acp_bench / acp_bench.py
harshakokel's picture
Update acp_bench.py
306bd24 verified
"""ACP Bench dataset."""
import json
import datasets
import itertools
_CITATION = """\
@article{kokel2024acp,
title={ACPBench: Reasoning about Action, Change, and Planning},
author={Kokel, Harsha and Katz, Michael and Srinivas, Kavitha and Sohrabi, Shirin},
journal={arXiv},
year={2024}
}
"""
_DESCRIPTION = """ACPBench consists of 7 reasoning tasks over 13 domains. The 13 domains include 11 classical planning domains, ALFWorld, and a novel Swap domain. The 7 tasks included in ACPBench are Action Applicability (app), Progression (prog), Atom Reachability (reach), Validation (val), Action Reachability (areach), Justification (just), and Landmarks (land)."""
_HOMEPAGE = "https://ibm.github.io/ACPBench/"
_LICENSE = "MIT"
_BASE_URL = "https://raw.github.com/ibm/ACPBench/main/dataset"
task_list = [
"app",
"areach",
"just",
"land",
"prog",
"reach",
"val"
]
format_list = [
"bool", "mcq"
]
class ACPConfig(datasets.BuilderConfig):
def __init__(self, urls, **kwargs):
"""
urls: *dict[string]*, the urls for each split of the ACPBench set.
"""
super().__init__(version=datasets.Version("1.0.0"), **kwargs)
self.urls = urls
class ACP(datasets.GeneratorBasedBuilder):
BUILDER_CONFIGS = [
ACPConfig(
name=f"acp_{task_name}_{format_name}",
urls={
"test": f"{_BASE_URL}/{task_name}/test.{format_name}.json.gz",
"val": f"{_BASE_URL}/{task_name}/dev.{format_name}.json",
},
)
for task_name, format_name in itertools.product(task_list,format_list)
]
def _info(self):
features = {
"context": datasets.Value("string"),
"question": datasets.Value("string"),
"answer": datasets.Value("string"),
"group": datasets.Value("string"),
"id": datasets.Value("string")
}
if 'mcq' in self.config.name:
features["query"]= datasets.Value("string")
features["choices"]= datasets.features.Sequence(feature={'text': datasets.Value(dtype='string', id=None), 'label': datasets.Value(dtype='string', id=None)}, length=-1, id=None)
return datasets.DatasetInfo(
description=_DESCRIPTION,
features=datasets.Features(features),
homepage=_HOMEPAGE,
license=_LICENSE,
citation=_CITATION,
)
def _split_generators(self, dl_manager):
data_dir = dl_manager.download_and_extract(self.config.urls)
return [
datasets.SplitGenerator(
name=datasets.Split.TEST,
gen_kwargs={
"filepath": data_dir["test"],
},
),datasets.SplitGenerator(
name=datasets.Split.VALIDATION,
gen_kwargs={
"filepath": data_dir["val"],
},
)
]
def _generate_examples(self, filepath):
with open(filepath) as f:
examples = json.load(f)
for i, instance in enumerate(examples):
yield i, instance