various documentation formatting (#9359)

Summary:
This is a grab-bag of documentation formatting fixes.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/9359

Differential Revision: D8831400

Pulled By: soumith

fbshipit-source-id: 8dac02303168b2ea365e23938ee528d8e8c9f9b7

docs/source/notes/windows.rst

@@ -48,8 +48,8 @@ build tasks. It can be used by typing only a few lines of code.
 One key install script
 ^^^^^^^^^^^^^^^^^^^^^^
 
-You can take a look at the script `here
-<https://github.com/peterjc123/pytorch-scripts>`_. 
+You can take a look at `this set of scripts
+<https://github.com/peterjc123/pytorch-scripts>`_.
 It will lead the way for you.
 
 Extension
@@ -176,8 +176,8 @@ You can resolve this by typing the following command.
 
 As for the wheels package, since we didn't pack some libaries and VS2017 
 redistributable files in, please make sure you install them manually.
-The VS 2017 redistributable installer can be downloaded `here
-<https://aka.ms/vs/15/release/VC_redist.x64.exe>`_.
+The `VS 2017 redistributable installer
+<https://aka.ms/vs/15/release/VC_redist.x64.exe>`_ can be downloaded.
 And you should also pay attention to your installation of Numpy. Make sure it
 uses MKL instead of OpenBLAS. You may type in the following command.
 

torch/_torch_docs.py

@@ -3782,10 +3782,11 @@ Args:
     end (Number): the ending value for the set of points
     step (Number): the gap between each pair of adjacent points. Default: ``1``.
     {out}
-    {dtype}  If `dtype` is not given, infer the data type from the other input arguments.
-             If any of `start`, `end`, or `stop` are floating-point,
-             the `dtype` is inferred to be the default dtype, see :meth:`~torch.get_default_dtype`.
-             Otherwise, the `dtype` is inferred to be `torch.int64`.
+    {dtype}
+        If `dtype` is not given, infer the data type from the other input arguments.
+        If any of `start`, `end`, or `stop` are floating-point,
+        the `dtype` is inferred to be the default dtype, see :meth:`~torch.get_default_dtype`.
+        Otherwise, the `dtype` is inferred to be `torch.int64`.
     {layout}
     {device}
     {requires_grad}

torch/distributions/categorical.py

@@ -22,7 +22,7 @@ class Categorical(Distribution):
     vectors.
 
     .. note:: :attr:`probs` must be non-negative, finite and have a non-zero sum,
-    and it will be normalized to sum to 1.
+              and it will be normalized to sum to 1.
 
     See also: :func:`torch.multinomial`
 

torch/distributions/multinomial.py

@@ -16,7 +16,7 @@ class Multinomial(Distribution):
     called (see example below)
 
     .. note:: :attr:`probs` must be non-negative, finite and have a non-zero sum,
-    and it will be normalized to sum to 1.
+              and it will be normalized to sum to 1.
 
     -   :meth:`sample` requires a single shared `total_count` for all
         parameters and samples.

torch/distributions/one_hot_categorical.py

@@ -12,7 +12,7 @@ class OneHotCategorical(Distribution):
     Samples are one-hot coded vectors of size ``probs.size(-1)``.
 
     .. note:: :attr:`probs` must be non-negative, finite and have a non-zero sum,
-    and it will be normalized to sum to 1.
+              and it will be normalized to sum to 1.
 
     See also: :func:`torch.distributions.Categorical` for specifications of
     :attr:`probs` and :attr:`logits`.

torch/utils/checkpoint.py

@@ -87,7 +87,7 @@ def checkpoint(function, *args):
         args: tuple containing inputs to the :attr:`function`
 
     Returns:
-        Output of running :attr:`function` on *:attr:`args`
+        Output of running :attr:`function` on :attr:`*args`
     """
     return CheckpointFunction.apply(function, *args)
 
@@ -120,7 +120,7 @@ def checkpoint_sequential(functions, segments, *inputs):
         inputs: tuple of Tensors that are inputs to :attr:`functions`
 
     Returns:
-        Output of running :attr:`functions` sequentially on *:attr:`inputs`
+        Output of running :attr:`functions` sequentially on :attr:`*inputs`
 
     Example:
         >>> model = nn.Sequential(...)

torch/utils/dlpack.py

@@ -7,8 +7,8 @@ torch._C._add_docstr(from_dlpack, r"""from_dlpack(dlpack) -> Tensor
 
 Decodes a DLPack to a tensor.
 
-Arguments::
-    dlpack - a PyCapsule object with the dltensor
+Args:
+    dlpack: a PyCapsule object with the dltensor
 
 The tensor will share the memory with the object represented
 in the dlpack.
@@ -19,8 +19,8 @@ torch._C._add_docstr(to_dlpack, r"""to_dlpack(tensor) -> PyCapsule
 
 Returns a DLPack representing the tensor.
 
-Arguments::
-    tensor - a tensor to be exported
+Args:
+    tensor: a tensor to be exported
 
 The dlpack shares the tensors memory.
 Note that each dlpack can only be consumed once.