Skip to main content Link Menu Expand (external link) Document Search 复制 已复制

Foundation Models

Packages Description Paper
MOMENT Open time-series foundation models from CMU/Auton Lab for forecasting, anomaly detection, classification, imputation, and embeddings. Paper Link
IBM Granite TSFM IBM’s open repository for time series foundation models, notebooks, demos, and TinyTimeMixer workflows. IBM Docs
Timer / Sundial THUML repository for large generative time series models such as Timer, Timer-XL, and Sundial. Timer Paper
OpenLTM Unified codebase for developing and evaluating large time-series models and recent LLM4TS implementations. N/A
Time-LLM Time-LLM is a reprogramming framework to repurpose LLMs for general time series forecasting with the backbone language models kept intact. Paper Link
MOMENT From CMU. A family of open-source foundation models for general-purpose time-series analysis. Paper Link
TimesFM From Google. A time-series foundation model for forecasting whose out-of-the-box zero-shot performance on a variety of public datasets comes close to the accuracy of state-of-the-art supervised forecasting models for each individual dataset. Paper Link
Lag-Llama A general-purpose foundation model for univariate probabilistic forecasting. Paper Link
MOIRAI From Salesforce. An encoder-only Transformer model, specialized in universal time-series forecasting. Paper Link
TimeGPT The authors claim it’s the first foundation model for time series forecasting. Paper Link
Chronos From Amazon. Chronos is a family of pretrained time series forecasting models based on language model architectures. Paper Link

GitHub Resources

Resource Description
Awesome Time Series Broad GitHub collection of libraries, datasets, and resources across the time series ecosystem.
TS-PTMs Survey Repo Survey companion repository summarizing time-series pre-trained models and their task coverage.
Awesome-time-series Papers Paper-centric GitHub summary of forecasting, classification, anomaly detection, and benchmark datasets.