首页 文章

TFLearn / Tensorflow:保存从自动编码器中提取的编码器的正确方法

提问于
浏览
0

这个问题最初发布在tflearn github repo上,但我没有运气:https://github.com/tflearn/tflearn/issues/682

我正在尝试从自动编码器中保存代表中间层的编码器模型 . 使用MNIST示例,当我运行此处的脚本时:

https://github.com/tflearn/tflearn/blob/master/examples/images/autoencoder.py

然后尝试使用保存encoding_model

encoding_model = tflearn.DNN(encoder, session=model.session)
encoding_model.save('encoder.tfl')

我收到以下错误消息:

回溯(最近一次调用最后一次):文件“”,第1行,文件“/usr/local/lib/python2.7/dist-packages/tflearn/models/dnn.py”,第260行,保存自我 . trainer.save(model_file)文件“/usr/local/lib/python2.7/dist-packages/tflearn/helpers/trainer.py”,第376行,保存self.saver.save(self.session,model_file,global_step = global_step)文件“/usr/local/lib/python2.7/dist-packages/tensorflow/python/training/saver.py”,第1363行,保存{self.saver_def.filename_tensor_name:checkpoint_file})文件“/ usr /local/lib/python2.7/dist-packages/tensorflow/python/client/session.py“,第767行,运行run_metadata_ptr)文件”/usr/local/lib/python2.7/dist-packages/tensorflow/ python / client / session.py“,第965行,在_run feed_dict_string,options,run_metadata中)文件”/usr/local/lib/python2.7/dist-packages/tensorflow/python/client/session.py“,第1015行,在_do_run target_list,options,run_metadata)文件“/usr/local/lib/python2.7/dist-packages/tensorflow/python/client/session.py”,第1035行,在_do_call ra中ise type(e)(node_def,op,message)tensorflow.python.framework.errors_impl.FailedPreconditionError:尝试使用未初始化的值Global_Step_1 [[Node:Global_Step_1 / _96 = SendT = DT_FLOAT,client_terminated = false,recv_device =“/ job: localhost / replica:0 / task:0 / cpu:0“,send_device =”/ job:localhost / replica:0 / task:0 / gpu:0“,send_device_incarnation = 1,tensor_name =”edge_31_Global_Step_1“, device =”/工作:本地主机/副本:0 /任务:0 / GPU:0" ]]

我认为ADAM优化器变量没有初始化 . 保存这样的模型的正确方法是什么?

1 回答

相关问题