python_ERROR日常出坑指南(SIMAIR背景下）

1、ModuleNotFoundError: No module named ‘tornado.platform.auto’

----> 3 from tornado.platform.auto import set_close_exec

解决：我的tornado版本为6.1，在官网https://www.tornadoweb.org/en/stable/releases了解到6.1版本中的set_close_exec已被移除，所以要对tornado版本降级。根据官网介绍，选择了tornado6.0解决了该问题。

2、AttributeError: ‘str’ object has no attribute ‘decode’

D:\anoconda\envs\simair\lib\site-packages\keras\models.py in load_model(filepath, custom_objects, compile)
    237         if model_config is None:
    238             raise ValueError('No model found in config file.')
--> 239         model_config = json.loads(model_config.decode('utf-8'))
    240         model = model_from_config(model_config, custom_objects=custom_objects)
    241 

AttributeError: 'str' object has no attribute 'decode'

解决：
我的h5py版本为3.2.0，更新为2.10.0

pip install h5py=2.10.0

3、TypeError: unsupported operand type(s) for *: ‘ZMQIOLoop’ and ‘float’

此错误与tornado版本有关，查到应该用tornado==4.5.3，所以pip install tornado==4.5.3
安装之后发现jupyter notebook会闪退，原因是jupyter notebook和tornado版本对应，如下图：

所以接着重新安装notebook,这里我选择了5.7.8，pip install notebook==5.7.8,安装完之后快捷方式jupyter notebook还是闪退，不过可以在prompt中得对应环境下输入 jupyter notebook来打开，代码可以运行了。

4、UnknownError (see above for traceback): Failed to get convolution algorithm. This is probably because cuDNN failed to initialize, so try looking to see if a warning log message was printed above.

在命令行显示器下显示如下错误：

Could not create cudnn handle: CUDNN_STATUS_ALLOC_FAILED

##方法一
config = tf.ConfigProto()
      config.gpu_options.allow_growth = True
      session = tf.Session(config=config)
##方法二     
      gpu_options = tf.GPUOptions(per_process_gpu_memory_fraction=args.gpu_memory_fraction,allow_growth=True)
      sess = tf.Session(config=tf.ConfigProto(gpu_options=gpu_options, log_device_placement=False))
      
##方法三：
把config.gpu_options.per_process_gpu_memory_fraction = 1
改成config.gpu_options.per_process_gpu_memory_fraction = 0.4