Name: Rickkk23/LegalPT
Creator: Rickkk23
Published: 2026-05-17 15:03:11
License: 暂无描述

下载链接：

https://hf-mirror.com/datasets/Rickkk23/LegalPT

下载链接

链接失效反馈

官方服务：

资源简介：

--- language: - pt size_categories: - 10M<n<100M task_categories: - text-generation tags: - legal dataset_info: - config_name: all features: - name: id dtype: int64 - name: source dtype: string - name: orig_id dtype: int64 - name: text dtype: string splits: - name: train num_bytes: 135151899572 num_examples: 24194918 download_size: 71423192838 dataset_size: 135151899572 - config_name: acordaos_tcu features: - name: id dtype: int64 - name: text dtype: string - name: meta struct: - name: dedup struct: - name: exact_norm struct: - name: cluster_main_idx dtype: int64 - name: cluster_size dtype: int64 - name: exact_hash_idx dtype: int64 - name: is_duplicate dtype: bool - name: minhash struct: - name: cluster_main_idx dtype: int64 - name: cluster_size dtype: int64 - name: is_duplicate dtype: bool - name: minhash_idx dtype: int64 splits: - name: train num_bytes: 3494790013 num_examples: 634711 download_size: 1653039356 dataset_size: 3494790013 - config_name: datastf features: - name: text dtype: string - name: meta struct: - name: dedup struct: - name: exact_norm struct: - name: cluster_main_idx dtype: int64 - name: cluster_size dtype: int64 - name: exact_hash_idx dtype: int64 - name: is_duplicate dtype: bool - name: minhash struct: - name: cluster_main_idx dtype: int64 - name: cluster_size dtype: int64 - name: is_duplicate dtype: bool - name: minhash_idx dtype: int64 - name: id dtype: int64 splits: - name: train num_bytes: 3699382656 num_examples: 737769 download_size: 1724245648 dataset_size: 3699382656 - config_name: iudicium_textum features: - name: text dtype: string - name: meta struct: - name: dedup struct: - name: exact_norm struct: - name: cluster_main_idx dtype: int64 - name: cluster_size dtype: int64 - name: exact_hash_idx dtype: int64 - name: is_duplicate dtype: bool - name: minhash struct: - name: cluster_main_idx dtype: int64 - name: cluster_size dtype: int64 - name: is_duplicate dtype: bool - name: minhash_idx dtype: int64 - name: id dtype: int64 splits: - name: train num_bytes: 896139675 num_examples: 198387 download_size: 408025309 dataset_size: 896139675 - config_name: mlp_pt_BRCAD-5 features: - name: id dtype: int64 - name: text dtype: string - name: meta struct: - name: dedup struct: - name: exact_norm struct: - name: cluster_main_idx dtype: int64 - name: cluster_size dtype: int64 - name: exact_hash_idx dtype: int64 - name: is_duplicate dtype: bool - name: minhash struct: - name: cluster_main_idx dtype: int64 - name: cluster_size dtype: int64 - name: is_duplicate dtype: bool - name: minhash_idx dtype: int64 splits: - name: train num_bytes: 20311710293 num_examples: 3128292 download_size: 9735599974 dataset_size: 20311710293 - config_name: mlp_pt_CJPG features: - name: id dtype: int64 - name: text dtype: string - name: meta struct: - name: dedup struct: - name: exact_norm struct: - name: cluster_main_idx dtype: int64 - name: cluster_size dtype: int64 - name: exact_hash_idx dtype: int64 - name: is_duplicate dtype: bool - name: minhash struct: - name: cluster_main_idx dtype: int64 - name: cluster_size dtype: int64 - name: is_duplicate dtype: bool - name: minhash_idx dtype: int64 splits: - name: train num_bytes: 63201157801 num_examples: 14068634 download_size: 30473107046 dataset_size: 63201157801 - config_name: mlp_pt_eurlex-caselaw features: - name: id dtype: int64 - name: text dtype: string - name: meta struct: - name: dedup struct: - name: exact_norm struct: - name: cluster_main_idx dtype: int64 - name: cluster_size dtype: int64 - name: exact_hash_idx dtype: int64 - name: is_duplicate dtype: bool - name: minhash struct: - name: cluster_main_idx dtype: int64 - name: cluster_size dtype: int64 - name: is_duplicate dtype: bool - name: minhash_idx dtype: int64 splits: - name: train num_bytes: 1499601545 num_examples: 104312 download_size: 627235870 dataset_size: 1499601545 - config_name: mlp_pt_eurlex-contracts features: - name: id dtype: int64 - name: text dtype: string - name: meta struct: - name: dedup struct: - name: exact_norm struct: - name: cluster_main_idx dtype: int64 - name: cluster_size dtype: int64 - name: exact_hash_idx dtype: int64 - name: is_duplicate dtype: bool - name: minhash struct: - name: cluster_main_idx dtype: int64 - name: cluster_size dtype: int64 - name: is_duplicate dtype: bool - name: minhash_idx dtype: int64 splits: - name: train num_bytes: 467200973 num_examples: 11581 download_size: 112805426 dataset_size: 467200973 - config_name: mlp_pt_eurlex-legislation features: - name: id dtype: int64 - name: text dtype: string - name: meta struct: - name: dedup struct: - name: exact_norm struct: - name: cluster_main_idx dtype: int64 - name: cluster_size dtype: int64 - name: exact_hash_idx dtype: int64 - name: is_duplicate dtype: bool - name: minhash struct: - name: cluster_main_idx dtype: int64 - name: cluster_size dtype: int64 - name: is_duplicate dtype: bool - name: minhash_idx dtype: int64 splits: - name: train num_bytes: 5669271303 num_examples: 232556 download_size: 1384571339 dataset_size: 5669271303 - config_name: mlp_pt_legal-mc4 features: - name: id dtype: int64 - name: text dtype: string - name: meta struct: - name: dedup struct: - name: exact_norm struct: - name: cluster_main_idx dtype: int64 - name: cluster_size dtype: int64 - name: exact_hash_idx dtype: int64 - name: is_duplicate dtype: bool - name: minhash struct: - name: cluster_main_idx dtype: int64 - name: cluster_size dtype: int64 - name: is_duplicate dtype: bool - name: minhash_idx dtype: int64 splits: - name: train num_bytes: 4483889482 num_examples: 191174 download_size: 2250422592 dataset_size: 4483889482 - config_name: parlamento-pt features: - name: id dtype: int64 - name: text dtype: string - name: meta struct: - name: dedup struct: - name: exact_norm struct: - name: cluster_main_idx dtype: int64 - name: cluster_size dtype: int64 - name: exact_hash_idx dtype: int64 - name: is_duplicate dtype: bool - name: minhash struct: - name: cluster_main_idx dtype: int64 - name: cluster_size dtype: int64 - name: is_duplicate dtype: bool - name: minhash_idx dtype: int64 splits: - name: train num_bytes: 2867291543 num_examples: 2670846 download_size: 1319479156 dataset_size: 2867291543 - config_name: tesemo_v2 features: - name: id dtype: int64 - name: text dtype: string - name: meta struct: - name: dedup struct: - name: exact_norm struct: - name: cluster_main_idx dtype: int64 - name: cluster_size dtype: int64 - name: exact_hash_idx dtype: int64 - name: is_duplicate dtype: bool - name: minhash struct: - name: cluster_main_idx dtype: int64 - name: cluster_size dtype: int64 - name: is_duplicate dtype: bool - name: minhash_idx dtype: int64 splits: - name: train num_bytes: 29158221995 num_examples: 2216656 download_size: 13543440397 dataset_size: 29158221995 configs: - config_name: all data_files: - split: train path: all/train-* - config_name: acordaos_tcu data_files: - split: train path: acordaos_tcu/train-* - config_name: datastf data_files: - split: train path: datastf/train-* - config_name: iudicium_textum data_files: - split: train path: iudicium_textum/train-* - config_name: mlp_pt_BRCAD-5 data_files: - split: train path: mlp_pt_BRCAD-5/train-* - config_name: mlp_pt_CJPG data_files: - split: train path: mlp_pt_CJPG/train-* - config_name: mlp_pt_eurlex-caselaw data_files: - split: train path: mlp_pt_eurlex-caselaw/train-* - config_name: mlp_pt_eurlex-contracts data_files: - split: train path: mlp_pt_eurlex-contracts/train-* - config_name: mlp_pt_eurlex-legislation data_files: - split: train path: mlp_pt_eurlex-legislation/train-* - config_name: mlp_pt_legal-mc4 data_files: - split: train path: mlp_pt_legal-mc4/train-* - config_name: parlamento-pt data_files: - split: train path: parlamento-pt/train-* - config_name: tesemo_v2 data_files: - split: train path: tesemo_v2/train-* --- # LegalPT LegalPT aggregates the maximum amount of publicly available legal data in Portuguese, drawing from varied sources including legislation, jurisprudence, legal articles, and government documents. This is the raw version. Deduplicated version is available [here](https://huggingface.co/datasets/eduagarcia/LegalPT_dedup). ## Dataset Details Dataset is composed by six corpora: [Ulysses-Tesemõ](https://github.com/ulysses-camara/ulysses-tesemo), [MultiLegalPile (PT)](https://arxiv.org/abs/2306.02069v2), [ParlamentoPT](http://arxiv.org/abs/2305.06721), [Iudicium Textum](https://www.inf.ufpr.br/didonet/articles/2019_dsw_Iudicium_Textum_Dataset.pdf), [Acordãos TCU](https://link.springer.com/chapter/10.1007/978-3-030-61377-8_46), and [DataSTF](https://legalhackersnatal.wordpress.com/2019/05/09/mais-dados-juridicos/). - [**MultiLegalPile**](https://huggingface.co/datasets/joelniklaus/Multi_Legal_Pile) ([Paper](https://arxiv.org/abs/2306.02069v2)): a multilingual corpus of legal texts comprising 689 GiB of data, covering 24 languages in 17 jurisdictions. The corpus is separated by language, and the subset in Portuguese contains 92GiB of data, containing 13.76 billion words. This subset includes the jurisprudence of the Court of Justice of São Paulo (CJPG), appeals from the [5th Regional Federal Court (BRCAD-5)](https://www.kaggle.com/datasets/eliasjacob/brcad5), the Portuguese subset of legal documents from the European Union, known as [EUR-Lex](https://huggingface.co/datasetsjoelniklaus/eurlex_resources), and a filter for legal documents from [MC4](http://arxiv.org/abs/2010.11934). - [**Ulysses-Tesemõ**](https://github.com/ulysses-camara/ulysses-tesemo): a legal corpus in Brazilian Portuguese, composed of 2.2 million documents, totaling about 26GiB of text obtained from 96 different data sources. These sources encompass legal, legislative, academic papers, news, and related comments. The data was collected through web scraping of government websites. - [**ParlamentoPT**](PORTULAN/parlamento-pt) ([Paper](http://arxiv.org/abs/2305.06721)): a corpus for training language models in European Portuguese. The data was collected from the Portuguese government portal and consists of 2.6 million documents of transcriptions of debates in the Portuguese Parliament. - [**Iudicium Textum**](https://dadosabertos.c3sl.ufpr.br/acordaos/) ([Paper](https://www.inf.ufpr.br/didonet/articles/2019_dsw_Iudicium_Textum_Dataset.pdf)): consists of rulings, votes, and reports from the Supreme Federal Court (STF) of Brazil, published between 2010 and 2018. The dataset contains 1GiB of data extracted from PDFs. - [**Acordãos TCU**](https://www.kaggle.com/datasets/ferraz/acordaos-tcu) ([Paper](https://link.springer.com/chapter/10.1007/978-3-030-61377-8_46)): an open dataset from the Tribunal de Contas da União (Brazilian Federal Court of Accounts), containing 600,000 documents obtained by web scraping government websites. The documents span from 1992 to 2019. - [**DataSTF**](https://legalhackersnatal.wordpress.com/2019/05/09/mais-dados-juridicos/)): a dataset of monocratic decisions from the Superior Court of Justice (STJ) in Brazil, containing 700,000 documents (5GiB of data). ### Dataset Description - **Language(s) (NLP):** Portuguese (pt-BR and pt-PT) - **Repository:** https://github.com/eduagarcia/roberta-legal-portuguese - **Paper:** https://aclanthology.org/2024.propor-1.38/ ## Citation ```bibtex @inproceedings{garcia-etal-2024-robertalexpt, title = "{R}o{BERT}a{L}ex{PT}: A Legal {R}o{BERT}a Model pretrained with deduplication for {P}ortuguese", author = "Garcia, Eduardo A. S. and Silva, Nadia F. F. and Siqueira, Felipe and Albuquerque, Hidelberg O. and Gomes, Juliana R. S. and Souza, Ellen and Lima, Eliomar A.", editor = "Gamallo, Pablo and Claro, Daniela and Teixeira, Ant{\'o}nio and Real, Livy and Garcia, Marcos and Oliveira, Hugo Gon{\c{c}}alo and Amaro, Raquel", booktitle = "Proceedings of the 16th International Conference on Computational Processing of Portuguese", month = mar, year = "2024", address = "Santiago de Compostela, Galicia/Spain", publisher = "Association for Computational Lingustics", url = "https://aclanthology.org/2024.propor-1.38", pages = "374--383", } ``` ## Acknowledgment This work has been supported by the AI Center of Excellence (Centro de Excelência em Inteligência Artificial – CEIA) of the Institute of Informatics at the Federal University of Goiás (INF-UFG).

应用场景：