IBM has released an open-source ensemble forecasting model called Granite Timeseries Ensemble, available on Hugging Face. This model combines four distinct Granite time-series architectures to provide more robust predictions without requiring users to train on their specific datasets. The ensemble offers both median and quantile forecasts, drawing inspiration from weather forecasting ensembles like the North American Multi-Model Ensemble to reduce the risk associated with selecting a single model. AI
IMPACT Provides a more robust and user-friendly approach to time-series forecasting by combining multiple models, reducing the need for individual model selection and training.
RANK_REASON Release of an open-source ensemble model for time-series forecasting.
- FlowState
- Granite Timeseries Ensemble
- Granite TSFM
- Hugging Face
- IBM Granite
- ibm-granite/granite-timeseries-ensemble-r1
- North American Multi-Model Ensemble
- PatchTST-FM-r1
- PatchTST-FM-r2
- QuantileEnsembleForecaster
- Tiny Time Mixers
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
