search

chilynn / sequence-labeling

307 stars 167 forks source link

关于y_train_weight_batch #22

Open chan4cc opened 7 years ago

chan4cc commented 7 years ago 
for iteration in range(num_iterations):
                # train
                X_train_batch, y_train_batch = helper.nextBatch(X_train, y_train, start_index=iteration * self.batch_size, batch_size=self.batch_size)
                y_train_weight_batch = 1 + np.array((y_train_batch == label2id['B']) | (y_train_batch == label2id['E']), float)
                transition_batch = helper.getTransition(y_train_batch)

你好，BILSTM_CRF.py文件中第173行的这个y_train_weight_batch的含义是什么？