LUMA 1 CHAT

This repository contains server code for the Luma-1 drum machine PDF documentation, enabling the ability to chat against it in real time.

Table of Contents

• Environment Variables
• Doing Embeddings
• Running the Server With Docker
• Running the Server Without Docker
• Sending Requests

Environment Variables

All functionality of this server requires:

1. An OpenAI API key (model access)
2. A Pinecone API key (vector store access)

Start by going to Pinecone to create a new account. Create a new index with the name of luma-1-index and a dimension of 1536. Next, create a new API key in Pinecone and then create an .env file at the root of this project with the following environment variable in it:

PINECONE_API_KEY="<api key here>"

Next, go to OpenAI's developer dashboard and get an API key. Put this in your .env file as well like this:

OPENAI_API_KEY="<api key here>"

Doing Embeddings

Before you can chat against the document, you'll need to embed it in a vector store. Assuming that you've done the Environment Variables section above, you can run the following command to embed the PDF document data:

make embed

Running the Server With Docker

NOTE: This section requires that you have Docker installed locally. Also, make sure you've completed the Environment Variables section above before continuing on.

First, set an environment variable in your .env file for your server host:

SERVER_HOST=0.0.0.0

Run the following commands to build a new docker container locally:

make build

Once that's done, run the docker container with this command:

make run-docker

You'll now be able to hit http://localhost:3200/v1/chat to ask questions against the docs.

Running the Server Without Docker

Make sure you've completed the Environment Variables section above before continuing on.

First, set an environment variable in your .env file for your server host:

SERVER_HOST=127.0.0.1

First, create a new virtual environment:

python3 -m venv venv

Next, activate it:

source /venv/bin/activate

Finally, go ahead and install the required dependencies with this make command:

make install

Once you've done this, you can run the server locally with this command:

make run

You'll now be able to hit http://localhost:3200/v1/chat to ask questions against the docs.

Sending Requests

Once you have the server running, you can send requests to http://localhost:3200/v1/chat like this:

Request:

{
  "prompt": "How can I add new sounds?"
}

Response:

{
  "result": "To add new sounds to the Luma-1, you can use the Load Voice Bank command to load downloaded sounds onto the device via its internal SD card or the Luma-1 Web Application. You can put your sample files into specific folders for different drum sounds, ensuring they are in u-law format and no larger than 32KB. You can also build a custom Voice Bank from EPROM and SysEx sounds and save it to the SD card. To load your new Drum Bank, navigate through the menu and select the folder containing your loaded sounds. The Luma-1 will process the new samples and return to normal drum machine operation."
}