Fix link: Add missing #

Author: MarkDaoust

site/en/tutorials/distribute/input.ipynb

@@ -93,7 +93,7 @@
         "id": "lNy9GxjSlMKQ"
       },
       "source": [
-        "To use `tf.distribute` APIs to scale, it is recommended that users use `tf.data.Dataset` to represent their input. `tf.distribute` has been made to work efficiently with `tf.data.Dataset` (for example, automatic prefetch of data onto each accelerator device) with performance optimizations being regularly incorporated into the implementation. If you have a use case for using something other than `tf.data.Dataset`, please refer a later [section](\"https://www.tensorflow.org/guide/distributed_training\") in this guide.\n",
+        "To use `tf.distribute` APIs to scale, it is recommended that users use `tf.data.Dataset` to represent their input. `tf.distribute` has been made to work efficiently with `tf.data.Dataset` (for example, automatic prefetch of data onto each accelerator device) with performance optimizations being regularly incorporated into the implementation. If you have a use case for using something other than `tf.data.Dataset`, please refer a later [section](#tensorinputs) in this guide.\n",
         "In a non distributed training loop, users first create a `tf.data.Dataset` instance and then iterate over the elements. For example:\n"
       ]
     },