Welcome to Wicker’s documentation!

Wicker is a Python framework for storing and serving Machine Learning datasets. It provides:

Abstraction of underlying storage

Read and write data as Python dictionaries - don’t worry about where/how it is stored.

Schematized Data

Enforce a schema when reading and writing your data, providing readability and useability for users of your datasets.

Metadata Indexing

Easily store, query and visualize your metadata in your own stores such as BigQuery or Snowflake for downstream analyses.

Efficient Dataloading

Built and optimized for high-throughput distributed training of deep learning models.

Indices and tables