isl-org/MultiObjectiveOptimization
PyTorch implementation of multi-objective optimization algorithms for multi-task neural network learning, from NeurIPS 2018.
Not currently ranked — collecting fresh signals.
star history
This repository provides the PyTorch implementation of the MGDA_UB algorithm from the NeurIPS 2018 paper “Multi-Task Learning as Multi-Objective Optimization” by Sener and Koltun. The code implements both Frank-Wolfe and projected gradient descent optimization methods for training neural networks with multiple objectives. A generic numpy-only version is also provided for portability to other deep learning frameworks.
Frequently asked
- What is isl-org/MultiObjectiveOptimization?
- PyTorch implementation of multi-objective optimization algorithms for multi-task neural network learning, from NeurIPS 2018.
- Is MultiObjectiveOptimization open source?
- Yes — isl-org/MultiObjectiveOptimization is open source, released under the MIT license.
- What language is MultiObjectiveOptimization written in?
- isl-org/MultiObjectiveOptimization is primarily written in Python.
- How popular is MultiObjectiveOptimization?
- isl-org/MultiObjectiveOptimization has 1.1k stars on GitHub.
- Where can I find MultiObjectiveOptimization?
- isl-org/MultiObjectiveOptimization is on GitHub at https://github.com/isl-org/MultiObjectiveOptimization.