Research map / Architectures and efficiency / Attention and transformers
O(n) Connections are Expressive Enough: Universal Approximability of Sparse Transformers
Chulhee Yun, Yin-Wen Chang, Srinadh Bhojanapalli, Ankit Singh Rawat, Sashank Reddi, Sanjiv Kumar
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 deterministic sparsity pattern instead of Random | Sparse Transformers · Synthetic copying task | accuracy | Improved |
| Use Star attention rather than Strided or Fixed attention | Sparse Transformers · One Billion Word Benchmark | perplexity | Improved |
| Use Star attention rather than the other tested patterns | Sparse Transformers · Synthetic copying task | accuracy | Hurt |
| Use Union rather than Sequential or Multihead configuration | Strided and Fixed sparse Transformers · WMT18 English-Czech, Newstest 2015 | BLEU | Improved |
| Increase sparsity in the Multihead configuration | Strided and Fixed sparse Transformers · WMT18 English-Czech, Newstest 2015 | BLEU | Improved |
| Use Union rather than the other head configurations | Strided and Fixed sparse Transformers · BERT tasks | accuracy | Hurt |
| Choose sparsity pattern based on task rather than assuming one pattern is best everywhere | Sparse Transformers · NLP tasks | task performance | Mixed |
Closest papers by meaning
- Transformers Learn Faster with Semantic Focus (NeurIPS 2025 · similarity 0.89)
- When Do Transformers Outperform Feedforward and Recurrent Networks? A Statistical Perspective (NeurIPS 2025 · similarity 0.89)
- How Transformers Learn Regular Language Recognition: A Theoretical Study on Training Dynamics and Implicit Bias (ICML 2025 · similarity 0.88)
- Low-Rank Approximation for Sparse Attention in Multi-Modal LLMs (CVPR 2024 · similarity 0.88)
- SEA: Sparse Linear Attention with Estimated Attention Mask (ICLR 2024 · similarity 0.88)
- Curse of High Dimensionality Issue in Transformer for Long Context Modeling (ICML 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.
