DCP supports hybrid attention (#40996)

Signed-off-by: YanXu <yancey.yx@alibaba-inc.com>
Signed-off-by: Jingyi Yang <girasoleyang@gmail.com>
Co-authored-by: Jingyi Yang <girasoleyang@gmail.com>
This commit is contained in:
Yan Xu
2026-07-09 21:34:45 -07:00
committed by GitHub
co-authored by Jingyi Yang
parent 2d814a0082
commit 95ed0feaa5
26 changed files with 783 additions and 125 deletions
@@ -38,6 +38,7 @@ def mock_model_runner_with_input_batch():
vocab_size=32000,
block_sizes=[16],
kernel_block_sizes=[16],
max_num_blocks_per_req=[64],
logitsprocs=None,
is_pooling_model=False,
)