Create an account on Google Cloud Platform.

Head over to ai.google.dev and get the Gemini API Key (Save the API somewhere safe for later).

Install Python 3.11 (Refer Python documentation).

Clone the repository and cd inside it,

$ git clone https://github.com/swarajpande5/genai-rag-primer.git
$ cd genai-rag-primer

Create a virtual environment and activate the same,

$ python3.11 -m venv venv
$ source venv/bin/activate

Install the project and its dependencies,

$ pip install -e .

Add the GEMINI_API_KEY as the environment variable,

$ export GEMINI_API_KEY=<PASTE THE API KEY>

OR

Create a .env file and specify the GEMINI_API_KEY inside it,

# .env file
GEMINI_API_KEY=<PASTE THE API KEY>

(Optional) The following environment variables can be specified in the same way as above.

• CHROMADB_PATH : The path on the local system where Chroma vector database will be created (Default: $HOME/chromadb).

• COLLECTION_NAME : The name of the collection inside Chroma vector database where vector embeddings will be stored (Default: gemini-rag).

• GEMINI_EMBEDDING_MODEL : The model which will be used to create embeddings (Default: models/embedding-001).

• GEMINI_MODEL : The LLM which will be used (Default: gemini-pro).

• N_RESULTS : The number of relevant results from vector database to fetch, which will be passed to the model later (Default: 3).

  Note: The number of matching embeddings can be less as compared to N_RESULTS.

Start the CLI application,

$ python3.11 src/__main__.py