Hyperdimensional Computing (HDC), also known as Vector Symbolic Architectures, is an alternative computing paradigm inspired by how the brain processes information. Instead of traditional numeric computation, HDC operates on high-dimensional vectors (called hypervectors), enabling fast and noise-robust learning, often without backpropagation.Torchhd is a library for HDC, built on top of PyTorch. It provides an easy-to-use, modular framework for researchers and developers to experiment with HDC models and applications, while leveraging GPU acceleration. Torchhd aims to make prototyping and scaling HDC algorithms effortless.