Inference-time techniques, such as repeated sampling or iterative revisions, are emerging as powerful ways to enhance large-language models (LLMs) at test time. However, best practices for developing systems that combine these techniques remain underdeveloped due to our limited understanding of the utility of each technique across models and tasks, the interactions between them, and the massive search space for combining them. To address these challenges, we introduce Archon, a modular and automated framework for optimizing the process of selecting and combining inference-time techniques and LLMs. Given a compute budget and a set of available LLMs, Archon explores a large design space to discover optimized configurations tailored to target benchmarks. It can design custom or general-purpose architectures that advance the Pareto frontier of accuracy vs. maximum token budget compared to top-performing baselines. Across instruction-following, reasoning, and coding tasks, we show that Archon can leverage additional inference compute budget to design systems that outperform frontier models such as OpenAI’s o1, GPT-4o, and Claude 3.5 Sonnet by an average of 15.1%.

Feature attribution, the ability to localize regions of the input data that are relevant for classification, is an important capability for ML models in scientific and biomedical domains. Current methods for feature attribution, which rely on "explaining" the predictions of end-to-end classifiers, suffer from imprecise feature localization and are inadequate for use with small sample sizes and high-dimensional datasets due to computational challenges. We introduce prospector heads, an efficient and interpretable alternative to explanation-based attribution methods that can be applied to any encoder and any data modality. Prospector heads generalize across modalities through experiments on sequences (text), images (pathology), and graphs (protein structures), outperforming baseline attribution methods by up to 26.3 points in mean localization AUPRC. We also demonstrate how prospector heads enable improved interpretation and discovery of class-specific patterns in input data. Through their high performance, flexibility, and generalizability, prospectors provide a framework for improving trust and transparency for ML models in complex domains.

Retrieval pipelines are an integral component of many machine learning systems. However, they perform poorly in domains where documents are long (e.g., 10K tokens or more) and where identifying the relevant document requires synthesizing information across the entire text. Developing long-context retrieval encoders suitable for these domains raises three challenges: (1) how to evaluate long-context retrieval performance, (2) how to pretrain a base language model to represent both short contexts (corresponding to queries) and long contexts (corresponding to documents), and (3) how to finetune this model for retrieval under the batch size limitations imposed by GPU memory constraints. To address these challenges, we first introduce LoCoV1, a 12 task benchmark constructed to measure long-context retrieval where chunking is not possible or not effective. We next present the M2-BERT retrieval encoder, an 80M parameter state-space encoder model built from the Monarch Mixer architecture, capable of scaling to documents up to 32K tokens long. We describe a pretraining data mixture which allows this encoder to process both short and long context sequences, and a finetuning approach that adapts this base model to retrieval with only single-sample batches. Finally, we validate the M2-BERT retrieval encoder on LoCoV1, finding that it outperforms competitive Transformer-based models by at least 22.2 points, despite containing 90× fewer parameters.

Large language models (LLMs) are increasingly used in applications where LLM inputs may span many different tasks. Recent work has found that the choice of LLM is consequential, and different LLMs may be good for different input samples. Prior approaches have thus explored how engineers might select an LLM to use for each sample (i.e. _routing_). While existing routing methods mostly require training auxiliary models on human-annotated data, our work explores whether it is possible to perform _unsupervised_ routing. We propose Smoothie, a weak supervision-inspired routing approach that requires no labeled data. Given a set of outputs from different LLMs, Smoothie constructs a latent variable graphical model over embedding representations of observable LLM outputs and unknown “true” outputs. Using this graphical model, we estimate sample-dependent quality scores for each LLM, and route each sample to the LLM with the highest corresponding score. We find that Smoothie's LLM quality-scores correlate with ground-truth model quality (correctly identifying the optimal model on 9/14 tasks), and that Smoothie outperforms baselines for routing by up to 10 points accuracy.

Weak supervision (WS) is a popular approach for label-efficient learning, leveraging diverse sources of noisy but inexpensive weak labels to automatically annotate training data. Despite its wide usage, WS and its practical value are challenging to benchmark due to the many knobs in its setup, including: data sources, labeling functions (LFs), aggregation techniques (called label models), and end model pipelines. Existing evaluation suites tend to be limited, focusing on particular components or specialized use cases. Moreover, they often involve simplistic benchmark tasks or de-facto LF sets that are suboptimally written, producing insights that may not generalize to real-world settings. We address these limitations by introducing a new benchmark, BOXWRENCH, designed to more accurately reflect real-world usages of WS. This benchmark features tasks with (1) higher class cardinality and imbalance, (2) notable domain expertise requirements, and (3) opportunities to re-use LFs across parallel multilingual corpora. For all tasks, LFs are written using a careful procedure aimed at mimicking real-world settings. In contrast to existing WS benchmarks, we show that supervised learning requires substantial amounts (1000+) of labeled examples to match WS in many settings.

The computational cost of transformer-based models has a quadratic dependence on the length of the input sequence. This makes it challenging to deploy these models in domains in which long documents are especially lengthy, such as the legal domain. To address this issue, we propose a three-stage cascading approach for long document classification. We begin by filtering out likely irrelevant information with a lightweight logistic regression model before passing the more challenging inputs to the transformer-based model. We evaluate our approach using CUAD, a legal dataset with 510 manually-annotated, long contracts. We find that the cascading approach reduces training time by up to 80% while improving baseline performance. We hypothesize that the gains in performance stem from localizing the classification task of the transformer model to particularly difficult examples.

Recent work has shown that language models' (LMs) prompt-based learning capabilities make them well suited for automating data labeling in domains where manual annotation is expensive. The challenge is that while writing an initial prompt is cheap, improving a prompt is costly---practitioners often require significant labeled data in order to evaluate the impact of prompt modifications. Our work asks whether it is possible to improve prompt-based learning _without_ additional labeled data. We approach this problem by attempting to modify the predictions of a prompt, rather than the prompt itself. Our intuition is that accurate predictions should also be consistent: samples which are similar under some feature representation should receive the same prompt prediction. We propose Embroid, a method which computes multiple representations of a dataset under different embedding functions, and uses the consistency between the LM predictions for neighboring samples to identify mispredictions. Embroid then uses these neighborhoods to create additional predictions for each sample, and combines these predictions with a simple latent variable graphical model in order to generate a final corrected prediction. In addition to providing a theoretical analysis of Embroid, we conduct a rigorous empirical evaluation across six different LMs and up to 95 different tasks. We find that (1) Embroid substantially improves performance over original prompts (e.g., by an average of 7.3 points on GPT-JT), (2) also realizes improvements for more sophisticated prompting strategies (e.g., chain-of-thought), and (3) can be specialized to domains like law through the embedding functions.

The advent of large language models (LLMs) and their adoption by the legal community has given rise to the question: what types of legal reasoning can LLMs perform? To enable greater study of this question, we present LegalBench: a collaboratively constructed legal reasoning benchmark consisting of 162 tasks covering six different types of legal reasoning. LegalBench was built through an interdisciplinary process, in which we collected tasks designed and hand-crafted by legal professionals. Because these subject matter experts took a leading role in construction, tasks either measure legal reasoning capabilities that are practically useful, or measure reasoning skills that lawyers find interesting. To enable cross-disciplinary conversations about LLMs in the law, we additionally show how popular legal frameworks for describing legal reasoning—which distinguish between its many forms—correspond to LegalBench tasks, thus giving lawyers and LLM developers a common vocabulary. This paper describes LegalBench, presents an empirical evaluation of 20 open-source and commercial LLMs, and illustrates the types of research explorations LegalBench enables.

One concern with the rise of large language models lies with their potential for significant harm, particularly from pretraining on biased, obscene, copyrighted, and private information. Emerging ethical approaches have attempted to filter pretraining material, but such approaches have been ad hoc and failed to take context into account. We offer an approach to filtering grounded in law, which has directly addressed the tradeoffs in filtering material. First, we gather and make available the Pile of Law, a ~256GB (and growing) dataset of open-source English-language legal and administrative data, covering court opinions, contracts, administrative rules, and legislative records. Pretraining on the Pile of Law may help with legal tasks that have the promise to improve access to justice. Second, we distill the legal norms that governments have developed to constrain the inclusion of toxic or private content into actionable lessons for researchers and discuss how our dataset reflects these norms. Third, we show how the Pile of Law offers researchers the opportunity to learn such filtering rules directly from the data, providing an exciting new research direction in model-based processing.

While self-supervised learning has made rapid advances in natural language processing, it remains unclear when researchers should engage in resource-intensive domain-specific pretraining (domain pretraining). The law, puzzlingly, has yielded few documented instances of substantial gains to domain pretraining in spite of the fact that legal language is widely seen to be unique. We hypothesize that these existing results stem from the fact that existing legal NLP tasks are too easy and fail to meet conditions for when domain pretraining can help. To address this, we first present CaseHOLD (Case Holdings On Legal Decisions), a new dataset comprised of over 53,000+ multiple choice questions to identify the relevant holding of a cited case. This dataset presents a fundamental task to lawyers and is both legally meaningful and difficult from an NLP perspective (F1 of 0.4 with a BiLSTM baseline). Second, we assess performance gains on CaseHOLD and existing legal NLP datasets. While a Transformer architecture (BERT) pretrained on a general corpus (Google Books and Wikipedia) improves performance, domain pretraining (on a corpus of ≈3.5M decisions across all courts in the U.S. that is larger than BERT's) with a custom legal vocabulary exhibits the most substantial performance gains with CaseHOLD (gain of 7.2% on F1, representing a 12% improvement on BERT) and consistent performance gains across two other legal tasks. Third, we show that domain pretraining may be warranted when the task exhibits sufficient similarity to the pretraining corpus: the level of performance increase in three legal tasks was directly tied to the domain specificity of the task. Our findings inform when researchers should engage in resource-intensive pretraining and show that Transformer-based architectures, too, learn embeddings suggestive of distinct legal language.