Update intermediate_source/char_rnn_classification_tutorial.py

mgs28 · svekars · web-flow · commit c2cca947be9a · 2024-10-30T21:22:21.000-04:00
Co-authored-by: Svetlana Karslioglu &lt;svekars@meta.com&gt;
diff --git a/intermediate_source/char_rnn_classification_tutorial.py b/intermediate_source/char_rnn_classification_tutorial.py
@@ -226,7 +226,7 @@ def __getitem__(self, idx):
 # 
 # This CharRNN class implements an RNN with three components. 
 # First, we use the `nn.RNN implementation <https://pytorch.org/docs/stable/generated/torch.nn.RNN.html>`__.
-# , next we define a layer that maps the RNN hidden layers to our output and finally we apply a softmax. Using nn.RNN 
+# Next, we define a layer that maps the RNN hidden layers to our output. And finally, we apply a ``softmax`` function. Using ``nn.RNN``
 # leads to a significant improvement in performance (e.g. cuDNN-accelerated kernals) versus implementing 
 # each layer as a nn.Linear. It also simplifies the implementation in forward().
 #

Original file line number	Diff line number	Diff line change
`@@ -226,7 +226,7 @@ def __getitem__(self, idx):`
`226`	`226`	`#`
`227`	`227`	`# This CharRNN class implements an RNN with three components.`
`228`	`228`	# First, we use the `nn.RNN implementation <https://pytorch.org/docs/stable/generated/torch.nn.RNN.html>`__.
`229`		`-# , next we define a layer that maps the RNN hidden layers to our output and finally we apply a softmax. Using nn.RNN`
	`229`	+# Next, we define a layer that maps the RNN hidden layers to our output. And finally, we apply a ``softmax`` function. Using ``nn.RNN``
`230`	`230`	`# leads to a significant improvement in performance (e.g. cuDNN-accelerated kernals) versus implementing`
`231`	`231`	`# each layer as a nn.Linear. It also simplifies the implementation in forward().`
`232`	`232`	`#`