Commit 8cbd1ff3 authored by burkhardt's avatar burkhardt
Browse files

Update proBert.py

parent 4c2c4b17

probing/proBert.py

+4 −4
Original line number Diff line number Diff line
@@ -27,7 +27,7 @@ def open_ncs(tsv_list_local): 


# Liste mit Templates öffnen

def open_templates():

    return pd.read_csv('data/templates.tsv', sep="\t").dropna(subset=['Relation'])

    return pd.read_csv('../data/templates.tsv', sep="\t").dropna(subset=['Relation'])





# Verben und dazugehörige Wahrscheinlichkeiten mit MLM suchen
@@ -95,16 +95,16 @@ def clustering(all_guesses_cluster, templates_cluster): 


        pbar.close()



    all_clusters.to_csv('data/fine/all_clusters.tsv', sep='\t')

    all_clusters.to_csv('results/fine/all_clusters.tsv', sep='\t')





def main():

    tic = time.perf_counter()

    tsv_list = ['data/data.tsv']

    tsv_list = ['../data/probing/data.tsv']

    ncs_with_rel = open_ncs(tsv_list)

    templates = open_templates()

    all_guesses = mlm(ncs_with_rel, templates).dropna()

    all_guesses.to_csv('data/fine/all_guesses.tsv', sep='\t')

    all_guesses.to_csv('results/fine/all_guesses.tsv', sep='\t')

    clustering(all_guesses, templates)

    toc = time.perf_counter()

    print(f"\nTook me some time: {toc - tic:0.4f} seconds")