I’m having an issue in which I’m trying to get Rasa to pick up a different version of Python.
In my terminal window when I run the python command I can. see the correct version i.e. 3.8.12, however, when I run Rasa it shows up 3.7.12. I’m using Virtualenv to set the version of Python.
How can I fix this? Also is there some way to get Rasa to explicitly use a particular version or does it just get it from the environment variables?
@qamir alright! In your, system Python Verison is 3.8.12 which is fine as you have checked on the terminal window using the command python --version, is that you have install Anconda Version?
@qamir When you created the Virtual Environment for Rasa installation the virtualenv picked the latest supported version of python i.e I guess 3.7.12 (I can be totally wrong on this) as when I created my virtualenv it’s 3.7.7 and system 3.6.8.
So, if you delete the virtualenv or update the virtualenv or try create the new virtaulenv may be it will pic the latest python version.
When I run python --version I see 3.8.12 as expected. I ended up upgrading to Rasa 3.0 which fixed the issue.
As a summary of what happened:
I was on version Rasa 2.7.0 and while setting up another ML project unrelated to Rasa Tensorflow became updated which caused Rasa to throw this error when I ran my Rasa project:
[AttributeError: Can’t get attribute ‘FeatureArray’ on <module ‘rasa.utils.tensorflow.model_data’…
I didn’t have virtual environments set for both projects so paths and libraries got mixed up, I cleaned everything up and created virtual environments for both projects but the problem for Rasa remained & upgrading Rasa fixed it.
@nik202 Yes all is good, thanks! My bad for not having my virtual environments set up in the first place, for deployment to production I’m using Docker so that part is isolated.