Specific file for validation set

[ninpnin]

Currently, the validation set is randomly selected from the train set. We want to be able to use a specific CSV in the validation process.

python3 train_binary_bert.py --data_path traindata.csv --val_data_path valdata.csv

Old functionality should be retained as an option, so if val_data_path is not provided, the validation set should be randomly selected from whatever is loaded from data_data_path.