| import pandas as pd | |
| from datasets import load_dataset | |
| data = load_dataset('relbert/semeval2012_relational_similarity_v4') | |
| 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()) | |