algorithms/deepCORAL.py
- 1 F401: 'torch_geometric.nn.global_mean_pool' imported but unused
-
7
E501: line too long (90 > 79 characters)
(and 5 similar)
-
Line too long (90 > 79 characters):
11 This algorithm was originally proposed as an unsupervised domain adaptation algorithm. -
Line too long (80 > 79 characters):
21 The CORAL penalty function below is adapted from DomainBed's implementation: -
Line too long (90 > 79 characters):
31 featurizer, classifier = initialize_model(config, d_out=d_out, is_featurizer=True) -
Line too long (91 > 79 characters):
54 # featurizers output Tensors of size (batch_size, ..., feature dimensionality). -
Line too long (117 > 79 characters):
105 penalty += self.coral_penalty(features[group_indices[i_group]], features[group_indices[j_group]]) -
Line too long (102 > 79 characters):
107 penalty /= (n_groups_per_batch * (n_groups_per_batch - 1) / 2) # get the mean penalty -
Line too long (93 > 79 characters):
117 avg_loss = self.loss.compute(results['y_pred'], results['y_true'], return_dict=False)
-
- 1 E302: expected 2 blank lines, found 1