Adding additional sanity check

Signed-off-by: Priyanka Dangi <quic_pdangi@quicinc.com>

TrainingExtensions/torch/src/python/aimet_torch/v2/nn/modules/custom.py

@@ -590,6 +590,9 @@ def batch_norm_wrapper(
             momentum: float = 0.1,
             eps: float = 1e-5,
         ) -> Tensor:
+            if training:
+                if self.input_quantizers[1] is not None or self.input_quantizers[2] is not None:
+                    raise RuntimeError(f"{self.__class__} doesn't support quantizing running_mean or running_var in training mode")
 
             input = _quantize_if_applicable(input, self.input_quantizers[0])
             running_mean = _quantize_if_applicable(running_mean, self.input_quantizers[1])