SB-Bench: Stereotype Bias Benchmark for Large Multimodal Models

Stereotype biases in Large Multimodal Models (LMMs) perpetuate harmful societal prejudices, undermining the fairness and equity of AI applications. As LMMs grow increasingly influential, addressing and mitigating inherent biases related to stereotypes, harmful generations, and ambiguous assumptions in real-world scenarios has become essential. However, existing datasets evaluating stereotype biases in LMMs often lack diversity and rely on synthetic images, leaving a gap in bias evaluation for real-world visual contexts. To address this, we introduce the Stereotype Bias Benchmark (SB-bench), the most comprehensive framework to date for assessing stereotype biases across nine diverse categories with non-synthetic images. SB-bench rigorously evaluates LMMs through carefully curated, visually grounded scenarios, challenging them to reason accurately about visual stereotypes. It offers a robust evaluation framework featuring real-world visual samples, image variations, and multiple-choice question formats. By introducing visually grounded queries that isolate visual biases from textual ones, SB-bench enables a precise and nuanced assessment of a model's reasoning capabilities across varying levels of difficulty. Through rigorous testing of state-of-the-art open-source and closed-source LMMs, SB-bench provides a systematic approach to assessing stereotype biases in LMMs across key social dimensions. This benchmark represents a significant step toward fostering fairness in AI systems and reducing harmful biases, laying the groundwork for more equitable and socially responsible LMMs. Our code and dataset are publicly available.

TimeLogic: A Temporal Logic Benchmark for Video QA

Temporal logical understanding, a core facet of human cognition, plays a pivotal role in capturing complex sequential events and their temporal relationships within videos. This capability is particularly crucial in tasks like Video Question Answering (VideoQA), where the goal is to process visual data over time together with textual data to provide coherent answers. However, current VideoQA benchmarks devote little focus to evaluating this critical skill due to the challenge of annotating temporal logic. Despite the advancement of vision-language models, assessing their temporal logical reasoning powers remains a challenge, primarily due to the lack QA pairs that demand formal, complex temporal reasoning. To bridge this gap, we introduce the TimeLogic QA (TLQA) framework to automatically generate the QA pairs, specifically designed to evaluate the temporal logical understanding. To this end, TLQA leverages temporal annotations from existing video datasets together with temporal operators derived from logic theory to construct questions that test understanding of event sequences and their temporal relationships. TLQA framework is generic and scalable, capable of leveraging both, existing video action datasets with temporal action segmentation annotations, or video datasets with temporal scene graph annotations, to automatically generate temporal logical questions. We leverage 4 datasets, STAR, Breakfast, AGQA, and CrossTask, and generate two VideoQA dataset variants - small (TLQA-S) and large (TLQA-L) - containing 2k and 10k QA pairs for each category, resulting in 32k and 160k total pairs per dataset. We undertake a comprehensive evaluation of leading-edge VideoQA models, employing the TLQA to benchmark their temporal logical understanding capabilities. We assess the VideoQA model's temporal reasoning performance on 16 categories of temporal logic with varying temporal complexity.

LlamaV-o1: Rethinking Step-by-step Visual Reasoning in LLMs

Reasoning is a fundamental capability for solving complex multi-step problems, particularly in visual contexts where sequential step-wise understanding is essential. Existing approaches lack a comprehensive framework for evaluating visual reasoning and do not emphasize step-wise problem-solving. To this end, we propose a comprehensive framework for advancing step-by-step visual reasoning in large language models (LMMs) through three key contributions. First, we introduce a visual reasoning benchmark specifically designed to evaluate multi-step reasoning tasks. The benchmark presents a diverse set of challenges with eight different categories ranging from complex visual perception to scientific reasoning with over 4k reasoning steps in total, enabling robust evaluation of LLMs' abilities to perform accurate and interpretable visual reasoning across multiple steps. Second, we propose a novel metric that assesses visual reasoning quality at the granularity of individual steps, emphasizing both correctness and logical coherence. The proposed metric offers deeper insights into reasoning performance compared to traditional end-task accuracy metrics. Third, we present a new multimodal visual reasoning model, named LlamaV-o1, trained using a multi-step curriculum learning approach, where tasks are progressively organized to facilitate incremental skill acquisition and problem-solving. The proposed LlamaV-o1 is designed for multi-step reasoning and learns step-by-step through a structured training paradigm. Extensive experiments show that our LlamaV-o1 outperforms existing open-source models and performs favorably against close-source proprietary models. Compared to the recent Llava-CoT, our LlamaV-o1 achieves an average score of 67.3 with an absolute gain of 3.8\% across six benchmarks while being 5 times faster during inference scaling. Our benchmark, model, and code are publicly available.

LIAR: Leveraging Alignment (Best-of-N) to Jailbreak LLMs in Seconds

Many existing jailbreak techniques rely on solving discrete combinatorial optimization, while more recent approaches involve training LLMs to generate multiple adversarial prompts. However, both approaches require significant computational resources to produce even a single adversarial prompt. We hypothesize that the inefficiency of current approaches stems from an inadequate characterization of the jailbreak problem. To address this gap, we formulate the jailbreak problem in terms of alignment. By starting from an available safety-aligned model, we leverage an unsafe reward to guide the safe model towards generating unsafe outputs using alignment techniques (e.g., reinforcement learning from human feedback), effectively performing jailbreaking via alignment. We propose a novel jailbreak method called LIAR (LeveragIng Alignment to jailbReak). To demonstrate the simplicity and effectiveness of our approach, we employ a best-of-N method to solve the alignment problem. LIAR offers significant advantages: lower computational requirements without additional training, fully black-box operation, competitive attack success rates, and more human-readable prompts. We provide theoretical insights into the possibility of jailbreaking a safety-aligned model, revealing inherent vulnerabilities in current alignment strategies for LLMs. We also provide sub-optimality guarantees for the proposed \algo. Experimentally, we achieve ASR comparable to the SoTA with a 10x improvement to perplexity and a Time-to-Attack measured in seconds rather than tens of hours.

Deepfake Media Generation and Detection in the Generative AI Era: A Survey and Outlook

With the recent advancements in generative modeling, the realism of deepfake content has been increasing at a steady pace, even reaching the point where people often fail to detect manipulated media content online, thus being deceived into various kinds of scams. In this paper, we survey deepfake generation and detection techniques, including the most recent developments in the field, such as diffusion models and Neural Radiance Fields. Our literature review covers all deepfake media types, comprising image, video, audio and multimodal (audio-visual) content. We identify various kinds of deepfakes, according to the procedure used to alter or generate the fake content. We further construct a taxonomy of deepfake generation and detection methods, illustrating the important groups of methods and the domains where these methods are applied. Next, we gather datasets used for deepfake detection and provide updated rankings of the best performing deepfake detectors on the most popular datasets. In addition, we develop a novel multimodal benchmark to evaluate deepfake detectors on out-of-distribution content. The results indicate that state-of-the-art detectors fail to generalize to deepfake content generated by unseen deepfake generators. Finally, we propose future directions to obtain robust and powerful deepfake detectors. Our project page and new benchmark are available at https://github.com/CroitoruAlin/biodeep.

All Languages Matter: Evaluating LMMs on Culturally Diverse 100 Languages

Existing Large Multimodal Models (LMMs) generally focus on only a few regions and languages. As LMMs continue to improve, it is increasingly important to ensure they understand cultural contexts, respect local sensitivities, and support low-resource languages, all while effectively integrating corresponding visual cues. In pursuit of culturally diverse global multimodal models, our proposed All Languages Matter Benchmark (ALM-bench) represents the largest and most comprehensive effort to date for evaluating LMMs across 100 languages. ALM-bench challenges existing models by testing their ability to understand and reason about culturally diverse images paired with text in various languages, including many low-resource languages traditionally underrepresented in LMM research. The benchmark offers a robust and nuanced evaluation framework featuring various question formats, including true/false, multiple choice, and open-ended questions, which are further divided into short and long-answer categories. ALM-bench design ensures a comprehensive assessment of a model's ability to handle varied levels of difficulty in visual and linguistic reasoning. To capture the rich tapestry of global cultures, ALM-bench carefully curates content from 13 distinct cultural aspects, ranging from traditions and rituals to famous personalities and celebrations. Through this, ALM-bench not only provides a rigorous testing ground for state-of-the-art open and closed-source LMMs but also highlights the importance of cultural and linguistic inclusivity, encouraging the development of models that can serve diverse global populations effectively. Our benchmark is publicly available.

DLCR: A Generative Data Expansion Framework via Diffusion for Clothes-Changing Person Re-ID

With the recent exhibited strength of generative diffusion models, an open research question is \textit{if images generated by these models can be used to learn better visual representations}. While this generative data expansion may suffice for easier visual tasks, we explore its efficacy on a more difficult discriminative task: clothes-changing person re-identification (CC-ReID). CC-ReID aims to match people appearing in non-overlapping cameras, even when they change their clothes across cameras. Not only are current CC-ReID models constrained by the limited diversity of clothing in current CC-ReID datasets, but generating additional data that retains important personal features for accurate identification is a current challenge. To address this issue we propose DLCR, a novel data expansion framework that leverages pre-trained diffusion and large language models (LLMs) to accurately generate diverse images of individuals in varied attire. We generate additional data for five benchmark CC-ReID datasets (PRCC, CCVID, LaST, VC-Clothes, and LTCC) and \textbf{increase their clothing diversity by \boldmath{$10$}x, totaling over \boldmath{$2.1$}M images generated}. DLCR employs diffusion-based text-guided inpainting, conditioned on clothing prompts constructed using LLMs, to generate synthetic data that only modifies a subject's clothes while preserving their personally identifiable features. With this massive increase in data, we introduce two novel strategies - progressive learning and test-time prediction refinement - that respectively reduce training time and further boosts CC-ReID performance. On the PRCC dataset, we obtain a large top-1 accuracy improvement of $11.3\%$ by training CAL, a previous state of the art (SOTA) method, with DLCR-generated data. We publicly release our code and generated data for each dataset here: \url{https://github.com/CroitoruAlin/dlcr}.

CityGuessr: City-Level Video Geo-Localization on a Global Scale

Video geolocalization is a crucial problem in current times. Given just a video, ascertaining where it was captured from can have a plethora of advantages. The problem of worldwide geolocalization has been tackled before, but only using the image modality. Its video counterpart remains relatively unexplored. Meanwhile, video geolocalization has also garnered some attention in the recent past, but the existing methods are all restricted to specific regions. This motivates us to explore the problem of video geolocalization at a global scale. Hence, we propose a novel problem of worldwide video geolocalization with the objective of hierarchically predicting the correct city, state/province, country, and continent, given a video. However, no large scale video datasets that have extensive worldwide coverage exist, to train models for solving this problem. To this end, we introduce a new dataset, CityGuessr68k comprising of 68,269 videos from 166 cities all over the world. We also propose a novel baseline approach to this problem, by designing a transformer-based architecture comprising of an elegant Self-Cross Attention module for incorporating scenes as well as a TextLabel Alignment strategy for distilling knowledge from textlabels in feature space. To further enhance our location prediction, we also utilize soft-scene labels. Finally we demonstrate the performance of our method on our new dataset as well as Mapillary(MSLS). Our code and datasets are available at: https://github.com/ParthPK/CityGuessr

Hierarchical Preference Optimization: Learning to achieve goals via feasible subgoals prediction

This work introduces Hierarchical Preference Optimization (HPO), a novel approach to hierarchical reinforcement learning (HRL) that addresses non-stationarity and infeasible subgoal generation issues when solving complex robotic control tasks. HPO leverages maximum entropy reinforcement learning combined with token-level Direct Preference Optimization (DPO), eliminating the need for pre-trained reference policies that are typically unavailable in challenging robotic scenarios. Mathematically, we formulate HRL as a bi-level optimization problem and transform it into a primitive-regularized DPO formulation, ensuring feasible subgoal generation and avoiding degenerate solutions. Extensive experiments on challenging robotic navigation and manipulation tasks demonstrate impressive performance of HPO, where it shows an improvement of up to 35% over the baselines. Furthermore, ablation studies validate our design choices, and quantitative analyses confirm the ability of HPO to mitigate non-stationarity and infeasible subgoal generation issues in HRL.

Investigating Memorization in Video Diffusion Models

Diffusion models, widely used for image and video generation, face a significant limitation: the risk of memorizing and reproducing training data during inference, potentially generating unauthorized copyrighted content. While prior research has focused on image diffusion models (IDMs), video diffusion models (VDMs) remain underexplored. To address this gap, we first formally define the two types of memorization in VDMs (content memorization and motion memorization) in a practical way that focuses on privacy preservation and applies to all generation types. We then introduce new metrics specifically designed to separately assess content and motion memorization in VDMs. Additionally, we curate a dataset of text prompts that are most prone to triggering memorization when used as conditioning in VDMs. By leveraging these prompts, we generate diverse videos from various open-source VDMs, successfully extracting numerous training videos from each tested model. Through the application of our proposed metrics, we systematically analyze memorization across various pretrained VDMs, including text-conditional and unconditional models, on a variety of datasets. Our comprehensive study reveals that memorization is widespread across all tested VDMs, indicating that VDMs can also memorize image training data in addition to video datasets. Finally, we propose efficient and effective detection strategies for both content and motion memorization, offering a foundational approach for improving privacy in VDMs.