Risk-Conditioned Fine-Tuning of Large Language Models

• URL: http://arxiv.org/abs/2609.08064v1
• Date: Tue, 08 Sep 2026 00:12:05 GMT
• Title: Risk-Conditioned Fine-Tuning of Large Language Models
• Abstract summary: We propose risk-conditioned RLHF, a framework that trains a single policy that provides a continuous risk-control interface.<n> Experiments across multiple benchmarks demonstrate that a single risk-conditioned policy can adapt to different risk levels at inference time.
• Score: 7.332941410825274
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large Language Models (LLMs) are increasingly deployed in settings where rare but severe harmful generations can have significant consequences. Existing Risk-Averse RLHF addresses this issue by optimizing Conditional Value-at-Risk (CVaR), but it trains policies for fixed risk levels and therefore cannot adjust the desired degree of risk aversion at inference time. In this paper, we propose risk-conditioned RLHF, a framework that trains a single policy that provides a continuous risk-control interface, enabling users to select different degrees of risk aversion without retraining or deploying multiple risk-specific models. Experiments across multiple benchmarks demonstrate that a single risk-conditioned policy can adapt to different risk levels at inference time, enabling more flexible and risk-aware LLM deployment.

Related papers

• Behavioural Signatures of Risk-Sensitive Decision-Making in Large Language Models [121.70781851874035]
  Large language models (LLMs) are increasingly used in decision support.<n>It is important to understand whether their choices under uncertainty exhibit stable and interpretable behavioural regularities.<n>Here we examine this question using a controlled multi-model framework based on no-limit Texas Hold'em.
  arXiv  Detail & Related papers  (2026-07-11T10:33:39Z)
• Constrained Language Model Policy Optimization via Risk-aware Stepwise Alignment [49.2305683068875]
  We propose Risk-aware Stepwise Alignment (RSA), a novel alignment method that incorporates risk awareness into the policy optimization process.<n> RSA mitigates risks induced by excessive model shift away from a reference policy, and it explicitly suppresses low-probability yet high-impact harmful behaviors.<n> Experimental results demonstrate that our method achieves high levels of helpfulness while ensuring strong safety.
  arXiv  Detail & Related papers  (2025-12-30T14:38:02Z)
• Risk-sensitive Actor-Critic with Static Spectral Risk Measures for Online and Offline Reinforcement Learning [4.8342038441006805]
  We propose a novel framework for optimizing static Spectral Risk Measures (SRM)<n>Our algorithms consistently outperform existing risk-sensitive methods in both online and offline environments across diverse domains.
  arXiv  Detail & Related papers  (2025-07-05T04:41:54Z)
• Answer, Refuse, or Guess? Investigating Risk-Aware Decision Making in Language Models [63.559461750135334]
  Language models (LMs) are increasingly used to build agents that can act autonomously to achieve goals.<n>We study this "answer-or-defer" problem with an evaluation framework that systematically varies human-specified risk structures.<n>We find that a simple skill-decomposition method, which isolates the independent skills required for answer-or-defer decision making, can consistently improve LMs' decision policies.
  arXiv  Detail & Related papers  (2025-03-03T09:16:26Z)
• C-RAG: Certified Generation Risks for Retrieval-Augmented Language Models [57.10361282229501]
  We propose C-RAG, the first framework to certify generation risks for RAG models. Specifically, we provide conformal risk analysis for RAG models and certify an upper confidence bound of generation risks. We prove that RAG achieves a lower conformal generation risk than that of a single LLM when the quality of the retrieval model and transformer is non-trivial.
  arXiv  Detail & Related papers  (2024-02-05T16:46:16Z)
• RiskQ: Risk-sensitive Multi-Agent Reinforcement Learning Value Factorization [49.26510528455664]
  We introduce the Risk-sensitive Individual-Global-Max (RIGM) principle as a generalization of the Individual-Global-Max (IGM) and Distributional IGM (DIGM) principles. We show that RiskQ can obtain promising performance through extensive experiments.
  arXiv  Detail & Related papers  (2023-11-03T07:18:36Z)
• DRL-ORA: Distributional Reinforcement Learning with Online Risk Adaption [9.191326295161725]
  We propose a new framework, Distributional RL with Online Risk Adaptation (DRL-ORA)<n>The framework unifies the existing variants of risk adaption approaches and offers better explainability and flexibility.<n>We show that DRL-ORA outperforms existing methods that rely on fixed risk levels or manually designed risk level adaptation in multiple classes of tasks.
  arXiv  Detail & Related papers  (2023-10-08T14:32:23Z)
• Risk-Averse Reinforcement Learning via Dynamic Time-Consistent Risk Measures [10.221369785560785]
  In this paper, we consider the problem of maximizing dynamic risk of a sequence of rewards in Markov Decision Processes (MDPs) Using a convex combination of expectation and conditional value-at-risk (CVaR) as a special one-step conditional risk measure, we reformulate the risk-averse MDP as a risk-neutral counterpart with augmented action space and manipulation on the immediate rewards. Our numerical studies show that the risk-averse setting can reduce the variance and enhance robustness of the results.
  arXiv  Detail & Related papers  (2023-01-14T21:43:18Z)
• RASR: Risk-Averse Soft-Robust MDPs with EVaR and Entropic Risk [28.811725782388688]
  We propose and analyze a new framework to jointly model the risk associated with uncertainties in finite-horizon and discounted infinite-horizon MDPs. We show that when the risk-aversion is defined using either EVaR or the entropic risk, the optimal policy in RASR can be computed efficiently using a new dynamic program formulation with a time-dependent risk level.
  arXiv  Detail & Related papers  (2022-09-09T00:34:58Z)
• Efficient Risk-Averse Reinforcement Learning [79.61412643761034]
  In risk-averse reinforcement learning (RL), the goal is to optimize some risk measure of the returns. We prove that under certain conditions this inevitably leads to a local-optimum barrier, and propose a soft risk mechanism to bypass it. We demonstrate improved risk aversion in maze navigation, autonomous driving, and resource allocation benchmarks.
  arXiv  Detail & Related papers  (2022-05-10T19:40:52Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.