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 2, 20:45 UTC
[chores][log] change registry log from `warning` to `debug` (#43045)
Signed-off-by: Hank <hcc.mayday@gmail.com>
Hank_ committed
May 26, 2026 at 15:13 UTC
b3269454b1eff2ce651b8ff3d0858c40ff81bc6a
3 files changed
+5
-5
tests/quantization/test_register_quantization_config.py
+3
-3
@@ -108,14 +108,14 @@ def test_register_quantization_config(caplog_vllm):
108
assert get_quantization_config("custom_quant") == CustomQuantConfig
109
110
# The quantization method `custom_quant` is already exists,
111
- # should raise a warning when re-registering it.
112
- with caplog_vllm.at_level(logging.WARNING):
111
+ # should raise a debug message when re-registering it.
112
+ with caplog_vllm.at_level(logging.DEBUG, logger="vllm"):
113
register_quantization_config("custom_quant")(CustomQuantConfig)
114
115
assert any(
116
"The quantization method 'custom_quant' already exists" in message
117
for message in caplog_vllm.messages
118
- ), "Expected a warning when re-registering custom_quant"
118
+ ), "Expected a debug message when re-registering custom_quant"
119
120
121
@pytest.mark.parametrize(
vllm/model_executor/layers/quantization/__init__.py
+1
-1
@@ -84,7 +84,7 @@ def register_quantization_config(quantization: str):
84
85
def _wrapper(quant_config_cls):
86
if quantization in QUANTIZATION_METHODS:
87
- logger.warning(
87
+ logger.debug(
88
"The quantization method '%s' already exists and will be "
89
"overwritten by the quantization config %s.",
90
quantization,
vllm/model_executor/models/registry.py
+1
-1
@@ -980,7 +980,7 @@ class _ModelRegistry:
980
raise TypeError(msg)
981
982
if model_arch in self.models:
983
- logger.warning(
983
+ logger.debug(
984
"Model architecture %s is already registered, and will be "
985
"overwritten by the new model class %s.",
986
model_arch,