You can run SQL queries on your datasets to explore and analyze your data.

​
How does it work?

When you open the SQL Notebook, a local SQLite database is created in your browser’s local storage. All your queries are run against this database.

You can run most SQL queries supported by SQLite.

Queries will be saved for each dataset so your team can share queries and results.

Note: Since the dataset is stored in your browser’s local storage, it may consume a lot of memory for large datasets.

​
Available Tables & Columns

The sidebar on the left will indicate which tables and columns are accessible.

Pro tip: You can also run SQL queries when comparing datasets.

​
Example Queries

-- Select all
select * from dataset;

-- Select all with filters
select * from dataset where "Response Faithfulness" > 0.9;

-- Get average eval score grouped by topic
select topic, avg("Factual Accuracy")
from dataset
group by topic;

Note: Any changes you make to the database are not persisted to the dataset when you close the browser tab.