Merge pull request #904 from Titus-von-Koeller/fix-failing-peft-tests

Fix issue #901: Failing PEFT integration tests by reversing PR#867 from jph00/patch-2

bitsandbytes/nn/modules.py

@@ -165,8 +165,6 @@ class Params4bit(torch.nn.Parameter):
         return self
 
     def cuda(self, device):
-        if self.quant_state is not None:
-            return self
         w = self.data.contiguous().half().cuda(device)
         w_4bit, quant_state = bnb.functional.quantize_4bit(w, blocksize=self.blocksize, compress_statistics=self.compress_statistics, quant_type=self.quant_type)
         self.data = w_4bit