https://github.com/MachineLearningSystem/Fluid

A Generic Resource-Aware Hyperparameter Tuning Execution Engine
https://github.com/MachineLearningSystem/Fluid

Last synced: about 2 months ago
JSON representation

A Generic Resource-Aware Hyperparameter Tuning Execution Engine

• Host: GitHub
• URL: https://github.com/MachineLearningSystem/Fluid
• Owner: MachineLearningSystem
• License: apache-2.0
• Fork: true (SymbioticLab/Fluid)
• Created: 2023-01-25T13:36:03.000Z (over 1 year ago)
• Default Branch: main
• Last Pushed: 2022-01-08T22:23:01.000Z (over 2 years ago)
• Last Synced: 2024-07-27T17:55:15.608Z (2 months ago)
• Homepage:
• Size: 292 KB
• Stars: 0
• Watchers: 0
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

• awesome-AI-system - Fluid: Resource-Aware Hyperparameter Tuning Engine MLSYS'21

README

        
# Fluid: Resource-Aware Hyperparameter Tuning Engine

[![PyPI version](https://img.shields.io/pypi/v/fluidexec.svg)](https://pypi.org/project/fluidexec)

[![Python package](https://github.com/SymbioticLab/Fluid/actions/workflows/python-package.yml/badge.svg?event=release)](https://github.com/SymbioticLab/Fluid/actions/workflows/python-package.yml)

`Fluid` is an alternative [Ray](https://ray.io) executor that intelligently manages trial executions on behalf of hyperparameter tuning algorithms, in order to increase the resource utilization, and improve end-to-end makespan.

This is the implementation of our MLSys'21 [paper](https://symbioticlab.org/publications/#/venue:MLSys) "Fluid: Resource-Aware Hyperparameter Tuning Engine".

## Get Started

First follow the [instruction](https://docs.ray.io/en/master/tune/index.html) in Ray Tune to setup the Ray cluster and a tuning environment as usual.

Then make sure [Nvidia MPS](https://docs.nvidia.com/deploy/mps/index.html#topic_6_1) is correctly setup on all worker nodes.

`Fluid` itself is a normal python package that can be installed by `pip install fluidexec`. Note that the pypi package name is `fluidexec` because the name `fluid` is already taken.

To use `Fluid` in Ray Tune, pass an instance of it as the trial executor to `tune.run`. It should work with any other schedulers:

```python

from fluid.fliud_executor import FluidExecutor

tune.run(

    MyTrainable,

    trial_executor=FluidExecutor(),

    ...

)

```

## Reproduce Experiments

See the README in [`workloads`](workloads/) for more information.

## Notes

Please consider to cite our paper if you find this useful in your research project.

```bibtex

@inproceedings{fluid:mlsys21,

    author    = {Peifeng Yu and Jiachen Liu and Mosharaf Chowdhury},

    booktitle = {MLSys},

    title     = {Fluid: Resource-Aware Hyperparameter Tuning Engine},

    year      = {2021},

}

```