Parseltongue is a Python application that enables LLM-powered chat with your Obsidian notes. It allows you to create and persist embeddings of your Obsidian notes in a local Qdrant vector database. You can then query these embeddings and use GPT-3.5-turbo to generate a response.
- Embedding Creation: Parseltongue uses a local Qdrant vector database to create embeddings of your Obsidian notes. These embeddings capture the semantic meaning of your notes.
- Querying: You can query the Qdrant vector database to retrieve relevant embeddings based on a given query. This allows you to find notes that are related to a specific topic or keyword.
- Response Generation: Once the relevant embeddings are retrieved, Parseltongue utilizes GPT-3.5-turbo, a powerful language model, to generate a response based on the query and the retrieved embeddings. This response can provide insights, suggestions, or answers related to your query.
To use Parseltongue, follow these steps:
- Make sure you have Python 3.12 installed.
- Create a virtual environment:
python -m venv .venv - Activate the virtual environment:
source .venv/bin/activate - Install required dependencies by running
pip install -r requirements.txt. - Set up a local Qdrant vector database. You can refer to the Qdrant quickstart guide.
- Configure the connection details in the application
.envfile. You can refer to.env.example, as it contains all configurations needed. - Run your Qdrant service:
docker run -p 6333:6333 -p 6334:6334 \
-v $/path/to/local/storage:/qdrant/storage:z \
qdrant/qdrant- Run the application with
uvicorn app:app --reload. - Start playing!
Here's an example cURL of how you can use Parseltongue:
curl --location 'http://localhost:8000/chat' \
--header 'Accept: application/json' \
--header 'Content-Type: application/json' \
--data '{
"user_input": "organising knowledge"
}'