Large Language Model
A Survey on LLM-as-a-Judge
Gu, Jiawei, Jiang, Xuhui, Shi, Zhichao, Tan, Hexiang, Zhai, Xuehao, Xu, Chengjin, Li, Wei, Shen, Yinghan, Ma, Shengjie, Liu, Honghao, Wang, Yuanzhuo, Guo, Jian
Accurate and consistent evaluation is crucial for decision-making across numerous fields, yet it remains a challenging task due to inherent subjectivity, variability, and scale. Large Language Models (LLMs) have achieved remarkable success across diverse domains, leading to the emergence of "LLM-as-a-Judge," where LLMs are employed as evaluators for complex tasks. With their ability to process diverse data types and provide scalable, cost-effective, and consistent assessments, LLMs present a compelling alternative to traditional expert-driven evaluations. However, ensuring the reliability of LLM-as-a-Judge systems remains a significant challenge that requires careful design and standardization. This paper provides a comprehensive survey of LLM-as-a-Judge, addressing the core question: How can reliable LLM-as-a-Judge systems be built? We explore strategies to enhance reliability, including improving consistency, mitigating biases, and adapting to diverse assessment scenarios. Additionally, we propose methodologies for evaluating the reliability of LLM-as-a-Judge systems, supported by a novel benchmark designed for this purpose. To advance the development and real-world deployment of LLM-as-a-Judge systems, we also discussed practical applications, challenges, and future directions. This survey serves as a foundational reference for researchers and practitioners in this rapidly evolving field.
BudgetMLAgent: A Cost-Effective LLM Multi-Agent system for Automating Machine Learning Tasks
Gandhi, Shubham, Patwardhan, Manasi, Vig, Lovekesh, Shroff, Gautam
Large Language Models (LLMs) excel in diverse applications including generation of code snippets, but often struggle with generating code for complex Machine Learning (ML) tasks. Although existing LLM single-agent based systems give varying performance depending on the task complexity, they purely rely on larger and expensive models such as GPT-4. Our investigation reveals that no-cost and low-cost models such as Gemini-Pro, Mixtral and CodeLlama perform far worse than GPT-4 in a single-agent setting. With the motivation of developing a cost-efficient LLM based solution for solving ML tasks, we propose an LLM Multi-Agent based system which leverages combination of experts using profiling, efficient retrieval of past observations, LLM cascades, and ask-the-expert calls. Through empirical analysis on ML engineering tasks in the MLAgentBench benchmark, we demonstrate the effectiveness of our system, using no-cost models, namely Gemini as the base LLM, paired with GPT-4 in cascade and expert to serve occasional ask-the-expert calls for planning. With 94.2\% reduction in the cost (from \$0.931 per run cost averaged over all tasks for GPT-4 single agent system to \$0.054), our system is able to yield better average success rate of 32.95\% as compared to GPT-4 single-agent system yielding 22.72\% success rate averaged over all the tasks of MLAgentBench.
Mixture of Knowledge Minigraph Agents for Literature Review Generation
Zhang, Zhi, Liu, Yan, Zhong, Sheng-hua, Chen, Gong, Yang, Yu, Cao, Jiannong
Literature reviews play a crucial role in scientific research for understanding the current state of research, identifying gaps, and guiding future studies on specific topics. However, the process of conducting a comprehensive literature review is yet time-consuming. This paper proposes a novel framework, collaborative knowledge minigraph agents (CKMAs), to automate scholarly literature reviews. A novel prompt-based algorithm, the knowledge minigraph construction agent (KMCA), is designed to identify relations between concepts from academic literature and automatically constructs knowledge minigraphs. By leveraging the capabilities of large language models on constructed knowledge minigraphs, the multiple path summarization agent (MPSA) efficiently organizes concepts and relations from different viewpoints to generate literature review paragraphs. We evaluate CKMAs on three benchmark datasets. Experimental results show the effectiveness of the proposed method, further revealing promising applications of LLMs in scientific research.
Literature Meets Data: A Synergistic Approach to Hypothesis Generation
Liu, Haokun, Zhou, Yangqiaoyu, Li, Mingxuan, Yuan, Chenfei, Tan, Chenhao
AI holds promise for transforming scientific processes, including hypothesis generation. Prior work on hypothesis generation can be broadly categorized into theory-driven and data-driven approaches. While both have proven effective in generating novel and plausible hypotheses, it remains an open question whether they can complement each other. To address this, we develop the first method that combines literature-based insights with data to perform LLM-powered hypothesis generation. We apply our method on five different datasets and demonstrate that integrating literature and data outperforms other baselines (8.97\% over few-shot, 15.75\% over literature-based alone, and 3.37\% over data-driven alone). Additionally, we conduct the first human evaluation to assess the utility of LLM-generated hypotheses in assisting human decision-making on two challenging tasks: deception detection and AI generated content detection. Our results show that human accuracy improves significantly by 7.44\% and 14.19\% on these tasks, respectively. These findings suggest that integrating literature-based and data-driven approaches provides a comprehensive and nuanced framework for hypothesis generation and could open new avenues for scientific inquiry.
Representation Shattering in Transformers: A Synthetic Study with Knowledge Editing
Nishi, Kento, Okawa, Maya, Ramesh, Rahul, Khona, Mikail, Tanaka, Hidenori, Lubana, Ekdeep Singh
Knowledge Editing (KE) algorithms alter models' weights to perform targeted updates to incorrect, outdated, or otherwise unwanted factual associations. To better identify the possibilities and limitations of these approaches, recent work has shown that applying KE can adversely affect models' factual recall accuracy and diminish their general reasoning abilities. While these studies give broad insights into the potential harms of KE algorithms, e.g., via performance evaluations on benchmarks, we argue little is understood as to why such destructive failures occur. Is it possible KE methods distort representations of concepts beyond the targeted fact, hence hampering abilities at broad? If so, what is the extent of this distortion? Motivated by such questions, we define a novel synthetic task wherein a Transformer is trained from scratch to internalize a "structured" knowledge graph. The structure enforces relationships between entities of the graph, such that editing a factual association has "trickling effects" on other entities in the graph (e.g., altering X's parent is Y to Z affects who X's siblings' parent is). Through evaluations of edited models and analysis of extracted representations, we show that KE inadvertently affects representations of entities beyond the targeted one, distorting relevant structures that allow a model to infer unseen knowledge about an entity. We call this phenomenon representation shattering and demonstrate that it results in degradation of factual recall and reasoning performance more broadly. To corroborate our findings in a more naturalistic setup, we perform preliminary experiments with pre-trained Llama and Mamba models, reproducing the representation shattering effect therein as well. Overall, our work yields a precise mechanistic hypothesis to explain why KE has adverse effects on model abilities. However, the static nature of their training pipelines implies that as our world evolves, models' internalized knowledge can become incorrect or outdated. To address this, recent work has proposed several protocols for knowledge editing (KE), wherein the goal is to minimally and precisely alter model weights such that only the targeted information (and its relevant associations) are updated, but all unrelated information remains (ideally) unaffected (Mitchell et al., 2022; Meng et al., 2022a; 2023; Dai et al., 2021; Cheng et al., 2023; De Cao et al., 2021; Sinitsin et al., 2020).
Towards Realistic Evaluation of Commit Message Generation by Matching Online and Offline Settings
Tsvetkov, Petr, Eliseeva, Aleksandra, Dig, Danny, Bezzubov, Alexander, Golubev, Yaroslav, Bryksin, Timofey, Zharov, Yaroslav
When a Commit Message Generation (CMG) system is integrated into the IDEs and other products at JetBrains, we perform online evaluation based on user acceptance of the generated messages. However, performing online experiments with every change to a CMG system is troublesome, as each iteration affects users and requires time to collect enough statistics. On the other hand, offline evaluation, a prevalent approach in the research literature, facilitates fast experiments but employs automatic metrics that are not guaranteed to represent the preferences of real users. In this work, we describe a novel way we employed to deal with this problem at JetBrains, by leveraging an online metric - the number of edits users introduce before committing the generated messages to the VCS - to select metrics for offline experiments. To support this new type of evaluation, we develop a novel markup collection tool mimicking the real workflow with a CMG system, collect a dataset with 57 pairs consisting of commit messages generated by GPT-4 and their counterparts edited by human experts, and design and verify a way to synthetically extend such a dataset. Then, we use the final dataset of 656 pairs to study how the widely used similarity metrics correlate with the online metric reflecting the real users' experience. Our results indicate that edit distance exhibits the highest correlation with the online metric, whereas commonly used similarity metrics such as BLEU and METEOR demonstrate low correlation. This contradicts the previous studies on similarity metrics for CMG, suggesting that user interactions with a CMG system in real-world settings differ significantly from the responses by human labelers within controlled environments. We release all the code and the dataset to support future research in the field: https://jb.gg/cmg-evaluation.
Generative AI Policies under the Microscope: How CS Conferences Are Navigating the New Frontier in Scholarly Writing
Nahar, Mahjabin, Lee, Sian, Guillen, Becky, Lee, Dongwon
While Gen-AI offers significant benefits in content generation and task automation [9], it can be also misused and abused in nefarious applications [7], with more significant risks toward long-tail populations and regions [6]. Professionals in fields like journalism and law still remain cautious due to concerns over hallucinations and ethical issues but scholars in Computer Science (CS), the field where Gen-AI originated, appear to be cautiously but actively exploring its use. For instance, [3] reports the increased use of large language models (LLMs) in the CS scholarly articles (up to 17.5%), compared to Mathematics articles (up to 6.3%), and [2] reports that between 6.5% and 16.9% of peer reviews at ICLR 2024, NeurIPS 2023, CoRL 2023, and EMNLP 2023 may have been significantly altered by LLMs beyond minor revisions. Considering researchers' increasing adoption of Gen-AI, it is crucial to establish usage guidelines and well-defined policies to promote fair and ethical practices in scholarly writing and peer reviews. Previous research also examined Gen-AI policies by major publishers like Elsevier, Springer, etc. [5], but there is still a lack of clear understanding of how CS conferences are adapting to this paradigm shift.
PalmBench: A Comprehensive Benchmark of Compressed Large Language Models on Mobile Platforms
Li, Yilong, Liu, Jingyu, Zhang, Hao, Narayanan, M Badri, Sharma, Utkarsh, Zhang, Shuai, Hu, Pan, Zeng, Yijing, Raghuram, Jayaram, Banerjee, Suman
Deploying large language models (LLMs) locally on mobile devices is advantageous in scenarios where transmitting data to remote cloud servers is either undesirable due to privacy concerns or impractical due to network connection. Recent advancements (MLC, 2023a; Gerganov, 2023) have facilitated the local deployment of LLMs. However, local deployment also presents challenges, particularly in balancing quality (generative performance), latency, and throughput within the hardware constraints of mobile devices. In this paper, we introduce our lightweight, all-in-one automated benchmarking framework that allows users to evaluate LLMs on mobile devices. We provide a comprehensive benchmark of various popular LLMs with different quantization configurations (both weights and activations) across multiple mobile platforms with varying hardware capabilities. Unlike traditional benchmarks that assess full-scale models on high-end GPU clusters, we focus on evaluating resource efficiency (memory and power consumption) and harmful output for compressed models on mobile devices. Our key observations include i) differences in energy efficiency and throughput across mobile platforms; ii) the impact of quantization on memory usage, GPU execution time, and power consumption; and iii) accuracy and performance degradation of quantized models compared to their non-quantized counterparts; and iv) the frequency of hallucinations and toxic content generated by compressed LLMs on mobile devices.
A Zero-Shot Open-Vocabulary Pipeline for Dialogue Understanding
Safa, Abdulfattah, Şahin, Gözde Gül
Dialogue State Tracking (DST) is crucial for understanding user needs and executing appropriate system actions in task-oriented dialogues. Majority of existing DST methods are designed to work within predefined ontologies and assume the availability of gold domain labels, struggling with adapting to new slots values. While Large Language Models (LLMs)-based systems show promising zero-shot DST performance, they either require extensive computational resources or they underperform existing fully-trained systems, limiting their practicality. To address these limitations, we propose a zero-shot, open-vocabulary system that integrates domain classification and DST in a single pipeline. Our approach includes reformulating DST as a question-answering task for less capable models and employing self-refining prompts for more adaptable ones. Our system does not rely on fixed slot values defined in the ontology allowing the system to adapt dynamically. We compare our approach with existing SOTA, and show that it provides up to 20% better Joint Goal Accuracy (JGA) over previous methods on datasets like Multi-WOZ 2.1, with up to 90% fewer requests to the LLM API.
TS-HTFA: Advancing Time Series Forecasting via Hierarchical Text-Free Alignment with Large Language Models
Wang, Pengfei, Zheng, Huanran, Xu, Qi'ao, Dai, Silong, Wang, Yiqiao, Yue, Wenjing, Zhu, Wei, Qian, Tianwen, Wang, Xiaoling
Given the significant potential of large language models (LLMs) in sequence modeling, emerging studies have begun applying them to time-series forecasting. Despite notable progress, existing methods still face two critical challenges: 1) their reliance on large amounts of paired text data, limiting the model applicability, and 2) a substantial modality gap between text and time series, leading to insufficient alignment and suboptimal performance. In this paper, we introduce \textbf{H}ierarchical \textbf{T}ext-\textbf{F}ree \textbf{A}lignment (\textbf{TS-HTFA}), a novel method that leverages hierarchical alignment to fully exploit the representation capacity of LLMs while eliminating the dependence on text data. Specifically, we replace paired text data with adaptive virtual text based on QR decomposition word embeddings and learnable prompt. Furthermore, we establish comprehensive cross-modal alignment at three levels: input, feature, and output. Extensive experiments on multiple time-series benchmarks demonstrate that HTFA achieves state-of-the-art performance, significantly improving prediction accuracy and generalization.