The paper investigates how to make Mixture-of-Experts (MoE) reward models more interpretable and useful for personalization by focusing on sparsity and diversity regularizers. By keeping the architecture and training data constant, the authors show that conventional MoE reward models learn entangled components that hinder both interpretability and personalization. Introducing a sparse MoE objective reduces entanglement, yielding experts that align with semantic categories and functional strengths, and enables composable, few‑shot personalization with a 25.81‑point improvement over entangled models.
arXiv:2606. 04284v1 Announce Type: cross Abstract: Preference modeling plays a central role in reinforcement learning from human feedback (RLHF), enabling large language models (LLMs) to align with human values.
arXiv:2608. 06400v1 Announce Type: new Abstract: Reward models are central to learning from human preferences, yet identifying what drives their predictions remains challenging.
arXiv:2606. 12360v1 Announce Type: new Abstract: Language-model post-training is the main stage at which model behavior is shaped, yet it still largely involves optimization of scalar rewards that summarize diverse desiderata.
Language-model post-training is the main stage at which model behavior is shaped, yet it still largely involves optimization of scalar rewards that summarize diverse desiderata. This abstraction gives practitioners little visibility into what their data actually teaches models, allowing spurious correlations to be learned by a model and inducing undesirable behaviors such as over-stylization and sycophancy.
The paper investigates how to personalize Large Language Models by leveraging Mixture-of-Experts (MoE) architectures, which route tokens to a few specialized experts, thereby reducing polysemanticity compared to dense models. The authors adapt three open-weight MoE LLMs on a large personalization dataset, train crosscoders to learn interpretable personalized latent representations, and identify experts that encode distinct personalization features. Experiments show that personalized MoE LLMs match dense LLM performance while activating fewer parameters, exhibit lower polysemanticity, and that causal interventions on the identified experts can steer a non-personalized model toward personalized behavior.
The paper investigates federated learning where each client’s data consists of unknown mixtures of distinct tasks, a scenario termed compound heterogeneity. It shows that when tasks share a common feature geometry, the optimal model for a mixed client is a convex combination of task‑specific models, motivating input‑dependent routing to specialized experts. The authors propose FedSEE, a method that recovers task experts via a convex program and achieves better performance than baselines, reducing negative transfer by 2.9 points overall and 3.7 points for the worst‑served quartile.
CausalDreamer enhances generative world models by freezing the original video tokenizer and re-encoding its latent space into a factored representation that separates controllable versus uncontrollable dynamics and reward-relevant versus reward-irrelevant information. The pretrained dynamics model is fine‑tuned to predict this factored representation, enabling more accurate model‑predictive planning. On the MMBench2 benchmark, CausalDreamer improves normalized scores by 14% on clean tasks and 25% on manipulated variants compared to the baseline, while both models perform poorly on entirely new environments.
arXiv:2601. 21944v3 Announce Type: replace Abstract: The widespread adoption of deep learning models in computer vision has intensified concerns about interpretability.
arXiv:2610.05370v2 Announce Type: replace Abstract: Generative reward models (GRMs) are important for LLM optimization. Unlike scalar reward models, GRMs generate natural-language critiques alongside...
The paper presents a method for fine‑tuning a large language model (LLM) recommender to generate personalized, non‑harmful explanations for its recommendations. By training two LLM‑judge reward models and using constrained GRPO, the authors achieve a significant increase in the PASS rate for all three criteria, from 0.649 to 0.956 on their own judges and from 0.677 to 0.931 on an independent judge. The fine‑tuned model maintains its original recommendation performance, demonstrating that LLM‑based recommenders can be adapted to complex tasks without loss of effectiveness.
Personalization of Large Language Models (LLMs) is increasingly important for user-centric applications, yet prior methods mainly adapt dense LLMs, whose polysemantic neurons, each encoding multiple u...
arXiv:2605. 28969v2 Announce Type: replace-cross Abstract: If an AI agent makes decisions on a person's behalf, those decisions must align with its user.