search

thuml / Autoformer

About Code release for "Autoformer: Decomposition Transformers with Auto-Correlation for Long-Term Series Forecasting" (NeurIPS 2021), https://arxiv.org/abs/2106.13008
MIT License
2k stars 430 forks source link

Snapshot of input data #109

Closed oremnirv closed 1 year ago

oremnirv commented 1 year ago

Thank you very much for the lovely paper.

I'm trying to implement your model.

Can I kindly ask for a snapshot of what the (x_enc, x_mark_enc, x_dec, x_mark_dec) look like? i.e., an example of one row from the data frame (preferably in a pandas view with column names)?

Thank you!

wuhaixu2016 commented 1 year ago

Hi, sorry for the delayed response. Let us take the encoder inputs as the example: x_enc is in the shape of BLC, where B is the batch dimension, L is the temporal dimension and C is the variate dimension. x_mark_enc is in the shape of BL5, which is the date information of each time point. For each time point of each sample, the 115 dimension organized in (month, day, weekday, hour, minute).