尝试在 tf.keras 上重命名预训练模型时出错

Posted 2023-02-23

【中文标题】尝试在 tf.keras 上重命名预训练模型时出错

【英文标题】：Error when trying to rename a pretrained model on tf.keras

【发布时间】：2019-11-15 02:34:22

【问题描述】：

我训练了两个模型来集成它们， 当我尝试使用此代码加载它们时：

  from tensorflow.keras.models import load_model
  models=[]
  modelTemp=load_model('models/full.h5')
  modelTemp.name = "inception1"
  models.append(modelTemp)

发生错误：

  AttributeError: Can't set the attribute "name", likely because it conflicts with an existing read-only @property of the object. Please choose a different name.

完整的错误信息：

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/base_layer.py in __setattr__(self, name, value)
   1968       try:
-> 1969         super(tracking.AutoTrackable, self).__setattr__(name, value)
   1970       except AttributeError:

AttributeError: can't set attribute

During handling of the above exception, another exception occurred:

AttributeError                            Traceback (most recent call last)
2 frames
/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/base_layer.py in __setattr__(self, name, value)
   1972             ('Can\'t set the attribute "", likely because it conflicts with '
   1973              'an existing read-only @property of the object. Please choose a '
-> 1974              'different name.').format(name))
   1975       return
   1976 

AttributeError: Can't set the attribute "name", likely because it conflicts with an existing read-only @property of the object. Please choose a different name.

【问题讨论】：

你试过其他名字吗？

是的，我试过了，但没有出现同样的错误

Keras rename model and layers的可能重复

【参考方案1】：

根据this question here on *** 你需要使用：

modelTemp._name = 'inception'

【讨论】：

这应该是公认的答案。显然他们没有包含名称设置器装饰器，必须手动更改内部属性

这是正确的方法吗？弄乱“_name”似乎很危险。也许 tensorflow 团队这样做是有原因的。

【参考方案2】：

我在这里回答了同样的话题How to rename Pre-Trained model ? ValueError 'Trained Model' is not a valid scope name

解决办法是：

model = load_model(r"C:\Master\Learning\Agri_Intelligence\Models\Model.h5")
model._name = "New_Model_Name"
model.save(r"C:\Master\Learning\Agri_Intelligence\Models\New_Model.h5")

【讨论】：

你自己的问题（你链接到这里）和这个问题一样吗？如果是这样，那么在这里添加您的详细答案而不是发布类似的问题会更合适。

@FluffyKitten 是的，没错！