epsilon-compositionality

Code for our paper Measuring Idiomaticity in Text Embedding Models with 𝜀-compositionality
git clone https://esimon.eu/repos/epsilon-compositionality.git
Log | Files | Refs | Submodules | README

commit 9983c50a268c33035c7aabb8b12a3e02017505b8
parent bb9a7a5fc4e66784095d6b8fefd891596691742f
Author: Sondre Wold <sondrewold1@gmail.com>
Date:   Thu, 25 Sep 2025 16:07:09 +0200

reproduce file updates

Diffstat:
Mreproduce.sh | 13++++++-------
1 file changed, 6 insertions(+), 7 deletions(-)

diff --git a/reproduce.sh b/reproduce.sh @@ -1,12 +1,11 @@ -results_folder=./reproduction_model_results_24 -epsilon_file=./distributions_24.json -final_result_file=./result_24.txt +results_folder=./reproduction_model_results_25 +epsilon_file=./distributions_25.json +final_result_file=./result_25.txt echo "Running all the embedding models on NCIMP" -python experiment.py -data_path ncimp/ncip_clean.csv -experiment_type ncip -normalization_value 0.0 -result_path $results_folder -python experiment.py -data_path ncimp/ncip_clean.csv -experiment_type ncip-non -normalization_value 0.0 -result_path $results_folder - +python experiment.py -data_path ncimp/classed_data.csv -experiment_type ncip -result_path $results_folder +python experiment.py -data_path ncimp/classed_data.csv -experiment_type ncip-non -result_path $results_folder echo "Calculating epsilon values..." python get_epsilon.py -points_path $results_folder -result_path $epsilon_file echo "Running statistical testing..." -python statistical_testing.py -epsilons_path $epsilon_file > $final_result_file +python statistical_testing.py -epsilons_path $epsilon_file -class_type PC NC > $final_result_file