Nathaniel Weir

Nathaniel Weir, Ryan Thomas, Randolph D'Amore et al. · microsoft-research

We introduce a language generation task grounded in a popular video game environment. KNUDGE (KNowledge Constrained User-NPC Dialogue GEneration) requires models to produce trees of dialogue between video game characters that accurately reflect quest and entity specifications stated in natural language. KNUDGE is constructed from side quest dialogues drawn directly from game data of Obsidian Entertainment's The Outer Worlds, leading to real-world complexities in generation: (1) dialogues are branching trees as opposed to linear chains of utterances; (2) utterances must remain faithful to the game lore -- character personas, backstories, and entity relationships; and (3) a dialogue must accurately reveal new quest details to the human player. We report results for a set of neural generation models using supervised and in-context learning techniques; we find competent performance but room for future work addressing the challenges of creating realistic, game-quality dialogues.

Nathaniel Weir, Xingdi Yuan, Marc-Alexandre Côté et al. · microsoft-research

Humans have the capability, aided by the expressive compositionality of their language, to learn quickly by demonstration. They are able to describe unseen task-performing procedures and generalize their execution to other contexts. In this work, we introduce DescribeWorld, an environment designed to test this sort of generalization skill in grounded agents, where tasks are linguistically and procedurally composed of elementary concepts. The agent observes a single task demonstration in a Minecraft-like grid world, and is then asked to carry out the same task in a new map. To enable such a level of generalization, we propose a neural agent infused with hierarchical latent language--both at the level of task inference and subtask planning. Our agent first generates a textual description of the demonstrated unseen task, then leverages this description to replicate it. Through multiple evaluation scenarios and a suite of generalization tests, we find that agents that perform text-based inference are better equipped for the challenge under a random split of tasks.

Nathaniel Weir, Peter Clark, Benjamin Van Durme

Our goal is a modern approach to answering questions via systematic reasoning where answers are supported by human interpretable proof trees grounded in an NL corpus of authoritative facts. Such a system would help alleviate the challenges of interpretability and hallucination with modern LMs, and the lack of grounding of current explanation methods (e.g., Chain-of-Thought). This paper proposes a new take on Prolog-based inference engines, where we replace handcrafted rules with a combination of neural language modeling, guided generation, and semiparametric dense retrieval. Our implementation, NELLIE, is the first system to demonstrate fully interpretable, end-to-end grounded QA as entailment tree proof search, going beyond earlier work explaining known-to-be-true facts from text. In experiments, NELLIE outperforms a similar-sized state-of-the-art reasoner [Tafjord et al., 2022] while producing knowledge-grounded explanations. We also find NELLIE can exploit both semi-structured and NL text corpora to guide reasoning. Together these suggest a new way to jointly reap the benefits of both modern neural methods and traditional symbolic reasoning.

Jiefu Ou, Sapana Chaudhary, Kaj Bostrom et al. · amazon-science

Large Language Models (LLMs) demonstrate strong capabilities in general coding tasks but encounter two key challenges when optimizing code: (i) the complexity of writing optimized code (such as performant CUDA kernels and competition-level CPU code) requires expertise in systems, algorithms and specific languages and (ii) requires interpretation of performance metrics like timing and device utilization beyond binary correctness. In this work, we explore inference-time search algorithms that guide the LLM to discover better solutions through iterative refinement based on execution feedback. Our approach, called MaxCode unifies existing search methods under a max-reward reinforcement learning framework, making the observation and action-value functions modular for modification. To enhance the observation space, we integrate a natural language critique model that converts raw execution feedback into diagnostic insights about errors and performance bottlenecks, and the best-discounted reward seen so far. Together, these provide richer input to the code proposal function. To improve exploration during search, we train a generative reward-to-go model using action values from rollouts to rerank potential solutions. Testing on the KernelBench (CUDA) and PIE (C++) optimization benchmarks shows that MaxCode improves optimized code performance compared to baselines, achieving 20.3% and 10.1% relative improvements in absolute speedup value and relative speedup ranking, respectively.

Orion Weller, Aleem Khan, Nathaniel Weir et al.

Recent work in open-domain question answering (ODQA) has shown that adversarial poisoning of the search collection can cause large drops in accuracy for production systems. However, little to no work has proposed methods to defend against these attacks. To do so, we rely on the intuition that redundant information often exists in large corpora. To find it, we introduce a method that uses query augmentation to search for a diverse set of passages that could answer the original question but are less likely to have been poisoned. We integrate these new passages into the model through the design of a novel confidence method, comparing the predicted answer to its appearance in the retrieved contexts (what we call Confidence from Answer Redundancy, i.e. CAR). Together these methods allow for a simple but effective way to defend against poisoning attacks that provides gains of nearly 20% exact match across varying levels of data poisoning/knowledge conflicts.

Zhengping Jiang, Jingyu Zhang, Nathaniel Weir et al.

Hallucinations pose a challenge to the application of large language models (LLMs) thereby motivating the development of metrics to evaluate factual precision. We observe that popular metrics using the Decompose-Then-Verify framework, such as \FActScore, can be manipulated by adding obvious or repetitive subclaims to artificially inflate scores. This observation motivates our new customizable plug-and-play subclaim selection component called Core, which filters down individual subclaims according to their uniqueness and informativeness. We show that many popular factual precision metrics augmented by Core are substantially more robust on a wide range of knowledge domains. We release an evaluation framework supporting easy and modular use of Core and various decomposition strategies, which we recommend adoption by the community. We also release an expansion of the FActScore biography dataset to facilitate further studies of decomposition-based factual precision evaluation.

Kate Sanders, Nathaniel Weir, Sapana Chaudhary et al.

An impediment to using Large Language Models (LLMs) for reasoning output verification is that LLMs struggle to reliably identify errors in thinking traces, particularly in long outputs, domains requiring expert knowledge, and problems without verifiable rewards. We propose a data-driven approach to automatically construct highly granular reasoning error taxonomies to enhance LLM-driven error detection on unseen reasoning traces. Our findings indicate that classification approaches that leverage these error taxonomies, or "rubrics", demonstrate strong error identification compared to baseline methods in technical domains like coding, math, and chemical engineering. These rubrics can be used to build stronger LLM-as-judge reward functions for reasoning model training via reinforcement learning. Experimental results show that these rewards have the potential to improve models' task accuracy on difficult domains over models trained by general LLMs-as-judges by +45%, and approach performance of models trained by verifiable rewards while using as little as 20% as many gold labels. Through our approach, we extend the usage of reward rubrics from assessing qualitative model behavior to assessing quantitative model correctness on tasks typically learned via RLVR rewards. This extension opens the door for teaching models to solve complex technical problems without a full dataset of gold labels, which are often highly costly to procure.

Andre He, Nathaniel Weir, Kaj Bostrom et al.

Reinforcement learning with verifiable rewards (RLVR) has emerged as a promising approach for training reasoning language models (RLMs) by leveraging supervision from verifiers. Although verifier implementation is easier than solution annotation for many tasks, existing synthetic data generation methods remain largely solution-centric, while verifier-based methods rely on a few hand-crafted procedural environments. In this work, we scale RLVR by introducing ReSyn, a pipeline that generates diverse reasoning environments equipped with instance generators and verifiers, covering tasks such as constraint satisfaction, algorithmic puzzles, and spatial reasoning. A Qwen2.5-7B-Instruct model trained with RL on ReSyn data achieves consistent gains across reasoning benchmarks and out-of-domain math benchmarks, including a 27\% relative improvement on the challenging BBEH benchmark. Ablations show that verifier-based supervision and increased task diversity both contribute significantly, providing empirical evidence that generating reasoning environments at scale can enhance reasoning abilities in RLMs

Yu Feng, Nathaniel Weir, Kaj Bostrom et al.

LLMs can perform multi-step reasoning through Chain-of-Thought (CoT), but they cannot reliably verify their own logic. Even when they reach correct answers, the underlying reasoning may be flawed, undermining trust in high-stakes scenarios. To mitigate this issue, we introduce VeriCoT, a neuro-symbolic method that extracts and verifies formal logical arguments from CoT reasoning. VeriCoT formalizes each CoT reasoning step into first-order logic and identifies premises that ground the argument in source context, commonsense knowledge, or prior reasoning steps. The symbolic representation enables automated solvers to verify logical validity while the NL premises allow humans and systems to identify ungrounded or fallacious reasoning steps. Experiments on the ProofWriter, LegalBench, and BioASQ datasets show VeriCoT effectively identifies flawed reasoning, and serves as a strong predictor of final answer correctness. We also leverage VeriCoT's verification signal for (1) inference-time self-reflection, (2) supervised fine-tuning (SFT) on VeriCoT-distilled datasets and (3) preference fine-tuning (PFT) with direct preference optimization (DPO) using verification-based pairwise rewards, further improving reasoning validity and accuracy.

Sam Bayless, Stefano Buliani, Darion Cassel et al.

Large Language Models perform well at natural language interpretation and reasoning, but their inherent stochasticity limits their adoption in regulated industries like finance and healthcare that operate under strict policies. To address this limitation, we present a two-stage neurosymbolic framework that (1) uses LLMs with optional human guidance to formalize natural language policies, allowing fine-grained control of the formalization process, and (2) uses inference-time autoformalization to validate logical correctness of natural language statements against those policies. When correctness is paramount, we perform multiple redundant formalization steps at inference time, cross checking the formalizations for semantic equivalence. Our benchmarks demonstrate that our approach exceeds 99% soundness, indicating a near-zero false positive rate in identifying logical validity. Our approach produces auditable logical artifacts that substantiate the verification outcomes and can be used to improve the original text.

Dongwei Jiang, Jingyu Zhang, Orion Weller et al.

Can LLMs consistently improve their previous outputs for better results? For this to be true, LLMs would need to be better at discriminating among previously-generated alternatives, than generating initial responses. We explore the validity of this hypothesis in practice. We first formulate a unified framework that allows us to compare the generative and discriminative capability of any model on any task. In our resulting experimental analysis of several open-source and industrial LLMs, we observe that models are not reliably better at discriminating among previously-generated alternatives than generating initial responses. This finding challenges the notion that LLMs may be able to enhance their performance only through their own judgment.

Nathaniel Weir, Kate Sanders, Orion Weller et al.

Recent language models enable new opportunities for structured reasoning with text, such as the construction of intuitive, proof-like textual entailment trees without relying on brittle formal logic. However, progress in this direction has been hampered by a long-standing lack of a clear protocol for determining what valid compositional entailment is. This absence causes noisy datasets and limited performance gains by modern neuro-symbolic engines. To address these problems, we formulate a consistent and theoretically grounded approach to annotating decompositional entailment and evaluate its impact on LLM-based textual inference. We find that our new dataset, RDTE (Recognizing Decompositional Textual Entailment), has a substantially higher internal consistency (+9%) than prior decompositional entailment datasets. We also find that training an RDTE-oriented entailment classifier via knowledge distillation and employing it in an entailment tree reasoning engine significantly improves both accuracy and proof quality, illustrating the practical benefit of this advance for textual inference.

Kate Sanders, Nathaniel Weir, Benjamin Van Durme

It is challenging for models to understand complex, multimodal content such as television clips, and this is in part because video-language models often rely on single-modality reasoning and lack interpretability. To combat these issues we propose TV-TREES, the first multimodal entailment tree generator. TV-TREES serves as an approach to video understanding that promotes interpretable joint-modality reasoning by searching for trees of entailment relationships between simple text-video evidence and higher-level conclusions that prove question-answer pairs. We also introduce the task of multimodal entailment tree generation to evaluate reasoning quality. Our method's performance on the challenging TVQA benchmark demonstrates interpretable, state-of-the-art zero-shot performance on full clips, illustrating that multimodal entailment tree generation can be a best-of-both-worlds alternative to black-box systems.

Nathaniel Weir, Bhavana Dalvi Mishra, Orion Weller et al.

Recent reasoning methods (e.g., chain-of-thought, entailment reasoning) help users understand how language models (LMs) answer a single question, but they do little to reveal the LM's overall understanding, or "theory," about the question's topic, making it still hard to trust the model. Our goal is to materialize such theories - here called microtheories (a linguistic analog of logical microtheories) - as a set of sentences encapsulating an LM's core knowledge about a topic. These statements systematically work together to entail answers to a set of questions to both engender trust and improve performance. Our approach is to first populate a knowledge store with (model-generated) sentences that entail answers to training questions and then distill those down to a core microtheory that is concise, general, and non-redundant. We show that, when added to a general corpus (e.g., Wikipedia), microtheories can supply critical, topical information not necessarily present in the corpus, improving both a model's ability to ground its answers to verifiable knowledge (i.e., show how answers are systematically entailed by documents in the corpus, fully grounding up to +8% more answers), and the accuracy of those grounded answers (up to +8% absolute). We also show that, in a human evaluation in the medical domain, our distilled microtheories contain a significantly higher concentration of topically critical facts than the non-distilled knowledge store. Finally, we show we can quantify the coverage of a microtheory for a topic (characterized by a dataset) using a notion of $p$-relevance. Together, these suggest that microtheories are an efficient distillation of an LM's topic-relevant knowledge, that they can usefully augment existing corpora, and can provide both performance gains and an interpretable, verifiable window into the model's knowledge of a topic.

Vikash Singh, Darion Cassel, Nathaniel Weir et al.

Despite the syntactic fluency of Large Language Models (LLMs), ensuring their logical correctness in high-stakes domains remains a fundamental challenge. We present a neurosymbolic framework that combines LLMs with SMT solvers to produce verification-guided answers through iterative refinement. Our approach decomposes LLM outputs into atomic claims, autoformalizes them into first-order logic, and verifies their logical consistency using automated theorem proving. We introduce three key innovations: (1) multi-model consensus via formal semantic equivalence checking to ensure logic-level alignment between candidates, eliminating the syntactic bias of surface-form metrics, (2) semantic routing that directs different claim types to appropriate verification strategies: symbolic solvers for logical claims and LLM ensembles for commonsense reasoning, and (3) precise logical error localization via Minimal Correction Subsets (MCS), which pinpoint the exact subset of claims to revise, transforming binary failure signals into actionable feedback. Our framework classifies claims by their logical status and aggregates multiple verification signals into a unified score with variance-based penalty. The system iteratively refines answers using structured feedback until acceptance criteria are met or convergence is achieved. This hybrid approach delivers formal guarantees where possible and consensus verification elsewhere, advancing trustworthy AI. With the GPT-OSS-120B model, VERGE demonstrates an average performance uplift of 18.7% at convergence across a set of reasoning benchmarks compared to single-pass approaches.

Xinrui Zou, Ming Zhang, Nathaniel Weir et al.

Statutory reasoning refers to the application of legislative provisions to a series of case facts described in natural language. We re-frame statutory reasoning as an analogy task, where each instance of the analogy task involves a combination of two instances of statutory reasoning. This increases the dataset size by two orders of magnitude, and introduces an element of interpretability. We show that this task is roughly as difficult to Natural Language Processing models as the original task. Finally, we come back to statutory reasoning, solving it with a combination of a retrieval mechanism and analogy models, and showing some progress on prior comparable work.

Orion Weller, Marc Marone, Nathaniel Weir et al.

Large Language Models (LLMs) may hallucinate and generate fake information, despite pre-training on factual data. Inspired by the journalistic device of "according to sources", we propose according-to prompting: directing LLMs to ground responses against previously observed text. To quantify this grounding, we propose a novel evaluation metric (QUIP-Score) that measures the extent to which model-produced answers are directly found in underlying text corpora. We illustrate with experiments on three corpora (Wikipedia, PubMed, and the U.S. legal tax code) that these prompts improve grounding under our metrics, with the additional benefit of often improving end-task performance. Furthermore, prompts that ask the model to decrease grounding (or to ground to other corpora) indeed decrease QUIP-Score, indicating the ability of LLMs to increase or decrease grounded generations on request.

Jiefu Ou, Nathaniel Weir, Anton Belyy et al.

We propose a structured extension to bidirectional-context conditional language generation, or "infilling," inspired by Frame Semantic theory (Fillmore, 1976). Guidance is provided through two approaches: (1) model fine-tuning, conditioning directly on observed symbolic frames, and (2) a novel extension to disjunctive lexically constrained decoding that leverages frame semantic lexical units. Automatic and human evaluations confirm that frame-guided generation allows for explicit manipulation of intended infill semantics, with minimal loss in distinguishability from human-generated text. Our methods flexibly apply to a variety of use scenarios, and we provide a codebase and interactive demo available from https://nlp.jhu.edu/demos/infillmore.

Nathaniel Weir, João Sedoc, Benjamin Van Durme

We present COD3S, a novel method for generating semantically diverse sentences using neural sequence-to-sequence (seq2seq) models. Conditioned on an input, seq2seq models typically produce semantically and syntactically homogeneous sets of sentences and thus perform poorly on one-to-many sequence generation tasks. Our two-stage approach improves output diversity by conditioning generation on locality-sensitive hash (LSH)-based semantic sentence codes whose Hamming distances highly correlate with human judgments of semantic textual similarity. Though it is generally applicable, we apply COD3S to causal generation, the task of predicting a proposition's plausible causes or effects. We demonstrate through automatic and human evaluation that responses produced using our method exhibit improved diversity without degrading task performance.

Nathaniel Weir, Adam Poliak, Benjamin Van Durme

Humans carry stereotypic tacit assumptions (STAs) (Prince, 1978), or propositional beliefs about generic concepts. Such associations are crucial for understanding natural language. We construct a diagnostic set of word prediction prompts to evaluate whether recent neural contextualized language models trained on large text corpora capture STAs. Our prompts are based on human responses in a psychological study of conceptual associations. We find models to be profoundly effective at retrieving concepts given associated properties. Our results demonstrate empirical evidence that stereotypic conceptual representations are captured in neural models derived from semi-supervised linguistic exposure.

Prasetya Utama, Nathaniel Weir, Fuat Basik et al.

The ability to extract insights from new data sets is critical for decision making. Visual interactive tools play an important role in data exploration since they provide non-technical users with an effective way to visually compose queries and comprehend the results. Natural language has recently gained traction as an alternative query interface to databases with the potential to enable non-expert users to formulate complex questions and information needs efficiently and effectively. However, understanding natural language questions and translating them accurately to SQL is a challenging task, and thus Natural Language Interfaces for Databases (NLIDBs) have not yet made their way into practical tools and commercial products. In this paper, we present DBPal, a novel data exploration tool with a natural language interface. DBPal leverages recent advances in deep models to make query understanding more robust in the following ways: First, DBPal uses a deep model to translate natural language statements to SQL, making the translation process more robust to paraphrasing and other linguistic variations. Second, to support the users in phrasing questions without knowing the database schema and the query features, DBPal provides a learned auto-completion model that suggests partial query extensions to users during query formulation and thus helps to write complex queries.