taxonomy node · level 4
sparse attention indexer
22 methods filed at this node or below it, from the sources of 7 models.
model architecture :: token mixer :: sparse attention :: sparse attention indexer
Matching aids for the classifier: lightning indexer; index branch; indexer warmup; IndexShare.
In this branch 22
Everything filed at this node or below it, with one collapsible heading per child node.
filed here 22
By model
Which of this branch's techniques each model's own documents describe, and how strongly. Under each model: its strongest treatment anywhere in the branch.
| Model | techniques |
|---|---|
| GLM-5.3-Flash core | IndexPool core— |
| DeepSeek-V4.1-Flash core | Compressed Sparse Attention 2 coreHierarchical Sparse Indexer coreReindex Mode coreReuse Mode coreCross-stage shared-state management for attention reuse used— |
| Hy4-preview core | IndexCache core— |
| GLM-5.2 core | IndexShare coreLightning Indexer used— |
| MiniMax-M3 core | Index Branch coreIndexer Warmup usedSingle-head index key usedIndex Branch output not usedIndex Branch value head not used— |
| DeepSeek-V3.2 core | Fine-grained token selection coreLightning Indexer coreDense Warm-up Stage usedDetached indexer-input optimization used— |
| Qwen3.8-Flash-Next core | Compressed lightweight indexer coreMQA indexer coreReuse QSA index selection across speculative decoding steps coreAverage pooling usedBlock-causal scoring used— |