Skip to content

Latest commit

 

History

History
68 lines (42 loc) · 2.15 KB

using_docker.md

File metadata and controls

68 lines (42 loc) · 2.15 KB
Raw

Installation using Docker

You can directly pull the pre-build docker images for DELTA and DELTANN. We have created the following docker images:

  • delta-gpu-py3

  • delta-cpu-py3

  • deltann-gpu-py3

  • deltann-cpu-py3

Install Docker

Make sure docker has been installed. You can refer to the official tutorial.

Pull Docker Image

You can build DETLA or DETLANN locally as Build Images, or using pre-build images as belows:

All avaible image tags list in here, please choose one as needed.

If you choose delta-cpu-py3, then download the image as below:

docker pull zh794390558/delta:delta-cpu-py3

Create Container

After the image downloaded, create a container.

For delta usage (model development):

cd /path/to/detla && docker run -v `pwd`:/delta -it zh794390558/delta:delta-cpu-py3 /bin/bash

The basic version of delta (except Kaldi) was already installed in this container. You can develop in this container like:

# Add DELTA enviornment
source env.sh

# Generate mock data for text classification.
pushd egs/mock_text_cls_data/text_cls/v1
./run.sh
popd

# Train the model
python3 delta/main.py --cmd train_and_eval --config egs/mock_text_cls_data/text_cls/v1/config/han-cls.yml

For deltann usage (model deployment):

cd /path/to/detla 
WORKSPACE=$PWD
docker run -it -v $WORKSPACE:$WORKSPACE zh794390558/delta:deltann-cpu-py3 /bin/bash

We recommend using a high-end machine to develop DELTANN, since it needs to compile Tensorflow which is time-consuming.