mehrdadh commented on code in PR #13627: URL: https://github.com/apache/tvm/pull/13627#discussion_r1061966967
########## gallery/how_to/work_with_microtvm/micro_tflite.py: ########## @@ -26,109 +26,36 @@ """ ###################################################################### -# .. note:: -# If you want to run this tutorial on the microTVM Reference VM, download the Jupyter -# notebook using the link at the bottom of this page and save it into the TVM directory. Then: # -# #. Login to the reference VM with a modified ``vagrant ssh`` command: +# .. include:: ../../../../gallery/how_to/work_with_microtvm/install_dependencies.rst # -# ``$ vagrant ssh -- -L8888:localhost:8888`` -# -# #. Install jupyter: ``pip install jupyterlab`` -# #. ``cd`` to the TVM directory. -# #. Install tflite: poetry install -E importer-tflite -# #. Launch Jupyter Notebook: ``jupyter notebook`` -# #. Copy the localhost URL displayed, and paste it into your browser. -# #. Navigate to saved Jupyter Notebook (``.ipynb`` file). -# -# -# Setup -# ----- -# -# Install TFLite -# ^^^^^^^^^^^^^^ -# -# To get started, TFLite package needs to be installed as prerequisite. You can do this in two ways: -# -# 1. Install tflite with ``pip`` -# -# .. code-block:: bash -# -# pip install tflite=2.1.0 --user -# -# 2. Generate the TFLite package yourself. The steps are the following: -# -# Get the flatc compiler. -# Please refer to https://github.com/google/flatbuffers for details -# and make sure it is properly installed. -# -# .. code-block:: bash -# -# flatc --version -# -# Get the TFLite schema. -# -# .. code-block:: bash -# -# wget https://raw.githubusercontent.com/tensorflow/tensorflow/r1.13/tensorflow/lite/schema/schema.fbs -# -# Generate TFLite package. -# -# .. code-block:: bash -# -# flatc --python schema.fbs -# -# Add the current folder (which contains generated tflite module) to PYTHONPATH. -# -# .. code-block:: bash -# -# export PYTHONPATH=${PYTHONPATH:+$PYTHONPATH:}$(pwd) -# -# To validate that the TFLite package was installed successfully, ``python -c "import tflite"`` -# -# Install Zephyr (physical hardware only) -# ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ -# -# When running this tutorial with a host simulation (the default), you can use the host ``gcc`` to -# build a firmware image that simulates the device. When compiling to run on physical hardware, you -# need to install a *toolchain* plus some target-specific dependencies. microTVM allows you to -# supply any compiler and runtime that can launch the TVM RPC server, but to get started, this -# tutorial relies on the Zephyr RTOS to provide these pieces. -# -# You can install Zephyr by following the -# `Installation Instructions <https://docs.zephyrproject.org/latest/getting_started/index.html>`_. -# -# Aside: Recreating your own Pre-Trained TFLite model -# The tutorial downloads a pretrained TFLite model. When working with microcontrollers -# you need to be mindful these are highly resource constrained devices as such standard -# models like MobileNet may not fit into their modest memory. -# -# For this tutorial, we'll make use of one of the TF Micro example models. -# -# If you wish to replicate the training steps see: -# https://github.com/tensorflow/tensorflow/tree/master/tensorflow/lite/micro/examples/hello_world/train -# -# .. note:: -# -# If you accidentally download the example pretrained model from: -# -# ``wget https://storage.googleapis.com/download.tensorflow.org/models/tflite/micro/hello_world_2020_04_13.zip`` -# -# this will fail due to an unimplemented opcode (114) -# -# Load and prepare the Pre-Trained Model -# -------------------------------------- -# -# Load the pretrained TFLite model from a file in your current -# directory into a buffer # sphinx_gallery_start_ignore from tvm import testing testing.utils.install_request_hook(depth=3) # sphinx_gallery_end_ignore +# You can skip the following two sections (installing Zephyr and CMSIS-NN) if the following flag is False. Review Comment: This is because it will run on CRT, so Zephyr and CMSIS-NN dependencies are not required -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: commits-unsubscr...@tvm.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
