|
|
|
|
|
from datasets import load_dataset
|
|
|
import pandas as pd
|
|
|
import os
|
|
|
os.chdir(os.path.dirname(__file__))
|
|
|
clean_math = pd.read_json(
|
|
|
"deepmind_math.jsonl",
|
|
|
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')
|
|
|
|
|
|
|
