UW CLMS program project for ling-573 by Paige Finkelstein, Jacob Hoffman, Wesley Rose, and Joshua Tanner
All that is needed to run the end-to-end system on the devtest and evaltest datasets on patas is to submit D4.cmd to condor:
condor_submit D4.cmd
This will run the run.sh script, which will set up a conda environment and install all of the necessary dependencies.
It will write the generated summaries for devtest to the output/devtest/ directory, and write the generated summaries for evaltest to the output/evaltest/ directory.
The ROUGE evalulation score reports for both will be saved in the results directory to as rouge_scores_devtest.out and rouge_scores_evaltest.out.
To run the summarization system and output summaries (to outputs):
python3 src/baseline.py devtest|train|evaltest
Once you have summary results in the outputs/x directory, adjust the dataset name in the eval.sh file, and run:
sh eval.sh
If not running on patas, be sure to switch the file paths in eval.sh to point at your correct local paths.