TIF_E41211115_lstm-quiz-gen.../dataset_converter.py

14 lines
403 B
Python

import json
with open("dataset/dataset_ner_srl.json", encoding="utf-8") as f:
data = json.load(f)
with open("dataset/dataset_ner_srl.tsv", "w", encoding="utf-8") as f:
for entry in data:
for tok, ner, srl in zip(
entry["tokens"], entry["labels_ner"], entry["labels_srl"]
):
f.write(f"{tok}\t{ner}\t{srl}\n")
f.write("\n") # Separate sentences