pittisl/ElasticTrainer

Code for paper "ElasticTrainer: Speeding Up On-Device Training with Runtime Elastic Tensor Selection" (MobiSys'23)

31
/ 100
Emerging

This project helps machine learning practitioners speed up the training of neural networks directly on edge devices like Raspberry Pi or Nvidia Jetson TX2. It takes your existing TensorFlow model and dataset, and outputs a trained model that runs significantly faster on-device, potentially doubling the speed, without losing accuracy. It's designed for embedded systems developers or ML engineers deploying models to resource-constrained hardware.

No commits in the last 6 months.

Use this if you need to train image classification models faster on resource-limited embedded devices while maintaining model accuracy.

Not ideal if you are working with PyTorch models or primarily training large language models (LLMs), although a related project, GreenTrainer, addresses LLM fine-tuning.

edge-ai embedded-systems on-device-training neural-network-optimization computer-vision
Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 5 / 25
Maturity 16 / 25
Community 10 / 25

How are scores calculated?

Stars

14

Forks

2

Language

Python

License

MIT

Last pushed

Nov 01, 2023

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/pittisl/ElasticTrainer"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.