[Bugfix][Compile] Replace Transformers AOT caching with graph caching (#58918)

Signed-off-by: fangyao fan <2281693949@qq.com>
Signed-off-by: Taimys <ffy2281693949@gmail.com>
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Co-authored-by: Taimys <ffy2281693949@gmail.com>
Co-authored-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
This commit is contained in:
Taimys
2026-10-13 19:17:29 +00:00
committed by GitHub
co-authored by Taimys Harry Mellor
parent 3f550129e8
commit c068db0e1d
9 changed files with 387 additions and 54 deletions
@@ -206,7 +206,9 @@ def test_segmented_triton_cache_survives_standalone_compilation(
RocmSegmentedAttentionImpl(8, 128, 128**-0.5, 2, None, None, "auto")
assert os.environ["TRITON_CACHE_DIR"] == str(expected)
assert expected.is_dir()
InductorStandaloneAdaptor("binary").initialize_cache(str(tmp_path / model))
InductorStandaloneAdaptor(
use_aot_compile=True, save_format="binary"
).initialize_cache(str(tmp_path / model))
assert os.environ["TRITON_CACHE_DIR"] == str(expected)