prepare_data.py

import ast

import pandas as pd

import os
from typing import Union

import datasets
from datasets import load_dataset



def save_and_compress(dataset: Union[datasets.Dataset, pd.DataFrame], name: str, idx=None):
    if idx:
        path = f"{name}_{idx}.jsonl"
    else:
        path = f"{name}.jsonl"

    print("Saving to", path)
    dataset.to_json(path, force_ascii=False, orient='records', lines=True)

    print("Compressing...")
    os.system(f'xz -zkf -T0 {path}')  # -TO to use multithreading

def parse_to_list(example):
    example["tokens"] = ast.literal_eval(example["tokens"])
    example["labels"] = ast.literal_eval(example["labels"])
    return example

for subset in ["agent", "argType"]:
    for split in ["train", "val", "test"]:
        dataset = load_dataset("csv", data_dir=f"original_data/{split}/{subset}", sep="\t", split="train")
        dataset = dataset.map(parse_to_list)
        print(dataset[0])
        save_and_compress(dataset, f"data/{subset}/{split}")