Multimodal Large Language Models (MLLMs) have demonstrated remarkable capabilities by integrating visual and textual inputs, yet modality alignment remains one of the most challenging aspects. Current MLLMs typically rely on simple adapter architectures and pretraining approaches to bridge vision encoders with large language models (LLM), guided by image-level supervision. We identify this paradigm often leads to suboptimal alignment between modalities, significantly constraining the LLM’s ability to properly interpret and reason with visual features particularly for smaller language models. To address this fundamental limitation, we propose Supervised Embedding Alignment (SEA), a token-level supervision alignment method that enables more precise visual-text alignment during pretraining. SEA introduces minimal computational overhead while preserving language capabilities and substantially improving cross-modal understanding. Our comprehensive analyses reveal critical insights into the adapter’s role in multimodal integration, and extensive experiments demonstrate that SEA consistently improves performance across various model sizes, with smaller models benefiting the most (average performance gain of 7.61% for Gemma-2B). This work establishes a foundation for developing more effective alignment strategies for future multimodal systems.

The proliferation of large language models (LLMs) in generating content raises concerns about text copyright. Watermarking methods, particularly logit-based approaches, embed imperceptible identifiers into text to address these challenges. However, the widespread usage of watermarking across diverse LLMs has led to an inevitable issue known as watermark collision during common tasks, such as paraphrasing or translation.In this paper, we introduce watermark collision as a novel and general philosophy for watermark attacks, aimed at enhancing attack performance on top of any other attacking methods. We also provide a comprehensive demonstration that watermark collision poses a threat to all logit-based watermark algorithms, impacting not only specific attack scenarios but also downstream applications.

Generative linguistic steganography attempts to hide secret messages into covertext. Previous studies have generally focused on the statistical differences between the covertext and stegotext, however, ill-formed stegotext can readily be identified by humans. In this paper, we propose a novel zero-shot approach based on in-context learning for linguistic steganography to achieve better perceptual and statistical imperceptibility. We also design several new metrics and reproducible language evaluations to measure the imperceptibility of the stegotext. Our experimental results indicate that our method produces 1.926× more innocent and intelligible stegotext than any other method.

Deep neural networks have made great success on video captioning in supervised learning setting. However, annotating videos with descriptions is very expensive and time-consuming. If the video captioning algorithm can benefit from a large number of unlabeled videos, the cost of annotation can be reduced. In the proposed study, we make the first attempt to train the video captioning model on labeled data and unlabeled data jointly, in a semi-supervised learning manner. For labeled data, we train them with the traditional cross-entropy loss. For unlabeled data, we leverage a self-critical policy gradient method with the difference between the scores obtained by Monte-Carlo sampling and greedy decoding as the reward function, while the scores are the negative K-L divergence between output distributions of original video data and augmented video data. The final loss is the weighted sum of losses obtained by labeled data and unlabeled data. Experiments conducted on VATEX, MSR-VTT and MSVD dataset demonstrate that the introduction of unlabeled data can improve the performance of the video captioning model. The proposed semi-supervised learning algorithm also outperforms several state-of-the-art semi-supervised learning approaches.

We introduce the URIEL knowledge base for massively multilingual NLP and the lang2vec utility, which provides information-rich vector identifications of languages drawn from typological, geographical, and phylogenetic databases and normalized to have straightforward and consistent formats, naming, and semantics. The goal of URIEL and lang2vec is to enable multilingual NLP, especially on less-resourced languages and make possible types of experiments (especially but not exclusively related to NLP tasks) that are otherwise difficult or impossible due to the sparsity and incommensurability of the data sources. lang2vec vectors have been shown to reduce perplexity in multilingual language modeling, when compared to one-hot language identification vectors.