由于numpy问题，无法从控制台启动张量板

Question

我想使用TensorBoard。我已经使用tensorflow.keras.callbacks库将训练和验证数据保存在与项目数据的其余部分相同的目录中。

我已经重新安装了numpy，没有成功。然后我删除了tb-nightly和tensorboard，因为我有他们两个由于某种原因，只是重新安装tb-nightly。

(Tensorflow) C:\Users\alias>python -c "print(__import__('numpy').__version__)"
1.16.2

(Tensorflow) C:\Users\alias>python -c "print(__import__('tensorflow').__version__)"
2.0.0-alpha0

(Tensorflow) C:\Users\alias>python -c "print(__import__('tensorboard.version').version.VERSION)"
1.14.0a20190301

在Win10-OS上通过Anaconda提示调用时，会发生以下情况：

(Tensorflow) C:\Users\alias>tensorboard --logdir=logs\
TensorBoard 1.14.0a20190301 at http://LAPTOP-4E1BJCAV:6006 (Press CTRL+C to quit)
Traceback (most recent call last):
  File "c:\users\alias\anaconda3\envs\tensorflow\lib\runpy.py", line 193, in _run_module_as_main
    "__main__", mod_spec)
  File "c:\users\alias\anaconda3\envs\tensorflow\lib\runpy.py", line 85, in _run_code
    exec(code, run_globals)
  File "C:\Users\alias\Anaconda3\envs\Tensorflow\Scripts\tensorboard.exe\__main__.py", line 9, in <module>
  File "c:\users\alias\anaconda3\envs\tensorflow\lib\site-packages\tensorboard\main.py", line 58, in run_main
    app.run(tensorboard.main, flags_parser=tensorboard.configure)
  File "c:\users\alias\anaconda3\envs\tensorflow\lib\site-packages\absl\app.py", line 300, in run
    _run_main(main, args)
  File "c:\users\alias\anaconda3\envs\tensorflow\lib\site-packages\absl\app.py", line 251, in _run_main
    sys.exit(main(argv))
  File "c:\users\alias\anaconda3\envs\tensorflow\lib\site-packages\tensorboard\program.py", line 228, in main
    self._register_info(server)
  File "c:\users\alias\anaconda3\envs\tensorflow\lib\site-packages\tensorboard\program.py", line 274, in _register_info
    manager.write_info_file(info)
  File "c:\users\alias\anaconda3\envs\tensorflow\lib\site-packages\tensorboard\manager.py", line 269, in write_info_file
    payload = "%s\n" % _info_to_string(tensorboard_info)
  File "c:\users\alias\anaconda3\envs\tensorflow\lib\site-packages\tensorboard\manager.py", line 129, in _info_to_string
    for k in _TENSORBOARD_INFO_FIELDS
  File "c:\users\alias\anaconda3\envs\tensorflow\lib\site-packages\tensorboard\manager.py", line 129, in <dictcomp>
    for k in _TENSORBOARD_INFO_FIELDS
  File "c:\users\alias\anaconda3\envs\tensorflow\lib\site-packages\tensorboard\manager.py", line 51, in <lambda>
    (dt - datetime.datetime.fromtimestamp(0)).total_seconds()),
OSError: [Errno 22] Invalid argument

一些帮助会很高兴。

提前致谢！

Answer 1

您最近是否进行了全新安装或更新？

不确定这是否适用于您的情况？昨天我从头开始为python 2.7安装了最新的conda，并在安装后更新到最新的软件包。从pycharm或windows powershell运行python并导入numpy将导致多阵列导入错误。对我来说，解决方法是将numpy降级为？1.16.12？到？1.5.14？对不起，但是我离开我的电脑是因为我遇到了错误，但我认为那些版本是正确的。

Answer 2

我在当地的Jupyter笔记本中遇到了同样的错误。升级Numpy包对我有用。尝试升级Numpy如下

pip install numpy==1.16

如果上述操作无效，请使用以下命令升级tensorflow，然后重试

pip install tensorflow --upgrade

Answer 3

问题解决了@ Github https://github.com/tensorflow/tensorboard/issues/2092

感谢您的支持