Despite substantial efforts in safety alignment, recent research indicates that Large Language Models (LLMs) remain highly susceptible to jailbreak attacks.

We study the problem of computing an optimal large language model (LLM) policy for the constrained alignment problem, where the goal is to maximize a primary reward objective while satisfying constraints on secondary utilities. Despite the popularity of Lagrangian-based LLM policy search in constrained alignment, iterative primal-dual methods often fail to converge, and non-iterative dual-based methods do not achieve optimality in the LLM parameter space. To address these challenges, we employ Lagrangian duality to develop an iterative dual-based alignment method that alternates between updating the LLM policy via Lagrangian maximization and updating the dual variable via dual descent. In theory, we characterize the primal-dual gap between the primal value in the distribution space and the dual value in the LLM parameter space. We further quantify the optimality gap of the learned LLM policies at near-optimal dual variables with respect to both the objective and the constraint functions. These results prove that dual-based alignment methods can find an optimal constrained LLM policy, up to an LLM parametrization gap. We demonstrate the effectiveness and merits of our approach through extensive experiments conducted on the PKU-SafeRLHF and Anthropic HH-RLHF datasets.

Recently, slow-thinking systems like GPT-o1 and DeepSeek-R1 have demonstrated great potential in solving challenging problems through explicit reflection. They significantly outperform the best fast-thinking models, such as GPT-4o, on various math and science benchmarks. However, their multimodal reasoning capabilities remain on par with fast-thinking models. For instance, GPT-o1's performance on benchmarks like MathVista, MathVerse, and MathVision is similar to fast-thinking models. In this paper, we showcase how to enhance the slow-thinking capabilities of vision-language models using reinforcement learning, to advance the state of the art, without relying on costly distillation. First, we adapt the GRPO algorithm with a novel technique called Selective Sample Replay (SSR) to address the vanishing advantages problem.

Recent advancements in multimodal large language models (MLLMs) have aimed to integrate and interpret data across diverse modalities. However, the capacity of these models to concurrently process and reason about multiple modalities remains underexplored, partly due to the lack of comprehensive modality-wise benchmarks. We introduce OmniBench, a novel benchmark designed to rigorously evaluate models' ability to recognize, interpret, and reason across visual, acoustic, and textual inputs simultaneously. We define language models capable of such tri-modal processing as the omni-language models (OLMs). OmniBench is distinguished by high-quality human annotations, ensuring that accurate responses require integrated understanding and reasoning across all three modalities.

Due to the size and complexity of modern large language models (LLMs), it has proven challenging to uncover the underlying mechanisms that models use to solve reasoning problems. For instance, is their reasoning for a specific problem localized to certain parts of the network? Do they break down the reasoning problem into modular components that are then executed as sequential steps as we go deeper in the model? To better understand the reasoning capability of LLMs, we study a minimal propositional logic problem that requires combining multiple facts to arrive at a solution. By studying this problem on Mistral and Gemma models, up to 27B parameters, we illuminate the core components the models use to solve such logic problems. From a mechanistic interpretability point of view, we use causal mediation analysis to uncover the pathways and components of the LLMs' reasoning processes. Then, we offer fine-grained insights into the functions of attention heads in different layers. We not only find a sparse circuit that computes the answer, but we decompose it into sub-circuits that have four distinct and modular uses. Finally, we reveal that three distinct models - Mistral-7B, Gemma2-9B and Gemma-2-27B - contain analogous but not identical mechanisms.

Current camouflaged object detection methods predominantly follow discriminative segmentation paradigms and heavily rely on predefined categories present in the training data, limiting their generalization to unseen or emerging camouflage objects. This limitation is further compounded by the labor-intensive and time-consuming nature of collecting camouflage imagery. Although Large VisionLanguage Models (LVLMs) show potential to improve such issues with their powerful generative capabilities, their understanding of camouflage scenes is still insufficient. To bridge this gap, we introduce MMCSBench, the first comprehensive multimodal benchmark designed to evaluate and advance LVLM capabilities in camouflage scenes. MMCSBench comprises 22,537 images and 76,843 corresponding image-text pairs across five fine-grained camouflage tasks. Additionally, we propose a new task, Camouflage Efficacy Assessment (CEA), aimed at quantitatively evaluating the camouflage effectiveness of objects in images and enabling automated collection of camouflage images from large-scale databases. Extensive experiments on 26 LVLMs reveal significant shortcomings in models' ability to perceive and interpret camouflage scenes. These findings highlight the fundamental differences between natural and camouflaged visual inputs, offering insights for future research in advancing LVLM capabilities within this challenging domain.

The urban environment is characterized by complex spatio-temporal dynamics arising from diverse human activities and interactions. Effectively modeling these dynamics is essential for understanding and optimizing urban systems. In this work, we introduce UrbanDiT, a foundation model for open-world urban spatiotemporal learning that successfully scales up diffusion transformers in this field.

Supplementary Material for BLINK-Twice: You see, but do you observe? This supplementary material provides additional details omitted from the main paper due to space1 limitations. It includes a more comprehensive description of the dataset (Section A), covering2 data collection, comparisons with existing datasets, and additional visualizations. We also present3 extended experimental details (Section B), including the full list of evaluated models, the computation4 of evaluation metrics, analysis of multimodal reasoning paradigms, and more qualitative visual results.5 Finally, we discuss the limitations of our method (Section C).6 A.1 Data Collection8 Figure 3 illustrates our data collection pipeline.

Recently, Multimodal Large Language Models (MLLMs) have made rapid progress, particularly in enhancing their reasoning capabilities. However, existing reasoning benchmarks still primarily assess language-based reasoning, often treating visual input as replaceable context. To address this gap, we introduce BLINK-Twice, a vision-centric reasoning benchmark grounded in challenging perceptual tasks. Instead of relying on external knowledge, our tasks require models to reason from visual content alone, shifting the focus from language-based to image-grounded reasoning. Compared to prior perception benchmarks, it moves beyond shallow perception ("see") and requires fine-grained observation and analytical reasoning ("observe").

Modern tokenizers employ deterministic algorithms to map text into a single "canonical" token sequence, yet the same string can be encoded as many noncanonical tokenizations using the tokenizer vocabulary. In this work, we investigate the robustness of LMs to text encoded with non-canonical tokenizations entirely unseen during training. Surprisingly, when evaluated across 20 benchmarks, we find that instruction-tuned models retain up to 93.4% of their original performance when given a randomly sampled tokenization, and 90.8% with character-level tokenization. We see that overall stronger models tend to be more robust, and robustness diminishes as the tokenization departs farther from the canonical form. Motivated by these results, we then identify settings where non-canonical tokenization schemes can improve performance, finding that character-level segmentation improves string manipulation and code understanding tasks by up to +14%, and right-aligned digit grouping enhances large-number arithmetic by +33%. Finally, we investigate the source of this robustness, finding that it arises in the instructiontuning phase. We show that while both base and post-trained models grasp the semantics of non-canonical tokenizations (perceiving them as containing misspellings), base models try to mimic the imagined mistakes and degenerate into nonsensical output, while post-trained models are committed to fluent responses. Overall, our findings suggest that models are less tied to their tokenizer than previously believed, and demonstrate the promise of intervening on tokenization at inference time to boost performance.1