Vibhav Vineet

Saachi Jain, Hadi Salman, Eric Wong et al. · mit

Missingness, or the absence of features from an input, is a concept fundamental to many model debugging tools. However, in computer vision, pixels cannot simply be removed from an image. One thus tends to resort to heuristics such as blacking out pixels, which may in turn introduce bias into the debugging process. We study such biases and, in particular, show how transformer-based architectures can enable a more natural implementation of missingness, which side-steps these issues and improves the reliability of model debugging in practice. Our code is available at https://github.com/madrylab/missingness

Christopher Agia, Krishna Murthy Jatavallabhula, Mohamed Khodeir et al. · mit, stanford

3D scene graphs (3DSGs) are an emerging description; unifying symbolic, topological, and metric scene representations. However, typical 3DSGs contain hundreds of objects and symbols even for small environments; rendering task planning on the full graph impractical. We construct TASKOGRAPHY, the first large-scale robotic task planning benchmark over 3DSGs. While most benchmarking efforts in this area focus on vision-based planning, we systematically study symbolic planning, to decouple planning performance from visual representation learning. We observe that, among existing methods, neither classical nor learning-based planners are capable of real-time planning over full 3DSGs. Enabling real-time planning demands progress on both (a) sparsifying 3DSGs for tractable planning and (b) designing planners that better exploit 3DSG hierarchies. Towards the former goal, we propose SCRUB, a task-conditioned 3DSG sparsification method; enabling classical planners to match and in some cases surpass state-of-the-art learning-based planners. Towards the latter goal, we propose SEEK, a procedure enabling learning-based planners to exploit 3DSG structure, reducing the number of replanning queries required by current best approaches by an order of magnitude. We will open-source all code and baselines to spur further research along the intersections of robot task planning, learning and 3DSGs.

Vidhisha Balachandran, Jingya Chen, Neel Joshi et al. · cmu

Rigorous and reproducible evaluation is critical for assessing the state of the art and for guiding scientific advances in Artificial Intelligence. Evaluation is challenging in practice due to several reasons, including benchmark saturation, lack of transparency in methods used for measurement, development challenges in extracting measurements for generative tasks, and, more generally, the extensive number of capabilities required for a well-rounded comparison across models. We make three contributions to alleviate the above challenges. First, we present Eureka, an open-source framework for standardizing evaluations of large foundation models beyond single-score reporting and rankings. Second, we introduce Eureka-Bench as an extensible collection of benchmarks testing capabilities that (i) are still challenging for state-of-the-art models and (ii) represent fundamental but overlooked language and multimodal capabilities. The inherent space for improvement in non-saturated benchmarks enables us to discover meaningful differences between models at a capability level. Third, using Eureka, we conduct an analysis of 12 state-of-the-art models, providing in-depth insights into failure understanding and model comparison, which can be leveraged to plan targeted improvements. In contrast to recent trends in reports and leaderboards showing absolute rankings and claims for one model or another to be the best, our analysis shows that there is no such best model. Different models have different strengths, but there are models that appear more often than others as best performers for some capabilities. Despite the recent improvements, current models still struggle with several fundamental capabilities including detailed image understanding, benefiting from multimodal input when available rather than fully relying on language, factuality and grounding for information retrieval, and over refusals.

Yunhao Ge, Jiashu Xu, Brian Nlong Zhao et al. · harvard

We propose a new paradigm to automatically generate training data with accurate labels at scale using the text-to-image synthesis frameworks (e.g., DALL-E, Stable Diffusion, etc.). The proposed approach1 decouples training data generation into foreground object generation, and contextually coherent background generation. To generate foreground objects, we employ a straightforward textual template, incorporating the object class name as input prompts. This is fed into a text-to-image synthesis framework, producing various foreground images set against isolated backgrounds. A foreground-background segmentation algorithm is then used to generate foreground object masks. To generate context images, we begin by creating language descriptions of the context. This is achieved by applying an image captioning method to a small set of images representing the desired context. These textual descriptions are then transformed into a diverse array of context images via a text-to-image synthesis framework. Subsequently, we composite these with the foreground object masks produced in the initial step, utilizing a cut-and-paste method, to formulate the training data. We demonstrate the advantages of our approach on five object detection and segmentation datasets, including Pascal VOC and COCO. We found that detectors trained solely on synthetic data produced by our method achieve performance comparable to those trained on real data (Fig. 1). Moreover, a combination of real and synthetic data yields even much better results. Further analysis indicates that the synthetic data distribution complements the real data distribution effectively. Additionally, we emphasize the compositional nature of our data generation approach in out-of-distribution and zero-shot data generation scenarios. We open-source our code at https://github.com/gyhandy/Text2Image-for-Detection

Yunhao Ge, Harkirat Behl, Jiashu Xu et al. · harvard, microsoft-research

Training computer vision models usually requires collecting and labeling vast amounts of imagery under a diverse set of scene configurations and properties. This process is incredibly time-consuming, and it is challenging to ensure that the captured data distribution maps well to the target domain of an application scenario. Recently, synthetic data has emerged as a way to address both of these issues. However, existing approaches either require human experts to manually tune each scene property or use automatic methods that provide little to no control; this requires rendering large amounts of random data variations, which is slow and is often suboptimal for the target domain. We present the first fully differentiable synthetic data pipeline that uses Neural Radiance Fields (NeRFs) in a closed-loop with a target application's loss function. Our approach generates data on-demand, with no human labor, to maximize accuracy for a target task. We illustrate the effectiveness of our method on synthetic and real-world object detection tasks. We also introduce a new "YCB-in-the-Wild" dataset and benchmark that provides a test scenario for object detection with varied poses in real-world environments.

Yash Jain, Harkirat Behl, Zsolt Kira et al. · gatech, microsoft-research

Construction of a universal detector poses a crucial question: How can we most effectively train a model on a large mixture of datasets? The answer lies in learning dataset-specific features and ensembling their knowledge but do all this in a single model. Previous methods achieve this by having separate detection heads on a common backbone but that results in a significant increase in parameters. In this work, we present Mixture-of-Experts as a solution, highlighting that MoEs are much more than a scalability tool. We propose Dataset-Aware Mixture-of-Experts, DAMEX where we train the experts to become an `expert' of a dataset by learning to route each dataset tokens to its mapped expert. Experiments on Universal Object-Detection Benchmark show that we outperform the existing state-of-the-art by average +10.2 AP score and improve over our non-MoE baseline by average +2.0 AP score. We also observe consistent gains while mixing datasets with (1) limited availability, (2) disparate domains and (3) divergent label sets. Further, we qualitatively show that DAMEX is robust against expert representation collapse.

Yunhao Ge, Jiashu Xu, Brian Nlong Zhao et al. · harvard

We propose a new paradigm to automatically generate training data with accurate labels at scale using the text-toimage synthesis frameworks (e.g., DALL-E, Stable Diffusion, etc.). The proposed approach decouples training data generation into foreground object mask generation and background (context) image generation. For foreground object mask generation, we use a simple textual template with object class name as input to DALL-E to generate a diverse set of foreground images. A foreground-background segmentation algorithm is then used to generate foreground object masks. Next, in order to generate context images, first a language description of the context is generated by applying an image captioning method on a small set of images representing the context. These language descriptions are then used to generate diverse sets of context images using the DALL-E framework. These are then composited with object masks generated in the first step to provide an augmented training set for a classifier. We demonstrate the advantages of our approach on four object detection datasets including on Pascal VOC and COCO object detection tasks. Furthermore, we also highlight the compositional nature of our data generation approach on out-of-distribution and zero-shot data generation scenarios.

Tyler LaBonte, Yale Song, Xin Wang et al.

A critical object detection task is finetuning an existing model to detect novel objects, but the standard workflow requires bounding box annotations which are time-consuming and expensive to collect. Weakly supervised object detection (WSOD) offers an appealing alternative, where object detectors can be trained using image-level labels. However, the practical application of current WSOD models is limited, as they only operate at small data scales and require multiple rounds of training and refinement. To address this, we propose the Weakly Supervised Detection Transformer, which enables efficient knowledge transfer from a large-scale pretraining dataset to WSOD finetuning on hundreds of novel objects. Additionally, we leverage pretrained knowledge to improve the multiple instance learning (MIL) framework often used in WSOD methods. Our experiments show that our approach outperforms previous state-of-the-art models on large-scale novel object detection datasets, and our scaling study reveals that class quantity is more important than image quantity for WSOD pretraining. The code is available at https://github.com/tmlabonte/weakly-supervised-DETR.

Garrett Thomas, Ching-An Cheng, Ricky Loynd et al. · microsoft-research

A rich representation is key to general robotic manipulation, but existing approaches to representation learning require large amounts of multimodal demonstrations. In this work we propose PLEX, a transformer-based architecture that learns from a small amount of task-agnostic visuomotor trajectories and a much larger amount of task-conditioned object manipulation videos -- a type of data available in quantity. PLEX uses visuomotor trajectories to induce a latent feature space and to learn task-agnostic manipulation routines, while diverse video-only demonstrations teach PLEX how to plan in the induced latent feature space for a wide variety of tasks. Experiments showcase PLEX's generalization on Meta-World and SOTA performance in challenging Robosuite environments. In particular, using relative positional encoding in PLEX's transformers greatly helps in low-data regimes of learning from human-collected demonstrations. The paper's accompanying code and data are available at https://microsoft.github.io/PLEX.

Tejas Gokhale, Hamid Palangi, Besmira Nushi et al.

Spatial understanding is a fundamental aspect of computer vision and integral for human-level reasoning about images, making it an important component for grounded language understanding. While recent text-to-image synthesis (T2I) models have shown unprecedented improvements in photorealism, it is unclear whether they have reliable spatial understanding capabilities. We investigate the ability of T2I models to generate correct spatial relationships among objects and present VISOR, an evaluation metric that captures how accurately the spatial relationship described in text is generated in the image. To benchmark existing models, we introduce a dataset, $\mathrm{SR}_{2D}$, that contains sentences describing two or more objects and the spatial relationships between them. We construct an automated evaluation pipeline to recognize objects and their spatial relationships, and employ it in a large-scale evaluation of T2I models. Our experiments reveal a surprising finding that, although state-of-the-art T2I models exhibit high image quality, they are severely limited in their ability to generate multiple objects or the specified spatial relations between them. Our analyses demonstrate several biases and artifacts of T2I models such as the difficulty with generating multiple objects, a bias towards generating the first object mentioned, spatially inconsistent outputs for equivalent relationships, and a correlation between object co-occurrence and spatial understanding capabilities. We conduct a human study that shows the alignment between VISOR and human judgement about spatial understanding. We offer the $\mathrm{SR}_{2D}$ dataset and the VISOR metric to the community in support of T2I reasoning research.

Shehreen Azad, Yash Jain, Rishit Garg et al. · gatech, microsoft-research

Geometric understanding - including depth and height perception - is fundamental to intelligence and crucial for navigating our environment. Despite the impressive capabilities of large Vision Language Models (VLMs), it remains unclear how well they possess the geometric understanding required for practical applications in visual perception. In this work, we focus on evaluating the geometric understanding of these models, specifically targeting their ability to perceive the depth and height of objects in an image. To address this, we introduce GeoMeter, a suite of benchmark datasets - encompassing 2D and 3D scenarios - to rigorously evaluate these aspects. By benchmarking 18 state-of-the-art VLMs, we found that although they excel in perceiving basic geometric properties like shape and size, they consistently struggle with depth and height perception. Our analysis reveal that these challenges stem from shortcomings in their depth and height reasoning capabilities and inherent biases. This study aims to pave the way for developing VLMs with enhanced geometric understanding by emphasizing depth and height perception as critical components necessary for real-world applications.

Yunhao Ge, Jiashu Xu, Brian Nlong Zhao et al. · harvard

We propose EM-PASTE: an Expectation Maximization(EM) guided Cut-Paste compositional dataset augmentation approach for weakly-supervised instance segmentation using only image-level supervision. The proposed method consists of three main components. The first component generates high-quality foreground object masks. To this end, an EM-like approach is proposed that iteratively refines an initial set of object mask proposals generated by a generic region proposal method. Next, in the second component, high-quality context-aware background images are generated using a text-to-image compositional synthesis method like DALL-E. Finally, the third component creates a large-scale pseudo-labeled instance segmentation training dataset by compositing the foreground object masks onto the original and generated background images. The proposed approach achieves state-of-the-art weakly-supervised instance segmentation results on both the PASCAL VOC 2012 and MS COCO datasets by using only image-level, weak label information. In particular, it outperforms the best baseline by +7.4 and +2.8 mAP0.50 on PASCAL and COCO, respectively. Further, the method provides a new solution to the long-tail weakly-supervised instance segmentation problem (when many classes may only have few training samples), by selectively augmenting under-represented classes.

Eric Heiden, Ziang Liu, Vibhav Vineet et al.

Being able to reproduce physical phenomena ranging from light interaction to contact mechanics, simulators are becoming increasingly useful in more and more application domains where real-world interaction or labeled data are difficult to obtain. Despite recent progress, significant human effort is needed to configure simulators to accurately reproduce real-world behavior. We introduce a pipeline that combines inverse rendering with differentiable simulation to create digital twins of real-world articulated mechanisms from depth or RGB videos. Our approach automatically discovers joint types and estimates their kinematic parameters, while the dynamic properties of the overall mechanism are tuned to attain physically accurate simulations. Control policies optimized in our derived simulation transfer successfully back to the original system, as we demonstrate on a simulated system. Further, our approach accurately reconstructs the kinematic tree of an articulated mechanism being manipulated by a robot, and highly nonlinear dynamics of a real-world coupled pendulum mechanism. Website: https://eric-heiden.github.io/video2sim

Benoit Guillard, Sai Vemprala, Jayesh K. Gupta et al.

Simulating realistic sensors is a challenging part in data generation for autonomous systems, often involving carefully handcrafted sensor design, scene properties, and physics modeling. To alleviate this, we introduce a pipeline for data-driven simulation of a realistic LiDAR sensor. We propose a model that learns a mapping between RGB images and corresponding LiDAR features such as raydrop or per-point intensities directly from real datasets. We show that our model can learn to encode realistic effects such as dropped points on transparent surfaces or high intensity returns on reflective materials. When applied to naively raycasted point clouds provided by off-the-shelf simulator software, our model enhances the data by predicting intensities and removing points based on the scene's appearance to match a real LiDAR sensor. We use our technique to learn models of two distinct LiDAR sensors and use them to improve simulated LiDAR data accordingly. Through a sample task of vehicle segmentation, we show that enhancing simulated point clouds with our technique improves downstream task performance.

Madeline Chantry Schiappa, Naman Biyani, Prudvi Kamtam et al.

We have seen a great progress in video action recognition in recent years. There are several models based on convolutional neural network (CNN) and some recent transformer based approaches which provide top performance on existing benchmarks. In this work, we perform a large-scale robustness analysis of these existing models for video action recognition. We focus on robustness against real-world distribution shift perturbations instead of adversarial perturbations. We propose four different benchmark datasets, HMDB51-P, UCF101-P, Kinetics400-P, and SSv2-P to perform this analysis. We study robustness of six state-of-the-art action recognition models against 90 different perturbations. The study reveals some interesting findings, 1) transformer based models are consistently more robust compared to CNN based models, 2) Pretraining improves robustness for Transformer based models more than CNN based models, and 3) All of the studied models are robust to temporal perturbations for all datasets but SSv2; suggesting the importance of temporal information for action recognition varies based on the dataset and activities. Next, we study the role of augmentations in model robustness and present a real-world dataset, UCF101-DS, which contains realistic distribution shifts, to further validate some of these findings. We believe this study will serve as a benchmark for future research in robust video action recognition.

Madeline C. Schiappa, Shruti Vyas, Hamid Palangi et al.

Joint visual and language modeling on large-scale datasets has recently shown good progress in multi-modal tasks when compared to single modal learning. However, robustness of these approaches against real-world perturbations has not been studied. In this work, we perform the first extensive robustness study of video-language models against various real-world perturbations. We focus on text-to-video retrieval and propose two large-scale benchmark datasets, MSRVTT-P and YouCook2-P, which utilize 90 different visual and 35 different text perturbations. The study reveals some interesting initial findings from the studied models: 1) models are generally more susceptible when only video is perturbed as opposed to when only text is perturbed, 2) models that are pre-trained are more robust than those trained from scratch, 3) models attend more to scene and objects rather than motion and action. We hope this study will serve as a benchmark and guide future research in robust video-language learning. The benchmark introduced in this study along with the code and datasets is available at https://bit.ly/3CNOly4.

Madeline Chantry Schiappa, Shehreen Azad, Sachidanand VS et al.

Due to the increase in computational resources and accessibility of data, an increase in large, deep learning models trained on copious amounts of multi-modal data using self-supervised or semi-supervised learning have emerged. These ``foundation'' models are often adapted to a variety of downstream tasks like classification, object detection, and segmentation with little-to-no training on the target dataset. In this work, we perform a robustness analysis of Visual Foundation Models (VFMs) for segmentation tasks and focus on robustness against real-world distribution shift inspired perturbations. We benchmark seven state-of-the-art segmentation architectures using 2 different perturbed datasets, MS COCO-P and ADE20K-P, with 17 different perturbations with 5 severity levels each. Our findings reveal several key insights: (1) VFMs exhibit vulnerabilities to compression-induced corruptions, (2) despite not outpacing all of unimodal models in robustness, multimodal models show competitive resilience in zero-shot scenarios, and (3) VFMs demonstrate enhanced robustness for certain object categories. These observations suggest that our robustness evaluation framework sets new requirements for foundational models, encouraging further advancements to bolster their adaptability and performance. The code and dataset is available at: \url{https://tinyurl.com/fm-robust}.

Sharath Girish, Debadeepta Dey, Neel Joshi et al.

The current literature on self-supervised learning (SSL) focuses on developing learning objectives to train neural networks more effectively on unlabeled data. The typical development process involves taking well-established architectures, e.g., ResNet demonstrated on ImageNet, and using them to evaluate newly developed objectives on downstream scenarios. While convenient, this does not take into account the role of architectures which has been shown to be crucial in the supervised learning literature. In this work, we establish extensive empirical evidence showing that a network architecture plays a significant role in SSL. We conduct a large-scale study with over 100 variants of ResNet and MobileNet architectures and evaluate them across 11 downstream scenarios in the SSL setting. We show that there is no one network that performs consistently well across the scenarios. Based on this, we propose to learn not only network weights but also architecture topologies in the SSL regime. We show that "self-supervised architectures" outperform popular handcrafted architectures (ResNet18 and MobileNetV2) while performing competitively with the larger and computationally heavy ResNet50 on major image classification benchmarks (ImageNet-1K, iNat2021, and more). Our results suggest that it is time to consider moving beyond handcrafted architectures in SSL and start thinking about incorporating architecture search into self-supervised learning objectives.

Nishant Jain, Harkirat Behl, Yogesh Singh Rawat et al.

A recent trend in deep learning algorithms has been towards training large scale models, having high parameter count and trained on big dataset. However, robustness of such large scale models towards real-world settings is still a less-explored topic. In this work, we first benchmark the performance of these models under different perturbations and datasets thereby representing real-world shifts, and highlight their degrading performance under these shifts. We then discuss on how complete model fine-tuning based existing robustification schemes might not be a scalable option given very large scale networks and can also lead them to forget some of the desired characterstics. Finally, we propose a simple and cost-effective method to solve this problem, inspired by knowledge transfer literature. It involves robustifying smaller models, at a lower computation cost, and then use them as teachers to tune a fraction of these large scale networks, reducing the overall computational overhead. We evaluate our proposed method under various vision perturbations including ImageNet-C,R,S,A datasets and also for transfer learning, zero-shot evaluation setups on different datasets. Benchmark results show that our method is able to induce robustness to these large scale models efficiently, requiring significantly lower time and also preserves the transfer learning, zero-shot properties of the original model which none of the existing methods are able to achieve.

Akash Kumar, Ashlesha Kumar, Vibhav Vineet et al.

Self-supervised learning is an effective way for label-free model pre-training, especially in the video domain where labeling is expensive. Existing self-supervised works in the video domain use varying experimental setups to demonstrate their effectiveness and comparison across approaches becomes challenging with no standard benchmark. In this work, we first provide a benchmark that enables a comparison of existing approaches on the same ground. Next, we study five different aspects of self-supervised learning important for videos; 1) dataset size, 2) complexity, 3) data distribution, 4) data noise, and, 5)feature analysis. To facilitate this study, we focus on seven different methods along with seven different network architectures and perform an extensive set of experiments on 5 different datasets with an evaluation of two different downstream tasks. We present several interesting insights from this study which span across different properties of pretraining and target datasets, pretext-tasks, and model architectures among others. We further put some of these insights to the real test and propose an approach that requires a limited amount of training data and outperforms existing state-of-the-art approaches which use 10x pretraining data. We believe this work will pave the way for researchers to a better understanding of self-supervised pretext tasks in video representation learning.

Jinoh Cho, Minguk Kang, Vibhav Vineet et al.

Image completion is a task that aims to fill in the missing region of a masked image with plausible contents. However, existing image completion methods tend to fill in the missing region with the surrounding texture instead of hallucinating a visual instance that is suitable in accordance with the context of the scene. In this work, we propose a novel image completion model, dubbed ImComplete, that hallucinates the missing instance that harmonizes well with - and thus preserves - the original context. ImComplete first adopts a transformer architecture that considers the visible instances and the location of the missing region. Then, ImComplete completes the semantic segmentation masks within the missing region, providing pixel-level semantic and structural guidance. Finally, the image synthesis blocks generate photo-realistic content. We perform a comprehensive evaluation of the results in terms of visual quality (LPIPS and FID) and contextual preservation scores (CLIPscore and object detection accuracy) with COCO-panoptic and Visual Genome datasets. Experimental results show the superiority of ImComplete on various natural images.

Jing Wu, Daphne Barretto, Yiye Chen et al.

Long-horizon, repetitive workflows are common in professional settings, such as processing expense reports from receipts and entering student grades from exam papers. These tasks are often tedious for humans since they can extend to extreme lengths proportional to the size of the data to process. However, they are ideal for Computer-Use Agents (CUAs) due to their structured, recurring sub-workflows with logic that can be systematically learned. Identifying the absence of an evaluation benchmark as a primary bottleneck, we establish OS-Marathon, comprising 242 long-horizon, repetitive tasks across 2 domains to evaluate state-of-the-art (SOTA) agents. We then introduce a cost-effective method to construct a condensed demonstration using only few-shot examples to teach agents the underlying workflow logic, enabling them to execute similar workflows effectively on larger, unseen data collections. Extensive experiments demonstrate both the inherent challenges of these tasks and the effectiveness of our proposed method. Project website: https://os-marathon.github.io/.

Yuhang He, Yash Jain, Xubo Liu et al.

Despite significant advancements in Text-to-Audio (TTA) generation models achieving high-fidelity audio with fine-grained context understanding, they struggle to model the relations between audio events described in the input text. However, previous TTA methods have not systematically explored audio event relation modeling, nor have they proposed frameworks to enhance this capability. In this work, we systematically study audio event relation modeling in TTA generation models. We first establish a benchmark for this task by: 1. proposing a comprehensive relation corpus covering all potential relations in real-world scenarios; 2. introducing a new audio event corpus encompassing commonly heard audios; and 3. proposing new evaluation metrics to assess audio event relation modeling from various perspectives. Furthermore, we propose a finetuning framework to enhance existing TTA models ability to model audio events relation. Code is available at: https://github.com/yuhanghe01/RiTTA

Rajat Modi, Vibhav Vineet, Yogesh Singh Rawat

This paper explores the impact of occlusions in video action detection. We facilitate this study by introducing five new benchmark datasets namely O-UCF and O-JHMDB consisting of synthetically controlled static/dynamic occlusions, OVIS-UCF and OVIS-JHMDB consisting of occlusions with realistic motions and Real-OUCF for occlusions in realistic-world scenarios. We formally confirm an intuitive expectation: existing models suffer a lot as occlusion severity is increased and exhibit different behaviours when occluders are static vs when they are moving. We discover several intriguing phenomenon emerging in neural nets: 1) transformers can naturally outperform CNN models which might have even used occlusion as a form of data augmentation during training 2) incorporating symbolic-components like capsules to such backbones allows them to bind to occluders never even seen during training and 3) Islands of agreement can emerge in realistic images/videos without instance-level supervision, distillation or contrastive-based objectives2(eg. video-textual training). Such emergent properties allow us to derive simple yet effective training recipes which lead to robust occlusion models inductively satisfying the first two stages of the binding mechanism (grouping/segregation). Models leveraging these recipes outperform existing video action-detectors under occlusion by 32.3% on O-UCF, 32.7% on O-JHMDB & 2.6% on Real-OUCF in terms of the vMAP metric. The code for this work has been released at https://github.com/rajatmodi62/OccludedActionBenchmark.

Siddharth Joshi, Besmira Nushi, Vidhisha Balachandran et al. · cmu

Vision-language models (VLMs) are highly effective but often underperform on specialized tasks; for example, Llava-1.5 struggles with chart and diagram understanding due to scarce task-specific training data. Existing training data, sourced from general-purpose datasets, fails to capture the nuanced details needed for these tasks. We introduce MM-Gen, a scalable method that generates task-specific, high-quality synthetic text for candidate images by leveraging stronger models. MM-Gen employs a three-stage targeted process: partitioning data into subgroups, generating targeted text based on task descriptions, and filtering out redundant and outlier data. Fine-tuning VLMs with data generated by MM-Gen leads to significant performance gains, including 29% on spatial reasoning and 15% on diagram understanding for Llava-1.5 (7B). Compared to human-curated caption data, MM-Gen achieves up to 1.6x better improvements for the original models, proving its effectiveness in enhancing task-specific VLM performance and bridging the gap between general-purpose datasets and specialized requirements. Code available at https://github.com/sjoshi804/MM-Gen.

Amirhossein Abaskohi, Yuhang He, Peter West et al.

Computer-use agents~(CUAs) rely on visual observations of graphical user interfaces, where each screenshot is encoded into a large number of visual tokens. As interaction trajectories grow, the token cost increases rapidly, limiting the amount of history that can be incorporated under fixed context and compute budgets. This has resulted in no or very limited improvement in the performance when using history unlike other domains. We address this inefficiency by introducing ReVision, which is used to train multimodal language models on trajectories where redundant visual patches are removed using a learned patch selector that compares patch representations across consecutive screenshots while preserving spatial structure required by the model. Across three benchmarks, OSWorld, WebTailBench, and AgentNetBench, when processing trajectories with 5 history screenshots using Qwen2.5-VL-7B, ReVision reduces token usage by approximately 46% on average while improving success rate by 3% over the no drop baseline. This establishes a clear efficiency gain, enabling agents to process longer trajectories with fewer tokens. With this improved efficiency, we revisit the role of history in CUAs and find that performance continues to improve as more past observations are incorporated when redundancy is removed. This suggests that the commonly observed saturation in visual history is not due to limited usefulness of past information, but rather a consequence of inefficient token representations.

Guillaume Leclerc, Hadi Salman, Andrew Ilyas et al.

We introduce 3DB: an extendable, unified framework for testing and debugging vision models using photorealistic simulation. We demonstrate, through a wide range of use cases, that 3DB allows users to discover vulnerabilities in computer vision systems and gain insights into how models make decisions. 3DB captures and generalizes many robustness analyses from prior work, and enables one to study their interplay. Finally, we find that the insights generated by the system transfer to the physical world. We are releasing 3DB as a library (https://github.com/3db/3db) alongside a set of example analyses, guides, and documentation: https://3db.github.io/3db/ .

Stuart Golodetz, Michael Sapienza, Julien P. C. Valentin et al.

We present an open-source, real-time implementation of SemanticPaint, a system for geometric reconstruction, object-class segmentation and learning of 3D scenes. Using our system, a user can walk into a room wearing a depth camera and a virtual reality headset, and both densely reconstruct the 3D scene and interactively segment the environment into object classes such as 'chair', 'floor' and 'table'. The user interacts physically with the real-world scene, touching objects and using voice commands to assign them appropriate labels. These user-generated labels are leveraged by an online random forest-based machine learning algorithm, which is used to predict labels for previously unseen parts of the scene. The entire pipeline runs in real time, and the user stays 'in the loop' throughout the process, receiving immediate feedback about the progress of the labelling and interacting with the scene as necessary to refine the predicted segmentation.

Yash Jain, Anshul Nasery, Vibhav Vineet et al. · gatech, microsoft-research

Modern video generation models like Sora have achieved remarkable success in producing high-quality videos. However, a significant limitation is their inability to offer interactive control to users, a feature that promises to open up unprecedented applications and creativity. In this work, we introduce the first solution to equip diffusion-based video generation models with spatio-temporal control. We present Peekaboo, a novel masked attention module, which seamlessly integrates with current video generation models offering control without the need for additional training or inference overhead. To facilitate future research, we also introduce a comprehensive benchmark for interactive video generation. This benchmark offers a standardized framework for the community to assess the efficacy of emerging interactive video generation models. Our extensive qualitative and quantitative assessments reveal that Peekaboo achieves up to a 3.8x improvement in mIoU over baseline models, all while maintaining the same latency. Code and benchmark are available on the webpage.

Marah Abdin, Sahaj Agarwal, Ahmed Awadallah et al. · cmu

We introduce Phi-4-reasoning, a 14-billion parameter reasoning model that achieves strong performance on complex reasoning tasks. Trained via supervised fine-tuning of Phi-4 on carefully curated set of "teachable" prompts-selected for the right level of complexity and diversity-and reasoning demonstrations generated using o3-mini, Phi-4-reasoning generates detailed reasoning chains that effectively leverage inference-time compute. We further develop Phi-4-reasoning-plus, a variant enhanced through a short phase of outcome-based reinforcement learning that offers higher performance by generating longer reasoning traces. Across a wide range of reasoning tasks, both models outperform significantly larger open-weight models such as DeepSeek-R1-Distill-Llama-70B model and approach the performance levels of full DeepSeek-R1 model. Our comprehensive evaluations span benchmarks in math and scientific reasoning, coding, algorithmic problem solving, planning, and spatial understanding. Interestingly, we observe a non-trivial transfer of improvements to general-purpose benchmarks as well. In this report, we provide insights into our training data, our training methodologies, and our evaluations. We show that the benefit of careful data curation for supervised fine-tuning (SFT) extends to reasoning language models, and can be further amplified by reinforcement learning (RL). Finally, our evaluation points to opportunities for improving how we assess the performance and robustness of reasoning models.

Vidhisha Balachandran, Jingya Chen, Lingjiao Chen et al. · cmu

Inference-time scaling can enhance the reasoning capabilities of large language models (LLMs) on complex problems that benefit from step-by-step problem solving. Although lengthening generated scratchpads has proven effective for mathematical tasks, the broader impact of this approach on other tasks remains less clear. In this work, we investigate the benefits and limitations of scaling methods across nine state-of-the-art models and eight challenging tasks, including math and STEM reasoning, calendar planning, NP-hard problems, navigation, and spatial reasoning. We compare conventional models (e.g., GPT-4o) with models fine-tuned for inference-time scaling (e.g., o1) through evaluation protocols that involve repeated model calls, either independently or sequentially with feedback. These evaluations approximate lower and upper performance bounds and potential for future performance improvements for each model, whether through enhanced training or multi-model inference systems. Our extensive empirical analysis reveals that the advantages of inference-time scaling vary across tasks and diminish as problem complexity increases. In addition, simply using more tokens does not necessarily translate to higher accuracy in these challenging regimes. Results from multiple independent runs with conventional models using perfect verifiers show that, for some tasks, these models can achieve performance close to the average performance of today's most advanced reasoning models. However, for other tasks, a significant performance gap remains, even in very high scaling regimes. Encouragingly, all models demonstrate significant gains when inference is further scaled with perfect verifiers or strong feedback, suggesting ample potential for future improvements.

Shehreen Azad, Vibhav Vineet, Yogesh Singh Rawat

Despite advancements in multimodal large language models (MLLMs), current approaches struggle in medium-to-long video understanding due to frame and context length limitations. As a result, these models often depend on frame sampling, which risks missing key information over time and lacks task-specific relevance. To address these challenges, we introduce HierarQ, a task-aware hierarchical Q-Former based framework that sequentially processes frames to bypass the need for frame sampling, while avoiding LLM's context length limitations. We introduce a lightweight two-stream language-guided feature modulator to incorporate task awareness in video understanding, with the entity stream capturing frame-level object information within a short context and the scene stream identifying their broader interactions over longer period of time. Each stream is supported by dedicated memory banks which enables our proposed Hierachical Querying transformer (HierarQ) to effectively capture short and long-term context. Extensive evaluations on 10 video benchmarks across video understanding, question answering, and captioning tasks demonstrate HierarQ's state-of-the-art performance across most datasets, proving its robustness and efficiency for comprehensive video analysis.

Brian Nlong Zhao, Yuhang Xiao, Jiashu Xu et al. · harvard

The popularization of Text-to-Image (T2I) diffusion models enables the generation of high-quality images from text descriptions. However, generating diverse customized images with reference visual attributes remains challenging. This work focuses on personalizing T2I diffusion models at a more abstract concept or category level, adapting commonalities from a set of reference images while creating new instances with sufficient variations. We introduce a solution that allows a pretrained T2I diffusion model to learn a set of soft prompts, enabling the generation of novel images by sampling prompts from the learned distribution. These prompts offer text-guided editing capabilities and additional flexibility in controlling variation and mixing between multiple distributions. We also show the adaptability of the learned prompt distribution to other tasks, such as text-to-3D. Finally we demonstrate effectiveness of our approach through quantitative analysis including automatic evaluation and human assessment. Project website: https://briannlongzhao.github.io/DreamDistribution

Jen-Hao Cheng, Vivian Wang, Huayu Wang et al.

Understanding causal event relationships and achieving fine-grained temporal grounding in videos remain challenging for vision-language models. Existing methods either compress video tokens to reduce temporal resolution, or treat videos as unsegmented streams, which obscures fine-grained event boundaries and limits the modeling of causal dependencies. We propose TEMPURA (Temporal Event Masked Prediction and Understanding for Reasoning in Action), a two-stage training framework that enhances video temporal understanding. TEMPURA first applies masked event prediction reasoning to reconstruct missing events and generate step-by-step causal explanations from dense event annotations, drawing inspiration from effective infilling techniques. TEMPURA then learns to perform video segmentation and dense captioning to decompose videos into non-overlapping events with detailed, timestamp-aligned descriptions. We train TEMPURA on VER, a large-scale dataset curated by us that comprises 1M training instances and 500K videos with temporally aligned event descriptions and structured reasoning steps. Experiments on temporal grounding and highlight detection benchmarks demonstrate that TEMPURA outperforms strong baseline models, confirming that integrating causal reasoning with fine-grained temporal segmentation leads to improved video understanding.

Mazda Moayeri, Vidhisha Balachandran, Varun Chandrasekaran et al. · cmu

With models getting stronger, evaluations have grown more complex, testing multiple skills in one benchmark and even in the same instance at once. However, skill-wise performance is obscured when inspecting aggregate accuracy, under-utilizing the rich signal modern benchmarks contain. We propose an automatic approach to recover the underlying skills relevant for any evaluation instance, by way of inspecting model-generated rationales. After validating the relevance of rationale-parsed skills and inferring skills for $46$k instances over $12$ benchmarks, we observe many skills to be common across benchmarks, resulting in the curation of hundreds of skill-slices (i.e. sets of instances testing a common skill). Inspecting accuracy over these slices yields novel insights on model trade-offs: e.g., compared to GPT-4o and Claude 3.5 Sonnet, on average, Gemini 1.5 Pro is $18\%$ more accurate in "computing molar mass", but $19\%$ less accurate in "applying constitutional law", despite the overall accuracies of the three models differing by a mere $0.4\%$. Furthermore, we demonstrate the practical utility of our approach by showing that insights derived from skill slice analysis can generalize to held-out instances: when routing each instance to the model strongest on the relevant skills, we see a $3\%$ accuracy improvement over our $12$ dataset corpus. Our skill-slices and framework open a new avenue in model evaluation, leveraging skill-specific analyses to unlock a more granular and actionable understanding of model capabilities.

Mohammad Abdul Hafeez Khan, Yash Jain, Siddhartha Bhattacharyya et al.

Text-to-image (T2I) generation models have made significant strides but still struggle with prompt sensitivity: even minor changes in prompt wording can yield inconsistent or inaccurate outputs. To address this challenge, we introduce a closed-loop, test-time prompt refinement framework that requires no additional training of the underlying T2I model, termed TIR. In our approach, each generation step is followed by a refinement step, where a pretrained multimodal large language model (MLLM) analyzes the output image and the user's prompt. The MLLM detects misalignments (e.g., missing objects, incorrect attributes) and produces a refined and physically grounded prompt for the next round of image generation. By iteratively refining the prompt and verifying alignment between the prompt and the image, TIR corrects errors, mirroring the iterative refinement process of human artists. We demonstrate that this closed-loop strategy improves alignment and visual coherence across multiple benchmark datasets, all while maintaining plug-and-play integration with black-box T2I models.

Akash Kumar, Ashlesha Kumar, Vibhav Vineet et al.

Self-supervised learning has emerged as a powerful paradigm for label-free model pretraining, particularly in the video domain, where manual annotation is costly and time-intensive. However, existing self-supervised approaches employ diverse experimental setups, making direct comparisons challenging due to the absence of a standardized benchmark. In this work, we establish a unified benchmark that enables fair comparisons across different methods. Additionally, we systematically investigate five critical aspects of self-supervised learning in videos: (1) dataset size, (2) model complexity, (3) data distribution, (4) data noise, and (5) feature representations. To facilitate this study, we evaluate six self-supervised learning methods across six network architectures, conducting extensive experiments on five benchmark datasets and assessing performance on two distinct downstream tasks. Our analysis reveals key insights into the interplay between pretraining strategies, dataset characteristics, pretext tasks, and model architectures. Furthermore, we extend these findings to Video Foundation Models (ViFMs), demonstrating their relevance in large-scale video representation learning. Finally, leveraging these insights, we propose a novel approach that significantly reduces training data requirements while surpassing state-of-the-art methods that rely on 10% more pretraining data. We believe this work will guide future research toward a deeper understanding of self-supervised video representation learning and its broader implications.

Shresth Grover, Vibhav Vineet, Yogesh S Rawat

In this work we present a novel task of understanding unintentional human activities in videos. We formalize this problem as a reasoning task under zero-shot scenario, where given a video of an unintentional activity we want to know why it transitioned from intentional to unintentional. We first evaluate the effectiveness of current state-of-the-art Large Multimodal Models on this reasoning task and observe that they suffer from hallucination. We further propose a novel prompting technique,termed as Dream of Thoughts (DoT), which allows the model to navigate through hallucinated thoughts to achieve better reasoning. To evaluate the performance on this task, we also introduce three different specialized metrics designed to quantify the models reasoning capability. We perform our experiments on two different datasets, OOPs and UCF-Crimes, and our findings show that DOT prompting technique is able to outperform standard prompting, while minimizing hallucinations.

Jiwan Chung, Neel Joshi, Pratyusha Sharma et al.

Multimodal reasoning models have recently shown promise on challenging domains such as olympiad-level geometry, yet their evaluation remains dominated by aggregate accuracy, a single score that obscures where and how models are improving. We introduce MathLens, a benchmark designed to disentangle the subskills of multimodal reasoning while preserving the complexity of textbook-style geometry problems. The benchmark separates performance into three components: Perception: extracting information from raw inputs, Reasoning: operating on available information, and Integration: selecting relevant perceptual evidence and applying it within reasoning. To support each test, we provide annotations: visual diagrams, textual descriptions to evaluate reasoning in isolation, controlled questions that require both modalities, and probes for fine-grained perceptual skills, all derived from symbolic specifications of the problems to ensure consistency and robustness. Our analysis reveals that different training approaches have uneven effects: First, reinforcement learning chiefly strengthens perception, especially when supported by textual supervision, while textual SFT indirectly improves perception through reflective reasoning. Second, reasoning improves only in tandem with perception. Third, integration remains the weakest capacity, with residual errors concentrated there once other skills advance. Finally, robustness diverges: RL improves consistency under diagram variation, whereas multimodal SFT reduces it through overfitting. We will release all data and experimental logs.

Sahithya Ravi, Gabriel Sarch, Vibhav Vineet et al.

An embodied AI assistant operating on egocentric video must integrate spatial cues across time - for instance, determining where an object A, glimpsed a few moments ago lies relative to an object B encountered later. We introduce Disjoint-3DQA , a generative QA benchmark that evaluates this ability of VLMs by posing questions about object pairs that are not co-visible in the same frame. We evaluated seven state-of-the-art VLMs and found that models lag behind human performance by 28%, with steeper declines in accuracy (60% to 30 %) as the temporal gap widens. Our analysis further reveals that providing trajectories or bird's-eye-view projections to VLMs results in only marginal improvements, whereas providing oracle 3D coordinates leads to a substantial 20% performance increase. This highlights a core bottleneck of multi-frame VLMs in constructing and maintaining 3D scene representations over time from visual signals. Disjoint-3DQA therefore sets a clear, measurable challenge for long-horizon spatial reasoning and aims to catalyze future research at the intersection of vision, language, and embodied AI.

Shresth Grover, Priyank Pathak, Akash Kumar et al.

Large-scale Vision-Language Models (VLMs) exhibit impressive complex reasoning capabilities but remain largely unexplored in visual sequential planning, i.e., executing multi-step actions towards a goal. Additionally, practical sequential planning often involves non-optimal (erroneous) steps, challenging VLMs to detect and correct such steps. We propose Corrective Sequential Planning Benchmark (CoSPlan) to evaluate VLMs in error-prone, vision-based sequential planning tasks across 4 domains: maze navigation, block rearrangement, image reconstruction,and object reorganization. CoSPlan assesses two key abilities: Error Detection (identifying non-optimal action) and Step Completion (correcting and completing action sequences to reach the goal). Despite using state-of-the-art reasoning techniques such as Chain-of-Thought and Scene Graphs, VLMs (e.g. Intern-VLM and Qwen2) struggle on CoSPlan, failing to leverage contextual cues to reach goals. Addressing this, we propose a novel training-free method, Scene Graph Incremental updates (SGI), which introduces intermediate reasoning steps between the initial and goal states. SGI helps VLMs reason about sequences, yielding an average performance gain of 5.2%. In addition to enhancing reliability in corrective sequential planning, SGI generalizes to traditional planning tasks such as Plan-Bench and VQA. Project Page : https://shroglck.github.io/cos_plan/

Ahmed Awadallah, Yash Lara, Raghav Magazine et al.

Progress in computer use agents (CUAs) has been constrained by the absence of large and high-quality datasets that capture how humans interact with a computer. While LLMs have thrived on abundant textual data, no comparable corpus exists for CUA trajectories. To address these gaps, we introduce FaraGen, a novel synthetic data generation system for multi-step web tasks. FaraGen can propose diverse tasks from frequently used websites, generate multiple solution attempts, and filter successful trajectories using multiple verifiers. It achieves high throughput, yield, and diversity for multi-step web tasks, producing verified trajectories at approximately $1 each. We use this data to train Fara-7B, a native CUA model that perceives the computer using only screenshots, executes actions via predicted coordinates, and is small enough to run on-device. We find that Fara-7B outperforms other CUA models of comparable size on benchmarks like WebVoyager, Online-Mind2Web, and WebTailBench -- our novel benchmark that better captures under-represented web tasks in pre-existing benchmarks. Furthermore, Fara-7B is competitive with much larger frontier models, illustrating key benefits of scalable data generation systems in advancing small efficient agentic models. We are making Fara-7B open-weight on Microsoft Foundry and HuggingFace, and we are releasing WebTailBench.

Ieva Bagdonaviciute, Vibhav Vineet

While recent Vision-Language Models (VLMs) have achieved impressive progress, it remains difficult to determine why they succeed or fail on complex reasoning tasks. Traditional benchmarks evaluate what models can answer correctly, not why they succeed or fail. In this work, we perform a failure-mode analysis of six frontier VLMs on three physics-based benchmarks - Physion, Physion++, and CLEVRER - by introducing custom subtests (for Physion and Physion++) and an integration of existing benchmark categories (for CLEVRER) to factor benchmark performance into distinct, testable capabilities. These subtests isolate perception (object, color, and occlusion recognition) and physics understanding (motion prediction and spatial reasoning), enabling us to test whether models attend to the correct entities and dynamics underlying their answers. Counterintuitively, subtest mastery correlates only weakly with benchmark accuracy: models often answer correctly without grounding in perception or physics. This suggests that current VLMs sometimes achieve benchmark scores for the wrong reasons, underscoring the need for diagnostics that expose hidden failure modes beyond aggregate metrics.

Erfan Shayegani, Keegan Hines, Yue Dong et al. · cmu, microsoft-research

Computer-Use Agents (CUAs) are an increasingly deployed class of agents that take actions on GUIs to accomplish user goals. In this paper, we show that CUAs consistently exhibit Blind Goal-Directedness (BGD): a bias to pursue goals regardless of feasibility, safety, reliability, or context. We characterize three prevalent patterns of BGD: (i) lack of contextual reasoning, (ii) assumptions and decisions under ambiguity, and (iii) contradictory or infeasible goals. We develop BLIND-ACT, a benchmark of 90 tasks capturing these three patterns. Built on OSWorld, BLIND-ACT provides realistic environments and employs LLM-based judges to evaluate agent behavior, achieving 93.75% agreement with human annotations. We use BLIND-ACT to evaluate nine frontier models, including Claude Sonnet and Opus 4, Computer-Use-Preview, and GPT-5, observing high average BGD rates (80.8%) across them. We show that BGD exposes subtle risks that arise even when inputs are not directly harmful. While prompting-based interventions lower BGD levels, substantial risk persists, highlighting the need for stronger training- or inference-time interventions. Qualitative analysis reveals observed failure modes: execution-first bias (focusing on how to act over whether to act), thought-action disconnect (execution diverging from reasoning), and request-primacy (justifying actions due to user request). Identifying BGD and introducing BLIND-ACT establishes a foundation for future research on studying and mitigating this fundamental risk and ensuring safe CUA deployment.

Aayush Rana, Akash Kumar, Vibhav Vineet et al.

Video action detection requires dense spatio-temporal annotations, which are both challenging and expensive to obtain. However, real-world videos often vary in difficulty and may not require the same level of annotation. This paper analyzes the appropriate annotation types for each sample and their impact on spatio-temporal video action detection. It focuses on two key aspects: 1) how to obtain varying levels of annotation for videos, and 2) how to learn action detection from different annotation types. The study explores video-level tags, points, scribbles, bounding boxes, and pixel-level masks. First, a simple active learning strategy is proposed to estimate the necessary annotation type for each video. Then, a novel spatio-temporal 3D-superpixel approach is introduced to generate pseudo-labels from these annotations, enabling effective training. The approach is validated on UCF101-24 and JHMDB-21 datasets, significantly cutting annotation costs with minimal performance loss.

Gabriel Sarch, Balasaravanan Thoravi Kumaravel, Sahithya Ravi et al.

A person's demonstration often serves as a key reference for others learning the same task. However, RGB video, the dominant medium for representing these demonstrations, often fails to capture fine-grained contextual cues such as intent, safety-critical environmental factors, and subtle preferences embedded in human behavior. This sensory gap fundamentally limits the ability of Vision Language Models (VLMs) to reason about why actions occur and how they should adapt to individual users. To address this, we introduce MICA (Multimodal Interactive Contextualized Assistance), a framework that improves conversational agents for task assistance by integrating eye gaze and speech cues. MICA segments demonstrations into meaningful sub-tasks and extracts keyframes and captions that capture fine-grained intent and user-specific cues, enabling richer contextual grounding for visual question answering. Evaluations on questions derived from real-time chat-assisted task replication show that multimodal cues significantly improve response quality over frame-based retrieval. Notably, gaze cues alone achieves 93% of speech performance, and their combination yields the highest accuracy. Task type determines the effectiveness of implicit (gaze) vs. explicit (speech) cues, underscoring the need for adaptable multimodal models. These results highlight the limitations of frame-based context and demonstrate the value of multimodal signals for real-world AI task assistance.

Jiayu Wang, Yifei Ming, Zhenmei Shi et al.

Large language models (LLMs) and vision-language models (VLMs) have demonstrated remarkable performance across a wide range of tasks and domains. Despite this promise, spatial understanding and reasoning -- a fundamental component of human cognition -- remains under-explored. We propose SpatialEval, a novel benchmark that covers diverse aspects of spatial reasoning such as relationship understanding, navigation, and counting. We conduct a comprehensive evaluation of competitive language and vision-language models. Our findings reveal several counter-intuitive insights that have been overlooked in the literature: (1) Spatial reasoning poses significant challenges where competitive models can fall behind random guessing; (2) Despite additional visual input, VLMs often under-perform compared to their LLM counterparts; (3) When both textual and visual information is available, multi-modal language models become less reliant on visual information if sufficient textual clues are provided. Additionally, we demonstrate that leveraging redundancy between vision and text can significantly enhance model performance. We hope our study will inform the development of multimodal models to improve spatial intelligence and further close the gap with human intelligence.

Joykirat Singh, Akshay Nambi, Vibhav Vineet

Large Language Models (LLMs) have been applied to Math Word Problems (MWPs) with transformative impacts, revolutionizing how these complex problems are approached and solved in various domains including educational settings. However, the evaluation of these models often prioritizes final accuracy, overlooking the crucial aspect of reasoning capabilities. This work addresses this gap by focusing on the ability of LLMs to detect and correct reasoning mistakes. We introduce a novel dataset MWP-MISTAKE, incorporating MWPs with both correct and incorrect reasoning steps generated through rule-based methods and smaller language models. Our comprehensive benchmarking reveals significant insights into the strengths and weaknesses of state-of-the-art models, such as GPT-4o, GPT-4, GPT-3.5Turbo, and others. We highlight GPT-$o's superior performance in mistake detection and rectification and the persistent challenges faced by smaller models. Additionally, we identify issues related to data contamination and memorization, impacting the reliability of LLMs in real-world applications. Our findings emphasize the importance of rigorous evaluation of reasoning processes and propose future directions to enhance the generalization and robustness of LLMs in mathematical problem-solving.

Tianjun Zhang, Yi Zhang, Vibhav Vineet et al.

Current text-to-image generation models often struggle to follow textual instructions, especially the ones requiring spatial reasoning. On the other hand, Large Language Models (LLMs), such as GPT-4, have shown remarkable precision in generating code snippets for sketching out text inputs graphically, e.g., via TikZ. In this work, we introduce Control-GPT to guide the diffusion-based text-to-image pipelines with programmatic sketches generated by GPT-4, enhancing their abilities for instruction following. Control-GPT works by querying GPT-4 to write TikZ code, and the generated sketches are used as references alongside the text instructions for diffusion models (e.g., ControlNet) to generate photo-realistic images. One major challenge to training our pipeline is the lack of a dataset containing aligned text, images, and sketches. We address the issue by converting instance masks in existing datasets into polygons to mimic the sketches used at test time. As a result, Control-GPT greatly boosts the controllability of image generation. It establishes a new state-of-art on the spatial arrangement and object positioning generation and enhances users' control of object positions, sizes, etc., nearly doubling the accuracy of prior models. Our work, as a first attempt, shows the potential for employing LLMs to enhance the performance in computer vision tasks.

Benno Krojer, Vaibhav Adlakha, Vibhav Vineet et al.

The ability to integrate context, including perceptual and temporal cues, plays a pivotal role in grounding the meaning of a linguistic utterance. In order to measure to what extent current vision-and-language models master this ability, we devise a new multimodal challenge, Image Retrieval from Contextual Descriptions (ImageCoDe). In particular, models are tasked with retrieving the correct image from a set of 10 minimally contrastive candidates based on a contextual description. As such, each description contains only the details that help distinguish between images. Because of this, descriptions tend to be complex in terms of syntax and discourse and require drawing pragmatic inferences. Images are sourced from both static pictures and video frames. We benchmark several state-of-the-art models, including both cross-encoders such as ViLBERT and bi-encoders such as CLIP, on ImageCoDe. Our results reveal that these models dramatically lag behind human performance: the best variant achieves an accuracy of 20.9 on video frames and 59.4 on static pictures, compared with 90.8 in humans. Furthermore, we experiment with new model variants that are better equipped to incorporate visual and temporal context into their representations, which achieve modest gains. Our hope is that ImageCoDE will foster progress in grounded language understanding by encouraging models to focus on fine-grained visual differences.