Xiangchi Yuan, Xingjian Diao, Zheyuan Liu, Chunhui Zhang, Dachuan Shi, Kejing Xia, Zhenbang Du, Shenglong Yao 等 10 位作者
LLM post-training often optimizes multiple objectives with unequal priorities, such as correctness and reasoning efficiency. However, the widely used additive reward aggregation in RL can violate these priorities: when primary and auxiliary advantages conflict, a sufficiently large auxiliary signal can overturn the preference induced by the primary objective. We term this problem \emph{advantage priority inversion}. Although reward-level gating can protect priority more directly, it still removes potentially useful auxiliary distinctions among rollout samples that fail the primary condition. To address this trade-off, we propose Primary Advantage Policy Optimization (\textbf{PAPO}), a priority-aware advantage aggregation method that asymmetrically combines a primary and an auxiliary advantage. PAPO retains auxiliary supervision when the objectives agree and smoothly attenuates conflicting auxiliary signals, preserving the primary sign during fusion for \emph{primary-consistent ordering}. This rule applies to sequence-level advantages in GRPO and token-level advantages in actor--critic PPO. Experiments on multiple domains and models show that PAPO consistently achieves the best primary objective while maintaining best or competitive auxiliary-objective performance across both GRPO- and PPO-based optimization. \href{https://xiangchi-yuan.github.io/papo-project/}{Code is available}.
Can AI agents autonomously train useful language models and improve them using feedback from users? We introduce RSI Arena, a two-stage study of autonomous post-training as a step toward recursive self-improvement. Ten frontier models act as training agents, each starting from NVIDIA Nemotron 3.5 Lightning 30B-A3B Base. During a six-day first stage, each agent has 1,000 GPU-hours and $300 in API credit on a shared 64-GPU cluster, independently selecting data, implementing training procedures, and choosing checkpoints. We evaluate their models on 20 benchmarks and through blind human evaluations. All ten agents produce working assistants, with the strongest model, trained by GPT-6 Astra, closing a median of 76% of the base-to-Instruct gap across 12 capability benchmarks. Its human-preference rating is also close to NVIDIA Instruct after statistical adjustment for answer length and formatting. In Stage 2, each agent receives the comparison records and votes involving its Stage 1 model and is given three additional days for further training. Using NVIDIA Instruct as a common reference across the two evaluation rounds, we observe an improvement from Stage 1 to Stage 2 in the style-adjusted relative rating of eight of the ten models. We show that benchmark gains were modest, with no clear link to preference gains, while some agents encouraged longer responses and one model showed increased harmful compliance on StrongREJECT. RSI Arena provides an auditable setting for studying how human feedback can guide autonomous model development. It takes a step toward recursive self-improvement with human feedback (RSIHF), in which people help define what improvement means for the AI systems they use.
Zhaoxuan Tan University of Notre Dame, Zixuan Zhang Amazon, Haoyang Wen Amazon, Zheng Li, Rongzhi Zhang Georgia Institute of Technology, Zhejiang University and Amazon, Pei Chen Amazon, Fengran Mo, Zheyuan Liu University of Notre Dame 等 11 位作者
Personalized large language models (LLMs) tailor content to individual preferences using user profiles or histories. However, existing parameter-efficient fine-tuning (PEFT) methods, such as the “One-PEFT-Per-User” (OPPU) paradigm, require training a separate adapter for each user, making them computationally expensive and impractical for real-time updates. We introduce Profile-to-PEFT, a scalable framework that employs a hypernetwork, trained end-to-end, to map a user’s encoded profile directly to a full set of adapter parameters (e.g., LoRA), eliminating per-user training at deployment. This design enables instant adaptation, generalization to unseen users, and privacy-preserving local deployment. Experimental results demonstrate that our method outperforms both prompt-based personalization and OPPU while using substantially fewer computational resources at deployment. The framework exhibits strong generalization to out-of-distribution users and maintains robustness across varying user activity levels and different embedding backbones. The proposed Profile-to-PEFT framework enables efficient, scalable, and adaptive LLM personalization suitable for large-scale applications.
尚无 AI 解读，展示原文摘要
cs.CL
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表
Xiangchi Yuan ByteDance Inc., Dachuan Shi Meta Superintelligence Labs and Georgia Institute of Technology, Chunhui Zhang Dartmouth College, Zheyuan Liu University of Notre Dame, Shenglong Yao, Soroush Vosoughi Dartmouth College, Wenke Lee Georgia Institute of Technology
Reinforcement learning (RL) is central to post-training, particularly for agentic models that require specialized reasoning behaviors. In this setting, model merging offers a practical mechanism for integrating multiple RL-trained agents from different tasks into a single generalist model. However, existing merging methods are designed for supervised fine-tuning (SFT), and they are suboptimal to preserve task-specific capabilities on RL-trained agentic models. The root is a task-vector mismatch between RL and SFT: on-policy RL induces task vectors that are highly sparse and heterogeneous, whereas SFT-style merging implicitly assumes dense and globally comparable task vectors. When standard global averaging is applied under this mismatch, RL’s non-overlapping task vectors that encode critical task-specific behaviors are reduced and parameter updates are diluted. To address this issue, we propose Reinforced Agent Merging (RAM), a distribution-aware merging framework explicitly designed for RL-trained agentic models. RAM disentangles shared and task-specific unique parameter updates, averaging shared components while selectively preserving and rescaling unique ones to counteract parameter update dilution. Experiments across multiple agent domains and model architectures demonstrate that RAM not only surpasses merging baselines, but also unlocks synergistic potential among agents to achieve performance superior to that of specialized agents in their domains.
尚无 AI 解读，展示原文摘要
cs.CL · cs.AI
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表
Zheyuan Liu University of Notre Dame, Yixin Wan University of California, Los Angeles, Kai-Wei Chang University of California, Los Angeles, Meng Jiang University of Notre Dame, Jieyu Zhao University of Southern California, Nouha Dziri Cohere Labs, Yuning Mao Meta Superintelligence Labs, Jia-Chen Gu University of California, Los Angeles 等 9 位作者
Controlling the knowledge and behavior of generative AI systems, including large language models (LLMs), multimodal LLMs (MLLMs), and text-to-image (T2I) models, has become critical as they are increasingly used in safety-sensitive and socially impactful applications. These models often encode unintended, biased, or private content, leading to harmful or unethical outputs. Post-training knowledge control has thus emerged as a practical framework for selectively modifying or removing model behaviors without full retraining, offering scalable and interpretable interventions for improving safety, privacy, and fairness. This tutorial introduces the foundations of post-training knowledge control and showcases recent frontier methods, bridging research insights with real-world practices from both academia and industry. We cover: (i) key motivations and failure modes, such as harmful generation and stereotype reinforcement; (ii) core methods such as machine unlearning, knowledge editing, and inference-time interventions for targeted behavior adjustment; and (iii) evaluation protocols for balancing forgetting, retention, and fairness. Case studies will span text and vision–language generation, including privacy preservation, bias mitigation, and factual correction.
尚无 AI 解读，展示原文摘要
cs.CL · cs.CV
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 5: Tutorial Abstracts) · 2026 · 正式发表