mwcvitkovic / open-vocabulary-learning-on-source-code-with-a-graph-structured-cache Goto Github PK

Hi @mwcvitkovic ,

Thank you for a wonderful reproduction package!

I've been working on using models with vanilla AST-s generated with JavaParser for bug prediction task with some success. However, I believe your work (GGNN+GSC) opens up opportunities to improve. As you are the author of this source code, perhaps you can guide me a bit.

1. What would be required to use the approach as a binary classification model for method level tasks?

(The aim is to have a supervised model and feed examples of good code and bad code in order to train. Later, use this trained model to get a prediction (bad/good) at a method level)

I tried to use "conda env create -f environment.yml" to create an environment but got the following problem:

Collecting package metadata (repodata.json): done
Solving environment: failed

ResolvePackageNotFound:
  - pykerberos=1.1.14
  - graphite2=1.3.10
  - gstreamer=1.8.0
  - pango=1.40.3
  - gmp=6.1.2
  - libtool=2.4.6
  - harfbuzz=1.3.4
  - libgfortran=3.0.0
  - libgfortran-ng=7.2.0
  - libedit=3.1
  - jasper=1.900.1
  - readline=7.0
  - fontconfig=2.12.1
  - ncurses=6.0
  - glib=2.51.4
  - dbus=1.10.22
  - sparkmagic=0.12.5
  - libstdcxx-ng=7.2.0
  - gst-plugins-base=1.8.0
  - ptyprocess=0.5.2
  - x264=20131217
  - libgcc-ng=7.2.0 

I wonder how it is possible that there are conflicts since you successfully used an environment with the specification.