qq_36857472的博客

原创 Early stopping conditioned on metric `val_loss` which is not available

Early stopping conditioned on metric val_loss which is not available这个错误是keras在执行keras.callbacks.EarlyStopping(monitor=‘loss’,min_delta=1e-3,patience=2),时由于model.fit参数中未设置validation_split=0.1（验证集）导致的。...