TTI-Worker

Serving Text To Image Model Using FastAPI and Celery

How to start

Using docker-compose(recommended)

1. Create RabbitMQ and FastAPI container(refer to TTI-FastAPI)

2. Clone repository

git clone https://github.com/ainize-team/TTI-Worker
cd TTI-Worker

3. Edit docker-compose.yml and .env file for your project.

• Vhost for each worker is in docker-compose.yml and common RabbitMQ config is in .env file.

4. Run worker container

docker-compose up -d

# If you want to run a specific worker container, write service name.
docker-compose up -d <service name>

Using docker

1. Run RabbitMQ comtainer as a broker

docker run -d --name tti-rabbitmq -p 5672:5672 -p 15672:15672 --restart=unless-stopped rabbitmq:3.11.2-management

2. Clone repository

git clone https://github.com/ainize-team/TTI-Worker
cd TTI-Worker

3. Build docker image

docker build -t tti-worker .

4. Create docker container

docker run -d --name <worker_container_name> \
--gpus='"device=0"' -e BROKER_URI=<broker_uri> \
-e DATABASE_URL=<firebase_realtime_database_url> \
-e STORAGE_BUCKET=<firebase_storage_url> \
-v <firebase_credential_path>:/app/key -v <model_local_path>:/app/model \
tti-worker

Or, you can use the .env file to run as follows.

docker run -d --name <worker_container_name> \
--gpus='"device=0"' \
--env-file <env filename> \
-v <firebase_credential_path>:/app/key -v <model_local_path>:/app/model \
tti-worker

Test with FastAPI

• Check our TTI-FastAPI Repo.

For Developers

1. Install dev package

pip install -r requirements.txt
pip install -r requirements-dev.txt

2. Install pre-commit

pre-commit install