LEGAL-BERT: The Muppets Straight Out of Law School

ABSTRACT: “BERT has achieved impressive performance in several NLP tasks. However, there has been limited investigation on its adaptation guidelines in specialised domains. Here we focus on the legal domain, where we explore several approaches for applying BERT models to downstream legal tasks, evaluating on multiple datasets. Our findings indicate that the previous guidelines for pre-training and fine-tuning, often blindly followed, do not always generalize well in the legal domain. Thus we propose a systematic investigation of the available strategies when applying BERT in specialised domains. These are: (a) use the original BERT out of the box, (b) adapt BERT by additional pre-training on domain-specific corpora, and (c) pre-train BERT from scratch on domain-specific corpora. We also propose a broader hyper-parameter search space when fine-tuning for downstream tasks and we release LEGAL-BERT, a family of BERT models intended to assist legal NLP research, computational law, and legal technology applications.”

Congrats to all of the authors on their acceptance in the Empirical Methods in Natural Language Processing Conference in November.

In the legal scientific community, we are witnessing increasing efforts to connect general purpose NLP Advances to domain specific applications within law. First, we saw Word Embeddings (i.e. word2Vec, etc.) now Transformers (i.e BERT, etc.). (And dont forget about GPT-3, etc.) Indeed, the development of LexNLP is centered around the idea that in order to have better performing Legal AI – we will need to connect broader NLP developments to the domain specific needs within law. Stay tuned!

Rethinking Attention with Performers (Important New Paper on arXiv)

Transformers (such as BERT, etc.) have suffered quadratic complexity in the number of tokens in the input sequence … which makes training incredibly laborious / expensive… so this is an important paper by researchers from Google, Cambridge and DeepMind …

ABSTRACT: “We introduce Performers, Transformer architectures which can estimate regular (softmax) full-rank-attention Transformers with provable accuracy, but using only linear (as opposed to quadratic) space and time complexity, without relying on any priors such as sparsity or low-rankness. To approximate softmax attention-kernels, Performers use a novel Fast Attention Via positive Orthogonal Random features approach (FAVOR+), which may be of independent interest for scalable kernel methods. FAVOR+ can be also used to efficiently model kernelizable attention mechanisms beyond softmax. This representational power is crucial to accurately compare softmax with other kernels for the first time on large-scale tasks, beyond the reach of regular Transformers, and investigate optimal attention-kernels. Performers are linear architectures fully compatible with regular Transformers and with strong theoretical guarantees: unbiased or nearly-unbiased estimation of the attention matrix, uniform convergence and low estimation variance. We tested Performers on a rich set of tasks stretching from pixel-prediction through text models to protein sequence modeling. We demonstrate competitive results with other examined efficient sparse and dense attention methods, showcasing effectiveness of the novel attention-learning paradigm leveraged by Performers.” ACCESS THE PAPER from arXiv.