error occured when re_training graph classifier

[callmewanzi]

Hi, thanks for your code.🙂
when I try to train the node classifier through the instruction
python3 ./molgrapher/scripts/train/train_graph_classifier.py
the below error accurred
Original Traceback (most recent call last): ... File "MolGrapher/molgrapher/datasets/dataset_graph.py", line 162, in __getitem__ molecular_graph = self.get_molecular_graph(image, keypoints, molecule) File "MolGrapher/molgrapher/datasets/dataset_graph.py", line 46, in get_molecular_gr aph molecular_graph = MolecularGraph(self.config).from_rdkit_molecule(molecule, keypoints) File "MolGrapher/molgrapher/utils/utils_graph.py", line 153, in from_rdkit_molecule "class": self.types_classes_bonds[bond_type], KeyError: 'SOLID'

The error will be disappeared when i replace "nb_bonds_classes": 6 with "nb_bonds_classes": 8 in MolGrapher/data/config_dataset_graph_2.json
But i don't know whether it's correct to replace.

Besides, there seems to be some problem with tensorboard files, when visualizing the file, the number of epoch is not changed with the number of training step. (like below)

[lucas-morin]

Hello @callmewanzi,

Thanks for your interest in MolGrapher.
If I am not mistaken, I think you can train with both 6 or 8 bond classes, but it needs to be consistent with the configuration you used to generate the training dataset.

I only used the number of training steps but I think you could log self.trainer.current_epoch (as used here) to get the number of epoch.