为什么keras嵌入层没有密钥

时间:2018-10-09 14:18:20

标签: keras embedding

self.embed =顺序([Embedding(9488,output_dim = 512,input_length = 14),
   Activation('relu'),
  辍学(0.5)],名称='embed.0')

inputs_bedding = Input(shape =(10,))
xt = self.embed(inputs_bedding)

============================
keras_model.save('temp.h5')
使用h5py.File('temp.h5','a')为f:
    model_weights = f ['model_weights']
params = util.dig_to_params(model_weights [layer])

def dig_to_params(keras_h5_layer):
    而不是_contains_weights(keras_h5_layer):
        keras_h5_layer = keras_h5_layer [list(keras_h5_layer.keys())[0]]

return keras_h5_layer <br>

错误: keras_h5_layer = keras_h5_layer [list(keras_h5_layer.keys())[0]]
AttributeError:“数据集”对象没有属性“键”

0 个答案:

没有答案