Michael James Bommarito

Michael Bommarito, Daniel Martin Katz

Nearly all jurisdictions in the United States require a professional license exam, commonly referred to as "the Bar Exam," as a precondition for law practice. To even sit for the exam, most jurisdictions require that an applicant completes at least seven years of post-secondary education, including three years at an accredited law school. In addition, most test-takers also undergo weeks to months of further, exam-specific preparation. Despite this significant investment of time and capital, approximately one in five test-takers still score under the rate required to pass the exam on their first try. In the face of a complex task that requires such depth of knowledge, what, then, should we expect of the state of the art in "AI?" In this research, we document our experimental evaluation of the performance of OpenAI's `text-davinci-003` model, often-referred to as GPT-3.5, on the multistate multiple choice (MBE) section of the exam. While we find no benefit in fine-tuning over GPT-3.5's zero-shot performance at the scale of our training data, we do find that hyperparameter optimization and prompt engineering positively impacted GPT-3.5's zero-shot performance. For best prompt and parameters, GPT-3.5 achieves a headline correct rate of 50.3% on a complete NCBE MBE practice exam, significantly in excess of the 25% baseline guessing rate, and performs at a passing rate for both Evidence and Torts. GPT-3.5's ranking of responses is also highly-correlated with correctness; its top two and top three choices are correct 71% and 88% of the time, respectively, indicating very strong non-entailment performance. While our ability to interpret these results is limited by nascent scientific understanding of LLMs and the proprietary nature of GPT, we believe that these results strongly suggest that an LLM will pass the MBE component of the Bar Exam in the near future.

Daniel Martin Katz, Dirk Hartung, Lauritz Gerlach et al.

In this paper, we summarize the current state of the field of NLP & Law with a specific focus on recent technical and substantive developments. To support our analysis, we construct and analyze a nearly complete corpus of more than six hundred NLP & Law related papers published over the past decade. Our analysis highlights several major trends. Namely, we document an increasing number of papers written, tasks undertaken, and languages covered over the course of the past decade. We observe an increase in the sophistication of the methods which researchers deployed in this applied context. Slowly but surely, Legal NLP is beginning to match not only the methodological sophistication of general NLP but also the professional standards of data availability and code reproducibility observed within the broader scientific community. We believe all of these trends bode well for the future of the field, but many questions in both the academic and commercial sphere still remain open.

Jillian Bommarito, Michael Bommarito, Daniel Martin Katz et al.

The global economy is increasingly dependent on knowledge workers to meet the needs of public and private organizations. While there is no single definition of knowledge work, organizations and industry groups still attempt to measure individuals' capability to engage in it. The most comprehensive assessment of capability readiness for professional knowledge workers is the Uniform CPA Examination developed by the American Institute of Certified Public Accountants (AICPA). In this paper, we experimentally evaluate OpenAI's `text-davinci-003` and prior versions of GPT on both a sample Regulation (REG) exam and an assessment of over 200 multiple-choice questions based on the AICPA Blueprints for legal, financial, accounting, technology, and ethical tasks. First, we find that `text-davinci-003` achieves a correct rate of 14.4% on a sample REG exam section, significantly underperforming human capabilities on quantitative reasoning in zero-shot prompts. Second, `text-davinci-003` appears to be approaching human-level performance on the Remembering & Understanding and Application skill levels in the Exam absent calculation. For best prompt and parameters, the model answers 57.6% of questions correctly, significantly better than the 25% guessing rate, and its top two answers are correct 82.1% of the time, indicating strong non-entailment. Finally, we find that recent generations of GPT-3 demonstrate material improvements on this assessment, rising from 30% for `text-davinci-001` to 57% for `text-davinci-003`. These findings strongly suggest that large language models have the potential to transform the quality and efficiency of future knowledge work.

Michael J. Bommarito

Deep learning research for binary analysis faces a critical infrastructure gap. Today, existing datasets target single platforms, require specialized tooling, or provide only hand-engineered features incompatible with modern neural architectures; no single dataset supports accessible research and pedagogy on realistic use cases. To solve this, we introduce Binary-30K, the first heterogeneous binary dataset designed for sequence-based models like transformers. Critically, Binary-30K covers Windows, Linux, macOS, and Android across 15+ CPU architectures. With 29,793 binaries and approximately 26.93% malware representation, Binary-30K enables research on platform-invariant detection, cross-target transfer learning, and long-context binary understanding. The dataset provides pre-computed byte-level BPE tokenization alongside comprehensive structural metadata, supporting both sequence modeling and structure-aware approaches. Platform-first stratified sampling ensures representative coverage across operating systems and architectures, while distribution via Hugging Face with official train/validation/test splits enables reproducible benchmarking. The dataset is publicly available at https://huggingface.co/datasets/mjbommar/binary-30k, providing an accessible resource for researchers, practitioners, and students alike.

Michael J Bommarito, Daniel Martin Katz, Eric M Detterman

OpenEDGAR is an open source Python framework designed to rapidly construct research databases based on the Electronic Data Gathering, Analysis, and Retrieval (EDGAR) system operated by the US Securities and Exchange Commission (SEC). OpenEDGAR is built on the Django application framework, supports distributed compute across one or more servers, and includes functionality to (i) retrieve and parse index and filing data from EDGAR, (ii) build tables for key metadata like form type and filer, (iii) retrieve, parse, and update CIK to ticker and industry mappings, (iv) extract content and metadata from filing documents, and (v) search filing document contents. OpenEDGAR is designed for use in both academic research and industrial applications, and is distributed under MIT License at https://github.com/LexPredict/openedgar.

Michael J Bommarito, Daniel Martin Katz, Eric M Detterman

LexNLP is an open source Python package focused on natural language processing and machine learning for legal and regulatory text. The package includes functionality to (i) segment documents, (ii) identify key text such as titles and section headings, (iii) extract over eighteen types of structured information like distances and dates, (iv) extract named entities such as companies and geopolitical entities, (v) transform text into features for model training, and (vi) build unsupervised and supervised models such as word embedding or tagging models. LexNLP includes pre-trained models based on thousands of unit tests drawn from real documents available from the SEC EDGAR database as well as various judicial and regulatory proceedings. LexNLP is designed for use in both academic research and industrial applications, and is distributed at https://github.com/LexPredict/lexpredict-lexnlp.

Michael J Bommarito, Daniel Martin Katz, Jillian Bommarito

We present the KL3M tokenizers, a family of specialized tokenizers for legal, financial, and governmental text. Despite established work on tokenization, specialized tokenizers for professional domains remain understudied. Our paper offers two main contributions to this area. First, we introduce domain-specific BPE tokenizers for legal, financial, and governmental text. Our kl3m-004-128k-cased tokenizer uses 9-17% fewer tokens than GPT-4o and Llama3 for domain-specific documents, despite having a smaller vocabulary. For specialized terminology, our cased tokenizer is even more efficient, using up to 83% fewer tokens for legal terms and 39% fewer tokens for financial terms. Second, we develop character-level BPE tokenizers (4K, 8K, and 16K vocabulary sizes) for text correction tasks like OCR post-processing. These tokenizers keep consistent token boundaries between error-containing and correct text, making it easier for models to learn correction patterns. These tokenizers help professional applications by fitting more text in context windows, reducing computational needs, and preserving the meaning of domain-specific terms. Our analysis shows these efficiency gains directly benefit the processing of long legal and financial documents. We release all tokenizers and code through GitHub and Hugging Face to support further research in specialized tokenization.

Michael J Bommarito, Daniel Martin Katz, Jillian Bommarito

We present NUPunkt and CharBoundary, two sentence boundary detection libraries optimized for high-precision, high-throughput processing of legal text in large-scale applications such as due diligence, e-discovery, and legal research. These libraries address the critical challenges posed by legal documents containing specialized citations, abbreviations, and complex sentence structures that confound general-purpose sentence boundary detectors. Our experimental evaluation on five diverse legal datasets comprising over 25,000 documents and 197,000 annotated sentence boundaries demonstrates that NUPunkt achieves 91.1% precision while processing 10 million characters per second with modest memory requirements (432 MB). CharBoundary models offer balanced and adjustable precision-recall tradeoffs, with the large model achieving the highest F1 score (0.782) among all tested methods. Notably, NUPunkt provides a 29-32% precision improvement over general-purpose tools while maintaining exceptional throughput, processing multi-million document collections in minutes rather than hours. Both libraries run efficiently on standard CPU hardware without requiring specialized accelerators. NUPunkt is implemented in pure Python with zero external dependencies, while CharBoundary relies only on scikit-learn and optional ONNX runtime integration for optimized performance. Both libraries are available under the MIT license, can be installed via PyPI, and can be interactively tested at https://sentences.aleainstitute.ai/. These libraries address critical precision issues in retrieval-augmented generation systems by preserving coherent legal concepts across sentences, where each percentage improvement in precision yields exponentially greater reductions in context fragmentation, creating cascading benefits throughout retrieval pipelines and significantly enhancing downstream reasoning quality.

Ilias Chalkidis, Abhik Jana, Dirk Hartung et al.

Laws and their interpretations, legal arguments and agreements\ are typically expressed in writing, leading to the production of vast corpora of legal text. Their analysis, which is at the center of legal practice, becomes increasingly elaborate as these collections grow in size. Natural language understanding (NLU) technologies can be a valuable tool to support legal practitioners in these endeavors. Their usefulness, however, largely depends on whether current state-of-the-art models can generalize across various tasks in the legal domain. To answer this currently open question, we introduce the Legal General Language Understanding Evaluation (LexGLUE) benchmark, a collection of datasets for evaluating model performance across a diverse set of legal NLU tasks in a standardized way. We also provide an evaluation and analysis of several generic and legal-oriented models demonstrating that the latter consistently offer performance improvements across multiple tasks.

Ethan Bommarito, Michael J Bommarito

In this research, we present a comprehensive, longitudinal empirical summary of the R package ecosystem, including not just CRAN, but also Bioconductor and GitHub. We analyze more than 25,000 packages, 150,000 releases, and 15 million files across two decades, providing comprehensive counts and trends for common metrics across packages, releases, authors, licenses, and other important metadata. We find that the historical growth of the ecosystem has been robust under all measures, with a compound annual growth rate of 29% for active packages, 28% for new releases, and 26% for active maintainers. As with many similar social systems, we find a number of highly right-skewed distributions with practical implications, including the distribution of releases per package, packages and releases per author or maintainer, package and maintainer dependency in-degree, and size per package and release. For example, the top five packages are imported by nearly 25% of all packages, and the top ten maintainers support packages that are imported by over half of all packages. We also highlight the dynamic nature of the ecosystem, recording both dramatic acceleration and notable deceleration in the growth of R. From a licensing perspective, we find a notable majority of packages are distributed under copyleft licensing or omit licensing information entirely. The data, methods, and calculations herein provide an anchor for public discourse and industry decisions related to R and CRAN, serving as a foundation for future research on the R software ecosystem and "data science" more broadly.

Ethan Bommarito, Michael Bommarito

In this research, we provide a comprehensive empirical summary of the Python Package Repository, PyPI, including both package metadata and source code covering 178,592 packages, 1,745,744 releases, 76,997 contributors, and 156,816,750 import statements. We provide counts and trends for packages, releases, dependencies, category classifications, licenses, and package imports, as well as authors, maintainers, and organizations. As one of the largest and oldest software repositories as of publication, PyPI provides insight not just into the Python ecosystem today, but also trends in software development and licensing more broadly over time. Within PyPI, we find that the growth of the repository has been robust under all measures, with a compound annual growth rate of 47% for active packages, 39% for new authors, and 61% for new import statements over the last 15 years. As with many similar social systems, we find a number of highly right-skewed distributions, including the distribution of releases per package, packages and releases per author, imports per package, and size per package and release. However, we also find that most packages are contributed by single individuals, not multiple individuals or organizations. The data, methods, and calculations herein provide an anchor for public discourse on PyPI and serve as a foundation for future research on the Python software ecosystem.