l_map)
test_f1, test_acc = eval_batch(ner_model, test_dataset_loader, pack,
l_map)
print(
'(checkpoint: dev F1 = %.4f, dev acc = %.4f, F1 on test = %.4f, acc on test= %.4f)'
% (dev_f1, dev_acc, test_f1, test_acc))
tot_length = sum(map(lambda t: len(t), dataset_loader))
best_f1 = float('-inf')
best_acc = float('-inf')
track_list = list()
start_time = time.time()
epoch_list = range(args.start_epoch, args.start_epoch + args.epoch)
patience_count = 0
evaluator = eval_w(packer, l_map, args.eva_matrix)
for epoch_idx, args.start_epoch in enumerate(epoch_list):
epoch_loss = 0
ner_model.train()
for feature, tg, mask in tqdm(
itertools.chain.from_iterable(dataset_loader),
mininterval=2,
desc=' - Tot it %d (epoch %d)' %
(tot_length, args.start_epoch),
leave=False,
file=sys.stdout):
fea_v, tg_v, mask_v = packer.repack_vb(feature, tg, mask)
test_f1, test_acc = eval_batch(ner_model, test_dataset_loader, pack, l_map)
print('(checkpoint: dev F1 = %.4f, dev acc = %.4f, F1 on test = %.4f, acc on test= %.4f)' %
(dev_f1,
dev_acc,
test_f1,
test_acc))
tot_length = sum(map(lambda t: len(t), dataset_loader))
best_f1 = float('-inf')
best_acc = float('-inf')
track_list = list()
start_time = time.time()
epoch_list = range(args.start_epoch, args.start_epoch + args.epoch)
patience_count = 0
evaluator = eval_w(packer, l_map, args.eva_matrix)
for epoch_idx, args.start_epoch in enumerate(epoch_list):
epoch_loss = 0
ner_model.train()
for feature, tg, mask in tqdm(
itertools.chain.from_iterable(dataset_loader), mininterval=2,
desc=' - Tot it %d (epoch %d)' % (tot_length, args.start_epoch), leave=False, file=sys.stdout):
fea_v, tg_v, mask_v = packer.repack_vb(feature, tg, mask)
ner_model.zero_grad()
scores, hidden = ner_model.forward(fea_v)
loss = crit.forward(scores, tg_v, mask_v)
loss.backward()