Eduard Hovy

AAAI Conference 2026 Conference Paper

Control Illusion: The Failure of Instruction Hierarchies in Large Language Models

• Yilin Geng
• Haonan Li
• Honglin Mu
• Xudong Han
• Timothy Baldwin
• Omri Abend
• Eduard Hovy
• Lea Frermann

Large language models (LLMs) are increasingly deployed with hierarchical instruction schemes, where certain instructions (e.g., system-level directives) are expected to take precedence over others (e.g., user messages). Yet, we lack a systematic understanding of how effectively these hierarchical control mechanisms work. We introduce a systematic evaluation framework based on constraint prioritization to assess how well LLMs enforce instruction hierarchies. Our experiments across six state-of-the-art LLMs reveal that models struggle with consistent instruction prioritization, even for simple formatting conflicts. We find that the widely-adopted system/user prompt separation fails to establish a reliable instruction hierarchy, and models exhibit strong inherent biases toward certain constraint types regardless of their priority designation. Interestingly, we also find that societal hierarchy framings (e.g., authority, expertise, consensus) show stronger influence on model behavior than system/user roles, suggesting that pretraining-derived social structures function as latent behavioral priors with potentially greater impact than post-training guardrails.

NeurIPS Conference 2025 Conference Paper

What is Your Data Worth to GPT? LLM-Scale Data Valuation with Influence Functions

• Sang Choe
• Hwijeen Ahn
• Juhan Bae
• Kewen Zhao
• Youngseog Chung
• Adithya Pratapa
• Willie Neiswanger
• Emma Strubell

Large language models (LLMs) are trained on a vast amount of human-written data, but data providers often remain uncredited. In response to this issue, data valuation (or data attribution), which quantifies the contribution or value of each data to the model output, has been discussed as a potential solution. Nevertheless, applying existing data valuation methods to recent LLMs and their vast training datasets has been largely limited by prohibitive compute and memory costs. In this work, we focus on influence functions, a popular gradient-based data valuation method, and significantly improve its scalability with an efficient gradient projection strategy called LoGra that leverages the gradient structure in backpropagation. We then provide a theoretical motivation of gradient projection approaches to influence functions to promote trust in the data valuation process. Lastly, we lower the barrier to implementing data valuation systems by introducing LogIX, a software package that can transform existing training code into data valuation code with minimal effort. In our data valuation experiments, LoGra achieves competitive accuracy against more expensive baselines while showing up to 6, 500x improvement in throughput and 5x reduction in GPU memory usage when applied to Llama3-8B-Instruct and the 1B-token dataset.

AAAI Conference 2022 Conference Paper

NAREOR: The Narrative Reordering Problem

• Varun Gangal
• Steven Y. Feng
• Malihe Alikhani
• Teruko Mitamura
• Eduard Hovy

Many implicit inferences exist in text depending on how it is structured that can critically impact the text’s interpretation and meaning. One such structural aspect present in text with chronology is the order of its presentation. For narratives or stories, this is known as the narrative order. Reordering a narrative can impact the temporal, causal, event-based, and other inferences readers draw from it, which in turn can have strong effects both on its interpretation and interestingness. In this paper, we propose and investigate the task of Narrative Reordering (NAREOR) which involves rewriting a given story in a different narrative order while preserving its plot. We present a dataset, NAREORC, with human rewritings of stories within ROCStories in non-linear orders, and conduct a detailed analysis of it. Further, we propose novel task-specific training methods with suitable evaluation metrics. We perform experiments on NAREORC using state-of-the-art models such as BART and T5 and conduct extensive automatic and human evaluations. We demonstrate that although our models can perform decently, NAREOR is a challenging task with potential for further exploration. We also investigate two applications of NAREOR: generation of more interesting variations of stories and serving as adversarial sets for temporal/event-related tasks, besides discussing other prospective ones, such as for pedagogical setups related to language skills like essay writing and applications to medicine involving clinical narratives.

AAAI Conference 2022 Conference Paper

Retrieve, Caption, Generate: Visual Grounding for Enhancing Commonsense in Text Generation Models

• Steven Y. Feng
• Kevin Lu
• Zhuofu Tao
• Malihe Alikhani
• Teruko Mitamura
• Eduard Hovy
• Varun Gangal

We investigate the use of multimodal information contained in images as an effective method for enhancing the commonsense of Transformer models for text generation. We perform experiments using BART and T5 on concept-to-text generation, specifically the task of generative commonsense reasoning, or CommonGen. We call our approach VisCTG: Visually Grounded Concept-to-Text Generation. VisCTG involves captioning images representing appropriate everyday scenarios, and using these captions to enrich and steer the generation process. Comprehensive evaluation and analysis demonstrate that VisCTG noticeably improves model performance while successfully addressing several issues of the baseline generations, including poor commonsense, fluency, and specificity.

NeurIPS Conference 2020 Conference Paper

Unsupervised Data Augmentation for Consistency Training

• Qizhe Xie
• Zihang Dai
• Eduard Hovy
• Thang Luong
• Quoc Le

Semi-supervised learning lately has shown much promise in improving deep learning models when labeled data is scarce. Common among recent approaches is the use of consistency training on a large amount of unlabeled data to constrain model predictions to be invariant to input noise. In this work, we present a new perspective on how to effectively noise unlabeled examples and argue that the quality of noising, specifically those produced by advanced data augmentation methods, plays a crucial role in semi-supervised learning. By substituting simple noising operations with advanced data augmentation methods such as RandAugment and back-translation, our method brings substantial improvements across six language and three vision tasks under the same consistency training framework. On the IMDb text classification dataset, with only 20 labeled examples, our method achieves an error rate of 4. 20, outperforming the state-of-the-art model trained on 25, 000 labeled examples. On a standard semi-supervised learning benchmark, CIFAR-10, our method outperforms all previous approaches and achieves an error rate of 5. 43 with only 250 examples. Our method also combines well with transfer learning, e. g. , when finetuning from BERT, and yields improvements in high-data regime, such as ImageNet, whether when there is only 10% labeled data or when a full labeled set with 1. 3M extra unlabeled examples is used. Code is available at https: //github. com/google-research/uda.

AAAI Conference 2019 Conference Paper

Fast and Simple Mixture of Softmaxes with BPE and Hybrid-LightRNN for Language Generation

• Xiang Kong
• Qizhe Xie
• Zihang Dai
• Eduard Hovy

Mixture of Softmaxes (MoS) has been shown to be effective at addressing the expressiveness limitation of Softmax-based models. Despite the known advantage, MoS is practically sealed by its large consumption of memory and computational time due to the need of computing multiple Softmaxes. In this work, we set out to unleash the power of MoS in practical applications by investigating improved word coding schemes, which could effectively reduce the vocabulary size and hence relieve the memory and computation burden. We show both BPE and our proposed Hybrid-LightRNN lead to improved encoding mechanisms that can halve the time and memory consumption of MoS without performance losses. With MoS, we achieve an improvement of 1. 5 BLEU scores on IWSLT 2014 German-to-English corpus and an improvement of 0. 76 CIDEr score on image captioning. Moreover, on the larger WMT 2014 machine translation dataset, our MoSboosted Transformer yields 29. 6 BLEU score for English-to- German and 42. 1 BLEU score for English-to-French, outperforming the single-Softmax Transformer by 0. 9 and 0. 4 BLEU scores respectively and achieving the state-of-the-art result on WMT 2014 English-to-German task.

NeurIPS Conference 2019 Conference Paper

MaCow: Masked Convolutional Generative Flow

• Xuezhe Ma
• Xiang Kong
• Shanghang Zhang
• Eduard Hovy

Flow-based generative models, conceptually attractive due to tractability of both the exact log-likelihood computation and latent-variable inference, and efficiency of both training and sampling, has led to a number of impressive empirical successes and spawned many advanced variants and theoretical investigations. Despite their computational efficiency, the density estimation performance of flow-based generative models significantly falls behind those of state-of-the-art autoregressive models. In this work, we introduce masked convolutional generative flow (MaCow), a simple yet effective architecture of generative flow using masked convolution. By restricting the local connectivity in a small kernel, MaCow enjoys the properties of fast and stable training, and efficient sampling, while achieving significant improvements over Glow for density estimation on standard image benchmarks, considerably narrowing the gap to autoregressive models.

AAAI Conference 2019 Conference Paper

Neural Machine Translation with Adequacy-Oriented Learning

• Xiang Kong
• Zhaopeng Tu
• Shuming Shi
• Eduard Hovy
• Tong Zhang

Although Neural Machine Translation (NMT) models have advanced state-of-the-art performance in machine translation, they face problems like the inadequate translation. We attribute this to that the standard Maximum Likelihood Estimation (MLE) cannot judge the real translation quality due to its several limitations. In this work, we propose an adequacyoriented learning mechanism for NMT by casting translation as a stochastic policy in Reinforcement Learning (RL), where the reward is estimated by explicitly measuring translation adequacy. Benefiting from the sequence-level training of RL strategy and a more accurate reward designed specifically for translation, our model outperforms multiple strong baselines, including (1) standard and coverage-augmented attention models with MLE-based training, and (2) advanced reinforcement and adversarial training strategies with rewards based on both word-level BLEU and character-level CHRF3. Quantitative and qualitative analyses on different language pairs and NMT architectures demonstrate the effectiveness and universality of the proposed approach.

AAAI Conference 2018 Conference Paper

SPINE: SParse Interpretable Neural Embeddings

• Anant Subramanian
• Danish Pruthi
• Harsh Jhamtani
• Taylor Berg-Kirkpatrick
• Eduard Hovy

Prediction without justification has limited utility. Much of the success of neural models can be attributed to their ability to learn rich, dense and expressive representations. While these representations capture the underlying complexity and latent trends in the data, they are far from being interpretable. We propose a novel variant of denoising k-sparse autoencoders that generates highly efficient and interpretable distributed word representations (word embeddings), beginning with existing word representations from state-of-the-art methods like GloVe and word2vec. Through large scale human evaluation, we report that our resulting word embedddings are much more interpretable than the original GloVe and word2vec embeddings. Moreover, our embeddings outperform existing popular word embeddings on a diverse suite of benchmark downstream tasks1.

NeurIPS Conference 2017 Conference Paper

Controllable Invariance through Adversarial Feature Learning

• Qizhe Xie
• Zihang Dai
• Yulun Du
• Eduard Hovy
• Graham Neubig

Learning meaningful representations that maintain the content necessary for a particular task while filtering away detrimental variations is a problem of great interest in machine learning. In this paper, we tackle the problem of learning representations invariant to a specific factor or trait of data. The representation learning process is formulated as an adversarial minimax game. We analyze the optimal equilibrium of such a game and find that it amounts to maximizing the uncertainty of inferring the detrimental factor given the representation while maximizing the certainty of making task-specific predictions. On three benchmark tasks, namely fair and bias-free classification, language-independent generation, and lighting-independent image classification, we show that the proposed framework induces an invariant representation, and leads to better generalization evidenced by the improved performance.

IJCAI Conference 2015 Conference Paper

An Active Learning Approach to Coreference Resolution

• Mrinmaya Sachan
• Eduard Hovy
• Eric P. Xing

In this paper, we define the problem of coreference resolution in text as one of clustering with pairwise constraints where human experts are asked to provide pairwise constraints (pairwise judgments of coreferentiality) to guide the clustering process. Positing that these pairwise judgments are easy to obtain from humans given the right context, we show that with significantly lower number of pairwise judgments and feature-engineering effort, we can achieve competitive coreference performance. Further, we describe an active learning strategy that minimizes the overall number of such pairwise judgments needed by asking the most informative questions to human experts at each step of coreference resolution. We evaluate this hypothesis and our algorithms on both entity and event coreference tasks and on two languages.

AIJ Journal 2013 Journal Article

Collaboratively built semi-structured content and Artificial Intelligence: The story so far

• Eduard Hovy
• Roberto Navigli
• Simone Paolo Ponzetto

Recent years have seen a great deal of work that exploits collaborative, semi-structured content for Artificial Intelligence (AI) and Natural Language Processing (NLP). This special issue of the Artificial Intelligence Journal presents a variety of state-of-the-art contributions, each of which illustrates the substantial impact that work on leveraging semi-structured content is having on AI and NLP as it continuously fosters new directions of cutting-edge research. We contextualize the papers collected in this special issue by providing a detailed overview of previous work on collaborative, semi-structured resources. The survey is made up of two main logical parts: in the first part, we present the main characteristics of collaborative resources that make them attractive for AI and NLP research; in the second part, we present an overview of how these features have been exploited to tackle a variety of long-standing issues in the two fields, in particular the acquisition of large amounts of machine-readable knowledge, and its application to a wide range of tasks. The overall picture shows that not only are semi-structured resources enabling a renaissance of knowledge-rich AI techniques, but also that significant advances in high-end applications that require deep understanding capabilities can be achieved by synergistically exploiting large amounts of machine-readable structured knowledge in combination with sound statistical AI and NLP techniques.

AIJ Journal 2013 Journal Article

Editorial

• Eduard Hovy
• Roberto Navigli
• Simone Paolo Ponzetto

IJCAI Conference 2007 Conference Paper

• Rahul Bhagat
• Eduard Hovy

Proper names, whether English or non-English, have several different spellings when transliterated from a non-English source language into English. Knowing the different variations can significantly improve the results of name-searches on various source texts, especially when recall is important. In this paper we propose two novel phonetic models to generate numerous candidate variant spellings of a name. Our methods show threefold improvement over the baseline and generate four times as many good name variants compared to a human while maintaining a respectable precision of 0. 68.