Mistral AI releases Timestral-123M time series forecasting model on Hugging Face
Overview
Mistral AI releases Timestral-123M, a 123M-parameter time series foundation model for zero-shot forecasting, on Hugging Face.
The model accepts context of up to 32,768 time steps and forecasts up to 1,024 steps per pass, outputting 21 quantiles. Mistral reports a MASE of 0.659 and CRPS of 0.448 on the GIFT-Eval benchmark. The checkpoint is released under a Modified MIT license, with the license terms linked from the model repository.
Written by AI from the articles below · updated Oct 15, 12:07 PM ET
Check the sources:
Article timeline
The articles in this story. Times are ET.
- Mistral AI · new models on Hugging FaceOfficialMistral releases Timestral-123M, a 123M-parameter time series forecasting model
AIMistral AI releases Timestral-123M, a 123M-parameter time series foundation model for zero-shot forecasting, on Hugging Face. The model accepts context of up to 32,768 time steps and forecasts up to 1,024 steps per pass, outputting 21 quantiles. On the GIFT-Eval benchmark, it scores a MASE of 0.659 and CRPS of 0.448, and it is released under a Modified MIT license.
Heat trend
Not enough continuous observations to show a trend yet.