Zero-Shot Forecasting
The ability of a machine learning model to make predictions directly on target data that it has not seen and has not been fine-tuned on, relying on general pattern recognition capabilities acquired through large-scale, diverse pretraining.
Timeline (last 90 days)
HotPOT showed performance gains in zero-shot prediction under unseen physical conditions, indicating that it learned transferable numerical-physical representations rather than overfitting to the training equation families.
TimesFM draws on the pre-training + zero-shot inference paradigm of large language models, enabling direct forecasting of unseen time series after pre-training without targeted training.
All Facts (2)
HotPOT showed performance gains in zero-shot prediction under unseen physical conditions, indicating that it learned transferable numerical-physical representations rather than overfitting to the training equation families.
50%UnverifiedTimesFM draws on the pre-training + zero-shot inference paradigm of large language models, enabling direct forecasting of unseen time series after pre-training without targeted training.
50%