Pile of Law BERT large model 2 (uncased)

Pretrained model on English language legal and administrative text using the RoBERTa pretraining objective. This model was trained with the same setup as pile-of-law/legalbert-large-1.7M-1 , but with a different seed.

Model description

Pile of Law BERT large 2 is a transformers model with the BERT large model (uncased) architecture pretrained on the Pile of Law , a dataset consisting of ~256GB of English language legal and administrative text for language model pretraining.

Intended uses & limitations

You can use the raw model for masked language modeling or fine-tune it for a downstream task. Since this model was pretrained on a English language legal and administrative text corpus, legal downstream tasks will likely be more in-domain for this model.

How to use

You can use the model directly with a pipeline for masked language modeling:

>>> from transformers import pipeline
>>> pipe = pipeline(task='fill-mask', model='pile-of-law/legalbert-large-1.7M-2')
>>> pipe("An [MASK] is a request made after a trial by a party that has lost on one or more issues that a higher court review the decision to determine if it was correct.")

[{'sequence': 'an exception is a request made after a trial by a party that has lost on one or more issues that a higher court review the decision to determine if it was correct.', 
  'score': 0.5218929052352905, 
  'token': 4028, 
  'token_str': 'exception'}, 
  {'sequence': 'an appeal is a request made after a trial by a party that has lost on one or more issues that a higher court review the decision to determine if it was correct.', 
  'score': 0.11434809118509293, 
  'token': 1151, 
  'token_str': 'appeal'}, 
  {'sequence': 'an exclusion is a request made after a trial by a party that has lost on one or more issues that a higher court review the decision to determine if it was correct.', 
  'score': 0.06454459577798843, 
  'token': 5345, 
  'token_str': 'exclusion'}, 
  {'sequence': 'an example is a request made after a trial by a party that has lost on one or more issues that a higher court review the decision to determine if it was correct.', 
  'score': 0.043593790382146835, 
  'token': 3677, 
  'token_str': 'example'}, 
  {'sequence': 'an objection is a request made after a trial by a party that has lost on one or more issues that a higher court review the decision to determine if it was correct.', 
  'score': 0.03758585825562477, 
  'token': 3542, 
  'token_str': 'objection'}]

Here is how to use this model to get the features of a given text in PyTorch:

from transformers import BertTokenizer, BertModel
tokenizer = BertTokenizer.from_pretrained('pile-of-law/legalbert-large-1.7M-2')
model = BertModel.from_pretrained('pile-of-law/legalbert-large-1.7M-2')
text = "Replace me by any text you'd like."
encoded_input = tokenizer(text, return_tensors='pt')
output = model(**encoded_input)

and in TensorFlow:

from transformers import BertTokenizer, TFBertModel
tokenizer = BertTokenizer.from_pretrained('pile-of-law/legalbert-large-1.7M-2')
model = TFBertModel.from_pretrained('pile-of-law/legalbert-large-1.7M-2')
text = "Replace me by any text you'd like."
encoded_input = tokenizer(text, return_tensors='tf')
output = model(encoded_input)

Limitations and bias

Please see Appendix G of the Pile of Law paper for copyright limitations related to dataset and model use.

This model can have biased predictions. In the following example where the model is used with a pipeline for masked language modeling, for the race descriptor of the criminal, the model predicts a higher score for "black" than "white".

>>> from transformers import pipeline
>>> pipe = pipeline(task='fill-mask', model='pile-of-law/legalbert-large-1.7M-2')
>>> pipe("The transcript of evidence reveals that at approximately 7:30 a. m. on January 22, 1973, the prosecutrix was awakened in her home in DeKalb County by the barking of the family dog, and as she opened her eyes she saw a [MASK] man standing beside her bed with a gun.", targets=["black", "white"])

[{'sequence': 'the transcript of evidence reveals that at approximately 7 : 30 a. m. on january 22, 1973, the prosecutrix was awakened in her home in dekalb county by the barking of the family dog, and as she opened her eyes she saw a black man standing beside her bed with a gun.', 
  'score': 0.02685137465596199, 
  'token': 4311, 
  'token_str': 'black'}, 
  {'sequence': 'the transcript of evidence reveals that at approximately 7 : 30 a. m. on january 22, 1973, the prosecutrix was awakened in her home in dekalb county by the barking of the family dog, and as she opened her eyes she saw a white man standing beside her bed with a gun.', 
  'score': 0.013632853515446186, 
  'token': 4249, 
  'token_str': 'white'}]

This bias will also affect all fine-tuned versions of this model.

Training data

The Pile of Law BERT large model was pretrained on the Pile of Law, a dataset consisting of ~256GB of English language legal and administrative text for language model pretraining. The Pile of Law consists of 35 data sources, including legal analyses, court opinions and filings, government agency publications, contracts, statutes, regulations, casebooks, etc. We describe the data sources in detail in Appendix E of the Pile of Law paper. The Pile of Law dataset is placed under a CreativeCommons Attribution-NonCommercial-ShareAlike 4.0 International license.

Training procedure

Preprocessing

The model vocabulary consists of 29,000 tokens from a custom word-piece vocabulary fit to Pile of Law using the HuggingFace WordPiece tokenizer and 3,000 randomly sampled legal terms from Black's Law Dictionary, for a vocabulary size of 32,000 tokens. The 80-10-10 masking, corruption, leave split, as in BERT , is used, with a replication rate of 20 to create different masks for each context. To generate sequences, we use the LexNLP sentence segmenter , which handles sentence segmentation for legal citations (which are often falsely mistaken as sentences). The input is formatted by filling sentences until they comprise 256 tokens, followed by a [SEP] token, and then filling sentences such that the entire span is under 512 tokens. If the next sentence in the series is too large, it is not added, and the remaining context length is filled with padding tokens.

Pretraining

The model was trained on a SambaNova cluster, with 8 RDUs, for 1.7 million steps. We used a smaller learning rate of 5e-6 and batch size of 128, to mitigate training instability, potentially due to the diversity of sources in our training data. The masked language modeling (MLM) objective without NSP loss, as described in RoBERTa , was used for pretraining. The model was pretrained with 512 length sequence lengths for all steps.

We trained two models with the same setup in parallel model training runs, with different random seeds. We selected the lowest log likelihood model, pile-of-law/legalbert-large-1.7M-1 , which we refer to as PoL-BERT-Large, for experiments, but also release the second model, pile-of-law/legalbert-large-1.7M-2 .

Evaluation results

See the model card for pile-of-law/legalbert-large-1.7M-1 for finetuning results on the CaseHOLD variant provided by the LexGLUE paper .

BibTeX entry and citation info

@misc{hendersonkrass2022pileoflaw,
  url = {https://arxiv.org/abs/2207.00220},
  author = {Henderson, Peter and Krass, Mark S. and Zheng, Lucia and Guha, Neel and Manning, Christopher D. and Jurafsky, Dan and Ho, Daniel E.},
  title = {Pile of Law: Learning Responsible Data Filtering from the Law and a 256GB Open-Source Legal Dataset},
  publisher = {arXiv},
  year = {2022}
}