Fastai causes pandas AttributeError: 'LazyConfigValue'

Part 1 (2019)
1

I think the current fastai build might be breaking pandas?

In tutorial 3, under the satellite imagery it causes the following error after df.head()

_AttributeError                            Traceback (most recent call last)_
_D:\Anaconda3\envs\fastai_v3\lib\site-packages\IPython\core\formatters.py in __call__(self, obj)_
_    343             method = get_real_method(obj, self.print_method)_
_    344             if method is not None:_
_--> 345                 return method()_
_    346             return None_
_    347         else:_

_D:\Anaconda3\envs\fastai_v3\lib\site-packages\pandas\core\frame.py in _repr_html_(self)_
_    647         # display HTML, so this check can be removed when support for_
_    648         # IPython 2.x is no longer needed._
_--> 649         if console.in_qtconsole():_
_    650             # 'HTML output is disabled in QtConsole'_
_    651             return None_

_D:\Anaconda3\envs\fastai_v3\lib\site-packages\pandas\io\formats\console.py in in_qtconsole()_
_    121             ip.config.get('KernelApp', {}).get('parent_appname', "") or_
_    122             ip.config.get('IPKernelApp', {}).get('parent_appname', ""))_
_--> 123         if 'qtconsole' in front_end.lower():_
_    124             return True_
_    125     except NameError:_

_AttributeError: 'LazyConfigValue' object has no attribute 'lower'_

This error does not exist if I restart the notebook, don’t import fastai and re-read the csv to df. Has anyone been able to fix this?

2

I encountered this AttributeError, too, but it wasn’t a breaking error. The df still loaded just fine and the remainder of the notebook ran smoothly. It has been reported as an issue on the pandas GitHub repo, so I don’t think it’s unique to fastai…FWIW.

3

Solution in this post.

2 Likes