Demand forecasting with the Temporal Fusion Transformer

Hello everyone,

I want to use TFT model for my use case. I am able to train the model using the tutorial provided in the below link

However, I am facing issues with the shape of the output that the TFT Model is producing.

I am using a tabular data where there are 6 columns. I am using these 6 columns to predict the values of a single column among these 6. I am trying to test the trained TFT model on 5000 samples but when I am checking the shape of ouput it is only 2851x1.

Also I am not sure whether this model is built to predict multiple targets i.e. values of multiple columns. Because the same output shape issue is coming in this case as well.

Is there any blog or help where I can understand the solution to this issue?

Please help me it is little urgent.