Research map / Architectures and efficiency / Attention and transformers
Noise Stability of Transformer Models
Themistoklis Haris, Zihan Zhang, Yuichi Yoshida
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 |
|---|---|---|---|
| add noise stability regularization | decoder-only Transformer · Modular addition | iterations to emergence: 4500 -> 3300 (36% acceleration) | Improved |
| add noise stability regularization | decoder-only Transformer · Noisy k-sparse parity | training iterations to grokking: ~35% speed-up | Improved |
| add noise stability regularization | decoder-only Transformer · WikiText-2 next-token prediction | iterations to high validation accuracy: ~75% fewer iterations | Improved |
| use noise stability bound instead of average sensitivity bound | spectral tail mass prediction · GPT-2, BERT, RoBERTa, Gemma | tail mass bound: RoBERTa 0.19 -> 0.02; GPT-2 0.003 -> 0.0005 | Improved |
Closest papers by meaning
- Transformers Learn Low Sensitivity Functions: Investigations and Implications (ICLR 2025 · similarity 0.88)
- GaLileo: General Linear Relaxation Framework for Tightening Robustness Certification of Transformers (AAAI 2024 · similarity 0.87)
- Deep Active Learning with Noise Stability (AAAI 2024 · similarity 0.86)
- The Quantization Benefits of Residual-Free Transformers (NeurIPS 2026 · similarity 0.86)
- Peri-LN: Revisiting Normalization Layer in the Transformer Architecture (ICML 2025 · similarity 0.85)
- Normalization Layer Per-Example Gradients are Sufficient to Predict Gradient Noise Scale in Transformers (NeurIPS 2024 · similarity 0.85)
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.
