Research map / Architectures and efficiency / Attention and transformers
Big Bird: Transformers for Longer Sequences
Manzil Zaheer, Guru Guruganesh, Kumar Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed
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 |
|---|---|---|---|
| combine random and window attention instead of window only (ablation) | sparse attention @512 · SQuAD | F1: 76.4 -> 85.1 | Improved |
| combine random and window attention instead of random only (ablation) | sparse attention @512 · MNLI | accuracy: 80.2 -> 80.5 | Improved |
| replace full attention with random+window sparse attention (no global) at length 512 (ablation) | BERT-base · MNLI | accuracy: 83.4 -> 80.5 | Hurt |
| use extended global tokens (etc) instead of internal (itc) (ablation) | BigBird base · Natural Questions dev LA | F1: 70.8 -> 73.9 | Improved |
| use extended global tokens (etc) instead of internal (itc) (ablation) | BigBird base · TriviaQA dev | F1: 79.5 -> 78.7 | Hurt |
| longer context via BigBird sparse attention vs RoBERTa | base QA model · HotpotQA dev Joint | F1: 63.5 -> 67.8 | Improved |
| BigBird sparse encoder for long input | Pegasus large · Arxiv summarization | ROUGE-1: 43.85 -> 46.63 | Improved |
| BigBird sparse encoder for long input | Pegasus large · BigPatent summarization | ROUGE-1: 52.25 -> 60.64 | Improved |
| increase context from 512 to 4096 with BigBird | DNA MLM · human reference genome | BPC: 1.23 -> 1.12 | Improved |
| BigBird vs DeepSea | chromatin-profile prediction · DHS | AUC: 92.3 -> 92.1 | No clear effect |
| BigBird long-context classification | RoBERTa · IMDb | accuracy | No clear effect |
Later papers that compared with this method
16 later accepted papers report BigBird as a baseline in their results tables; 12 of them report a better number on most of the results they share with it.
- Making Every Head Count: Sparse Attention Without the Speed-Performance Trade-off (AAAI 2026 · better on 3 of 3 shared results)
- Training Free Exponential Context Extension via Cascading KV Cache (ICLR 2025 · better on 2 of 2 shared results)
- Cached Transformers: Improving Transformers with Differentiable Memory Cachde (AAAI 2024 · better on 1 of 1 shared result)
- Proxyformer: Nyström-Based Linear Transformer with Trainable Proxy Tokens (AAAI 2024 · better on 0 of 1 shared result)
- LeaPformer: Enabling Linear Transformers for Autoregressive and Simultaneous Tasks via Learned Proportions (ICML 2024 · better on 1 of 1 shared result)
- Diffuser: Efficient Transformers with Multi-Hop Attention Diffusion for Long Sequences (AAAI 2023 · better on 4 of 4 shared results)
- Treeformer: Dense Gradient Trees for Efficient Attention Computation (ICLR 2023 · better on 2 of 4 shared results)
- VCC: Scaling Transformers to 128K Tokens or More by Prioritizing Important Tokens (NeurIPS 2023 · better on 4 of 4 shared results)
8 more not listed.
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- FlexPrefill: A Context-Aware Sparse Attention Mechanism for Efficient Long-Sequence Inference (ICLR 2025 · similarity 0.88)
- RAT: Bridging RNN Efficiency and Attention Accuracy via Chunk-based Sequence Modeling (NeurIPS 2025 · similarity 0.88)
- Star Attention: Efficient LLM Inference over Long Sequences (ICML 2025 · similarity 0.88)
- Rodimus*: Breaking the Accuracy-Efficiency Trade-Off with Efficient Attentions (ICLR 2025 · similarity 0.87)
- SEA: Sparse Linear Attention with Estimated Attention Mask (ICLR 2024 · similarity 0.87)
- Curse of High Dimensionality Issue in Transformer for Long Context Modeling (ICML 2025 · similarity 0.87)
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.
