implementation detail · filed under model architecture
Top-512 block selection
Keeps the 512 best blocks and expands them back to 2048 logical token positions in Qwen Sparse Attention.
Also called QSA keeps the best 512 blocks.
- source
- 1
- model
- 1
- lab adopt it
- 1
- strongest
- core
How sources treat it
One count per evidence span, weakest treatment to strongest.
core 1
Documented in
Evidence
1 span quoted from the sources, strongest treatment first.
QSA keeps the best 512 blocks, expands them back to 2048 logical token positions
coreunclearin Qwen3.8-Flash-NextQwen
Filed alongside
Other methods under model architecture :: token mixer :: sparse attention :: block selection.