LanceViewer

View your lance datasets as interactive dataframes.

Installation

pip install lanceviewer

Usage

Once installed, simply launch the app using lanceviewer command

lanceviewer

There are additional settings you can pass that are applied golabally.

Multi-Modal data support [WIP]

You can visualize multimodal data stored as bytes in the lance dataset. Currently Image bytes are handled with support for more types coming soon

Transforms for MultiModal data

You need to write configuration maps to handle the columns that contain multi-modal bytes they won't be rendered in the browser There are 2 configs written to handle img columns:

• MJ: This configuration assumes that the "img" column contains the image bytes as ["extension", "bytes"]
• Default: This configuration assumes that the "img" column contains image bytes as the top level, i.e, ["bytes"]

NOTE: Currently, MJ is the default golbal setting

Changing configuration settings

• To view the global settings

lanceviewer settings

Changing Default Transformation

• To set the a different column configuration, you can just pass --transforms {NAME}. The settings are persisted in a global settings file so next time you don't to pass it again

Working with Local Lance Datasets

• Simply pass the path of the of the lance dataset as the URI.
• And then pass the SQL query. Loading dataset and running queries locally are almost real-time.

Working with remote buckets

GCS

When working with remote gcs bucket data, you need to make sure the dataset uri you're passing is accessible to you. If not already set up, please follow ADC as described in [https://cloud.google.com/docs/authentication/external/set-up-adc]

Note: Loading the querying speed depends directly on bandwidth.