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 14, 23:45 UTC
[Cleanup] Only log MoE DP setup warning if DP is enabled (#21315)
Signed-off-by: mgoin <mgoin64@gmail.com>
Michael Goin committed
Jul 23, 2025 at 03:02 UTC
f002e9a8700c9970a1ac21711c8673f246bbaf1b
1 file changed
+5
-4
vllm/model_executor/layers/fused_moe/config.py
+5
-4
@@ -464,10 +464,11 @@ class FusedMoEConfig:
464
)
465
else:
466
_quant_config = FusedMoEQuantConfig()
467
- logger.warning_once("MoE DP setup unable to determine "
468
- "quantization scheme or unsupported "
469
- "quantization type. This model will "
470
- "not run with DP enabled.")
467
+ if moe_parallel_config.dp_size > 1:
468
+ logger.warning_once("MoE DP setup unable to determine "
469
+ "quantization scheme or unsupported "
470
+ "quantization type. This model will "
471
+ "not run with DP enabled.")
472
else:
473
_quant_config = quant_config
474