SimplexLab/TorchJD

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

What it solves

TorchJD addresses the challenge of training neural networks with multiple loss functions, which often occurs in multi-task learning. It provides tools to handle conflicts between different losses, preventing one task from dominating others or conflicting gradients from hindering optimization.

Related

  • Project
    UM-ARM-Lab/pytorch_kinematics

    A parallel and differentiable robotics kinematics library for PyTorch that provides fast forward and inverse kinematics and Jacobian calculations.

  • Project
    metaopt/torchopt

    An efficient PyTorch library for differentiable optimization that supports explicit, implicit, and zero-order differentiation for bi-level optimization and meta-learning.

  • Project
    facebookresearch/multimodal

    A PyTorch library for training state-of-the-art multimodal models at scale, providing modular building blocks and pretrained weights for models like CLIP and BLIP-2.

  • Project
    Xilinx/brevitas

    A PyTorch library for neural network quantization that supports both post-training quantization and quantization-aware training to optimize models for efficiency.

  • Project
    pytorch/ao

    A PyTorch-native library for model optimization that provides quantization and sparsity tools to accelerate training and inference while reducing memory usage.