# Time-Series-Library Benchmark Collection

Canonical source: https://huggingface.co/datasets/thuml/Time-Series-Library

Official code: https://github.com/thuml/Time-Series-Library

Paper: https://arxiv.org/abs/2407.13278

## Dataset Type

Time-series benchmark collection covering long-term forecasting, short-term forecasting, imputation, anomaly detection, and classification.

## Temporal Structure

The long-term forecasting subset includes ETT, Electricity, Traffic, Weather, Exchange, and ILI with MSE/MAE horizons such as 96 to 720 steps. Toto's LSF evaluation uses ETTh1, ETTh2, ETTm1, ETTm2, Electricity, and Weather.

## Actions Or Interventions

None. These are passive forecasting and representation/evaluation tasks.

## Suitability Note

This entry is the wiki's durable handle for the legacy LSF/LTSF benchmark family referenced by Toto. These datasets are useful for continuity with older forecasting literature, but they are low-dimensional relative to BOOM and Time-HD and can be saturated by modern supervised baselines.

## Access And License Notes

The Hugging Face dataset page lists CC BY 4.0. Component-source terms should still be checked before downstream redistribution or operational use.