Vanna AI/Ollama/DuckDB: Learn to Connect Database and Query in Natural Language with Vanna AI+Ollama and automated Visualization with Plotly, Other Important Tools/Database/LLM in this Article are ChromaDB/Jupyter Notebook/Mistral.
Vanna AI
Chat with your SQL database. Accurate Text-to-SQL Generation via LLMs using RAG
Let Vanna.AI write your SQL for you
The fastest way to get actionable insights from your database just by asking questions
Why Vanna
Open-Source
The Vanna Python package and the various frontend integrations are all open-source. You can run Vanna on your own infrastructure.
High accuracy on complex datasets
Vanna’s capabilities are tied to the training data you give it. More training data means better accuracy for large and complex datasets.
Designed for security
Your database contents are never sent to the LLM unless you specifically enable features that require it. The metadata storage layer only sees schemas, documentation, and queries.
Self learning
As you use Vanna more, your model continuously improves as we augment your training data.
Supports many databases
Snowflake , BigQuery, Postgres, and many others. You can easily make a connector for any database.
Choose your front end
Start in a Jupyter Notebook. Expose to business users via Slackbot, web app, Streamlit app, any other frontend. Even integrate in your web app for customers.
Ollama: Large Language Model Runner.
ChromaDB:
the AI-native open-source embedding database
Simple: Fully-typed, fully-tested, fully-documented == happiness
Integrations: LangChain (python and js), LlamaIndex and more soon
Dev, Test, Prod: the same API that runs in your python notebook, scales to your cluster
Feature-rich: Queries, filtering, density estimation and more
Free & Open Source: Apache 2.0 Licensed
Use case: Chat
For example, the “Chat your data” use case:
1. Add documents to your database. You can pass in your own embeddings, embedding function, or let Chroma embed them for you.
2. Query relevant documents with natural language.
3. Compose documents into the context window of an LLM like GPT3 for additional summarisation or analysis.
DuckDB is a fast in-process analytical database
DuckDB supports a feature-rich SQL dialect complemented with deep integrations into client APIs
Simple
DuckDB is easy to install and deploy. It has zero external dependencies and runs in-process in its host application or as a single binary.
Portable
DuckDB runs on Linux, macOS, Windows, and all popular hardware architectures. It has idiomatic client APIs for major programming languages.
Feature-rich
DuckDB offers a rich SQL dialect. It can read and write file formats such as CSV, Parquet, and JSON, to and from the local file system and remote endpoints such as S3 buckets.
Fast
DuckDB runs analytical queries at blazing speed thanks to its columnar engine, which supports parallel execution and can process larger-than-memory workloads.
Extensible
DuckDB is extensible by third-party features such as new data types, functions, file formats and new SQL syntax.
Free
DuckDB and its core extensions are open-source under the permissive MIT License.
JupyterLab is the latest web-based interactive development environment for notebooks, code, and data.
Its flexible interface allows users to configure and arrange workflows in data science, scientific computing, computational journalism, and machine learning. A modular design invites extensions to expand and enrich functionality.
Plotly’s Python graphing library makes interactive, publication-quality graphs. Examples of how to make line plots, scatter plots, area charts, bar charts, error bars, box plots, histograms, heatmaps, subplots, multiple-axes, polar charts, and bubble charts. Plotly.py is free and open source and you can view the source, report issues or contribute on GitHub
Now here is How to get started.
Step 01: Open Jupyter Notebook and type below command
%pip install 'vanna[chromadb,duckdb]'
Step 02: Now after running first command type below command in Jupyter Notebook.
from vanna.ollama import Ollama
from vanna.chromadb import ChromaDB_VectorStore
Step 03: Now Connect Vanna/Chromadb/Ollama-Mistral Model by typing and running below command in Jupyter Notebook
class MyVanna(ChromaDB_VectorStore, Ollama):
def __init__(self, config=None):
ChromaDB_VectorStore.__init__(self, config=config)
Ollama.__init__(self, config=config)
vn = MyVanna(config={'model': 'mistral'})
Step 04: Now Connect with DuckDB with below command,
vn.connect_to_duckdb(url='abc.duckdb')
Step 05: You can training you data once unless you have added other data with below command.
# The information schema query may need some tweaking depending on your database. This is a good starting point.
df_information_schema = vn.run_sql("SELECT * FROM INFORMATION_SCHEMA.COLUMNS")
# This will break up the information schema into bite-sized chunks that can be referenced by the LLM
plan = vn.get_training_plan_generic(df_information_schema)
plan
# If you like the plan, then uncomment this and run it to train
# vn.train(plan=plan)
Step 06: At any time you can inspect what training data the package is able to reference
training_data = vn.get_training_data()
training_data
Step 07: You can remove training data if there’s obsolete/incorrect information.
vn.remove_training_data(id='1-ddl')Step 08: Now You can start flask app by typing and running below command in Jupyter Notebook, App will be available at localhost:8084 and start Asking your question in Natural Language.
from vanna.flask import VannaFlaskApp
app = VannaFlaskApp(vn)
app.run()
Step 09: Now Start asking question in Natural Language First it will show results and then generate automated Visualization based on Plotly like show below.
Results
Then Automated Visualization
Step 10: You can ask other question
Q2
Answer
Visualization
Other Articles for Vanna AI+Ollama are here
