#COOLTERM TUTORIAL HOW TO#
#COOLTERM TUTORIAL FOR MAC#
#COOLTERM TUTORIAL INSTALL#
#COOLTERM TUTORIAL UPDATE#
#COOLTERM TUTORIAL SOFTWARE#

Compile : uTensor_util.cpp Compile : quantization_utils.cpp Link: my_uTensor Elf2Bin: my_uTensor | Module |. Expect to see the similar compilation message. We are also using a custom build profile here to enable C++11 support.

#COOLTERM TUTORIAL UPDATE#

You may want to update this to the target name of your board. The Mbed-cli needs to know what board it is compiling for, in this case, K66F. Now, we can compile the whole project by issuing: $ mbed compile -m K66F -t GCC_ARM The application has to keep the input memory-block safe for during inferencing. The data is arranged in row-major layout in memory (same as any C array). In practice, this would be buffered sensor data or any memory-block containing the input data. In this example, the static array defined in the input_data.h is being used as the input for inferencing. The name of the output tensor is the same as the output node’s as specified by your training script. The Context class, now contains the inference graph, can be evaluated to produce an output tensor containing the inference result. It automatically populates a Context object with the inference graph, takes a Tensor class as input. The get_deep_mlp() is a generated function. The Context class is the playground where the inference takes place. It depends on how the network is setup.Īt this stage, we should have an Mbed project containing:Īll we need is a main.cpp to tie everything together: The name of the output node is shown in the training message in the previous section. Specifying the output node helps uTensor-cli to traverse the graph and apply optimisations. Generate source file: models/deep_mlp.cpp Generate header file: models/deep_mlp.hpp. Generate weight file: models/deep_mlp_weight.hpp. Transforming graph: mnist_model/deep_mlp.pb. The command for generating the C++ file is: $ utensor-cli convert mnist_model/ deep_mlp.pb -output-nodes= y_pred.

deep_mlp_weight.hpp Contains the weight of the neural network.

#COOLTERM TUTORIAL HOW TO#

In this case, a function signature: get_deep_mlp_ctx(…) We shall see how to use this in the main.cpp later.

deep_mlp.hp p This is the interface to the embedded program.

deep_mlp.cpp Contains the implementation of the model.

Turning a graph, deep_mlp.pb, into C++ files : $ ls mnist_model/ deep_mlp.pbĭeep_mlp.pb is what we will supply to uTensor-cli for C++ code generation in the next step. It is what we will supply to uTensor-cli for C++ code generation in the next step. written graph to: mnist_model/deep_mlp.pb the output nodes: Ī protocol buffer that contains the trained model will be saved to the file system. step 19000, training accuracy 0.92 step 20000, training accuracy 0.94 test accuracy 0.9274 saving checkpoint: chkps/mnist_model Converted 6 variables to const ops. Running the script you should see something like: $ python3 deep_mlp.py. The script defines the MLP and its training parameters. $ mbed add $ ls mbed-os mbed_settings.py uTensor.lib mbed-os.lib uTensor It contains all the function implementations that will be linked during the compilation time. We will need the uTensor runtime library. $ mbed new my_uTensor $ cd my_uTensor/ $ ls mbed-os mbed-os.lib mbed_settings.py

This may take a couple minutes as the entire Mbed-OS’s source history will be downloaded. Let’s start with creating a new Mbed project.

#COOLTERM TUTORIAL INSTALL#

Getting uTensor-cli: $ pip3 install utensor_cgenįinally, grab CoolTerm from Roger Meier’s website. $ pip3 install mbed-cli $ mbed -version 1.10.1 Check out the official doc for more details. Mbed-cli will need Mercural and Git: $ brew install mercurial git

However, do not work with your system Python directly, it is a good idea to use a virtual environment. If you already have Python installed, skip this step. Next, install a working version of GCC-arm cross-compiler: $ brew install In your terminal: $ /usr/bin/ruby -e “$(curl -fsSL )' $ brew update

#COOLTERM TUTORIAL SOFTWARE#

Brew is handy for installing software packages on your Mac. These tools allow you to compile MCUs projects with your favorite IDE/editor on Windows/Mac/Linux.

#COOLTERM TUTORIAL FOR MAC#

Although these instructions are for Mac OS, they are applicable to other operating systems. It covers tool installations, training of the neural network, generating C++ files, setting up Mbed projects and the deployment. In this tutorial, we will be using uTensor with Mbed and TensorFlow.

Coolterm tutorial