BaseReward: A Strong Baseline for Multimodal Reward Model

The rapid advancement of Multimodal Large Language Models (MLLMs) has made aligning them with human preferences a critical challenge. Reward Models (RMs) are a core technology for achieving this goal, but a systematic guide for building state-of-the-art Multimodal Reward Models (MRMs) is currently lacking in both academia and industry. Through exhaustive experimental analysis, this paper aims to provide a clear ``recipe'' for constructing high-performance MRMs. We systematically investigate every crucial component in the MRM development pipeline, including \textit{reward modeling paradigms} (e.g., Naive-RM, Critic-based RM, and Generative RM), \textit{reward head architecture}, \textit{training strategies}, \textit{data curation} (covering over ten multimodal and text-only preference datasets), \textit{backbone model} and \textit{model scale}, and \textit{ensemble methods}. Based on these experimental insights, we introduce \textbf{BaseReward}, a powerful and efficient baseline for multimodal reward modeling. BaseReward adopts a simple yet effective architecture, built upon a {Qwen2.5-VL} backbone, featuring an optimized two-layer reward head, and is trained on a carefully curated mixture of high-quality multimodal and text-only preference data. Our results show that BaseReward establishes a new SOTA on major benchmarks such as MM-RLHF-Reward Bench, VL-Reward Bench, and Multimodal Reward Bench, outperforming previous models. Furthermore, to validate its practical utility beyond static benchmarks, we integrate BaseReward into a real-world reinforcement learning pipeline, successfully enhancing an MLLM's performance across various perception, reasoning, and conversational tasks. This work not only delivers a top-tier MRM but, more importantly, provides the community with a clear, empirically-backed guide for developing robust reward models for the next generation of MLLMs.

Aligning Audio Captions with Human Preferences

Current audio captioning systems rely heavily on supervised learning with paired audio-caption datasets, which are expensive to curate and may not reflect human preferences in real-world scenarios. To address this limitation, we propose a preference-aligned audio captioning framework based on Reinforcement Learning from Human Feedback (RLHF). To effectively capture nuanced human preferences, we train a Contrastive Language-Audio Pretraining (CLAP)-based reward model using human-labeled pairwise preference data. This reward model is integrated into a reinforcement learning framework to fine-tune any baseline captioning system without relying on ground-truth caption annotations. Extensive human evaluations across multiple datasets show that our method produces captions preferred over those from baseline models, particularly in cases where the baseline models fail to provide correct and natural captions. Furthermore, our framework achieves performance comparable to supervised approaches with ground-truth data, demonstrating its effectiveness in aligning audio captioning with human preferences and its scalability in real-world scenarios.

Towards personalized, precise and survey-free environment recognition: AI-enhanced sensor fusion without pre-deployment

Accurate and personalized environment recognition is essential for seamless indoor positioning and optimized connectivity, yet traditional fingerprinting requires costly site surveys and lacks user-level adaptation. We present a survey-free, on-device sensor-fusion framework that builds a personalized, lightweight multi-source fingerprint (FP) database from pedestrian dead reckoning (PDR), WiFi/cellular, GNSS, and interaction time tags. Matching is performed by an AI-enhanced dynamic time warping module (AIDTW) that aligns noisy, asynchronous sequences. To turn perception into continually improving actions, a cloud-edge online Reinforcement Learning from Human Feedback (RLHF) loop aggregates desensitized summaries and human feedback in the cloud to optimize a policy via proximal policy optimization (PPO), and periodically distills updates to devices. Across indoor/outdoor scenarios, our system reduces network-transition latency (measured by time-to-switch, TTS) by 32-65% in daily environments compared with conventional baselines, without site-specific pre-deployment.

The Anatomy of Alignment: Decomposing Preference Optimization by Steering Sparse Features

Aligning large language models is critical for their usability and safety. However, the prevailing approach of Reinforcement Learning from Human Feedback (RLHF) induces diffuse, opaque parameter changes, making it difficult to discern what the model has internalized. Hence, we introduce Feature Steering with Reinforcement Learning (FSRL), a transparent alignment framework that trains a lightweight adapter to steer behavior by modulating interpretable features from a Sparse Autoencoder (SAE). First, we demonstrate that FSRL is an effective method for preference optimization and is comparable with current RLHF methods. We then perform mechanistic analysis on the trained adapter, and find that its policy systematically promotes style features over explicit alignment concepts, suggesting that the preference optimization process rewards stylistic presentation as a proxy for quality. Ultimately, we hope that FSRL provides a tool for both interpretable model control and diagnosing the internal mechanisms of alignment.

Interpretability as Alignment: Making Internal Understanding a Design Principle

Large neural models are increasingly deployed in high-stakes settings, raising concerns about whether their behavior reliably aligns with human values. Interpretability provides a route to internal transparency by revealing the computations that drive outputs. We argue that interpretability especially mechanistic approaches should be treated as a design principle for alignment, not an auxiliary diagnostic tool. Post-hoc methods such as LIME or SHAP offer intuitive but correlational explanations, while mechanistic techniques like circuit tracing or activation patching yield causal insight into internal failures, including deceptive or misaligned reasoning that behavioral methods like RLHF, red teaming, or Constitutional AI may overlook. Despite these advantages, interpretability faces challenges of scalability, epistemic uncertainty, and mismatches between learned representations and human concepts. Our position is that progress on safe and trustworthy AI will depend on making interpretability a first-class objective of AI research and development, ensuring that systems are not only effective but also auditable, transparent, and aligned with human intent.

HumanAgencyBench: Scalable Evaluation of Human Agency Support in AI Assistants

As humans delegate more tasks and decisions to artificial intelligence (AI), we risk losing control of our individual and collective futures. Relatively simple algorithmic systems already steer human decision-making, such as social media feed algorithms that lead people to unintentionally and absent-mindedly scroll through engagement-optimized content. In this paper, we develop the idea of human agency by integrating philosophical and scientific theories of agency with AI-assisted evaluation methods: using large language models (LLMs) to simulate and validate user queries and to evaluate AI responses. We develop HumanAgencyBench (HAB), a scalable and adaptive benchmark with six dimensions of human agency based on typical AI use cases. HAB measures the tendency of an AI assistant or agent to Ask Clarifying Questions, Avoid Value Manipulation, Correct Misinformation, Defer Important Decisions, Encourage Learning, and Maintain Social Boundaries. We find low-to-moderate agency support in contemporary LLM-based assistants and substantial variation across system developers and dimensions. For example, while Anthropic LLMs most support human agency overall, they are the least supportive LLMs in terms of Avoid Value Manipulation. Agency support does not appear to consistently result from increasing LLM capabilities or instruction-following behavior (e.g., RLHF), and we encourage a shift towards more robust safety and alignment targets.

Improving LLM Safety and Helpfulness using SFT and DPO: A Study on OPT-350M

This research investigates the effectiveness of alignment techniques, Supervised Fine-Tuning (SFT), Direct Preference Optimization (DPO), and a combined SFT+DPO approach on improving the safety and helpfulness of the OPT-350M language model. Utilizing the Anthropic Helpful-Harmless RLHF dataset, we train and evaluate four models: the base OPT350M, an SFT model, a DPO model, and a model trained with both SFT and DPO. We introduce three key evaluation metrics: Harmlessness Rate (HmR), Helpfulness Rate (HpR), and a Combined Alignment Score (CAS), all derived from reward model outputs. The results show that while SFT outperforms DPO, The combined SFT+DPO model outperforms all others across all metrics, demonstrating the complementary nature of these techniques. Our findings also highlight challenges posed by noisy data, limited GPU resources, and training constraints. This study offers a comprehensive view of how fine-tuning strategies affect model alignment and provides a foundation for more robust alignment pipelines in future work.

RewardDance: Reward Scaling in Visual Generation

Reward Models (RMs) are critical for improving generation models via Reinforcement Learning (RL), yet the RM scaling paradigm in visual generation remains largely unexplored. It primarily due to fundamental limitations in existing approaches: CLIP-based RMs suffer from architectural and input modality constraints, while prevalent Bradley-Terry losses are fundamentally misaligned with the next-token prediction mechanism of Vision-Language Models (VLMs), hindering effective scaling. More critically, the RLHF optimization process is plagued by Reward Hacking issue, where models exploit flaws in the reward signal without improving true quality. To address these challenges, we introduce RewardDance, a scalable reward modeling framework that overcomes these barriers through a novel generative reward paradigm. By reformulating the reward score as the model's probability of predicting a "yes" token, indicating that the generated image outperforms a reference image according to specific criteria, RewardDance intrinsically aligns reward objectives with VLM architectures. This alignment unlocks scaling across two dimensions: (1) Model Scaling: Systematic scaling of RMs up to 26 billion parameters; (2) Context Scaling: Integration of task-specific instructions, reference examples, and chain-of-thought (CoT) reasoning. Extensive experiments demonstrate that RewardDance significantly surpasses state-of-the-art methods in text-to-image, text-to-video, and image-to-video generation. Crucially, we resolve the persistent challenge of "reward hacking": Our large-scale RMs exhibit and maintain high reward variance during RL fine-tuning, proving their resistance to hacking and ability to produce diverse, high-quality outputs. It greatly relieves the mode collapse problem that plagues smaller models.

The Personality Illusion: Revealing Dissociation Between Self-Reports & Behavior in LLMs

Personality traits have long been studied as predictors of human behavior.Recent advances in Large Language Models (LLMs) suggest similar patterns may emerge in artificial systems, with advanced LLMs displaying consistent behavioral tendencies resembling human traits like agreeableness and self-regulation. Understanding these patterns is crucial, yet prior work primarily relied on simplified self-reports and heuristic prompting, with little behavioral validation. In this study, we systematically characterize LLM personality across three dimensions: (1) the dynamic emergence and evolution of trait profiles throughout training stages; (2) the predictive validity of self-reported traits in behavioral tasks; and (3) the impact of targeted interventions, such as persona injection, on both self-reports and behavior. Our findings reveal that instructional alignment (e.g., RLHF, instruction tuning) significantly stabilizes trait expression and strengthens trait correlations in ways that mirror human data. However, these self-reported traits do not reliably predict behavior, and observed associations often diverge from human patterns. While persona injection successfully steers self-reports in the intended direction, it exerts little or inconsistent effect on actual behavior. By distinguishing surface-level trait expression from behavioral consistency, our findings challenge assumptions about LLM personality and underscore the need for deeper evaluation in alignment and interpretability.

SharedRep-RLHF: A Shared Representation Approach to RLHF with Diverse Preferences

Uniform-reward reinforcement learning from human feedback (RLHF), which trains a single reward model to represent the preferences of all annotators, fails to capture the diversity of opinions across sub-populations, inadvertently favoring dominant groups. The state-of-the-art, MaxMin-RLHF, addresses this by learning group-specific reward models, and by optimizing for the group receiving the minimum reward, thereby promoting fairness. However, we identify that a key limitation of MaxMin-RLHF is its poor performance when the minimum-reward group is a minority. To mitigate this drawback, we introduce a novel framework, termed {\em SharedRep-RLHF}. At its core, SharedRep-RLHF learns and leverages {\em shared traits} in annotations among various groups, in contrast to learning separate reward models across groups. We first show that MaxMin-RLHF is provably suboptimal in learning shared traits, and then quantify the sample complexity of SharedRep-RLHF. Experiments across diverse natural language tasks showcase the effectiveness of SharedRep-RLHF compared to MaxMin-RLHF with a gain of up to 20% in win rate.