Model techniques map
Techniquesmodel architecturetoken mixersparse attentionsparse attention indexer

implementation detail · filed under model architecture

Block-causal scoring

Scores query-to-compressed-block importance using block-causal scoring.

source
1
model
1
lab adopt it
1
strongest
used

How sources treat it

One count per evidence span, weakest treatment to strongest.

used 1

Documented in

Qwen3.8-Flash-Next

Evidence

1 span quoted from the sources, strongest treatment first.

Block-level importance scores are then obtained through block-causal scoring.

usedunclearin Qwen3.8-Flash-NextQwen
On the Design of Qwen3.8-Next Architecture: Evaluation, Efficiency, and Training Stability·Qwen·2026-08-31

Filed alongside

Other methods under model architecture :: token mixer :: sparse attention :: sparse attention indexer.

Compressed Sparse Attention 2Hierarchical Sparse IndexerIndexCacheIndexShareLightning IndexerReindex ModeReuse ModeDense Warm-up StageFine-grained token selectionIndex BranchIndexer WarmupReuse QSA index selection across speculative decoding stepsAverage poolingCompressed lightweight indexerCross-stage shared-state management for attention reuseDetached indexer-input optimizationIndex Branch outputIndex Branch value headIndexPoolMQA indexerSingle-head index key