[V1] port xformers backend to v1 (#21342)

Signed-off-by: Giancarlo Delfin <gdelfin@meta.com>
This commit is contained in:
Giancarlo Delfin
2025-08-05 10:04:46 -07:00
committed by GitHub
parent ae87ddd040
commit 469b3ffaaa
6 changed files with 438 additions and 1 deletions

View File

@@ -128,6 +128,8 @@ def get_attention_backend(backend_name: _Backend):
"vllm.v1.attention.backends.triton_attn.TritonAttentionBackend",
_Backend.TREE_ATTN:
"vllm.v1.attention.backends.tree_attn.TreeAttentionBackend",
_Backend.XFORMERS_VLLM_V1:
"vllm.v1.attention.backends.xformers.XFormersAttentionBackend",
}
if backend_name not in backend_map: