Error reporting [project error reporting]

No module named ‘XXX’

reason ： No libraries installed
solve ：

pip install XXX

ModuleNotFoundError: No module named utils.util utils is not a package

explain ：utils It's a library written by myself , Import the modules in your library , Need to be in utils Create a new folder named __init__.py file

TensorBoard logging requires TensorBoard with Python summary writer installed. This should be available in 1.14 or above.
reason ： No installation tensorboard
solve ：

pip install tensorboard

No module named ‘past’

solve ： install future modular

pip install future

Remote file /home/chenkaixuan/projects/ is mapped to the local path D:\PycharmProjects\D-LSG-Video-Caption-main\ and can’t be found. You can continue debugging, but without the source.

• explain ： long-range debug When next step Then this interface appears , Then jump to a file several times data.py, I think there is a repetition when importing the Library （ Guess only ）
• solve ： Regardless of him , continue debug That's it

OSError: Unable to open file (truncated file: eof = 724698112, sblock->base_addr = 0, stored_eof = 15142273088)

• The background that ： After changing some network codes , Re execution train_debug.py There is such an error reading the file . Implemented before modification , There is no wrong report .
• Error reason ： I found it on the Internet , Roughly speaking , Downloaded files before , It didn't download successfully , Cause error in reading file . When I think of changing the network , There are files that are not automatically uploaded to the server , So I was silly and uploaded the project again , But the data is too large, which leads to slow uploading, and I think I can upload the changed files separately , So the upload was canceled in the middle , This may be the cause of file download errors .
• solve ： Upload the project or data again

UserWarning: This overload of nonzero is deprecated

• Error reason ： Version of the problem , Use torch.nonzero() It is better to add as_tuple=True/False Parameters . If True, Said to return to tuple,False return tensor. Generally speaking, we all need to return tensor Of .
• solve ：torch.nonzero(input, as_tuple = False)

127.0.0.1 Access denied

• The background that ：3 Use in the month PyCharm After the remote connection server runs, the code is opened tensorboard, It's normal . Today, I changed part of the code （ Mainly the network part ） All of a sudden 127.0.0.1 Access denied , And there was no response when opening the tunnel , Cannot be opened , Performance in Start/Stop The button below doesn't respond after clicking
  
• solve ：
  （1） Violence ： Close the tunnel , Build a new one . That is, when resetting the environment, it is found that the port number of the server needs to be filled in the configuration , I think the port number of the server has changed since the last tunnel was established , Lead to tensorboard Can't connect to the server .
  （2） Reset the connection settings .