added cuad dataset

dataset_infos.json

@@ -1 +1 @@
-{"german_ler": {"description": "description", "citation": "@inproceedings{leitner2019fine,\nauthor = {Elena Leitner and Georg Rehm and Julian Moreno-Schneider},\ntitle = {{Fine-grained Named Entity Recognition in Legal Documents}},\nbooktitle = {Semantic Systems. The Power of AI and Knowledge\n  Graphs. Proceedings of the 15th International Conference\n  (SEMANTiCS 2019)},\nyear = 2019,\neditor = {Maribel Acosta and Philippe Cudr\u00e9-Mauroux and Maria\n  Maleshkova and Tassilo Pellegrini and Harald Sack and York\n  Sure-Vetter},\nkeywords = {aip},\npublisher = {Springer},\nseries = {Lecture Notes in Computer Science},\nnumber = {11702},\naddress = {Karlsruhe, Germany},\nmonth = 9,\nnote = {10/11 September 2019},\npages = {272--287},\npdf = {https://link.springer.com/content/pdf/10.1007%2F978-3-030-33220-4_20.pdf}}\n", "homepage": "https://github.com/elenanereiss/Legal-Entity-Recognition", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 39, "names": ["B-AN", "B-EUN", "B-GRT", "B-GS", "B-INN", "B-LD", "B-LDS", "B-LIT", "B-MRK", "B-ORG", "B-PER", "B-RR", "B-RS", "B-ST", "B-STR", "B-UN", "B-VO", "B-VS", "B-VT", "I-AN", "I-EUN", "I-GRT", "I-GS", "I-INN", "I-LD", "I-LDS", "I-LIT", "I-MRK", "I-ORG", "I-PER", "I-RR", "I-RS", "I-ST", "I-STR", "I-UN", "I-VO", "I-VS", "I-VT", "O"], "names_file": null, "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "legal_glue", "config_name": "german_ler", "version": {"version_str": "1.1.0", "description": "", "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 38853928, "num_examples": 66723, "dataset_name": "legal_glue"}}, "download_checksums": {"https://raw.githubusercontent.com/elenanereiss/Legal-Entity-Recognition/master/data/dataset_courts.zip": {"num_bytes": 4392913, "checksum": "f0427df5fb8bfdefe5228bc0fa0e75e9cfa782d1a78e32582cce096473c88567"}}, "download_size": 4392913, "post_processing_size": null, "dataset_size": 38853928, "size_in_bytes": 43246841}, "lener_br": {"description": "LeNER-Br is a Portuguese language dataset for named entity recognition\napplied to legal documents. LeNER-Br consists entirely of manually annotated\nlegislation and legal cases texts and contains tags for persons, locations,\ntime entities, organizations, legislation and legal cases.\nTo compose the dataset, 66 legal documents from several Brazilian Courts were\ncollected. Courts of superior and state levels were considered, such as Supremo\nTribunal Federal, Superior Tribunal de Justi\u00e7a, Tribunal de Justi\u00e7a de Minas\nGerais and Tribunal de Contas da Uni\u00e3o. In addition, four legislation documents\nwere collected, such as \"Lei Maria da Penha\", giving a total of 70 documents\n", "citation": "@inproceedings{luz_etal_propor2018,\nauthor = {Pedro H. {Luz de Araujo} and Te'{o}filo E. {de Campos} and\nRenato R. R. {de Oliveira} and Matheus Stauffer and\nSamuel Couto and Paulo Bermejo},\ntitle = {{LeNER-Br}: a Dataset for Named Entity Recognition in {Brazilian} Legal Text},\nbooktitle = {International Conference on the Computational Processing of Portuguese ({PROPOR})},\npublisher = {Springer},\nseries = {Lecture Notes on Computer Science ({LNCS})},\npages = {313--323},\nyear = {2018},\nmonth = {September 24-26},\naddress = {Canela, RS, Brazil},\ndoi = {10.1007/978-3-319-99722-3_32},\nurl = {https://cic.unb.br/~teodecampos/LeNER-Br/},\n}\n", "homepage": "https://cic.unb.br/~teodecampos/LeNER-Br/", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 13, "names": ["O", "B-ORGANIZACAO", "I-ORGANIZACAO", "B-PESSOA", "I-PESSOA", "B-TEMPO", "I-TEMPO", "B-LOCAL", "I-LOCAL", "B-LEGISLACAO", "I-LEGISLACAO", "B-JURISPRUDENCIA", "I-JURISPRUDENCIA"], "names_file": null, "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "legal_glue", "config_name": "lener_br", "version": {"version_str": "1.1.0", "description": "", "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 7153474101, "num_examples": 7828, "dataset_name": "legal_glue"}, "test": {"name": "test", "num_bytes": 195782089, "num_examples": 1177, "dataset_name": "legal_glue"}, "validation": {"name": "validation", "num_bytes": 280965196, "num_examples": 1390, "dataset_name": "legal_glue"}}, "download_checksums": {"https://github.com/peluz/lener-br/raw/master/leNER-Br/train/train.conll": {"num_bytes": 2142199, "checksum": "6fdf9066333c84565f9e3d28ee8f0f519336bece69b63f8d78b8de0fe96dcd47"}, "https://github.com/peluz/lener-br/raw/master/leNER-Br/test/test.conll": {"num_bytes": 438441, "checksum": "f90cd26a31afc2d1f132c4473d40c26d2283a98b374025fa5b5985b723dce825"}, "https://github.com/peluz/lener-br/raw/master/leNER-Br/dev/dev.conll": {"num_bytes": 402497, "checksum": "7e350feb828198031e57c21d6aadbf8dac92b19a684e45d7081c6cb491e2063b"}}, "download_size": 2983137, "post_processing_size": null, "dataset_size": 7630221386, "size_in_bytes": 7633204523}}
+{"german_ler": {"description": "description", "citation": "@inproceedings{leitner2019fine,\nauthor = {Elena Leitner and Georg Rehm and Julian Moreno-Schneider},\ntitle = {{Fine-grained Named Entity Recognition in Legal Documents}},\nbooktitle = {Semantic Systems. The Power of AI and Knowledge\n  Graphs. Proceedings of the 15th International Conference\n  (SEMANTiCS 2019)},\nyear = 2019,\neditor = {Maribel Acosta and Philippe Cudr\u00e9-Mauroux and Maria\n  Maleshkova and Tassilo Pellegrini and Harald Sack and York\n  Sure-Vetter},\nkeywords = {aip},\npublisher = {Springer},\nseries = {Lecture Notes in Computer Science},\nnumber = {11702},\naddress = {Karlsruhe, Germany},\nmonth = 9,\nnote = {10/11 September 2019},\npages = {272--287},\npdf = {https://link.springer.com/content/pdf/10.1007%2F978-3-030-33220-4_20.pdf}}\n", "homepage": "https://github.com/elenanereiss/Legal-Entity-Recognition", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 39, "names": ["B-AN", "B-EUN", "B-GRT", "B-GS", "B-INN", "B-LD", "B-LDS", "B-LIT", "B-MRK", "B-ORG", "B-PER", "B-RR", "B-RS", "B-ST", "B-STR", "B-UN", "B-VO", "B-VS", "B-VT", "I-AN", "I-EUN", "I-GRT", "I-GS", "I-INN", "I-LD", "I-LDS", "I-LIT", "I-MRK", "I-ORG", "I-PER", "I-RR", "I-RS", "I-ST", "I-STR", "I-UN", "I-VO", "I-VS", "I-VT", "O"], "names_file": null, "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "legal_glue", "config_name": "german_ler", "version": {"version_str": "1.1.0", "description": "", "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 38853928, "num_examples": 66723, "dataset_name": "legal_glue"}}, "download_checksums": {"https://raw.githubusercontent.com/elenanereiss/Legal-Entity-Recognition/master/data/dataset_courts.zip": {"num_bytes": 4392913, "checksum": "f0427df5fb8bfdefe5228bc0fa0e75e9cfa782d1a78e32582cce096473c88567"}}, "download_size": 4392913, "post_processing_size": null, "dataset_size": 38853928, "size_in_bytes": 43246841}, "lener_br": {"description": "LeNER-Br is a Portuguese language dataset for named entity recognition\napplied to legal documents. LeNER-Br consists entirely of manually annotated\nlegislation and legal cases texts and contains tags for persons, locations,\ntime entities, organizations, legislation and legal cases.\nTo compose the dataset, 66 legal documents from several Brazilian Courts were\ncollected. Courts of superior and state levels were considered, such as Supremo\nTribunal Federal, Superior Tribunal de Justi\u00e7a, Tribunal de Justi\u00e7a de Minas\nGerais and Tribunal de Contas da Uni\u00e3o. In addition, four legislation documents\nwere collected, such as \"Lei Maria da Penha\", giving a total of 70 documents\n", "citation": "@inproceedings{luz_etal_propor2018,\nauthor = {Pedro H. {Luz de Araujo} and Te'{o}filo E. {de Campos} and\nRenato R. R. {de Oliveira} and Matheus Stauffer and\nSamuel Couto and Paulo Bermejo},\ntitle = {{LeNER-Br}: a Dataset for Named Entity Recognition in {Brazilian} Legal Text},\nbooktitle = {International Conference on the Computational Processing of Portuguese ({PROPOR})},\npublisher = {Springer},\nseries = {Lecture Notes on Computer Science ({LNCS})},\npages = {313--323},\nyear = {2018},\nmonth = {September 24-26},\naddress = {Canela, RS, Brazil},\ndoi = {10.1007/978-3-319-99722-3_32},\nurl = {https://cic.unb.br/~teodecampos/LeNER-Br/},\n}\n", "homepage": "https://cic.unb.br/~teodecampos/LeNER-Br/", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 13, "names": ["O", "B-ORGANIZACAO", "I-ORGANIZACAO", "B-PESSOA", "I-PESSOA", "B-TEMPO", "I-TEMPO", "B-LOCAL", "I-LOCAL", "B-LEGISLACAO", "I-LEGISLACAO", "B-JURISPRUDENCIA", "I-JURISPRUDENCIA"], "names_file": null, "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "legal_glue", "config_name": "lener_br", "version": {"version_str": "1.1.0", "description": "", "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 7153474101, "num_examples": 7828, "dataset_name": "legal_glue"}, "test": {"name": "test", "num_bytes": 195782089, "num_examples": 1177, "dataset_name": "legal_glue"}, "validation": {"name": "validation", "num_bytes": 280965196, "num_examples": 1390, "dataset_name": "legal_glue"}}, "download_checksums": {"https://github.com/peluz/lener-br/raw/master/leNER-Br/train/train.conll": {"num_bytes": 2142199, "checksum": "6fdf9066333c84565f9e3d28ee8f0f519336bece69b63f8d78b8de0fe96dcd47"}, "https://github.com/peluz/lener-br/raw/master/leNER-Br/test/test.conll": {"num_bytes": 438441, "checksum": "f90cd26a31afc2d1f132c4473d40c26d2283a98b374025fa5b5985b723dce825"}, "https://github.com/peluz/lener-br/raw/master/leNER-Br/dev/dev.conll": {"num_bytes": 402497, "checksum": "7e350feb828198031e57c21d6aadbf8dac92b19a684e45d7081c6cb491e2063b"}}, "download_size": 2983137, "post_processing_size": null, "dataset_size": 7630221386, "size_in_bytes": 7633204523}, "cuad": {"description": "Contract Understanding Atticus Dataset (CUAD) v1 is a corpus of more than 13,000 labels in 510\ncommercial legal contracts that have been manually labeled to identify 41 categories of important\nclauses that lawyers look for when reviewing contracts in connection with corporate transactions.\n", "citation": "@article{hendrycks2021cuad,\ntitle={CUAD: An Expert-Annotated NLP Dataset for Legal Contract Review},\nauthor={Dan Hendrycks and Collin Burns and Anya Chen and Spencer Ball},\njournal={arXiv preprint arXiv:2103.06268},\nyear={2021}\n}\n", "homepage": "https://www.atticusprojectai.org/cuad", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"text": {"dtype": "string", "id": null, "_type": "Value"}, "answer_start": {"dtype": "int32", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "legal_glue", "config_name": "cuad", "version": {"version_str": "1.1.0", "description": "", "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 1466037640, "num_examples": 22450, "dataset_name": "legal_glue"}, "test": {"name": "test", "num_bytes": 198543467, "num_examples": 4182, "dataset_name": "legal_glue"}}, "download_checksums": {"https://github.com/TheAtticusProject/cuad/raw/main/data.zip": {"num_bytes": 18309308, "checksum": "f8161d18bea4e9c05e78fa6dda61c19c846fb8087ea969c172753bc2f45b999a"}}, "download_size": 18309308, "post_processing_size": null, "dataset_size": 1664581107, "size_in_bytes": 1682890415}}

legalglue.py

@@ -181,6 +181,29 @@ class LegalGLUE(datasets.GeneratorBasedBuilder):
                 url = {https://cic.unb.br/~teodecampos/LeNER-Br/},
                 }
             """)
+        ),
+        LegalGlueConfig(
+            name="cuad",
+            description=textwrap.dedent(
+            """\
+            Contract Understanding Atticus Dataset (CUAD) v1 is a corpus of more than 13,000 labels in 510
+            commercial legal contracts that have been manually labeled to identify 41 categories of important
+            clauses that lawyers look for when reviewing contracts in connection with corporate transactions.
+            """
+            ),
+            label_classes=None,
+            multi_label=False,
+            data_url="https://github.com/TheAtticusProject/cuad/raw/main/data.zip",
+            data_files=["train_separate_questions.json", "test.json"],
+            homepage="https://www.atticusprojectai.org/cuad",
+            citation=textwrap.dedent("""\
+            @article{hendrycks2021cuad,
+            title={CUAD: An Expert-Annotated NLP Dataset for Legal Contract Review},
+            author={Dan Hendrycks and Collin Burns and Anya Chen and Spencer Ball},
+            journal={arXiv preprint arXiv:2103.06268},
+            year={2021}
+            }
+            """)
         )
     ]
 
@@ -205,6 +228,22 @@ class LegalGLUE(datasets.GeneratorBasedBuilder):
                     )
                 )
             }
+        elif self.config.name == "cuad":
+            features = datasets.Features(
+                {
+                    "id": datasets.Value("string"),
+                    "title": datasets.Value("string"),
+                    "context": datasets.Value("string"),
+                    "question": datasets.Value("string"),
+                    "answers": datasets.features.Sequence(
+                        {
+                            "text": datasets.Value("string"),
+                            "answer_start": datasets.Value("int32"),
+                        }
+                    ),
+                }
+            )
+
         return datasets.DatasetInfo(
             description=self.config.description,
             features=datasets.Features(features),
@@ -222,7 +261,7 @@ class LegalGLUE(datasets.GeneratorBasedBuilder):
                 gen_kwargs={
                     "filepath": self.config.data_files,
                     "split": "train",
-                    "files": [os.path.join(archive,file) for file in self.config.data_files]#dl_manager.iter_archive(archive),
+                    "files": [os.path.join(archive,file) for file in self.config.data_files],
                 },
             )]
         elif self.config.name == "lener_br":
@@ -261,6 +300,28 @@ class LegalGLUE(datasets.GeneratorBasedBuilder):
                         },
                     ),
                 ]
+        elif self.config.name == "cuad":
+            archive = dl_manager.download_and_extract(self.config.data_url)
+            return [
+                datasets.SplitGenerator(
+                    name=datasets.Split.TRAIN,
+                    # These kwargs will be passed to _generate_examples
+                    gen_kwargs={
+                        "filepath": self.config.data_files ,
+                        "split": "train",
+                        "files": os.path.join(archive, self.config.data_files[0])
+                        },
+                ),
+                datasets.SplitGenerator(
+                    name=datasets.Split.TEST,
+                    # These kwargs will be passed to _generate_examples
+                    gen_kwargs={
+                        "filepath": self.config.data_files,
+                        "split": "test",
+                        "files":  os.path.join(archive, self.config.data_files[1])
+                        },
+                ),
+            ]
 
         # else:
         #     return [
@@ -324,6 +385,7 @@ class LegalGLUE(datasets.GeneratorBasedBuilder):
                         "tokens": tokens,
                         "ner_tags": ner_tags,
                         }
+
         elif self.config.name == "lener_br":
             with open(files, encoding="utf-8") as f:
                 id = 0
@@ -350,3 +412,28 @@ class LegalGLUE(datasets.GeneratorBasedBuilder):
                     "tokens": tokens,
                     "ner_tags": ner_tags,
                     }
+
+        elif self.config.name == "cuad":
+            with open(files, encoding="utf-8") as f:
+                cuad = json.load(f)
+                for example in cuad["data"]:
+                    title = example.get("title", "").strip()
+                    for paragraph in example["paragraphs"]:
+                        context = paragraph["context"].strip()
+                        for qa in paragraph["qas"]:
+                            question = qa["question"].strip()
+                            id_ = qa["id"]
+
+                            answer_starts = [answer["answer_start"] for answer in qa["answers"]]
+                            answers = [answer["text"].strip() for answer in qa["answers"]]
+
+                            yield id_, {
+                                "title": title,
+                                "context": context,
+                                "question": question,
+                                "id": id_,
+                                "answers": {
+                                    "answer_start": answer_starts,
+                                    "text": answers,
+                                },
+                            }