SimplexLab/TorchJD

Library for Jacobian descent with PyTorch. It enables the optimization of neural networks with multiple losses (e.g. multi-task learning).

/ 100

Established

This library helps machine learning engineers and researchers efficiently train neural networks that have multiple goals or "loss functions" simultaneously, like in multi-task learning. Instead of combining these goals into a single average and risking poorer performance on some, it takes all individual loss functions as input and produces an optimized set of model parameters where each goal is better balanced. It's designed for anyone building advanced AI models with PyTorch.

306 stars. Available on PyPI.

Use this if you are training neural networks with PyTorch and encounter scenarios where you need to optimize for multiple conflicting objectives or loss functions at the same time, such as in multi-task learning.

Not ideal if your neural network only has a single objective to optimize or if you are not working with PyTorch.

multi-task learning neural network training machine learning research model optimization deep learning

Maintenance 10 / 25

Adoption 10 / 25

Maturity 25 / 25

Community 10 / 25

How are scores calculated?

Stars

306

Forks

Language

Python

License

MIT

Related frameworks

nschaetti/EchoTorch

A Python toolkit for Reservoir Computing and Echo State Network experimentation based on...

metaopt/torchopt

TorchOpt is an efficient library for differentiable optimization built upon PyTorch.

gpauloski/kfac-pytorch

Distributed K-FAC preconditioner for PyTorch

opthub-org/pytorch-bsf

PyTorch implementation of Bezier simplex fitting

pytorch/xla

Enabling PyTorch on XLA Devices (e.g. Google TPU)

Explore ML Frameworks

All categories Trending ML Framework directory Insights