TensorFlow:恢复模型

时间:2017-07-30 18:22:38

标签: python machine-learning tensorflow deep-learning restore

我试图在结束时保存我的模型,并在每次训练开始时恢复它。我只是按照this link做了什么。

    saver = tf.train.Saver()
    with tf.Session(graph=graph) as session:
        # Initializate the weights and biases
        tf.global_variables_initializer().run()
        new_saver = tf.train.import_meta_graph('model.meta')
        new_saver.restore(sess,tf.train.latest_checkpoint('./'))
        W1 = session.run(W)
        print(W1)

        for curr_epoch in range(num_epochs):
            train_cost = train_ler = 0
            start = time.time()
            for batch in range(num_batches_per_epoch):
                ...Some training...

        W2 = session.run(W)
        print(W2)
        save_path = saver.save(session, "models/model")

但它给出了以下错误:

--->  new_saver.restore(session, tf.train.latest_checkpoint('./'))
SystemError: <built-in function TF_Run> returned a result with an error set

有人能帮帮我吗?非常感谢!

1 个答案:

答案 0 :(得分:0)

如果您要加载./,您必须确保您的控制台(用于启动python程序)实际上是在该目录(models /)上设置的。 但在这种情况下,它会将您的新数据保存在新目录中。所以用./models/加载

(您也不需要启动变量,恢复会为您执行此操作。)