Lib.rs

› Science › Machine learning › Burn
#machine-learning #data-pipeline #deep-learning #data-transformation #data-source #data

burn-dataset

Library with simple dataset APIs for creating ML data pipelines

by Nathaniel Simard and 97 contributors

19 releases (breaking)

0.15.0 Oct 28, 2024
0.13.2 May 3, 2024
0.12.1 Feb 1, 2024
0.11.1 Dec 4, 2023
0.3.0 Nov 20, 2022

#1071 in Machine learning

Download history 1669/week @ 2024-08-20 1774/week @ 2024-08-27 1401/week @ 2024-09-03 1547/week @ 2024-09-10 1264/week @ 2024-09-17 1648/week @ 2024-09-24 1407/week @ 2024-10-01 1401/week @ 2024-10-08 1520/week @ 2024-10-15 1756/week @ 2024-10-22 1827/week @ 2024-10-29 1388/week @ 2024-11-05 1177/week @ 2024-11-12 1315/week @ 2024-11-19 3144/week @ 2024-11-26 3148/week @ 2024-12-03

8,933 downloads per month
Used in 25 crates (via burn-core)

MIT/Apache

130KB
3K SLoC

Burn Dataset

Burn dataset library

Current Crates.io Version license

The Burn Dataset library is designed to streamline your machine learning (ML) data pipeline creation process. It offers a variety of dataset implementations, transformation functions, and data sources.

Feature Flags

  • audio - enables audio dataset (SpeechCommandsDataset). Run the following example to try it out:

    cargo run --example speech_commands --features audio

lib.rs:

Burn Dataset

Burn Dataset is a library for creating and loading datasets.

Dependencies

~7–32MB
~456K SLoC