How?
- We'll try to keep this as simple as possible and not get lost in the weeds
- You will be able to ask a question and the NL2SQL engine will be able to figure out the schema of the db and generate the SQL
- The SQL will be executed by the agent. For that, we'll use LangChain
We'll try to do all of this in Jupyter notebooks. Hopefully you can follow along:
We'll load up and use a subset of the titanic dataset from Kaggle.
You'll need to install:
- vscode
- jupyter notebook extension
- open the nl2sql jupyter notebook and follow along
...and you are ok with Docker...
you may need to adjust these commands if you do NOT use WSL
docker pull mcr.microsoft.com/mssql/server:2019-latest
# basic syntax
docker run -e 'ACCEPT_EULA=Y' \
-e 'SA_PASSWORD=Password01!!' \
-p 1433:1433 \
-h DockerSQL \
--name DockerSQL \
-d \
mcr.microsoft.com/mssql/server:2019-latest
docker ps -a