minor spelling tweaks

docs/source/concepts.rst

@@ -143,7 +143,7 @@ you need to do is add the class to the registry using the `@register_plugin` dec
             super(MyClassifier, self).__init__()
 
         def forward(self, input_tensor):
-            # Do complex transofmrations
+            # Do complex transformations
             return result
 
 Finally, to enable the sharing of experiment configuration files, we can use environment variables for paths parameters,

docs/source/data-components.rst

@@ -5,20 +5,20 @@ Data Management Components
 Vocabularies
 ------------
 We provide classes to build vocabularies over datasets.
-These classes do not take into account the nature of the symbols whith which you are filling a dictionary.
+These classes do not take into account the nature of the symbols which which you are filling a dictionary.
 Hence, whether you want to use vocabularies for tokens, characters, BPE, etc.., you can still use the vocabulary classes
 coupled with a vectorizer of your choice.
 
 
 Vectorizers
 -----------
-Vectorizers take string inputs and converts hem to lists of symbnols.
+Vectorizers take string inputs and converts hem to lists of symbols.
 When implementing your vectorizer, you need to build the vocabularies that you need for your experiment, and set these
 vocabularies as vectorizer attributes. You also need to implement the `vectorize` method, which turns a string input
 into a list of numbers representing the symbols you choose to use to represent the text.
 
 Loaders
 -------
-Data Loaders splits te dataset into train, validation and test sets, and creates the appropriate PyTorch DataLoaders.
+Data Loaders splits the dataset into train, validation and test sets, and creates the appropriate PyTorch DataLoaders.
 
 

docs/source/modeling-components.rst

@@ -24,11 +24,11 @@ parameters yield by the PyTorch batch iterator. For example:
             super(MyClassifier, self).__init__()
 
         def forward(self, input_tensor: torch.tensor):
-            # Do complex transofmrations
+            # Do complex transformations
             return result
 
 In this example, you need to set your data loader to yield batches with the key `"input_tensor"`.
-If the `forward` method has default parameters that do not appear in the batch, they will be used, otherwise tyey will
+If the `forward` method has default parameters that do not appear in the batch, they will be used, otherwise they will
 be replaced by the values from the batch
 
 Optimizers

docs/source/surnames.rst

@@ -6,7 +6,7 @@ from different countries is provided and the task is to predict the country.
 
 Vectorizer
 ----------
-The most straigthforward to represent a surname is to get its one-hot character encoding:
+The most straightforward to represent a surname is to get its one-hot character encoding:
 
 
 .. code-block:: python