[doc] Autodiff: fix typo + add reference to torch example script

docs/src/inverse_rendering/diff_render.rst

@@ -162,7 +162,7 @@ samples per pixel (``spp``).
 .. code-block:: python
 
     # Render a reference image (no derivatives used yet)
-    from mitsuba.python.autodiff import render, render, write_bitmap
+    from mitsuba.python.autodiff import render, write_bitmap
     image_ref = render(scene, spp=8)
     crop_size = scene.sensors()[0].film().crop_size()
     write_bitmap('out_ref.png', image_ref, crop_size)
@@ -261,6 +261,7 @@ gradient steps.
 
 .. code-block:: python
 
+    # for loop body (continued)
         # Objective: MSE between 'image' and 'image_ref'
         ob_val = ek.hsum(ek.sqr(image - image_ref)) / len(image)
 

docs/src/inverse_rendering/pytorch.rst

@@ -112,3 +112,8 @@ The main optimization loop looks as follows:
         # Compare iterate against ground-truth value
         err_ref = objective(params_torch['red.reflectance.value'], param_ref)
         print('Iteration %03i: error=%g' % (it, err_ref * 3))
+
+.. note::
+
+    The full Python script of this tutorial can be found in the file:
+    :file:`docs/examples/10_diff_render/invert_cbox_torch.py`.