MLDatasets.jl: Access to Common Machine Learning Benchmarks in Julia

JuliaML maintained MLDatasets.jl as a utility for accessing common machine learning benchmarks, with the most recent update in March 2026. The package provides a standardized interface for loading data directly into Julia-native structures for model training and evaluation.

Use Cases

• Training convolutional neural networks using the MNIST or CIFAR-10 image loaders
• Evaluating clustering algorithms on the Iris or Wine tabular datasets
• Developing graph-based models using the Cora or CiteSeer graph datasets

Strengths

• MIT licensed for unrestricted commercial and academic use
• Maintained by the official JuliaML organization
• Provides native Julia data structures like Arrays and DataFrames

Limitations

• Acts as a metadata and downloader wrapper rather than a primary data host
• Requires internet access for the initial download of each dataset
• Dataset availability depends on the stability of external third-party servers

Provenance

Source

JuliaML Organization

Collection Method

Aggregated from various public machine learning repositories

Freshness

Updated as of March 2026.