paper-with-me

Papers

Self-Play Preference Optimization for Language Model Alignment

2024-05-01 · Yue Wu, Zhiqing Sun, Huizhuo Yuan, Kaixuan Ji, Yiming Yang, Quanquan Gu

Standard reinforcement learning from human feedback (RLHF) approaches relying on parametric models like the Bradley-Terry model fall short in capturing the intransitivity and irrationality in human preferences. Recent advancements suggest that directly working with preference probabilities can yield a more accurate reflection of human preferences, enabling more flexible and accurate language model alignment. In this paper, we propose a self-play-based method for language model alignment, which treats the problem as a constant-sum two-player game aimed at identifying the Nash equilibrium policy. Our approach, dubbed Self-Play Preference Optimization (SPPO), utilizes iterative policy updates to provably approximate the Nash equilibrium. Additionally, we propose a new SPPO objective which is both strongly motivated by theory and is simple and effective in practice. In our experiments, using only 60k prompts (without responses) from the UltraFeedback dataset and without any prompt augmentation, by leveraging a pre-trained preference model PairRM with only 0.4B parameters, SPPO can obtain a model from fine-tuning Mistral-7B-Instruct-v0.2 that achieves the state-of-the-art length-controlled win-rate of 28.53% against GPT-4-Turbo on AlpacaEval 2.0. It also outperforms the (iterative) DPO and IPO on MT-Bench, Arena-Hard, and the Open LLM Leaderboard. Starting from a stronger base model Llama-3-8B-Instruct, we are able to achieve a length-controlled win rate of 38.77%. Notably, the strong performance of SPPO is achieved without additional external supervision (e.g., responses, preferences, etc.) from GPT-4 or other stronger language models. Codes are available at https://github.com/uclaml/SPPO.

📄 PDF Abstract BibTeX arXiv:2405.00675

Code (1)

uclaml/sppo 공식 구현 pytorch

Tasks

Language ModelingLanguage Modellingmodel

Methods 이 논문이 사용한 방법론

Attention 설명 없음
DPO 설명 없음
BASE 설명 없음
Dense Connections Dense Connections, or Fully Connected Connections, are a type of layer in a deep neural network that use a linear operation where every input is connected to every output…
Dropout Dropout is a regularization technique for neural networks that drops a unit (along with connections) at training time with a specified probability $p$ (a common value is…
Label Smoothing Label Smoothing is a regularization technique that introduces noise for the labels. This accounts for the fact that datasets may have mistakes in them, so maximizing the…
Residual Connection 설명 없음
Softmax The Softmax output function transforms a previous layer's output into a vector of probabilities. It is commonly used for multiclass classification. Given an input vector $x$…

Similar Papers 제목 키워드 기반

Self-Augmented Preference Optimization: Off-Policy Paradigms for Language Model Alignment

2024-05-31 · Yueqin Yin, Zhendong Wang, Yujia Xie, Weizhu Chen 외

Traditional language model alignment methods, such as Direct Preference Optimization (DPO), are limited by their dependence on static, pre-collected paired preference data, which hampers their adaptability and practical …

Language ModelingLanguage Modelling

Stackelberg Game Preference Optimization for Data-Efficient Alignment of Language Models

2025-02-25 · Xu Chu, Zhixin Zhang, Tianyu Jia, Yujie Jin

Aligning language models with human preferences is critical for real-world deployment, but existing methods often require large amounts of high-quality human annotations. Aiming at a data-efficient alignment method, we p…

2kModels Alignment

Self-Play with Adversarial Critic: Provable and Scalable Offline Alignment for Language Models

2024-06-06 · Xiang Ji, Sanjeev Kulkarni, Mengdi Wang, Tengyang Xie

This work studies the challenge of aligning large language models (LLMs) with offline preference data. We focus on alignment by Reinforcement Learning from Human Feedback (RLHF) in particular. While popular preference op…

Offline RLreinforcement-learningReinforcement LearningReinforcement Learning (RL)

Magnetic Preference Optimization: Achieving Last-iterate Convergence for Language Model Alignment

2024-10-22 · Mingzhi Wang, Chengdong Ma, Qizhi Chen, Linjian Meng 외

Self-play methods have demonstrated remarkable success in enhancing model capabilities across various domains. In the context of Reinforcement Learning from Human Feedback (RLHF), self-play not only boosts Large Language…

Language ModelingLanguage ModellingLarge Language ModelModels Alignment

AMPO: Active Multi-Preference Optimization

2025-02-25 · Taneesh Gupta, Rahul Madhavan, Xuchao Zhang, Chetan Bansal 외

Multi-preference optimization enriches language-model alignment beyond pairwise preferences by contrasting entire sets of helpful and undesired responses, thereby enabling richer training signals for large language model…

Language ModelingLanguage Modelling