Update files from the datasets library (from 1.0.0)

Release notes: https://github.com/huggingface/datasets/releases/tag/1.0.0

.gitattributes

@@ -0,0 +1,27 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bin.* filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zstandard filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text

dataset_infos.json

@@ -0,0 +1 @@
+{"snli_format": {"description": "The SciTail dataset is an entailment dataset created from multiple-choice science exams and web sentences. Each question \nand the correct answer choice are converted into an assertive statement to form the hypothesis. We use information \nretrieval to obtain relevant text from a large text corpus of web sentences, and use these sentences as a premise P. We \ncrowdsource the annotation of such premise-hypothesis pair as supports (entails) or not (neutral), in order to create \nthe SciTail dataset. The dataset contains 27,026 examples with 10,101 examples with entails label and 16,925 examples \nwith neutral label\n", "citation": "inproceedings{scitail,\n     Author = {Tushar Khot and Ashish Sabharwal and Peter Clark},\n     Booktitle = {AAAI},\n     Title = {{SciTail}: A Textual Entailment Dataset from Science Question Answering},\n     Year = {2018}\n}\n", "homepage": "https://allenai.org/data/scitail", "license": "", "features": {"sentence1_binary_parse": {"dtype": "string", "id": null, "_type": "Value"}, "sentence1_parse": {"dtype": "string", "id": null, "_type": "Value"}, "sentence1": {"dtype": "string", "id": null, "_type": "Value"}, "sentence2_parse": {"dtype": "string", "id": null, "_type": "Value"}, "sentence2": {"dtype": "string", "id": null, "_type": "Value"}, "annotator_labels": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "gold_label": {"dtype": "string", "id": null, "_type": "Value"}}, "supervised_keys": null, "builder_name": "scitail", "config_name": "snli_format", "version": {"version_str": "1.1.0", "description": "", "datasets_version_to_prepare": null, "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 22495833, "num_examples": 23596, "dataset_name": "scitail"}, "test": {"name": "test", "num_bytes": 2008631, "num_examples": 2126, "dataset_name": "scitail"}, "validation": {"name": "validation", "num_bytes": 1266529, "num_examples": 1304, "dataset_name": "scitail"}}, "download_checksums": {"http://data.allenai.org.s3.amazonaws.com/downloads/SciTailV1.1.zip": {"num_bytes": 14174621, "checksum": "3fccd37350a94ca280b75998568df85fc2fc62843a3198d644fcbf858e6943d5"}}, "download_size": 14174621, "dataset_size": 25770993, "size_in_bytes": 39945614}, "tsv_format": {"description": "The SciTail dataset is an entailment dataset created from multiple-choice science exams and web sentences. Each question \nand the correct answer choice are converted into an assertive statement to form the hypothesis. We use information \nretrieval to obtain relevant text from a large text corpus of web sentences, and use these sentences as a premise P. We \ncrowdsource the annotation of such premise-hypothesis pair as supports (entails) or not (neutral), in order to create \nthe SciTail dataset. The dataset contains 27,026 examples with 10,101 examples with entails label and 16,925 examples \nwith neutral label\n", "citation": "inproceedings{scitail,\n     Author = {Tushar Khot and Ashish Sabharwal and Peter Clark},\n     Booktitle = {AAAI},\n     Title = {{SciTail}: A Textual Entailment Dataset from Science Question Answering},\n     Year = {2018}\n}\n", "homepage": "https://allenai.org/data/scitail", "license": "", "features": {"premise": {"dtype": "string", "id": null, "_type": "Value"}, "hypothesis": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"dtype": "string", "id": null, "_type": "Value"}}, "supervised_keys": null, "builder_name": "scitail", "config_name": "tsv_format", "version": {"version_str": "1.1.0", "description": "", "datasets_version_to_prepare": null, "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 4618115, "num_examples": 23097, "dataset_name": "scitail"}, "test": {"name": "test", "num_bytes": 411343, "num_examples": 2126, "dataset_name": "scitail"}, "validation": {"name": "validation", "num_bytes": 261086, "num_examples": 1304, "dataset_name": "scitail"}}, "download_checksums": {"http://data.allenai.org.s3.amazonaws.com/downloads/SciTailV1.1.zip": {"num_bytes": 14174621, "checksum": "3fccd37350a94ca280b75998568df85fc2fc62843a3198d644fcbf858e6943d5"}}, "download_size": 14174621, "dataset_size": 5290544, "size_in_bytes": 19465165}, "dgem_format": {"description": "The SciTail dataset is an entailment dataset created from multiple-choice science exams and web sentences. Each question \nand the correct answer choice are converted into an assertive statement to form the hypothesis. We use information \nretrieval to obtain relevant text from a large text corpus of web sentences, and use these sentences as a premise P. We \ncrowdsource the annotation of such premise-hypothesis pair as supports (entails) or not (neutral), in order to create \nthe SciTail dataset. The dataset contains 27,026 examples with 10,101 examples with entails label and 16,925 examples \nwith neutral label\n", "citation": "inproceedings{scitail,\n     Author = {Tushar Khot and Ashish Sabharwal and Peter Clark},\n     Booktitle = {AAAI},\n     Title = {{SciTail}: A Textual Entailment Dataset from Science Question Answering},\n     Year = {2018}\n}\n", "homepage": "https://allenai.org/data/scitail", "license": "", "features": {"premise": {"dtype": "string", "id": null, "_type": "Value"}, "hypothesis": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"dtype": "string", "id": null, "_type": "Value"}, "hypothesis_graph_structure": {"dtype": "string", "id": null, "_type": "Value"}}, "supervised_keys": null, "builder_name": "scitail", "config_name": "dgem_format", "version": {"version_str": "1.1.0", "description": "", "datasets_version_to_prepare": null, "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 6832104, "num_examples": 23088, "dataset_name": "scitail"}, "test": {"name": "test", "num_bytes": 608213, "num_examples": 2126, "dataset_name": "scitail"}, "validation": {"name": "validation", "num_bytes": 394040, "num_examples": 1304, "dataset_name": "scitail"}}, "download_checksums": {"http://data.allenai.org.s3.amazonaws.com/downloads/SciTailV1.1.zip": {"num_bytes": 14174621, "checksum": "3fccd37350a94ca280b75998568df85fc2fc62843a3198d644fcbf858e6943d5"}}, "download_size": 14174621, "dataset_size": 7834357, "size_in_bytes": 22008978}, "predictor_format": {"description": "The SciTail dataset is an entailment dataset created from multiple-choice science exams and web sentences. Each question \nand the correct answer choice are converted into an assertive statement to form the hypothesis. We use information \nretrieval to obtain relevant text from a large text corpus of web sentences, and use these sentences as a premise P. We \ncrowdsource the annotation of such premise-hypothesis pair as supports (entails) or not (neutral), in order to create \nthe SciTail dataset. The dataset contains 27,026 examples with 10,101 examples with entails label and 16,925 examples \nwith neutral label\n", "citation": "inproceedings{scitail,\n     Author = {Tushar Khot and Ashish Sabharwal and Peter Clark},\n     Booktitle = {AAAI},\n     Title = {{SciTail}: A Textual Entailment Dataset from Science Question Answering},\n     Year = {2018}\n}\n", "homepage": "https://allenai.org/data/scitail", "license": "", "features": {"answer": {"dtype": "string", "id": null, "_type": "Value"}, "sentence2_structure": {"dtype": "string", "id": null, "_type": "Value"}, "sentence1": {"dtype": "string", "id": null, "_type": "Value"}, "sentence2": {"dtype": "string", "id": null, "_type": "Value"}, "gold_label": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}}, "supervised_keys": null, "builder_name": "scitail", "config_name": "predictor_format", "version": {"version_str": "1.1.0", "description": "", "datasets_version_to_prepare": null, "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 8884823, "num_examples": 23587, "dataset_name": "scitail"}, "test": {"name": "test", "num_bytes": 797161, "num_examples": 2126, "dataset_name": "scitail"}, "validation": {"name": "validation", "num_bytes": 511305, "num_examples": 1304, "dataset_name": "scitail"}}, "download_checksums": {"http://data.allenai.org.s3.amazonaws.com/downloads/SciTailV1.1.zip": {"num_bytes": 14174621, "checksum": "3fccd37350a94ca280b75998568df85fc2fc62843a3198d644fcbf858e6943d5"}}, "download_size": 14174621, "dataset_size": 10193289, "size_in_bytes": 24367910}}

dummy/dgem_format/1.1.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9780d78e6912f6c0eaff9c195ff2b436ab51b16c2ce701518f797b36a8ef0726
+size 2335

dummy/predictor_format/1.1.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9f6ef64a1229f7fda023409e0aac23bdbedfb031659ac9483afda7ad5aa8a1fa
+size 2618

dummy/snli_format/1.1.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dd9ec34873ade4d199c1e6cb4bf6b7517c919019fbb374bd3af456dfd121c41a
+size 4903

dummy/tsv_format/1.1.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:02ae4a4663505680821a2a09264e7a06294eb7aaf16e91ec8c38ca29a82b4015
+size 2834

scitail.py

@@ -0,0 +1,299 @@
+"""TODO(sciTail): Add a description here."""
+
+from __future__ import absolute_import, division, print_function
+
+import csv
+import json
+import os
+import textwrap
+
+import datasets
+
+
+# TODO(sciTail): BibTeX citation
+_CITATION = """\
+inproceedings{scitail,
+     Author = {Tushar Khot and Ashish Sabharwal and Peter Clark},
+     Booktitle = {AAAI},
+     Title = {{SciTail}: A Textual Entailment Dataset from Science Question Answering},
+     Year = {2018}
+}
+"""
+
+# TODO(sciTail):
+_DESCRIPTION = """\
+The SciTail dataset is an entailment dataset created from multiple-choice science exams and web sentences. Each question
+and the correct answer choice are converted into an assertive statement to form the hypothesis. We use information
+retrieval to obtain relevant text from a large text corpus of web sentences, and use these sentences as a premise P. We
+crowdsource the annotation of such premise-hypothesis pair as supports (entails) or not (neutral), in order to create
+the SciTail dataset. The dataset contains 27,026 examples with 10,101 examples with entails label and 16,925 examples
+with neutral label
+"""
+
+_URL = "http://data.allenai.org.s3.amazonaws.com/downloads/SciTailV1.1.zip"
+
+
+class ScitailConfig(datasets.BuilderConfig):
+
+    """ BuilderConfig for Xquad"""
+
+    def __init__(self, **kwargs):
+        """
+
+        Args:
+            **kwargs: keyword arguments forwarded to super.
+        """
+        super(ScitailConfig, self).__init__(version=datasets.Version("1.1.0", ""), **kwargs)
+
+
+class Scitail(datasets.GeneratorBasedBuilder):
+    """TODO(sciTail): Short description of my dataset."""
+
+    # TODO(sciTail): Set up version.
+    VERSION = datasets.Version("1.1.0")
+    BUILDER_CONFIGS = [
+        ScitailConfig(
+            name="snli_format",
+            description="JSONL format used by SNLI with a JSON object corresponding to each entailment example in each line.",
+        ),
+        ScitailConfig(
+            name="tsv_format", description="Tab-separated format with three columns: premise hypothesis label"
+        ),
+        ScitailConfig(
+            name="dgem_format",
+            description="Tab-separated format used by the DGEM model: premise hypothesis label hypothesis graph structure",
+        ),
+        ScitailConfig(
+            name="predictor_format",
+            description=textwrap.dedent(
+                """\
+          AllenNLP predictors work only with JSONL format. This folder contains the SciTail train/dev/test in JSONL format
+        so that it can be loaded into the predictors. Each line is a JSON object with the following keys:
+        gold_label : the example label from {entails, neutral}
+        sentence1: the premise
+        sentence2: the hypothesis
+        sentence2_structure: structure from the hypothesis """
+            ),
+        ),
+    ]
+
+    def _info(self):
+        # TODO(sciTail): Specifies the datasets.DatasetInfo object
+        if self.config.name == "snli_format":
+            return datasets.DatasetInfo(
+                # This is the description that will appear on the datasets page.
+                description=_DESCRIPTION,
+                # datasets.features.FeatureConnectors
+                features=datasets.Features(
+                    {
+                        "sentence1_binary_parse": datasets.Value("string"),
+                        "sentence1_parse": datasets.Value("string"),
+                        "sentence1": datasets.Value("string"),
+                        "sentence2_parse": datasets.Value("string"),
+                        "sentence2": datasets.Value("string"),
+                        "annotator_labels": datasets.features.Sequence(datasets.Value("string")),
+                        "gold_label": datasets.Value("string")
+                        # These are the features of your dataset like images, labels ...
+                    }
+                ),
+                # If there's a common (input, target) tuple from the features,
+                # specify them here. They'll be used if as_supervised=True in
+                # builder.as_dataset.
+                supervised_keys=None,
+                # Homepage of the dataset for documentation
+                homepage="https://allenai.org/data/scitail",
+                citation=_CITATION,
+            )
+        elif self.config.name == "tsv_format":
+            return datasets.DatasetInfo(
+                # This is the description that will appear on the datasets page.
+                description=_DESCRIPTION,
+                # datasets.features.FeatureConnectors
+                features=datasets.Features(
+                    {
+                        "premise": datasets.Value("string"),
+                        "hypothesis": datasets.Value("string"),
+                        "label": datasets.Value("string")
+                        # These are the features of your dataset like images, labels ...
+                    }
+                ),
+                # If there's a common (input, target) tuple from the features,
+                # specify them here. They'll be used if as_supervised=True in
+                # builder.as_dataset.
+                supervised_keys=None,
+                # Homepage of the dataset for documentation
+                homepage="https://allenai.org/data/scitail",
+                citation=_CITATION,
+            )
+        elif self.config.name == "predictor_format":
+            return datasets.DatasetInfo(
+                # This is the description that will appear on the datasets page.
+                description=_DESCRIPTION,
+                # datasets.features.FeatureConnectors
+                features=datasets.Features(
+                    {
+                        "answer": datasets.Value("string"),
+                        "sentence2_structure": datasets.Value("string"),
+                        "sentence1": datasets.Value("string"),
+                        "sentence2": datasets.Value("string"),
+                        "gold_label": datasets.Value("string"),
+                        "question": datasets.Value("string")
+                        # These are the features of your dataset like images, labels ...
+                    }
+                ),
+                # If there's a common (input, target) tuple from the features,
+                # specify them here. They'll be used if as_supervised=True in
+                # builder.as_dataset.
+                supervised_keys=None,
+                # Homepage of the dataset for documentation
+                homepage="https://allenai.org/data/scitail",
+                citation=_CITATION,
+            )
+        elif self.config.name == "dgem_format":
+            return datasets.DatasetInfo(
+                # This is the description that will appear on the datasets page.
+                description=_DESCRIPTION,
+                # datasets.features.FeatureConnectors
+                features=datasets.Features(
+                    {
+                        "premise": datasets.Value("string"),
+                        "hypothesis": datasets.Value("string"),
+                        "label": datasets.Value("string"),
+                        "hypothesis_graph_structure": datasets.Value("string")
+                        # These are the features of your dataset like images, labels ...
+                    }
+                ),
+                # If there's a common (input, target) tuple from the features,
+                # specify them here. They'll be used if as_supervised=True in
+                # builder.as_dataset.
+                supervised_keys=None,
+                # Homepage of the dataset for documentation
+                homepage="https://allenai.org/data/scitail",
+                citation=_CITATION,
+            )
+
+    def _split_generators(self, dl_manager):
+        """Returns SplitGenerators."""
+        # TODO(sciTail): Downloads the data and defines the splits
+        # dl_manager is a datasets.download.DownloadManager that can be used to
+        # download and extract URLs
+        dl_dir = dl_manager.download_and_extract(_URL)
+        data_dir = os.path.join(dl_dir, "SciTailV1.1")
+        snli = os.path.join(data_dir, "snli_format")
+        dgem = os.path.join(data_dir, "dgem_format")
+        tsv = os.path.join(data_dir, "tsv_format")
+        predictor = os.path.join(data_dir, "predictor_format")
+        if self.config.name == "snli_format":
+            return [
+                datasets.SplitGenerator(
+                    name=datasets.Split.TRAIN,
+                    # These kwargs will be passed to _generate_examples
+                    gen_kwargs={"filepath": os.path.join(snli, "scitail_1.0_train.txt")},
+                ),
+                datasets.SplitGenerator(
+                    name=datasets.Split.TEST,
+                    # These kwargs will be passed to _generate_examples
+                    gen_kwargs={"filepath": os.path.join(snli, "scitail_1.0_test.txt")},
+                ),
+                datasets.SplitGenerator(
+                    name=datasets.Split.VALIDATION,
+                    # These kwargs will be passed to _generate_examples
+                    gen_kwargs={"filepath": os.path.join(snli, "scitail_1.0_dev.txt")},
+                ),
+            ]
+        elif self.config.name == "tsv_format":
+            return [
+                datasets.SplitGenerator(
+                    name=datasets.Split.TRAIN,
+                    # These kwargs will be passed to _generate_examples
+                    gen_kwargs={"filepath": os.path.join(tsv, "scitail_1.0_train.tsv")},
+                ),
+                datasets.SplitGenerator(
+                    name=datasets.Split.TEST,
+                    # These kwargs will be passed to _generate_examples
+                    gen_kwargs={"filepath": os.path.join(tsv, "scitail_1.0_test.tsv")},
+                ),
+                datasets.SplitGenerator(
+                    name=datasets.Split.VALIDATION,
+                    # These kwargs will be passed to _generate_examples
+                    gen_kwargs={"filepath": os.path.join(tsv, "scitail_1.0_dev.tsv")},
+                ),
+            ]
+        elif self.config.name == "predictor_format":
+            return [
+                datasets.SplitGenerator(
+                    name=datasets.Split.TRAIN,
+                    # These kwargs will be passed to _generate_examples
+                    gen_kwargs={"filepath": os.path.join(predictor, "scitail_1.0_structure_train.jsonl")},
+                ),
+                datasets.SplitGenerator(
+                    name=datasets.Split.TEST,
+                    # These kwargs will be passed to _generate_examples
+                    gen_kwargs={"filepath": os.path.join(predictor, "scitail_1.0_structure_test.jsonl")},
+                ),
+                datasets.SplitGenerator(
+                    name=datasets.Split.VALIDATION,
+                    # These kwargs will be passed to _generate_examples
+                    gen_kwargs={"filepath": os.path.join(predictor, "scitail_1.0_structure_dev.jsonl")},
+                ),
+            ]
+        elif self.config.name == "dgem_format":
+            return [
+                datasets.SplitGenerator(
+                    name=datasets.Split.TRAIN,
+                    # These kwargs will be passed to _generate_examples
+                    gen_kwargs={"filepath": os.path.join(dgem, "scitail_1.0_structure_train.tsv")},
+                ),
+                datasets.SplitGenerator(
+                    name=datasets.Split.TEST,
+                    # These kwargs will be passed to _generate_examples
+                    gen_kwargs={"filepath": os.path.join(dgem, "scitail_1.0_structure_test.tsv")},
+                ),
+                datasets.SplitGenerator(
+                    name=datasets.Split.VALIDATION,
+                    # These kwargs will be passed to _generate_examples
+                    gen_kwargs={"filepath": os.path.join(dgem, "scitail_1.0_structure_dev.tsv")},
+                ),
+            ]
+
+    def _generate_examples(self, filepath):
+        """Yields examples."""
+        # TODO(sciTail): Yields (key, example) tuples from the dataset
+        with open(filepath, encoding="utf-8") as f:
+            if self.config.name == "snli_format":
+                for id_, row in enumerate(f):
+                    data = json.loads(row)
+
+                    yield id_, {
+                        "sentence1_binary_parse": data["sentence1_binary_parse"],
+                        "sentence1_parse": data["sentence1_parse"],
+                        "sentence1": data["sentence1"],
+                        "sentence2_parse": data["sentence2_parse"],
+                        "sentence2": data["sentence2"],
+                        "annotator_labels": data["annotator_labels"],
+                        "gold_label": data["gold_label"],
+                    }
+            elif self.config.name == "tsv_format":
+                data = csv.reader(f, delimiter="\t")
+                for id_, row in enumerate(data):
+                    yield id_, {"premise": row[0], "hypothesis": row[1], "label": row[2]}
+            elif self.config.name == "dgem_format":
+                data = csv.reader(f, delimiter="\t")
+                for id_, row in enumerate(data):
+                    yield id_, {
+                        "premise": row[0],
+                        "hypothesis": row[1],
+                        "label": row[2],
+                        "hypothesis_graph_structure": row[3],
+                    }
+            elif self.config.name == "predictor_format":
+                for id_, row in enumerate(f):
+                    data = json.loads(row)
+                    yield id_, {
+                        "answer": data["answer"],
+                        "sentence2_structure": data["sentence2_structure"],
+                        "sentence1": data["sentence1"],
+                        "sentence2": data["sentence2"],
+                        "gold_label": data["gold_label"],
+                        "question": data["question"],
+                    }