Beyond Reverse KL: Generalizing Direct Preference Optimization with Diverse Divergence Constraints

• URL: http://arxiv.org/abs/2309.16240v1
• Date: Thu, 28 Sep 2023 08:29:44 GMT
• Title: Beyond Reverse KL: Generalizing Direct Preference Optimization with Diverse Divergence Constraints
• Authors: Chaoqi Wang, Yibo Jiang, Chenghao Yang, Han Liu, Yuxin Chen
• Abstract summary: The increasing capabilities of large language models (LLMs) raise opportunities for artificial general intelligence but amplify safety concerns. RLHF has emerged as a promising pathway towards AI alignment but brings forth challenges due to its complexity and dependence on a separate reward model. DPO has been proposed as an alternative, and it remains equivalent to RLHF under the reverse KL regularization constraint. We show that under certain $f$-divergences, including Jensen-Shannon divergence, forward KL divergences and $alpha$-divergences, the complex relationship between the reward and optimal policy can also be simplified
• Score: 26.274786600234876
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: The increasing capabilities of large language models (LLMs) raise opportunities for artificial general intelligence but concurrently amplify safety concerns, such as potential misuse of AI systems, necessitating effective AI alignment. Reinforcement Learning from Human Feedback (RLHF) has emerged as a promising pathway towards AI alignment but brings forth challenges due to its complexity and dependence on a separate reward model. Direct Preference Optimization (DPO) has been proposed as an alternative, and it remains equivalent to RLHF under the reverse KL regularization constraint. This paper presents $f$-DPO, a generalized approach to DPO by incorporating diverse divergence constraints. We show that under certain $f$-divergences, including Jensen-Shannon divergence, forward KL divergences and $\alpha$-divergences, the complex relationship between the reward and optimal policy can also be simplified by addressing the Karush-Kuhn-Tucker conditions. This eliminates the need for estimating the normalizing constant in the Bradley-Terry model and enables a tractable mapping between the reward function and the optimal policy. Our approach optimizes LLMs to align with human preferences in a more efficient and supervised manner under a broad set of divergence constraints. Empirically, adopting these divergences ensures a balance between alignment performance and generation diversity. Importantly, $f$-DPO outperforms PPO-based methods in divergence efficiency, and divergence constraints directly influence expected calibration error (ECE).

Related papers

• SALSA: Soup-based Alignment Learning for Stronger Adaptation in RLHF [22.88031166401938]
  This paper presents SALSA, a novel approach designed to overcome limitations by creating a more flexible and better located reference model. We show that SALSA fosters better exploration, achieving higher rewards and improving model robustness, out-of-distribution, and performance.
  arXiv  Detail & Related papers  (2024-11-04T04:53:43Z)
• Hierarchical Preference Optimization: Learning to achieve goals via feasible subgoals prediction [71.81851971324187]
  This work introduces Hierarchical Preference Optimization (HPO), a novel approach to hierarchical reinforcement learning (HRL) HPO addresses non-stationarity and infeasible subgoal generation issues when solving complex robotic control tasks. Experiments on challenging robotic navigation and manipulation tasks demonstrate impressive performance of HPO, where it shows an improvement of up to 35% over the baselines.
  arXiv  Detail & Related papers  (2024-11-01T04:58:40Z)
• $α$-DPO: Adaptive Reward Margin is What Direct Preference Optimization Needs [45.46582930202524]
  $alpha$-DPO is an adaptive preference optimization algorithm for large language models. It balances the policy model and the reference model to achieve personalized reward margins. It consistently outperforms DPO and SimPO across various model settings.
  arXiv  Detail & Related papers  (2024-10-14T04:29:57Z)
• Generalizing Alignment Paradigm of Text-to-Image Generation with Preferences through $f$-divergence Minimization [9.242941447250969]
  This study focuses on extending reverse Kullback-Leibler divergence in the alignment paradigm of text-to-image models to $f$-divergence. We conduct comprehensive evaluation on image-text alignment performance, human value alignment performance and generation diversity performance under different divergence constraints.
  arXiv  Detail & Related papers  (2024-09-15T15:46:03Z)
• Joint Demonstration and Preference Learning Improves Policy Alignment with Human Feedback [58.049113055986375]
  We develop a single stage approach named Alignment with Integrated Human Feedback (AIHF) to train reward models and the policy. The proposed approach admits a suite of efficient algorithms, which can easily reduce to, and leverage, popular alignment algorithms. We demonstrate the efficiency of the proposed solutions with extensive experiments involving alignment problems in LLMs and robotic control problems in MuJoCo.
  arXiv  Detail & Related papers  (2024-06-11T01:20:53Z)
• Provably Mitigating Overoptimization in RLHF: Your SFT Loss is Implicitly an Adversarial Regularizer [52.09480867526656]
  We identify the source of misalignment as a form of distributional shift and uncertainty in learning human preferences. To mitigate overoptimization, we first propose a theoretical algorithm that chooses the best policy for an adversarially chosen reward model. Using the equivalence between reward models and the corresponding optimal policy, the algorithm features a simple objective that combines a preference optimization loss and a supervised learning loss.
  arXiv  Detail & Related papers  (2024-05-26T05:38:50Z)
• Diffusion-based Reinforcement Learning via Q-weighted Variational Policy Optimization [55.97310586039358]
  Diffusion models have garnered widespread attention in Reinforcement Learning (RL) for their powerful expressiveness and multimodality. We propose a novel model-free diffusion-based online RL algorithm, Q-weighted Variational Policy Optimization (QVPO) Specifically, we introduce the Q-weighted variational loss, which can be proved to be a tight lower bound of the policy objective in online RL under certain conditions. We also develop an efficient behavior policy to enhance sample efficiency by reducing the variance of the diffusion policy during online interactions.
  arXiv  Detail & Related papers  (2024-05-25T10:45:46Z)
• Model-Based Epistemic Variance of Values for Risk-Aware Policy Optimization [59.758009422067]
  We consider the problem of quantifying uncertainty over expected cumulative rewards in model-based reinforcement learning. We propose a new uncertainty Bellman equation (UBE) whose solution converges to the true posterior variance over values. We introduce a general-purpose policy optimization algorithm, Q-Uncertainty Soft Actor-Critic (QU-SAC) that can be applied for either risk-seeking or risk-averse policy optimization.
  arXiv  Detail & Related papers  (2023-12-07T15:55:58Z)
• Permutation Invariant Policy Optimization for Mean-Field Multi-Agent Reinforcement Learning: A Principled Approach [128.62787284435007]
  We propose the mean-field proximal policy optimization (MF-PPO) algorithm, at the core of which is a permutation-invariant actor-critic neural architecture. We prove that MF-PPO attains the globally optimal policy at a sublinear rate of convergence. In particular, we show that the inductive bias introduced by the permutation-invariant neural architecture enables MF-PPO to outperform existing competitors.
  arXiv  Detail & Related papers  (2021-05-18T04:35:41Z)

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.