ValueError: 'a' cannot be empty unless no samples are taken

[PotatoGan]

plz help me, error came out when running "python -m scripts.train_diffusion ../config/training/diff_pos0_10_pos1.e-7_0.01_6_v001_bondTrue_scalar128_vec32_layer8.yml --logdir modelsave"
/root/miniconda3/envs/diffsmol/lib/python3.9/site-packages/oddt/surface.py:21: UserWarning: scikit-image could not be imported and is required forgenerating molecular surfaces.
warnings.warn('scikit-image could not be imported and is required for'
[2025-06-04 15:41:02,253::train::INFO] Namespace(config='../config/training/diff_pos0_10_pos1.e-7_0.01_6_v001_bondTrue_scalar128_vec32_layer8.yml', device='cuda', logdir='modelsave', change_log_dir=None, tag='', continue_train_iter=-1, train_report_iter=200)
[2025-06-04 15:41:02,253::train::INFO] {'data': {'name': 'shapemol', 'dataset': 'moses2', 'version': 'dgcnn_signeddist_512', 'datasize': 300, 'chunk_size': 50000, 'num_workers': 20, 'processed_path': '../data/MOSES2/', 'path': '../data/MOSES2/MOSES2_training_val_dataset.pkl', 'split': '../data/MOSES2/MOSES2_training_val_dataset_split.pt', 'transform': {'ligand_atom_mode': 'add_aromatic', 'random_rot': False}, 'shape': {'use_shape': True, 'shape_type': 'pointAE_shape', 'point_cloud_samples': 512, 'shape_parallel': False, 'num_workers': 1, 'batch_size': 8, 'checkpoint': '../models/se.pt'}}, 'model': {'denoise_type': 'diffusion', 'model_mean_type': 'C0', 'gt_noise_type': 'origin', 'schedule_pos': {'beta_schedule': 'sigmoid', 'beta_start': 1e-07, 'beta_end': 0.01, 's': 6}, 'schedule_v': {'beta_schedule': 'cosine', 's': 0.01}, 'num_diffusion_timesteps': 1000, 'loss_v_weight': 200.0, 'v_mode': 'uniform', 'v_net_type': 'mlp', 'pred_bond_type': True, 'loss_bond_weight': 0.1, 'loss_pos_type': 'mse', 'use_bond_dist_loss': True, 'loss_bond_dist_weight': 0.1, 'use_bond_angle_loss': False, 'loss_bond_angle_weight': 0.1, 'use_torsion_angle_loss': False, 'loss_torsion_angle_weight': 0.01, 'loss_torsion_angle_type': 'one', 'sample_time_method': 'symmetric', 'loss_weight_type': 'noise_level', 'loss_pos_min_weight': 0, 'loss_pos_max_weight': 10, 'time_emb_dim': 8, 'time_emb_mode': 'simple', 'center_pos_mode': 'none', 'atom_enc_mode': 'add_aromatic', 'node_indicator': True, 'model_type': 'uni_o2', 'num_blocks': 1, 'num_layers': 8, 'scalar_hidden_dim': 128, 'vec_hidden_dim': 32, 'n_heads': 16, 'edge_feat_dim': 5, 'edge_feat': 'covalent_bond', 'num_r_gaussian': 20, 'knn': 8, 'num_node_types': 8, 'act_fn': 'relu', 'norm': True, 'cutoff_mode': 'knn', 'r_feat_mode': 'sparse', 'energy_h_mode': 'basic', 'r_max': 10.0, 'x2h_out_fc': False, 'sync_twoup': False, 'shape_dim': 128, 'shape_latent_dim': 128, 'shape_mode': 'None', 'shape_type': 'pointAE_shape', 'cond_mask_prob': 0.1, 'use_shape_vec_mul': False, 'use_residue': True}, 'train': {'seed': 2023, 'batch_size': 32, 'num_workers': 10, 'max_iters': 10000000, 'val_freq': 2000, 'pos_noise_std': 0.1, 'max_grad_norm': 8.0, 'bond_loss_weight': 1.0, 'optimizer': {'type': 'adam', 'lr': 0.001, 'weight_decay': 0, 'beta1': 0.95, 'beta2': 0.999}, 'scheduler': {'type': 'plateau', 'factor': 0.6, 'patience': 10, 'min_lr': '1e-05'}, 'n_acc_batch': 1}}
[2025-06-04 15:41:02,259::train::INFO] Loading dataset...
Traceback (most recent call last):
File "/root/miniconda3/envs/diffsmol/lib/python3.9/runpy.py", line 197, in _run_module_as_main
return _run_code(code, main_globals, None,
File "/root/miniconda3/envs/diffsmol/lib/python3.9/runpy.py", line 87, in _run_code
exec(code, run_globals)
File "/root/DiffSMol/source/scripts/train_diffusion.py", line 244, in
dataset, subsets = get_dataset(
File "/root/DiffSMol/source/datasets/init.py", line 39, in get_dataset
random_valid_indices = np.random.choice(v, 1000).tolist()
File "mtrand.pyx", line 915, in numpy.random.mtrand.RandomState.choice
ValueError: 'a' cannot be empty unless no samples are taken
(diffsmol) root@DESKTOP-QME9DA7:~/DiffSMol/source# python -m scripts.train_diffusion ../config/training/diff_pos0_10_pos1.e-7_0.01_6_v001_bondTrue_scalar128_vec32_layer8.yml --logdir modelsave
/root/miniconda3/envs/diffsmol/lib/python3.9/site-packages/oddt/surface.py:21: UserWarning: scikit-image could not be imported and is required forgenerating molecular surfaces.
warnings.warn('scikit-image could not be imported and is required for'
[2025-06-05 10:06:40,845::train::INFO] Namespace(config='../config/training/diff_pos0_10_pos1.e-7_0.01_6_v001_bondTrue_scalar128_vec32_layer8.yml', device='cuda', logdir='modelsave', change_log_dir=None, tag='', continue_train_iter=-1, train_report_iter=200)
[2025-06-05 10:06:40,845::train::INFO] {'data': {'name': 'shapemol', 'dataset': 'moses2', 'version': 'dgcnn_signeddist_512', 'datasize': 300, 'chunk_size': 50000, 'num_workers': 20, 'processed_path': '../data/MOSES2/', 'path': '../data/MOSES2/MOSES2_training_val_dataset.pkl', 'split': '../data/MOSES2/MOSES2_training_val_dataset_split.pt', 'transform': {'ligand_atom_mode': 'add_aromatic', 'random_rot': False}, 'shape': {'use_shape': True, 'shape_type': 'pointAE_shape', 'point_cloud_samples': 512, 'shape_parallel': False, 'num_workers': 1, 'batch_size': 8, 'checkpoint': '../models/se.pt'}}, 'model': {'denoise_type': 'diffusion', 'model_mean_type': 'C0', 'gt_noise_type': 'origin', 'schedule_pos': {'beta_schedule': 'sigmoid', 'beta_start': 1e-07, 'beta_end': 0.01, 's': 6}, 'schedule_v': {'beta_schedule': 'cosine', 's': 0.01}, 'num_diffusion_timesteps': 1000, 'loss_v_weight': 200.0, 'v_mode': 'uniform', 'v_net_type': 'mlp', 'pred_bond_type': True, 'loss_bond_weight': 0.1, 'loss_pos_type': 'mse', 'use_bond_dist_loss': True, 'loss_bond_dist_weight': 0.1, 'use_bond_angle_loss': False, 'loss_bond_angle_weight': 0.1, 'use_torsion_angle_loss': False, 'loss_torsion_angle_weight': 0.01, 'loss_torsion_angle_type': 'one', 'sample_time_method': 'symmetric', 'loss_weight_type': 'noise_level', 'loss_pos_min_weight': 0, 'loss_pos_max_weight': 10, 'time_emb_dim': 8, 'time_emb_mode': 'simple', 'center_pos_mode': 'none', 'atom_enc_mode': 'add_aromatic', 'node_indicator': True, 'model_type': 'uni_o2', 'num_blocks': 1, 'num_layers': 8, 'scalar_hidden_dim': 128, 'vec_hidden_dim': 32, 'n_heads': 16, 'edge_feat_dim': 5, 'edge_feat': 'covalent_bond', 'num_r_gaussian': 20, 'knn': 8, 'num_node_types': 8, 'act_fn': 'relu', 'norm': True, 'cutoff_mode': 'knn', 'r_feat_mode': 'sparse', 'energy_h_mode': 'basic', 'r_max': 10.0, 'x2h_out_fc': False, 'sync_twoup': False, 'shape_dim': 128, 'shape_latent_dim': 128, 'shape_mode': 'None', 'shape_type': 'pointAE_shape', 'cond_mask_prob': 0.1, 'use_shape_vec_mul': False, 'use_residue': True}, 'train': {'seed': 2023, 'batch_size': 32, 'num_workers': 10, 'max_iters': 10000000, 'val_freq': 2000, 'pos_noise_std': 0.1, 'max_grad_norm': 8.0, 'bond_loss_weight': 1.0, 'optimizer': {'type': 'adam', 'lr': 0.001, 'weight_decay': 0, 'beta1': 0.95, 'beta2': 0.999}, 'scheduler': {'type': 'plateau', 'factor': 0.6, 'patience': 10, 'min_lr': '1e-05'}, 'n_acc_batch': 1}}
[2025-06-05 10:06:40,847::train::INFO] Loading dataset...
Traceback (most recent call last):
File "/root/miniconda3/envs/diffsmol/lib/python3.9/runpy.py", line 197, in _run_module_as_main
return _run_code(code, main_globals, None,
File "/root/miniconda3/envs/diffsmol/lib/python3.9/runpy.py", line 87, in _run_code
exec(code, run_globals)
File "/root/DiffSMol/source/scripts/train_diffusion.py", line 244, in
dataset, subsets = get_dataset(
File "/root/DiffSMol/source/datasets/init.py", line 39, in get_dataset
random_valid_indices = np.random.choice(v, 1000).tolist()
File "mtrand.pyx", line 915, in numpy.random.mtrand.RandomState.choice
ValueError: 'a' cannot be empty unless no samples are taken

[ziqi92]

Hi,

Thank you for your interest in our work!

Based on the provided log, I think the dataset is not correctly loaded. can you double-check whether you have downloaded our processed training data in the link https://zenodo.org/records/14855728/files/MOSES2_training_val_dataset.pkl?download=1 and the downloaded path matches with the one specified in the config

DiffSMol/config/training/diff_pos0_10_pos1.e-7_0.01_6_v001_bondTrue_scalar128_vec32_layer8.yml

Line 9 in 2ccade3

path: ../data/MOSES2/MOSES2_training_val_dataset.pkl

? Please let us know if you have any other issues.