AttentionIndex: Accelerating Learned Sparse Attention with a Causal Inverted-File Index
cs.AI, cs.LG
Submitted: 2026-11-02
Updated: 2026-11-02
License: http://creativecommons.org/licenses/by/4.0/
The gist: Learned sparse attention, used in production models such as DeepSeek-V4 and GLM-5, lets each query attend to only k keys chosen by a lightweight indexer…
Terminology
Abstract
Learned sparse attention, used in production models such as DeepSeek-V4 and GLM-5, lets each query attend to only k keys chosen by a lightweight indexer, but the indexer still scores every key in the prefix. At one million tokens, this selection takes more than half of prefill time. Two-stage selection over contiguous blocks, as in HiSA, reduces this cost, but its recall declines as context grows at a fixed candidate budget. We present AttentionIndex, which routes each query over an inverted-file index that groups indexer keys by similarity and refines the candidates with the native indexer score. AttentionIndex builds the index during prefill, making new keys searchable immediately, reusing previous clustering to keep maintenance cheap, and exposing each query only to clusters built from its own prefix. On GLM-5.3 and DeepSeek-V4-Flash at one million tokens, AttentionIndex delivers 1.3-1.8 times the prefill throughput of exhaustive selection, and at throughput within 5% of HiSA, it recovers 17.0 and 10.0 percentage points more of the exhaustive top-k, respectively. AttentionIndex also improves accuracy on long-context benchmarks by up to 18.4 percentage points over HiSA at the same candidate budget.
Sources
- DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence
- Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context
- Squeezed Attention: Accelerating Long Context Length LLM Inference
- StreamIndex: Memory-Bounded Compressed Sparse Attention via Streaming Top-k
- Inference-time sparse attention with asymmetric indexing
Related papers
- Multi-Objective Structured Pruning of LLMs for Latency and Model Size Optimization
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- Embeddings for Preferences, Not Semantics
- Can We Trust Item Response Theory for AI Evaluation?
- Diagnosing Task Insensitivity in Language Agents