HelloPDF - Deployable GPT-4 Powered PDF Chat

Run and deploy a GPT-4 powered chatbot in minutes!

Utilizes ChromaDB for its vectorstore, with a Next.js frontend.

Development

1. Install Docker Desktop for your platform.

2. Clone the repo or download the ZIP

git clone [github https url]

3. Install packages

First run npm install yarn -g to install yarn globally (if you haven't already).

Then run:

yarn install

After installation, you should now see a node_modules folder.

4. Set up your .env file

• Copy .env.example into .env Your .env file should look like this:

OPENAI_API_KEY=
CHROMA_AUTH_BASIC=
CHROMA_AUTH_TOKEN=
CHROMA_URL=
COLLECTION_NAME=[optional]

• Visit openai to retrieve API keys and insert into your .env file.
• Choose a collection name where you'd like to store your embeddings in Chroma. This collection will later be used for queries and retrieval.
• Chroma details

5. Depending on your setup, you may need to modify app/api/files/utilities.ts to connect to the right ChromaDB instance.

6. In a new terminal window, run Chroma in the Docker container:

docker run -p 8000:8000 ghcr.io/chroma-core/chroma:latest

Run the app

You can run the app with npm run dev to launch the local dev environment, and then upload one or many PDF files to chat with. After uploading, you'll be able to chat with the model.

ChromaDB Deployment

The terraform folder contains scripts originally from chromadb/examples. To deploy your ChromaDB to GCP, do as follows:

1. Install GCP CLI, log in via CLI, and create a new project. Note the project ID.

2. Install terraform CLI.

3. Update terraform/exportapply.sh with your project ID variable.

4. (Optional) Generate a keypair if you want to be able to SSH in to the GCP instance.

5. Run exportapply.sh in your terminal.

6. Run terraform output instance_public_ip. Take note of the output IP, and update your .env.

7. Run terraform output chroma_auth_token. Take note of your auth token, and update your .env.

8. (optional) It takes some time for the GCP instance to come up, so you can check on the status with

% export instance_public_ip=$(terraform output instance_public_ip | sed 's/"//g')
% curl -v http://$instance_public_ip:8000/api/v1/heartbeat

9. The API should now be able to connect with your GCP-deployed ChromaDB instance for vector store operations.

Troubleshooting

In general, keep an eye out in the issues and discussions section of this repo for solutions.

General errors

• Make sure you're running the latest Node version. Run node -v
• Try a different PDF or convert your PDF to text first. It's possible your PDF is corrupted, scanned, or requires OCR to convert to text.
• Console.log the env variables and make sure they are exposed.
• Check that you've created an .env file that contains your valid (and working) API keys, environment and index name.
• If you change modelName in OpenAI, make sure you have access to the api for the appropriate model.
• Make sure you have enough OpenAI credits and a valid card on your billings account.
• Check that you don't have multiple OPENAPI keys in your global environment. If you do, the local env file from the project will be overwritten by systems env variable.
• Try to hard code your API keys into the process.env variables if there are still issues.

Credits

Originally forked from https://github.com/mayooear/gpt4-pdf-chatbot-langchain/tree/feat/chroma