#RAG Chatbot Modifications

 [LlamaIndex](https://www.llamaindex.ai/) and [Together.ai](https://www.together.ai/) RAG chatbot

## Getting Started

Copy your `.example.env` file into a `.env` and add values to the env variables

1. Install the dependencies.

```
npm install
```

2. Generate the embeddings and store them locally in the `cache` folder. You can also provide a PDF in the `data` folder instead of the default one.

```
npm run generate
```

3. Run the app and send messages to your chatbot. It will use context from the embeddings to answer questions.

```
npm run dev
```

## How to deploy
Make sure you have `docker`, `terraform`, and `aws` cli tools installed locally AND have a dockerhub account and an AWS account AND your `aws` cli tool is configured properly with credentials as `terraform` will need the credentials to deploy resources.

### Prepare docker image

#### Build docker image
```
docker buildx build --platform linux/amd64 . -t <your_docker_hub_name>/headstarter-ai-chatbot
```

#### Log-in your dockerhub account
```
docker login
```

#### Push your image to your repository, make sure the repository public (by default it is)
```
docker push <your_docker_hub_name>/headstarter-ai-chatbot 
```

### Deploy EC2 and run container inside using Terraform

Before running `terraform` commands, make sure 
- your AWS credentials `AWS Access Key ID` and `AWS Secret Access Key` are congiured in `aws` cli locally.
- have a SSH pub/private key pair ready at `~/.ssh/id_rsa` and `~/.ssh/id_rsa.pub`
- your local env var file is at root dir and callled `.env.local`

```
cd deploy
terraform init
terraform apply --auto-approve
```

### Destroy EC2

```
terraform destroy --auto-approve
```

## Documentation
- [Together AI Documentation](https://docs.together.ai/docs) - learn about Together.ai (inference, finetuning, embeddings)
- [LlamaIndex Documentation](https://docs.llamaindex.ai) - learn about LlamaIndex (Python features).
- [LlamaIndexTS Documentation](https://ts.llamaindex.ai) - learn about LlamaIndex (Typescript features).