Follow-Your-Preference++: Rethinking Preference Alignment for Image Inpainting

We study preference alignment for image inpainting. Rather than proposing yet another method, we revisit the problem from first principles and reassess its core challenges. We adopt the widely used direct preference optimization framework and construct preference training data with publicly available reward models. Our empirical study spans nine reward models, two benchmarks, and two baseline inpainting models that differ in architecture and generative mechanism. Our main findings are: (1) Most reward models provide valid signals for preference data construction, although some are unreliable as evaluators. (2) Across models and benchmarks, preference data exhibits consistent trends under both candidate and sample scaling. (3) Reward models display pronounced biases--particularly in brightness, composition, and color scheme--that make them prone to inducing reward hacking. (4) A simple ensemble of reward models mitigates such biases and yields robust, generalizable performance. {\color{rebuttal_blue}(5) Preference alignment is transferable to the object removal task, where the goal shifts from open-ended creative generation to coherent background completion. (6) Further analysis reveals that a calibrated ensemble method further mitigates hacking and improves robustness.} Without modifying model architectures or introducing additional datasets, our models substantially outperform prior state-of-the-art models on standard metrics, large vision-language model evaluations, and human assessments. Our code is available at: https://github.com/shenytzzz/Follow-Your-Preference.

MangaFlow: An End-to-End Agentic Framework for Controllable Story to Manga Generation

End-to-end manga generation is a structured visual storytelling task that requires story decomposition, recurring character and scene grounding, page layout design, panel rendering, page composition, and lettering. However, existing generative models often perform direct page synthesis, entangling these factors in a single visual output and limiting precise control over layout geometry, visual references, and cross-panel consistency. To address these limitations, we propose MangaFlow, an agentic framework for controllable long-form manga generation that decomposes manga creation into planning, grounding, layout construction, reference-conditioned rendering, composition, and text placement. By treating layout and visual references as explicit intermediate variables, MangaFlow enables both simple text-to-manga generation and more precise user-controlled manga creation. This design exposes layout, visual assets, and lettering as editable intermediate controls for refining panel geometry, references, and text placement. To support long-form consistency, MangaFlow introduces a story section memory that links section descriptions with corresponding character, scene, and object references for reuse across panels. We further present a meta-benchmark for evaluating layout controllability, visual consistency, and generation quality. Experiments show that MangaFlow improves layout adherence and cross-panel consistency over direct generation baselines while supporting flexible human control.

Refining Multidimensional Video Reward Models via Disentangled Influence Functions

As Text-to-Video (T2V) generation models continue to evolve, the complexity of video evaluation necessitates a fine-grained assessment across various axes. To address this, recent works have focused on developing Multidimensional Video Reward Models (MVRMs), which decompose the evaluation process to better align with the multifaceted nature of human visual perception. However, training effective MVRMs is fundamentally challenged by the complex nature of video data. In this work, we identify a critical phenomenon termed Dimensional Heterogeneity: the reliability of a training sample can vary substantially across evaluation dimensions, meaning that a sample may provide reliable supervision for one objective while inducing high supervision risk for another. Consequently, prevailing data-centric methods that filter based on global scalar metrics are ill-posed for T2V tasks. To address this, we propose a disentangled influence framework that that efficiently estimates dimension-specific supervision risk. Leveraging this framework, we introduce two dimension-disentangled refinement strategies: Dimension-Disentangled Pruning, which removes extreme high-risk samples, and Dimension-Disentangled Reweighting, which softly down-weights high-risk supervision. Extensive experiments demonstrate that our disentangled strategies significantly outperform global filtering baselines, yielding reward models with superior alignment to ground truth.

A Comprehensive Information-Decomposition Analysis of Large Vision-Language Models

Large vision-language models (LVLMs) achieve impressive performance, yet their internal decision-making processes remain opaque, making it difficult to determine if the success stems from true multimodal fusion or from reliance on unimodal priors. To address this attribution gap, we introduce a novel framework using partial information decomposition (PID) to quantitatively measure the "information spectrum" of LVLMs -- decomposing a model's decision-relevant information into redundant, unique, and synergistic components. By adapting a scalable estimator to modern LVLM outputs, our model-agnostic pipeline profiles 26 LVLMs on four datasets across three dimensions -- breadth (cross-model & cross-task), depth (layer-wise information dynamics), and time (learning dynamics across training). Our analysis reveals two key results: (i) two task regimes (synergy-driven vs. knowledge-driven) and (ii) two stable, contrasting family-level strategies (fusion-centric vs. language-centric). We also uncover a consistent three-phase pattern in layer-wise processing and identify visual instruction tuning as the key stage where fusion is learned. Together, these contributions provide a quantitative lens beyond accuracy-only evaluation and offer insights for analyzing and designing the next generation of LVLMs. Code and data are available at https://github.com/RiiShin/pid-lvlm-analysis .

Post Persona Alignment for Multi-Session Dialogue Generation

Multi-session persona-based dialogue generation presents challenges in maintaining long-term consistency and generating diverse, personalized responses. While large language models (LLMs) excel in single-session dialogues, they struggle to preserve persona fidelity and conversational coherence across extended interactions. Existing methods typically retrieve persona information before response generation, which can constrain diversity and result in generic outputs. We propose Post Persona Alignment (PPA), a novel two-stage framework that reverses this process. PPA first generates a general response based solely on dialogue context, then retrieves relevant persona memories using the response as a query, and finally refines the response to align with the speaker's persona. This post-hoc alignment strategy promotes naturalness and diversity while preserving consistency and personalization. Experiments on multi-session LLM-generated dialogue data demonstrate that PPA significantly outperforms prior approaches in consistency, diversity, and persona relevance, offering a more flexible and effective paradigm for long-term personalized dialogue generation.

Exploring and Controlling Diversity in LLM-Agent Conversation

Diversity is a critical aspect of multi-agent communication. In this paper, we focus on controlling and exploring diversity in the context of open-domain multi-agent conversations, particularly for world simulation applications. We propose Adaptive Prompt Pruning (APP), a novel method that dynamically adjusts the content of the utterance generation prompt to control diversity using a single parameter, lambda. Through extensive experiments, we show that APP effectively controls the output diversity across models and datasets, with pruning more information leading to more diverse output. We comprehensively analyze the relationship between prompt content and conversational diversity. Our findings reveal that information from all components of the prompt generally constrains the diversity of the output, with the Memory block exerting the most significant influence. APP is compatible with established techniques like temperature sampling and top-p sampling, providing a versatile tool for diversity management. To address the trade-offs of increased diversity, such as inconsistencies with omitted information, we incorporate a post-generation correction step, which effectively balances diversity enhancement with output consistency. Additionally, we examine how prompt structure, including component order and length, impacts diversity. This study addresses key questions surrounding diversity in multi-agent world simulation, offering insights into its control, influencing factors, and associated trade-offs. Our contributions lay the foundation for systematically engineering diversity in LLM-based multi-agent collaborations, advancing their effectiveness in real-world applications.

NEMO: Can Multimodal LLMs Identify Attribute-Modified Objects?

Multimodal Large Language Models (MLLMs) have made notable advances in visual understanding, yet their abilities to recognize objects modified by specific attributes remain an open question. To address this, we explore MLLMs' reasoning capabilities in object recognition, ranging from commonsense to beyond-commonsense scenarios. We introduce a novel benchmark, NEMO, which comprises 900 images of origiNal fruits and their corresponding attributE-MOdified ones; along with a set of 2,700 questions including open-, multiple-choice-, unsolvable types. We assess 26 recent open-sourced and commercial models using our benchmark. The findings highlight pronounced performance gaps in recognizing objects in NEMO and reveal distinct answer preferences across different models. Although stronger vision encoders improve performance, MLLMs still lag behind standalone vision encoders. Interestingly, scaling up the model size does not consistently yield better outcomes, as deeper analysis reveals that larger LLMs can weaken vision encoders during fine-tuning. These insights shed light on critical limitations in current MLLMs and suggest potential pathways toward developing more versatile and resilient multimodal models.

BrainCodec: Neural fMRI codec for the decoding of cognitive brain states

Recently, leveraging big data in deep learning has led to significant performance improvements, as confirmed in applications like mental state decoding using fMRI data. However, fMRI datasets remain relatively small in scale, and the inherent issue of low signal-to-noise ratios (SNR) in fMRI data further exacerbates these challenges. To address this, we apply compression techniques as a preprocessing step for fMRI data. We propose BrainCodec, a novel fMRI codec inspired by the neural audio codec. We evaluated BrainCodec's compression capability in mental state decoding, demonstrating further improvements over previous methods. Furthermore, we analyzed the latent representations obtained through BrainCodec, elucidating the similarities and differences between task and resting state fMRI, highlighting the interpretability of BrainCodec. Additionally, we demonstrated that fMRI reconstructions using BrainCodec can enhance the visibility of brain activity by achieving higher SNR, suggesting its potential as a novel denoising method. Our study shows that BrainCodec not only enhances performance over previous methods but also offers new analytical possibilities for neuroscience. Our codes, dataset, and model weights are available at https://github.com/amano-k-lab/BrainCodec.

HALL-E: Hierarchical Neural Codec Language Model for Minute-Long Zero-Shot Text-to-Speech Synthesis

Recently, Text-to-speech (TTS) models based on large language models (LLMs) that translate natural language text into sequences of discrete audio tokens have gained great research attention, with advances in neural audio codec (NAC) models using residual vector quantization (RVQ). However, long-form speech synthesis remains a significant challenge due to the high frame rate, which increases the length of audio tokens and makes it difficult for autoregressive language models to generate audio tokens for even a minute of speech. To address this challenge, this paper introduces two novel post-training approaches: 1) Multi-Resolution Requantization (MReQ) and 2) HALL-E. MReQ is a framework to reduce the frame rate of pre-trained NAC models. Specifically, it incorporates multi-resolution residual vector quantization (MRVQ) module that hierarchically reorganizes discrete audio tokens through teacher-student distillation. HALL-E is an LLM-based TTS model designed to predict hierarchical tokens of MReQ. Specifically, it incorporates the technique of using MRVQ sub-modules and continues training from a pre-trained LLM-based TTS model. Furthermore, to promote TTS research, we create MinutesSpeech, a new benchmark dataset consisting of 40k hours of filtered speech data for training and evaluating speech synthesis ranging from 3s up to 180s. In experiments, we demonstrated the effectiveness of our approaches by applying our post-training framework to VALL-E. We achieved the frame rate down to as low as 8 Hz, enabling the stable minitue-long speech synthesis in a single inference step. Audio samples, dataset, codes and pre-trained models are available at https://yutonishimura-v2.github.io/HALL-E_DEMO/.

Harnessing the Latent Diffusion Model for Training-Free Image Style Transfer

Diffusion models have recently shown the ability to generate high-quality images. However, controlling its generation process still poses challenges. The image style transfer task is one of those challenges that transfers the visual attributes of a style image to another content image. Typical obstacle of this task is the requirement of additional training of a pre-trained model. We propose a training-free style transfer algorithm, Style Tracking Reverse Diffusion Process (STRDP) for a pretrained Latent Diffusion Model (LDM). Our algorithm employs Adaptive Instance Normalization (AdaIN) function in a distinct manner during the reverse diffusion process of an LDM while tracking the encoding history of the style image. This algorithm enables style transfer in the latent space of LDM for reduced computational cost, and provides compatibility for various LDM models. Through a series of experiments and a user study, we show that our method can quickly transfer the style of an image without additional training. The speed, compatibility, and training-free aspect of our algorithm facilitates agile experiments with combinations of styles and LDMs for extensive application.