Hy4-preview
Tencent (Hunyuan)with IndexCache for cross-layer sparse index reuse
2 more from this model
with IndexCache for cross-layer sparse index reuse
with IndexCache for cross-layer sparse index reuse
implementation detail · filed under model architecture
A cache mechanism for reusing sparse-attention indices across layers.
The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See sparse attention indexer for the methods filed beside it.
3 quotes from 3 sources, one card per model, each opening with its strongest.
with IndexCache for cross-layer sparse index reuse
with IndexCache for cross-layer sparse index reuse
with IndexCache for cross-layer sparse index reuse
Picked by hand, not extracted: where to read more, not evidence for anything on this page.