hypernetwork

Package for working with hypernetworks in PyTorch.

[WWW2023] PyTorch implementation of "DUET: A Tuning-Free Device-Cloud Collaborative Parameters Generation Framework for Efficient Device Model Generalization".

Hypernetwork training considerations and implementation types in PyTorch. Includes classification and time-series examples alongside 1D GroupConv Parallelization.

Code for ICLR 2022 Paper (HyperDQN: A Randomized Exploration Method for Deep Reinforcement Learning)

AAAI'26 Oral: "WeightFlow: Learning Stochastic Dynamics via Evolving Weight of Neural Network"

Hypernetwork-Ensemble Learning of Segmentation Probability for Medical Image Segmentation with Ambiguous Labels

Database for "HyperSOR: Context-aware graph hypernetwork for salient object ranking", TPAMI 2024

Official repo for the "HyDA: A Hypernetwork Framework for Unsupervised Domain Adaptation for Medical Images" MICCAI 2025 paper.

Imaging tools CLI for preprocessing datasets before model training.

Using teacher assistant networks to distill recommender systems

Hypergraph-based Investigation of Perturbation Effects and Resilience (HIPER) provides optimized data structures and algorithms for hypernetwork analysis and attack simulation.

Hypernetwork-driven LoRA distillation from Qwen3-Coder teacher trajectories

Instant Knowledge Internalization — Convert documents into LoRA adapter weights using Text-to-LoRA hypernetworks. No fine-tuning, no gradient descent — just inference.

Zero-shot semantic HyperLoRA generation on constrained SVD manifolds

3D Real Scene Style Transfer of Neural Radiance Fields with Hash Encoding and HyperNet

Charaka Samhita Sutra for 2.8T intelligence in 6GB — 13 Ayurveda bets, 1473x compression, one file no GPU