle-toolbox
experiment infrastructure: