Chronos Bolt Tiny is a time series forecasting model hosted on Hugging Face. It belongs to the class of pretrained foundation models for time series and is made available by Amazon under the apache-2.0 license.
The model uses the t5 architecture and is distributed in safetensors format. It supports probabilistic forecasting by producing outputs at specified quantile levels. Integration occurs through the Chronos library, which supplies a BaseChronosPipeline class for loading the model and generating predictions from pandas dataframes.
Usage begins with installation of the chronos-forecasting package followed by creation of a pipeline object that can target cuda devices. The pipeline accepts a context dataframe containing historical observations, a prediction length parameter that defines how many future steps to forecast, a list of quantile levels such as 0.1 0.5 and 0.9, and an identifier column name. An example workflow loads the AirPassengers dataset from a public S3 bucket and calls predict_df to obtain a forecast dataframe.
The repository lists related arxiv papers 1910.10683 and 2403.07815 that describe the underlying research. It is one of several chronos variants offered for time series foundation model tasks.
Chronos Bolt Tiny sits in PulseGate's Data science & ML workbench category. It provides accurate time series forecasting for predictive analytics and business decision-making. It is built as an open-source project for data scientists and analysts. Chronos Bolt Tiny is open source under the Open Source license. It runs on the web, the command line, and API, and it can be self-hosted.
It is developed by Amazon, and the product first shipped in 2024. Key capabilities include time series forecasting, predictive analytics, and chronos integration.
Latest indexed changes and source events
amazon/chronos-bolt-tiny verified by the PulseGate indexer
Other apps tracked under the same category.
