WeiChengTseng/Pytorch-PCGrad
A PyTorch reimplementation of PCGrad (gradient surgery) for multi-task learning.
Not currently ranked — collecting fresh signals.
star history
This repository provides a PyTorch implementation of the gradient surgery technique for multi-task learning. It wraps standard PyTorch optimizers with a PCGrad class that modifies conflicting gradients between tasks before applying updates. The implementation supports multi-task scenarios including Multi-MNIST and is designed to prevent gradient interference when training a single model on multiple objectives.
Frequently asked
- What is WeiChengTseng/Pytorch-PCGrad?
- A PyTorch reimplementation of PCGrad (gradient surgery) for multi-task learning.
- Is Pytorch-PCGrad open source?
- Yes — WeiChengTseng/Pytorch-PCGrad is open source, released under the BSD-3-Clause license.
- What language is Pytorch-PCGrad written in?
- WeiChengTseng/Pytorch-PCGrad is primarily written in Python.
- How popular is Pytorch-PCGrad?
- WeiChengTseng/Pytorch-PCGrad has 403 stars on GitHub.
- Where can I find Pytorch-PCGrad?
- WeiChengTseng/Pytorch-PCGrad is on GitHub at https://github.com/WeiChengTseng/Pytorch-PCGrad.