Research map / Architectures and efficiency / RNNs and normalization
A Modern Look at the Relationship between Sharpness and Generalization
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 |
|---|---|---|---|
| use worst-case l_inf adaptive sharpness with logit normalization instead of without | ViT-B/16 trained from scratch on ImageNet-1k · ImageNet-1k and OOD variants | Kendall tau with test error | Mixed |
| sharpness measured without logit normalization (worst-case l_inf) | CLIP ViT-B/32 fine-tuned on ImageNet · OOD distribution shifts (ImageNet-A, ImageNet-R) | Kendall tau with test error | Hurt |
| adding logit normalization to adaptive sharpness (ablation) | ResNet-18 on CIFAR-10 · CIFAR-10 from scratch, adaptive worst-case l_inf | Kendall tau with test error | Mixed |
| learning rate: sharpness negatively correlates with learning rate | ResNet and ViT trained on CIFAR-10 · CIFAR-10 from scratch | correlation of sharpness with learning rate | No clear effect |
| sharpness evaluated on diagonal linear networks with whitened data: average-case adaptive sharpness vs worst-case (ablation) | diagonal linear networks sparse regression · sparse regression | Kendall tau with test loss | Mixed |
| BERT MNLI: no measured sharpness variant predicts generalization | BERT fine-tuned on MNLI · MNLI and HANS | Kendall tau with test error | No clear effect |
Closest papers by meaning
- Towards Robust Out-of-Distribution Generalization Bounds via Sharpness (ICLR 2024 · similarity 0.90)
- Friendly Sharpness-Aware Minimization (CVPR 2024 · similarity 0.89)
- Changing the Training Data Distribution to Reduce Simplicity Bias Improves In-distribution Generalization (NeurIPS 2024 · similarity 0.89)
- Sharpness-Aware Minimization Enhances Feature Quality via Balanced Learning (ICLR 2024 · similarity 0.88)
- Sharpness-Aware Minimization: General Analysis and Improved Rates (ICLR 2025 · similarity 0.88)
- Sharpness-Aware Minimization Efficiently Selects Flatter Minima Late In Training (ICLR 2025 · similarity 0.88)
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.