Torchhd is a Python library for Hyperdimensional Computing and Vector Symbolic Architectures
-
Updated
Jun 19, 2025 - Python
Torchhd is a Python library for Hyperdimensional Computing and Vector Symbolic Architectures
Local, gradient-free neuro-symbolic memory engine combining Hyperdimensional Computing (HDC/VSA), Hebbian plasticity, and graph triples for offline AI.
Hyperdimensional Computing Library for building Vector-Symbolic Architectures in Python 3
Hyperdimensional computing in JAX, with statistical guarantees.
Boolean Hypervectors with various operators for experiments in hyperdimensional computing (HDC).
Intro to Hyperdimensional Computing (HDC) using a tea dataset
A computational foundation for AGI, based on hyperdimensional computing and framed in set theory.
HD3C: A lightweight classification framework designed for low-power devices.
Holographic vectors you can compute with. Bind structure, bundle sets, unbind components cross NumPy, PyTorch, and JAX.
Your language model forgets everything. Sillage gives it a 4 MB memory that persists across sessions, no gradients, no fine-tuning, no growing index. Beats an unbounded kNN-LM at 1/13 the storage (perplexity 31 → 17). CPU-only, fully reproducible, four preprints with Zenodo DOIs.
Repository for HYPERDOA: Robust and Efficient DoA Estimation using Hyperdimensional Computing
Contributions to my area of research as a research assistant at ULL in Hyperdimensional Computing (HDC)
Sub-linear knowledge retrieval via quantum-inspired hyperdimensional folded space (0.88ms @ 100% accuracy)
NeuroHDC: A framework for high-efficiency Edge-AI ECG classification using Spiking-Hyperdimensional Computing (HDC). Features benchmarking against CNN/LSTM baselines across MIT-BIH, PTB, STT, and PhysioNet datasets to enable low-power cardiac diagnostics.
A fixed-size holographic (FHRR) associative memory that ages: decay, reinforcement, exact contradiction, dated queries. With a reproducible capacity benchmark.
CPU-native neuro-symbolic cognitive architecture on 10,240-bit hypervectors: spiking perception, memory, reasoning and language, with glass-box decision traces. Python and Rust.
Training-free behavioral anomaly detector for Mantle DeFi. HDC engine + Z.ai explainability. Zero false positives, on-chain proof.
Neural "networks" with zero multiplications at inference. AdderNet for embedded systems.
Local, deterministic semantic search built from textbook parts - measured, defeats included. A didactic experiment that runs in production.
To associate your repository with the hyperdimensional-computing topic, visit your repo's landing page and select "manage topics."