有点晚了，但我希望它仍然有帮助，
从我的研究中，我看到Kera 2.0中删除了F1分数指标，因为分数是分批近似的。我认为即使使用Tensorflow Addon，您也会面临这个问题。
欲了解更多详情，请访问：https://github.com/keras-team/keras/issues/5794#issuecomment-287641301
如果你想要F1分数，你可以在我实现的epoch结束时计算它：

model.compile(optimizer=tf.keras.optimizers.Adam(),
              loss=tf.keras.losses.BinaryCrossentropy(from_logits=False),
              metrics=[
                tf.keras.metrics.BinaryAccuracy(),
                tf.keras.metrics.Precision(),
                tf.keras.metrics.Recall(),
              ])

class MetricsF1(tf.keras.callbacks.Callback):
    def on_epoch_end(self, epoch, logs=None):
        precision = logs['precision']
        recall = logs['recall']
        f1_score = 2 * (precision * recall) / (precision + recall)
        print("F1 SCORE TRAIN:", round(f1_score, 4))

        val_precision = logs['val_precision']
        val_recall = logs['val_recall']
        val_f1_score = 2 * (val_precision * val_recall) / (val_precision + val_recall)
        print("F1 SCORE VALIDATION:", round(val_f1_score, 4))

model.fit(train_dataset,
          epochs=10,
          callbacks=[
            MetricsF1()
          ],
          validation_data=val_dataset
)

字符串