diff --git a/src/peft/tuners/adalora/gptq.py b/src/peft/tuners/adalora/gptq.py
index 32d8f5c2d8..c1708c7bfc 100644
--- a/src/peft/tuners/adalora/gptq.py
+++ b/src/peft/tuners/adalora/gptq.py
@@ -24,6 +24,7 @@ def __init__(
         adapter_name,
         config: AdaLoraConfig,
         r: int = 0,
+        lora_alpha: int = 1,
         **kwargs,
     ) -> None:
         super().__init__()
@@ -33,7 +34,7 @@ def __init__(
         # for backwards compatibility
         self.quant_linear_module = base_layer
         self._active_adapter = adapter_name
-        self.update_layer(adapter_name, r, config=config)
+        self.update_layer(adapter_name, r, lora_alpha=lora_alpha, config=config)
 
     def forward(self, x: torch.Tensor) -> torch.Tensor:
         result = self.quant_linear_module(x)
diff --git a/src/peft/tuners/adalora/model.py b/src/peft/tuners/adalora/model.py
index 72e76cf587..4b170461b4 100644
--- a/src/peft/tuners/adalora/model.py
+++ b/src/peft/tuners/adalora/model.py
@@ -189,7 +189,7 @@ def _create_new_module(lora_config, adapter_name, target, device_map=None, **kwa
             )
             new_module = SVDLinear4bit(target, adapter_name, config=lora_config, **fourbit_kwargs)
         elif QuantLinear is not None and isinstance(target, QuantLinear):
-            new_module = SVDQuantLinear(target, adapter_name, **kwargs)
+            new_module = SVDQuantLinear(target, adapter_name, config=lora_config, **kwargs)
         else:
             if isinstance(target_base_layer, torch.nn.Linear):
                 if lora_config.fan_in_fan_out: