Question on documentation

[enikidis]

Hello ! I managed succesfully to install allegro and I am currently running a job on a gpu. Is there any place I can get a general idea on the output and what they actually mean.?

Train # Epoch wal LR loss_f loss_e loss f_mae f_rmse e_mae e_rmse
! Train 24 32637.932 0.001 0.00142 0.00137 0.0156 0.0319 0.0596 0.752 1.26
! Validation 24 32637.932 0.001 0.00128 0.00114 0.0139 0.0283 0.0583 0.695 1.2
Wall time: 32637.932265501004
! Best model 24 0.014

As I understand this 0.014 is important and needs to decrease with every best model found.
How should someone start and understanding this and not use it as a "blackbox" approach?

[Linux-cpp-lisp]

Best model is reporting whichever metric was set as metrics_key; from full.yaml where all trainer options are documented:

metrics_key: validation_loss                                                       # metrics used for scheduling and saving best model. Options: `set`_`quantity`, set can be either "train" or "validation, "quantity" can be loss or anything that appears in the validation batch step header, such as f_mae, f_rmse, e_mae, e_rmse

Default is validation_loss.