Norod78's picture
Update load_ds.py
00e5f76
raw
history blame
891 Bytes
import pandas as pd
from datasets import Dataset
from collections import defaultdict
import textwrap
def chunk_examples(txt_list):
chunks = []
for sentence in txt_list:
chunks += textwrap.wrap(sentence, width=1000)
return chunks
txt = open('HebrewStageAndLyrics_clean.txt').read()
parsed = txt.split("\n\n")
df = pd.DataFrame(parsed)
parsed_lines = df[0].values.tolist()
data = defaultdict(list)
chunked_lines = chunk_examples(parsed_lines)
for line in chunked_lines:
sc_line = line.replace(";", ",").replace("\n", " ; ")
data['text'].append(sc_line)
dataset = Dataset.from_dict(data).shuffle(seed=42)
dataset = dataset.train_test_split(test_size=0.05)
print(dataset)
print(dataset.column_names)
for i in range(20, 22, 1):
print(dataset['train'][i])
print(dataset['test'][i])
dataset.push_to_hub("Norod78/HebrewStageAndLyricsWithNewLines")