TorchJD
A PyTorch-based multi-objective optimization library that applies Jacobian descent to handle multiple conflicting loss functions. It is suited for multi-task learning and meta-learning scenarios and has been incorporated into the official PyTorch ecosystem.
Timeline (last 90 days)
The PyTorch ecosystem working group announced the addition of 10 new projects to the PyTorch Ecosystem Landscape: Perforated, AReaL, TorchJD, RLinf, Miles, SMG, FiftyOne, TokenSpeed, VisualTorch, and TorchSurv.
TorchJD (Torch Jacobian Descent) is a project targeting multi-objective optimization scenarios, based on Jacobian matrix descent methods.
All Facts (2)
TorchJD (Torch Jacobian Descent) is a project targeting multi-objective optimization scenarios, based on Jacobian matrix descent methods.
50%UnverifiedThe PyTorch ecosystem working group announced the addition of 10 new projects to the PyTorch Ecosystem Landscape: Perforated, AReaL, TorchJD, RLinf, Miles, SMG, FiftyOne, TokenSpeed, VisualTorch, and TorchSurv.
50%