add small notes to autograd tc tutorial

Author: prigoyal

docs/source/framework/pytorch_integration/autograd_with_tc.rst

@@ -3,8 +3,8 @@ Autograd with TC
 
 We provide the TC intergation with PyTorch `autograd` so that it is easy to write
 a training layer with TC and be able to run backwards as well if the layer is part
-of a network. In order to write a training layer with TC, you need to follow the
-steps below:
+of a network. We do not support double backwards right now. In order to write a
+training layer with TC, you need to follow the steps below:
 
 1. Define your TC language that has two definitions: one for the forward layer and the other for the backward layer and pass it to :code:`tc.define` call. In addition, also pass :code:`training=True` and the name of the backward TC :code:`backward`.
 

docs/source/tutorials/tutorial_tensordot_with_tc.rst

@@ -36,7 +36,7 @@ to TensorDot operation.
 
 A simple 2D matrix multiply operation in TC is expressed as:
 
-.. code-block:: python
+.. code::
 
      def matmul(float(M, N) X, float(N, K) W) -> (output) {
          output(m, k) +=! X(m, nn) * W(nn, k)
@@ -47,7 +47,7 @@ The variable :code:`nn` is being reduced in above expression. Now, let's write a
 **batched matrix-multiply** operation using above expression. For that, we need to
 add a batch dimension to it and the expression becomes:
 
-.. code-block:: python
+.. code::
 
      def batch_matmul(float(B, M, N) X, float(B, N, K) W) -> (output) {
          output(b, m, k) +=! X(b, m, nn) * W(b, nn, k)
@@ -56,7 +56,7 @@ add a batch dimension to it and the expression becomes:
 Now, for the tensordot operation, we need to add spatial dimensions :code:`H` and :code:`W`
 to the batched matrix multiply, and the expression for TensorDot becomes:
 
-.. code-block:: python
+.. code::
 
      def tensordot(float(B, C1, C2, H, W) I0, float(B, C2, C3, H, W) I1) -> (O) {
          O(b, c1, c3, h, w) +=! I0(b, c1, c2, h, w) * I1(b, c2, c3, h, w)