Google’s research team has introduced TimesFM-3, a multivariate time-series forecasting foundation model with 330 million parameters. The model was pretrained on more than 1 trillion time points drawn from real-world and synthetic series, and it can handle multiple targets, historical covariates, and known future covariates without task-specific fine-tuning. According to MarkTechPost, TimesFM-3 ranked first on average among pretrained foundation models across point forecasting and probabilistic metrics on the GIFT-Eval, fev-bench, and TIME leaderboards. The release also comes with a usage limitation: weights for version 3.0 are under a non-commercial, non-production-use license, which means they cannot be used in production forecasting APIs. The update was reported by Techub News in a brief citing MarkTechPost.
Google’s research team has released TimesFM-3, a multivariate time-series forecasting foundation model with 330 million parameters, according to a Techub News brief citing MarkTechPost.
The model was pretrained on more than 1 trillion time points from real-world and synthetic series. It can handle multiple targets, historical covariates, and known future covariates without task-specific fine-tuning.
On performance, TimesFM-3 ranked first on average among pretrained foundation models in both point forecasting and probabilistic metrics on the GIFT-Eval, fev-bench, and TIME leaderboards.
Still, the weights for TimesFM version 3.0 are restricted by a non-commercial, non-production-use license and cannot be used in production forecasting APIs.
This article was originally published by Bit.Fan. For more cryptocurrency news and market insights, visit www.bit.fan. Disclaimer:
The market information, project data, and third-party content displayed on this platform are for industry information sharing only and do not constitute any form of investment advice or return commitment.
Cryptocurrency trading carries high risks. Users should fully assess their risk tolerance and make independent decisions. All profits, losses, and legal responsibilities are borne by the users themselves.
