Get Relay tutorials working with Colab

gallery/how_to/work_with_relay/build_gcn.py

@@ -25,6 +25,15 @@
 Cora dataset is a common benchmark for Graph Neural Networks (GNN) and frameworks that support GNN training and inference.
 We directly load the dataset from DGL library to do the apples to apples comparison against DGL.
 
+To start, we'll install Apache TVM, DGL, and PyTorch using Pip:
+
+.. code-block:: bash
+
+    %%shell
+    pip install apache-tvm
+    pip install torch==1.9.0
+    pip install dgl==v0.7.2 -f https://data.dgl.ai/wheels/repo.html
+
 Please refer to DGL doc for DGL installation at
 https://docs.dgl.ai/install/index.html.
 

gallery/how_to/work_with_relay/using_external_lib.py

@@ -29,7 +29,13 @@
 Before we can use external libraries from Relay, your TVM needs to be built with libraries you want to use.
 For example, to use cuDNN, USE_CUDNN option in `cmake/config.cmake` needs to be enabled, and cuDNN include and library directories need to be specified if necessary.
 
-To begin with, we import Relay and TVM.
+To start, we'll install Apache TVM and import it:
+
+.. code-block:: bash
+
+    %%shell
+    pip install apache-tvm
+
 """
 
 # sphinx_gallery_start_ignore

gallery/how_to/work_with_relay/using_pipeline_executor.py

@@ -20,6 +20,12 @@
 **Author**: `Hua Jiang <https://github.com/huajsj>`_
 
 This is a short tutorial on how to use "Pipeline Executor" with Relay.
+
+.. code-block:: bash
+
+    %%shell
+    pip install apache-tvm
+
 """
 import tvm
 from tvm import te

gallery/how_to/work_with_relay/using_relay_viz.py

@@ -34,6 +34,12 @@
 We will introduce how to implement customized parsers and renderers through interface classes.
 
 For more details, please refer to :py:mod:`tvm.contrib.relay_viz`.
+
+.. code-block:: bash
+
+    %%shell
+    pip install apache-tvm
+
 """
 
 # sphinx_gallery_start_ignore