ArgLab / AWE_Workbench

This repository contains the web service API for calling AWE Workbench to support automated writing evaluation, plus examples of use.
Other
0 stars 0 forks source link

Testing (without holmes) #12

Open duckduckdoof opened 2 months ago

duckduckdoof commented 2 months ago

Changes involve:

bradley-erickson commented 3 weeks ago

This needs documentation on how to install and run tests. Much of the original documentation is not very clean (no code-blocks for commands and a bunch of unneeded =s).

I tried to go and test this code to make sure it works on my machine as well and immediately ran into issues regarding installation. The requirements.txt file specifies awe_languagetool which I know where to find and install, but pip does not. For the additional AWE items, we should specify them similar to how its specified in Learning Observer.

AWE_SpellCorrect @ git+https://github.com/ETS-Next-Gen/AWE_SpellCorrect.git
AWE_Components @ git+https://github.com/ETS-Next-Gen/AWE_Components.git
AWE_Lexica @ git+https://github.com/ETS-Next-Gen/AWE_Lexica.git
AWE_LanguageTool @ git+https://github.com/ETS-Next-Gen/AWE_LanguageTool.git
duckduckdoof commented 2 weeks ago

Yeah tbh I'm not sure why the requirements.txt file and setup.cfg are both there; the installation of Workbench doesn't explicitly refer to the former. Currently working to resolve those dependencies.

Additionally, a numpy update has stopped the python -m awe_workbench.setup.data from working... I've added a version lock and PR in AWE_Components (https://github.com/ArgLab/AWE_Components/pull/6).

bradley-erickson commented 1 week ago

Whenever I try to install the workbench, it winds up checking a bunch of different spacy versions. The download and installation of each one takes a bit (which heavily adds up over time). This should be investigated so the installation immediately installs an appropriate version.