Skip to content

tmu-nlp/CodaLab_DoTS_on_Wiki

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

DoTS/Wiki: Automatic Program and Document for CodaLab Bundle

We added the following corpora repos via:

  • SWiPE: git submodule add https://github.com/salesforce/simplification.git Corpora/SWiPE
  • Klexikon: git submodule add https://github.com/dennlinger/klexikon.git Corpora/Klexikon
  • JADOS: git submodule add https://github.com/tmu-nlp/JADOS.git Corpora/JADOS-Wiki

Usage:

So you need to clone to the local via:

git clone --recurse-submodules https://github.com/tmu-nlp/CodaLab_DoTS_on_Wiki.git
cd CodaLab_DoTS_on_Wiki

Requirements:

pip install -r requirements.txt    # for Python
Rscript requirements.R             # for R

Setup Docker for CodaLab workers:

docker buildx build -t zchen0420/sari:0 . --push
# refer to the same image name in competition.yaml

Build:

cd Bundle
./create.sh
  • Upload the generated bundle.zip to CodaLab.
  • Check or delete the tmp folder for your confirmation.

How it works?

Scoring Program:

  • CodaLab workers accept a submission and unzip it into a temporary directory $input as res folder (along with ref as the reference to execute score.py, as defined in metadata).
  • score.py (requiring the above docker image) writes the scores to $output/scores.txt, which are projected on to the leaderboard defined in competition.yaml.

If you want to modify the scores, check both score.py and competition.yaml.

Interface:

  • You need to configure the competition.yaml (e.g., start_date and end_date) and modify the corresponding html files in Bundle/html.
  • Note that images (i.e., svg files) have been uploaded to a separate server, accessible by <a href="https://cl.sd.tmu.ac.jp/~zchen/...">.

About

Automatic program and document for CodaLab bundle

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published