add detail about set_epoch (#2066)

Author: subramen

beginner_source/ddp_series_multigpu.rst

@@ -123,6 +123,18 @@ Distributing input data
    +   sampler=DistributedSampler(train_dataset),
    )
 
+-  Calling the ``set_epoch()`` method on the ``DistributedSampler`` at the beginning of each epoch is necessary to make shuffling work 
+   properly across multiple epochs. Otherwise, the same ordering will be used in each epoch.
+
+.. code:: diff
+
+   def _run_epoch(self, epoch):
+       b_sz = len(next(iter(self.train_data))[0])
+   +   self.train_data.sampler.set_epoch(epoch)
+       for source, targets in self.train_data:
+         ...
+         self._run_batch(source, targets)
+
 
 Saving model checkpoints
 ~~~~~~~~~~~~~~~~~~~~~~~~