Research map / Architectures and efficiency / Neural architecture search
NAS-Bench-1Shot1: Benchmarking and Dissecting One-shot Neural Architecture Search
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Tune one-shot NAS hyperparameters with BOHB using NAS-Bench-101 validation error as the objective | DARTS, GDAS, and PC-DARTS · NAS-Bench-1Shot1 search spaces 1–3 | test regret: up to a factor of 10 | Helped |
| Include learning rate in the BOHB configuration space (ablation) | DARTS, GDAS, and PC-DARTS · NAS-Bench-1Shot1 search spaces | test regret | Helped |
| Tune in CS1 without learning rate (ablation) | DARTS, GDAS, and PC-DARTS · NAS-Bench-1Shot1 search spaces | test regret | Hurt |
| Tune in the larger CS3 configuration space (ablation) | DARTS, GDAS, and PC-DARTS · NAS-Bench-1Shot1 search spaces | test regret | Mixed |
| Enable Cutout during architecture search (ablation) | DARTS, GDAS, and PC-DARTS · NAS-Bench-1Shot1 search spaces 1–3 | test regret | Hurt |
| Increase L2 regularization (ablation) | GDAS and PC-DARTS · NAS-Bench-1Shot1 search spaces 1–3 | test regret | Helped |
| Increase L2 regularization beyond the default (ablation) | DARTS · NAS-Bench-1Shot1 search spaces 1–3 | test regret | Hurt |
| Transfer the best Search space 3 hyperparameter configuration (ablation) | DARTS · Search space 1, 50 search epochs | test regret: 1.165e-2 ± 0.162e-2 -> 1.103e-2 ± 0.220e-2 | Helped |
| Transfer the best Search space 3 hyperparameter configuration (ablation) | DARTS · Search space 2, 50 search epochs | test regret: 0.933e-2 ± 0.138e-2 -> 1.393e-2 ± 0.635e-2 | Hurt |
| Transfer the best Search space 3 hyperparameter configuration (ablation) | GDAS · Search space 1, 50 search epochs | test regret: 1.155e-2 ± 0.0 -> 4.619e-2 ± 4.718e-2 | Hurt |
| Transfer the best Search space 3 hyperparameter configuration (ablation) | GDAS · Search space 2, 50 search epochs | test regret: 0.871e-2 ± 0.0 -> 1.019e-2 ± 0.277e-2 | Hurt |
| Transfer the best Search space 3 hyperparameter configuration (ablation) | PC-DARTS · Search spaces 1 and 2, 50 search epochs | test regret: 1.246e-2 ± 0.173e-2 -> 1.107e-2 ± 0.129e-2; 1.622e-2 ± 0.364e-2 -> 1.069e-2 ± 0.123e-2 | Helped |
Closest papers by meaning
- Generalizing Few-Shot NAS with Gradient Matching (ICLR 2022 · similarity 0.91)
- Few-Shot Neural Architecture Search (ICML 2021 · similarity 0.91)
- NAS-Bench-201: Extending the Scope of Reproducible Neural Architecture Search (ICLR 2020 · similarity 0.91)
- NAS-Bench-101: Towards Reproducible Neural Architecture Search (ICML 2019 · similarity 0.91)
- Efficient Few-Shot Neural Architecture Search by Counting the Number of Nonlinear Functions (AAAI 2025 · similarity 0.90)
- HEP-NAS: Towards Efficient Few-shot Neural Architecture Search via Hierarchical Edge Partitioning (AAAI 2025 · similarity 0.89)
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.