Bhavana Dalvi Mishra

Linxin Song, Jiefeng Chen, Yue Huang et al.

Coding agents increasingly act as codebase-scale collaborators that can assist with codebase conversion, but this progress has exposed a critical weakness: agents often over-trust their own local validation routines and declare success on artifacts that satisfy surface checks while violating the semantic contracts users actually care about. This problem is especially acute in codebase conversion, where prior evaluation is largely outcome-driven and therefore unstable: two implementations can match on a shallow outcome, such as a single forward loss, while diverging in gradients, optimizer behavior, or short-horizon training dynamics. We introduce T2J-Bench, a benchmark for codebase conversion that reformulates conversion as transfer under a fixed equivalence contract. A fixed verifier then compares source and converted codebases through three ordered stages: Spec (interface admissibility), Numeric (forward outputs, losses, gradients, and objective-specific tensors), and Behavioral (short training dynamics under fixed seeds). Across 355 blind conversion attempts, the best system reaches only 26.7--28.9% overall pass rate despite Spec pass rates up to 91.1%; a 4.7x token-budget spread yields only a 2.2x pass-rate spread; and all systems overestimate success by 66.6--97.8 points relative to the fixed evaluator. This suggests that failures stem more from contract-misaligned self-validation than from limited budget or backbone strength.

Oyvind Tafjord, Bhavana Dalvi Mishra, Peter Clark · allen-ai

Our goal is a question-answering (QA) system that can show how its answers are implied by its own internal beliefs via a systematic chain of reasoning. Such a capability would allow better understanding of why a model produced the answer it did. Our approach is to recursively combine a trained backward-chaining model, capable of generating a set of premises entailing an answer hypothesis, with a verifier that checks that the model itself believes those premises (and the entailment itself) through self-querying. To our knowledge, this is the first system to generate multistep chains that are both faithful (the answer follows from the reasoning) and truthful (the chain reflects the system's own internal beliefs). In evaluation using two different datasets, users judge that a majority (70%+) of generated chains clearly show how an answer follows from a set of facts - substantially better than a high-performance baseline - while preserving answer accuracy. By materializing model beliefs that systematically support an answer, new opportunities arise for understanding the model's system of belief, and diagnosing and correcting its misunderstandings when an answer is wrong.

Bhavana Dalvi Mishra, Oyvind Tafjord, Peter Clark · allen-ai

Our goal is a teachable reasoning system for question-answering (QA), where a user can interact with faithful answer explanations, and correct its errors so that the system improves over time. Our approach is to augment a QA model with a dynamic memory of user feedback, containing user-supplied corrections to erroneous model beliefs that users identify during interaction. Retrievals from memory are used as additional context for QA, to help avoid previous mistakes in similar new situations - a novel application of memory-based continuous learning. With simulated feedback, we find that our system (called TeachMe) continually improves with time, and without model retraining, requiring feedback on only 25% of training examples to reach within 1% of the upper-bound (feedback on all examples). Similarly, in experiments with real users, we observe a similar trend, with performance improving by over 15% on a hidden test set after teaching. This suggests new opportunities for using frozen language models in an interactive setting where users can inspect, debug, and correct the model's beliefs, leading to improved system's performance over time.

Yuling Gu, Bhavana Dalvi Mishra, Peter Clark · allen-ai

When people think of everyday things like an egg, they typically have a mental image associated with it. This allows them to correctly judge, for example, that "the yolk surrounds the shell" is a false statement. Do language models similarly have a coherent picture of such everyday things? To investigate this, we propose a benchmark dataset consisting of 100 everyday things, their parts, and the relationships between these parts, expressed as 11,720 "X relation Y?" true/false questions. Using these questions as probes, we observe that state-of-the-art pre-trained language models (LMs) like GPT-3 and Macaw have fragments of knowledge about these everyday things, but do not have fully coherent "parts mental models" (54-59% accurate, 19-43% conditional constraint violation). We propose an extension where we add a constraint satisfaction layer on top of the LM's raw predictions to apply commonsense constraints. As well as removing inconsistencies, we find that this also significantly improves accuracy (by 16-20%), suggesting how the incoherence of the LM's pictures of everyday things can be significantly reduced.

Yuling Gu, Yao Fu, Valentina Pyatkin et al. · allen-ai, uw

Figurative language (e.g., "he flew like the wind") is challenging to understand, as it is hard to tell what implicit information is being conveyed from the surface form alone. We hypothesize that to perform this task well, the reader needs to mentally elaborate the scene being described to identify a sensible meaning of the language. We present DREAM-FLUTE, a figurative language understanding system that does this, first forming a "mental model" of situations described in a premise and hypothesis before making an entailment/contradiction decision and generating an explanation. DREAM-FLUTE uses an existing scene elaboration model, DREAM, for constructing its "mental model." In the FigLang2022 Shared Task evaluation, DREAM-FLUTE achieved (joint) first place (Acc@60=63.3%), and can perform even better with ensemble techniques, demonstrating the effectiveness of this approach. More generally, this work suggests that adding a reflective component to pretrained language models can improve their performance beyond standard fine-tuning (3.3% improvement in Acc@60).

Bodhisattwa Prasad Majumder, Bhavana Dalvi Mishra, Peter Jansen et al.

Language agents have shown some ability to interact with an external environment, e.g., a virtual world such as ScienceWorld, to perform complex tasks, e.g., growing a plant, without the startup costs of reinforcement learning. However, despite their zero-shot capabilities, these agents to date do not continually improve over time beyond performance refinement on a specific task. Here we present CLIN, the first language-based agent to achieve this, so that it continually improves over multiple trials, including when both the environment and task are varied, and without requiring parameter updates. Our approach is to use a persistent, dynamic, textual memory centered on causal abstractions (rather than general "helpful hints") that is regularly updated after each trial so that the agent gradually learns useful knowledge for new trials. In the ScienceWorld benchmark, CLIN is able to continually improve on repeated trials on the same task and environment, outperforming state-of-the-art reflective language agents like Reflexion by 23 absolute points. CLIN can also transfer its learning to new environments (or new tasks), improving its zero-shot performance by 4 points (13 for new tasks) and can further improve performance there through continual memory updates, enhancing performance by an additional 17 points (7 for new tasks). This suggests a new architecture for agents built on frozen models that can still continually and rapidly improve over time.

Bodhisattwa Prasad Majumder, Harshit Surana, Dhruv Agarwal et al.

Can the rapid advances in code generation, function calling, and data analysis using large language models (LLMs) help automate the search and verification of hypotheses purely from a set of provided datasets? To evaluate this question, we present DiscoveryBench, the first comprehensive benchmark that formalizes the multi-step process of data-driven discovery. The benchmark is designed to systematically assess current model capabilities in discovery tasks and provide a useful resource for improving them. Our benchmark contains 264 tasks collected across 6 diverse domains, such as sociology and engineering, by manually deriving discovery workflows from published papers to approximate the real-world challenges faced by researchers, where each task is defined by a dataset, its metadata, and a discovery goal in natural language. We additionally provide 903 synthetic tasks to conduct controlled evaluations across task complexity. Furthermore, our structured formalism of data-driven discovery enables a facet-based evaluation that provides useful insights into different failure modes. We evaluate several popular LLM-based reasoning frameworks using both open and closed LLMs as baselines on DiscoveryBench and find that even the best system scores only 25%. Our benchmark, thus, illustrates the challenges in autonomous data-driven discovery and serves as a valuable resource for the community to make progress.

Rui Meng, Bhavana Dalvi Mishra, Jiefeng Chen et al.

Autonomous research agents produce competitive solutions and professional-looking manuscripts, yet their outputs contain verifiability failures undetectable by surface-level evaluation: fabricated citations, unreproducible scores, and method descriptions that diverge from the implementation. We address this through three contributions. First, Chain-of-Evidence (CoE), a verifiability framework requiring every claim to be traceable to its evidence source. Second, ScientistOne, an end-to-end autonomous research system that maintains evidence chains by construction throughout literature review, solution discovery, and paper writing. Third, CoE Audit, a post-hoc audit whose four integrity checks -- score verification, specification violation, reference verification, and method-code alignment -- apply uniformly to all systems. Across 75 papers spanning five systems and five frontier research tasks, every baseline exhibits at least one systematic failure mode: hallucinated reference rates reach 21%, score verification passes in as few as 42% of papers, and method-code alignment ranges from 20% to 80%. ScientistOne achieves zero hallucinated references (0/337), perfect score verification (12/12), and the highest method-code alignment (14/15), while matching or exceeding human expert performance on all five tasks. ScientistOne further generalizes to six additional tasks spanning medical imaging, fine-grained recognition, 3D perception, and language modeling, achieving state-of-the-art on Parameter Golf and gold medals on MLE-Bench tasks where baselines fail entirely.

Hita Kambhamettu, Bhavana Dalvi Mishra, Andrew Head et al.

Developing a novel research idea is hard. It must be distinct enough from prior work to claim a contribution while also building on it. This requires iteratively reviewing literature and refining an idea based on what a researcher reads; yet when an idea changes, the literature that matters often changes with it. Most tools offer limited support for this interplay: literature tools help researchers understand a fixed body of work, while ideation tools evaluate ideas against a static, pre-curated set of papers. We introduce literature-initiated pivots, a mechanism where engagement with literature prompts revision to a developing idea, and where that revision changes which literature is relevant. We operationalize this in LitPivot, where researchers concurrently draft and vet an idea. LitPivot dynamically retrieves clusters of papers relevant to a selected part of the idea and proposes literature-informed critiques for how to revise it. A lab study ($n{=}17$) shows researchers produced higher-rated ideas with stronger self-reported understanding of the literature space; an open-ended study ($n{=}5$) reveals how researchers use LitPivot to iteratively evolve their own ideas.

Minbeom Kim, Lesly Miculicich, Bhavana Dalvi Mishra et al.

As AI agents move from chat interfaces to systems that read private data, call tools, and execute multi-step workflows, guardrails become a last line of defense against concrete deployment harms. In these settings, guardrail failures are no longer merely answer-quality errors: they can leak secrets, authorize unsafe actions, or block legitimate work. The hardest failures are often contextual: whether an action is acceptable depends on local privacy norms, organizational policies, and user expectations that resist pre-deployment specification. This creates a practical gap: guardrails must adapt to their own operating environments, yet deployment feedback is typically limited to sparse, noisy user-reported failures, and repeated fine-tuning is often impractical. To address this gap, we propose LiSA (Lifelong Safety Adaptation), a conservative policy induction framework that improves a fixed base guardrail through structured memory. LiSA converts occasional failures into reusable policy abstractions so that sparse reports can generalize beyond individual cases, adds conflict-aware local rules to prevent overgeneralization in mixed-label contexts, and applies evidence-aware confidence gating via a posterior lower bound, so that memory reuse scales with accumulated evidence rather than empirical accuracy alone. Across PrivacyLens+, ConFaide+, and AgentHarm, LiSA consistently outperforms strong memory-based baselines under sparse feedback, remains robust under noisy user feedback even at 20% label-flip rates, and pushes the latency--performance frontier beyond backbone model scaling. Ultimately, LiSA offers a practical path to secure AI agents against the unpredictable long tail of real-world edge risks.

Jiefeng Chen, Bhavana Dalvi Mishra, Jaehyun Nam et al.

Automating AI research differs from general software engineering due to computationally expensive evaluation (e.g., model training) and opaque performance attribution. Current LLM-based agents struggle here, often generating monolithic scripts that ignore execution costs and causal factors. We introduce MARS (Modular Agent with Reflective Search), a framework optimized for autonomous AI research. MARS relies on three pillars: (1) Budget-Aware Planning via cost-constrained Monte Carlo Tree Search (MCTS) to explicitly balance performance with execution expense; (2) Modular Construction, employing a "Design-Decompose-Implement" pipeline to manage complex research repositories; and (3) Comparative Reflective Memory, which addresses credit assignment by analyzing solution differences to distill high-signal insights. MARS achieves state-of-the-art performance among open-source frameworks on MLE-Bench under comparable settings, maintaining competitiveness with the global leaderboard's top methods. Furthermore, the system exhibits qualitative "Aha!" moments, where 63% of all utilized lessons originate from cross-branch transfer, demonstrating that the agent effectively generalizes insights across search paths.

Siru Ouyang, Jun Yan, Yanfei Chen et al.

LLM-based agents are increasingly deployed to handle streaming tasks, yet they often remain one-off problem solvers that fail to learn from past interactions. Reusable skills distilled from experience provide a natural substrate for self-evolution, where high-quality skill curation serves as the key bottleneck. Existing approaches either rely on manual skill curation, prescribe heuristic skill operations, or train for short-horizon skill operations. However, they still struggle to learn complex long-term curation policies from indirect and delayed feedback. To tackle this challenge, we propose SkillOS, an experience-driven RL training recipe for learning skill curation in self-evolving agents. SkillOS pairs a frozen agent executor that retrieves and applies skills with a trainable skill curator that updates an external SkillRepo from accumulated experience. To provide learning signals for curation, we design composite rewards and train on grouped task streams based on skill-relevant task dependencies, where earlier trajectories update the SkillRepo, and later related tasks evaluate these updates. Across multi-turn agentic tasks and single-turn reasoning tasks, SkillOS consistently outperforms memory-free and strong memory-based baselines in both effectiveness and efficiency, with the learned skill curator generalizing across different executor backbones and task domains. Further analyses show that the learned curator produces more targeted skill use, while the skills in SkillRepo evolve into more richly structured Markdown files that encode higher-level meta-skills over time.

Gaotang Li, Bhavana Dalvi Mishra, Zifeng Wang et al.

Training deep research agents, namely systems that plan, search, evaluate evidence, and synthesize long-form reports, pushes reinforcement learning beyond the regime of verifiable rewards. Their outputs lack ground-truth answers, their trajectories span many tool-augmented decisions, and standard post-training offers little mechanism for turning past attempts into reusable experience. In this work, we argue that rubrics should serve not merely as final-answer evaluators, but as the shared interface that structures policy execution, judge feedback, and agent memory. Based on this view, we introduce RubricEM, a rubric-guided reinforcement learning framework that combines stagewise policy decomposition with reflection-based meta-policy evolution. RubricEM first makes research trajectories stage-aware by conditioning planning, evidence gathering, review, and synthesis on self-generated rubrics. It then assigns credit with Stage-Structured GRPO, which uses stagewise rubric judgments to provide denser semantic feedback for long-horizon optimization. In parallel, RubricEM trains a shared-backbone reflection meta-policy that distills judged trajectories into reusable rubric-grounded guidance for future attempts. The resulting RubricEM-8B achieves strong performance across four long-form research benchmarks, outperforming comparable open models and approaching proprietary deep-research systems. Beyond final performance, we perform thorough analyses to understand the key ingredients of RubricEM.

Peter Jansen, Marc-Alexandre Côté, Tushar Khot et al.

Automated scientific discovery promises to accelerate progress across scientific domains. However, developing and evaluating an AI agent's capacity for end-to-end scientific reasoning is challenging as running real-world experiments is often prohibitively expensive or infeasible. In this work we introduce DISCOVERYWORLD, the first virtual environment for developing and benchmarking an agent's ability to perform complete cycles of novel scientific discovery. DISCOVERYWORLD contains a variety of different challenges, covering topics as diverse as radioisotope dating, rocket science, and proteomics, to encourage development of general discovery skills rather than task-specific solutions. DISCOVERYWORLD itself is an inexpensive, simulated, text-based environment (with optional 2D visual overlay). It includes 120 different challenge tasks, spanning eight topics each with three levels of difficulty and several parametric variations. Each task requires an agent to form hypotheses, design and run experiments, analyze results, and act on conclusions. DISCOVERYWORLD further provides three automatic metrics for evaluating performance, based on (a) task completion, (b) task-relevant actions taken, and (c) the discovered explanatory knowledge. We find that strong baseline agents, that perform well in prior published environments, struggle on most DISCOVERYWORLD tasks, suggesting that DISCOVERYWORLD captures some of the novel challenges of discovery, and thus that DISCOVERYWORLD may help accelerate near-term development and assessment of scientific discovery competency in agents. Code available at: www.github.com/allenai/discoveryworld

Yuling Gu, Bhavana Dalvi Mishra, Peter Clark

When people answer questions about a specific situation, e.g., "I cheated on my mid-term exam last week. Was that wrong?", cognitive science suggests that they form a mental picture of that situation before answering. While we do not know how language models (LMs) answer such questions, we conjecture that they may answer more accurately if they are also provided with additional details about the question situation, elaborating the "scene". To test this conjecture, we train a new model, DREAM, to answer questions that elaborate the scenes that situated questions are about, and then provide those elaborations as additional context to a question-answering (QA) model. We find that DREAM is able to create better scene elaborations (more accurate, useful, and consistent) than a representative state-of-the-art, zero-shot model (Macaw). We also find that using the scene elaborations as additional context improves the answer accuracy of a downstream QA system, including beyond that obtainable by simply further finetuning the QA system on DREAM's training data. These results suggest that adding focused elaborations about a situation can improve a system's reasoning about it, and may serve as an effective way of injecting new scenario based knowledge into QA models. Finally, our approach is dataset-neutral; we observe improved QA performance across different models, with even bigger gains on models with fewer parameters. We make our dataset and model publicly available at https://github.com/allenai/dream.

Peter Jansen, Oyvind Tafjord, Marissa Radensky et al. · allen-ai

Despite the surge of interest in autonomous scientific discovery (ASD) of software artifacts (e.g., improved ML algorithms), current ASD systems face two key limitations: (1) they largely explore variants of existing codebases or similarly constrained design spaces, and (2) they produce large volumes of research artifacts (such as automatically generated papers and code) that are typically evaluated using conference-style paper review with limited evaluation of code. In this work we introduce CodeScientist, a novel ASD system that frames ideation and experiment construction as a form of genetic search jointly over combinations of research articles and codeblocks defining common actions in a domain (like prompting a language model). We use this paradigm to conduct hundreds of automated experiments on machine-generated ideas broadly in the domain of agents and virtual environments, with the system returning 19 discoveries, 6 of which were judged as being both at least minimally sound and incrementally novel after a multi-faceted evaluation beyond that typically conducted in prior work, including external (conference-style) review, code review, and replication attempts. Moreover, the discoveries span new tasks, agents, metrics, and data, suggesting a qualitative shift from benchmark optimization to broader discoveries.

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.

Kolby Nottingham, Bodhisattwa Prasad Majumder, Bhavana Dalvi Mishra et al.

Large language models (LLMs) have recently been used for sequential decision making in interactive environments. However, leveraging environment reward signals for continual LLM actor improvement is not straightforward. We propose Skill Set Optimization (SSO) for improving LLM actor performance through constructing and refining sets of transferable skills. SSO constructs skills by extracting common subtrajectories with high rewards and generating subgoals and instructions to represent each skill. These skills are provided to the LLM actor in-context to reinforce behaviors with high rewards. Then, SSO further refines the skill set by pruning skills that do not continue to result in high rewards. We evaluate our method in the classic videogame NetHack and the text environment ScienceWorld to demonstrate SSO's ability to optimize a set of skills and perform in-context policy improvement. SSO outperforms baselines by 40% in our custom NetHack task and outperforms the previous state-of-the-art in ScienceWorld by 35%.

Dhruv Agarwal, Bodhisattwa Prasad Majumder, Reece Adamson et al.

The promise of autonomous scientific discovery (ASD) hinges not only on answering questions, but also on knowing which questions to ask. Most recent works in ASD explore the use of large language models (LLMs) in goal-driven settings, relying on human-specified research questions to guide hypothesis generation. However, scientific discovery may be accelerated further by allowing the AI system to drive exploration by its own criteria. The few existing approaches in open-ended ASD select hypotheses based on diversity heuristics or subjective proxies for human interestingness, but the former struggles to meaningfully navigate the typically vast hypothesis space, and the latter suffers from imprecise definitions. This paper presents AutoDS -- a method for open-ended ASD that instead drives scientific exploration using Bayesian surprise. Here, we quantify the epistemic shift from the LLM's prior beliefs about a hypothesis to its posterior beliefs after gathering experimental results. To efficiently explore the space of nested hypotheses, our method employs a Monte Carlo tree search (MCTS) strategy with progressive widening using surprisal as the reward function. We evaluate AutoDS in the setting of data-driven discovery across 21 real-world datasets spanning domains such as biology, economics, finance, and behavioral science. Our results demonstrate that under a fixed budget, AutoDS substantially outperforms competitors by producing 5--29\% more discoveries deemed surprising by the LLM. Our human evaluation further finds that two-thirds of AutoDS discoveries are surprising to the domain experts, suggesting this is an important step forward towards building open-ended ASD systems.

Rosni Vasu, Chandrayee Basu, Bhavana Dalvi Mishra et al.

Large Language models have demonstrated promising performance in research ideation across scientific domains. Hypothesis development, the process of generating a highly specific declarative statement connecting a research idea with empirical validation, has received relatively less attention. Existing approaches trivially deploy retrieval augmentation and focus only on the quality of the final output ignoring the underlying reasoning process behind ideation. We present $\texttt{HypER}$ ($\textbf{Hyp}$othesis Generation with $\textbf{E}$xplanation and $\textbf{R}$easoning), a small language model (SLM) trained for literature-guided reasoning and evidence-based hypothesis generation. $\texttt{HypER}$ is trained in a multi-task setting to discriminate between valid and invalid scientific reasoning chains in presence of controlled distractions. We find that $\texttt{HypER}$ outperformes the base model, distinguishing valid from invalid reasoning chains (+22\% average absolute F1), generates better evidence-grounded hypotheses (0.327 vs. 0.305 base model) with high feasibility and impact as judged by human experts ($>$3.5 on 5-point Likert scale).

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.

Peter Clark, Bhavana Dalvi Mishra, Oyvind Tafjord

While there are numerous benchmarks comparing the performance of modern language models (LMs), end-task evaluations often conflate notions of *factual accuracy* ("truth") and *reasoning ability* ("rationality", or "honesty" in the sense of correctly reporting implications of beliefs). Our goal is a dataset that clearly distinguishes these two notions. Our approach is to leverage and extend a collection of human-annotated *entailment trees*, engineered to express both good and bad chains of reasoning, and using a mixture of true and false facts, in particular including counterfactual examples, to avoid belief bias (also known as the "content effect"). The resulting dataset, called BaRDa, contains 3000 entailments (1787 valid, 1213 invalid), using 6681 true and 2319 false statements. Testing on four GPT-series models, GPT3(curie)/GPT3(davinici)/3.5/4, we find factual accuracy (truth) scores of 74.1/80.6/82.6/87.1 and reasoning accuracy scores of 63.1/78.0/71.8/79.2. This shows the clear progression of models towards improved factual accuracy and entailment reasoning, and the dataset provides a new benchmark that more cleanly separates and quantifies these two notions.

Rosni Vasu, Peter Jansen, Pao Siangliulue et al. · allen-ai

While there has been a surge of interest in automated scientific discovery (ASD), especially with the emergence of LLMs, it remains challenging for tools to generate hypotheses that are both testable and grounded in the scientific literature. Additionally, existing ideation tools are not adaptive to prior experimental outcomes. We developed HARPA to address these challenges by incorporating the ideation workflow inspired by human researchers. HARPA first identifies emerging research trends through literature mining, then explores hypothesis design spaces, and finally converges on precise, testable hypotheses by pinpointing research gaps and justifying design choices. Our evaluations show that HARPA-generated hypothesis-driven research proposals perform comparably to a strong baseline AI-researcher across most qualitative dimensions (e.g., specificity, novelty, overall quality), but achieve significant gains in feasibility(+0.78, p$<0.05$, bootstrap) and groundedness (+0.85, p$<0.01$, bootstrap) on a 10-point Likert scale. When tested with the ASD agent (CodeScientist), HARPA produced more successful executions (20 vs. 11 out of 40) and fewer failures (16 vs. 21 out of 40), showing that expert feasibility judgments track with actual execution success. Furthermore, to simulate how researchers continuously refine their understanding of what hypotheses are both testable and potentially interesting from experience, HARPA learns a reward model that scores new hypotheses based on prior experimental outcomes, achieving approx. a 28\% absolute gain over HARPA's untrained baseline scorer. Together, these methods represent a step forward in the field of AI-driven scientific discovery.

Zhouhang Xie, Tushar Khot, Bhavana Dalvi Mishra et al.

Instruction-following LLMs have recently allowed systems to discover hidden concepts from a collection of unstructured documents based on a natural language description of the purpose of the discovery (i.e., goal). Still, the quality of the discovered concepts remains mixed, as it depends heavily on LLM's reasoning ability and drops when the data is noisy or beyond LLM's knowledge. We present Instruct-LF, a goal-oriented latent factor discovery system that integrates LLM's instruction-following ability with statistical models to handle large, noisy datasets where LLM reasoning alone falls short. Instruct-LF uses LLMs to propose fine-grained, goal-related properties from documents, estimates their presence across the dataset, and applies gradient-based optimization to uncover hidden factors, where each factor is represented by a cluster of co-occurring properties. We evaluate latent factors produced by Instruct-LF on movie recommendation, text-world navigation, and legal document categorization tasks. These interpretable representations improve downstream task performance by 5-52% than the best baselines and were preferred 1.8 times as often as the best alternative, on average, in human evaluation.

Sumithra Bhakthavatsalam, Daniel Khashabi, Tushar Khot et al.

We present the ARC-DA dataset, a direct-answer ("open response", "freeform") version of the ARC (AI2 Reasoning Challenge) multiple-choice dataset. While ARC has been influential in the community, its multiple-choice format is unrepresentative of real-world questions, and multiple choice formats can be particularly susceptible to artifacts. The ARC-DA dataset addresses these concerns by converting questions to direct-answer format using a combination of crowdsourcing and expert review. The resulting dataset contains 2985 questions with a total of 8436 valid answers (questions typically have more than one valid answer). ARC-DA is one of the first DA datasets of natural questions that often require reasoning, and where appropriate question decompositions are not evident from the questions themselves. We describe the conversion approach taken, appropriate evaluation metrics, and several strong models. Although high, the best scores (81% GENIE, 61.4% F1, 63.2% ROUGE-L) still leave considerable room for improvement. In addition, the dataset provides a natural setting for new research on explanation, as many questions require reasoning to construct answers. We hope the dataset spurs further advances in complex question-answering by the community. ARC-DA is available at https://allenai.org/data/arc-da

Oyvind Tafjord, Bhavana Dalvi Mishra, Peter Clark

Transformers have been shown to emulate logical deduction over natural language theories (logical rules expressed in natural language), reliably assigning true/false labels to candidate implications. However, their ability to generate implications of a theory has not yet been demonstrated, and methods for reconstructing proofs of answers are imperfect. In this work we show that a generative model, called ProofWriter, can reliably generate both implications of a theory and the natural language proof(s) that support them. In particular, iterating a 1-step implication generator results in proofs that are highly reliable, and represent actual model decisions (rather than post-hoc rationalizations). On the RuleTaker dataset, the accuracy of ProofWriter's proofs exceed previous methods by +9% absolute, and in a way that generalizes to proof depths unseen in training and on out-of-domain problems. We also show that generative techniques can perform a type of abduction with high precision: Given a theory and an unprovable conclusion, identify a missing fact that allows the conclusion to be proved, along with a proof. These results significantly improve the viability of neural methods for systematically reasoning over natural language.

Niket Tandon, Keisuke Sakaguchi, Bhavana Dalvi Mishra et al.

We present the first dataset for tracking state changes in procedural text from arbitrary domains by using an unrestricted (open) vocabulary. For example, in a text describing fog removal using potatoes, a car window may transition between being foggy, sticky,opaque, and clear. Previous formulations of this task provide the text and entities involved,and ask how those entities change for just a small, pre-defined set of attributes (e.g., location), limiting their fidelity. Our solution is a new task formulation where given just a procedural text as input, the task is to generate a set of state change tuples(entity, at-tribute, before-state, after-state)for each step,where the entity, attribute, and state values must be predicted from an open vocabulary. Using crowdsourcing, we create OPENPI1, a high-quality (91.5% coverage as judged by humans and completely vetted), and large-scale dataset comprising 29,928 state changes over 4,050 sentences from 810 procedural real-world paragraphs from WikiHow.com. A current state-of-the-art generation model on this task achieves 16.1% F1 based on BLEU metric, leaving enough room for novel model architectures.

Aida Amini, Antoine Bosselut, Bhavana Dalvi Mishra et al.

Procedural texts often describe processes (e.g., photosynthesis and cooking) that happen over entities (e.g., light, food). In this paper, we introduce an algorithm for procedural reading comprehension by translating the text into a general formalism that represents processes as a sequence of transitions over entity attributes (e.g., location, temperature). Leveraging pre-trained language models, our model obtains entity-aware and attribute-aware representations of the text by joint prediction of entity attributes and their transitions. Our model dynamically obtains contextual encodings of the procedural text exploiting information that is encoded about previous and current states to predict the transition of a certain attribute which can be identified as a span of text or from a pre-defined set of classes. Moreover, our model achieves state of the art results on two procedural reading comprehension datasets, namely ProPara and npn-cooking

Bhavana Dalvi Mishra, Niket Tandon, Antoine Bosselut et al.

Our goal is to better comprehend procedural text, e.g., a paragraph about photosynthesis, by not only predicting what happens, but why some actions need to happen before others. Our approach builds on a prior process comprehension framework for predicting actions' effects, to also identify subsequent steps that those effects enable. We present our new model (XPAD) that biases effect predictions towards those that (1) explain more of the actions in the paragraph and (2) are more plausible with respect to background knowledge. We also extend an existing benchmark dataset for procedural text comprehension, ProPara, by adding the new task of explaining actions by predicting their dependencies. We find that XPAD significantly outperforms prior systems on this task, while maintaining the performance on the original task in ProPara. The dataset is available at http://data.allenai.org/propara

Niket Tandon, Bhavana Dalvi Mishra, Keisuke Sakaguchi et al.

We introduce WIQA, the first large-scale dataset of "What if..." questions over procedural text. WIQA contains three parts: a collection of paragraphs each describing a process, e.g., beach erosion; a set of crowdsourced influence graphs for each paragraph, describing how one change affects another; and a large (40k) collection of "What if...?" multiple-choice questions derived from the graphs. For example, given a paragraph about beach erosion, would stormy weather result in more or less erosion (or have no effect)? The task is to answer the questions, given their associated paragraph. WIQA contains three kinds of questions: perturbations to steps mentioned in the paragraph; external (out-of-paragraph) perturbations requiring commonsense knowledge; and irrelevant (no effect) perturbations. We find that state-of-the-art models achieve 73.8% accuracy, well below the human performance of 96.3%. We analyze the challenges, in particular tracking chains of influences, and present the dataset as an open challenge to the community.

Peter Clark, Oren Etzioni, Daniel Khashabi et al.

AI has achieved remarkable mastery over games such as Chess, Go, and Poker, and even Jeopardy, but the rich variety of standardized exams has remained a landmark challenge. Even in 2016, the best AI system achieved merely 59.3% on an 8th Grade science exam challenge. This paper reports unprecedented success on the Grade 8 New York Regents Science Exam, where for the first time a system scores more than 90% on the exam's non-diagram, multiple choice (NDMC) questions. In addition, our Aristo system, building upon the success of recent language models, exceeded 83% on the corresponding Grade 12 Science Exam NDMC questions. The results, on unseen test questions, are robust across different test years and different variations of this kind of test. They demonstrate that modern NLP methods can result in mastery on this task. While not a full solution to general question-answering (the questions are multiple choice, and the domain is restricted to 8th Grade science), it represents a significant milestone for the field.

Xinya Du, Bhavana Dalvi Mishra, Niket Tandon et al.

Our goal is procedural text comprehension, namely tracking how the properties of entities (e.g., their location) change with time given a procedural text (e.g., a paragraph about photosynthesis, a recipe). This task is challenging as the world is changing throughout the text, and despite recent advances, current systems still struggle with this task. Our approach is to leverage the fact that, for many procedural texts, multiple independent descriptions are readily available, and that predictions from them should be consistent (label consistency). We present a new learning framework that leverages label consistency during training, allowing consistency bias to be built into the model. Evaluation on a standard benchmark dataset for procedural text, ProPara (Dalvi et al., 2018), shows that our approach significantly improves prediction performance (F1) over prior state-of-the-art systems.

Niket Tandon, Bhavana Dalvi Mishra, Joel Grus et al.

Comprehending procedural text, e.g., a paragraph describing photosynthesis, requires modeling actions and the state changes they produce, so that questions about entities at different timepoints can be answered. Although several recent systems have shown impressive progress in this task, their predictions can be globally inconsistent or highly improbable. In this paper, we show how the predicted effects of actions in the context of a paragraph can be improved in two ways: (1) by incorporating global, commonsense constraints (e.g., a non-existent entity cannot be destroyed), and (2) by biasing reading with preferences from large-scale corpora (e.g., trees rarely move). Unlike earlier methods, we treat the problem as a neural structured prediction task, allowing hard and soft constraints to steer the model away from unlikely predictions. We show that the new model significantly outperforms earlier systems on a benchmark dataset for procedural text comprehension (+8% relative gain), and that it also avoids some of the nonsensical predictions that earlier systems make.

Bhavana Dalvi Mishra, Lifu Huang, Niket Tandon et al.

We present a new dataset and models for comprehending paragraphs about processes (e.g., photosynthesis), an important genre of text describing a dynamic world. The new dataset, ProPara, is the first to contain natural (rather than machine-generated) text about a changing world along with a full annotation of entity states (location and existence) during those changes (81k datapoints). The end-task, tracking the location and existence of entities through the text, is challenging because the causal effects of actions are often implicit and need to be inferred. We find that previous models that have worked well on synthetic data achieve only mediocre performance on ProPara, and introduce two new neural models that exploit alternative mechanisms for state prediction, in particular using LSTM input encoding and span prediction. The new models improve accuracy by up to 19%. The dataset and models are available to the community at http://data.allenai.org/propara.