1.msl任务也会赋值domain_mask为None
2.validation 阶段给入的val_x有问题，predict报错；
best_model = model.fit(train_model_input, train[target].values, batch_size=train_batch_size, epochs=epochs, validation_data=(test_model_input, test[target].values))
改为：
best_model = model.fit(train_model_input, train[target].values, batch_size=train_batch_size, epochs=epochs, validation_split=0.)
3.未引入sklearn.metrics的roc_auc_score, log_loss
在main.py新增from sklearn.metrics import roc_auc_score, log_loss
1.msl任务也会赋值domain_mask为None
2.validation 阶段给入的val_x有问题，predict报错；
best_model = model.fit(train_model_input, train[target].values, batch_size=train_batch_size, epochs=epochs, validation_data=(test_model_input, test[target].values))
改为：
best_model = model.fit(train_model_input, train[target].values, batch_size=train_batch_size, epochs=epochs, validation_split=0.)
3.未引入sklearn.metrics的roc_auc_score, log_loss
在main.py新增from sklearn.metrics import roc_auc_score, log_loss