Add a simple torch native attention backend (#2241)

2024-12-01 19:01:25 +08:00
parent fc78640e00
commit 62c516ac45
7 changed files with 388 additions and 26 deletions
--- a/test/srt/run_suite.py
+++ b/test/srt/run_suite.py
@@ -34,6 +34,7 @@ suites = {
        "test_srt_endpoint.py",
        "test_torch_compile.py",
        "test_torch_compile_moe.py",
+        "test_torch_native_attention_backend.py",
        "test_torchao.py",
        "test_triton_attention_kernels.py",
        "test_triton_attention_backend.py",