thuml/Large-Time-Series-Model
Timer is a generative pre-trained transformer foundation model for time series forecasting, zero-shot prediction, and unified temporal analysis tasks.

This repository provides official code, pre-trained checkpoints, and datasets for Timer and subsequent time series foundation models including Sundial and Timer-XL. It supports zero-shot probabilistic forecasting across multiple time series domains using transformer-based generative architecture. Models are released on HuggingFace for download and the repo includes quickstart notebooks and data loaders for the UTSD dataset.
Frequently asked
- What is thuml/Large-Time-Series-Model?
- Timer is a generative pre-trained transformer foundation model for time series forecasting, zero-shot prediction, and unified temporal analysis tasks.
- Is Large-Time-Series-Model open source?
- Yes — thuml/Large-Time-Series-Model is open source, released under the MIT license.
- What language is Large-Time-Series-Model written in?
- thuml/Large-Time-Series-Model is primarily written in Python.
- How popular is Large-Time-Series-Model?
- thuml/Large-Time-Series-Model has 1k stars on GitHub.
- Where can I find Large-Time-Series-Model?
- thuml/Large-Time-Series-Model is on GitHub at https://github.com/thuml/Large-Time-Series-Model.