(1/2) Moirai - Salesforce's Foundation Forecasting Model 🚀
Salesforce recently released Moirari - a new #Python 🐍 library with a foundation model for time series forecasting applications. According to the release blog - the model comes with universal forecasting capabilities and can handle multiple scenarios and different frequencies.
(2/2) The model was trained on a large-scale dataset with 27 billion observations using data from nine distinct domains.
Following the release of Moirai and Amazon's Chronos model, I expect to see the release of additional foundation models for time series forecasting in the near future.
@ramikrispin any idea if there are time series embedding models too? seems like it would be powerful to say, “what you’re currently seeing is like these 4 other instances here”
@kellogh I believe that this is the idea behind this (and similar) models. They are trained on huge dataset and classify by their characteristics. For new time series the model idenityf based on the series characteristics what would be the best approach (based on what the model was train on).
Add comment