Training Classifier Problem

[drorhunvural]

You are giving code block below as an example for Training Classifier

python train.py -m model.py --train_types scPDB_train0.types --test_types scPDB_test0.types -i 200000 --train_recmolcache scPDB_new.molcache2 --test_recmolcache scPDB_new.molcache2 -r val0 -o /model_saves/val9 --base_lr 0.001 --solver Adam

You are using --data_dir (-d) in train.py as below

 eptrain = molgrid.ExampleProvider(shuffle=True, stratify_receptor=True, labelpos=0,balanced=True,
                                      data_root=args.data_dir,recmolcache=args.train_recmolcache)

Where are you reading data_dir from? The training classifier example you shared does not have data_dir in the code block ? '

[RishalAggarwal]

The data_dir argument is ignored (despite being passed) for the classification model as all the required data is present in the molcache files. It becomes relevant for the segmentation models