tatoeba_large / script.py
mawadalla's picture
Create script.py
54a09a9 verified
raw
history blame
551 Bytes
import pyarrow.parquet as pq
class TatoebaLarge:
def __init__(self, file_path):
self.file_path = file_path
self.parquet_file = pq.ParquetFile(self.file_path)
def __iter__(self):
return self.parquet_file.iter_batches()
def __len__(self):
return self.parquet_file.metadata.num_rows
# Usage example
dataset = ParquetDataset("path/to/your/dataset.parquet")
for batch in dataset:
for example in batch.to_pydict():
# Do something with the example, e.g., preprocess the data
print(example)