Welcome to torchcache!

torchcache offers an effortless way to cache PyTorch module outputs on-the-fly. By caching the outputs of a module, you can save time and resources when running the same pre-trained model on the same inputs multiple times.

Note that gradients will not flow through the cached outputs.

Citation

If you use our work, please consider citing our paper:

@inproceedings{akbiyik2023routeformer,
    title={Leveraging Driver Field-of-View for Multimodal Ego-Trajectory Prediction},
    author={M. Eren Akbiyik, Nedko Savov, Danda Pani Paudel, Nikola Popovic, Christian Vater, Otmar Hilliges, Luc Van Gool, Xi Wang},
    booktitle={International Conference on Learning Representations},
    year={2025}
}

Table of Contents:

• Installation
  • Assumptions
• Usage
  • Limitations
  • Persistent caching
  • Adjusting cache size
  • Optimizing cache size
  • Setting parameters directly in the module
  • Selectively enabling torchcache for module instances
  • When is cache invalidated?
  • Avoiding cache invalidation
• How it works
  • Automatic cache management
  • Tensor hashing
• Environment variables
• Contribution
• API documentation
  • torchcache.torchcache
  • torchcache.set_logger_config