diff --git a/modules/sdnq/quantizer.py b/modules/sdnq/quantizer.py
index d39523e5b..b069d470b 100644
--- a/modules/sdnq/quantizer.py
+++ b/modules/sdnq/quantizer.py
@@ -230,8 +230,11 @@ def sdnq_quantize_layer(layer, weights_dtype="int8", torch_dtype=None, group_siz
             use_quantized_matmul=use_quantized_matmul,
             re_quantize_for_matmul=re_quantize_for_matmul,
         )
+
         layer.weight.data = layer.sdnq_dequantizer.pack_weight(layer.weight).to(return_device, non_blocking=non_blocking)
         layer.sdnq_dequantizer = layer.sdnq_dequantizer.to(return_device, non_blocking=non_blocking)
+        if layer.bias is not None and layer.bias.device != "meta":
+            layer.bias.data = layer.bias.to(return_device, non_blocking=non_blocking)
 
         layer.forward = get_forward_func(layer_class_name, use_quantized_matmul, dtype_dict[weights_dtype]["is_integer"], use_tensorwise_fp8_matmul)
         layer.forward = layer.forward.__get__(layer, layer.__class__)