TensorFlow错误

Question

我正在尝试运行这个示例(https://github.com/tensorflow/models/tree/master/textsum)。当我运行以下语句时：

bazel-bin/textsum/seq2seq_attention \
  --mode=train \
  --article_key=article \
  --abstract_key=abstract \
  --data_path=data/training-* \
  --vocab_path=data/vocab \
  --log_root=textsum/log_root \
  --train_dir=textsum/log_root/train

我看到以下输出

  File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/platform/app.py", line 30, in run
    sys.exit(main(sys.argv[:1] + flags_passthrough))

  File "/home/depiano/Scrivania/TextSummarization/bazel-bin/textsum/seq2seq_attention.runfiles/__main__/textsum/seq2seq_attention.py", line 196, in main
    _Train(model, batcher)

  File "/home/depiano/Scrivania/TextSummarization/bazel-bin/textsum/seq2seq_attention.runfiles/__main__/textsum/seq2seq_attention.py", line 98, in _Train
    allow_soft_placement=True))

  File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/training/supervisor.py", line 715, in prepare_or_wait_for_session
    init_feed_dict=self._init_feed_dict, init_fn=self._init_fn)

  File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/training/session_manager.py", line 227, in prepare_session
    config=config)

  File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/training/session_manager.py", line 173, in _restore_checkpoint
    saver.restore(sess, ckpt.model_checkpoint_path)

  File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/training/saver.py", line 1342, in restore
    "File path is: %r" % (save_path, file_path))

ValueError: Restore called with invalid save path: u'textsum/log_root/model.ckpt-0'. File path is: u'textsum/log_root/model.ckpt-0'
Errore di segmentazione (core dump creato)

Answer 1

该错误指示您的模型文件textsum/log_root/model.ckpt-0不存在或无法创建。确保目录textsum/log_root存在。

Answer 2

这里发生的是tensorflow试图将先前保存的参数加载到图中，module.They可以是体系结构细节和其他初始化内容。这就是tf测试是否存在预训练模块的方法

ckpt =tf.train.get_checkpoint_state(os.path.dirname('c/checkpointsq'))
    # if that checkpoint exists, restore from checkpoint
    if ckpt and ckpt.model_checkpoint_path:
        print("Restoring the checkpoins")
        saver.restore(sess, ckpt.model_checkpoint_path)   

在这里，第一个函数将返回一个ckpt文件对象，如果有任何名为checkpoint的文件。这是文本文件。在它里面会有你的网络的细节。因此，在您的情况下，您必须更正日志目录。