TAROT: Targeted Data Selection via Optimal Transport

December 12, 2024 · View on GitHub

TAROT: Targeted Data Selection via Optimal Transport

Lan Feng, Fan Nie, Yuejiang Liu, Alexandre Alahi

Visual Intelligence for Transportation Lab, EPFL

Stanford University

Paper

System Overview

🛠 Environment Setup

To get started with TAROT, follow these steps:


🚀 Quick Start

🚗Motion Prediction

📖Instruction Tuning

🏞️Semantic Segmentation

📊Qualitative Results of WFD

system

Acknowledgement

This repo relies on the TRAK implementation to compute the projected gradients. We are immensely grateful to the authors of that project.

For citation:

@article{feng2024tarot,
  title={TAROT: Targeted Data Selection via Optimal Transport},
  author={Feng, Lan and Nie, Fan and Liu, Yuejiang and Alahi, Alexandre},
  journal={arXiv preprint arXiv:2412.00420},
  year={2024}
}