From ec7aa63b3a511475f1c86efa7b0f1b8454923e78 Mon Sep 17 00:00:00 2001 From: anujj Date: Tue, 15 Oct 2024 09:02:59 +0530 Subject: [PATCH] nvidia awq only use QuantFormat.QDQ quant format (#22429) nvidia awq only use QuantFormat.QDQ quant format --- onnxruntime/python/tools/quantization/matmul_4bits_quantizer.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/onnxruntime/python/tools/quantization/matmul_4bits_quantizer.py b/onnxruntime/python/tools/quantization/matmul_4bits_quantizer.py index 91819a2078..1d91141a11 100644 --- a/onnxruntime/python/tools/quantization/matmul_4bits_quantizer.py +++ b/onnxruntime/python/tools/quantization/matmul_4bits_quantizer.py @@ -280,7 +280,7 @@ class NVAWQWeightOnlyQuantConfig(WeightOnlyQuantConfig): super().__init__( algorithm="nvidia_awq", - quant_format=quant_format, + quant_format=QuantFormat.QDQ, op_types_to_quantize=None, # Assuming op_types_to_quantize is handled elsewhere quant_axes=None, # Assuming quant_axes is handled elsewhere )