Skip to content

Latest commit

 

History

History
85 lines (62 loc) · 3.63 KB

README.md

File metadata and controls

85 lines (62 loc) · 3.63 KB

AIMET model zoo package

This page provides information on the installable python pip packages for the AIMET model zoo and the instructions to generate them.

Table of Contents

Install package

NOTE: Follow this section in order to install and use a released version of the AIMET model zoo package (recommended).

Go to https://github.com/quic/aimet-model-zoo/releases and identify the release tag of the package you want to install.

Set the <variant_string> to ONE of the following depending on your desired variant

  • For the PyTorch variant, use "torch"
  • For the TensorFlow variant, use "tensorflow"
export aimet_zoo_variant=<variant_string>

Replace <release_version> in the steps below with the appropriate tag:

export release_version=<release_version>

Set the package download URL as follows:

export download_url="https://github.com/quic/aimet-model-zoo/releases/download/${release_version}"

Install the AIMET packages in the order specified below:

python3 -m pip install ${download_url}/aimet_zoo_${aimet_zoo_variant}-${release_version}-py3-none-any.whl

NOTE: Dependencies (if any) will NOT get installed automatically. For any additional pre-requisite packages, please check the appropriate .md within the subfolder in TensorFlow or PyTorch folders corresponding to your model(s) of interest.

Build new package

NOTE: Follow this section in order to locally build a new AIMET model zoo package(s) from sources.

Please read this section for the host pre-requisities and this section for the recommended environment setup.

Upgrade the setuptools package to the latest version as follows:
python3 -m pip install --upgrade setuptools

To obtain the code, first define a workspace and follow these instructions:

WORKSPACE="<absolute_path_to_workspace>"
mkdir $WORKSPACE && cd $_
git clone https://github.com/quic/aimet-model-zoo.git

Follow these instructions to build the AIMET model zoo code:

NOTE: If you are inside the docker, set WORKSPACE="<absolute_path_to_workspace>" again.

cd $WORKSPACE/aimet
mkdir build && cd build

# Run cmake (set flags in the below command depending on your needs)
# To include torch, use -DENABLE_TORCH=ON. To exclude torch, use -DENABLE_TORCH=OFF.
# To include tensorflow, use -DENABLE_TENSORFLOW=ON. To exclude tensorflow, use -DENABLE_TENSORFLOW=OFF.
cmake .. -DENABLE_TORCH=ON -DENABLE_TENSORFLOW=ON -DDEPLOYMENT_PATH="<path_to_deploy_tarball>" -DPIP_CONFIG_FILE="<pip_repo_config_file>" -DPIP_INDEX="<name_of_repo_index>" -DPIP_CERT_FILE="<ca_certificate_file>"

make

make install

Generate the package as follows:
make packagemodelzoo

Once the installation step is complete, the wheel files AND the consolidated tarball package is created at: $WORKSPACE/aimet-model-zoo/build/packaging/.

Deploy the package tarball to the location specified (via the cmake command above):
make deploy

Upload the python pip package to your repository (using the destination URL specified via the cmake command above):
make upload

NOTE: The upload make target does NOT work yet. You may need to upload manually until this is fixed.

Install the package(s) that you just built as follows:

python3 -m pip install $WORKSPACE/aimet-model-zoo/build/packaging/aimet_zoo_${aimet_zoo_variant}-${release_version}-py3-none-any.whl