Unintended Impacts of LLM Alignment on Global Representation
- URL: http://arxiv.org/abs/2402.15018v2
- Date: Thu, 6 Jun 2024 22:31:48 GMT
- Title: Unintended Impacts of LLM Alignment on Global Representation
- Authors: Michael J. Ryan, William Held, Diyi Yang,
- Abstract summary: We show that developers align Large Language Models (LLMs) to user preferences through a variety of procedures, such as Reinforcement Learning From Human Feedback (RLHF) and Direct Preference Optimization (DPO)
We explore how alignment impacts performance along three axes of global representation: English dialects, multilingualism, and opinions from and about countries worldwide.
We conclude by discussing design decisions that led to these unintended impacts and recommendations for more equitable preference tuning.
- Score: 62.6579934112071
- License: http://creativecommons.org/licenses/by/4.0/
- Abstract: Before being deployed for user-facing applications, developers align Large Language Models (LLMs) to user preferences through a variety of procedures, such as Reinforcement Learning From Human Feedback (RLHF) and Direct Preference Optimization (DPO). Current evaluations of these procedures focus on benchmarks of instruction following, reasoning, and truthfulness. However, human preferences are not universal, and aligning to specific preference sets may have unintended effects. We explore how alignment impacts performance along three axes of global representation: English dialects, multilingualism, and opinions from and about countries worldwide. Our results show that current alignment procedures create disparities between English dialects and global opinions. We find alignment improves capabilities in several languages. We conclude by discussing design decisions that led to these unintended impacts and recommendations for more equitable preference tuning. We make our code and data publicly available on Github.
Related papers
- Assessing Code Generation with Intermediate Languages [6.999311675957218]
This study explores the utilization of intermediate languages, including various programming languages, natural language solutions, and pseudo-code.
Our findings reveal that intermediate languages generally exhibit greater efficacy in larger models that have not yet achieved state-of-the-art performance.
arXiv Detail & Related papers (2024-07-07T15:35:41Z) - BAPO: Base-Anchored Preference Optimization for Personalized Alignment in Large Language Models [26.526171463511332]
This paper examines the impact of personalized preference optimization on Large Language Models (LLMs)
BAPO effectively adapts to diverse user preferences while minimally affecting global knowledge or general alignment.
arXiv Detail & Related papers (2024-06-30T13:30:04Z) - Fairer Preferences Elicit Improved Human-Aligned Large Language Model Judgments [41.25558612970942]
We show that large language models (LLMs) exhibit preference biases and worrying sensitivity to prompt designs.
Motivated by this phenomenon, we propose an automatic Zero-shot Evaluation-oriented Prompt Optimization framework, ZEPO.
arXiv Detail & Related papers (2024-06-17T09:48:53Z) - Controllable Preference Optimization: Toward Controllable
Multi-Objective Alignment [107.63756895544842]
Alignment in artificial intelligence pursues consistency between model responses and human preferences as well as values.
Existing alignment techniques are mostly unidirectional, leading to suboptimal trade-offs and poor flexibility over various objectives.
We introduce controllable preference optimization (CPO), which explicitly specifies preference scores for different objectives.
arXiv Detail & Related papers (2024-02-29T12:12:30Z) - MaxMin-RLHF: Towards Equitable Alignment of Large Language Models with
Diverse Human Preferences [101.57443597426374]
Reinforcement Learning from Human Feedback (RLHF) aligns language models to human preferences by employing a singular reward model derived from preference data.
We learn a mixture of preference distributions via an expectation-maximization algorithm to better represent diverse human preferences.
Our algorithm achieves an average improvement of more than 16% in win-rates over conventional RLHF algorithms.
arXiv Detail & Related papers (2024-02-14T03:56:27Z) - Linear Alignment: A Closed-form Solution for Aligning Human Preferences without Tuning and Feedback [70.32795295142648]
Linear alignment is a novel algorithm that aligns language models with human preferences in one single inference step.
Experiments on both general and personalized preference datasets demonstrate that linear alignment significantly enhances the performance and efficiency of LLM alignment.
arXiv Detail & Related papers (2024-01-21T10:46:23Z) - MAPO: Advancing Multilingual Reasoning through Multilingual Alignment-as-Preference Optimization [65.31411639849516]
We propose a Multilingual-Alignment-as-Preference Optimization framework (MAPO) to align the reasoning processes in other languages with the dominant language.
Specifically, we harness an off-the-shelf translation model for the consistency between answers in non-dominant and dominant languages.
Experiments show that MAPO stably achieves significant improvements in the multilingual reasoning of various models.
arXiv Detail & Related papers (2024-01-12T18:03:54Z) - ULMA: Unified Language Model Alignment with Human Demonstration and
Point-wise Preference [16.73260713938154]
A typical alignment procedure consists of supervised fine-tuning and preference learning.
We introduce Point-wise Direct Preference Optimization, a novel preference learning method designed to harness point-wise feedback effectively.
Our work also uncovers a novel connection between supervised fine-tuning and point-wise preference learning, culminating in Unified Language Model Alignment.
arXiv Detail & Related papers (2023-12-05T07:52:12Z) - Learning Universal Representations from Word to Sentence [89.82415322763475]
This work introduces and explores the universal representation learning, i.e., embeddings of different levels of linguistic unit in a uniform vector space.
We present our approach of constructing analogy datasets in terms of words, phrases and sentences.
We empirically verify that well pre-trained Transformer models incorporated with appropriate training settings may effectively yield universal representation.
arXiv Detail & Related papers (2020-09-10T03:53:18Z)
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.