Open cloner174 opened 1 month ago
Assume you initialize the model with name 'exp'. After training , you can use these: exp.trues_during_training exp.preds_during_training exp.train_losses exp.test_losses
Just let me know if you faced with any error. or if you have any questions .
fixing axis=1 in the part with dropping 'date' column with pandas . it kept gave error duo to not specified the param 'axis' for 1. -> in Custom dataset. adding MinMaxScaler, and can be access via: kind_of_scaler = 'MinMax' adding name_of_col_with_date which is very useful since the title of the date column may not be exactly the 'date'! in some cases. kind_of_optim let you choose between 'AdamW', 'SparseAdam', 'SGD', 'RMSprop', 'RAdam', 'NAdam' ,'LBFGS','Adamax' 'ASGD' 'Adadelta' 'Adagrad' , for optimizing and of course the default is Adam! test_size let you decide the size of test! which force the train and validate data to obey! train_size = 0.9 - test_size default is 0.2 let the arg.criter be 'default' (which is MSE) as it is provided in run.py. I am not sure about that. But for now it also be 'MAE'