https://github.com/MachineLearningSystem/ModelKeeper

A Cluster-Wide Model Manager to Accelerate DNN Training via Automated Training Warmup
https://github.com/MachineLearningSystem/ModelKeeper

Last synced: about 1 month ago
JSON representation

A Cluster-Wide Model Manager to Accelerate DNN Training via Automated Training Warmup

• Host: GitHub
• URL: https://github.com/MachineLearningSystem/ModelKeeper
• Owner: MachineLearningSystem
• Fork: true (SymbioticLab/ModelKeeper)
• Created: 2022-10-10T07:22:02.000Z (about 2 years ago)
• Default Branch: master
• Last Pushed: 2022-10-10T03:59:10.000Z (about 2 years ago)
• Last Synced: 2024-08-02T19:33:31.983Z (5 months ago)
• Homepage:
• Size: 16.2 MB
• Stars: 0
• Watchers: 0
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

• awesome-AI-system - ModelKeeper: Accelerating DNN Training via Automated Training Warmup NSDI'23

README

        
# ModelKeeper

This repository contains the evvaluation artifacts of our NSDI '23 paper "[ModelKeeper: Accelerating DNN Training via Automated Training Warmup](https://www.usenix.org/conference/nsdi23/presentation/lai)".

**ModelKeeper has been merged as part of [FedScale](https://github.com/SymbioticLab/FedScale) and is actively maintained there. Please try it!**

# Overview

* [Getting Started](#getting-started)

* [Run Experiments](#run-experiments)

* [Repo Structure](#repo-structure)

* [Contact](#contact)

# Getting Started

Our ```install.sh``` will install the following automatically:

* Anaconda Package Manager

* CUDA 10.2

Note: if you prefer different versions of conda and CUDA, please check  comments in `install.sh` for details.

Run the following commands to install ModelKeeper. 

```

source install.sh 

pip install -e .

```

# Run Experiments

# Repo Structure

```

Repo Root

|---- modelkeeper   # Core implementation (e.g., Matcher).

|---- evals         # MK support for different training backends

    |---- ray_tune      # Ray experiments

    |---- nni           # Retiarii experiments

|---- examples      # Toy experiments of model transformation

```

# Notes

please consider to cite our paper if you use the code or data in your research project.

```bibtex

@inproceedings{modelkeeper-nsdi23,

  title={ModelKeeper: Accelerating DNN Training via Automated Training Warmup},

  author={Fan Lai and Yinwei Dai and Harsha V. Madhyastha and Mosharaf Chowdhury},

  booktitle={USENIX Symposium on Networked Systems Design and Implementation (NSDI)},

  year={2023}

}

```

# Contact

Fan Lai ([email protected]) and Yinwei Dai ([email protected]).