Datasets:
metadata
license: apache-2.0
task_categories:
- time-series-forecasting
pretty_name: BLAST
size_categories:
- 10M<n<100M
tags:
- time-series
- pre-training
- universal-forecasting
- foundation-model
This repository hosts the BLAST dataset introduced in the paper BLAST: Balanced Sampling Time Series Corpus for Universal Forecasting Models.
To learn how to generate this dataset, please refer to the instructions provided here.
For guidance on using BLAST to train universal forecasting models—such as TimeMoE or ChronosBolt—please see the BLAST tutorial in BasicTS repository.
The BLAST-trained checkpoints can be found here.