Bridging the LLM Accessibility Divide? Performance, Fairness, and Cost of Closed versus Open LLMs for Automated Essay Scoring

Closed large language models (LLMs) such as GPT-4 have set state-of-the-art results across a number of NLP tasks and have become central to NLP and machine learning (ML)-driven solutions. Closed LLMs' performance and wide adoption has sparked considerable debate about their accessibility in terms of availability, cost, and transparency. In this study, we perform a rigorous comparative analysis of nine leading LLMs, spanning closed, open, and open-source LLM ecosystems, across text assessment and generation tasks related to automated essay scoring. Our findings reveal that for few-shot learning-based assessment of human generated essays, open LLMs such as Llama 3 and Qwen2.5 perform comparably to GPT-4 in terms of predictive performance, with no significant differences in disparate impact scores when considering age- or race-related fairness. Moreover, Llama 3 offers a substantial cost advantage, being up to 37 times more cost-efficient than GPT-4. For generative tasks, we find that essays generated by top open LLMs are comparable to closed LLMs in terms of their semantic composition/embeddings and ML assessed scores. Our findings challenge the dominance of closed LLMs and highlight the democratizing potential of open LLMs, suggesting they can effectively bridge accessibility divides while maintaining competitive performance and fairness.

Predicting Practically? Domain Generalization for Predictive Analytics in Real-world Environments

Predictive machine learning models are widely used in customer relationship management (CRM) to forecast customer behaviors and support decision-making. However, the dynamic nature of customer behaviors often results in significant distribution shifts between training data and serving data, leading to performance degradation in predictive models. Domain generalization, which aims to train models that can generalize to unseen environments without prior knowledge of their distributions, has become a critical area of research. In this work, we propose a novel domain generalization method tailored to handle complex distribution shifts, encompassing both covariate and concept shifts. Our method builds upon the Distributionally Robust Optimization framework, optimizing model performance over a set of hypothetical worst-case distributions rather than relying solely on the training data. Through simulation experiments, we demonstrate the working mechanism of the proposed method. We also conduct experiments on a real-world customer churn dataset, and validate its effectiveness in both temporal and spatial generalization settings. Finally, we discuss the broader implications of our method for advancing Information Systems (IS) design research, particularly in building robust predictive models for dynamic managerial environments.

Empirical Guidelines for Deploying LLMs onto Resource-constrained Edge Devices

The scaling laws have become the de facto guidelines for designing large language models (LLMs), but they were studied under the assumption of unlimited computing resources for both training and inference. As LLMs are increasingly used as personalized intelligent assistants, their customization (i.e., learning through fine-tuning) and deployment onto resource-constrained edge devices will become more and more prevalent. An urging but open question is how a resource-constrained computing environment would affect the design choices for a personalized LLM. We study this problem empirically in this work. In particular, we consider the tradeoffs among a number of key design factors and their intertwined impacts on learning efficiency and accuracy. The factors include the learning methods for LLM customization, the amount of personalized data used for learning customization, the types and sizes of LLMs, the compression methods of LLMs, the amount of time afforded to learn, and the difficulty levels of the target use cases. Through extensive experimentation and benchmarking, we draw a number of surprisingly insightful guidelines for deploying LLMs onto resource-constrained devices. For example, an optimal choice between parameter learning and RAG may vary depending on the difficulty of the downstream task, the longer fine-tuning time does not necessarily help the model, and a compressed LLM may be a better choice than an uncompressed LLM to learn from limited personalized data.

FL-NAS: Towards Fairness of NAS for Resource Constrained Devices via Large Language Models

Neural Architecture Search (NAS) has become the de fecto tools in the industry in automating the design of deep neural networks for various applications, especially those driven by mobile and edge devices with limited computing resources. The emerging large language models (LLMs), due to their prowess, have also been incorporated into NAS recently and show some promising results. This paper conducts further exploration in this direction by considering three important design metrics simultaneously, i.e., model accuracy, fairness, and hardware deployment efficiency. We propose a novel LLM-based NAS framework, FL-NAS, in this paper, and show experimentally that FL-NAS can indeed find high-performing DNNs, beating state-of-the-art DNN models by orders-of-magnitude across almost all design considerations.

Enabling On-Device Large Language Model Personalization with Self-Supervised Data Selection and Synthesis

After a large language model (LLM) is deployed on edge devices, it is desirable for these devices to learn from user-generated conversation data to generate user-specific and personalized responses in real-time. However, user-generated data usually contains sensitive and private information, and uploading such data to the cloud for annotation is not preferred if not prohibited. While it is possible to obtain annotation locally by directly asking users to provide preferred responses, such annotations have to be sparse to not affect user experience. In addition, the storage of edge devices is usually too limited to enable large-scale fine-tuning with full user-generated data. It remains an open question how to enable on-device LLM personalization, considering sparse annotation and limited on-device storage. In this paper, we propose a novel framework to select and store the most representative data online in a self-supervised way. Such data has a small memory footprint and allows infrequent requests of user annotations for further fine-tuning. To enhance fine-tuning quality, multiple semantically similar pairs of question texts and expected responses are generated using the LLM. Our experiments show that the proposed framework achieves the best user-specific content-generating capability (accuracy) and fine-tuning speed (performance) compared with vanilla baselines. To the best of our knowledge, this is the very first on-device LLM personalization framework.

Bias A-head? Analyzing Bias in Transformer-Based Language Model Attention Heads

Transformer-based pretrained large language models (PLM) such as BERT and GPT have achieved remarkable success in NLP tasks. However, PLMs are prone to encoding stereotypical biases. Although a burgeoning literature has emerged on stereotypical bias mitigation in PLMs, such as work on debiasing gender and racial stereotyping, how such biases manifest and behave internally within PLMs remains largely unknown. Understanding the internal stereotyping mechanisms may allow better assessment of model fairness and guide the development of effective mitigation strategies. In this work, we focus on attention heads, a major component of the Transformer architecture, and propose a bias analysis framework to explore and identify a small set of biased heads that are found to contribute to a PLM's stereotypical bias. We conduct extensive experiments to validate the existence of these biased heads and to better understand how they behave. We investigate gender and racial bias in the English language in two types of Transformer-based PLMs: the encoder-based BERT model and the decoder-based autoregressive GPT model. Overall, the results shed light on understanding the bias behavior in pretrained language models.

Exploring the Relationship between In-Context Learning and Instruction Tuning

In-Context Learning (ICL) and Instruction Tuning (IT) are two primary paradigms of adopting Large Language Models (LLMs) to downstream applications. However, they are significantly different. In ICL, a set of demonstrations are provided at inference time but the LLM's parameters are not updated. In IT, a set of demonstrations are used to tune LLM's parameters in training time but no demonstrations are used at inference time. Although a growing body of literature has explored ICL and IT, studies on these topics have largely been conducted in isolation, leading to a disconnect between these two paradigms. In this work, we explore the relationship between ICL and IT by examining how the hidden states of LLMs change in these two paradigms. Through carefully designed experiments conducted with LLaMA-2 (7B and 13B), we find that ICL is implicit IT. In other words, ICL changes an LLM's hidden states as if the demonstrations were used to instructionally tune the model. Furthermore, the convergence between ICL and IT is largely contingent upon several factors related to the provided demonstrations. Overall, this work offers a unique perspective to explore the connection between ICL and IT and sheds light on understanding the behaviors of LLM.

When Automated Assessment Meets Automated Content Generation: Examining Text Quality in the Era of GPTs

The use of machine learning (ML) models to assess and score textual data has become increasingly pervasive in an array of contexts including natural language processing, information retrieval, search and recommendation, and credibility assessment of online content. A significant disruption at the intersection of ML and text are text-generating large-language models such as generative pre-trained transformers (GPTs). We empirically assess the differences in how ML-based scoring models trained on human content assess the quality of content generated by humans versus GPTs. To do so, we propose an analysis framework that encompasses essay scoring ML-models, human and ML-generated essays, and a statistical model that parsimoniously considers the impact of type of respondent, prompt genre, and the ML model used for assessment model. A rich testbed is utilized that encompasses 18,460 human-generated and GPT-based essays. Results of our benchmark analysis reveal that transformer pretrained language models (PLMs) more accurately score human essay quality as compared to CNN/RNN and feature-based ML methods. Interestingly, we find that the transformer PLMs tend to score GPT-generated text 10-15\% higher on average, relative to human-authored documents. Conversely, traditional deep learning and feature-based ML models score human text considerably higher. Further analysis reveals that although the transformer PLMs are exclusively fine-tuned on human text, they more prominently attend to certain tokens appearing only in GPT-generated text, possibly due to familiarity/overlap in pre-training. Our framework and results have implications for text classification settings where automated scoring of text is likely to be disrupted by generative AI.

Alternating minimization algorithm with initialization analysis for r-local and k-sparse unlabeled sensing

The unlabeled sensing problem is to recover an unknown signal from permuted linear measurements. We propose an alternating minimization algorithm with a suitable initialization for the widely considered k-sparse permutation model. Assuming either a Gaussian measurement matrix or a sub-Gaussian signal, we upper bound the initialization error for the r-local and k-sparse permutation models in terms of the block size $r$ and the number of shuffles k, respectively. Our algorithm is computationally scalable and, compared to baseline methods, achieves superior performance on real and synthetic datasets.

Constructing a Testbed for Psychometric Natural Language Processing

Psychometric measures of ability, attitudes, perceptions, and beliefs are crucial for understanding user behaviors in various contexts including health, security, e-commerce, and finance. Traditionally, psychometric dimensions have been measured and collected using survey-based methods. Inferring such constructs from user-generated text could afford opportunities for timely, unobtrusive, collection and analysis. In this paper, we describe our efforts to construct a corpus for psychometric natural language processing (NLP). We discuss our multi-step process to align user text with their survey-based response items and provide an overview of the resulting testbed which encompasses survey-based psychometric measures and accompanying user-generated text from over 8,500 respondents. We report preliminary results on the use of the text to categorize/predict users' survey response labels. We also discuss the important implications of our work and resulting testbed for future psychometric NLP research.