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 15, 12:15 UTC
fix typo in logging (#17605)
Eric Hartford committed
May 2, 2025 at 21:04 UTC
9b103a1d7664b10901b893d054eaa15dde56c48d
2 files changed
+2
-2
vllm/model_executor/layers/quantization/awq_marlin.py
+1
-1
@@ -140,7 +140,7 @@ class AWQMarlinConfig(QuantizationConfig):
140
from vllm.model_executor.layers.quantization.moe_wna16 import (
141
MoeWNA16Config)
142
if not check_moe_marlin_supports_layer(layer, self.group_size):
143
- logger.warning_one(
143
+ logger.warning_once(
144
f"Layer '{prefix}' is not supported by AWQMoeMarlin. "
145
"Falling back to Moe WNA16 kernels.")
146
return MoeWNA16Config.from_config(
vllm/model_executor/layers/quantization/gptq_marlin.py
+1
-1
@@ -157,7 +157,7 @@ class GPTQMarlinConfig(QuantizationConfig):
157
from vllm.model_executor.layers.quantization.moe_wna16 import (
158
MoeWNA16Config)
159
if not check_moe_marlin_supports_layer(layer, self.group_size):
160
- logger.warning_one(
160
+ logger.warning_once(
161
f"Layer '{prefix}' is not supported by GPTQMoeMarlin. "
162
"Falling back to Moe WNA16 kernels.")
163
return MoeWNA16Config.from_config(