arxiv works

citations_and_descriptions.py

@@ -1,56 +0,0 @@
-_FS_CITATION = """
-TBD
-"""
-
-_FS_DESCRIPTION = """
-TBD
-"""
-
-_SUMM_SCREEN_DESCRIPTION = """
-SummScreenFD (Chen  et al., 2021) is a summarization dataset in the domain of TV shows (e.g. Friends, Game of Thrones).
-Given a transcript of a specific episode, the goal is to produce the episode's recap.
-The original dataset is divided into two complementary subsets, based on the source of its community contributed transcripts. 
-For SCROLLS, we use the ForeverDreaming (FD) subset, as it incorporates 88 different shows, 
-making it a more diverse alternative to the TV MegaSite (TMS) subset, which has only 10 shows. 
-Community-authored recaps for the ForeverDreaming transcripts were collected from English Wikipedia and TVMaze."""
-
-_GOV_REPORT_DESCRIPTION = """
-GovReport (Huang et al., 2021) is a summarization dataset of reports addressing various national policy issues published by the 
-Congressional Research Service and the U.S. Government Accountability Office, where each document is paired with a hand-written executive summary.
-The reports and their summaries are longer than their equivalents in other popular long-document summarization datasets; 
-for example, GovReport's documents are approximately 1.5 and 2.5 times longer than the documents in Arxiv and PubMed, respectively."""
-
-_ARXIV_DESCRIPTION = """
-"""
-
-_SUMM_SCREEN_CITATION = r"""
-@misc{chen2021summscreen,
-      title={SummScreen: A Dataset for Abstractive Screenplay Summarization}, 
-      author={Mingda Chen and Zewei Chu and Sam Wiseman and Kevin Gimpel},
-      year={2021},
-      eprint={2104.07091},
-      archivePrefix={arXiv},
-      primaryClass={cs.CL}
-}"""
-
-_GOV_REPORT_CITATION = r"""
-@inproceedings{huang-etal-2021-efficient,
-    title = "Efficient Attentions for Long Document Summarization",
-    author = "Huang, Luyang  and
-      Cao, Shuyang  and
-      Parulian, Nikolaus  and
-      Ji, Heng  and
-      Wang, Lu",
-    booktitle = "Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies",
-    month = jun,
-    year = "2021",
-    address = "Online",
-    publisher = "Association for Computational Linguistics",
-    url = "https://aclanthology.org/2021.naacl-main.112",
-    doi = "10.18653/v1/2021.naacl-main.112",
-    pages = "1419--1436",
-    abstract = "The quadratic computational and memory complexities of large Transformers have limited their scalability for long document summarization. In this paper, we propose Hepos, a novel efficient encoder-decoder attention with head-wise positional strides to effectively pinpoint salient information from the source. We further conduct a systematic study of existing efficient self-attentions. Combined with Hepos, we are able to process ten times more tokens than existing models that use full attentions. For evaluation, we present a new dataset, GovReport, with significantly longer documents and summaries. Results show that our models produce significantly higher ROUGE scores than competitive comparisons, including new state-of-the-art results on PubMed. Human evaluation also shows that our models generate more informative summaries with fewer unfaithful errors.",
-}"""
-
-_ARXIV_CITATION = r"""
-}"""

configs/arxiv.py

@@ -1,37 +0,0 @@
-from typing import NoReturn
-
-from configs.fs import FSConfig
-
-
-class ArxivConfig(FSConfig):
-    def __init__(self, **kwargs):
-        super().__init__(**kwargs)
-
-    @property
-    def id_key(self) -> str:
-        return "article_id"
-
-    @property
-    def source_key(self) -> str:
-        return "article_text"
-
-    @property
-    def target_key(self) -> str:
-        return "abstract_text"
-
-    @property
-    def train_file(self) -> str:
-        return "train.txt"
-
-    @property
-    def validation_file(self) -> str:
-        return "val.txt"
-
-    @property
-    def test_file(self) -> str:
-        return "test.txt"
-
-    def process(self, example) -> NoReturn:
-        example[self.source_key] = " ".join(example[self.source_key])
-        example[self.target_key] = " ".join(example[self.target_key]).replace("<S>", "").replace("</S>", "")
-        del example["labels"]

configs/fs.py

@@ -1,66 +0,0 @@
-from abc import abstractmethod
-from typing import Optional, NoReturn, Union
-
-import datasets
-
-
-class FSConfig(datasets.BuilderConfig):
-    """BuilderConfig for FS."""
-
-    def __init__(self, additional_features, data_url, citation, url, **kwargs):
-        """BuilderConfig for FS.
-        Args:
-          additional_features: `list[string]`, list of the features that will appear in the feature dict
-           additionally to the self.id_key, self.source_key and self.target_key. Should not include "label".
-          data_url: `string`, url to download the zip file from.
-          citation: `string`, citation for the data set.
-          url: `string`, url for information about the data set.
-          label_classes: `list[string]`, the list of classes for the label if the
-            label is present as a string. Non-string labels will be cast to either
-            'False' or 'True'.
-          **kwargs: keyword arguments forwarded to super.
-        """
-        super(FSConfig, self).__init__(version=datasets.Version("1.0.0"), **kwargs)
-        self.features = [self.id_key, self.source_key, self.target_key] + additional_features
-        if self.question_key:
-            self.features += [self.question_key]
-        self.data_url = data_url
-        self.citation = citation
-        self.url = url
-
-    @property
-    @abstractmethod
-    def id_key(self) -> str:
-        pass
-
-    @property
-    @abstractmethod
-    def train_file(self) -> str:
-        pass
-
-    @property
-    @abstractmethod
-    def validation_file(self) -> str:
-        pass
-
-    @property
-    @abstractmethod
-    def test_file(self) -> str:
-        pass
-
-    @property
-    @abstractmethod
-    def source_key(self) -> str:
-        pass
-
-    @property
-    def question_key(self) -> Union[str, None]:
-        return None
-
-    @property
-    @abstractmethod
-    def target_key(self) -> str:
-        pass
-
-    def process(self, example) -> NoReturn:
-        pass

configs/scrolls.py

@@ -1,30 +0,0 @@
-from configs.fs import FSConfig
-
-
-class ScrollsConfig(FSConfig):
-    def __init__(self, **kwargs):
-        super().__init__(**kwargs)
-
-    @property
-    def source_key(self) -> str:
-        return "input"
-
-    @property
-    def target_key(self) -> str:
-        return "output"
-
-    @property
-    def train_file(self) -> str:
-        return "train.jsonl"
-
-    @property
-    def validation_file(self) -> str:
-        return "validation.jsonl"
-
-    @property
-    def test_file(self) -> str:
-        return "test.jsonl"
-
-    @property
-    def id_key(self) -> str:
-        return "pid"

configs/super_glue.py

@@ -1,48 +0,0 @@
-from typing import Optional, Union
-
-from configs.fs import FSConfig
-
-
-class SuperGLUEConfig(FSConfig):
-    @property
-    def id_key(self) -> str:
-        return "idx"
-
-    @property
-    def target_key(self) -> str:
-        return "label"
-
-    @property
-    def train_file(self) -> str:
-        return "train.jsonl"
-
-    @property
-    def validation_file(self) -> str:
-        return "val.jsonl"
-
-    @property
-    def test_file(self) -> str:
-        return "test.jsonl"
-
-
-class BoolQConfig(SuperGLUEConfig):
-
-    @property
-    def source_key(self) -> str:
-        return "passage"
-
-    @property
-    def question_key(self) -> Union[str, None]:
-        return "question"
-
-
-class RTEConfig(SuperGLUEConfig):
-
-    # TODO HACK - we treat premise == source, hypothesis == question
-    @property
-    def source_key(self) -> str:
-        return "premise"
-
-    @property
-    def question_key(self) -> Union[str, None]:
-        return "hypothesis"

data/arxiv_debug.zip

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:51575dfb34c29cc1646f444cce45f1e47f36839682c9e6c78a68fc53e40ce915
-size 954416

data/summ_screen_fd_debug.zip

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:735c0ee602901d0e6a548d104812fd60733a904d264d4a36d2a494920de747c3
-size 685706

debug.py

@@ -1,13 +1,7 @@
-from transformers import AutoTokenizer
-from datasets import load_dataset
-
-def main():
-    # dataset = load_dataset("tau/fs",name="summ_screen_fd", max_source_length=512, tokenizer=tokenizer, prompt="Summary:")
-    ssfd_debug = load_dataset("/Users/yuvalkirstain/repos/fs", name="summ_screen_fd")
-    x = 5
-    # arxiv_debug = load_dataset("/Users/yuvalkirstain/repos/fs", name="arxiv_debug", max_source_length=512,
-    #                            tokenizer=tokenizer, prompt="Summarize the above:")
-
+import datasets
 
 if __name__ == '__main__':
-    main()
+    dataset = datasets.load_dataset("fs.py", 'arxiv', streaming=True, split="validation")
+    it = iter(dataset)
+    a = next(it)
+    x = 5

fs.py

@@ -1,74 +1,94 @@
+# coding=utf-8
+# Lint as: python3
+"""The SCROLLS benchmark."""
+
 import json
 import os
+from abc import abstractmethod
 
 import datasets
-from citations_and_descriptions import (
-    _SUMM_SCREEN_DESCRIPTION, _SUMM_SCREEN_CITATION,
-    _GOV_REPORT_CITATION, _GOV_REPORT_DESCRIPTION,
-    _ARXIV_CITATION, _ARXIV_DESCRIPTION,
-    _FS_DESCRIPTION, _FS_CITATION,
-)
-from configs.arxiv import ArxivConfig
-from configs.scrolls import ScrollsConfig
-from configs.super_glue import BoolQConfig, RTEConfig
-
-
-class FS(datasets.GeneratorBasedBuilder):
-    """The SCROLLS benchmark."""
+
+
+class FewsionConfig(datasets.BuilderConfig):
+    """BuilderConfig for SCROLLS."""
+
+    def __init__(self, data_url, **kwargs):
+        """BuilderConfig for SCROLLS.
+        Args:
+          features: `list[string]`, list of the features that will appear in the
+            feature dict. Should not include "label".
+          data_url: `string`, url to download the zip file from.
+          citation: `string`, citation for the data set.
+          url: `string`, url for information about the data set.
+          label_classes: `list[string]`, the list of classes for the label if the
+            label is present as a string. Non-string labels will be cast to either
+            'False' or 'True'.
+          **kwargs: keyword arguments forwarded to super.
+        """
+        super(FewsionConfig, self).__init__(version=datasets.Version("1.0.0"), **kwargs)
+        self.data_url = data_url
+        self.features = [self.source_column_name, self.target_column_name, self.id_column_name]
+        if self.question_column_name:
+            self.features.append(self.question_column_name)
+
+    @property
+    @abstractmethod
+    def source_column_name(self) -> str:
+        pass
+
+    @property
+    @abstractmethod
+    def target_column_name(self) -> str:
+        pass
+
+    @property
+    @abstractmethod
+    def question_column_name(self) -> str:
+        pass
+
+    @property
+    @abstractmethod
+    def id_column_name(self) -> str:
+        pass
+
+
+class ArxivConfig(FewsionConfig):
+
+    @property
+    def source_column_name(self) -> str:
+        return "article"
+
+    @property
+    def target_column_name(self) -> str:
+        return "abstract"
+
+    @property
+    def question_column_name(self) -> str:
+        pass
+
+    @property
+    def id_column_name(self) -> str:
+        return "article_id"
+
+
+class Fewsion(datasets.GeneratorBasedBuilder):
+
     DEFAULT_WRITER_BATCH_SIZE = 1000  # because Narrative QA is a rather large dataset
     BUILDER_CONFIGS = [
-        # word level
-        BoolQConfig(
-            additional_features=[],
-            name="boolq",
-            description="",  # TODO
-            data_url="https://dl.fbaipublicfiles.com/glue/superglue/data/v2/BoolQ.zip",
-            citation="",  # TODO
-            url=""  # TODO
-        ),
-        RTEConfig(
-            additional_features=[],
-            name="rte",
-            description="",  # TODO
-            data_url="https://dl.fbaipublicfiles.com/glue/superglue/data/v2/RTE.zip",
-            citation="",  # TODO
-            url=""
-        ),
-        # paragraph level
-        ScrollsConfig(
-            additional_features=["id"],
-            name="summ_screen_fd_debug",
-            description=_SUMM_SCREEN_DESCRIPTION,
-            data_url="https://huggingface.co/datasets/tau/fs/resolve/main/data/summ_screen_fd_debug.zip",
-            citation=_SUMM_SCREEN_CITATION,
-            url="https://github.com/mingdachen/SummScreen"
-        ),
-        ScrollsConfig(
-            additional_features=["id"],
-            name="gov_report",
-            description=_GOV_REPORT_CITATION,
-            data_url="https://huggingface.co/datasets/tau/fs/resolve/main/data/gov_report.zip",
-            citation=_GOV_REPORT_DESCRIPTION,
-            url="https://gov-report-data.github.io/"
-        ),
         ArxivConfig(
-            additional_features=['section_names', 'sections'],
-            name="arxiv_debug",
-            description=_ARXIV_CITATION,
-            data_url="https://huggingface.co/datasets/tau/fs/resolve/main/data/arxiv_debug.zip",
-            citation=_ARXIV_DESCRIPTION,
-            url="https://github.com/armancohan/long-summarization"
-        ),
+            name="arxiv",
+            data_url="https://fewsion.s3.us-east-2.amazonaws.com/arxiv.zip",
+        )
     ]
 
     def _info(self):
         features = {feature: datasets.Value("string") for feature in self.config.features}
 
         return datasets.DatasetInfo(
-            description=_FS_DESCRIPTION + self.config.description,
+            description="",
             features=datasets.Features(features),
-            homepage=self.config.url,
-            citation=self.config.citation + "\n" + _FS_CITATION,
+            homepage="",
+            citation="",
         )
 
     def _split_generators(self, dl_manager):
@@ -85,31 +105,31 @@ class FS(datasets.GeneratorBasedBuilder):
             datasets.SplitGenerator(
                 name=datasets.Split.TRAIN,
                 gen_kwargs={
-                    "data_file": os.path.join(dl_dir, self.config.train_file),
+                    "data_file": os.path.join(dl_dir, "train.jsonl"),
+                    "split": datasets.Split.TRAIN,
                 },
             ),
             datasets.SplitGenerator(
                 name=datasets.Split.VALIDATION,
                 gen_kwargs={
-                    "data_file": os.path.join(dl_dir, self.config.validation_file),
+                    "data_file": os.path.join(dl_dir, "val.jsonl"),
+                    "split": datasets.Split.VALIDATION,
                 },
             ),
             datasets.SplitGenerator(
                 name=datasets.Split.TEST,
                 gen_kwargs={
-                    "data_file": os.path.join(dl_dir, self.config.test_file),
+                    "data_file": os.path.join(dl_dir, "test.jsonl") if data_files is None else data_files["test"],
+                    "split": datasets.Split.TEST,
                 },
             ),
         ]
 
-    def _generate_examples(self, data_file):
+    def _generate_examples(self, data_file, split):
         with open(data_file, encoding="utf-8") as f:
             for line in f:
                 row = json.loads(line)
-                self.config.process(row)
-                if self.config.target_key not in row:
-                    row[self.config.target_key] = None
-                yield row[self.config.id_key], row
+                yield row[self.config.id_column_name], row
 
 
 def _get_task_name_from_data_url(data_url):

normalize_raw_data/normalize_scrolls.py

@@ -1,26 +0,0 @@
-import os
-import shutil
-from fire import Fire
-from datasets import load_dataset
-from icecream import ic
-
-def normalize_example(example):
-    return {"source": example["input"], "target": example["output"]}
-
-
-def main(dataset_name, num_proc=5, data_dir="../data/"):
-    dataset = load_dataset("tau/scrolls", dataset_name)
-    dataset = dataset.map(normalize_example, num_proc=num_proc, remove_columns=["input", "output"])
-    # ic(dataset_name, dataset["train"][0])
-    dir_name = os.path.join(data_dir, dataset_name)
-    os.makedirs(dir_name, exist_ok=True)
-    for split in dataset:
-        dataset[split].to_json(os.path.join(dir_name, f"{split}.jsonl"))
-    shutil.make_archive(base_name=dir_name,
-                        format='zip',
-                        root_dir=dir_name)
-    shutil.rmtree(dir_name)
-
-
-if __name__ == '__main__':
-    Fire(main)