So I wanted to do some grid searching for my model to improve its prediction. I decided to try talos, so I used
pip install talos
And wrote the import statement
import talos as ta
This is where it went wrong. I got the output listed below the checkboxes. I noticed that matplotlib had been downgraded to 2.2.3 after installing talos, so I force upgraded it to the latest stable again. The error message before and after were, I believe, slightly different, but both involved the backend in some way. I'm using python 3.6.0 on ubuntu 16.04 and running my model in a notebook using jupyterlab.
[x] I'm up-to-date with the latest release:
pip install -U talos
[x] I've confirmed that my Keras model works outside of Talos.
---------------------------------------------------------------------------
AttributeError Traceback (most recent call last)
<ipython-input-12-1f96ca05ae20> in <module>()
5 import datetime
6 import numpy as np
----> 7 import talos as ta
8 import matplotlib.pyplot as plt
9 from sklearn.model_selection import GridSearchCV
~/.pyenv/versions/3.6.0/lib/python3.6/site-packages/talos/__init__.py in <module>()
1 # import commands
2 from .scan.Scan import Scan
----> 3 from .commands.reporting import Reporting
4 from .commands.predict import Predict
5 from .commands.deploy import Deploy
~/.pyenv/versions/3.6.0/lib/python3.6/site-packages/talos/commands/reporting.py in <module>()
3
4 if is_connected() is True:
----> 5 from astetik import line, hist, corr, regs, bargrid, kde, box
6
7 from ..metrics.names import metric_names
~/.pyenv/versions/3.6.0/lib/python3.6/site-packages/astetik/__init__.py in <module>()
1 # PLOTS
----> 2 from .plots.corr import corr
3 from .plots.hist import hist
4 from .plots.kde import kde
5 from .plots.pie import pie
~/.pyenv/versions/3.6.0/lib/python3.6/site-packages/astetik/plots/corr.py in <module>()
4 import numpy as np
5 import matplotlib
----> 6 matplotlib.use('Agg')
7 # ENDS #
8 import matplotlib.pyplot as plt
~/.pyenv/versions/3.6.0/lib/python3.6/site-packages/matplotlib/__init__.py in use(arg, warn, force)
1389 if force:
1390 from matplotlib.pyplot import switch_backend
-> 1391 switch_backend(name)
1392 # Finally if pyplot is not imported update both rcParams and
1393 # rcDefaults so restoring the defaults later with rcdefaults
~/.pyenv/versions/3.6.0/lib/python3.6/site-packages/matplotlib/pyplot.py in switch_backend(newbackend)
211 newbackend, Backend.backend_version)
212
--> 213 required_framework = Backend.required_interactive_framework
214 if required_framework is not None:
215 current_framework = \
AttributeError: type object 'Backend' has no attribute 'required_interactive_framework'
Turns out my issue was that I was trying to import talos into an already running kernel that already had matplotlib.pyplot loaded. Restarting the kernel removed the error.
So I wanted to do some grid searching for my model to improve its prediction. I decided to try talos, so I used
And wrote the import statement
This is where it went wrong. I got the output listed below the checkboxes. I noticed that matplotlib had been downgraded to 2.2.3 after installing talos, so I force upgraded it to the latest stable again. The error message before and after were, I believe, slightly different, but both involved the backend in some way. I'm using python 3.6.0 on ubuntu 16.04 and running my model in a notebook using jupyterlab.
[x] I'm up-to-date with the latest release:
pip install -U talos
[x] I've confirmed that my Keras model works outside of Talos.