gaia_dataset.py

from smolagents import FinalAnswerTool
from datasets import load_dataset, Dataset
import json
import os

HF_ACCESS_TOKEN = os.getenv("HF_ACCESS_TOKEN") 

gaia_dataset = load_dataset("gaia-benchmark/GAIA", "2023_level1", trust_remote_code=True, split="validation", token=HF_ACCESS_TOKEN)

def get_example_by_feature_value(dataset: Dataset, feature_name: str, feature_value: str):

    for example in dataset:
        if example[feature_name] == feature_value:
            return example
        
    return None


def get_question(task_id: str) -> str:

    question_data = get_example_by_feature_value(gaia_dataset, "task_id", task_id)

    question_text = "Question: " + question_data["Question"] + "\n\n"

    if question_data["file_name"]:
        question_text = question_text + "File path: " + question_data["file_path"] + "\n\n"

    question_text = question_text + "Tools required:\n" + question_data["Annotator Metadata"]['Tools'] + "\n\n"
    question_text = question_text + "Approximately, the problem can be solved as follows::\n" + question_data["Annotator Metadata"]["Steps"] + "\n\n"

    return question_text



if __name__ == "__main__":
    id = "a1e91b78-d3d8-4675-bb8d-62741b4b68a6" # Question without file
    # id = "cca530fc-4052-43b2-b130-b30968d8aa44" # Question with file

    print(get_question(id))