为什么tf.keras.models.load_model再次构建模型？

Question

ValueError: Sequential model 'sequential_2' has already been configured to use 
input shape (None, 224, 224, 3). You cannot build it with input_shape [None, 224, 224, 3]

def load_model(model_path):
  """
  Loads a saved model from a specified path.
  """

  tf.keras.config.enable_unsafe_deserialization()
  print(f"Loading saved model from: {model_path}")
  model = tf.keras.models.load_model(model_path)
  return model

loaded_1000_image_model = load_model('/content/drive/MyDrive/Dog Vision/models/20241002-16491727887796-1000-images-mobilenetv2-Adam.h5')

我期望它会加载模型而不会出现任何错误，但由于某种原因，即使我没有尝试重建或再次给出任何输入形状，它也会给出值错误。

Answer 1

模型在保存之前是否已编译？如果没有，您应该在保存之前尝试编译。您还可以尝试将编译设置为 false，并查看编译步骤是否出现问题。您还可以通过设置 safe_mode=False 来跳过“tf.keras.config.enable_unsafe_deserialization()”。

tf.keras.models.load_model(
    filepath, custom_objects=None, compile=True, safe_mode=True
)