Claire Cardie

Menglin Jia, Luming Tang, Bor-Chun Chen et al. · deepmind

The current modus operandi in adapting pre-trained models involves updating all the backbone parameters, ie, full fine-tuning. This paper introduces Visual Prompt Tuning (VPT) as an efficient and effective alternative to full fine-tuning for large-scale Transformer models in vision. Taking inspiration from recent advances in efficiently tuning large language models, VPT introduces only a small amount (less than 1% of model parameters) of trainable parameters in the input space while keeping the model backbone frozen. Via extensive experiments on a wide variety of downstream recognition tasks, we show that VPT achieves significant performance gains compared to other parameter efficient tuning protocols. Most importantly, VPT even outperforms full fine-tuning in many cases across model capacities and training data scales, while reducing per-task storage cost.

Yuntian Deng, Wenting Zhao, Jack Hessel et al. · allen-ai

The increasing availability of real-world conversation data offers exciting opportunities for researchers to study user-chatbot interactions. However, the sheer volume of this data makes manually examining individual conversations impractical. To overcome this challenge, we introduce WildVis, an interactive tool that enables fast, versatile, and large-scale conversation analysis. WildVis provides search and visualization capabilities in the text and embedding spaces based on a list of criteria. To manage million-scale datasets, we implemented optimizations including search index construction, embedding precomputation and compression, and caching to ensure responsive user interactions within seconds. We demonstrate WildVis' utility through three case studies: facilitating chatbot misuse research, visualizing and comparing topic distributions across datasets, and characterizing user-specific conversation patterns. WildVis is open-source and designed to be extendable, supporting additional datasets and customized search and visualization functionalities.

Aliva Das, Xinya Du, Barry Wang et al. · cmu

Document-level information extraction (IE) tasks have recently begun to be revisited in earnest using the end-to-end neural network techniques that have been successful on their sentence-level IE counterparts. Evaluation of the approaches, however, has been limited in a number of dimensions. In particular, the precision/recall/F1 scores typically reported provide few insights on the range of errors the models make. We build on the work of Kummerfeld and Klein (2013) to propose a transformation-based framework for automating error analysis in document-level event and (N-ary) relation extraction. We employ our framework to compare two state-of-the-art document-level template-filling approaches on datasets from three domains; and then, to gauge progress in IE since its inception 30 years ago, vs. four systems from the MUC-4 (1992) evaluation.

Wenting Zhao, Tanya Goyal, Yu Ying Chiu et al. · cmu, uw

While hallucinations of large language models (LLMs) prevail as a major challenge, existing evaluation benchmarks on factuality do not cover the diverse domains of knowledge that the real-world users of LLMs seek information about. To bridge this gap, we introduce WildHallucinations, a benchmark that evaluates factuality. It does so by prompting LLMs to generate information about entities mined from user-chatbot conversations in the wild. These generations are then automatically fact-checked against a systematically curated knowledge source collected from web search. Notably, half of these real-world entities do not have associated Wikipedia pages. We evaluate 118,785 generations from 15 LLMs on 7,919 entities. We find that LLMs consistently hallucinate more on entities without Wikipedia pages and exhibit varying hallucination rates across different domains. Finally, given the same base models, adding a retrieval component only slightly reduces hallucinations but does not eliminate hallucinations.

Wenting Zhao, Konstantine Arkoudas, Weiqi Sun et al.

Task-oriented parsing (TOP) aims to convert natural language into machine-readable representations of specific tasks, such as setting an alarm. A popular approach to TOP is to apply seq2seq models to generate linearized parse trees. A more recent line of work argues that pretrained seq2seq models are better at generating outputs that are themselves natural language, so they replace linearized parse trees with canonical natural-language paraphrases that can then be easily translated into parse trees, resulting in so-called naturalized parsers. In this work we continue to explore naturalized semantic parsing by presenting a general reduction of TOP to abstractive question answering that overcomes some limitations of canonical paraphrasing. Experimental results show that our QA-based technique outperforms state-of-the-art methods in full-data settings while achieving dramatic improvements in few-shot settings.

Jinyan Su, Claire Cardie, Preslav Nakov

In the age of large language models (LLMs) and the widespread adoption of AI-driven content creation, the landscape of information dissemination has witnessed a paradigm shift. With the proliferation of both human-written and machine-generated real and fake news, robustly and effectively discerning the veracity of news articles has become an intricate challenge. While substantial research has been dedicated to fake news detection, this either assumes that all news articles are human-written or abruptly assumes that all machine-generated news are fake. Thus, a significant gap exists in understanding the interplay between machine-(paraphrased) real news, machine-generated fake news, human-written fake news, and human-written real news. In this paper, we study this gap by conducting a comprehensive evaluation of fake news detectors trained in various scenarios. Our primary objectives revolve around the following pivotal question: How to adapt fake news detectors to the era of LLMs? Our experiments reveal an interesting pattern that detectors trained exclusively on human-written articles can indeed perform well at detecting machine-generated fake news, but not vice versa. Moreover, due to the bias of detectors against machine-generated texts \cite{su2023fake}, they should be trained on datasets with a lower machine-generated news ratio than the test set. Building on our findings, we provide a practical strategy for the development of robust fake news detectors.

Barry Wang, Xinya Du, Claire Cardie · cmu

The probing classifiers framework has been employed for interpreting deep neural network models for a variety of natural language processing (NLP) applications. Studies, however, have largely focused on sentencelevel NLP tasks. This work is the first to apply the probing paradigm to representations learned for document-level information extraction (IE). We designed eight embedding probes to analyze surface, semantic, and event-understanding capabilities relevant to document-level event extraction. We apply them to the representations acquired by learning models from three different LLM-based document-level IE approaches on a standard dataset. We found that trained encoders from these models yield embeddings that can modestly improve argument detections and labeling but only slightly enhance event-level tasks, albeit trade-offs in information helpful for coherence and event-type prediction. We further found that encoder models struggle with document length and cross-sentence discourse.

Wenting Zhao, Ge Gao, Claire Cardie et al.

When seeking information from unfamiliar documents, users frequently pose questions that cannot be answered by the documents. While existing large language models (LLMs) identify these unanswerable questions, they do not assist users in reformulating their questions, thereby reducing their overall utility. We curate CouldAsk, an evaluation benchmark composed of existing and new datasets for document-grounded question answering, specifically designed to study reformulating unanswerable questions. We evaluate state-of-the-art open-source and proprietary LLMs on CouldAsk. The results demonstrate the limited capabilities of these models in reformulating questions. Specifically, GPT-4 and Llama2-7B successfully reformulate questions only 26% and 12% of the time, respectively. Error analysis shows that 62% of the unsuccessful reformulations stem from the models merely rephrasing the questions or even generating identical questions. We publicly release the benchmark and the code to reproduce the experiments.

Jinyan Su, Claire Cardie

User queries are often underspecified and may admit multiple valid interpretations. Rather than silently making assumptions about the user's intent, a helpful assistant should surface such ambiguity by asking a clarifying question. Doing so requires two abilities: recognizing that a query is ambiguous, and acting on that recognition by seeking clarification instead of answering directly. To study these abilities, we evaluate models on ambiguous, unambiguous, and disambiguated questions in three settings: standard question answering, explicit ambiguity judgment, and behavioral analysis, where a judge model classifies responses as direct answers, refusals, or clarifying questions. We find a clear gap between recognition and behavior: models often identify ambiguity when explicitly asked to judge it, yet in the QA setting they overwhelmingly default to direct answers. Retrieved context further widens this gap by improving answerability while making models even less likely to ask clarifying questions.

Chengyu Huang, Sheng-Yen Chou, Zhengxin Zhang et al.

Self-play has recently emerged as a promising paradigm to train Large Language Models (LLMs). In self-play, the target LLM creates the task input (e.g., ask a question), which it then addresses itself by producing a task output (e.g., give an answer). A reward model evaluates the output, and the rewards are then used to train the LLM, typically via Reinforcement Learning (RL). Self-play incurs minimal supervision costs, and this is especially helpful for post-training LLMs, which require high-quality input-output pairs that traditionally have to be written by humans or expensive proprietary models. However, existing work explores self-play only for verifiable tasks such as math and coding. Instead, we seek to extend it to more realistic open-ended tasks. In particular, we propose POP, a self-play framework that uses the same LLM to synthesize evaluation rubrics, along with input-output pairs, for each example. The rubric is then used to evaluate outputs and train the model. We further ground the framework on a content-rich pretraining corpus to (1) ensure a generation-verification gap and reduce reward hacking, and (2) prevent mode collapse. On Qwen-2.5-7B, POP increases performance of both pretrained and instruction-tuned models, across different tasks ranging from long-form Healthcare QA to creative writing and instruction following.

Ge Gao, Jonathan D. Chang, Claire Cardie et al.

Text retrieval plays a crucial role in incorporating factual knowledge for decision making into language processing pipelines, ranging from chat-based web search to question answering systems. Current state-of-the-art text retrieval models leverage pre-trained large language models (LLMs) to achieve competitive performance, but training LLM-based retrievers via typical contrastive losses requires intricate heuristics, including selecting hard negatives and using additional supervision as learning signals. This reliance on heuristics stems from the fact that the contrastive loss itself is heuristic and does not directly optimize the downstream metrics of decision quality at the end of the processing pipeline. To address this issue, we introduce Neural PG-RANK, a novel training algorithm that learns to rank by instantiating a LLM as a Plackett-Luce ranking policy. Neural PG-RANK provides a principled method for end-to-end training of retrieval models as part of larger decision systems via policy gradient, with little reliance on complex heuristics, and it effectively unifies the training objective with downstream decision-making quality. We conduct extensive experiments on various text retrieval benchmarks. The results demonstrate that when the training objective aligns with the evaluation setup, Neural PG-RANK yields remarkable in-domain performance improvement, with substantial out-of-domain generalization to some critical datasets employed in downstream question answering tasks.

Chengyu Huang, Zhuohang Li, Sheng-Yen Chou et al.

Direct Preference Optimization (DPO) aligns Large Language Models with human preferences without the need for a separate reward model. However, DPO treats all tokens in responses equally, neglecting the differing importance of individual tokens. Existing token-level PO methods compute the token weights using either token-position-based heuristic functions or probability estimates given by a separately trained model, which lacks robustness and incurs extra training cost. In contrast, we propose Token-weighted DPO (TwDPO) -- a novel training objective grounded on token-weighted RL -- and AttentionPO -- an instantiation of TwDPO that uses attention from the LLM itself to estimate token weights. AttentionPO prompts the LLM to serve as a pairwise judge and check where the model attends when comparing the responses. This design makes AttentionPO content-aware, adjusting weights based on response content, and efficient, incurring only two extra forward passes per example. Experiment results show that AttentionPO significantly improves performance on AlpacaEval, MT-Bench, and ArenaHard, surpassing existing Preference Optimization methods.

Zhengxin Zhang, Ning Wang, Sainyam Galhotra et al.

Recent auto-research systems can produce complete papers, but feasibility is not the same as quality, and the field still lacks a systematic study of how good agent-generated papers actually are. We introduce ResearchArena, a minimal scaffold that lets off-the-shelf agents (Claude Code using Opus 4.6, Codex using GPT-5.4, and Kimi Code using K2.5) carry out the full research loop themselves (ideation, experimentation, paper writing, self-refinement) under only lightweight guidance. Across 13 computer science seeds and 3 trials per agent-domain pair, ResearchArena yields 117 agent-generated papers, each evaluated under three complementary lenses: a manuscript-only reviewer (SAR), an artifact-aware peer review (PR) in which agents inspect the workspace alongside the manuscript, and an human conducted meta-review. Under SAR alone the picture is optimistic: Claude Code obtains the highest score, outperforms Analemma's FARS, and matches the weighted-average human ICLR 2025 submission, suggesting that minimally scaffolded agents can produce papers that look competitive on manuscript-only review. Manual inspection, however, reveals this picture is overstated: SAR scores are poorly aligned with its actual acceptance decisions and reward plausible framing without verifying experimental substance. Under artifact-aware PR scores drop sharply, and manual auditing identifies experimental rigor as the major bottleneck, decomposing into three failure modes (fabricated results, underpowered experiments, and plan/execution mismatch) that are highly agent-dependent: Codex 5%/8% paper-vs-artifact mismatch / fabricated references versus Kimi Code 77%/72%, a $\sim$15$\times$ spread that tracks distinct research personas the agents develop. None of the 117 agent-generated papers reaches the acceptance bar of a top-tier venue. This suggests that we are still gapped from the true auto-research.

Zhengxin Zhang, Chengyu Huang, Aochong Oliver Li et al.

Large Language Models (LLMs) have achieved remarkable progress through Reinforcement Learning with Verifiable Rewards (RLVR), yet still rely heavily on external supervision (e.g., curated labels). Adversarial learning, particularly through self-play, offers a promising alternative that enables models to iteratively learn from themselves - thus reducing reliance on external supervision. Dual-play extends adversarial learning by assigning specialized roles to two models and training them against each other, fostering sustained competition and mutual evolution. Despite its promise, adapting dual-play training to LLMs remains limited, largely due to their susceptibility to reward hacking and training instability. In this paper, we introduce PasoDoble, a novel LLM dual-play framework. PasoDoble adversarially trains two models initialized from the same base model: a Proposer, which generates challenging questions with ground-truth answers, and a Solver, which attempts to solve them. We enrich the Proposer with knowledge from a pre-training dataset to ensure the questions' quality and diversity. To avoid reward hacking, the Proposer is rewarded for producing only valid questions that push the Solver's limit, while the Solver is rewarded for solving them correctly, and both are updated jointly. To further enhance training stability, we introduce an optional offline paradigm that decouples Proposer and Solver updates, alternately updating each for several steps while holding the other fixed. Notably, PasoDoble operates without supervision during training. Experimental results show that PasoDoble can improve the reasoning performance of LLMs. Our project page is available at https://hcy123902.github.io/PasoDoble.

Wenting Zhao, Xiang Ren, Jack Hessel et al.

Chatbots such as GPT-4 and ChatGPT are now serving millions of users. Despite their widespread use, there remains a lack of public datasets showcasing how these tools are used by a population of users in practice. To bridge this gap, we offered free access to ChatGPT for online users in exchange for their affirmative, consensual opt-in to anonymously collect their chat transcripts and request headers. From this, we compiled WildChat, a corpus of 1 million user-ChatGPT conversations, which consists of over 2.5 million interaction turns. We compare WildChat with other popular user-chatbot interaction datasets, and find that our dataset offers the most diverse user prompts, contains the largest number of languages, and presents the richest variety of potentially toxic use-cases for researchers to study. In addition to timestamped chat transcripts, we enrich the dataset with demographic data, including state, country, and hashed IP addresses, alongside request headers. This augmentation allows for more detailed analysis of user behaviors across different geographical regions and temporal dimensions. Finally, because it captures a broad range of use cases, we demonstrate the dataset's potential utility in fine-tuning instruction-following models. WildChat is released at https://wildchat.allen.ai under AI2 ImpACT Licenses.

Jinyan Su, Jennifer Healey, Preslav Nakov et al.

Retrieval-Augmented Generation (RAG) has emerged as a powerful approach to mitigate large language model (LLM) hallucinations by incorporating external knowledge retrieval. However, existing RAG frameworks often apply retrieval indiscriminately,leading to inefficiencies-over-retrieving when unnecessary or failing to retrieve iteratively when required for complex reasoning. Recent adaptive retrieval strategies, though adaptively navigates these retrieval strategies, predict only based on query complexity and lacks user-driven flexibility, making them infeasible for diverse user application needs. In this paper, we introduce a novel user-controllable RAG framework that enables dynamic adjustment of the accuracy-cost trade-off. Our approach leverages two classifiers: one trained to prioritize accuracy and another to prioritize retrieval efficiency. Via an interpretable control parameter $α$, users can seamlessly navigate between minimal-cost retrieval and high-accuracy retrieval based on their specific requirements. We empirically demonstrate that our approach effectively balances accuracy, retrieval cost, and user controllability, making it a practical and adaptable solution for real-world applications. Code is available at https://github.com/JinyanSu1/Flare-Aug.

Shaden Shaar, Bradon Thymes, Sirawut Chaixanien et al.

Understanding real-world videos such as movies requires integrating visual and dialogue cues to answer complex questions. Yet existing VideoQA benchmarks struggle to capture this multimodal reasoning and are largely not open-ended, given the difficulty of evaluating free-form answers. In this paper, we introduce a novel open-ended multi-modal VideoQA benchmark, MovieRecapsQA created using movie recap videos--a distinctive type of YouTube content that summarizes a film by presenting its key events through synchronized visual (recap video) and textual (recap summary) modalities. Using the recap summary, we generate $\approx 8.2$ K question-answer (QA) pairs (aligned with movie-subtitles) and provide the necessary "facts" needed to verify an answer in a reference-free manner. To our knowledge, this is the first open-ended VideoQA benchmark that supplies explicit textual context of the input (video and/or text); which we use for evaluation. Our benchmark provides videos of multiple lengths (i.e., recap-segments, movie-segments) and categorizations of questions (by modality and type) to enable fine-grained analysis. We evaluate the performance of seven state-of-the-art MLLMs using our benchmark and observe that: 1) visual-only questions remain the most challenging; 2) models default to textual inputs whenever available; 3) extracting factually accurate information from video content is still difficult for all models; and 4) proprietary and open-source models perform comparably on video-dependent questions.

Mengyun Shi, Claire Cardie, Serge Belongie

Consumers are exposed to advertisements across many different domains on the internet, such as fashion, beauty, car, food, and others. On the other hand, fashion represents second highest e-commerce shopping category. Does consumer digital record behavior on various fashion ad images reveal their fashion taste? Does ads from other domains infer their fashion taste as well? In this paper, we study the correlation between advertisements and fashion taste. Towards this goal, we introduce a new dataset, Fashionpedia-Ads, which asks subjects to provide their preferences on both ad (fashion, beauty, car, and dessert) and fashion product (social network and e-commerce style) images. Furthermore, we exhaustively collect and annotate the emotional, visual and textual information on the ad images from multi-perspectives (abstractive level, physical level, captions, and brands). We open-source Fashionpedia-Ads to enable future studies and encourage more approaches to interpretability research between advertisements and fashion taste.

Menglin Jia, Bor-Chun Chen, Zuxuan Wu et al.

Neural network classifiers have become the de-facto choice for current "pre-train then fine-tune" paradigms of visual classification. In this paper, we investigate k-Nearest-Neighbor (k-NN) classifiers, a classical model-free learning method from the pre-deep learning era, as an augmentation to modern neural network based approaches. As a lazy learning method, k-NN simply aggregates the distance between the test image and top-k neighbors in a training set. We adopt k-NN with pre-trained visual representations produced by either supervised or self-supervised methods in two steps: (1) Leverage k-NN predicted probabilities as indications for easy vs. hard examples during training. (2) Linearly interpolate the k-NN predicted distribution with that of the augmented classifier. Via extensive experiments on a wide range of classification tasks, our study reveals the generality and flexibility of k-NN integration with additional insights: (1) k-NN achieves competitive results, sometimes even outperforming a standard linear classifier. (2) Incorporating k-NN is especially beneficial for tasks where parametric classifiers perform poorly and / or in low-data regimes. We hope these discoveries will encourage people to rethink the role of pre-deep learning, classical methods in computer vision. Our code is available at: https://github.com/KMnP/nn-revisit.

Kai Sun, Dian Yu, Dong Yu et al.

Reading strategies have been shown to improve comprehension levels, especially for readers lacking adequate prior knowledge. Just as the process of knowledge accumulation is time-consuming for human readers, it is resource-demanding to impart rich general domain knowledge into a deep language model via pre-training. Inspired by reading strategies identified in cognitive science, and given limited computational resources -- just a pre-trained model and a fixed number of training instances -- we propose three general strategies aimed to improve non-extractive machine reading comprehension (MRC): (i) BACK AND FORTH READING that considers both the original and reverse order of an input sequence, (ii) HIGHLIGHTING, which adds a trainable embedding to the text embedding of tokens that are relevant to the question and candidate answers, and (iii) SELF-ASSESSMENT that generates practice questions and candidate answers directly from the text in an unsupervised manner. By fine-tuning a pre-trained language model (Radford et al., 2018) with our proposed strategies on the largest general domain multiple-choice MRC dataset RACE, we obtain a 5.8% absolute increase in accuracy over the previous best result achieved by the same pre-trained model fine-tuned on RACE without the use of strategies. We further fine-tune the resulting model on a target MRC task, leading to an absolute improvement of 6.2% in average accuracy over previous state-of-the-art approaches on six representative non-extractive MRC datasets from different domains (i.e., ARC, OpenBookQA, MCTest, SemEval-2018 Task 11, ROCStories, and MultiRC). These results demonstrate the effectiveness of our proposed strategies and the versatility and general applicability of our fine-tuned models that incorporate these strategies. Core code is available at https://github.com/nlpdata/strategy/.

Jinyan Su, Jinpeng Zhou, Claire Cardie et al.

Personalized LLMs can significantly enhance user experiences by tailoring responses to preferences such as helpfulness, conciseness, and humor. However, fine-tuning models to address all possible combinations of user preferences is computationally expensive and impractical. In this paper, we introduce \textbf{CLIPer}(\textbf{Cl}assifier-guided \textbf{I}nference-time \textbf{Per}sonalization), a lightweight personalization approach that leverages a classifier model to steer LLM generation dynamically to different user preferences at inference time. Our method eliminates the need for extensive fine-tuning, inducing negligible additional computational overhead while enabling more controllable and nuanced personalization across single and multi-dimensional preferences. Comprehensive empirical analyses demonstrate the scalability and effectiveness of our approach in delivering personalized language generation.

Jinyan Su, Jennifer Healey, Preslav Nakov et al.

Large language models (LLMs) are increasingly optimized for long reasoning, under the assumption that more reasoning leads to better performance. However, emerging evidence suggests that longer responses can sometimes degrade accuracy rather than improve it. In this paper, we conduct a systematic empirical study of the relationship between reasoning length and answer correctness. We find that LLMs tend to overthink simple problems, generating unnecessarily long outputs, and underthink harder ones, failing to extend their reasoning when it is most needed. This indicates that models might misjudge problem difficulty and fail to calibrate their response length appropriately. Furthermore, we investigate the effects of length reduction with a preference optimization algorithm when simply preferring the shorter responses regardless of answer correctness. Experiments show that the generation length can be significantly reduced while maintaining acceptable accuracy. Our findings highlight generation length as a meaningful signal for reasoning behavior and motivate further exploration into LLMs' self-awareness in reasoning length adaptation.

Wenting Zhao, Nan Jiang, Celine Lee et al.

With the goal of benchmarking generative systems beyond expert software development ability, we introduce Commit0, a benchmark that challenges AI agents to write libraries from scratch. Agents are provided with a specification document outlining the library's API as well as a suite of interactive unit tests, with the goal of producing an implementation of this API accordingly. The implementation is validated through running these unit tests. As a benchmark, Commit0 is designed to move beyond static one-shot code generation towards agents that must process long-form natural language specifications, adapt to multi-stage feedback, and generate code with complex dependencies. Commit0 also offers an interactive environment where models receive static analysis and execution feedback on the code they generate. Our experiments demonstrate that while current agents can pass some unit tests, none can yet fully reproduce full libraries. Results also show that interactive feedback is quite useful for models to generate code that passes more unit tests, validating the benchmarks that facilitate its use.

Jinyan Su, Claire Cardie

Large language models (LLMs) have demonstrated strong reasoning abilities in mathematical tasks, often enhanced through reinforcement learning (RL). However, RL-trained models frequently produce unnecessarily long reasoning traces -- even for simple queries -- leading to increased inference costs and latency. While recent approaches attempt to control verbosity by adding length penalties to the reward function, these methods rely on fixed penalty terms that are hard to tune and cannot adapt as the model's reasoning capability evolves, limiting their effectiveness. In this work, we propose an adaptive reward-shaping method that enables LLMs to "think fast and right" -- producing concise outputs without sacrificing correctness. Our method dynamically adjusts the reward trade-off between accuracy and response length based on model performance: when accuracy is high, the length penalty increases to encourage faster length reduction; when accuracy drops, the penalty is relaxed to preserve correctness. This adaptive reward accelerates early-stage length reduction while avoiding over-compression in later stages. Experiments across multiple datasets show that our approach consistently and dramatically reduces reasoning length while largely maintaining accuracy, offering a new direction for cost-efficient adaptive reasoning in large-scale language models.

Zhen Wang, Fan Bai, Zhongyan Luo et al.

Autonomous agents powered by large language models (LLMs) promise to accelerate scientific discovery end-to-end, but rigorously evaluating their capacity for verifiable discovery remains a central challenge. Existing benchmarks face a trade-off: they either heavily rely on LLM-as-judge evaluations of automatically generated research outputs or optimize convenient yet isolated performance metrics that provide coarse proxies for scientific insight. To address this gap, we introduce FIRE-Bench (Full-cycle Insight Rediscovery Evaluation), a benchmark that evaluates agents through the rediscovery of established findings from recent, high-impact machine learning research. Agents are given only a high-level research question extracted from a published, verified study and must autonomously explore ideas, design experiments, implement code, execute their plans, and derive conclusions supported by empirical evidence. We evaluate a range of state-of-the-art agents with frontier LLMs backbones like gpt-5 on FIRE-Bench. Our results show that full-cycle scientific research remains challenging for current agent systems: even the strongest agents achieve limited rediscovery success (<50 F1), exhibit high variance across runs, and display recurring failure modes in experimental design, execution, and evidence-based reasoning. FIRE-Bench provides a rigorous and diagnostic framework for measuring progress toward reliable agent-driven scientific discovery.

Jingtian Wu, Claire Cardie

While large language models (LLMs) have demonstrated strong capabilities in tasks like question answering and fact verification, they continue to suffer from hallucinations and reasoning errors, especially in multi-hop tasks that require integration of multiple information sources. Current methods address these issues through retrieval-based techniques (grounding reasoning in external evidence), reasoning-based approaches (enhancing coherence via improved prompting), or hybrid strategies combining both elements. One prominent hybrid method, ReAct, has outperformed purely retrieval-based or reasoning-based approaches; however, it lacks internal verification of intermediate reasoning steps, allowing potential errors to propagate through complex reasoning tasks. In this paper, we introduce Reasoning Court (RC), a novel framework that extends iterative reasoning-and-retrieval methods, such as ReAct, with a dedicated LLM judge. Unlike ReAct, RC employs this judge to independently evaluate multiple candidate answers and their associated reasoning generated by separate LLM agents. The judge is asked to select the answer that it considers the most factually grounded and logically coherent based on the presented reasoning and evidence, or synthesizes a new answer using available evidence and its pre-trained knowledge if all candidates are inadequate, flawed, or invalid. Evaluations on multi-hop benchmarks (HotpotQA, MuSiQue) and fact-verification (FEVER) demonstrate that RC consistently outperforms state-of-the-art few-shot prompting methods without task-specific fine-tuning.

Mustafa Omer Gul, Claire Cardie, Tanya Goyal

LLMs cannot reliably recognize their parametric knowledge boundaries and often hallucinate answers to outside-of-boundary questions. In contrast, humans recognize their limitations and can either seek external help for such questions or abstain. In this paper, we introduce MASH (Modeling Abstention via Selective Help-seeking), a training framework that readily extracts abstentions from LLMs. Our key idea is that any external help-seeking by an LLM, i.e. search tool use, can serve as a proxy for abstention if the external help (search) is appropriately penalized while simultaneously rewarding answer accuracy. MASH operationalizes this idea using reinforcement learning with a pay-per-search reward. We run experiments on three knowledge-intensive QA datasets. Our results show that MASH substantially improves upon the selective help-seeking performance of prior efficient search approaches; on multi-hop datasets, MASH improves answer accuracy by 7.6%. Furthermore, MASH demonstrates strong off-the-shelf abstention -- it can distinguish between unanswerable/answerable questions and selectively generate responses for answerable questions -- showcasing behavior analogous to specialized abstention approaches. We emphasize that contrary to prior abstention methods, MASH does not require pre-determining knowledge boundaries to construct training data. Instead, MASH's abstentions are a by-product of training for the auxiliary selective help-seeking task. Overall, we show that MASH training effectively aligns search tool use with parametric knowledge, which can be successfully leveraged for making abstention decisions.

Shaden Shaar, Wayne Chen, Maitreyi Chatterjee et al. · cmu

Most existing work on event extraction has focused on sentence-level texts and presumes the identification of a trigger-span -- a word or phrase in the input that evokes the occurrence of an event of interest. Event arguments are then extracted with respect to the trigger. Indeed, triggers are treated as integral to, and trigger detection as an essential component of, event extraction. In this paper, we provide the first investigation of the role of triggers for the more difficult and much less studied task of document-level event extraction. We analyze their usefulness in multiple end-to-end and pipelined transformer-based event extraction models for three document-level event extraction datasets, measuring performance using triggers of varying quality (human-annotated, LLM-generated, keyword-based, and random). We find that whether or not systems benefit from explicitly extracting triggers depends both on dataset characteristics (i.e. the typical number of events per document) and task-specific information available during extraction (i.e. natural language event schemas). Perhaps surprisingly, we also observe that the mere existence of triggers in the input, even random ones, is important for prompt-based in-context learning approaches to the task.

Wenting Zhao, Justin T. Chiu, Claire Cardie et al.

Abductive reasoning aims to find plausible explanations for an event. This style of reasoning is critical for commonsense tasks where there are often multiple plausible explanations. Existing approaches for abductive reasoning in natural language processing (NLP) often rely on manually generated annotations for supervision; however, such annotations can be subjective and biased. Instead of using direct supervision, this work proposes an approach for abductive commonsense reasoning that exploits the fact that only a subset of explanations is correct for a given context. The method uses posterior regularization to enforce a mutual exclusion constraint, encouraging the model to learn the distinction between fluent explanations and plausible ones. We evaluate our approach on a diverse set of abductive reasoning datasets; experimental results show that our approach outperforms or is comparable to directly applying pretrained language models in a zero-shot manner and other knowledge-augmented zero-shot methods.

Wenting Zhao, Justin T. Chiu, Claire Cardie et al.

Explainable multi-hop question answering (QA) not only predicts answers but also identifies rationales, i. e. subsets of input sentences used to derive the answers. This problem has been extensively studied under the supervised setting, where both answer and rationale annotations are given. Because rationale annotations are expensive to collect and not always available, recent efforts have been devoted to developing methods that do not rely on supervision for rationales. However, such methods have limited capacities in modeling interactions between sentences, let alone reasoning across multiple documents. This work proposes a principled, probabilistic approach for training explainable multi-hop QA systems without rationale supervision. Our approach performs multi-hop reasoning by explicitly modeling rationales as sets, enabling the model to capture interactions between documents and sentences within a document. Experimental results show that our approach is more accurate at selecting rationales than the previous methods, while maintaining similar accuracy in predicting answers.

Mengyun Shi, Serge Belongie, Claire Cardie

Existing fashion datasets do not consider the multi-facts that cause a consumer to like or dislike a fashion image. Even two consumers like a same fashion image, they could like this image for total different reasons. In this paper, we study the reason why a consumer like a certain fashion image. Towards this goal, we introduce an interpretability dataset, Fashionpedia-taste, consist of rich annotation to explain why a subject like or dislike a fashion image from the following 3 perspectives: 1) localized attributes; 2) human attention; 3) caption. Furthermore, subjects are asked to provide their personal attributes and preference on fashion, such as personality and preferred fashion brands. Our dataset makes it possible for researchers to build computational models to fully understand and interpret human fashion taste from different humanistic perspectives and modalities.

Menglin Jia, Austin Reiter, Ser-Nam Lim et al.

We introduce Classification with Alternating Normalization (CAN), a non-parametric post-processing step for classification. CAN improves classification accuracy for challenging examples by re-adjusting their predicted class probability distribution using the predicted class distributions of high-confidence validation examples. CAN is easily applicable to any probabilistic classifier, with minimal computation overhead. We analyze the properties of CAN using simulated experiments, and empirically demonstrate its effectiveness across a diverse set of classification tasks.

Faisal Ladhak, Esin Durmus, He He et al.

Despite recent progress in abstractive summarization, systems still suffer from faithfulness errors. While prior work has proposed models that improve faithfulness, it is unclear whether the improvement comes from an increased level of extractiveness of the model outputs as one naive way to improve faithfulness is to make summarization models more extractive. In this work, we present a framework for evaluating the effective faithfulness of summarization systems, by generating a faithfulnessabstractiveness trade-off curve that serves as a control at different operating points on the abstractiveness spectrum. We then show that the Maximum Likelihood Estimation (MLE) baseline as well as a recently proposed method for improving faithfulness, are both worse than the control at the same level of abstractiveness. Finally, we learn a selector to identify the most faithful and abstractive summary for a given document, and show that this system can attain higher faithfulness scores in human evaluations while being more abstractive than the baseline system on two datasets. Moreover, we show that our system is able to achieve a better faithfulness-abstractiveness trade-off than the control at the same level of abstractiveness.

Menglin Jia, Zuxuan Wu, Austin Reiter et al.

Visual engagement in social media platforms comprises interactions with photo posts including comments, shares, and likes. In this paper, we leverage such visual engagement clues as supervisory signals for representation learning. However, learning from engagement signals is non-trivial as it is not clear how to bridge the gap between low-level visual information and high-level social interactions. We present VisE, a weakly supervised learning approach, which maps social images to pseudo labels derived by clustered engagement signals. We then study how models trained in this way benefit subjective downstream computer vision tasks such as emotion recognition or political bias detection. Through extensive studies, we empirically demonstrate the effectiveness of VisE across a diverse set of classification tasks beyond the scope of conventional recognition.

Dian Yu, Kai Sun, Dong Yu et al.

In spite of much recent research in the area, it is still unclear whether subject-area question-answering data is useful for machine reading comprehension (MRC) tasks. In this paper, we investigate this question. We collect a large-scale multi-subject multiple-choice question-answering dataset, ExamQA, and use incomplete and noisy snippets returned by a web search engine as the relevant context for each question-answering instance to convert it into a weakly-labeled MRC instance. We then propose a self-teaching paradigm to better use the generated weakly-labeled MRC instances to improve a target MRC task. Experimental results show that we can obtain +5.1% in accuracy on a multiple-choice MRC dataset, C^3, and +3.8% in exact match on an extractive MRC dataset, CMRC 2018 over state-of-the-art MRC baselines, demonstrating the effectiveness of our framework and the usefulness of large-scale subject-area question-answering data for different types of machine reading comprehension tasks.

Menglin Jia, Zuxuan Wu, Austin Reiter et al.

An image is worth a thousand words, conveying information that goes beyond the physical visual content therein. In this paper, we study the intent behind social media images with an aim to analyze how visual information can help the recognition of human intent. Towards this goal, we introduce an intent dataset, Intentonomy, comprising 14K images covering a wide range of everyday scenes. These images are manually annotated with 28 intent categories that are derived from a social psychology taxonomy. We then systematically study whether, and to what extent, commonly used visual information, i.e., object and context, contribute to human motive understanding. Based on our findings, we conduct further study to quantify the effect of attending to object and context classes as well as textual information in the form of hashtags when training an intent classifier. Our results quantitatively and qualitatively shed light on how visual and textual information can produce observable effects when predicting intent.

Zonglin Yang, Xinya Du, Alexander Rush et al.

End-to-end models in NLP rarely encode external world knowledge about length of time. We introduce two effective models for duration prediction, which incorporate external knowledge by reading temporal-related news sentences (time-aware pre-training). Specifically, one model predicts the range/unit where the duration value falls in (R-pred); and the other predicts the exact duration value E-pred. Our best model -- E-pred, substantially outperforms previous work, and captures duration information more accurately than R-pred. We also demonstrate our models are capable of duration prediction in the unsupervised setting, outperforming the baselines.

Kai Sun, Seungwhan Moon, Paul Crook et al.

Existing dialogue corpora and models are typically designed under two disjoint motives: while task-oriented systems focus on achieving functional goals (e.g., booking hotels), open-domain chatbots aim at making socially engaging conversations. In this work, we propose to integrate both types of systems by Adding Chit-Chat to ENhance Task-ORiented dialogues (ACCENTOR), with the goal of making virtual assistant conversations more engaging and interactive. Specifically, we propose a Human <-> AI collaborative data collection approach for generating diverse chit-chat responses to augment task-oriented dialogues with minimal annotation effort. We then present our new chit-chat-based annotations to 23.8K dialogues from two popular task-oriented datasets (Schema-Guided Dialogue and MultiWOZ 2.1) and demonstrate their advantage over the originals via human evaluation. Lastly, we propose three new models for adding chit-chat to task-oriented dialogues, explicitly trained to predict user goals and to generate contextually relevant chit-chat responses. Automatic and human evaluations show that, compared with the state-of-the-art task-oriented baseline, our models can code-switch between task and chit-chat to be more engaging, interesting, knowledgeable, and humanlike, while maintaining competitive task performance.

Jialu Li, Esin Durmus, Claire Cardie

Online debate forums provide users a platform to express their opinions on controversial topics while being exposed to opinions from diverse set of viewpoints. Existing work in Natural Language Processing (NLP) has shown that linguistic features extracted from the debate text and features encoding the characteristics of the audience are both critical in persuasion studies. In this paper, we aim to further investigate the role of discourse structure of the arguments from online debates in their persuasiveness. In particular, we use the factor graph model to obtain features for the argument structure of debates from an online debating platform and incorporate these features to an LSTM-based model to predict the debater that makes the most convincing arguments. We find that incorporating argument structure features play an essential role in achieving the better predictive performance in assessing the persuasiveness of the arguments in online debates.

Faisal Ladhak, Esin Durmus, Claire Cardie et al.

We introduce WikiLingua, a large-scale, multilingual dataset for the evaluation of crosslingual abstractive summarization systems. We extract article and summary pairs in 18 languages from WikiHow, a high quality, collaborative resource of how-to guides on a diverse set of topics written by human authors. We create gold-standard article-summary alignments across languages by aligning the images that are used to describe each how-to step in an article. As a set of baselines for further studies, we evaluate the performance of existing cross-lingual abstractive summarization methods on our dataset. We further propose a method for direct crosslingual summarization (i.e., without requiring translation at inference time) by leveraging synthetic data and Neural Machine Translation as a pre-training step. Our method significantly outperforms the baseline approaches, while being more cost efficient during inference.

Kai Sun, Dian Yu, Jianshu Chen et al.

In this paper, we aim to extract commonsense knowledge to improve machine reading comprehension. We propose to represent relations implicitly by situating structured knowledge in a context instead of relying on a pre-defined set of relations, and we call it contextualized knowledge. Each piece of contextualized knowledge consists of a pair of interrelated verbal and nonverbal messages extracted from a script and the scene in which they occur as context to implicitly represent the relation between the verbal and nonverbal messages, which are originally conveyed by different modalities within the script. We propose a two-stage fine-tuning strategy to use the large-scale weakly-labeled data based on a single type of contextualized knowledge and employ a teacher-student paradigm to inject multiple types of contextualized knowledge into a student machine reader. Experimental results demonstrate that our method outperforms a state-of-the-art baseline by a 4.3% improvement in accuracy on the machine reading comprehension dataset C^3, wherein most of the questions require unstated prior knowledge.

Xinya Du, Alexander M. Rush, Claire Cardie

We revisit the classic problem of document-level role-filler entity extraction (REE) for template filling. We argue that sentence-level approaches are ill-suited to the task and introduce a generative transformer-based encoder-decoder framework (GRIT) that is designed to model context at the document level: it can make extraction decisions across sentence boundaries; is implicitly aware of noun phrase coreference structure, and has the capacity to respect cross-role dependencies in the template structure. We evaluate our approach on the MUC-4 dataset, and show that our model performs substantially better than prior work. We also show that our modeling choices contribute to model performance, e.g., by implicitly capturing linguistic knowledge such as recognizing coreferent entity mentions.

Xinya Du, Claire Cardie

Few works in the literature of event extraction have gone beyond individual sentences to make extraction decisions. This is problematic when the information needed to recognize an event argument is spread across multiple sentences. We argue that document-level event extraction is a difficult task since it requires a view of a larger context to determine which spans of text correspond to event role fillers. We first investigate how end-to-end neural sequence models (with pre-trained language model representations) perform on document-level role filler extraction, as well as how the length of context captured affects the models' performance. To dynamically aggregate information captured by neural representations learned at different levels of granularity (e.g., the sentence- and paragraph-level), we propose a novel multi-granularity reader. We evaluate our models on the MUC-4 event extraction dataset, and show that our best system performs substantially better than prior work. We also report findings on the relationship between context length and neural model performance on the task.

Xinya Du, Claire Cardie

The problem of event extraction requires detecting the event trigger and extracting its corresponding arguments. Existing work in event argument extraction typically relies heavily on entity recognition as a preprocessing/concurrent step, causing the well-known problem of error propagation. To avoid this issue, we introduce a new paradigm for event extraction by formulating it as a question answering (QA) task that extracts the event arguments in an end-to-end manner. Empirical results demonstrate that our framework outperforms prior methods substantially; in addition, it is capable of extracting event arguments for roles not seen at training time (zero-shot learning setting).

Menglin Jia, Mengyun Shi, Mikhail Sirotenko et al.

In this work we explore the task of instance segmentation with attribute localization, which unifies instance segmentation (detect and segment each object instance) and fine-grained visual attribute categorization (recognize one or multiple attributes). The proposed task requires both localizing an object and describing its properties. To illustrate the various aspects of this task, we focus on the domain of fashion and introduce Fashionpedia as a step toward mapping out the visual aspects of the fashion world. Fashionpedia consists of two parts: (1) an ontology built by fashion experts containing 27 main apparel categories, 19 apparel parts, 294 fine-grained attributes and their relationships; (2) a dataset with everyday and celebrity event fashion images annotated with segmentation masks and their associated per-mask fine-grained attributes, built upon the Fashionpedia ontology. In order to solve this challenging task, we propose a novel Attribute-Mask RCNN model to jointly perform instance segmentation and localized attribute recognition, and provide a novel evaluation metric for the task. We also demonstrate instance segmentation models pre-trained on Fashionpedia achieve better transfer learning performance on other fashion datasets than ImageNet pre-training. Fashionpedia is available at: https://fashionpedia.github.io/home/index.html.

Dian Yu, Kai Sun, Claire Cardie et al.

We present the first human-annotated dialogue-based relation extraction (RE) dataset DialogRE, aiming to support the prediction of relation(s) between two arguments that appear in a dialogue. We further offer DialogRE as a platform for studying cross-sentence RE as most facts span multiple sentences. We argue that speaker-related information plays a critical role in the proposed task, based on an analysis of similarities and differences between dialogue-based and traditional RE tasks. Considering the timeliness of communication in a dialogue, we design a new metric to evaluate the performance of RE methods in a conversational setting and investigate the performance of several representative RE methods on DialogRE. Experimental results demonstrate that a speaker-aware extension on the best-performing model leads to gains in both the standard and conversational evaluation settings. DialogRE is available at https://dataset.org/dialogre/.

Esin Durmus, Faisal Ladhak, Claire Cardie

Research in the social sciences and psychology has shown that the persuasiveness of an argument depends not only the language employed, but also on attributes of the source/communicator, the audience, and the appropriateness and strength of the argument's claims given the pragmatic and discourse context of the argument. Among these characteristics of persuasive arguments, prior work in NLP does not explicitly investigate the effect of the pragmatic and discourse context when determining argument quality. This paper presents a new dataset to initiate the study of this aspect of argumentation: it consists of a diverse collection of arguments covering 741 controversial topics and comprising over 47,000 claims. We further propose predictive models that incorporate the pragmatic and discourse context of argumentative claims and show that they outperform models that rely only on claim-specific linguistic features for predicting the perceived impact of individual claims within a particular line of argument.

Esin Durmus, Faisal Ladhak, Claire Cardie

Systems for automatic argument generation and debate require the ability to (1) determine the stance of any claims employed in the argument and (2) assess the specificity of each claim relative to the argument context. Existing work on understanding claim specificity and stance, however, has been limited to the study of argumentative structures that are relatively shallow, most often consisting of a single claim that directly supports or opposes the argument thesis. In this paper, we tackle these tasks in the context of complex arguments on a diverse set of topics. In particular, our dataset consists of manually curated argument trees for 741 controversial topics covering 95,312 unique claims; lines of argument are generally of depth 2 to 6. We find that as the distance between a pair of claims increases along the argument path, determining the relative specificity of a pair of claims becomes easier and determining their relative stance becomes harder.

Esin Durmus, Claire Cardie

Existing argumentation datasets have succeeded in allowing researchers to develop computational methods for analyzing the content, structure and linguistic features of argumentative text. They have been much less successful in fostering studies of the effect of "user" traits -- characteristics and beliefs of the participants -- on the debate/argument outcome as this type of user information is generally not available. This paper presents a dataset of 78, 376 debates generated over a 10-year period along with surprisingly comprehensive participant profiles. We also complete an example study using the dataset to analyze the effect of selected user traits on the debate outcome in comparison to the linguistic features typically employed in studies of this kind.

Esin Durmus, Claire Cardie

Public debate forums provide a common platform for exchanging opinions on a topic of interest. While recent studies in natural language processing (NLP) have provided empirical evidence that the language of the debaters and their patterns of interaction play a key role in changing the mind of a reader, research in psychology has shown that prior beliefs can affect our interpretation of an argument and could therefore constitute a competing alternative explanation for resistance to changing one's stance. To study the actual effect of language use vs. prior beliefs on persuasion, we provide a new dataset and propose a controlled setting that takes into consideration two reader level factors: political and religious ideology. We find that prior beliefs affected by these reader level factors play a more important role than language use effects and argue that it is important to account for them in NLP studies of persuasion.