How to use the docker image

The UTGen Docker files can be found under the Docker directiory of the UTGen repository. The Docker image is built using the Dockerfile and the docker-compose.yml file is used to run the experiment.

We have multiple variants of the Docker image:

1. client.11.Dockerfile - This Dockerfile is used to build the UTGen client image with the version 11 of the OpenJDK.
2. client.8.Dockerfile - This Dockerfile is used to build the UTGen client image with the version 8 of the OpenJDK.
3. llm-server.cpu.Dockerfile - This Dockerfile is used to build the LLM Server image with the CPU compute engine.
4. llm-server.nvidia.Dockerfile - This Dockerfile is used to build the LLM Server image with the Nvidia GPU compute engine.
5. docker-compose.yml - This file is used to run the experiment with the UTGen client and the LLM Server.
6. docker-compose.nvidia.yml - This file is used to run the experiment with the UTGen client and the LLM Server with the Nvidia GPU compute engine.

How to build the Docker image

Start the LLM Server

Run the following command to start the ollama llm-server service:

docker-compose -f docker-compose.yml start ollama llm-server

Start the UTGen Client

Once the llm-server service is running, and the model is running, you can start the utgen-client service:

docker-compose -f docker-compose.yml start utgen-client