TIME Forecasting Benchmark
Test zero-shot forecasting on tasks configured around real operational needs.
Explore the original evidence
Explore the original evaluation, its methods, and reported findings.
Open evalCoverage
The paper introduces 50 datasets and 98 forecasting tasks. Authors: Zhongzheng Qiao, Sheng Pan, Anni Wang, Viktoriya Zhukova, Yong Liu, Xudong Jiang, Qingsong Wen, Mingsheng Long, Ming Jin and Chenghao Liu.
Methodology
Keep each task’s frequency, horizon and variate settings fixed. Aggregate window-level forecasts against the Seasonal Naive reference and inspect results by dataset and temporal pattern. The repository records May 2026 data corrections and a CC BY-NC 4.0 dataset license. Its data-loading code adapts GIFT-Eval, but the tasks and benchmark identity are distinct. No dataset files or model scores are reproduced.