Closed tentrillion closed 7 years ago
Thanks for all the info! Never would've known there was a problem without you.
The fix I made was to add dots (relative paths) in __init__.py
like below.
https://github.com/bstriner/keras-tqdm/blob/master/keras_tqdm/__init__.py
from .tqdm_callback import TQDMCallback
from .tqdm_notebook_callback import TQDMNotebookCallback
Whereas this is the broken version:
from tqdm_callback import TQDMCallback
from tqdm_notebook_callback import TQDMNotebookCallback
The code in your stacktrace doesn't have the dots, so for some reason the install didn't get the latest updates. Might have to pip uninstall keras-tqdm
and reinstall after pulling the latest changes.
I'll try to get this up on PyPI at some point soon.
Cheers, Ben
Up on PyPI! Tested on a clean machine and appears to be working fine. Can now install with just pip install keras-tqdm
.
I installed from pip on a different machine today and can verify everything works great. Nice work!
I'm using Python 3.5.2 from Anaconda on a Windows 7 machine in a Jupyter notebook.
'3.5.2 |Continuum Analytics, Inc.| (default, Jul 5 2016, 11:41:13) [MSC v.1900 64 bit (AMD64)]'
After I installed (by
git clone
and thenpython setup.py install
, it seems as if an.egg
file appeared in the expected place in my conda tensorflow environment, and onfrom keras_tqdm import TQDMCallback
it tries to look there for the installation but is not finding it correctly.This can be fixed by doing an
sys.path.insert(1, 'C:/<PATH>/keras-tqdm/keras_tqdm/')
to put the correct directory ahead of the non-working one in the sys.path, but that isn't ideal.Note that this is after pulling the updates from yesterday.
All in all, this is a minor problem to a very nice solution to the keras progbars freezing in Jupyter problem. Thank you!