core.BaseTrainer¶

Module Contents¶

class core.BaseTrainer.BaseTrainer(model, dataloaders, optimizer, scheduler=None, criterion=None, num_epochs=100, device='cuda:0', no_valid=False, valid_frequency=1, amp=False, algo_params=None)[source]¶

Abstract base class for trainer implementations

model[source]¶

dataloaders[source]¶

optimizer[source]¶

scheduler = None[source]¶

criterion = None[source]¶

num_epochs = 100[source]¶

no_valid = False[source]¶

valid_frequency = 1[source]¶

device = 'cuda:0'[source]¶

amp = False[source]¶

best_weights = None[source]¶

best_f1_score = 0.1[source]¶

scaler = None[source]¶

loss_metric[source]¶

f1_metric[source]¶

post_pred[source]¶

post_gt[source]¶

train()[source]¶: Train the model