A collection of LLM services you can self host via docker or modal labs to support your applications development

The goal of this repo is to provide a series of docker containers, or modal labs deployments of common patterns when using LLMs and provide endpoints that allows you to intergrate easily with existing codebases that use the popular openai api.

• Support GPT4all's embedding api and match it to openai.com/v1/embedding
• Support JSONFormer api to match it to chatcompletion with function_calls
• Support Cross Encoders based on sentence transformers for any huggingface model
• Provide great documentation and runbooks using MkDocs

Before you begin, ensure that you have the following installed on your system:

1. Git: Visit the Git website to download and install Git on your operating system.
2. Docker: Visit the Docker website to download and install Docker on your operating system.

To clone the repository to your local machine, follow these steps:

1. Open a terminal or command prompt.

2. Change the current directory to where you want to store the repository.

3. Run the following command to clone the repository:

git clone https://github.com/jxnl/fastllm.git

This will create a local copy of the repository on your machine.

Once you have cloned the repository, navigate to the root directory of the application where the Dockerfile is located.

To build the Docker image, use the docker build command. The syntax is as follows:

docker build -t <image_name>:<tag> .

• <image_name>: Choose a meaningful name for your Docker image, e.g., my-app.
• <tag>: Specify a version or tag for your image, e.g., v1.0.

Replace <image_name> and <tag> with your desired values. The . at the end of the command refers to the current directory, which is where the Dockerfile resides.

Example:

docker build -t my-app:v1.0 .

The Docker build process will read the Dockerfile instructions and create an image containing your application and its dependencies.

Once the Docker image is built, you can run a container based on that image:

docker run -d -p <host_port>:8000 <image_name>:<tag>

• <host_port>: The port number on your host machine where you want to access the application.
• 8000: The port number exposed by the application inside the Docker container.

Example:

docker run -d -p 8000:8000 my-app:v1.0

After running the container, you can access the application running inside the container through your web browser or any HTTP client by visiting http://localhost:8000.

To start using a Docker image from Docker Hub, follow these steps:

1. Open a terminal or command prompt.

2. Run the following command to pull the desired Docker image from Docker Hub:

docker pull jxnl/embed-gpt4all
docker run -d -p 8000:8000 jxnl/embed-gpt4all:latest

import openai

openai.api_base = "http://localhost:8000"

This will download the xnl/embed-gpt4all image from Docker Hub with the "latest" tag.

Once the Docker image is pulled, you can run a container based on that image:

This command will run the jxnl/embed-gpt4all container in detached mode (-d) and forward port 8000 on the host to port 8000 inside the container.

After running the container, you can access the application running inside the container through your web browser or any HTTP client by visiting http://localhost:8000.

To stop a running container, use the docker stop command followed by the container ID or name:

docker stop <container_id_or_name>

To remove a stopped container, use the docker rm command followed by the container ID or name:

docker rm <container_id_or_name>

To stop a running container, use the docker stop command followed by the container ID or name:

docker stop <container_id_or_name>

To remove a stopped container, use the docker rm command followed by the container ID or name:

docker rm <container_id_or_name>

To remove the Docker image, use the docker rmi command followed by the image ID or name:

docker rmi <image_id_or_name>

Remember that removing an image is irreversible and cannot be undone.

Contributions are welcome! If you have any suggestions, improvements, or bug fixes, please open an issue or submit a pull request.

This project is licensed under the MIT License