Update oss.rst (#107)

docs/source/tutorials/oss.rst

@@ -18,11 +18,12 @@ Let's suppose that your trainer looks like
         dist_init(rank, world_size)
 
         # Problem statement
-        model = myAwesomeModel()
+        model = myAwesomeModel().to(rank)
         dataloader = mySuperFastDataloader()
-        loss = myVeryRelevantLoss()
+        loss_ln = myVeryRelevantLoss()
 
-        base_optimizer_arguments = {} # any optimizer specific arguments, LR, momentum, etc...
+        # optimizer specific arguments e.g. LR, momentum, etc...
+        base_optimizer_arguments = { "lr": 1e-4} 
         optimizer = torch.optim.SGD(
             params=model.parameters(),
             **base_optimizer_arguments)
@@ -30,11 +31,12 @@ Let's suppose that your trainer looks like
         # Any relevant training loop, nothing specific to OSS. For example:
         model.train()
         for e in range(epochs):
-            for batch in dataloader:
+            for (data, target) in dataloader:
+                data, target = data.to(rank), target.to(rank)
                 # Train
                 model.zero_grad()
-                outputs = model(batch["inputs"])
-                loss = loss_fn(outputs, batch["label"])
+                outputs = model(data)
+                loss = loss_fn(outputs, target)
                 loss /= world_size
                 loss.backward()
                 torch.distributed.all_reduce(loss, op=torch.distributed.ReduceOp.SUM)
@@ -58,11 +60,12 @@ Then sharding the optimizer state is merely a matter of wrapping your optimizer
         dist_init(rank, world_size)
 
         # Problem statement
-        model = myAwesomeModel()
+        model = myAwesomeModel().to(rank)
         dataloader = mySuperFastDataloader()
-        loss = myVeryRelevantLoss()
+        loss_ln = myVeryRelevantLoss()
 
-        base_optimizer_arguments = {}  # any optimizer specific arguments, LR, momentum, etc...
+        # optimizer specific arguments e.g. LR, momentum, etc...
+        base_optimizer_arguments = { "lr": 1e-4} 
 
         # ** NEW ** Wrap a base optimizer into OSS
         base_optimizer = torch.optim.SGD  # any pytorch compliant optimizer
@@ -74,12 +77,28 @@ Then sharding the optimizer state is merely a matter of wrapping your optimizer
         # Any relevant training loop, nothing specific to OSS. For example:
         model.train()
         for e in range(epochs):
-            for batch in dataloader:
+            for (data, target) in dataloader:
+                data, target = data.to(rank), target.to(rank)
                 # Train
                 model.zero_grad()
-                outputs = model(batch["inputs"])
-                loss = loss_fn(outputs, batch["label"])
+                outputs = model(data)
+                loss = loss_fn(outputs, target)
                 loss /= world_size
                 loss.backward()
                 torch.distributed.all_reduce(loss, op=torch.distributed.ReduceOp.SUM)
                 optimizer.step()
+
+
+The above `train` function will then need to be run via a `multiprocessing.spawn` function.
+
+.. code-block:: python
+
+
+    mp.spawn(
+            train,
+            args=(WORLD_SIZE, EPOCHS),
+            nprocs=WORLD_SIZE,
+            join=True
+        )
+    
+to see it in action, you can test it with the following script _`tutorial_oss.py <../../../examples/tutorial_oss.py>`_