Commit ad30176

committed

[None][feat] add paged DSV4 sparse attention cache

Signed-off-by: Balamurugan Marimuthu <246387390+bmarimuthu-nv@users.noreply.github.com>

1 parent e3a1b95 commit ad30176Copy full SHA for ad30176

3 files changed

tensorrt_llm/_torch/auto_deploy/custom_ops
- attention_interface.py
- attention
  - deepseek_v4_sparse_attention.py
tests/unittest/auto_deploy/singlegpu/custom_ops/attention
- test_deepseek_v4_sparse_attention.py

Comments

(0)

Provide feedback