Christopher Pal

Gaurav Sahu, Abhay Puri, Juan Rodriguez et al. · mila

Data analytics is essential for extracting valuable insights from data that can assist organizations in making effective decisions. We introduce InsightBench, a benchmark dataset with three key features. First, it consists of 100 datasets representing diverse business use cases such as finance and incident management, each accompanied by a carefully curated set of insights planted in the datasets. Second, unlike existing benchmarks focusing on answering single queries, InsightBench evaluates agents based on their ability to perform end-to-end data analytics, including formulating questions, interpreting answers, and generating a summary of insights and actionable steps. Third, we conducted comprehensive quality assurance to ensure that each dataset in the benchmark had clear goals and included relevant and meaningful questions and analysis. Furthermore, we implement a two-way evaluation mechanism using LLaMA-3 as an effective, open-source evaluator to assess agents' ability to extract insights. We also propose AgentPoirot, our baseline data analysis agent capable of performing end-to-end data analytics. Our evaluation on InsightBench shows that AgentPoirot outperforms existing approaches (such as Pandas Agent) that focus on resolving single queries. We also compare the performance of open- and closed-source LLMs and various evaluation strategies. Overall, this benchmark serves as a testbed to motivate further development in comprehensive automated data analytics and can be accessed here: https://github.com/ServiceNow/insight-bench.

Vikram Voleti, Alexia Jolicoeur-Martineau, Christopher Pal

Video prediction is a challenging task. The quality of video frames from current state-of-the-art (SOTA) generative models tends to be poor and generalization beyond the training data is difficult. Furthermore, existing prediction frameworks are typically not capable of simultaneously handling other video-related tasks such as unconditional generation or interpolation. In this work, we devise a general-purpose framework called Masked Conditional Video Diffusion (MCVD) for all of these video synthesis tasks using a probabilistic conditional score-based denoising diffusion model, conditioned on past and/or future frames. We train the model in a manner where we randomly and independently mask all the past frames or all the future frames. This novel but straightforward setup allows us to train a single model that is capable of executing a broad range of video tasks, specifically: future/past prediction -- when only future/past frames are masked; unconditional generation -- when both past and future frames are masked; and interpolation -- when neither past nor future frames are masked. Our experiments show that this approach can generate high-quality frames for diverse types of videos. Our MCVD models are built from simple non-recurrent 2D-convolutional architectures, conditioning on blocks of frames and generating blocks of frames. We generate videos of arbitrary lengths autoregressively in a block-wise manner. Our approach yields SOTA results across standard video prediction and interpolation benchmarks, with computation times for training models measured in 1-12 days using $\le$ 4 GPUs. Project page: https://mask-cond-video-diffusion.github.io ; Code : https://github.com/voletiv/mcvd-pytorch

Mahan Fathi, Jonathan Pilault, Orhan Firat et al. · mila

State space models (SSMs) have shown impressive results on tasks that require modeling long-range dependencies and efficiently scale to long sequences owing to their subquadratic runtime complexity. Originally designed for continuous signals, SSMs have shown superior performance on a plethora of tasks, in vision and audio; however, SSMs still lag Transformer performance in Language Modeling tasks. In this work, we propose a hybrid layer named Block-State Transformer (BST), that internally combines an SSM sublayer for long-range contextualization, and a Block Transformer sublayer for short-term representation of sequences. We study three different, and completely parallelizable, variants that integrate SSMs and block-wise attention. We show that our model outperforms similar Transformer-based architectures on language modeling perplexity and generalizes to longer sequences. In addition, the Block-State Transformer demonstrates more than tenfold increase in speed at the layer level compared to the Block-Recurrent Transformer when model parallelization is employed.

Benno Krojer, Dheeraj Vattikonda, Luis Lara et al. · mila

An image editing model should be able to perform diverse edits, ranging from object replacement, changing attributes or style, to performing actions or movement, which require many forms of reasoning. Current general instruction-guided editing models have significant shortcomings with action and reasoning-centric edits. Object, attribute or stylistic changes can be learned from visually static datasets. On the other hand, high-quality data for action and reasoning-centric edits is scarce and has to come from entirely different sources that cover e.g. physical dynamics, temporality and spatial reasoning. To this end, we meticulously curate the AURORA Dataset (Action-Reasoning-Object-Attribute), a collection of high-quality training data, human-annotated and curated from videos and simulation engines. We focus on a key aspect of quality training data: triplets (source image, prompt, target image) contain a single meaningful visual change described by the prompt, i.e., truly minimal changes between source and target images. To demonstrate the value of our dataset, we evaluate an AURORA-finetuned model on a new expert-curated benchmark (AURORA-Bench) covering 8 diverse editing tasks. Our model significantly outperforms previous editing models as judged by human raters. For automatic evaluations, we find important flaws in previous metrics and caution their use for semantically hard editing tasks. Instead, we propose a new automatic metric that focuses on discriminative understanding. We hope that our efforts : (1) curating a quality training dataset and an evaluation benchmark, (2) developing critical evaluations, and (3) releasing a state-of-the-art model, will fuel further progress on general image editing.

Jae Hyun Lim, Nikola B. Kovachki, Ricardo Baptista et al.

Diffusion models have recently emerged as a powerful framework for generative modeling. They consist of a forward process that perturbs input data with Gaussian white noise and a reverse process that learns a score function to generate samples by denoising. Despite their tremendous success, they are mostly formulated on finite-dimensional spaces, e.g., Euclidean, limiting their applications to many domains where the data has a functional form, such as in scientific computing and 3D geometric data analysis. This work introduces a mathematically rigorous framework called Denoising Diffusion Operators (DDOs) for training diffusion models in function space. In DDOs, the forward process perturbs input functions gradually using a Gaussian process. The generative process is formulated by a function-valued annealed Langevin dynamic. Our approach requires an appropriate notion of the score for the perturbed data distribution, which we obtain by generalizing denoising score matching to function spaces that can be infinite-dimensional. We show that the corresponding discretized algorithm generates accurate samples at a fixed cost independent of the data resolution. We theoretically and numerically verify the applicability of our approach on a set of function-valued problems, including generating solutions to the Navier-Stokes equation viewed as the push-forward distribution of forcings from a Gaussian Random Field (GRF), as well as volcano InSAR and MNIST-SDF.

Alexandre Piche, Valentin Thomas, Joseph Marino et al. · mila

Bootstrapping is behind much of the successes of Deep Reinforcement Learning. However, learning the value function via bootstrapping often leads to unstable training due to fast-changing target values. Target Networks are employed to stabilize training by using an additional set of lagging parameters to estimate the target values. Despite the popularity of Target Networks, their effect on the optimization is still misunderstood. In this work, we show that they act as an implicit regularizer. This regularizer has disadvantages such as being inflexible and non convex. To overcome these issues, we propose an explicit Functional Regularization that is a convex regularizer in function space and can easily be tuned. We analyze the convergence of our method theoretically and empirically demonstrate that replacing Target Networks with the more theoretically grounded Functional Regularization approach leads to better sample efficiency and performance improvements.

Jonathan Pilault, Michael Galkin, Bahare Fatemi et al. · deepmind, mila

Recent work on Graph Neural Networks has demonstrated that self-supervised pretraining can further enhance performance on downstream graph, link, and node classification tasks. However, the efficacy of pretraining tasks has not been fully investigated for downstream large knowledge graph completion tasks. Using a contextualized knowledge graph embedding approach, we investigate five different pretraining signals, constructed using several graph algorithms and no external data, as well as their combination. We leverage the versatility of our Transformer-based model to explore graph structure generation pretraining tasks (i.e. path and k-hop neighborhood generation), typically inapplicable to most graph embedding methods. We further propose a new path-finding algorithm guided by information gain and find that it is the best-performing pretraining task across three downstream knowledge graph completion datasets. While using our new path-finding algorithm as a pretraining signal provides 2-3% MRR improvements, we show that pretraining on all signals together gives the best knowledge graph completion results. In a multitask setting that combines all pretraining tasks, our method surpasses the latest and strong performing knowledge graph embedding methods on all metrics for FB15K-237, on MRR and Hit@1 for WN18RRand on MRR and hit@10 for JF17K (a knowledge hypergraph dataset).

Mattie Tesfaldet, Derek Nowrouzezahrai, Christopher Pal

Recent extensions of Cellular Automata (CA) have incorporated key ideas from modern deep learning, dramatically extending their capabilities and catalyzing a new family of Neural Cellular Automata (NCA) techniques. Inspired by Transformer-based architectures, our work presents a new class of $\textit{attention-based}$ NCAs formed using a spatially localized$\unicode{x2014}$yet globally organized$\unicode{x2014}$self-attention scheme. We introduce an instance of this class named $\textit{Vision Transformer Cellular Automata}$ (ViTCA). We present quantitative and qualitative results on denoising autoencoding across six benchmark datasets, comparing ViTCA to a U-Net, a U-Net-based CA baseline (UNetCA), and a Vision Transformer (ViT). When comparing across architectures configured to similar parameter complexity, ViTCA architectures yield superior performance across all benchmarks and for nearly every evaluation metric. We present an ablation study on various architectural configurations of ViTCA, an analysis of its effect on cell states, and an investigation on its inductive biases. Finally, we examine its learned representations via linear probes on its converged cell state hidden representations, yielding, on average, superior results when compared to our U-Net, ViT, and UNetCA baselines.

Julien Roy, Pierre-Luc Bacon, Christopher Pal et al.

In recent years, in-silico molecular design has received much attention from the machine learning community. When designing a new compound for pharmaceutical applications, there are usually multiple properties of such molecules that need to be optimised: binding energy to the target, synthesizability, toxicity, EC50, and so on. While previous approaches have employed a scalarization scheme to turn the multi-objective problem into a preference-conditioned single objective, it has been established that this kind of reduction may produce solutions that tend to slide towards the extreme points of the objective space when presented with a problem that exhibits a concave Pareto front. In this work we experiment with an alternative formulation of goal-conditioned molecular generation to obtain a more controllable conditional model that can uniformly explore solutions along the entire Pareto front.

Christopher Beckham, Martin Weiss, Florian Golemo et al.

Different types of mental rotation tests have been used extensively in psychology to understand human visual reasoning and perception. Understanding what an object or visual scene would look like from another viewpoint is a challenging problem that is made even harder if it must be performed from a single image. We explore a controlled setting whereby questions are posed about the properties of a scene if that scene was observed from another viewpoint. To do this we have created a new version of the CLEVR dataset that we call CLEVR Mental Rotation Tests (CLEVR-MRT). Using CLEVR-MRT we examine standard methods, show how they fall short, then explore novel neural architectures that involve inferring volumetric representations of a scene. These volumes can be manipulated via camera-conditioned transformations to answer the question. We examine the efficacy of different model variants through rigorous ablations and demonstrate the efficacy of volumetric representations.

Vikram Voleti, Christopher Pal, Adam Oberman

Generative models based on denoising diffusion techniques have led to an unprecedented increase in the quality and diversity of imagery that is now possible to create with neural generative models. However, most contemporary state-of-the-art methods are derived from a standard isotropic Gaussian formulation. In this work we examine the situation where non-isotropic Gaussian distributions are used. We present the key mathematical derivations for creating denoising diffusion models using an underlying non-isotropic Gaussian noise model. We also provide initial experiments with the CIFAR-10 dataset to help verify empirically that this more general modeling approach can also yield high-quality samples.

Mats L. Richter, Christopher Pal

Minimal changes to neural architectures (e.g. changing a single hyperparameter in a key layer), can lead to significant gains in predictive performance in Convolutional Neural Networks (CNNs). In this work, we present a new approach to receptive field analysis that can yield these types of theoretical and empirical performance gains across twenty well-known CNN architectures examined in our experiments. By further developing and formalizing the analysis of receptive field expansion in convolutional neural networks, we can predict unproductive layers in an automated manner before ever training a model. This allows us to optimize the parameter-efficiency of a given architecture at low cost. Our method is computationally simple and can be done in an automated manner or even manually with minimal effort for most common architectures. We demonstrate the effectiveness of this approach by increasing parameter efficiency across past and current top-performing CNN-architectures. Specifically, our approach is able to improve ImageNet1K performance across a wide range of well-known, state-of-the-art (SOTA) model classes, including: VGG Nets, MobileNetV1, MobileNetV3, NASNet A (mobile), MnasNet, EfficientNet, and ConvNeXt - leading to a new SOTA result for each model class.

Stefania Raimondo, Christopher Pal, Xiaotian Liu et al.

Task-oriented dialogue is difficult in part because it involves understanding user intent, collecting information from the user, executing API calls, and generating helpful and fluent responses. However, for complex tasks one must also correctly do all of these things over multiple steps, and in a specific order. While large pre-trained language models can be fine-tuned end-to-end to create multi-step task-oriented dialogue agents that generate fluent text, our experiments confirm that this approach alone cannot reliably perform new multi-step tasks that are unseen during training. To address these limitations, we augment the dialogue contexts given to \textmd{text2text} transformers with known \textit{valid workflow names} and \textit{action plans}. Action plans consist of sequences of actions required to accomplish a task, and are encoded as simple sequences of keywords (e.g. verify-identity, pull-up-account, reset-password, etc.). We perform extensive experiments on the Action-Based Conversations Dataset (ABCD) with T5-small, base and large models, and show that such models: a) are able to more readily generalize to unseen workflows by following the provided plan, and b) are able to generalize to executing unseen actions if they are provided in the plan. In contrast, models are unable to fully accomplish new multi-step tasks when they are not provided action plan information, even when given new valid workflow names.

Christopher Beckham, Christopher Pal

In this work we theoretically show that conservative objective models (COMs) for offline model-based optimisation (MBO) are a special kind of contrastive divergence-based energy model, one where the energy function represents both the unconditional probability of the input and the conditional probability of the reward variable. While the initial formulation only samples modes from its learned distribution, we propose a simple fix that replaces its gradient ascent sampler with a Langevin MCMC sampler. This gives rise to a special probabilistic model where the probability of sampling an input is proportional to its predicted reward. Lastly, we show that better samples can be obtained if the model is decoupled so that the unconditional and conditional probabilities are modelled separately.

Simon Guiroy, Christopher Pal, Gonçalo Mordido et al.

Meta-Learning algorithms for few-shot learning aim to train neural networks capable of generalizing to novel tasks using only a few examples. Early-stopping is critical for performance, halting model training when it reaches optimal generalization to the new task distribution. Early-stopping mechanisms in Meta-Learning typically rely on measuring the model performance on labeled examples from a meta-validation set drawn from the training (source) dataset. This is problematic in few-shot transfer learning settings, where the meta-test set comes from a different target dataset (OOD) and can potentially have a large distributional shift with the meta-validation set. In this work, we propose Activation Based Early-stopping (ABE), an alternative to using validation-based early-stopping for meta-learning. Specifically, we analyze the evolution, during meta-training, of the neural activations at each hidden layer, on a small set of unlabelled support examples from a single task of the target tasks distribution, as this constitutes a minimal and justifiably accessible information from the target problem. Our experiments show that simple, label agnostic statistics on the activations offer an effective way to estimate how the target generalization evolves over time. At each hidden layer, we characterize the activation distributions, from their first and second order moments, then further summarized along the feature dimensions, resulting in a compact yet intuitive characterization in a four-dimensional space. Detecting when, throughout training time, and at which layer, the target activation trajectory diverges from the activation trajectory of the source data, allows us to perform early-stopping and improve generalization in a large array of few-shot transfer learning settings, across different algorithms, source and target datasets.

Nicolas Gontier, Pau Rodriguez, Issam Laradji et al.

Text-based game environments are challenging because agents must deal with long sequences of text, execute compositional actions using text and learn from sparse rewards. We address these challenges by proposing Language Decision Transformers (LDTs), a framework that is based on transformer language models and decision transformers (DTs). Our LDTs extend DTs with 3 components: (1) exponential tilt to guide the agent towards high obtainable goals, (2) novel goal conditioning methods yielding better results than the traditional return-to-go (sum of all future rewards), and (3) a model of future observations that improves agent performance. LDTs are the first to address offline RL with DTs on these challenging games. Our experiments show that LDTs achieve the highest scores among many different types of agents on some of the most challenging Jericho games, such as Enchanter.

Vikram Voleti, Boris N. Oreshkin, Florent Bocquelet et al.

Inverse Kinematics (IK) systems are often rigid with respect to their input character, thus requiring user intervention to be adapted to new skeletons. In this paper we aim at creating a flexible, learned IK solver applicable to a wide variety of human morphologies. We extend a state-of-the-art machine learning IK solver to operate on the well known Skinned Multi-Person Linear model (SMPL). We call our model SMPL-IK, and show that when integrated into real-time 3D software, this extended system opens up opportunities for defining novel AI-assisted animation workflows. For example, pose authoring can be made more flexible with SMPL-IK by allowing users to modify gender and body shape while posing a character. Additionally, when chained with existing pose estimation algorithms, SMPL-IK accelerates posing by allowing users to bootstrap 3D scenes from 2D images while allowing for further editing. Finally, we propose a novel SMPL Shape Inversion mechanism (SMPL-SI) to map arbitrary humanoid characters to the SMPL space, allowing artists to leverage SMPL-IK on custom characters. In addition to qualitative demos showing proposed tools, we present quantitative SMPL-IK baselines on the H36M and AMASS datasets.

Arantxa Casanova, Marlène Careil, Adriana Romero-Soriano et al.

Recent advances in conditional generative image models have enabled impressive results. On the one hand, text-based conditional models have achieved remarkable generation quality, by leveraging large-scale datasets of image-text pairs. To enable fine-grained controllability, however, text-based models require long prompts, whose details may be ignored by the model. On the other hand, layout-based conditional models have also witnessed significant advances. These models rely on bounding boxes or segmentation maps for precise spatial conditioning in combination with coarse semantic labels. The semantic labels, however, cannot be used to express detailed appearance characteristics. In this paper, we approach fine-grained scene controllability through image collages which allow a rich visual description of the desired scene as well as the appearance and location of the objects therein, without the need of class nor attribute labels. We introduce "mixing and matching scenes" (M&Ms), an approach that consists of an adversarially trained generative image model which is conditioned on appearance features and spatial positions of the different elements in a collage, and integrates these into a coherent image. We train our model on the OpenImages (OI) dataset and evaluate it on collages derived from OI and MS-COCO datasets. Our experiments on the OI dataset show that M&Ms outperforms baselines in terms of fine-grained scene controllability while being very competitive in terms of image quality and sample diversity. On the MS-COCO dataset, we highlight the generalization ability of our model by outperforming DALL-E in terms of the zero-shot FID metric, despite using two magnitudes fewer parameters and data. Collage based generative models have the potential to advance content creation in an efficient and effective way as they are intuitive to use and yield high quality generations.

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.

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.

Zhi Hao Luo, Luis Lara, Ge Ya Luo et al.

Text conditioned generative models for images have yielded impressive results. Text conditioned floorplan generation as a special type of raster image generation task also received particular attention. However there are many use cases in floorpla generation where numerical properties of the generated result are more important than the aesthetics. For instance, one might want to specify sizes for certain rooms in a floorplan and compare the generated floorplan with given specifications Current approaches, datasets and commonly used evaluations do not support these kinds of constraints. As such, an attractive strategy is to generate an intermediate data structure that contains numerical properties of a floorplan which can be used to generate the final floorplan image. To explore this setting we (1) construct a new dataset for this data-structure to data-structure formulation of floorplan generation using two popular image based floorplan datasets RPLAN and ProcTHOR-10k, and provide the tools to convert further procedurally generated ProcTHOR floorplan data into our format. (2) We explore the task of floorplan generation given a partial or complete set of constraints and we design a series of metrics and benchmarks to enable evaluating how well samples generated from models respect the constraints. (3) We create multiple baselines by finetuning a large language model (LLM), Llama3, and demonstrate the feasibility of using floorplan data structure conditioned LLMs for the problem of floorplan generation respecting numerical constraints. We hope that our new datasets and benchmarks will encourage further research on different ways to improve the performance of LLMs and other generative modelling techniques for generating designs where quantitative constraints are only partially specified, but must be respected.

Christopher Beckham, Alexandre Piche, David Vazquez et al.

In model-based optimisation (MBO) we are interested in using machine learning to design candidates that maximise some measure of reward with respect to a black box function called the (ground truth) oracle, which is expensive to compute since it involves executing a real world process. In offline MBO we wish to do so without assuming access to such an oracle during training or validation, with makes evaluation non-straightforward. While an approximation to the ground oracle can be trained and used in place of it during model validation to measure the mean reward over generated candidates, the evaluation is approximate and vulnerable to adversarial examples. Measuring the mean reward of generated candidates over this approximation is one such `validation metric', whereas we are interested in a more fundamental question which is finding which validation metrics correlate the most with the ground truth. This involves proposing validation metrics and quantifying them over many datasets for which the ground truth is known, for instance simulated environments. This is encapsulated under our proposed evaluation framework which is also designed to measure extrapolation, which is the ultimate goal behind leveraging generative models for MBO. While our evaluation framework is model agnostic we specifically evaluate denoising diffusion models due to their state-of-the-art performance, as well as derive interesting insights such as ranking the most effective validation metrics as well as discussing important hyperparameters.

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.

Joao Monteiro, Pierre-Andre Noel, Etienne Marcotte et al.

Large Language Models (LLMs) are trained on vast amounts of data, most of which is automatically scraped from the internet. This data includes encyclopedic documents that harbor a vast amount of general knowledge (e.g., Wikipedia) but also potentially overlap with benchmark datasets used for evaluating LLMs. Consequently, evaluating models on test splits that might have leaked into the training set is prone to misleading conclusions. To foster sound evaluation of language models, we introduce a new test dataset named RepLiQA, suited for question-answering and topic retrieval tasks. RepLiQA is a collection of five splits of test sets, four of which have not been released to the internet or exposed to LLM APIs prior to this publication. Each sample in RepLiQA comprises (1) a reference document crafted by a human annotator and depicting an imaginary scenario (e.g., a news article) absent from the internet; (2) a question about the document's topic; (3) a ground-truth answer derived directly from the information in the document; and (4) the paragraph extracted from the reference document containing the answer. As such, accurate answers can only be generated if a model can find relevant content within the provided document. We run a large-scale benchmark comprising several state-of-the-art LLMs to uncover differences in performance across models of various types and sizes in a context-conditional language modeling setting. Released splits of RepLiQA can be found here: https://huggingface.co/datasets/ServiceNow/repliqa.

Jonathan Pilault, Amine Elhattami, Christopher Pal

Multi-Task Learning (MTL) networks have emerged as a promising method for transferring learned knowledge across different tasks. However, MTL must deal with challenges such as: overfitting to low resource tasks, catastrophic forgetting, and negative task transfer, or learning interference. Often, in Natural Language Processing (NLP), a separate model per task is needed to obtain the best performance. However, many fine-tuning approaches are both parameter inefficient, i.e., potentially involving one new model per task, and highly susceptible to losing knowledge acquired during pretraining. We propose a novel Transformer architecture consisting of a new conditional attention mechanism as well as a set of task-conditioned modules that facilitate weight sharing. Through this construction (a hypernetwork adapter), we achieve more efficient parameter sharing and mitigate forgetting by keeping half of the weights of a pretrained model fixed. We also use a new multi-task data sampling strategy to mitigate the negative effects of data imbalance across tasks. Using this approach, we are able to surpass single task fine-tuning methods while being parameter and data efficient (using around 66% of the data for weight updates). Compared to other BERT Large methods on GLUE, our 8-task model surpasses other Adapter methods by 2.8% and our 24-task model outperforms by 0.7-1.0% models that use MTL and single task fine-tuning. We show that a larger variant of our single multi-task model approach performs competitively across 26 NLP tasks and yields state-of-the-art results on a number of test and development sets. Our code is publicly available at https://github.com/CAMTL/CA-MTL.

Evan Racah, Christopher Beckham, Tegan Maharaj et al.

Then detection and identification of extreme weather events in large-scale climate simulations is an important problem for risk management, informing governmental policy decisions and advancing our basic understanding of the climate system. Recent work has shown that fully supervised convolutional neural networks (CNNs) can yield acceptable accuracy for classifying well-known types of extreme weather events when large amounts of labeled data are available. However, many different types of spatially localized climate patterns are of interest including hurricanes, extra-tropical cyclones, weather fronts, and blocking events among others. Existing labeled data for these patterns can be incomplete in various ways, such as covering only certain years or geographic areas and having false negatives. This type of climate data therefore poses a number of interesting machine learning challenges. We present a multichannel spatiotemporal CNN architecture for semi-supervised bounding box prediction and exploratory data analysis. We demonstrate that our approach is able to leverage temporal information and unlabeled data to improve the localization of extreme weather events. Further, we explore the representations learned by our model in order to better understand this important data. We present a dataset, ExtremeWeather, to encourage machine learning research in this area and to help facilitate further work in understanding and mitigating the effects of climate change. The dataset is available at extremeweatherdataset.github.io and the code is available at https://github.com/eracah/hur-detect.

Shubham Agarwal, Gaurav Sahu, Abhay Puri et al.

Conducting literature reviews for scientific papers is essential for understanding research, its limitations, and building on existing work. It is a tedious task which makes an automatic literature review generator appealing. Unfortunately, many existing works that generate such reviews using Large Language Models (LLMs) have significant limitations. They tend to hallucinate-generate non-factual information-and ignore the latest research they have not been trained on. To address these limitations, we propose a toolkit that operates on Retrieval Augmented Generation (RAG) principles, specialized prompting and instructing techniques with the help of LLMs. Our system first initiates a web search to retrieve relevant papers by summarizing user-provided abstracts into keywords using an off-the-shelf LLM. Authors can enhance the search by supplementing it with relevant papers or keywords, contributing to a tailored retrieval process. Second, the system re-ranks the retrieved papers based on the user-provided abstract. Finally, the related work section is generated based on the re-ranked results and the abstract. There is a substantial reduction in time and effort for literature review compared to traditional methods, establishing our toolkit as an efficient alternative. Our project page including the demo and toolkit can be accessed here: https://litllm.github.io

Juan A. Rodriguez, Abhay Puri, Shubham Agarwal et al. · mila

Scalable Vector Graphics (SVGs) are vital for modern image rendering due to their scalability and versatility. Previous SVG generation methods have focused on curve-based vectorization, lacking semantic understanding, often producing artifacts, and struggling with SVG primitives beyond path curves. To address these issues, we introduce StarVector, a multimodal large language model for SVG generation. It performs image vectorization by understanding image semantics and using SVG primitives for compact, precise outputs. Unlike traditional methods, StarVector works directly in the SVG code space, leveraging visual understanding to apply accurate SVG primitives. To train StarVector, we create SVG-Stack, a diverse dataset of 2M samples that enables generalization across vectorization tasks and precise use of primitives like ellipses, polygons, and text. We address challenges in SVG evaluation, showing that pixel-based metrics like MSE fail to capture the unique qualities of vector graphics. We introduce SVG-Bench, a benchmark across 10 datasets, and 3 tasks: Image-to-SVG, Text-to-SVG generation, and diagram generation. Using this setup, StarVector achieves state-of-the-art performance, producing more compact and semantically rich SVGs.

Xing Han Lù, Amirhossein Kazemnejad, Nicholas Meade et al. · eth-zurich, mila

Web agents enable users to perform tasks on web browsers through natural language interaction. Evaluating web agents trajectories is an important problem, since it helps us determine whether the agent successfully completed the tasks. Rule-based methods are widely used for this purpose, but they are challenging to extend to new tasks and may not always recognize successful trajectories. We may achieve higher accuracy through human evaluation, but the process would be substantially slower and more expensive. Automatic evaluations with LLMs may avoid the challenges of designing new rules and manually annotating trajectories, enabling faster and cost-effective evaluation. However, it is unclear how effective they are at evaluating web agents. To this end, we propose AgentRewardBench, the first benchmark to assess the effectiveness of LLM judges for evaluating web agents. AgentRewardBench contains 1302 trajectories across 5 benchmarks and 4 LLMs. Each trajectory in AgentRewardBench is reviewed by an expert, who answers questions pertaining to the success, side effects, and repetitiveness of the agent. Using our benchmark, we evaluate 12 LLM judges and find that no single LLM excels across all benchmarks. We also find that the rule-based evaluation used by common benchmarks tends to underreport the success rate of web agents, highlighting a key weakness of rule-based evaluation and the need to develop more flexible automatic evaluations. We release the benchmark at: https://agent-reward-bench.github.io

Luke Rowe, Roger Girgis, Anthony Gosselin et al.

Evaluating autonomous vehicle stacks (AVs) in simulation typically involves replaying driving logs from real-world recorded traffic. However, agents replayed from offline data are not reactive and hard to intuitively control. Existing approaches address these challenges by proposing methods that rely on heuristics or generative models of real-world data but these approaches either lack realism or necessitate costly iterative sampling procedures to control the generated behaviours. In this work, we take an alternative approach and propose CtRL-Sim, a method that leverages return-conditioned offline reinforcement learning (RL) to efficiently generate reactive and controllable traffic agents. Specifically, we process real-world driving data through a physics-enhanced Nocturne simulator to generate a diverse offline RL dataset, annotated with various rewards. With this dataset, we train a return-conditioned multi-agent behaviour model that allows for fine-grained manipulation of agent behaviours by modifying the desired returns for the various reward components. This capability enables the generation of a wide range of driving behaviours beyond the scope of the initial dataset, including adversarial behaviours. We show that CtRL-Sim can generate realistic safety-critical scenarios while providing fine-grained control over agent behaviours.

Ye Yuan, Can Chen, Christopher Pal et al.

In offline multi-objective optimization (MOO), we leverage an offline dataset of designs and their associated labels to simultaneously minimize multiple objectives. This setting more closely mirrors complex real-world problems compared to single-objective optimization. Recent works mainly employ evolutionary algorithms and Bayesian optimization, with limited attention given to the generative modeling capabilities inherent in such data. In this study, we explore generative modeling in offline MOO through flow matching, noted for its effectiveness and efficiency. We introduce ParetoFlow, specifically designed to guide flow sampling to approximate the Pareto front. Traditional predictor (classifier) guidance is inadequate for this purpose because it models only a single objective. In response, we propose a multi-objective predictor guidance module that assigns each sample a weight vector, representing a weighted distribution across multiple objective predictions. A local filtering scheme is introduced to address non-convex Pareto fronts. These weights uniformly cover the entire objective space, effectively directing sample generation towards the Pareto front. Since distributions with similar weights tend to generate similar samples, we introduce a neighboring evolution module to foster knowledge sharing among neighboring distributions. This module generates offspring from these distributions, and selects the most promising one for the next iteration. Our method achieves state-of-the-art performance across various tasks.

João Monteiro, Étienne Marcotte, Pierre-André Noël et al.

In-context learning (ICL) approaches typically leverage prompting to condition decoder-only language model generation on reference information. Just-in-time processing of a context is inefficient due to the quadratic cost of self-attention operations, and caching is desirable. However, caching transformer states can easily require almost as much space as the model parameters. When the right context isn't known in advance, caching ICL can be challenging. This work addresses these limitations by introducing models that, inspired by the encoder-decoder architecture, use cross-attention to condition generation on reference text without the prompt. More precisely, we leverage pre-trained decoder-only models and only train a small number of added layers. We use Question-Answering (QA) as a testbed to evaluate the ability of our models to perform conditional generation and observe that they outperform ICL, are comparable to fine-tuned prompted LLMs, and drastically reduce the space footprint relative to standard KV caching by two orders of magnitude.

Shubham Agarwal, Gaurav Sahu, Abhay Puri et al.

Literature reviews are an essential component of scientific research, but they remain time-intensive and challenging to write, especially due to the recent influx of research papers. This paper explores the zero-shot abilities of recent Large Language Models (LLMs) in assisting with the writing of literature reviews based on an abstract. We decompose the task into two components: 1. Retrieving related works given a query abstract, and 2. Writing a literature review based on the retrieved results. We analyze how effective LLMs are for both components. For retrieval, we introduce a novel two-step search strategy that first uses an LLM to extract meaningful keywords from the abstract of a paper and then retrieves potentially relevant papers by querying an external knowledge base. Additionally, we study a prompting-based re-ranking mechanism with attribution and show that re-ranking doubles the normalized recall compared to naive search methods, while providing insights into the LLM's decision-making process. In the generation phase, we propose a two-step approach that first outlines a plan for the review and then executes steps in the plan to generate the actual review. To evaluate different LLM-based literature review methods, we create test sets from arXiv papers using a protocol designed for rolling use with newly released LLMs to avoid test set contamination in zero-shot evaluations. We release this evaluation protocol to promote additional research and development in this regard. Our empirical results suggest that LLMs show promising potential for writing literature reviews when the task is decomposed into smaller components of retrieval and planning. Our project page including a demonstration system and toolkit can be accessed here: https://litllm.github.io.

Luke Rowe, Roger Girgis, Anthony Gosselin et al.

We introduce Scenario Dreamer, a fully data-driven generative simulator for autonomous vehicle planning that generates both the initial traffic scene - comprising a lane graph and agent bounding boxes - and closed-loop agent behaviours. Existing methods for generating driving simulation environments encode the initial traffic scene as a rasterized image and, as such, require parameter-heavy networks that perform unnecessary computation due to many empty pixels in the rasterized scene. Moreover, we find that existing methods that employ rule-based agent behaviours lack diversity and realism. Scenario Dreamer instead employs a novel vectorized latent diffusion model for initial scene generation that directly operates on the vectorized scene elements and an autoregressive Transformer for data-driven agent behaviour simulation. Scenario Dreamer additionally supports scene extrapolation via diffusion inpainting, enabling the generation of unbounded simulation environments. Extensive experiments show that Scenario Dreamer outperforms existing generative simulators in realism and efficiency: the vectorized scene-generation base model achieves superior generation quality with around 2x fewer parameters, 6x lower generation latency, and 10x fewer GPU training hours compared to the strongest baseline. We confirm its practical utility by showing that reinforcement learning planning agents are more challenged in Scenario Dreamer environments than traditional non-generative simulation environments, especially on long and adversarial driving environments.

Juan A. Rodriguez, Nicholas Botzer, David Vazquez et al. · mila

In today's digitally driven world, dialogue systems play a pivotal role in enhancing user interactions, from customer service to virtual assistants. In these dialogues, it is important to identify user's goals automatically to resolve their needs promptly. This has necessitated the integration of models that perform Intent Detection. However, users' intents are diverse and dynamic, making it challenging to maintain a fixed set of predefined intents. As a result, a more practical approach is to develop a model capable of identifying new intents as they emerge. We address the challenge of Intent Discovery, an area that has drawn significant attention in recent research efforts. Existing methods need to train on a substantial amount of data for correctly identifying new intents, demanding significant human effort. To overcome this, we introduce IntentGPT, a novel training-free method that effectively prompts Large Language Models (LLMs) such as GPT-4 to discover new intents with minimal labeled data. IntentGPT comprises an \textit{In-Context Prompt Generator}, which generates informative prompts for In-Context Learning, an \textit{Intent Predictor} for classifying and discovering user intents from utterances, and a \textit{Semantic Few-Shot Sampler} that selects relevant few-shot examples and a set of known intents to be injected into the prompt. Our experiments show that IntentGPT outperforms previous methods that require extensive domain-specific data and fine-tuning, in popular benchmarks, including CLINC and BANKING, among others.

Aditya Sharma, Aman Dalmia, Mehran Kazemi et al.

Geometry problem-solving demands advanced reasoning abilities to process multimodal inputs and employ mathematical knowledge effectively. Vision-language models (VLMs) have made significant progress in various multimodal tasks. Yet, they still struggle with geometry problems and are significantly limited by their inability to perform mathematical operations not seen during pre-training, such as calculating the cosine of an arbitrary angle, and by difficulties in correctly applying relevant geometry formulas. To overcome these challenges, we present GeoCoder, which leverages modular code-finetuning to generate and execute code using a predefined geometry function library. By executing the code, we achieve accurate and deterministic calculations, contrasting the stochastic nature of autoregressive token prediction, while the function library minimizes errors in formula usage. We also propose a multimodal retrieval-augmented variant of GeoCoder, named RAG-GeoCoder, which incorporates a non-parametric memory module for retrieving functions from the geometry library, thereby reducing reliance on parametric memory. Our modular code-finetuning approach enhances the geometric reasoning capabilities of VLMs, yielding an average improvement of over 16% across various question complexities on the GeomVerse dataset compared to other finetuning methods.

Aditya Sharma, Luis Lara, Christopher J. Pal et al.

The ability to generate SPARQL queries from natural language questions is crucial for ensuring efficient and accurate retrieval of structured data from knowledge graphs (KG). While large language models (LLMs) have been widely adopted for SPARQL query generation, they are often susceptible to hallucinations and out-of-distribution errors when producing KG elements like Uniform Resource Identifiers (URIs) based on internal parametric knowledge. This often results in content that appears plausible but is factually incorrect, posing significant challenges for their use in real-world information retrieval (IR) applications. This has led to increased research aimed at detecting and mitigating such errors. In this paper, we introduce PGMR (Post-Generation Memory Retrieval), a modular framework that incorporates a non-parametric memory module to retrieve KG elements and enhance LLM-based SPARQL query generation. Our experimental results indicate that PGMR consistently delivers strong performance across diverse datasets, data distributions, and LLMs. Notably, PGMR significantly mitigates URI hallucinations, nearly eliminating the problem in several scenarios.

Md Rifat Arefin, Gopeshh Subbaraj, Nicolas Gontier et al.

Decoder-only Transformers often struggle with complex reasoning tasks, particularly arithmetic reasoning requiring multiple sequential operations. In this work, we identify representation collapse in the model's intermediate layers as a key factor limiting their reasoning capabilities. To address this, we propose Sequential Variance-Covariance Regularization (Seq-VCR), which enhances the entropy of intermediate representations and prevents collapse. Combined with dummy pause tokens as substitutes for chain-of-thought (CoT) tokens, our method significantly improves performance in arithmetic reasoning problems. In the challenging $5 \times 5$ integer multiplication task, our approach achieves $99.5\%$ exact match accuracy, outperforming models of the same size (which yield $0\%$ accuracy) and GPT-4 with five-shot CoT prompting ($44\%$). We also demonstrate superior results on arithmetic expression and longest increasing subsequence (LIS) datasets. Our findings highlight the importance of preventing intermediate layer representation collapse to enhance the reasoning capabilities of Transformers and show that Seq-VCR offers an effective solution without requiring explicit CoT supervision.

Issam Laradji, Perouz Taslakian, Sai Rajeswar et al.

The extraction of a small number of relevant insights from vast amounts of data is a crucial component of data-driven decision-making. However, accomplishing this task requires considerable technical skills, domain expertise, and human labor. This study explores the potential of using Large Language Models (LLMs) to automate the discovery of insights in data, leveraging recent advances in reasoning and code generation techniques. We propose a new evaluation methodology based on a "capture the flag" principle, measuring the ability of such models to recognize meaningful and pertinent information (flags) in a dataset. We further propose two proof-of-concept agents, with different inner workings, and compare their ability to capture such flags in a real-world sales dataset. While the work reported here is preliminary, our results are sufficiently interesting to mandate future exploration by the community.

Luke Rowe, Rodrigue de Schaetzen, Roger Girgis et al.

Maintaining good driving behavior in out-of-distribution scenarios remains a critical challenge in autonomous driving. A promising direction is to leverage the generalist knowledge and reasoning capabilities of large-language models by treating unusual driving scenarios as a logical reasoning task. In this work, we present Poutine, a method that uses an off-the-shelf 3B-parameter vision-language model (VLM) - without any additional components - to achieve robust end-to-end autonomous driving via a simple and scalable training recipe. To learn strong base driving capabilities, we first train Poutine-Base using self-supervised next-token prediction over vision, language, and trajectory (VLT) tokens, leveraging both nominal and long-tail driving data. In the second stage, we fine-tune Poutine-Base using Group Relative Policy Optimization (GRPO) with a small set of human preference-labeled examples. We evaluated our approach on the Waymo end-to-end driving benchmark curated for long-tail scenarios. The final Poutine model achieves an RFS of 7.99 on the test set, placing 1st in the 2025 Waymo Vision-Based End-to-End Driving Challenge by a significant margin. Our results suggest that handcrafted tokenizers or custom architectural components added to base VLMs in prior work are not necessary to achieve strong driving performance. Instead, this work highlights the potential of scalable VLT pretraining combined with lightweight RL fine-tuning to enable robust and generalizable autonomous driving.

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.

Simon Guiroy, Mats Richter, Sarath Chandar et al.

To create state-of-the-art models for many downstream tasks, it has become common practice to fine-tune a pre-trained large vision model. However, it remains an open question of how to best determine which of the many possible model checkpoints resulting from a large training run to use as the starting point. This becomes especially important when data for the target task of interest is scarce, unlabeled and out-of-distribution. In such scenarios, common methods relying on in-distribution validation data become unreliable or inapplicable. This work proposes a novel approach for model selection that operates reliably on just a few unlabeled examples from the target task. Our approach is based on a novel concept: Neural Coherence, which entails characterizing a model's activation statistics for source and target domains, allowing one to define model selection methods with high data-efficiency. We provide experiments where models are pre-trained on ImageNet1K and examine target domains consisting of Food-101, PlantNet-300K and iNaturalist. We also evaluate it in many meta-learning settings. Our approach significantly improves generalization across these different target domains compared to established baselines. We further demonstrate the versatility of Neural Coherence as a powerful principle by showing its effectiveness in training data selection.

Gaurav Sahu, Hugo Larochelle, Laurent Charlin et al.

Peer review is the cornerstone of scientific publishing, yet it suffers from inconsistencies, reviewer subjectivity, and scalability challenges. We introduce ReviewerToo, a modular framework for studying and deploying AI-assisted peer review to complement human judgment with systematic and consistent assessments. ReviewerToo supports systematic experiments with specialized reviewer personas and structured evaluation criteria, and can be partially or fully integrated into real conference workflows. We validate ReviewerToo on a carefully curated dataset of 1,963 paper submissions from ICLR 2025, where our experiments with the gpt-oss-120b model achieves 81.8% accuracy for the task of categorizing a paper as accept/reject compared to 83.9% for the average human reviewer. Additionally, ReviewerToo-generated reviews are rated as higher quality than the human average by an LLM judge, though still trailing the strongest expert contributions. Our analysis highlights domains where AI reviewers excel (e.g., fact-checking, literature coverage) and where they struggle (e.g., assessing methodological novelty and theoretical contributions), underscoring the continued need for human expertise. Based on these findings, we propose guidelines for integrating AI into peer-review pipelines, showing how AI can enhance consistency, coverage, and fairness while leaving complex evaluative judgments to domain experts. Our work provides a foundation for systematic, hybrid peer-review systems that scale with the growth of scientific publishing.

Shambhavi Mishra, Gaurav Sahu, Marco Pedersoli et al.

Large language models (LLMs) demonstrate impressive capabilities across a wide range of tasks, yet it remains unclear whether such success reflects genuine reasoning or sophisticated recall. We introduce AInstein, a framework for testing whether LLMs can generate valid solutions to AI research problems using only their pretrained parametric knowledge -- without domain-specific fine-tuning, retrieval augmentation, or other external aids. Our approach extracts distilled problem statements from high-quality ICLR 2025 submissions, then tasks specialized solver agents with proposing and refining technical solutions through iterative critique loops, mimicking the cycles of proposal, review, and revision central to scientific inquiry. We evaluate AInstein on 1,214 ICLR papers stratified by acceptance tier (Oral, Spotlight, Poster), using an LLM-as-a-judge paradigm guided by a structured rubric, complemented by targeted manual checks. Performance is assessed with three metrics: Success Rate (does the solution address the problem?), Rediscovery (does it align with human-proposed methods?), and Novelty (does it yield valid, original approaches?). Our results reveal that while LLMs can rediscover feasible solutions and occasionally propose creative alternatives, their problem-solving ability remains fragile and highly sensitive to framing. These findings provide the first large-scale evidence on the extent to which LLMs can act as autonomous scientific problem-solvers, highlighting both their latent potential and their current limitations.

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.

Anthony Gosselin, Ge Ya Luo, Luis Lara et al.

Video diffusion techniques have advanced significantly in recent years; however, they struggle to generate realistic imagery of car crashes due to the scarcity of accident events in most driving datasets. Improving traffic safety requires realistic and controllable accident simulations. To tackle the problem, we propose Ctrl-Crash, a controllable car crash video generation model that conditions on signals such as bounding boxes, crash types, and an initial image frame. Our approach enables counterfactual scenario generation where minor variations in input can lead to dramatically different crash outcomes. To support fine-grained control at inference time, we leverage classifier-free guidance with independently tunable scales for each conditioning signal. Ctrl-Crash achieves state-of-the-art performance across quantitative video quality metrics (e.g., FVD and JEDi) and qualitative measurements based on a human-evaluation of physical realism and video quality compared to prior diffusion-based methods.

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.