M1 Max Tensorflow model.fit() InvalidArgumentError

Question

這是我在 M1 Mac 上的第一個 CNN 項目，當我運行 model.fit() 時出現 InvalidArgumentError。

我正在使用 tensorflow-metal 3.0、tensorflow-macos 2.7 並運行 MacOS 12.1。

early_stop = EarlyStopping(monitor="val_accuracy", patience=10)
optimizer = tf.keras.optimizers.Adam(learning_rate=lr, amsgrad=True)

model.compile(optimizer=optimizer, loss='categorical_crossentropy', metrics=['accuracy'])

train_datagen = ImageDataGenerator(rescale = 1./255,
                                   shear_range = 0.2,
                                   zoom_range = 0.2,
                                   horizontal_flip = True)

train_set = train_datagen.flow_from_directory(train_path,
                                              target_size = (224, 224),
                                              batch_size = batch_size,
                                              class_mode = 'categorical')                                                                        

res = model.fit(train_set,
                validation_data=test_set,
                batch_size=batch_size,
                epochs=epochs,
                steps_per_epoch=len(train_set),
                validation_steps=len(test_set),
                callbacks=[ckpt_saver, early_stop]
                )

Answer 1

我通過使用另一個優化器（不是亞當）暫時解決了這個問題。