模型概述
模型特點
模型能力
使用案例
🚀 法律BERT:剛從法學院畢業的“布偶秀明星”
Legal-BERT是一系列用於法律領域的BERT模型,旨在助力法律自然語言處理研究、計算法學以及法律科技應用。為了預訓練不同版本的Legal-BERT,我們從多個公開資源中收集了12GB不同類型的英文法律文本,涵蓋立法、法院判例、合同等多個領域。特定子領域的變體模型(如合同版、歐盟法律版、歐洲人權法院版)和通用的Legal-BERT在特定領域任務中的表現優於直接使用BERT。這是一個輕量級版本的BERT-BASE(大小僅為BERT-BASE的33%),它基於法律數據從頭開始預訓練,在性能上可與更大的模型相媲美,同時效率更高(大約快4倍),對環境的影響也更小。
🚀 快速開始
加載預訓練模型
from transformers import AutoTokenizer, AutoModel
tokenizer = AutoTokenizer.from_pretrained("nlpaueb/legal-bert-small-uncased")
model = AutoModel.from_pretrained("nlpaueb/legal-bert-small-uncased")
將Legal-BERT變體用作語言模型
語料庫 | 模型 | 掩碼標記 | 預測結果 |
---|---|---|---|
BERT-BASE-UNCASED | |||
(合同) | This [MASK] Agreement is between General Motors and John Murray . | employment | ('new', '0.09'), ('current', '0.04'), ('proposed', '0.03'), ('marketing', '0.03'), ('joint', '0.02') |
(歐洲人權法院) | The applicant submitted that her husband was subjected to treatment amounting to [MASK] whilst in the custody of Adana Security Directorate | torture | ('torture', '0.32'), ('rape', '0.22'), ('abuse', '0.14'), ('death', '0.04'), ('violence', '0.03') |
(歐盟法律) | Establishing a system for the identification and registration of [MASK] animals and regarding the labelling of beef and beef products . | bovine | ('farm', '0.25'), ('livestock', '0.08'), ('draft', '0.06'), ('domestic', '0.05'), ('wild', '0.05') |
CONTRACTS-BERT-BASE | |||
(合同) | This [MASK] Agreement is between General Motors and John Murray . | employment | ('letter', '0.38'), ('dealer', '0.04'), ('employment', '0.03'), ('award', '0.03'), ('contribution', '0.02') |
(歐洲人權法院) | The applicant submitted that her husband was subjected to treatment amounting to [MASK] whilst in the custody of Adana Security Directorate | torture | ('death', '0.39'), ('imprisonment', '0.07'), ('contempt', '0.05'), ('being', '0.03'), ('crime', '0.02') |
(歐盟法律) | Establishing a system for the identification and registration of [MASK] animals and regarding the labelling of beef and beef products . | bovine | (('domestic', '0.18'), ('laboratory', '0.07'), ('household', '0.06'), ('personal', '0.06'), ('the', '0.04') |
EURLEX-BERT-BASE | |||
(合同) | This [MASK] Agreement is between General Motors and John Murray . | employment | ('supply', '0.11'), ('cooperation', '0.08'), ('service', '0.07'), ('licence', '0.07'), ('distribution', '0.05') |
(歐洲人權法院) | The applicant submitted that her husband was subjected to treatment amounting to [MASK] whilst in the custody of Adana Security Directorate | torture | ('torture', '0.66'), ('death', '0.07'), ('imprisonment', '0.07'), ('murder', '0.04'), ('rape', '0.02') |
(歐盟法律) | Establishing a system for the identification and registration of [MASK] animals and regarding the labelling of beef and beef products . | bovine | ('live', '0.43'), ('pet', '0.28'), ('certain', '0.05'), ('fur', '0.03'), ('the', '0.02') |
ECHR-BERT-BASE | |||
(合同) | This [MASK] Agreement is between General Motors and John Murray . | employment | ('second', '0.24'), ('latter', '0.10'), ('draft', '0.05'), ('bilateral', '0.05'), ('arbitration', '0.04') |
(歐洲人權法院) | The applicant submitted that her husband was subjected to treatment amounting to [MASK] whilst in the custody of Adana Security Directorate | torture | ('torture', '0.99'), ('death', '0.01'), ('inhuman', '0.00'), ('beating', '0.00'), ('rape', '0.00') |
(歐盟法律) | Establishing a system for the identification and registration of [MASK] animals and regarding the labelling of beef and beef products . | bovine | ('pet', '0.17'), ('all', '0.12'), ('slaughtered', '0.10'), ('domestic', '0.07'), ('individual', '0.05') |
LEGAL-BERT-BASE | |||
(合同) | This [MASK] Agreement is between General Motors and John Murray . | employment | ('settlement', '0.26'), ('letter', '0.23'), ('dealer', '0.04'), ('master', '0.02'), ('supplemental', '0.02') |
(歐洲人權法院) | The applicant submitted that her husband was subjected to treatment amounting to [MASK] whilst in the custody of Adana Security Directorate | torture | ('torture', '1.00'), ('detention', '0.00'), ('arrest', '0.00'), ('rape', '0.00'), ('death', '0.00') |
(歐盟法律) | Establishing a system for the identification and registration of [MASK] animals and regarding the labelling of beef and beef products . | bovine | ('live', '0.67'), ('beef', '0.17'), ('farm', '0.03'), ('pet', '0.02'), ('dairy', '0.01') |
LEGAL-BERT-SMALL | |||
(合同) | This [MASK] Agreement is between General Motors and John Murray . | employment | ('license', '0.09'), ('transition', '0.08'), ('settlement', '0.04'), ('consent', '0.03'), ('letter', '0.03') |
(歐洲人權法院) | The applicant submitted that her husband was subjected to treatment amounting to [MASK] whilst in the custody of Adana Security Directorate | torture | ('torture', '0.59'), ('pain', '0.05'), ('ptsd', '0.05'), ('death', '0.02'), ('tuberculosis', '0.02') |
(歐盟法律) | Establishing a system for the identification and registration of [MASK] animals and regarding the labelling of beef and beef products . | bovine | ('all', '0.08'), ('live', '0.07'), ('certain', '0.07'), ('the', '0.07'), ('farm', '0.05') |
✨ 主要特性
- 專為法律領域設計,基於大規模法律文本數據進行預訓練。
- 有多種變體模型,適用於不同的法律子領域,如合同、歐盟法律、歐洲人權法院判例等。
- 輕量級版本效率高,在性能相當的情況下,速度更快且對環境影響小。
📦 預訓練語料庫
Legal-BERT的預訓練語料庫包括:
- 116,062份歐盟立法文件,可從EURLEX(http://eur-lex.europa.eu)獲取,這是歐盟出版局運營的歐盟法律資源庫。
- 61,826份英國立法文件,可從英國立法門戶(http://www.legislation.gov.uk)獲取。
- 19,867份歐洲法院(ECJ)的判例,同樣可從EURLEX獲取。
- 12,554份歐洲人權法院(ECHR)的判例,來自HUDOC資源庫(http://hudoc.echr.coe.int/eng)。
- 164,141份美國各地法院的判例,託管在Case Law Access Project門戶(https://case.law)。
- 76,366份美國合同,來自美國證券交易委員會(SECOM)的EDGAR數據庫(https://www.sec.gov/edgar.shtml)。
🔧 預訓練細節
- 我們使用Google BERT的GitHub倉庫(https://github.com/google-research/bert)中提供的官方代碼來訓練BERT。
- 我們發佈了一個類似於英文BERT-BASE模型(12層、768隱藏層、12頭、1.1億參數)的模型。
- 我們選擇遵循相同的訓練設置:進行100萬步訓練,批次大小為256個長度為512的序列,初始學習率為1e - 4。
- 我們能夠使用由TensorFlow Research Cloud (TFRC)免費提供的單個Google Cloud TPU v3 - 8,同時還利用了GCP研究信用額度。非常感謝這兩個Google項目對我們的支持!
📚 模型列表
模型名稱 | 模型路徑 | 訓練語料庫 |
---|---|---|
CONTRACTS-BERT-BASE | nlpaueb/bert-base-uncased-contracts |
美國合同 |
EURLEX-BERT-BASE | nlpaueb/bert-base-uncased-eurlex |
歐盟立法 |
ECHR-BERT-BASE | nlpaueb/bert-base-uncased-echr |
歐洲人權法院判例 |
LEGAL-BERT-BASE * | nlpaueb/legal-bert-base-uncased |
全部 |
LEGAL-BERT-SMALL | nlpaueb/legal-bert-small-uncased |
全部 |
* LEGAL-BERT-BASE是Chalkidis等人(2020)中提到的LEGAL-BERT-SC模型;該模型是在下面提到的法律語料庫上從頭開始訓練的,使用了由在相同語料庫上訓練的sentence - piece分詞器創建的新詞彙表。
** 由於很多人對LEGAL-BERT-FP模型(那些基於原始BERT-BASE檢查點的模型)表示感興趣,這些模型已在Archive.org(https://archive.org/details/legal_bert_fp)上發佈,因為這些模型是次要的,可能僅對那些希望深入研究Chalkidis等人(2020)中開放性問題的人有吸引力。
📚 下游任務評估
可參考文章“LEGAL-BERT: The Muppets straight out of Law School”中的實驗。Chalkidis等人,2020年,(https://aclanthology.org/2020.findings-emnlp.261)
📄 許可證
本項目採用CC BY - SA 4.0許可證。
📖 引用
@inproceedings{chalkidis-etal-2020-legal,
title = "{LEGAL}-{BERT}: The Muppets straight out of Law School",
author = "Chalkidis, Ilias and
Fergadiotis, Manos and
Malakasiotis, Prodromos and
Aletras, Nikolaos and
Androutsopoulos, Ion",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2020",
month = nov,
year = "2020",
address = "Online",
publisher = "Association for Computational Linguistics",
doi = "10.18653/v1/2020.findings-emnlp.261",
pages = "2898--2904"
}
👥 關於我們
雅典經濟與商業大學自然語言處理小組致力於開發讓計算機處理和生成自然語言文本的算法、模型和系統。
該小組目前的研究興趣包括:
- 用於數據庫、本體、文檔集合和網絡的問答系統,特別是生物醫學問答。
- 從數據庫和本體,特別是語義網絡本體中進行自然語言生成。
- 文本分類,包括過濾垃圾郵件和濫用內容。
- 信息提取和觀點挖掘,包括法律文本分析和情感分析。
- 用於希臘語的自然語言處理工具,例如解析器和命名實體識別器。
- 自然語言處理中的機器學習,特別是深度學習。
該小組隸屬於雅典經濟與商業大學信息學系信息處理實驗室。
Ilias Chalkidis 代表 雅典經濟與商業大學自然語言處理小組
| GitHub: @ilias.chalkidis | Twitter: @KiddoThe2B |



