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 3, 22:45 UTC
[Bugfix] Accept canonicalized `modelopt_*` quant_method in `_extract_modelopt_quant_algo` (#42181)
Signed-off-by: Vadim Gimpelson <vadim.gimpelson@gmail.com>
Vadim Gimpelson committed
May 11, 2026 at 19:10 UTC
a2e776d7168e1a75fe32438aea4254542602c012
1 file changed
+1
-1
vllm/model_executor/layers/quantization/modelopt.py
+1
-1
@@ -250,7 +250,7 @@ class ModelOptQuantConfigBase(QuantizationConfig):
250
"""
251
if hf_quant_cfg is None:
252
return None
253
- if hf_quant_cfg.get("quant_method", "").lower() != "modelopt":
253
+ if not hf_quant_cfg.get("quant_method", "").lower().startswith("modelopt"):
254
return None
255
if "quantization" in hf_quant_cfg:
256
quant_config = hf_quant_cfg["quantization"]