Research map / Architectures and efficiency / Pruning and efficient tuning
Calibrating the Rigged Lottery: Making All Tickets Reliable
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 |
|---|---|---|---|
| Two masks (deterministic + random) with weight & mask averaging (CigL) vs RigL | sparse ResNet-50 · CIFAR-10 and CIFAR-100, 80-99% sparsity | test accuracy: +0.7 to +4.0 points on CIFAR-100; +0.7 to +1.1 on CIFAR-10 | Helped |
| Two masks (deterministic + random) with weight & mask averaging (CigL) vs RigL | sparse ResNet-50 · CIFAR-10 and CIFAR-100, 80-99% sparsity | ECE | Helped |
| Adding weight dropout to RigL | RigL sparse ResNet-50 · CIFAR-10 at 80-99% sparsity | test accuracy: about -0.8 to -1.3 points vs RigL | Hurt |
| Adding MC dropout to RigL | RigL sparse ResNet-50 · CIFAR-10 at 80-99% sparsity | ECE | Mixed |
| Removing random mask (CigL w/o RM) (ablation) | CigL · CIFAR-100 ablation at 80-99% sparsity | ECE | Hurt |
| Removing weight & mask averaging (CigL w/o WMA) (ablation) | CigL · sparsities 95% and 99% | test accuracy and ECE | Helped |
| Increasing sparsity in sparse training (ECE vs sparsity) | sparse ResNet-50 trained with RigL · CIFAR-100 | ECE | Hurt |
Closest papers by meaning
- Distributionally Robust Ensemble of Lottery Tickets Towards Calibrated Sparse Network Training (NeurIPS 2023 · similarity 0.90)
- Rigging the Lottery: Making All Tickets Winners (ICML 2020 · similarity 0.90)
- Sparse Training from Random Initialization: Aligning Lottery Ticket Masks using Weight Symmetry (ICML 2025 · similarity 0.89)
- Winning the Lottery with Continuous Sparsification (NeurIPS 2020 · similarity 0.88)
- No Free Prune: Information-Theoretic Barriers to Pruning at Initialization (ICML 2024 · similarity 0.87)
- Find A Winning Sign: Sign Is All We Need to Win the Lottery (ICLR 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.