Revert "[Bugfix] Fix GPT-OSS AR+NORM fusion (#28841)" (#29483)

Signed-off-by: Huamin Li <3ericli@gmail.com>
This commit is contained in:
Huamin Li
2025-11-26 06:27:26 -08:00
committed by GitHub
parent 3650a74ed8
commit 70d5953f82
4 changed files with 7 additions and 24 deletions

View File

@@ -111,17 +111,6 @@ if current_platform.is_cuda():
async_tp=96, # MLP is MoE, half the fusions of dense
),
),
ModelBackendTestCase(
model_name="openai/gpt-oss-20b",
model_kwargs=dict(max_model_len=1024, kv_cache_dtype="fp8"),
backend=AttentionBackendEnum.FLASHINFER,
matches=Matches(
attention_fusion=0,
allreduce_fusion=49,
sequence_parallel=49,
async_tp=48,
),
),
]
elif current_platform.is_rocm():