import pandas as pd | |
from datasets import load_dataset | |
data = load_dataset('relbert/semeval2012_relational_similarity_v3') | |
stats = [] | |
for k in data.keys(): | |
for i in data[k]: | |
stats.append( | |
{ | |
'relation_type': i['relation_type'], | |
'split': k, | |
'positives': len(i['positives']), | |
'negatives': len(i['negatives']), | |
'level': i['level'] | |
}) | |
df = pd.DataFrame(stats) | |
g = df.groupby(['relation_type', 'level', 'split']).sum() | |
g.to_csv('stats.csv') | |
with open('stats.md', 'w') as f: | |
f.write(g.to_markdown()) | |