This PR contains my changes in the /scripts folder together with eraserbenchmark and the error inspector as well as the graphical penman editor (streamlit app).
/scripts/evaluate_hatexplain.py : now able to calculate the ERASER metrics! predicts rationals from the rules
/scripts/inspect_hatexplain.py : inspector for FP/FP/TN/TP samples, random+seed and paging mode
/scripts/eraserbenchmark/ : needed by those two
/frontend/graphedit.py : generates UD penman from text on the fly, as well as graph images from UD penman
TODO:
Edit eraserbenchmark so that data input files can be named other than val, test & train
Own output folder for different classification projects (not only /hatexplain)
This PR contains my changes in the /scripts folder together with eraserbenchmark and the error inspector as well as the graphical penman editor (streamlit app).
/scripts/evaluate_hatexplain.py : now able to calculate the ERASER metrics! predicts rationals from the rules /scripts/inspect_hatexplain.py : inspector for FP/FP/TN/TP samples, random+seed and paging mode /scripts/eraserbenchmark/ : needed by those two
/frontend/graphedit.py : generates UD penman from text on the fly, as well as graph images from UD penman
TODO: Edit eraserbenchmark so that data input files can be named other than val, test & train Own output folder for different classification projects (not only /hatexplain)