ritual/projects/gpt4/gpt4.md

Running OpenAI's GPT-4 on Infernet

In this tutorial we are going to integrate OpenAI's GPT-4 into infernet. We will:

1. Obtain an API key from OpenAI
2. Configure the gpt4 service, build & deploy it with Infernet
3. Make a web-2 request by directly prompting the gpt4 service
4. Make a web-3 request by integrating a sample PromptsGPT.sol smart contract. This contract will make a request to Infernet with their prompt, and receive the result of the request.

Install Pre-requisites

For this tutorial you'll need to have the following installed.

1. Docker
2. Foundry

Get an API key from OpenAI

First, you'll need to get an API key from OpenAI. You can do this by making an OpenAI account. After signing in, head over to their platform to make an API key.

Note

You will need a paid account to use the GPT-4 API.

Ensure docker & foundry exist

To check for docker, run the following command in your terminal:

docker --version
# Docker version 25.0.2, build 29cf629 (example output)

You'll also need to ensure that docker-compose exists in your terminal:

which docker-compose
# /usr/local/bin/docker-compose (example output)

To check for foundry, run the following command in your terminal:

forge --version
# forge 0.2.0 (551bcb5 2024-02-28T07:40:42.782478000Z) (example output)

Clone the starter repository

Just like our other examples, we're going to clone this repository. All of the code and instructions for this tutorial can be found in the projects/gpt4 directory of the repository.

# Clone locally
git clone --recurse-submodules https://github.com/ritual-net/infernet-container-starter
# Navigate to the repository
cd infernet-container-starter

Configure the gpt4 container

Configure API key in config.json

This is where we'll use the API key we obtained from OpenAI.

cd projects/gpt4/container
cp config.sample.json config.json

In the containers field, you will see the following. Replace your-openai-key with your OpenAI API key.

"containers": [
    {
        // etc. etc.
        "env": {
            "OPENAI_API_KEY": "your-openai-key" // replace with your OpenAI API key
        }
    }
],

Build the gpt4 container

First, navigate back to the root of the repository. Then simply run the following command to build the gpt4 container:

cd ../../..
make build-container project=gpt4

Deploy infernet node locally

Much like our hello world project, deploying the infernet node is as simple as running:

make deploy-container project=gpt4

Making a Web2 Request

From here, you can directly make a request to the infernet node:

curl -X POST http://127.0.0.1:4000/api/jobs \
     -H "Content-Type: application/json" \
     -d '{"containers":["gpt4"], "data": {"prompt": "Hello, can shrimp actually fry rice?"}}'
# {"id":"cab6eea8-8b1e-4144-9a70-f905c5ef375b"}

If you have jq installed, you can pipe the output of the last command to a file:

curl -X POST http://127.0.0.1:4000/api/jobs \
     -H "Content-Type: application/json" \
     -d '{"containers":["gpt4"], "data": {"prompt": "Hello, can shrimp actually fry rice?"}}' | jq -r ".id" > last-job.uuid

You can then check the status of the job by running:

curl -X GET http://127.0.0.1:4000/api/jobs\?id\=cab6eea8-8b1e-4144-9a70-f905c5ef375b
# response [{"id":"07026571-edc8-42ab-b38c-6b3cf19971b6","result":{"container":"gpt4","output":{"message":"No, shrimps cannot fry rice by themselves. However, in culinary terms, shrimp fried rice is a popular dish in which cooked shrimp are added to fried rice along with other ingredients. Cooks or chefs prepare it by frying the rice and shrimps together usually in a wok or frying pan."}},"status":"success"}]

And if you have jq installed and piped the last output to a file, you can instead run:

curl -X GET "http://127.0.0.1:4000/api/jobs?id=$(cat last-job.uuid)" | jq .
# returns something like:
[
  {
    "id": "1b50e85b-2295-44eb-9c85-40ae5331bd14",
    "result": {
      "container": "gpt4",
      "output": {
        "output": "Yes, shrimp can be used to make fried rice. In many Asian cuisines, shrimp is a popular ingredient in fried rice dishes. The shrimp adds flavor and protein to the dish, and can be cooked along with the rice and other ingredients such as vegetables, eggs, and seasonings."
      }
    },
    "status": "success"
  }
]

Making a Web3 Request

Now let's bring this service onchain! First we'll have to deploy the contracts. The contracts directory contains a simple foundry project with a simple contract called PromptsGpt. This contract exposes a single function function promptGPT(string calldata prompt). Using this function you'll be able to make an infernet request.

Anvil Logs: First, it's useful to look at the logs of the anvil node to see what's going on. In a new terminal, run docker logs -f anvil-node.

Deploying the contracts: In another terminal, run the following command:

make deploy-contracts project=gpt4

Calling the contract

Now, let's call the contract. So far everything's been identical to the hello world project. The only difference here is that calling the contract requires an input. We'll pass that input in using an env var named prompt:

make call-contract project=gpt4 prompt="Can shrimps actually fry rice"

On your anvil logs, you should see something like this:

eth_sendRawTransaction

_____  _____ _______ _    _         _
|  __ \|_   _|__   __| |  | |  /\   | |
| |__) | | |    | |  | |  | | /  \  | |
|  _  /  | |    | |  | |  | |/ /\ \ | |
| | \ \ _| |_   | |  | |__| / ____ \| |____
|_|  \_\_____|  |_|   \____/_/    \_\______|


subscription Id 1
interval 1
redundancy 1
node 0x70997970C51812dc3A010C7d01b50e0d17dc79C8
output: {'output': 'Yes, shrimps can be used to make fried rice. Fried rice is a versatile dish that can be made with various ingredients, including shrimp. Shrimp fried rice is a popular dish in many cuisines, especially in Asian cuisine.'}

    Transaction: 0x9bcab42cf7348953eaf107ca0ca539cb27f3843c1bb08cf359484c71fcf44d2b
    Gas used: 93726

    Block Number: 3
    Block Hash: 0x1cc39d03bb1d69ea7f32db85d2ee684071e28b6d6de9eab6f57e011e11a7ed08
    Block Time: "Fri, 26 Jan 2024 02:30:37 +0000"

beautiful, isn't it? 🥰