biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

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

Browse Source 
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
Download Patch File Download Diff File Expand all files Collapse all files

17
vllm/model_executor/layers/fused_moe/layer.py
View File

@@ -1690,10 +1690,6 @@ class FusedMoE(CustomOp):
)
def reduce_output(states: torch.Tensor) -> torch.Tensor:
# Slice before all_reduce to enable possible fusion
if self.hidden_size != og_hidden_states:
states = states[..., :og_hidden_states]
if (
not self.is_sequence_parallel
and not self.use_dp_chunking
@@ -1716,12 +1712,11 @@ class FusedMoE(CustomOp):
if self.zero_expert_num is not None and self.zero_expert_num > 0:
assert isinstance(fused_output, tuple)
fused_output, zero_expert_result = fused_output
return (
reduce_output(fused_output)
+ zero_expert_result[..., :og_hidden_states]
)
return (reduce_output(fused_output) + zero_expert_result)[
..., :og_hidden_states
]
else:
return reduce_output(fused_output)
return reduce_output(fused_output)[..., :og_hidden_states]
else:
if current_platform.is_tpu():
# TODO: Once the OOM issue for the TPU backend is resolved, we
@@ -1734,8 +1729,8 @@ class FusedMoE(CustomOp):
hidden_states, router_logits, self.layer_name
)
return (
reduce_output(shared_output),
reduce_output(fused_output),
reduce_output(shared_output)[..., :og_hidden_states],
reduce_output(fused_output)[..., :og_hidden_states],
)
def forward_cuda(