[torch.compile] support all attention backends (#10558)

Signed-off-by: youkaichao <youkaichao@gmail.com>
2024-11-22 14:04:42 -08:00
parent db100c5cde
commit eebad39f26
77 changed files with 876 additions and 648 deletions
--- a/vllm/platforms/hpu.py
+++ b/vllm/platforms/hpu.py
@@ -13,6 +13,7 @@ else:
 class HpuPlatform(Platform):
    _enum = PlatformEnum.HPU
    device_type: str = "hpu"
+    dispatch_key: str = "HPU"

    @classmethod
    def get_default_attn_backend(cls, selected_backend: _Backend) -> _Backend: