A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 8, 17:45 UTC
[CI][AMD]Bugfix] Check that model_config is not None in enable_norm_pad_fusion (#34007)
Signed-off-by: Randall Smith <Randall.Smith@amd.com>
rasmith committed
Feb 6, 2026 at 20:43 UTC
ec28784fdc4dd1eb0141e9629b3f2d7cfce5c114
1 file changed
+1
vllm/config/vllm.py
+1
@@ -109,6 +109,7 @@ def enable_norm_pad_fusion(cfg: "VllmConfig") -> bool:
109
envs.VLLM_ROCM_USE_AITER
110
and envs.VLLM_ROCM_USE_AITER_RMSNORM
111
and envs.VLLM_ROCM_USE_AITER_TRITON_GEMM
112
+ and cfg.model_config is not None
113
and cfg.model_config.get_hidden_size() == 2880
114
)
115