lila.py

# Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

"""LILA: A Unified Benchmark for Mathematical Reasoning

Loading script author: Sean Welleck
"""


import json
import os
import datasets


_CITATION = """\
@INPROCEEDINGS{Mishra2022Lila,
  author = {
    Swaroop Mishra 
      and Matthew Finlayson 
      and Pan Lu 
      and Leonard Tang 
      and Sean Welleck 
      and Chitta Baral 
      and Tanmay Rajpurohit 
      and Oyvind Tafjord 
      and Ashish Sabharwal 
      and Peter Clark 
      and Ashwin Kalyan},
  title = {Lila: A Unified Benchmark for Mathematical Reasoning},
  booktitle = {Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing (EMNLP)},
  year = {2022}
}
"""

_DESCRIPTION = """\
Līla is a comprehensive benchmark for mathematical reasoning with over 140K natural language questions annotated with Python programs and natural language instructions. The data set comes with multiple splits: Līla-IID (train, dev, test), Līla-OOD (train, dev, test), and Līla-Robust.
"""

_HOMEPAGE = "https://lila.apps.allenai.org/"

_LICENSE = "Creative Commons Attribution 4.0 International"

_URL = "https://github.com/allenai/Lila/raw/b81117ac7e56cc1dfb0fcabf0005d1755177252b/lila.zip"

_BASEDIR = 'lila'
_MULTIDIR = 'multi'
_ALLDIR = 'all'

_NAMES = [
    'iid',
    'ood',
    'addsub',
    'amps_algebra',
    'amps_calculus',
    'amps_counting_and_stats',
    'amps_geometry',
    'amps_linear_algebra',
    'amps_number_theory',
    'APPS_structured',
    'asdiv',
    'conala_structured',
    'deepmind_mathematics_algebra',
    'deepmind_mathematics_basicmath',
    'deepmind_mathematics_calculus',
    'deepmind_mathematics_muldiv',
    'deepmind_mathematics_numbertheory',
    'dolphin_t2_final',
    'draw_structured',
    'GSM8k_structured',
    'MATH_algebra_crowdsourced',
    'MATH_counting_and_probability_crowdsourced',
    'MATH_intermediate_algebra_crowdsourced',
    'mathqa_gain',
    'mathqa_general',
    'mathqa_geometry',
    'mathqa_other',
    'mathqa_physics',
    'mathqa_probability',
    'mbpp_structured',
    'MCTaco_event_duration_structured',
    'MCTaco_event_ordering_structured',
    'MCTaco_event_typical_time_structured',
    'MCTaco_frequency_structured',
    'MCTaco_stationarity_structured',
    'multiarith',
    'Numersense_structured',
    'NumGLUE_Type_1_crowdsourced',
    'NumGLUE_Type_2_crowdsourced',
    'NumGLUE_Type_3_crowdsourced',
    'NumGLUE_Type_4_crowdsourced',
    'NumGLUE_Type_5_crowdsourced',
    'NumGLUE_Type_6_crowdsourced',
    'NumGLUE_Type_7_crowdsourced',
    'NumGLUE_Type_8_crowdsourced',
    'simuleq',
    'singleop',
    'singleq',
    'svamp_structured'
]

VERSION = "1.1.0"


class Lila(datasets.GeneratorBasedBuilder):
    """Lila: A Unified Benchmark for Mathematical Reasoning"""

    BUILDER_CONFIGS = [
        datasets.BuilderConfig(name=name, version=VERSION, description=name) for name in _NAMES
    ]

    DEFAULT_CONFIG_NAME = "iid"

    def _info(self):
        features = datasets.Features(
            {
                "input": datasets.Value("string"),
                "output_program": datasets.Value("string"),
                "output_answer": datasets.Value("string"),
                "split": datasets.Value("string"),
                "dataset": datasets.Value("string"),
            }
        )
        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=features,
            homepage=_HOMEPAGE,
            license=_LICENSE,
            citation=_CITATION,
        )

    def _split_generators(self, dl_manager):
        urls = _URL
        data_dir = dl_manager.download_and_extract(urls)
        if self.config.name in {'iid', 'ood'}:
            train_filepath = os.path.join(data_dir, _BASEDIR, _MULTIDIR, self.config.name, "train.json")
            dev_filepath = os.path.join(data_dir, _BASEDIR, _MULTIDIR, self.config.name, "dev.json")
            test_filepath = os.path.join(data_dir, _BASEDIR, _MULTIDIR, self.config.name, "test.json")
        else:
            train_filepath = os.path.join(data_dir, _BASEDIR, _ALLDIR, self.config.name + ".json")
            dev_filepath = train_filepath
            test_filepath = train_filepath
        return [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                gen_kwargs={
                    "filepath": train_filepath,
                    "split": "train",
                },
            ),
            datasets.SplitGenerator(
                name=datasets.Split.VALIDATION,
                gen_kwargs={
                    "filepath": dev_filepath,
                    "split": "dev",
                },
            ),
            datasets.SplitGenerator(
                name=datasets.Split.TEST,
                gen_kwargs={
                    "filepath": test_filepath,
                    "split": "test"
                },
            ),
        ]

    def _generate_examples(self, filepath, split):
        if self.config.name in {"iid", "ood"}:
            with open(filepath) as f:
                for key, row in enumerate(f.readlines()):
                    data = json.loads(row)
                    yield key, {
                        "input": data["Input"],
                        "output_program": data["Output Program"][0],
                        "output_answer": data["Output Answer"][0],
                        "split": data["split"],
                        "dataset": data["dataset"],
                    }
        else:
            for key, data in enumerate(json.load(open(filepath))["Instances"]):
                if data['split'] == split:
                    yield key, {
                        "input": data["Input"],
                        "output_program": data["Output Program"][0],
                        "output_answer": data["Output Answer"][0],
                        "split": data["split"],
                        "dataset": self.config.name,
                    }