tokenize_dataset_rows_belle.py

import argparse
import json
from tqdm import tqdm
import ast

import datasets
import transformers
from alps.pytorch.api.utils.web_access import patch_requests
patch_requests()


def preprocess(tokenizer, config, example, max_seq_length):
    prompt = example["input"]
    target = example["target"]
    prompt_ids = tokenizer.encode(prompt, max_length=max_seq_length, truncation=True)
    target_ids = tokenizer.encode(
        target,
        max_length=max_seq_length,
        truncation=True,
        add_special_tokens=False)
    input_ids = prompt_ids + target_ids + [config.eos_token_id]
    return {"input_ids": input_ids, "seq_len": len(prompt_ids)}


def read_jsonl(path, max_seq_length, skip_overlength=False):
    model_name = "THUDM/chatglm-6b"
    tokenizer = transformers.AutoTokenizer.from_pretrained(
        "THUDM/chatglm-6b", cache_dir ='./',trust_remote_code=True
    )
    config = transformers.AutoConfig.from_pretrained(		    
        model_name, cache_dir ='./', trust_remote_code=True, device_map='auto')
    with open(path, "r") as f:
        for line in tqdm(f.readlines()):
            example = ast.literal_eval(line)
            feature = preprocess(tokenizer,config, example, max_seq_length)
            if skip_overlength and len(feature["input_ids"]) > max_seq_length:
                continue
            feature["input_ids"] = feature["input_ids"][:max_seq_length]
            yield feature


def main():
    parser = argparse.ArgumentParser()
    parser.add_argument("--jsonl_path", type=str, default="data/Belle.train.json")
    parser.add_argument("--save_path", type=str, default="data/alpaca")
    parser.add_argument("--max_seq_length", type=int, default=2048)
    parser.add_argument("--skip_overlength", type=bool, default=False)
    args = parser.parse_args()

    dataset = datasets.Dataset.from_generator(lambda: read_jsonl(args.jsonl_path, args.max_seq_length, args.skip_overlength))
    dataset.save_to_disk(args.save_path)


if __name__ == "__main__":
    main()