Time Series Prediction

🤗Transformers
1

Greetings,
I would like to predict a time series via TimeSeriesTransformerForPrediction described here: Time Series Transformer but I only got past_values and future_values and I get the error:

TypeError: forward () missing 4 required positional arguments: ‘past_time_features’, ‘past_observed_mask’, ‘static_categorical_features’, and ‘static_real_features’

Where am I wrong?

This is the code :

import torch
import torch.nn as nn

x_train = torch.from_numpy(x_train).type(torch.Tensor)
x_test = torch.from_numpy(x_test).type(torch.Tensor)
y_train = torch.from_numpy(y_train).type(torch.Tensor)
y_test = torch.from_numpy(y_test).type(torch.Tensor)


from transformers import TimeSeriesTransformerForPrediction

model = TimeSeriesTransformerForPrediction.from_pretrained(

    "huggingface/time-series-transformer-tourism-monthly"

)

# during training, one provides both past and future values

# as well as possible additional features

outputs = model(

    past_values=x_train,

    #past_time_features=None,

    #past_observed_mask=None,

    #static_categorical_features=None,

    #static_real_features=None,

    future_values=x_test,

    #future_time_features=None

)

loss = outputs.loss

loss.backward()
2

it looks like all of the columns mentioned in the error message have to be provided, have you tried providing some dummy values like in batch example?