paper-with-me

Papers

BRAIn: Bayesian Reward-conditioned Amortized Inference for natural language generation from feedback

2024-02-04 · Gaurav Pandey, Yatin Nandwani, Tahira Naseem, Mayank Mishra, Guangxuan Xu, Dinesh Raghu, Sachindra Joshi, Asim Munawar, Ramón Fernandez Astudillo

Distribution matching methods for language model alignment such as Generation with Distributional Control (GDC) and Distributional Policy Gradient (DPG) have not received the same level of attention in reinforcement learning from human feedback (RLHF) as contrastive methods such as Sequence Likelihood Calibration (SLiC), Direct Preference Optimization (DPO) and its variants. We identify high variance of the gradient estimate as the primary reason for the lack of success of these methods and propose a self-normalized baseline to reduce the variance. We further generalize the target distribution in DPG, GDC and DPO by using Bayes' rule to define the reward-conditioned posterior. The resulting approach, referred to as BRAIn - Bayesian Reward-conditioned Amortized Inference acts as a bridge between distribution matching methods and DPO and significantly outperforms prior art in summarization and Antropic HH tasks.

📄 PDF Abstract BibTeX arXiv:2402.02479

Code (0)

등록된 구현이 없습니다.

Tasks

Language ModelingLanguage ModellingText Generation

Methods 이 논문이 사용한 방법론

DPO 설명 없음
DPG Deterministic Policy Gradient, or DPG, is a policy gradient method for reinforcement learning. Instead of the policy function $\pi\left(.\mid{s}\right)$ being modeled as a…
Entropy Regularization 설명 없음
PPO Proximal Policy Optimization, or PPO, is a policy gradient method for reinforcement learning. The motivation was to have an algorithm with the data efficiency and reliable…

Similar Papers 제목 키워드 기반

Mitigating Reward Hacking in RLHF via Bayesian Non-negative Reward Modeling

2026-02-11 · Zhibin Duan, Guowei Rong, Zhuo Li, Bo Chen 외 arxiv

Reward models learned from human preferences are central to aligning large language models (LLMs) via reinforcement learning from human feedback, yet they are often vulnerable to reward hacking due to noisy annotations a…

Reinforcement Learning

ALINE: Joint Amortization for Bayesian Inference and Active Data Acquisition

2025-06-08 · Daolang Huang, Xinyi Wen, Ayush Bharti, Samuel Kaski 외

Many critical applications, from autonomous scientific discovery to personalized medicine, demand systems that can both strategically acquire the most informative data and instantaneously perform inference based upon it.…

Active LearningBayesian InferenceExperimental Designscientific discovery

Amortized Simulation-Based Inference in Generalized Bayes via Neural Posterior Estimation

2026-01-29 · Shiyi Sun, Geoff K. Nicholls, Jeong Eun Lee arxiv

Generalized Bayesian Inference (GBI) tempers a loss with a temperature $β> 0$ to mitigate overconfidence and improve robustness under model misspecification, but existing GBI methods typically rely on costly MCMC or SDE-…

Bayesian Inference

A Statistical Assessment of Amortized Inference Under Signal-to-Noise Variation and Distribution Shift

2026-01-12 · Roy Shivam Ram Shreshtth, Arnab Hazra, Gourab Mukherjee arxiv

Since the turn of the century, approximate Bayesian inference has steadily evolved as new computational techniques have been incorporated to handle increasingly complex and large-scale predictive problems. The recent suc…

Bayesian Inference

MAVRL: Learning Reward Functions from Multiple Feedback Types with Amortized Variational Inference

2026-02-16 · Raphaël Baur, Yannick Metz, Maria Gkoulta, Mennatallah El-Assady 외 arxiv

Reward learning typically relies on a single feedback type or combines multiple feedback types using manually weighted loss terms. Currently, it remains unclear how to jointly learn reward functions from heterogeneous fe…

Bayesian Inference