Model techniques map
Techniquesmodel architecturetoken mixersparse attentionsparse attention indexer

specific method · filed under model architecture

IndexShare

Reuses one indexer across multiple sparse-attention layers.

sources
3
model
1
lab adopt it
1
strongest
core

How sources treat it

One count per evidence span, weakest treatment to strongest.

used 1core 2

Documented in

Further reading

Picked by hand, not extracted: where to read more, not evidence for anything on this page.

Evidence

3 spans quoted from the sources, strongest treatment first.

the most technically significant innovation in GLM-5.2 is a mechanism Z.ai calls IndexShare

coremodel architecturein GLM-5.2Z.ai

We propose IndexShare, which reuses the same indexer across every four sparse attention layers, reducing per-token FLOPs by 2.9× at a 1M context length.

coremodel architecturein GLM-5.2Z.ai

We propose IndexShare, which reuses the same indexer across every four sparse attention layers, reducing per-token FLOPs by 2.9× at a 1M context length.

usedmodel architecturein GLM-5.2Z.ai

Filed alongside

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