fixes the asserion error when num_beams > 1

modeling_phi4mm.py

@@ -2096,7 +2096,7 @@ class Phi4MMForCausalLM(Phi4MMPreTrainedModel, GenerationMixin):
         return_dict = return_dict if return_dict is not None else self.config.use_return_dict
 
         if isinstance(input_mode, torch.Tensor):
-            assert len(input_mode) == 1
+            # len(input_mode) == num_beams in beam search, and all elements of input_mode should have the same value
             input_mode = input_mode[0].item()
         input_mode = InputMode(input_mode)