[!IMPORTANT]
This is the training dataset for the ICAPS 2025 paper "Chasing Progress, Not Perfection: Revisiting Strategies for End-to-End LLM Plan Generation".
from pathlib import Path
import os
import jsonlines
from copy import deepcopy
from datasets import load_dataset
from icecream import ic
import enum
from enum import IntEnum
from enum import auto
class CONFIG_TYPES(enum.Enum):
# "type_id": ['t0', 'accu-t1', 'accu-t2', 'accu-t3', 'accu-t4', 'accu-t4', 'accu-t1+t4', 'accu-t2+t4']… See the full description on the dataset page:
https://huggingface.co/datasets/huangsukai/llm_plan_gen_dataset_accu_t1.
