I am trying Informer architecture and wanted to ask whether it is normal that my Train Loss is higher than Validation and Test Losses. I tried to split my data on 60%/20%/20%; 70%/10%/20% and other configurations. I am trying multivariate (M) setting with default parameters and seq_len=96, label_len=48, pred_len=24. Thanks for help.
Hi, thank you for your work!
I am trying Informer architecture and wanted to ask whether it is normal that my Train Loss is higher than Validation and Test Losses. I tried to split my data on 60%/20%/20%; 70%/10%/20% and other configurations. I am trying multivariate (M) setting with default parameters and seq_len=96, label_len=48, pred_len=24. Thanks for help.