As large language models (LLMs) continue to advance, accurately and comprehensively evaluating their performance becomes increasingly challenging. Ranking the relative performance of LLMs based on Elo ratings, according to human judgment, is gaining more popularity. However, the extent to which humans and LLMs are capable evaluators remains uncertain. This study investigates the behavior of crowd-sourced and expert annotators, as well as LLMs, when comparing outputs from different models. To achieve this, we curate a dataset of intentionally flawed machine-generated answers. Our findings reveal a concerning bias in the evaluation process, as answers with factual errors are rated more favorably than answers that are too short or contained grammatical errors. To address this issue, we propose independently evaluating machine-generated text across multiple dimensions, rather than merging all the evaluation aspects into a single score. We instantiate this idea with the Elo rating system, resulting in the Multi-Elo Rating System (MERS). Empirical results from our study reveal that this proposed approach significantly enhances the quality of LLM-based evaluations, particularly in terms of factual accuracy. However, there is no significant improvement in crowd-sourced-based evaluations, indicating the need for further investigation.
Existing language models excel at writing from scratch, but many real-world scenarios require rewriting an existing document to fit a set of constraints. Although sentence-level rewriting has been fairly well-studied, little work has addressed the challenge of rewriting an entire document coherently. In this work, we introduce the task of document-level targeted content transfer and address it in the recipe domain, with a recipe as the document and a dietary restriction (such as vegan or dairy-free) as the targeted constraint. We propose a novel model for this task based on the generative pre-…
Can Small-Scale Data Poisoning Exacerbate Dialect-Linked Biases in Large Language Models?
Department of Electrical and Computer Engineering, Maroun Semaan Faculty of Engineering and Architecture（电气与计算机工程系，马鲁恩·塞马安工程与建筑学院）；American University of Beirut（贝鲁特美国大学）
Inclusivity in Large Language Models: Personality Traits and Gender Bias in Scientific Abstracts
Information Sciences Institute, University of Southern California（南加州大学信息科学学院）；Iovine and Young Academy, University of Southern California（南加州大学约万与杨学院）；In Vivo Group
Impact of Decoding Methods on Human Alignment of Conversational LLMs
Birla Institute of Technology and Science, Pilani（比拉理工与科学学院（皮拉尼））；National University of Singapore（新加坡国立大学）；NUS Center for Trusted Internet and Community（新加坡国立大学可信互联网与社区中心）；Department of Communications and New Media, National University of Singapore（新加坡国立大学传播与新媒体系）
本文研究了Beam Search、Top K Sampling和Nucleus Sampling等解码方法对LLM与人类对话对齐的影响，提出了内容、风格和心理测量导向的新衡量标准，发现较少beam数和较低P值对齐效果更好，且任务导向与开放式数据集表现不同。
On the Hidden Objective Biases of Group-based Reinforcement Learning
Department of Excellence in Robotics and AI, TeCIP, Scuola Superiore Sant’Anna（机器人与人工智能卓越部门、TeCIP、圣安娜高等学院）；Institute of Informatics and Telematics, National Research Council of Italy（信息与电信研究所、意大利国家研究委员会）；National Doctorate on Artificial Intelligence, Sapienza Università di Roma（人工智能国家博士项目、萨皮恩扎罗马大学）
It is well known that textual data on the internet and other digital platforms contain significant levels of bias and stereotypes. Although many such texts contain stereotypes and biases that inherently exist in natural language for reasons that are not necessarily malicious, there are crucial reasons to mitigate these biases. For one, these texts are being used as training corpus to train language models for salient applications like cv-screening, search engines, and chatbots; such applications are turning out to produce discriminatory results. Also, several research findings have concluded t…
The central idea of this paper is to gain a deeper understanding of song lyrics computationally. We focus on two aspects: style and biases of song lyrics. All prior works to understand these two aspects are limited to manual analysis of a small corpus of song lyrics. In contrast, we analyzed more than half a million songs spread over five decades. We characterize the lyrics style in terms of vocabulary, length, repetitiveness, speed, and readability. We have observed that the style of popular songs significantly differs from other songs. We have used distributed representation methods and WEAT…
USC Information Sciences Institute（南加州大学信息科学研究所）；Google Research（谷歌研究院）
There is a growing collection of work analyzing and mitigating societal biases in language understanding, generation, and retrieval tasks, though examining biases in creative tasks remains underexplored. Creative language applications are meant for direct interaction with users, so it is important to quantify and mitigate societal biases in these applications. We introduce a novel study on a pipeline to mitigate societal biases when retrieving next verse suggestions in a poetry composition system. Our results suggest that data augmentation through sentiment style transfer has potential for mit…
State-of-the-art pretrained contextualized models (PCM) eg. BERT use tasks such as WiC and WSD to evaluate their word-in-context representations. This inherently assumes that performance in these tasks reflect how well a model represents the coupled word and context semantics. We question this assumption by presenting the first quantitative analysis on the context-word interaction being tested in major contextual lexical semantic tasks. To achieve this, we run probing baselines on masked input, and propose measures to calculate and visualize the degree of context or word biases in existing dat…
Evaluation of biases in language models is often limited to synthetically generated datasets. This dependence traces back to the need for a prompt-style dataset to trigger specific behaviors of language models. In this paper, we address this gap by creating a prompt dataset with respect to occupations collected from real-world natural sentences present in Wikipedia. We aim to understand the differences between using template-based prompts and natural sentence prompts when studying gender-occupation biases in language models. We find bias evaluations are very sensitive to the design choices of …
StylisticBias: A Few Human Visual Cues Drive Most Social Biases in MLLMs
Technical University of Munich（慕尼黑工业大学）；Munich Center for Machine Learning（慕尼黑机器学习中心）；Princeton Center for Information and Technology Policy（普林斯顿信息与技术政策中心）
SRI International（SRI国际研究院）；Rutgers University（罗格斯大学）
We focus on Multimodal Machine Reading Comprehension (M3C) where a model is expected to answer questions based on given passage (or context), and the context and the questions can be in different modalities. Previous works such as RecipeQA have proposed datasets and cloze-style tasks for evaluation. However, we identify three critical biases stemming from the question-answer generation process and memorization capabilities of large deep models. These biases makes it easier for a model to overfit by relying on spurious correlations or naive data patterns. We propose a systematic framework to ad…
