#%%
from datasets import load_dataset
import pandas as pd
from pathlib import Path
BASE_DIR = Path(__file__).resolve().parent
JSONL_PATH = BASE_DIR / "deepmind_math.jsonl"
clean_math = pd.read_json(
    JSONL_PATH,
    lines=True,        
    orient="records"   
)
GSM8k = load_dataset('openai/gsm8k','main', split= 'train')
MMMLU = load_dataset('cais/mmlu', 'college_mathematics', split='test+validation')
MMMU = load_dataset('MMMU/MMMU', 'Math', split='test+validation')
Olympiad_math = load_dataset('Hothan/OlympiadBench', 'TP_TO_maths_en_COMP', split='train')
Olympiad_math2 = load_dataset('Hothan/OlympiadBench', 'OE_TO_maths_en_COMP', split='train')
ScienceQA = load_dataset("derek-thomas/ScienceQA", split="train")
PubmedQA = load_dataset('qiaojin/PubMedQA','pqa_unlabeled', split='train')
# %%