search

moment-timeseries-foundation-model / moment

MOMENT: A Family of Open Time-series Foundation Models
https://moment-timeseries-foundation-model.github.io/
MIT License
216 stars 26 forks source link

Input features of the first layer #20

Open dara1400 opened 1 month ago

dara1400 commented 1 month ago

Hello I hope you are doing well.

model = MOMENTPipeline.from_pretrained( "AutonLab/MOMENT-1-large", model_kwargs={ "task_name": "forecasting", "forecast_horizon": 96, }, ) model.init()

When calling the model, I noticed that the input features of the first layer is 8. Dose it mean that the model use the 8 previous inputs to predict the next point?

If it is the case, how can I change it?

dara1400 commented 1 month ago

I understand that the model always use 512 previous data to predict new point. Can this be changed?