fix loss normalization for ddp training

funasr/models/e2e_uni_asr.py

@@ -442,7 +442,7 @@ class UniASR(FunASRModel):
         stats["loss"] = torch.clone(loss.detach())
         # force_gatherable: to-device and to-tensor if scalar for DataParallel
         if self.length_normalized_loss:
-            batch_size = (text_lengths + 1).sum().type_as(batch_size)
+            batch_size = int((text_lengths + 1).sum())
         loss, stats, weight = force_gatherable((loss, stats, batch_size), loss.device)
         return loss, stats, weight