I wish to build a Hockey team predictor for my first OpenAI Project. I will also be using langchain, openai and weaviate for this project.
I wish to store historical sport scores, stats, games, history inside a vector database. Also may store text from news sites, hockey forums, etc.
Data for the sport scores, stats, and games will be from a returned JSON object.
I would like it to be free to first build on and if it becomes scalable, then transfer to the cloud and start paying for necessary fees.
Can I lets say install:
weaviate to local virtual environment
use it with python and store/create new vectors to hold the sports data. The more the better.
If I installed weaviate to my local venv, that means I have that data saved somewhere where I can export just in case? Or transfer to cloud, etc. How much space do vector db’s take? Would data like this require lots of space or money to save?
Sorry for so many questions…any insight would be appreciated.
2 - If you configure the LLM API_KEYS (ex: OpenAi, Cohere, etc) within you docker-compose.yml, whenever you throw data in, Weaviate will vectorize it for you. You will be able to do that with python, or javascript/typescript, go… or even a simple rest api call.
So now you have all your data locally. Awesome!
If you ever want to move your data around (from one server to another, or one class to another one) and do not vectorize everything again, you can follow this doc as a guide:
Regarding choosing Weaviate:
I believe that if your project needs a vector database, I am pretty sure you need to choose Weaviate
