chensvm/keras-callback-f1

自定义Metrics，并保存在验证集上该性能最好的模型

keras-callback-f1's Issues

there is an AttributeError in F1_Metrics.py

class F1_Metrics(Callback):
    def on_train_begin(self, logs={}):
        if not ('f1' in self.params['metrics']):
            self.params['metrics'].append('f1')
    def on_epoch_end(self, epoch, logs={}):
        logs['f1'] = float('-inf')
        val_predict = (np.asarray(self.model.predict(self.model.validation_data[0]))).round()
        val_targ = self.validation_data[1]
        if (self.validation_data):
            logs['f1'] = f1_score(val_targ, val_predict, average='macro')

Error
AttributeError: "Model" object has no attribute 'validation_data'

That I have added 'validation_data' to model.fit() and also try to use 'validation_split' to replace 'validation_data' has reported that error.