Fix how_to/compile_models to work in Colab

gallery/how_to/compile_models/from_coreml.py

@@ -23,13 +23,12 @@
 
 This article is an introductory tutorial to deploy CoreML models with Relay.
 
-For us to begin with, coremltools module is required to be installed.
-
-A quick solution is to install via pip
+To begin, we must install coremltools and Apache TVM:
 
 .. code-block:: bash
 
-    pip install -U coremltools --user
+    %%shell
+    pip install apache-tvm coremltools
 
 or please refer to official site
 https://github.com/apple/coremltools

gallery/how_to/compile_models/from_darknet.py

@@ -27,8 +27,9 @@
 
 .. code-block:: bash
 
-  pip install cffi
-  pip install opencv-python
+  %%shell
+  pip install apache-tvm cffi opencv-python
+
 """
 
 # sphinx_gallery_start_ignore

gallery/how_to/compile_models/from_keras.py

@@ -19,7 +19,7 @@
 =====================
 **Author**: `Yuwei Hu <https://Huyuwei.github.io/>`_
 
-This article is an introductory tutorial to deploy keras models with Relay.
+This article is an introductory tutorial to deploy Keras models with Relay.
 
 For us to begin with, keras should be installed.
 Tensorflow is also required since it's used as the default backend of keras.
@@ -28,8 +28,8 @@
 
 .. code-block:: bash
 
-    pip install -U keras --user
-    pip install -U tensorflow --user
+    %%shell
+    pip install apache-tvm keras tensorflow
 
 or please refer to official site
 https://keras.io/#installation
@@ -102,8 +102,8 @@
 shape_dict = {"input_1": data.shape}
 mod, params = relay.frontend.from_keras(keras_resnet50, shape_dict)
 # compile the model
-target = "cuda"
-dev = tvm.cuda(0)
+target = "llvm"
+dev = tvm.cpu(0)
 
 # TODO(mbs): opt_level=3 causes nn.contrib_conv2d_winograd_weight_transform
 # to end up in the module which fails memory validation on cuda most likely

gallery/how_to/compile_models/from_mxnet.py

@@ -24,13 +24,12 @@
 
 This article is an introductory tutorial to deploy mxnet models with Relay.
 
-For us to begin with, mxnet module is required to be installed.
-
-A quick solution is
+To begin, we must install `mxnet` and `apache-tvm`:
 
 .. code-block:: bash
 
-    pip install mxnet --user
+    %%shell
+    pip install apache-tvm mxnet
 
 or please refer to official installation guide.
 https://mxnet.apache.org/versions/master/install/index.html
@@ -102,7 +101,7 @@ def transform_image(image):
 
 ######################################################################
 # now compile the graph
-target = "cuda"
+target = tvm.target.Target("llvm")
 with tvm.transform.PassContext(opt_level=3):
     lib = relay.build(func, target, params=params)
 
@@ -112,7 +111,7 @@ def transform_image(image):
 # Now, we would like to reproduce the same forward computation using TVM.
 from tvm.contrib import graph_executor
 
-dev = tvm.cuda(0)
+dev = tvm.cpu(0)
 dtype = "float32"
 m = graph_executor.GraphModule(lib["default"](dev))
 # set inputs

gallery/how_to/compile_models/from_oneflow.py

@@ -27,8 +27,10 @@
 
 .. code-block:: bash
 
+    %%shell
+    pip install apache-tvm
     pip install flowvision==0.1.0
-    python3 -m pip install -f https://release.oneflow.info oneflow==0.7.0+cpu
+    pip install -f https://release.oneflow.info oneflow==0.7.0+cpu
 
 or please refer to official site:
 https://github.com/Oneflow-Inc/oneflow
@@ -113,7 +115,7 @@ def build(self, x):
 # Relay Build
 # -----------
 # Compile the graph to llvm target with given input specification.
-target = tvm.target.Target("llvm", host="llvm")
+target = tvm.target.Target("llvm")
 dev = tvm.cpu(0)
 with tvm.transform.PassContext(opt_level=3):
     lib = relay.build(mod, target=target, params=params)
@@ -122,9 +124,8 @@ def build(self, x):
 # Execute the portable graph on TVM
 # ---------------------------------
 # Now we can try deploying the compiled model on target.
-target = "cuda"
 with tvm.transform.PassContext(opt_level=10):
-    intrp = relay.build_module.create_executor("graph", mod, tvm.cuda(0), target)
+    intrp = relay.build_module.create_executor("graph", mod, tvm.cpu(0), target)
 
 print(type(img))
 print(img.shape)

gallery/how_to/compile_models/from_onnx.py

@@ -27,7 +27,8 @@
 
 .. code-block:: bash
 
-    pip install --user onnx onnxoptimizer
+    %%shell
+    pip install apache-tvm onnx onnxoptimizer
 
 or please refer to official site.
 https://github.com/onnx/onnx

gallery/how_to/compile_models/from_paddle.py

@@ -20,14 +20,16 @@
 **Author**: `Ziyuan Ma <https://github.com/ZiyuanMa/>`_
 
 This article is an introductory tutorial to deploy PaddlePaddle models with Relay.
-For us to begin with, PaddlePaddle>=2.1.3 is required to be installed.
+To begin, we'll install Apache TVM and PaddlePaddle>=2.1.3.
 A quick solution is
 
 .. code-block:: bash
 
+    %%shell
+    pip install apache-tvm
     pip install paddlepaddle -i https://mirror.baidu.com/pypi/simple
 
-or please refer to official site.
+For more details, refer to the official install instructions at:
 https://www.paddlepaddle.org.cn/install/quick?docurl=/documentation/docs/zh/install/pip/linux-pip.html
 """
 

gallery/how_to/compile_models/from_pytorch.py

@@ -28,6 +28,8 @@
 
 .. code-block:: bash
 
+    %%shell
+    pip install apache-tvm
     pip install torch==1.7.0
     pip install torchvision==0.8.1
 

gallery/how_to/compile_models/from_tensorflow.py

@@ -21,6 +21,11 @@
 
 For us to begin with, tensorflow python module is required to be installed.
 
+.. code-block:: bash
+
+    %%bash
+    pip install apache-tvm tensorflow
+
 Please refer to https://www.tensorflow.org/install
 """
 

gallery/how_to/compile_models/from_tflite.py

@@ -25,9 +25,8 @@
 
 .. code-block:: bash
 
-    # install tflite
-    pip install tflite==2.1.0 --user
-
+    %%shell
+    pip install apache-tvm tflite==2.1.0
 
 or you could generate TFLite package yourself. The steps are the following:
 

gallery/how_to/work_with_microtvm/micro_train.py

@@ -60,7 +60,7 @@
 #
 #     .. code-block:: bash
 #
-#       %%bash
+#       %%shell
 #       pip install -q tensorflow tflite
 #       pip install -q tlcpack-nightly -f https://tlcpack.ai/wheels
 #       apt-get -qq install imagemagick curl
@@ -504,7 +504,7 @@ def representative_dataset():
 #
 #     .. code-block:: bash
 #
-#       %%bash
+#       %%shell
 #       mkdir -p ~/tests
 #       curl "https://i.imgur.com/JBbEhxN.png" -o ~/tests/car_224.png
 #       convert ~/tests/car_224.png -resize 64 ~/tests/car_64.png