Spandana Gella

Nicholas Meade, Spandana Gella, Devamanyu Hazarika et al. · cmu

While large neural-based conversational models have become increasingly proficient dialogue agents, recent work has highlighted safety issues with these systems. For example, these systems can be goaded into generating toxic content, which often perpetuates social biases or stereotypes. We investigate a retrieval-based method for reducing bias and toxicity in responses from chatbots. It uses in-context learning to steer a model towards safer generations. Concretely, to generate a response to an unsafe dialogue context, we retrieve demonstrations of safe responses to similar dialogue contexts. We find our method performs competitively with strong baselines without requiring training. For instance, using automatic evaluation, we find our best fine-tuned baseline only generates safe responses to unsafe dialogue contexts from DiaSafety 4.04% more than our approach. Finally, we also propose a re-ranking procedure which can further improve response safeness.

Alexander Gurung, Spandana Gella, Alexandre Drouin et al.

Deep research agents increasingly combine private local documents with external tools like web retrieval, creating a privacy risk: an agent's external queries may leak sensitive information from its local context. This risk is amplified by the mosaic effect, where individual queries may appear harmless but become revealing in aggregate. We introduce MosaicLeaks, a benchmark of 1,001 multi-hop deep research tasks that chain private enterprise documents and a public web corpus, forcing agents to make external queries that depend on local information. We evaluate leakage with an adversary LLM that observes only the agent's external queries and attempts to infer private information at three levels: the agent's research intent, answers to specific private questions and verifiable claims about the enterprise documents. We find that models across families and sizes frequently leak at all three levels, that zero-shot privacy prompting reduces but does not eliminate leakage and that reinforcement learning for task performance alone worsens leakage. To address this, we propose Privacy-Aware Deep Research (PA-DR), an RL framework that combines situational rewards for task success with a learned privacy classifier to provide dense credit assignment over both per-query and mosaic-level leakage. Training Qwen3-4B-Instruct with PA-DR improves accuracy from 48.7% to 58.7% and reduces answer and full-information leakage from 34.0% to 9.9%.

Prakhar Gupta, Yang Liu, Di Jin et al. · cmu

Dialogue models are able to generate coherent and fluent responses, but they can still be challenging to control and may produce non-engaging, unsafe results. This unpredictability diminishes user trust and can hinder the use of the models in the real world. To address this, we introduce DialGuide, a novel framework for controlling dialogue model behavior using natural language rules, or guidelines. These guidelines provide information about the context they are applicable to and what should be included in the response, allowing the models to generate responses that are more closely aligned with the developer's expectations and intent. We evaluate DialGuide on three tasks in open-domain dialogue response generation: guideline selection, response generation, and response entailment verification. Our dataset contains 10,737 positive and 15,467 negative dialogue context-response-guideline triplets across two domains - chit-chat and safety. We provide baseline models for the tasks and benchmark their performance. We also demonstrate that DialGuide is effective in the dialogue safety domain, producing safe and engaging responses that follow developer guidelines.

Spandana Gella, Aishwarya Padmakumar, Patrick Lange et al. · amazon-science

Embodied agents need to be able to interact in natural language understanding task descriptions and asking appropriate follow up questions to obtain necessary information to be effective at successfully accomplishing tasks for a wide range of users. In this work, we propose a set of dialog acts for modelling such dialogs and annotate the TEACh dataset that includes over 3,000 situated, task oriented conversations (consisting of 39.5k utterances in total) with dialog acts. TEACh-DA is one of the first large scale dataset of dialog act annotations for embodied task completion. Furthermore, we demonstrate the use of this annotated dataset in training models for tagging the dialog acts of a given utterance, predicting the dialog act of the next response given a dialog history, and use the dialog acts to guide agent's non-dialog behaviour. In particular, our experiments on the TEACh Execution from Dialog History task where the model predicts the sequence of low level actions to be executed in the environment for embodied task completion, demonstrate that dialog acts can improve end task success rate by up to 2 points compared to the system without dialog acts.

Xiangru Jian, Shravan Nayak, Kevin Qinghong Lin et al.

Computer-use agents (CUAs) hold great promise for automating complex desktop workflows, yet progress toward general-purpose agents is bottlenecked by the scarcity of continuous, high-quality human demonstration videos. Recent work emphasizes that continuous video, not sparse screenshots, is the critical missing ingredient for scaling these agents. However, the largest existing open dataset, ScaleCUA, contains only 2 million screenshots, equating to less than 20 hours of video. To address this bottleneck, we introduce CUA-Suite, a large-scale ecosystem of expert video demonstrations and dense annotations for professional desktop computer-use agents. At its core is VideoCUA, which provides approximately 10,000 human-demonstrated tasks across 87 diverse applications with continuous 30 fps screen recordings, kinematic cursor traces, and multi-layerfed reasoning annotations, totaling approximately 55 hours and 6 million frames of expert video. Unlike sparse datasets that capture only final click coordinates, these continuous video streams preserve the full temporal dynamics of human interaction, forming a superset of information that can be losslessly transformed into the formats required by existing agent frameworks. CUA-Suite further provides two complementary resources: UI-Vision, a rigorous benchmark for evaluating grounding and planning capabilities in CUAs, and GroundCUA, a large-scale grounding dataset with 56K annotated screenshots and over 3.6 million UI element annotations. Preliminary evaluation reveals that current foundation action models struggle substantially with professional desktop applications (~60% task failure rate). Beyond evaluation, CUA-Suite's rich multimodal corpus supports emerging research directions including generalist screen parsing, continuous spatial control, video-based reward modeling, and visual world models. All data and models are publicly released.

Shravan Nayak, Xiangru Jian, Kevin Qinghong Lin et al. · mila

Autonomous agents that navigate Graphical User Interfaces (GUIs) to automate tasks like document editing and file management can greatly enhance computer workflows. While existing research focuses on online settings, desktop environments, critical for many professional and everyday tasks, remain underexplored due to data collection challenges and licensing issues. We introduce UI-Vision, the first comprehensive, license-permissive benchmark for offline, fine-grained evaluation of computer use agents in real-world desktop environments. Unlike online benchmarks, UI-Vision provides: (i) dense, high-quality annotations of human demonstrations, including bounding boxes, UI labels, and action trajectories (clicks, drags, and keyboard inputs) across 83 software applications, and (ii) three fine-to-coarse grained tasks-Element Grounding, Layout Grounding, and Action Prediction-with well-defined metrics to rigorously evaluate agents' performance in desktop environments. Our evaluation reveals critical limitations in state-of-the-art models like UI-TARS-72B, including issues with understanding professional software, spatial reasoning, and complex actions like drag-and-drop. These findings highlight the challenges in developing fully autonomous computer use agents. By releasing UI-Vision as open-source, we aim to advance the development of more capable agents for real-world desktop tasks.

Xiaoqiang Wang, Chao Wang, Hadi Nekoei et al.

We present Mem-$π$, a framework for adaptive memory in large language model (LLM) agents, where useful guidance is generated on demand rather than retrieved from external memory stores. Existing memory-augmented agents typically rely on similarity-based retrieval from episodic memory banks or skill libraries, returning static entries that often misalign with the current context. In contrast, Mem-$π$ uses a dedicated language or vision-language model with its own parameters, separate from the downstream agent, to generate context-specific guidance for complex tasks. Conditioned on the current agent context, the model jointly decides when to produce guidance and what guidance to produce. We train it with a decision-content decoupled reinforcement learning (RL) objective, enabling it to abstain when generation would not help and otherwise produce concise, useful guidance. Across diverse agentic benchmarks spanning web navigation, terminal-based tool use, and text-based embodied interaction, Mem-$π$ consistently outperforms retrieval-based and prior RL-optimized memory baselines, achieving over 30% relative improvement on web navigation tasks.

Aarash Feizi, Shravan Nayak, Xiangru Jian et al.

Building reliable computer-use agents requires grounding: accurately connecting natural language instructions to the correct on-screen elements. While large datasets exist for web and mobile interactions, high-quality resources for desktop environments are limited. To address this gap, we introduce GroundCUA, a large-scale desktop grounding dataset built from expert human demonstrations. It covers 87 applications across 12 categories and includes 56K screenshots, with every on-screen element carefully annotated for a total of over 3.56M human-verified annotations. From these demonstrations, we generate diverse instructions that capture a wide range of real-world tasks, providing high-quality data for model training. Using GroundCUA, we develop the GroundNext family of models that map instructions to their target UI elements. At both 3B and 7B scales, GroundNext achieves state-of-the-art results across five benchmarks using supervised fine-tuning, while requiring less than one-tenth the training data of prior work. Reinforcement learning post-training further improves performance, and when evaluated in an agentic setting on the OSWorld benchmark using o3 as planner, GroundNext attains comparable or superior results to models trained with substantially more data,. These results demonstrate the critical role of high-quality, expert-driven datasets in advancing general-purpose computer-use agents.

Ahmed Masry, Megh Thakkar, Patrice Bechard et al.

Retrieval-augmented generation has proven practical when models require specialized knowledge or access to the latest data. However, existing methods for multimodal document retrieval often replicate techniques developed for text-only retrieval, whether in how they encode documents, define training objectives, or compute similarity scores. To address these limitations, we present ColMate, a document retrieval model that bridges the gap between multimodal representation learning and document retrieval. ColMate utilizes a novel OCR-based pretraining objective, a self-supervised masked contrastive learning objective, and a late interaction scoring mechanism more relevant to multimodal document structures and visual characteristics. ColMate obtains 3.61% improvements over existing retrieval models on the ViDoRe V2 benchmark, demonstrating stronger generalization to out-of-domain benchmarks.

Juan Rodriguez, Xiangru Jian, Siba Smarak Panigrahi et al. · mila

Multimodal AI has the potential to significantly enhance document-understanding tasks, such as processing receipts, understanding workflows, extracting data from documents, and summarizing reports. Code generation tasks that require long-structured outputs can also be enhanced by multimodality. Despite this, their use in commercial applications is often limited due to limited access to training data and restrictive licensing, which hinders open access. To address these limitations, we introduce BigDocs-7.5M, a high-quality, open-access dataset comprising 7.5 million multimodal documents across 30 tasks. We use an efficient data curation process to ensure our data is high-quality and license-permissive. Our process emphasizes accountability, responsibility, and transparency through filtering rules, traceable metadata, and careful content analysis. Additionally, we introduce BigDocs-Bench, a benchmark suite with 10 novel tasks where we create datasets that reflect real-world use cases involving reasoning over Graphical User Interfaces (GUI) and code generation from images. Our experiments show that training with BigDocs-Bench improves average performance up to 25.8% over closed-source GPT-4o in document reasoning and structured output tasks such as Screenshot2HTML or Image2Latex generation. Finally, human evaluations showed a preference for outputs from models trained on BigDocs over GPT-4o. This suggests that BigDocs can help both academics and the open-source community utilize and improve AI tools to enhance multimodal capabilities and document reasoning. The project is hosted at https://bigdocs.github.io .

Amirhossein Abaskohi, Tianyi Chen, Miguel Muñoz-Mármol et al. · mila

We introduce DRBench, a benchmark for evaluating AI agents on complex, open-ended deep research tasks in enterprise settings. Unlike prior benchmarks that focus on simple questions or web-only queries, DRBench evaluates agents on multi-step queries (for example, ``What changes should we make to our product roadmap to ensure compliance with this standard?") that require identifying supporting facts from both the public web and private company knowledge base. Each task is grounded in realistic user personas and enterprise context, spanning a heterogeneous search space that includes productivity software, cloud file systems, emails, chat conversations, and the open web. Tasks are generated through a carefully designed synthesis pipeline with human-in-the-loop verification, and agents are evaluated on their ability to recall relevant insights, maintain factual accuracy, and produce coherent, well-structured reports. We release 15 deep research tasks across 10 domains, such as Sales, Cybersecurity, and Compliance. We demonstrate the effectiveness of DRBench by evaluating diverse DR agents across open- and closed-source models (such as GPT, Llama, and Qwen) and DR strategies, highlighting their strengths, weaknesses, and the critical path for advancing enterprise deep research. Code is available at https://github.com/ServiceNow/drbench.

Ahmed Masry, Abhay Puri, Masoud Hashemi et al. · mila

Charts are essential to data analysis, transforming raw data into clear visual representations that support human decision-making. Although current vision-language models (VLMs) have made significant progress, they continue to struggle with chart comprehension due to training on datasets that lack diversity and real-world authenticity, or on automatically extracted underlying data tables of charts, which can contain numerous estimation errors. Furthermore, existing models only rely on supervised fine-tuning using these low-quality datasets, severely limiting their effectiveness. To address these issues, we first propose BigCharts, a dataset creation pipeline that generates visually diverse chart images by conditioning the rendering process on real-world charts sourced from multiple online platforms. Unlike purely synthetic datasets, BigCharts incorporates real-world data, ensuring authenticity and visual diversity, while still retaining accurate underlying data due to our proposed replotting process. Additionally, we introduce a comprehensive training framework that integrates supervised fine-tuning with Group Relative Policy Optimization (GRPO)-based reinforcement learning. By introducing novel reward signals specifically designed for chart reasoning, our approach enhances model robustness and generalization across diverse chart styles and domains, resulting in a state-of-the-art chart reasoning model, BigCharts-R1. Extensive experiments demonstrate that our models surpass existing methods on multiple chart question-answering benchmarks compared to even larger open-source and closed-source models.

Juan Rodriguez, Haotian Zhang, Abhay Puri et al.

We introduce VectorGym, a comprehensive benchmark suite for Scalable Vector Graphics (SVG) that spans generation from text and sketches, complex editing, and visual understanding. VectorGym addresses the lack of realistic, challenging benchmarks aligned with professional design workflows. Our benchmark comprises four tasks with expert human-authored annotations: the novel Sketch2SVG task (VG-Sketch); a new SVG editing dataset (VG-Edit) featuring complex, multi-step edits with higher-order primitives; Text2SVG generation (VG-Text); and SVG captioning (VG-Cap). Unlike prior benchmarks that rely on synthetic edits, VectorGym provides gold-standard human annotations that require semantic understanding and design intent. We also propose a multi-task reinforcement learning approach that jointly optimizes across all four tasks using rendering-based rewards. Our method, built on GRPO with curriculum learning, trains a Qwen3-VL 8B model that achieves state-of-the-art performance among open-source models, surpassing much larger models including Qwen3-VL 235B and matching GPT-4o. We also introduce a VLM-as-a-Judge metric for SVG generation, validated through human correlation studies. Our evaluation of frontier VLMs reveals significant performance gaps, positioning VectorGym as a rigorous framework for advancing visual code generation. VectorGym is publicly available on huggingface.co/datasets/ServiceNow/VectorGym.

Chao Zhao, Spandana Gella, Seokhwan Kim et al.

Task-oriented Dialogue (TOD) Systems aim to build dialogue systems that assist users in accomplishing specific goals, such as booking a hotel or a restaurant. Traditional TODs rely on domain-specific APIs/DBs or external factual knowledge to generate responses, which cannot accommodate subjective user requests (e.g., "Is the WIFI reliable?" or "Does the restaurant have a good atmosphere?"). To address this issue, we propose a novel task of subjective-knowledge-based TOD (SK-TOD). We also propose the first corresponding dataset, which contains subjective knowledge-seeking dialogue contexts and manually annotated responses grounded in subjective knowledge sources. When evaluated with existing TOD approaches, we find that this task poses new challenges such as aggregating diverse opinions from multiple knowledge snippets. We hope this task and dataset can promote further research on TOD and subjective content understanding. The code and the dataset are available at https://github.com/alexa/dstc11-track5.

Arjun R Akula, Spandana Gella, Yaser Al-Onaizan et al.

Visual referring expression recognition is a challenging task that requires natural language understanding in the context of an image. We critically examine RefCOCOg, a standard benchmark for this task, using a human study and show that 83.7% of test instances do not require reasoning on linguistic structure, i.e., words are enough to identify the target object, the word order doesn't matter. To measure the true progress of existing models, we split the test set into two sets, one which requires reasoning on linguistic structure and the other which doesn't. Additionally, we create an out-of-distribution dataset Ref-Adv by asking crowdworkers to perturb in-domain examples such that the target object changes. Using these datasets, we empirically show that existing methods fail to exploit linguistic structure and are 12% to 23% lower in performance than the established progress for this task. We also propose two methods, one based on contrastive learning and the other based on multi-task learning, to increase the robustness of ViLBERT, the current state-of-the-art model for this task. Our datasets are publicly available at https://github.com/aws/aws-refcocog-adv

Spandana Gella, Mirella Lapata, Frank Keller

We introduce a new task, visual sense disambiguation for verbs: given an image and a verb, assign the correct sense of the verb, i.e., the one that describes the action depicted in the image. Just as textual word sense disambiguation is useful for a wide range of NLP tasks, visual sense disambiguation can be useful for multimodal tasks such as image retrieval, image description, and text illustration. We introduce VerSe, a new dataset that augments existing multimodal datasets (COCO and TUHOI) with sense labels. We propose an unsupervised algorithm based on Lesk which performs visual sense disambiguation using textual, visual, or multimodal embeddings. We find that textual embeddings perform well when gold-standard textual annotations (object labels and image descriptions) are available, while multimodal embeddings perform well on unannotated images. We also verify our findings by using the textual and multimodal embeddings as features in a supervised setting and analyse the performance of visual sense disambiguation task. VerSe is made publicly available and can be downloaded at: https://github.com/spandanagella/verse.

Ada Defne Tur, Nicholas Meade, Xing Han Lù et al. · eth-zurich, mila

LLM-based agents are becoming increasingly proficient at solving web-based tasks. With this capability comes a greater risk of misuse for malicious purposes, such as posting misinformation in an online forum or selling illicit substances on a website. To evaluate these risks, we propose SafeArena, the first benchmark to focus on the deliberate misuse of web agents. SafeArena comprises 250 safe and 250 harmful tasks across four websites. We classify the harmful tasks into five harm categories -- misinformation, illegal activity, harassment, cybercrime, and social bias, designed to assess realistic misuses of web agents. We evaluate leading LLM-based web agents, including GPT-4o, Claude-3.5 Sonnet, Qwen-2-VL 72B, and Llama-3.2 90B, on our benchmark. To systematically assess their susceptibility to harmful tasks, we introduce the Agent Risk Assessment framework that categorizes agent behavior across four risk levels. We find agents are surprisingly compliant with malicious requests, with GPT-4o and Qwen-2 completing 34.7% and 27.3% of harmful requests, respectively. Our findings highlight the urgent need for safety alignment procedures for web agents. Our benchmark is available here: https://safearena.github.io

Rabiul Awal, Mahsa Massoud, Aarash Feizi et al.

We present WebMMU, a multilingual benchmark that evaluates three core web tasks: (1) website visual question answering, (2) code editing involving HTML/CSS/JavaScript, and (3) mockup-to-code generation. Unlike prior benchmarks that treat these tasks separately, WebMMU unifies them using expert-annotated, real-world web data to assess models' abilities in complex multi-step reasoning, precise element grounding, and functional UI comprehension and coding. Our evaluation shows that while multimodal large language models (MLLMs) perform well on basic information extraction, they struggle with reasoning and grounding, editing code to preserve functionality, and generating design-to-code that maintains hierarchy and supports multilingual content. These findings reveal key limitations in current MLLMs and underscore the need for improved multimodal and cross-lingual reasoning to build future web agents capable of automating diverse web development tasks.

Amirhossein Abaskohi, Spandana Gella, Giuseppe Carenini et al.

Multimodal multihop question answering (MMQA) requires reasoning over images and text from multiple sources. Despite advances in visual question answering, this multihop setting remains underexplored due to a lack of quality datasets. Existing methods focus on single-hop, single-modality, or short texts, limiting real-world applications like interpreting educational documents with long, multimodal content. To fill this gap, we introduce FM2DS, the first framework for creating a high-quality dataset for MMQA. Our approach consists of a 5-stage pipeline that involves acquiring relevant multimodal documents from Wikipedia, synthetically generating high-level questions and answers, and validating them through rigorous criteria to ensure data quality. We evaluate our methodology by training models on our synthesized dataset and testing on two benchmarks: MultimodalQA and WebQA. Our results demonstrate that, with an equal sample size, models trained on our synthesized data outperform those trained on human-collected data by 1.9 in exact match (EM) score on average. Additionally, we introduce M2QA-Bench with 1k samples, the first benchmark for MMQA on long documents, generated using FM2DS and refined by human annotators. We believe our data synthesis method will serve as a strong foundation for training and evaluating MMQA models.

Aarash Feizi, Sai Rajeswar, Adriana Romero-Soriano et al.

Understanding how effectively large vision language models (VLMs) compare visual inputs is crucial across numerous applications, yet this fundamental capability remains insufficiently assessed. While VLMs are increasingly deployed for tasks requiring comparative judgment, including automated evaluation, re-ranking, and retrieval-augmented generation, no systematic framework exists to measure their performance in these scenarios. We present PairBench, a simple framework that evaluates VLMs as customizable similarity tools using widely available image datasets. Our approach introduces four key metrics for reliable comparison: alignment with human annotations, consistency across pair ordering, distribution smoothness, and controllability through prompting. Our analysis reveals that no model consistently excels across all metrics, with each demonstrating distinct strengths and weaknesses. Most concerning is the widespread inability of VLMs to maintain symmetric similarity scores. Interestingly, we demonstrate that performance on our benchmark strongly correlates with popular benchmarks used for more complex tasks, while providing additional metrics into controllability, smoothness and ordering. This makes PairBench a unique and comprehensive framework to evaluate the performance of VLMs for automatic evaluation depending on the task.

Juan A. Rodriguez, Haotian Zhang, Abhay Puri et al. · mila

Scalable Vector Graphics (SVG) offer a powerful format for representing visual designs as interpretable code. Recent advances in vision-language models (VLMs) have enabled high-quality SVG generation by framing the problem as a code generation task and leveraging large-scale pretraining. VLMs are particularly suitable for this task as they capture both global semantics and fine-grained visual patterns, while transferring knowledge across vision, natural language, and code domains. However, existing VLM approaches often struggle to produce faithful and efficient SVGs because they never observe the rendered images during training. Although differentiable rendering for autoregressive SVG code generation remains unavailable, rendered outputs can still be compared to original inputs, enabling evaluative feedback suitable for reinforcement learning (RL). We introduce RLRF(Reinforcement Learning from Rendering Feedback), an RL method that enhances SVG generation in autoregressive VLMs by leveraging feedback from rendered SVG outputs. Given an input image, the model generates SVG roll-outs that are rendered and compared to the original image to compute a reward. This visual fidelity feedback guides the model toward producing more accurate, efficient, and semantically coherent SVGs. RLRF significantly outperforms supervised fine-tuning, addressing common failure modes and enabling precise, high-quality SVG generation with strong structural understanding and generalization.

Suyuchen Wang, Tianyu Zhang, Ahmed Masry et al.

GUI grounding, the task of mapping natural-language instructions to pixel coordinates, is crucial for autonomous agents, yet remains difficult for current VLMs. The core bottleneck is reliable patch-to-pixel mapping, which breaks when extrapolating to high-resolution displays unseen during training. Current approaches generate coordinates as text tokens directly from visual features, forcing the model to infer complex position-to-pixel mappings implicitly; as a result, accuracy degrades and failures proliferate on new resolutions. We address this with two complementary innovations. First, RULER tokens serve as explicit coordinate markers, letting the model reference positions similar to gridlines on a map and adjust rather than generate coordinates from scratch. Second, Interleaved MRoPE (I-MRoPE) improves spatial encoding by ensuring that width and height dimensions are represented equally, addressing the asymmetry of standard positional schemes. Experiments on ScreenSpot, ScreenSpot-V2, and ScreenSpot-Pro show consistent gains in grounding accuracy, with the largest improvements on high-resolution interfaces. By providing explicit spatial guidance rather than relying on implicit learning, our approach enables more reliable GUI automation across diverse resolutions and platforms.

Rishabh Maheshwary, Masoud Hashemi, Khyati Mahajan et al.

Iterative RAG for multi-hop question answering faces challenges with lengthy contexts and the buildup of irrelevant information. This hinders a model's capacity to process and reason over retrieved content and limits performance. While recent methods focus on compressing retrieved information, they are either restricted to single-round RAG, require finetuning or lack scalability in iterative RAG. To address these challenges, we propose Notes Writing, a method that generates concise and relevant notes from retrieved documents at each step, thereby reducing noise and retaining only essential information. This indirectly increases the effective context length of Large Language Models (LLMs), enabling them to reason and plan more effectively while processing larger volumes of input text. Notes Writing is framework agnostic and can be integrated with different iterative RAG methods. We demonstrate its effectiveness with three iterative RAG methods, across two models and four evaluation datasets. Notes writing yields an average improvement of 15.6 percentage points overall, with minimal increase in output tokens.

Amirhossein Abaskohi, Amrutha Varshini Ramesh, Shailesh Nanisetty et al.

We introduce AgentAda, the first LLM-powered analytics agent that can learn and use new analytics skills to extract more specialized insights. Unlike existing methods that require users to manually decide which data analytics method to apply, AgentAda automatically identifies the skill needed from a library of analytical skills to perform the analysis. This also allows AgentAda to use skills that existing LLMs cannot perform out of the box. The library covers a range of methods, including clustering, predictive modeling, and NLP techniques like BERT, which allow AgentAda to handle complex analytics tasks based on what the user needs. AgentAda's dataset-to-insight extraction strategy consists of three key steps: (I) a question generator to generate queries relevant to the user's goal and persona, (II) a hybrid Retrieval-Augmented Generation (RAG)-based skill matcher to choose the best data analytics skill from the skill library, and (III) a code generator that produces executable code based on the retrieved skill's documentation to extract key patterns. We also introduce KaggleBench, a benchmark of curated notebooks across diverse domains, to evaluate AgentAda's performance. We conducted a human evaluation demonstrating that AgentAda provides more insightful analytics than existing tools, with 48.78% of evaluators preferring its analyses, compared to 27.67% for the unskilled agent. We also propose a novel LLM-as-a-judge approach that we show is aligned with human evaluation as a way to automate insight quality evaluation at larger scale.

Patrice Bechard, Chao Wang, Amirhossein Abaskohi et al.

Workflows are a fundamental component of automation in enterprise platforms, enabling the orchestration of tasks, data processing, and system integrations. Despite being widely used, building workflows can be complex, often requiring manual configuration through low-code platforms or visual programming tools. To simplify this process, we explore the use of generative foundation models, particularly vision-language models (VLMs), to automatically generate structured workflows from visual inputs. Translating hand-drawn sketches or computer-generated diagrams into executable workflows is challenging due to the ambiguity of free-form drawings, variations in diagram styles, and the difficulty of inferring execution logic from visual elements. To address this, we introduce StarFlow, a framework for generating structured workflow outputs from sketches using vision-language models. We curate a diverse dataset of workflow diagrams -- including synthetic, manually annotated, and real-world samples -- to enable robust training and evaluation. We finetune and benchmark multiple vision-language models, conducting a series of ablation studies to analyze the strengths and limitations of our approach. Our results show that finetuning significantly enhances structured workflow generation, outperforming large vision-language models on this task.

Ahmed Masry, Juan A. Rodriguez, Tianyu Zhang et al. · mila

Aligning visual features with language embeddings is a key challenge in vision-language models (VLMs). The performance of such models hinges on having a good connector that maps visual features generated by a vision encoder to a shared embedding space with the LLM while preserving semantic similarity. Existing connectors, such as multilayer perceptrons (MLPs), lack inductive bias to constrain visual features within the linguistic structure of the LLM's embedding space, making them data-hungry and prone to cross-modal misalignment. In this work, we propose a novel vision-text alignment method, AlignVLM, that maps visual features to a weighted average of LLM text embeddings. Our approach leverages the linguistic priors encoded by the LLM to ensure that visual features are mapped to regions of the space that the LLM can effectively interpret. AlignVLM is particularly effective for document understanding tasks, where visual and textual modalities are highly correlated. Our extensive experiments show that AlignVLM achieves state-of-the-art performance compared to prior alignment methods, with larger gains on document understanding tasks and under low-resource setups. We provide further analysis demonstrating its efficiency and robustness to noise.

Mert İnan, Aishwarya Padmakumar, Spandana Gella et al.

Task planning is an important component of traditional robotics systems enabling robots to compose fine grained skills to perform more complex tasks. Recent work building systems for translating natural language to executable actions for task completion in simulated embodied agents is focused on directly predicting low level action sequences that would be expected to be directly executable by a physical robot. In this work, we instead focus on predicting a higher level plan representation for one such embodied task completion dataset - TEACh, under the assumption that techniques for high-level plan prediction from natural language are expected to be more transferable to physical robot systems. We demonstrate that better plans can be predicted using multimodal context, and that plan prediction and plan execution modules are likely dependent on each other and hence it may not be ideal to fully decouple them. Further, we benchmark execution of oracle plans to quantify the scope for improvement in plan prediction models.

Lisa Bauer, Karthik Gopalakrishnan, Spandana Gella et al.

Prompting inputs with natural language task descriptions has emerged as a popular mechanism to elicit reasonably accurate outputs from large-scale generative language models with little to no in-context supervision. This also helps gain insight into how well language models capture the semantics of a wide range of downstream tasks purely from self-supervised pre-training on massive corpora of unlabeled text. Such models have naturally also been exposed to a lot of undesirable content like racist and sexist language and there is limited work on awareness of models along these dimensions. In this paper, we define and comprehensively evaluate how well such language models capture the semantics of four tasks for bias: diagnosis, identification, extraction and rephrasing. We define three broad classes of task descriptions for these tasks: statement, question, and completion, with numerous lexical variants within each class. We study the efficacy of prompting for each task using these classes and the null task description across several decoding methods and few-shot examples. Our analyses indicate that language models are capable of performing these tasks to widely varying degrees across different bias dimensions, such as gender and political affiliation. We believe our work is an important step towards unbiased language models by quantifying the limits of current self-supervision objectives at accomplishing such sociologically challenging tasks.

Sashank Santhanam, Behnam Hedayatnia, Spandana Gella et al.

Recently neural response generation models have leveraged large pre-trained transformer models and knowledge snippets to generate relevant and informative responses. However, this does not guarantee that generated responses are factually correct. In this paper, we examine factual correctness in knowledge-grounded neural response generation models. We present a human annotation setup to identify three different response types: responses that are factually consistent with respect to the input knowledge, responses that contain hallucinated knowledge, and non-verifiable chitchat style responses. We use this setup to annotate responses generated using different stateof-the-art models, knowledge snippets, and decoding strategies. In addition, to facilitate the development of a factual consistency detector, we automatically create a new corpus called Conv-FEVER that is adapted from the Wizard of Wikipedia dataset and includes factually consistent and inconsistent responses. We demonstrate the benefit of our Conv-FEVER dataset by showing that the models trained on this data perform reasonably well to detect factually inconsistent responses with respect to the provided knowledge through evaluation on our human annotated data. We will release the Conv-FEVER dataset and the human annotated responses.

Aishwarya Padmakumar, Jesse Thomason, Ayush Shrivastava et al.

Robots operating in human spaces must be able to engage in natural language interaction with people, both understanding and executing instructions, and using conversation to resolve ambiguity and recover from mistakes. To study this, we introduce TEACh, a dataset of over 3,000 human--human, interactive dialogues to complete household tasks in simulation. A Commander with access to oracle information about a task communicates in natural language with a Follower. The Follower navigates through and interacts with the environment to complete tasks varying in complexity from "Make Coffee" to "Prepare Breakfast", asking questions and getting additional information from the Commander. We propose three benchmarks using TEACh to study embodied intelligence challenges, and we evaluate initial models' abilities in dialogue understanding, language grounding, and task execution.

Lifu Tu, Garima Lalwani, Spandana Gella et al.

Recent work has shown that pre-trained language models such as BERT improve robustness to spurious correlations in the dataset. Intrigued by these results, we find that the key to their success is generalization from a small amount of counterexamples where the spurious correlations do not hold. When such minority examples are scarce, pre-trained models perform as poorly as models trained from scratch. In the case of extreme minority, we propose to use multi-task learning (MTL) to improve generalization. Our experiments on natural language inference and paraphrase identification show that MTL with the right auxiliary tasks significantly improves performance on challenging examples without hurting the in-distribution performance. Further, we show that the gain from MTL mainly comes from improved generalization from the minority examples. Our results highlight the importance of data diversity for overcoming spurious correlations.

Sravan Babu Bodapati, Spandana Gella, Kasturi Bhattacharjee et al.

User generated text on social media often suffers from a lot of undesired characteristics including hatespeech, abusive language, insults etc. that are targeted to attack or abuse a specific group of people. Often such text is written differently compared to traditional text such as news involving either explicit mention of abusive words, obfuscated words and typological errors or implicit abuse i.e., indicating or targeting negative stereotypes. Thus, processing this text poses several robustness challenges when we apply natural language processing techniques developed for traditional text. For example, using word or token based models to process such text can treat two spelling variants of a word as two different words. Following recent work, we analyze how character, subword and byte pair encoding (BPE) models can be aid some of the challenges posed by user generated text. In our work, we analyze the effectiveness of each of the above techniques, compare and contrast various word decomposition techniques when used in combination with others. We experiment with finetuning large pretrained language models, and demonstrate their robustness to domain shift by studying Wikipedia attack, toxicity and Twitter hatespeech datasets

Shruti Palaskar, Jindrich Libovický, Spandana Gella et al.

In this paper, we study abstractive summarization for open-domain videos. Unlike the traditional text news summarization, the goal is less to "compress" text information but rather to provide a fluent textual summary of information that has been collected and fused from different source modalities, in our case video and audio transcripts (or text). We show how a multi-source sequence-to-sequence model with hierarchical attention can integrate information from different modalities into a coherent output, compare various models trained with different modalities and present pilot experiments on the How2 corpus of instructional videos. We also propose a new evaluation metric (Content F1) for abstractive summarization task that measures semantic adequacy rather than fluency of the summaries, which is covered by metrics like ROUGE and BLEU.

Spandana Gella, Desmond Elliott, Frank Keller

Recent work has shown that visual context improves cross-lingual sense disambiguation for nouns. We extend this line of work to the more challenging task of cross-lingual verb sense disambiguation, introducing the MultiSense dataset of 9,504 images annotated with English, German, and Spanish verbs. Each image in MultiSense is annotated with an English verb and its translation in German or Spanish. We show that cross-lingual verb sense disambiguation models benefit from visual context, compared to unimodal baselines. We also show that the verb sense predicted by our best disambiguation model can improve the results of a text-only machine translation system when used for a multimodal translation task.

Spandana Gella, Rico Sennrich, Frank Keller et al.

In this paper we propose a model to learn multimodal multilingual representations for matching images and sentences in different languages, with the aim of advancing multilingual versions of image search and image understanding. Our model learns a common representation for images and their descriptions in two different languages (which need not be parallel) by considering the image as a pivot between two languages. We introduce a new pairwise ranking loss function which can handle both symmetric and asymmetric similarity between the two modalities. We evaluate our models on image-description ranking for German and English, and on semantic textual similarity of image descriptions in English. In both cases we achieve state-of-the-art performance.

Spandana Gella, Frank Keller

A large amount of recent research has focused on tasks that combine language and vision, resulting in a proliferation of datasets and methods. One such task is action recognition, whose applications include image annotation, scene under- standing and image retrieval. In this survey, we categorize the existing ap- proaches based on how they conceptualize this problem and provide a detailed review of existing datasets, highlighting their di- versity as well as advantages and disad- vantages. We focus on recently devel- oped datasets which link visual informa- tion with linguistic resources and provide a fine-grained syntactic and semantic anal- ysis of actions in images.

Spandana Gella, Marc Dymetman, Jean Michel Renders et al.

We propose an approach for helping agents compose email replies to customer requests. To enable that, we use LDA to extract latent topics from a collection of email exchanges. We then use these latent topics to label our data, obtaining a so-called "silver standard" topic labelling. We exploit this labelled set to train a classifier to: (i) predict the topic distribution of the entire agent's email response, based on features of the customer's email; and (ii) predict the topic distribution of the next sentence in the agent's reply, based on the customer's email features and on features of the agent's current sentence. The experimental results on a large email collection from a contact center in the tele- com domain show that the proposed ap- proach is effective in predicting the best topic of the agent's next sentence. In 80% of the cases, the correct topic is present among the top five recommended topics (out of fifty possible ones). This shows the potential of this method to be applied in an interactive setting, where the agent is presented a small list of likely topics to choose from for the next sentence.