paper-with-me

홈 › Papers

The Policy Cliff: A Theoretical Analysis of Reward-Policy Maps in Large Language Models

2025-07-27 · Xingcheng Xu arxiv

Reinforcement learning (RL) plays a crucial role in shaping the behavior of large language and reasoning models (LLMs/LRMs). However, it often produces brittle and unstable policies, leading to critical failures such as spurious reasoning, deceptive alignment, and instruction disobedience that undermine the trustworthiness and safety of LLMs/LRMs. Currently, these issues lack a unified theoretical explanation and are typically addressed using ad-hoc heuristics. This paper presents a rigorous mathematical framework for analyzing the stability of the mapping from a reward function to the optimal policy. We show that policy brittleness often stems from non-unique optimal actions, a common occurrence when multiple valid traces exist in a reasoning task. This theoretical lens provides a unified explanation for a range of seemingly disparate failures, reframing them as rational outcomes of optimizing rewards that may be incomplete or noisy, especially in the presence of action degeneracy. We extend this analysis from the fundamental single-reward setting to the more realistic multi-reward RL across diverse domains, showing how stability is governed by an "effective reward" aggregation mechanism. We also prove that entropy regularization restores policy stability at the cost of increased stochasticity. Our framework provides a unified explanation for recent empirical findings on deceptive reasoning, instruction-following trade-offs, and RLHF-induced sophistry, and is further validated through perturbation experiments in multi-reward RL. This work advances policy-stability analysis from empirical heuristics towards a principled theory, offering essential insights for designing safer and more trustworthy AI systems.

📄 PDF Abstract BibTeX arXiv:2507.20150

Code (0)

등록된 구현이 없습니다.

Tasks

Reinforcement Learning

Similar Papers 제목 키워드 기반

Cliff: Learning Process Rewards from the First Mistake

2026-09-02 · Peixuan Han, Runhui Wang, Ketan Ramaneti, Jie Hao 외 hf

Reinforcement learning with verifiable rewards (RLVR) has emerged as a powerful paradigm for large language model (LLM) post-training, but its reliance on coarse outcome rewards leads to limited guidance on intermediate …

Reinforcement Learning

LoRA Scaffolded Policy Optimization (LSPO): A Sampling-Time Low-Rank Scaffold for Recovering Reinforcement-Learning Gradient on Zero-Reward Cliff Prompts

2026-07-30 · Ken Ding arxiv

Reinforcement learning from verifiable rewards (RLVR) for mathematical reasoning suffers from a structural blind spot: on "cliff" prompts-those on which every sampled rollout in a group fails-the group-normalized advanta…

Reinforcement LearningMathematical Reasoning

Cliff Diving: Exploring Reward Surfaces in Reinforcement Learning Environments

2022-05-14 · Ryan Sullivan, J. K. Terry, Benjamin Black, John P. Dickerson

Visualizing optimization landscapes has led to many fundamental insights in numeric optimization, and novel improvements to optimization techniques. However, visualizations of the objective that reinforcement learning op…

reinforcement-learningReinforcement LearningReinforcement Learning (RL)

Zero Collapse: A Failure Mode of Policy Gradient Methods in Discontinuous Reward Environments

2026-05-29 · Nishant Kumar, Enrique Areyan Viqueira, Amy Greenwald arxiv

Bidding in repeated auctions is a central challenge for reinforcement learning (RL), combining continuous control with the strategic complexities of digital advertising. While policy gradient and value-based methods seem…

Reinforcement LearningContinuous Control

Provably Efficient Policy-Reward Co-Pretraining for Adversarial Imitation Learning

2026-06-20 · Tian Xu, Zexuan Chen, Zhilong Zhang, Yi-Chen Li 외 arxiv

Adversarial imitation learning (AIL) achieves high-quality imitation compared to behavioral cloning (BC), but demands substantial online environment interaction. Recent empirical work has explored initializing AIL algori…