Research map / Learning theory and algorithms / Sample complexity
On the Modularity of Hypernetworks
Tomer Galanti, Lior Wolf
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use a hypernetwork instead of an embedding method | Synthetic function approximation · Synthetic target functions | MSE | Improved |
| Increase the hypernetwork generator depth (ablation) | Hypernetwork · Synthetic target functions | MSE | Improved |
| Increase the embedding network depth (ablation) | Embedding method · Synthetic target functions | MSE | No clear effect |
| Increase the embedding dimension (ablation) | Embedding method · Synthetic target functions | MSE | No clear effect |
| Use a hypernetwork instead of an embedding method | Image rotation prediction · MNIST and CIFAR10 | classification error | Improved |
| Increase the hypernetwork generator depth (ablation) | Hypernetwork · Image rotation prediction | classification error | Improved |
| Increase the embedding network depth (ablation) | Embedding method · Image rotation prediction | classification error | No clear effect |
| Increase the embedding dimension (ablation) | Embedding method · Image rotation prediction | classification error | No clear effect |
| Use a hypernetwork instead of an embedding method | Hypernetwork · CIFAR10 image colorization | error rate | Improved |
| Increase the hypernetwork generator depth (ablation) | Hypernetwork · CIFAR10 image colorization | error rate | Improved |
| Increase the embedding network depth (ablation) | Embedding method · Synthetic target-function experiments | MSE | No clear effect |
Closest papers by meaning
- Breaking Neural Network Scaling Laws with Modularity (ICLR 2025 · similarity 0.88)
- Neural networks are more modular than single neurons suggest (NeurIPS 2026 · similarity 0.87)
- Uncovering a Universal Abstract Algorithm for Modular Addition in Neural Networks (NeurIPS 2025 · similarity 0.87)
- Discovering modular solutions that generalize compositionally (ICLR 2024 · similarity 0.86)
- Magnitude Invariant Parametrizations Improve Hypernetwork Learning (ICLR 2024 · similarity 0.86)
- Sample complexity of data-driven tuning of model hyperparameters in neural networks with structured parameter-dependent dual function (NeurIPS 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
