SPO: Multi-Dimensional Preference Sequential Alignment With Implicit Reward Modeling
- URL: http://arxiv.org/abs/2405.12739v2
- Date: Fri, 11 Oct 2024 06:18:30 GMT
- Title: SPO: Multi-Dimensional Preference Sequential Alignment With Implicit Reward Modeling
- Authors: Xingzhou Lou, Junge Zhang, Jian Xie, Lifeng Liu, Dong Yan, Kaiqi Huang,
- Abstract summary: We propose a method that sequentially fine-tunes large language models to align with human preferences.
We theoretically derive closed-form optimal SPO policy and loss function.
Empirical results on LLMs of different size and multiple evaluation datasets demonstrate that SPO successfully aligns LLMs across multiple dimensions of human preferences.
- Score: 34.32744849352087
- License:
- Abstract: Human preference alignment is critical in building powerful and reliable large language models (LLMs). However, current methods either ignore the multi-dimensionality of human preferences (e.g. helpfulness and harmlessness) or struggle with the complexity of managing multiple reward models. To address these issues, we propose Sequential Preference Optimization (SPO), a method that sequentially fine-tunes LLMs to align with multiple dimensions of human preferences. SPO avoids explicit reward modeling, directly optimizing the models to align with nuanced human preferences. We theoretically derive closed-form optimal SPO policy and loss function. Gradient analysis is conducted to show how SPO manages to fine-tune the LLMs while maintaining alignment on previously optimized dimensions. Empirical results on LLMs of different size and multiple evaluation datasets demonstrate that SPO successfully aligns LLMs across multiple dimensions of human preferences and significantly outperforms the baselines.
Related papers
- MM-RLHF: The Next Step Forward in Multimodal LLM Alignment [59.536850459059856]
We introduce MM-RLHF, a dataset containing $mathbf120k$ fine-grained, human-annotated preference comparison pairs.
We propose several key innovations to improve the quality of reward models and the efficiency of alignment algorithms.
Our approach is rigorously evaluated across $mathbf10$ distinct dimensions and $mathbf27$ benchmarks.
arXiv Detail & Related papers (2025-02-14T18:59:51Z) - Few-shot Steerable Alignment: Adapting Rewards and LLM Policies with Neural Processes [50.544186914115045]
Large language models (LLMs) are increasingly embedded in everyday applications.
Ensuring their alignment with the diverse preferences of individual users has become a critical challenge.
We present a novel framework for few-shot steerable alignment.
arXiv Detail & Related papers (2024-12-18T16:14:59Z) - MetaAlign: Align Large Language Models with Diverse Preferences during Inference Time [50.41806216615488]
Large Language Models (LLMs) acquire extensive knowledge and remarkable abilities from extensive text corpora.
To make LLMs more usable, aligning them with human preferences is essential.
We propose an effective method, textbf MetaAlign, which aims to help LLMs dynamically align with various explicit or implicit preferences specified at inference time.
arXiv Detail & Related papers (2024-10-18T05:31:13Z) - Zeroth-Order Fine-Tuning of LLMs in Random Subspaces [66.27334633749734]
As language models grow in size, memory demands for backpropagation increase.
Zeroth-order (ZOZO) optimization methods offer a memory-efficient alternative.
We show that SubZero enhances fine-tuning and achieves faster results compared to standard ZOZO approaches.
arXiv Detail & Related papers (2024-10-11T17:01:43Z) - Ordinal Preference Optimization: Aligning Human Preferences via NDCG [28.745322441961438]
We develop an end-to-end preference optimization algorithm by approxing NDCG with a differentiable surrogate loss.
OPO outperforms existing pairwise and listwise approaches on evaluation sets and general benchmarks like AlpacaEval.
arXiv Detail & Related papers (2024-10-06T03:49:28Z) - Self-supervised Preference Optimization: Enhance Your Language Model with Preference Degree Awareness [27.43137305486112]
We propose a novel Self-supervised Preference Optimization (SPO) framework, which constructs a self-supervised preference degree loss combined with the alignment loss.
The results demonstrate that SPO can be seamlessly integrated with existing preference optimization methods to achieve state-of-the-art performance.
arXiv Detail & Related papers (2024-09-26T12:37:26Z) - Self-Exploring Language Models: Active Preference Elicitation for Online Alignment [88.56809269990625]
We propose a bilevel objective optimistically biased towards potentially high-reward responses to actively explore out-of-distribution regions.
Our experimental results demonstrate that when fine-tuned on Zephyr-7B-SFT and Llama-3-8B-Instruct models, Self-Exploring Language Models (SELM) significantly boosts the performance on instruction-following benchmarks.
arXiv Detail & Related papers (2024-05-29T17:59:07Z) - Multi-Reference Preference Optimization for Large Language Models [56.84730239046117]
We introduce a novel closed-form formulation for direct preference optimization using multiple reference models.
The resulting algorithm, Multi-Reference Preference Optimization (MRPO), leverages broader prior knowledge from diverse reference models.
Our experiments demonstrate that LLMs finetuned with MRPO generalize better in various preference data, regardless of data scarcity or abundance.
arXiv Detail & Related papers (2024-05-26T00:29:04Z) - Rewards-in-Context: Multi-objective Alignment of Foundation Models with Dynamic Preference Adjustment [46.44464839353993]
We introduce Rewards-in-Context (RiC), which conditions the response of a foundation model on multiple rewards in its prompt context.
RiC only requires supervised fine-tuning of a single foundation model and supports dynamic adjustment for user preferences during inference time.
arXiv Detail & Related papers (2024-02-15T18:58:31Z)
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.