p = Preprocessor()
model1 = TextClassifier(len(config.class_group[0]), 'lstm')
model2 = TextClassifier(len(config.class_group[1]), 'lstm')
model3 = TextClassifier(len(config.class_group[2]), 'lstm')
model4 = TextClassifier(len(config.class_group[3]), 'lstm')
model5 = TextClassifier(len(config.class_group[4]), 'lstm')
model6 = TextClassifier(len(config.class_group[5]), 'lstm')
F1 = []
for i in range(30):
# 训练模型
train_seq1, train_label1 = p.shuffle(Train_seq_set[0], Train_label_set[0])
model1.train(train_seq1,
[train_label1[l] for l in range(len(config.class_group[0]))],
Validation_seq, [
Validation_label[config.class_group[0][i]]
for i in range(len(config.class_group[0]))
])
train_seq2, train_label2 = p.shuffle(Train_seq_set[1], Train_label_set[1])
model2.train(train_seq2,
[train_label2[l] for l in range(len(config.class_group[1]))],
Validation_seq, [
Validation_label[config.class_group[1][i]]
for i in range(len(config.class_group[1]))
])
train_seq3, train_label3 = p.shuffle(Train_seq_set[2], Train_label_set[2])
model1.train(train_seq3,
[train_label3[l] for l in range(len(config.class_group[2]))],
Validation_seq, [
batchListTrain = utils.buildBatchList(len(corpus.trainData), batchSize)
batchListDev = utils.buildBatchList(len(corpus.devData), batchSize)
while epoch < maxEpoch:
aveLoss = 0.0
trainAcc = 0.0
epoch += 1
print('--- Epoch ' + str(epoch))
random.shuffle(corpus.trainData)
opt = optim.SGD(classifier.parameters(),
lr=initialLearningRate / (1.0 + lrDecay * epoch),
weight_decay=weightDecay)
classifier.train()
for batch in batchListTrain:
opt.zero_grad()
# build input for the batch
curBatchSize = batch[1] - batch[0] + 1
batchInput, batchTarget, lengths = utils.buildBatchInOutForClassifier(
corpus.voc.getTokenIndex(corpus.voc.PAD), batch, corpus.trainData)
target = Variable(batchTarget)
if biDirectional:
shape = 2, curBatchSize, hiddenDim
else:
shape = 1, curBatchSize, hiddenDim
h0 = c0 = Variable(torch.zeros(*shape), requires_grad=False)