fix iterable dataset (issue 9899) (#9952)

python/ray/util/sgd/torch/distributed_torch_runner.py

@@ -140,7 +140,7 @@ class DistributedTorchRunner(TorchRunner):
             if self.add_dist_sampler:
                 self.train_loader = with_sampler(self.train_loader)
 
-        if self.validation_loader and should_wrap_dataloader(
+        if self.validation_loader is not None and should_wrap_dataloader(
                 self.validation_loader):
             if self.add_dist_sampler:
                 self.validation_loader = with_sampler(self.validation_loader)