3LA Integration

This is to integrate all parts (e.g., TVM-BYOC and ILAng generated simulators) into a single Docker image.

Build the docker image

RSA key-pair is required to access the private repos. For password protected key-pair:

docker build --build-arg SSH_KEY="$(openssl rsa -in ~/.ssh/id_rsa)" --tag byo3la --file Dockerfile .

For non-password-protected key-pair:

docker build --build-arg SSH_KEY="$(cat ~/.ssh/id_rsa)" --tag byo3la --file Dockerfile .

To remove intermediate images (licensed packages and your private keys):

docker rmi -f $(docker images -q --filter label=stage=intermediate)

Examples

Run the speech-to-text example with FlexNLP linear layer (simulation):

docker run byo3la /bin/bash /root/scripts/run_speech_to_text.sh