llm - 2024_09
Navigation
Papers
Key-value (KV) caching has become the de-facto to accelerate generation speed for large language models (LLMs) inference. However, the growing cache demand with increasing sequence length has transformed LLM inference to be a memory bound problem, significantly constraining the system throughput. Existing methods rely on dropping unimportant tokens or quantizing all entries uniformly. Such methods, however, often incur high approximation errors to represent the compressed matrices. The autoregressive decoding process further compounds the error of each step, resulting in critical deviation in model generation and deterioration of performance. To tackle this challenge, we propose GEAR, an efficient KV cache compression framework that achieves near-lossless high-ratio compression. GEAR first applies quantization to majority of entries of similar magnitudes to ultra-low precision. It then employs a low rank matrix to approximate the quantization error, and a sparse matrix to remedy individual errors from outlier entries. By adeptly integrating three techniques, GEAR is able to fully exploit their synergistic potentials. Our experiments demonstrate that compared to alternatives, GEAR achieves near-lossless 4-bit KV cache compression with up to 2.38x throughput improvement, while reducing peak-memory size up to 2.29x. Our code is publicly available at https://github.com/HaoKang-Timmy/GEAR.
Decision-making is a complex process requiring diverse abilities, making it an excellent framework for evaluating Large Language Models (LLMs). Researchers have examined LLMs' decision-making through the lens of Game Theory. However, existing evaluation mainly focus on two-player scenarios where an LLM competes against another. Additionally, previous benchmarks suffer from test set leakage due to their static design. We introduce GAMA($\gamma$)-Bench, a new framework for evaluating LLMs' Gaming Ability in Multi-Agent environments. It includes eight classical game theory scenarios and a dynamic scoring scheme specially designed to quantitatively assess LLMs' performance. $\gamma$-Bench allows flexible game settings and adapts the scoring system to different game parameters, enabling comprehensive evaluation of robustness, generalizability, and strategies for improvement. Our results indicate that GPT-3.5 demonstrates strong robustness but limited generalizability, which can be enhanced using methods like Chain-of-Thought. We also evaluate twelve LLMs from six model families, including GPT-3.5, GPT-4, Gemini, LLaMA-3.1, Mixtral, and Qwen-2. Gemini-1.5-Pro outperforms others, scoring of $68.1$ out of $100$, followed by LLaMA-3.1-70B ($64.5$) and Mixtral-8x22B ($61.4$). All code and experimental results are publicly available via https://github.com/CUHK-ARISE/GAMABench.
Deaf and Hard of Hearing (DHH) learners face unique challenges in learning environments, often due to a lack of tailored educational materials that address their specific needs. This study explores the potential of Large Language Models (LLMs) to generate personalized quiz questions to enhance DHH students' video-based learning experiences. We developed a prototype leveraging LLMs to generate questions with emphasis on two unique strategies: Visual Questions, which identify video segments where visual information might be misrepresented, and Emotion Questions, which highlight moments where previous DHH learners experienced learning difficulty manifested in emotional responses. Through user studies with DHH undergraduates, we evaluated the effectiveness of these LLM-generated questions in supporting the learning experience. Our findings indicate that while LLMs offer significant potential for personalized learning, challenges remain in the interaction accessibility for the diverse DHH community. The study highlights the importance of considering language diversity and culture in LLM-based educational technology design.
This study evaluates the performance of large language models (LLMs) as medical agents in Portuguese, aiming to develop a reliable and relevant virtual assistant for healthcare professionals. The HealthCareMagic-100k-en and MedQuAD datasets, translated from English using GPT-3.5, were used to fine-tune the ChatBode-7B model using the PEFT-QLoRA method. The InternLM2 model, with initial training on medical data, presented the best overall performance, with high precision and adequacy in metrics such as accuracy, completeness and safety. However, DrBode models, derived from ChatBode, exhibited a phenomenon of catastrophic forgetting of acquired medical knowledge. Despite this, these models performed frequently or even better in aspects such as grammaticality and coherence. A significant challenge was low inter-rater agreement, highlighting the need for more robust assessment protocols. This work paves the way for future research, such as evaluating multilingual models specific to the medical field, improving the quality of training data, and developing more consistent evaluation methodologies for the medical field.
Probing learned concepts in large language models (LLMs) is crucial for understanding how semantic knowledge is encoded internally. Training linear classifiers on probing tasks is a principle approach to denote the vector of a certain concept in the representation space. However, the single vector identified for a concept varies with both data and training, making it less robust and weakening its effectiveness in real-world applications. To address this challenge, we propose an approach to approximate the subspace representing a specific concept. Built on linear probing classifiers, we extend the concept vectors into Gaussian Concept Subspace (GCS). We demonstrate GCS's effectiveness through measuring its faithfulness and plausibility across multiple LLMs with different sizes and architectures. Additionally, we use representation intervention tasks to showcase its efficacy in real-world applications such as emotion steering. Experimental results indicate that GCS concept vectors have the potential to balance steering performance and maintaining the fluency in natural language generation tasks.
Evaluating LLM-generated text has become a key challenge, especially in domain-specific contexts like the medical field. This work introduces a novel evaluation methodology for LLM-generated medical explanatory arguments, relying on Proxy Tasks and rankings to closely align results with human evaluation criteria, overcoming the biases typically seen in LLMs used as judges. We demonstrate that the proposed evaluators are robust against adversarial attacks, including the assessment of non-argumentative text. Additionally, the human-crafted arguments needed to train the evaluators are minimized to just one example per Proxy Task. By examining multiple LLM-generated arguments, we establish a methodology for determining whether a Proxy Task is suitable for evaluating LLM-generated medical explanatory arguments, requiring only five examples and two human experts.
Goal-oriented planning, or anticipating a series of actions that transition an agent from its current state to a predefined objective, is crucial for developing intelligent assistants aiding users in daily procedural tasks. The problem presents significant challenges due to the need for comprehensive knowledge of temporal and hierarchical task structures, as well as strong capabilities in reasoning and planning. To achieve this, prior work typically relies on extensive training on the target dataset, which often results in significant dataset bias and a lack of generalization to unseen tasks. In this work, we introduce VidAssist, an integrated framework designed for zero/few-shot goal-oriented planning in instructional videos. VidAssist leverages large language models (LLMs) as both the knowledge base and the assessment tool for generating and evaluating action plans, thus overcoming the challenges of acquiring procedural knowledge from small-scale, low-diversity datasets. Moreover, VidAssist employs a breadth-first search algorithm for optimal plan generation, in which a composite of value functions designed for goal-oriented planning is utilized to assess the predicted actions at each step. Extensive experiments demonstrate that VidAssist offers a unified framework for different goal-oriented planning setups, e.g., visual planning for assistance (VPA) and procedural planning (PP), and achieves remarkable performance in zero-shot and few-shot setups. Specifically, our few-shot model outperforms the prior fully supervised state-of-the-art method by +7.7% in VPA and +4.81% PP task on the COIN dataset while predicting 4 future actions. Code, and models are publicly available at https://sites.google.com/view/vidassist.
The financial domain presents a complex environment for stock market prediction, characterized by volatile patterns and the influence of multifaceted data sources. Traditional models have leveraged either Convolutional Neural Networks (CNN) for spatial feature extraction or Long Short-Term Memory (LSTM) networks for capturing temporal dependencies, with limited integration of external textual data. This paper proposes a novel Two-Level Conv-LSTM Neural Network integrated with a Large Language Model (LLM) for comprehensive stock advising. The model harnesses the strengths of Conv-LSTM for analyzing time-series data and LLM for processing and understanding textual information from financial news, social media, and reports. In the first level, convolutional layers are employed to identify local patterns in historical stock prices and technical indicators, followed by LSTM layers to capture the temporal dynamics. The second level integrates the output with an LLM that analyzes sentiment and contextual information from textual data, providing a holistic view of market conditions. The combined approach aims to improve prediction accuracy and provide contextually rich stock advising.
This paper investigates the subtle and often concealed biases present in Large Language Models (LLMs), focusing on implicit biases that may remain despite passing explicit bias tests. Implicit biases are significant because they influence the decisions made by these systems, potentially perpetuating stereotypes and discrimination, even when LLMs appear to function fairly. Traditionally, explicit bias tests or embedding-based methods are employed to detect bias, but these approaches can overlook more nuanced, implicit forms of bias. To address this, we introduce two novel psychological-inspired methodologies: the LLM Implicit Association Test (IAT) Bias and the LLM Decision Bias, designed to reveal and measure implicit biases through prompt-based and decision-making tasks. Additionally, open-ended generation tasks with thematic analysis of word generations and storytelling provide qualitative insights into the model's behavior. Our findings demonstrate that the LLM IAT Bias correlates with traditional methods and more effectively predicts downstream behaviors, as measured by the LLM Decision Bias, offering a more comprehensive framework for detecting subtle biases in AI systems. This research advances the field of AI ethics by proposing new methods to continually assess and mitigate biases in LLMs, highlighting the importance of qualitative and decision-focused evaluations to address challenges that previous approaches have not fully captured.
Non-Factoid (NF) Question Answering (QA) is challenging to evaluate due to diverse potential answers and no objective criterion. The commonly used automatic evaluation metrics like ROUGE or BERTScore cannot accurately measure semantic similarities or answers from different perspectives. Recently, Large Language Models (LLMs) have been resorted to for NFQA evaluation due to their compelling performance on various NLP tasks. Common approaches include pointwise scoring of each candidate answer and pairwise comparisons between answers. Inspired by the evolution from pointwise to pairwise to listwise in learning-to-rank methods, we propose a novel listwise NFQA evaluation approach, that utilizes LLMs to rank candidate answers in a list of reference answers sorted by descending quality. Moreover, for NF questions that do not have multi-grade or any golden answers, we leverage LLMs to generate the reference answer list of various quality to facilitate the listwise evaluation. Extensive experimental results on three NFQA datasets, i.e., ANTIQUE, the TREC-DL-NF, and WebGLM show that our method has significantly higher correlations with human annotations compared to automatic scores and common pointwise and pairwise approaches.
Large semantic knowledge bases are grounded in factual knowledge. However, recent approaches to dense text representations (i.e. embeddings) do not efficiently exploit these resources. Dense and robust representations of documents are essential for effectively solving downstream classification and retrieval tasks. This work demonstrates that injecting embedded information from knowledge bases can augment the performance of contemporary Large Language Model (LLM)-based representations for the task of text classification. Further, by considering automated machine learning (AutoML) with the fused representation space, we demonstrate it is possible to improve classification accuracy even if we use low-dimensional projections of the original representation space obtained via efficient matrix factorization. This result shows that significantly faster classifiers can be achieved with minimal or no loss in predictive performance, as demonstrated using five strong LLM baselines on six diverse real-life datasets. The code is freely available at \url{https://github.com/bkolosk1/bablfusion.git}.
As LLMs become capable of complex tasks, there is growing potential for personalized interactions tailored to the subtle and idiosyncratic preferences of the user. We present a public benchmark, PersonalLLM, focusing on adapting LLMs to provide maximal benefits for a particular user. Departing from existing alignment benchmarks that implicitly assume uniform preferences, we curate open-ended prompts paired with many high-quality answers over which users would be expected to display heterogeneous latent preferences. Instead of persona-prompting LLMs based on high-level attributes (e.g., user's race or response length), which yields homogeneous preferences relative to humans, we develop a method that can simulate a large user base with diverse preferences from a set of pre-trained reward models. Our dataset and generated personalities offer an innovative testbed for developing personalization algorithms that grapple with continual data sparsity--few relevant feedback from the particular user--by leveraging historical data from other (similar) users. We explore basic in-context learning and meta-learning baselines to illustrate the utility of PersonalLLM and highlight the need for future methodological development. Our dataset is available at https://huggingface.co/datasets/namkoong-lab/PersonalLLM
As mobile devices increasingly become focal points for advanced applications, edge computing presents a viable solution to their inherent computational limitations, particularly in deploying large language models (LLMs). However, despite the advancements in edge computing, significant challenges remain in efficient training and deploying LLMs due to the computational demands and data privacy concerns associated with these models. This paper explores a collaborative training framework that integrates mobile users with edge servers to optimize resource allocation, thereby enhancing both performance and efficiency. Our approach leverages parameter-efficient fine-tuning (PEFT) methods, allowing mobile users to adjust the initial layers of the LLM while edge servers handle the more demanding latter layers. Specifically, we formulate a multi-objective optimization problem to minimize the total energy consumption and delay during training. We also address the common issue of instability in model performance by incorporating stability enhancements into our objective function. Through novel fractional programming technique, we achieve a stationary point for the formulated problem. Simulations demonstrate that our method reduces the energy consumption as well as the latency, and increases the reliability of LLMs across various mobile settings.
LLM-based agents have been widely applied as personal assistants, capable of memorizing information from user messages and responding to personal queries. However, there still lacks an objective and automatic evaluation on their memory capability, largely due to the challenges in constructing reliable questions and answers (QAs) according to user messages. In this paper, we propose MemSim, a Bayesian simulator designed to automatically construct reliable QAs from generated user messages, simultaneously keeping their diversity and scalability. Specifically, we introduce the Bayesian Relation Network (BRNet) and a causal generation mechanism to mitigate the impact of LLM hallucinations on factual information, facilitating the automatic creation of an evaluation dataset. Based on MemSim, we generate a dataset in the daily-life scenario, named MemDaily, and conduct extensive experiments to assess the effectiveness of our approach. We also provide a benchmark for evaluating different memory mechanisms in LLM-based agents with the MemDaily dataset. To benefit the research community, we have released our project at https://github.com/nuster1128/MemSim.
Large language models (LLMs) are vulnerable to adversarial attacks that can elicit harmful responses. Defending against such attacks remains challenging due to the opacity of jailbreaking mechanisms and the high computational cost of training LLMs robustly. We demonstrate that adversarial attacks share a universal mechanism for circumventing LLM safeguards that works by ablating a dimension in the residual stream embedding space called the refusal feature. We further show that the operation of refusal feature ablation (RFA) approximates the worst-case perturbation of offsetting model safety. Based on these findings, we propose Refusal Feature Adversarial Training (ReFAT), a novel algorithm that efficiently performs LLM adversarial training by simulating the effect of input-level attacks via RFA. Experiment results show that ReFAT significantly improves the robustness of three popular LLMs against a wide range of adversarial attacks, with considerably less computational overhead compared to existing adversarial training methods.
Neural metrics for machine translation (MT) evaluation have become increasingly prominent due to their superior correlation with human judgments compared to traditional lexical metrics. Researchers have therefore utilized neural metrics through quality-informed decoding strategies, achieving better results than likelihood-based methods. With the rise of Large Language Models (LLMs), preference-based alignment techniques have gained attention for their potential to enhance translation quality by optimizing model weights directly on preferences induced by quality estimators. This study focuses on Contrastive Preference Optimization (CPO) and conducts extensive experiments to evaluate the impact of preference-based alignment on translation quality. Our findings indicate that while CPO consistently outperforms Supervised Fine-Tuning (SFT) on high-quality data with regard to the alignment metric, it may lead to instability across downstream evaluation metrics, particularly between neural and lexical ones. Additionally, we demonstrate that relying solely on the base model for generating candidate translations achieves performance comparable to using multiple external systems, while ensuring better consistency across downstream metrics.
Despite significant progress in model editing methods, their application in real-world scenarios remains challenging as they often cause large language models (LLMs) to collapse. Among them, ROME is particularly concerning, as it could disrupt LLMs with only a single edit. In this paper, we study the root causes of such collapse. Through extensive analysis, we identify two primary factors that contribute to the collapse: i) inconsistent handling of prefixed and unprefixed keys in the parameter update equation may result in very small denominators, causing excessively large parameter updates; ii) the subject of collapse cases is usually the first token, whose unprefixed key distribution significantly differs from the prefixed key distribution in autoregressive transformers, causing the aforementioned issue to materialize. To validate our findings, we propose a simple yet effective approach: uniformly using prefixed keys during editing phase and adding prefixes during testing phase to ensure the consistency between training and testing. The experimental results show that the proposed solution can prevent model collapse while maintaining the effectiveness of the edits.
This study explores the proactive ability of LLMs to seek user support. We propose metrics to evaluate the trade-off between performance improvements and user burden, and investigate whether LLMs can determine when to request help under varying information availability. Our experiments show that without external feedback, many LLMs struggle to recognize their need for user support. The findings highlight the importance of external signals and provide insights for future research on improving support-seeking strategies. Source code: https://github.com/appier-research/i-need-help
This paper proposes a novel method for constructing instruction-tuned large language models (LLMs) for finance without instruction data. Traditionally, developing such domain-specific LLMs has been resource-intensive, requiring a large dataset and significant computational power for continual pretraining and instruction tuning. Our study proposes a simpler approach that combines domain-specific continual pretraining with model merging. Given that general-purpose pretrained LLMs and their instruction-tuned LLMs are often publicly available, they can be leveraged to obtain the necessary instruction task vector. By merging this with a domain-specific pretrained vector, we can effectively create instruction-tuned LLMs for finance without additional instruction data. Our process involves two steps: first, we perform continual pretraining on financial data; second, we merge the instruction-tuned vector with the domain-specific pretrained vector. Our experiments demonstrate the successful construction of instruction-tuned LLMs for finance. One major advantage of our method is that the instruction-tuned and domain-specific pretrained vectors are nearly independent. This independence makes our approach highly effective. The Japanese financial instruction-tuned LLMs we developed in this study are available at https://huggingface.co/pfnet/nekomata-14b-pfn-qfin-inst-merge.
This study explores the potential of large language models (LLMs) for identifying and examining intertextual relationships within biblical, Koine Greek texts. By evaluating the performance of LLMs on various intertextuality scenarios the study demonstrates that these models can detect direct quotations, allusions, and echoes between texts. The LLM's ability to generate novel intertextual observations and connections highlights its potential to uncover new insights. However, the model also struggles with long query passages and the inclusion of false intertextual dependences, emphasizing the importance of expert evaluation. The expert-in-the-loop methodology presented offers a scalable approach for intertextual research into the complex web of intertextuality within and beyond the biblical corpus.
Continual pretraining of large language models on domain-specific data has been proposed to enhance performance on downstream tasks. In astronomy, the previous absence of astronomy-focused benchmarks has hindered objective evaluation of these specialized LLM models. Leveraging a recent initiative to curate high-quality astronomical MCQs, this study aims to quantitatively assess specialized LLMs in astronomy. We find that the previously released AstroLLaMA series, based on LLaMA-2-7B, underperforms compared to the base model. We demonstrate that this performance degradation can be partially mitigated by utilizing high-quality data for continual pretraining, such as summarized text from arXiv. Despite the observed catastrophic forgetting in smaller models, our results indicate that continual pretraining on the 70B model can yield significant improvements. However, the current supervised fine-tuning dataset still constrains the performance of instruct models. In conjunction with this study, we introduce a new set of models, AstroLLaMA-3-8B and AstroLLaMA-2-70B, building upon the previous AstroLLaMA series.
Brain-related research topics in artificial intelligence have recently gained popularity, particularly due to the expansion of what multimodal architectures can do from computer vision to natural language processing. Our main goal in this work is to explore the possibilities and limitations of these architectures in spoken text decoding from non-invasive fMRI recordings. Contrary to vision and textual data, fMRI data represent a complex modality due to the variety of brain scanners, which implies (i) the variety of the recorded signal formats, (ii) the low resolution and noise of the raw signals, and (iii) the scarcity of pretrained models that can be leveraged as foundation models for generative learning. These points make the problem of the non-invasive decoding of text from fMRI recordings very challenging. In this paper, we propose and end-to-end multimodal LLM for decoding spoken text from fMRI signals. The proposed architecture is founded on (i) an encoder derived from a specific transformer incorporating an augmented embedding layer for the encoder and a better-adjusted attention mechanism than that present in the state of the art, and (ii) a frozen large language model adapted to align the embedding of the input text and the encoded embedding of brain activity to decode the output text. A benchmark in performed on a corpus consisting of a set of interactions human-human and human-robot interactions where fMRI and conversational signals are recorded synchronously. The obtained results are very promising, as our proposal outperforms the evaluated models, and is able to generate text capturing more accurate semantics present in the ground truth. The implementation code is provided in https://github.com/Hmamouche/brain_decode.
The hallucination problem of Large Language Models (LLMs) significantly limits their reliability and trustworthiness. Humans have a self-awareness process that allows us to recognize what we don't know when faced with queries. Inspired by this, our paper investigates whether LLMs can estimate their own hallucination risk before response generation. We analyze the internal mechanisms of LLMs broadly both in terms of training data sources and across 15 diverse Natural Language Generation (NLG) tasks, spanning over 700 datasets. Our empirical analysis reveals two key insights: (1) LLM internal states indicate whether they have seen the query in training data or not; and (2) LLM internal states show they are likely to hallucinate or not regarding the query. Our study explores particular neurons, activation layers, and tokens that play a crucial role in the LLM perception of uncertainty and hallucination risk. By a probing estimator, we leverage LLM self-assessment, achieving an average hallucination estimation accuracy of 84.32\% at run time.
Large Language Models (LLMs) have exhibited exceptional performance across a broad range of tasks and domains. However, they still encounter difficulties in solving mathematical problems due to the rigorous and logical nature of mathematics. Previous studies have employed techniques such as supervised fine-tuning (SFT), prompt engineering, and search-based methods to improve the mathematical problem-solving abilities of LLMs. Despite these efforts, their performance remains suboptimal and demands substantial computational resources. To address this issue, we propose a novel approach, BEATS, to enhance mathematical problem-solving abilities. Our method leverages newly designed prompts that guide the model to iteratively rewrite, advance by one step, and generate answers based on previous steps. Additionally, we introduce a new back-verification technique that uses LLMs to validate the correctness of the generated answers. Furthermore, we employ a pruning tree search to optimize search time while achieving strong performance. Notably, our method improves Qwen2-7b-Instruct's score from 36.94 to 61.52, outperforming GPT4's 42.5 on the MATH benchmark.
Ensembling multiple models has always been an effective approach to push the limits of existing performance and is widely used in classification tasks by simply averaging the classification probability vectors from multiple classifiers to achieve better accuracy. However, in the thriving open-source Large Language Model (LLM) community, ensembling methods are rare and typically limited to ensembling the full-text outputs of LLMs, such as selecting the best output using a ranker, which leads to underutilization of token-level probability information. In this paper, we treat the Generation of each token by LLMs as a Classification (GaC) for ensembling. This approach fully exploits the probability information at each generation step and better prevents LLMs from producing early incorrect tokens that lead to snowballing errors. In experiments, we ensemble state-of-the-art LLMs on several benchmarks, including exams, mathematics and reasoning, and observe that our method breaks the existing community performance ceiling. Furthermore, we observed that most of the tokens in the answer are simple and do not affect the correctness of the final answer. Therefore, we also experimented with ensembling only key tokens, and the results showed better performance with lower latency across benchmarks.
Traditional methods for making software deployment decisions in the automotive industry typically rely on manual analysis of tabular software test data. These methods often lead to higher costs and delays in the software release cycle due to their labor-intensive nature. Large Language Models (LLMs) present a promising solution to these challenges. However, their application generally demands multiple rounds of human-driven prompt engineering, which limits their practical deployment, particularly for industrial end-users who need reliable and efficient results. In this paper, we propose GoNoGo, an LLM agent system designed to streamline automotive software deployment while meeting both functional requirements and practical industrial constraints. Unlike previous systems, GoNoGo is specifically tailored to address domain-specific and risk-sensitive systems. We evaluate GoNoGo's performance across different task difficulties using zero-shot and few-shot examples taken from industrial practice. Our results show that GoNoGo achieves a 100% success rate for tasks up to Level 2 difficulty with 3-shot examples, and maintains high performance even for more complex tasks. We find that GoNoGo effectively automates decision-making for simpler tasks, significantly reducing the need for manual intervention. In summary, GoNoGo represents an efficient and user-friendly LLM-based solution currently employed in our industrial partner's company to assist with software release decision-making, supporting more informed and timely decisions in the release process for risk-sensitive vehicle systems.
Despite the success of large language models (LLMs) in natural language generation, much evidence shows that LLMs may produce incorrect or nonsensical text. This limitation highlights the importance of discerning when to trust LLMs, especially in safety-critical domains. Existing methods often express reliability by confidence level, however, their effectiveness is limited by the lack of objective guidance. To address this, we propose CONfidence-Quality-ORDer-preserving alignment approach (CONQORD), which leverages reinforcement learning guided by a tailored dual-component reward function. This function integrates quality reward and order-preserving alignment reward functions. Specifically, the order-preserving reward incentivizes the model to verbalize greater confidence for responses of higher quality to align the order of confidence and quality. Experiments demonstrate that CONQORD significantly improves the alignment performance between confidence and response accuracy, without causing over-cautious. Furthermore, the aligned confidence provided by CONQORD informs when to trust LLMs, and acts as a determinant for initiating the retrieval process of external knowledge. Aligning confidence with response quality ensures more transparent and reliable responses, providing better trustworthiness.
Recent progress with LLM-based agents has shown promising results across various tasks. However, their use in answering questions from knowledge bases remains largely unexplored. Implementing a KBQA system using traditional methods is challenging due to the shortage of task-specific training data and the complexity of creating task-focused model structures. In this paper, we present Triad, a unified framework that utilizes an LLM-based agent with three roles for KBQA tasks. The agent is assigned three roles to tackle different KBQA subtasks: agent as a generalist for mastering various subtasks, as a decision maker for the selection of candidates, and as an advisor for answering questions with knowledge. Our KBQA framework is executed in four phases, involving the collaboration of the agent's multiple roles. We evaluated the performance of our framework using three benchmark datasets, and the results show that our framework outperforms state-of-the-art systems on the LC-QuAD and YAGO-QA benchmarks, yielding F1 scores of 11.8% and 20.7%, respectively.
Recent advancements in large language models (LLMs) have shown promising results in multilingual translation even with limited bilingual supervision. The major challenges are catastrophic forgetting and parameter interference for finetuning LLMs when provided parallel training data. To address these challenges, we propose LANDeRMT, a \textbf{L}anguage-\textbf{A}ware \textbf{N}euron \textbf{De}tecting and \textbf{R}outing framework that selectively finetunes LLMs to \textbf{M}achine \textbf{T}ranslation with diverse translation training data. In LANDeRMT, we evaluate the awareness of neurons to MT tasks and categorize them into language-general and language-specific neurons. This categorization enables selective parameter updates during finetuning, mitigating parameter interference and catastrophic forgetting issues. For the detected neurons, we further propose a conditional awareness-based routing mechanism to dynamically adjust language-general and language-specific capacity within LLMs, guided by translation signals. Experimental results demonstrate that the proposed LANDeRMT is very effective in learning translation knowledge, significantly improving translation quality over various strong baselines for multiple language pairs.
Speech Language Models (SLMs) have demonstrated impressive performance on speech translation tasks. However, existing research primarily focuses on direct instruction fine-tuning and often overlooks the inherent reasoning capabilities of SLMs. In this paper, we introduce a three-stage training framework designed to activate the chain-of-thought (CoT) capabilities of SLMs. We propose CoT-ST, a speech translation model that utilizes multimodal CoT to decompose speech translation into sequential steps of speech recognition and translation. We validated the effectiveness of our method on two datasets: the CoVoST-2 dataset and MuST-C dataset. The experimental results demonstrate that CoT-ST outperforms previous state-of-the-art methods, achieving higher BLEU scores (CoVoST-2 en-ja: 30.5->30.8, en-zh: 45.2->47.7, MuST-C en-zh: 19.6->21.2). This work is open sourced at https://github.com/X-LANCE/SLAM-LLM/tree/main/examples/st_covost2 .
Reasoning is a fundamental component of language understanding. Recent prompting techniques, such as chain of thought, have consistently improved LLMs' performance on various reasoning tasks. Nevertheless, there is still little understanding of what triggers reasoning abilities in LLMs in the inference stage. In this paper, we introduce code prompting, a chain of prompts that transforms a natural language problem into code and directly prompts the LLM using the generated code without resorting to external code execution. We hypothesize that code prompts can elicit certain reasoning capabilities of LLMs trained on text and code and utilize the proposed method to improve conditional reasoning, the ability to infer different conclusions depending on the fulfillment of certain conditions. We find that code prompting exhibits a high-performance boost for multiple LLMs (up to 22.52 percentage points on GPT 3.5, 7.75 on Mixtral, and 16.78 on Mistral) across multiple conditional reasoning datasets. We then conduct comprehensive experiments to understand how code prompts trigger reasoning abilities and which capabilities are elicited in the underlying models. Our analysis of GPT 3.5 reveals that the code formatting of the input problem is essential for performance improvement. Furthermore, code prompts improve sample efficiency of in-context learning and facilitate state tracking of variables or entities.
The impact of social media on critical issues such as echo chambers needs to be addressed, as these phenomena can have disruptive consequences for our society. Traditional research often oversimplifies emotional tendencies and opinion evolution into numbers and formulas, neglecting that news and communication are conveyed through text, which limits these approaches. Hence, in this work, we propose an LLM-based simulation for the social opinion network to evaluate and counter polarization phenomena. We first construct three typical network structures to simulate different characteristics of social interactions. Then, agents interact based on recommendation algorithms and update their strategies through reasoning and analysis. By comparing these interactions with the classic Bounded Confidence Model (BCM), the Friedkin Johnsen (FJ) model, and using echo chamber-related indices, we demonstrate the effectiveness of our framework in simulating opinion dynamics and reproducing phenomena such as opinion polarization and echo chambers. We propose two mitigation methods, active and passive nudges, that can help reduce echo chambers, specifically within language-based simulations. We hope our work will offer valuable insights and guidance for social polarization mitigation.
Emphasis is a crucial component in human communication, which indicates the speaker's intention and implication beyond pure text in dialogue. While Large Language Models (LLMs) have revolutionized natural language processing, their ability to understand emphasis in dialogue remains unclear. This paper introduces Emphasized-Talk, a benchmark with emphasis-annotated dialogue samples capturing the implications of emphasis. We evaluate various LLMs, both open-source and commercial, to measure their performance in understanding emphasis. Additionally, we propose an automatic evaluation pipeline using GPT-4, which achieves a high correlation with human rating. Our findings reveal that although commercial LLMs generally perform better, there is still significant room for improvement in comprehending emphasized sentences.
This study introduces a simple yet effective method for identifying similar data points across non-free text domains, such as tabular and image data, using Large Language Models (LLMs). Our two-step approach involves data point summarization and hidden state extraction. Initially, data is condensed via summarization using an LLM, reducing complexity and highlighting essential information in sentences. Subsequently, the summarization sentences are fed through another LLM to extract hidden states, serving as compact, feature-rich representations. This approach leverages the advanced comprehension and generative capabilities of LLMs, offering a scalable and efficient strategy for similarity identification across diverse datasets. We demonstrate the effectiveness of our method in identifying similar data points on multiple datasets. Additionally, our approach enables non-technical domain experts, such as fraud investigators or marketing operators, to quickly identify similar data points tailored to specific scenarios, demonstrating its utility in practical applications. In general, our results open new avenues for leveraging LLMs in data analysis across various domains
Extremely low-resource (XLR) languages lack substantial corpora for training NLP models, motivating the use of all available resources such as dictionaries and grammar books. Machine Translation from One Book (Tanzer et al., 2024) suggests prompting long-context LLMs with one grammar book enables English-Kalamang translation, an unseen XLR language - a noteworthy case of linguistic knowledge helping an NLP task. We investigate whether the book's grammatical explanations or its parallel examples are most effective for learning XLR translation, finding almost all improvement stems from the parallel examples. Further, we find similar results for Nepali, a seen low-resource language, and achieve performance comparable to an LLM with a grammar book by simply fine-tuning an encoder-decoder translation model. We then investigate where grammar books help by testing two linguistic tasks, grammaticality judgment and gloss prediction, and we explore what kind of grammatical knowledge helps by introducing a typological feature prompt that achieves leading results on these more relevant tasks. We thus emphasise the importance of task-appropriate data for XLR languages: parallel examples for translation, and grammatical data for linguistic tasks. As we find no evidence that long-context LLMs can make effective use of grammatical explanations for XLR translation, we suggest data collection for multilingual XLR tasks such as translation is best focused on parallel data over linguistic description.
Cancer patients often struggle to transition swiftly to treatment due to limited institutional resources, lack of sophisticated professional guidance, and low health literacy. The emergence of Large Language Models (LLMs) offers new opportunities for such patients to access the wealth of existing patient education materials. The current paper presents the development process for an LLM-based chatbot focused on prostate cancer education, including needs assessment, co-design, and usability studies. The resulting application, MedEduChat, integrates with patients' electronic health record data and features a closed-domain, semi-structured, patient-centered approach to address real-world needs. This paper contributes to the growing field of patient-LLM interaction by demonstrating the potential of LLM-based chatbots to enhance prostate cancer patient education and by offering co-design guidelines for future LLM-based healthcare downstream applications.
Important variables of processes are, in many occasions, categorical, i.e. names or labels representing, e.g. categories of inputs, or types of reactors or a sequence of steps. In this work, we use Large Language Models (LLMs) to derive embeddings of such inputs that represent their actual meaning, or reflect the ``distances" between categories, i.e. how similar or dissimilar they are. This is a marked difference from the current standard practice of using binary, or one-hot encoding to replace categorical variables with sequences of ones and zeros. Combined with dimensionality reduction techniques, either linear such as Principal Components Analysis (PCA), or nonlinear such as Uniform Manifold Approximation and Projection (UMAP), the proposed approach leads to a \textit{meaningful}, low-dimensional feature space. The significance of obtaining meaningful embeddings is illustrated in the context of an industrial coating process for cutting tools that includes both numerical and categorical inputs. The proposed approach enables feature importance which is a marked improvement compared to the current state-of-the-art (SotA) in the encoding of categorical variables.
Forecasting weather and climate events is crucial for making appropriate measures to mitigate environmental hazards and minimize associated losses. Previous research on environmental forecasting focuses on predicting numerical meteorological variables related to closed-set events rather than forecasting open-set events directly, which limits the comprehensiveness of event forecasting. We propose Weather and Climate Event Forecasting (WCEF), a new task that leverages meteorological raster data and textual event data to predict potential weather and climate events. However, due to difficulties in aligning multimodal data and the lack of sufficient supervised datasets, this task is challenging to accomplish. Therefore, we first propose a framework to align historical meteorological data with past weather and climate events using the large language model (LLM). In this framework, we construct a knowledge graph by using LLM to extract information about weather and climate events from a corpus of over 41k highly environment-focused news articles. Subsequently, we mapped these events with meteorological raster data, creating a supervised dataset, which is the largest and most novel for LLM tuning on the WCEF task. Finally, we introduced our aligned models, CLLMate (LLM for climate), a multimodal LLM to forecast weather and climate events using meteorological raster data. In evaluating CLLMate, we conducted extensive experiments. The results indicate that CLLMate surpasses both the baselines and other multimodal LLMs, showcasing the potential of utilizing LLM to align weather and climate events with meteorological data and highlighting the promising future for research on the WCEF task.
With the emergence of widely available powerful LLMs, disinformation generated by large Language Models (LLMs) has become a major concern. Historically, LLM detectors have been touted as a solution, but their effectiveness in the real world is still to be proven. In this paper, we focus on an important setting in information operations -- short news-like posts generated by moderately sophisticated attackers. We demonstrate that existing LLM detectors, whether zero-shot or purpose-trained, are not ready for real-world use in that setting. All tested zero-shot detectors perform inconsistently with prior benchmarks and are highly vulnerable to sampling temperature increase, a trivial attack absent from recent benchmarks. A purpose-trained detector generalizing across LLMs and unseen attacks can be developed, but it fails to generalize to new human-written texts. We argue that the former indicates domain-specific benchmarking is needed, while the latter suggests a trade-off between the adversarial evasion resilience and overfitting to the reference human text, with both needing evaluation in benchmarks and currently absent. We believe this suggests a re-consideration of current LLM detector benchmarking approaches and provides a dynamically extensible benchmark to allow it (https://github.com/Reliable-Information-Lab-HEVS/benchmark_llm_texts_detection).
Large language models (LLMs) have been shown to face hallucination issues due to the data they trained on often containing human bias; whether this is reflected in the decision-making process of LLM Agents remains under-explored. As LLM Agents are increasingly employed in intricate social environments, a pressing and natural question emerges: Can we utilize LLM Agents' systematic hallucinations to mirror human cognitive biases, thus exhibiting irrational social intelligence? In this paper, we probe the irrational behavior among contemporary LLM Agents by melding practical social science experiments with theoretical insights. Specifically, We propose CogMir, an open-ended Multi-LLM Agents framework that utilizes hallucination properties to assess and enhance LLM Agents' social intelligence through cognitive biases. Experimental results on CogMir subsets show that LLM Agents and humans exhibit high consistency in irrational and prosocial decision-making under uncertain conditions, underscoring the prosociality of LLM Agents as social entities and highlighting the significance of hallucination properties. Additionally, the CogMir framework demonstrates its potential as a valuable platform for encouraging more research into the social intelligence of LLM Agents.
Vision-and-Language Navigation (VLN) tasks require an agent to follow textual instructions to navigate through 3D environments. Traditional approaches use supervised learning methods, relying heavily on domain-specific datasets to train VLN models. Recent methods try to utilize closed-source large language models (LLMs) like GPT-4 to solve VLN tasks in zero-shot manners, but face challenges related to expensive token costs and potential data breaches in real-world applications. In this work, we introduce Open-Nav, a novel study that explores open-source LLMs for zero-shot VLN in the continuous environment. Open-Nav employs a spatial-temporal chain-of-thought (CoT) reasoning approach to break down tasks into instruction comprehension, progress estimation, and decision-making. It enhances scene perceptions with fine-grained object and spatial knowledge to improve LLM's reasoning in navigation. Our extensive experiments in both simulated and real-world environments demonstrate that Open-Nav achieves competitive performance compared to using closed-source LLMs.
We propose a novel framework that leverages Visual Question Answering (VQA) models to automate the evaluation of LLM-generated data visualizations. Traditional evaluation methods often rely on human judgment, which is costly and unscalable, or focus solely on data accuracy, neglecting the effectiveness of visual communication. By employing VQA models, we assess data representation quality and the general communicative clarity of charts. Experiments were conducted using two leading VQA benchmark datasets, ChartQA and PlotQA, with visualizations generated by OpenAI's GPT-3.5 Turbo and Meta's Llama 3.1 70B-Instruct models. Our results indicate that LLM-generated charts do not match the accuracy of the original non-LLM-generated charts based on VQA performance measures. Moreover, while our results demonstrate that few-shot prompting significantly boosts the accuracy of chart generation, considerable progress remains to be made before LLMs can fully match the precision of human-generated graphs. This underscores the importance of our work, which expedites the research process by enabling rapid iteration without the need for human annotation, thus accelerating advancements in this field.
Retrieval-augmented generation (RAG) has received much attention for Open-domain question-answering (ODQA) tasks as a means to compensate for the parametric knowledge of large language models (LLMs). While previous approaches focused on processing retrieved passages to remove irrelevant context, they still rely heavily on the quality of retrieved passages which can degrade if the question is ambiguous or complex. In this paper, we propose a simple yet efficient method called question and passage augmentation (QPaug) via LLMs for open-domain QA. QPaug first decomposes the original questions into multiple-step sub-questions. By augmenting the original question with detailed sub-questions and planning, we are able to make the query more specific on what needs to be retrieved, improving the retrieval performance. In addition, to compensate for the case where the retrieved passages contain distracting information or divided opinions, we augment the retrieved passages with self-generated passages by LLMs to guide the answer extraction. Experimental results show that QPaug outperforms the previous state-of-the-art and achieves significant performance gain over existing RAG methods. The source code is available at \url{https://github.com/kmswin1/QPaug}.
The sudden emergence of large language models (LLMs) such as ChatGPT has had a disruptive impact throughout the computing education community. LLMs have been shown to excel at producing correct code to CS1 and CS2 problems, and can even act as friendly assistants to students learning how to code. Recent work shows that LLMs demonstrate unequivocally superior results in being able to explain and resolve compiler error messages -- for decades, one of the most frustrating parts of learning how to code. However, LLM-generated error message explanations have only been assessed by expert programmers in artificial conditions. This work sought to understand how novice programmers resolve programming error messages (PEMs) in a more realistic scenario. We ran a within-subjects study with $n$ = 106 participants in which students were tasked to fix six buggy C programs. For each program, participants were randomly assigned to fix the problem using either a stock compiler error message, an expert-handwritten error message, or an error message explanation generated by GPT-4. Despite promising evidence on synthetic benchmarks, we found that GPT-4 generated error messages outperformed conventional compiler error messages in only 1 of the 6 tasks, measured by students' time-to-fix each problem. Handwritten explanations still outperform LLM and conventional error messages, both on objective and subjective measures.
The ability to understand causality significantly impacts the competence of large language models (LLMs) in output explanation and counterfactual reasoning, as causality reveals the underlying data distribution. However, the lack of a comprehensive benchmark currently limits the evaluation of LLMs' causal learning capabilities. To fill this gap, this paper develops CausalBench based on data from the causal research community, enabling comparative evaluations of LLMs against traditional causal learning algorithms. To provide a comprehensive investigation, we offer three tasks of varying difficulties, including correlation, causal skeleton, and causality identification. Evaluations of 19 leading LLMs reveal that, while closed-source LLMs show potential for simple causal relationships, they significantly lag behind traditional algorithms on larger-scale networks ($>50$ nodes). Specifically, LLMs struggle with collider structures but excel at chain structures, especially at long-chain causality analogous to Chains-of-Thought techniques. This supports the current prompt approaches while suggesting directions to enhance LLMs' causal reasoning capability. Furthermore, CausalBench incorporates background knowledge and training data into prompts to thoroughly unlock LLMs' text-comprehension ability during evaluation, whose findings indicate that, LLM understand causality through semantic associations with distinct entities, rather than directly from contextual information or numerical distributions.
Assessing the performance of systems to classify Multi-Party Conversations (MPC) is challenging due to the interconnection between linguistic and structural characteristics of conversations. Conventional evaluation methods often overlook variances in model behavior across different levels of structural complexity on interaction graphs. In this work, we propose a methodological pipeline to investigate model performance across specific structural attributes of conversations. As a proof of concept we focus on Response Selection and Addressee Recognition tasks, to diagnose model weaknesses. To this end, we extract representative diagnostic subdatasets with a fixed number of users and a good structural variety from a large and open corpus of online MPCs. We further frame our work in terms of data minimization, avoiding the use of original usernames to preserve privacy, and propose alternatives to using original text messages. Results show that response selection relies more on the textual content of conversations, while addressee recognition requires capturing their structural dimension. Using an LLM in a zero-shot setting, we further highlight how sensitivity to prompt variations is task-dependent.
Human intention-based systems enable robots to perceive and interpret user actions to interact with humans and adapt to their behavior proactively. Therefore, intention prediction is pivotal in creating a natural interaction with social robots in human-designed environments. In this paper, we examine using Large Language Models (LLMs) to infer human intention in a collaborative object categorization task with a physical robot. We propose a novel multimodal approach that integrates user non-verbal cues, like hand gestures, body poses, and facial expressions, with environment states and user verbal cues to predict user intentions in a hierarchical architecture. Our evaluation of five LLMs shows the potential for reasoning about verbal and non-verbal user cues, leveraging their context-understanding and real-world knowledge to support intention prediction while collaborating on a task with a social robot.
Large language models (LLMs) provide powerful means to leverage prior knowledge for predictive modeling when data is limited. In this work, we demonstrate how LLMs can use their compressed world knowledge to generate intrinsically interpretable machine learning models, i.e., decision trees, without any training data. We find that these zero-shot decision trees can surpass data-driven trees on some small-sized tabular datasets and that embeddings derived from these trees perform on par with data-driven tree-based embeddings on average. Our knowledge-driven decision tree induction and embedding approaches therefore serve as strong new baselines for data-driven machine learning methods in the low-data regime.
More Effective LLM Compressed Tokens with Uniformly Spread Position Identifiers and Compression Loss
Compressing Transformer inputs into compressd tokens allows running LLMs with improved speed and cost efficiency. Based on the compression method ICAE, we carefully examine the position identifier choices for compressed tokens and also propose a new compression loss. We demonstrate empirically that our proposed methods achieve significantly higher compression ratios (15x compared to 4x for ICAE), while being able to attain comparable reconstruction performance.
Many recent studies have investigated social biases in LLMs but brand bias has received little attention. This research examines the biases exhibited by LLMs towards different brands, a significant concern given the widespread use of LLMs in affected use cases such as product recommendation and market analysis. Biased models may perpetuate societal inequalities, unfairly favoring established global brands while marginalizing local ones. Using a curated dataset across four brand categories, we probe the behavior of LLMs in this space. We find a consistent pattern of bias in this space -- both in terms of disproportionately associating global brands with positive attributes and disproportionately recommending luxury gifts for individuals in high-income countries. We also find LLMs are subject to country-of-origin effects which may boost local brand preference in LLM outputs in specific contexts.
While generalization over tasks from easy to hard is crucial to profile language models (LLMs), the datasets with fine-grained difficulty annotations for each problem across a broad range of complexity are still blank. Aiming to address this limitation, we present Easy2Hard-Bench, a consistently formatted collection of 6 benchmark datasets spanning various domains, such as mathematics and programming problems, chess puzzles, and reasoning questions. Each problem within these datasets is annotated with numerical difficulty scores. To systematically estimate problem difficulties, we collect abundant performance data on attempts to each problem by humans in the real world or LLMs on the prominent leaderboard. Leveraging the rich performance data, we apply well-established difficulty ranking systems, such as Item Response Theory (IRT) and Glicko-2 models, to uniformly assign numerical difficulty scores to problems. Moreover, datasets in Easy2Hard-Bench distinguish themselves from previous collections by a higher proportion of challenging problems. Through extensive experiments with six state-of-the-art LLMs, we provide a comprehensive analysis of their performance and generalization capabilities across varying levels of difficulty, with the aim of inspiring future research in LLM generalization. The datasets are available at https://huggingface.co/datasets/furonghuang-lab/Easy2Hard-Bench.
Low-Rank Adaptation (LoRA), as a representative Parameter-Efficient Fine-Tuning (PEFT)method, significantly enhances the training efficiency by updating only a small portion of the weights in Large Language Models (LLMs). Recently, weight-only quantization techniques have also been applied to LoRA methods to reduce the memory footprint of fine-tuning. However, applying weight-activation quantization to the LoRA pipeline is under-explored, and we observe substantial performance degradation primarily due to the presence of activation outliers. In this work, we propose RoLoRA, the first LoRA-based scheme for effective weight-activation quantization. RoLoRA utilizes rotation for outlier elimination and proposes rotation-aware fine-tuning to preserve the outlier-free characteristics in rotated LLMs. Experimental results show RoLoRA consistently improves low-bit LoRA convergence and post-training quantization robustness in weight-activation settings. We evaluate RoLoRA across LLaMA2-7B/13B, LLaMA3-8B models, achieving up to 29.5% absolute accuracy gain of 4-bit weight-activation quantized LLaMA2- 13B on commonsense reasoning tasks compared to LoRA baseline. We further demonstrate its effectiveness on Large Multimodal Models (LLaVA-1.5-7B). Codes are available at https://github.com/HuangOwen/RoLoRA
Performing building information modeling (BIM) tasks is a complex process that imposes a steep learning curve and a heavy cognitive load due to the necessity of remembering sequences of numerous commands. With the rapid advancement of large language models (LLMs), it is foreseeable that BIM tasks, including querying and managing BIM data, 4D and 5D BIM, design compliance checking, or authoring a design, using written or spoken natural language (i.e., text-to-BIM or speech-to-BIM), will soon supplant traditional graphical user interfaces. This paper proposes a generalized LLM-augmented BIM framework to expedite the development of LLM-enhanced BIM applications by providing a step-by-step development process. The proposed framework consists of six steps: interpret-fill-match-structure-execute-check. The paper demonstrates the applicability of the proposed framework through implementing a speech-to-BIM application, NADIA-S (Natural-language-based Architectural Detailing through Interaction with Artificial Intelligence via Speech), using exterior wall detailing as an example.
In-basket message interactions play a crucial role in physician-patient communication, occurring during all phases (pre-, during, and post) of a patient's care journey. However, responding to these patients' inquiries has become a significant burden on healthcare workflows, consuming considerable time for clinical care teams. To address this, we introduce RadOnc-GPT, a specialized Large Language Model (LLM) powered by GPT-4 that has been designed with a focus on radiotherapeutic treatment of prostate cancer with advanced prompt engineering, and specifically designed to assist in generating responses. We integrated RadOnc-GPT with patient electronic health records (EHR) from both the hospital-wide EHR database and an internal, radiation-oncology-specific database. RadOnc-GPT was evaluated on 158 previously recorded in-basket message interactions. Quantitative natural language processing (NLP) analysis and two grading studies with clinicians and nurses were used to assess RadOnc-GPT's responses. Our findings indicate that RadOnc-GPT slightly outperformed the clinical care team in "Clarity" and "Empathy," while achieving comparable scores in "Completeness" and "Correctness." RadOnc-GPT is estimated to save 5.2 minutes per message for nurses and 2.4 minutes for clinicians, from reading the inquiry to sending the response. Employing RadOnc-GPT for in-basket message draft generation has the potential to alleviate the workload of clinical care teams and reduce healthcare costs by producing high-quality, timely responses.
Whether a large language model policy is an explicit constitution or an implicit reward model, it is challenging to assess coverage over the unbounded set of real-world situations that a policy must contend with. We introduce an AI policy design process inspired by mapmaking, which has developed tactics for visualizing and iterating on maps even when full coverage is not possible. With Policy Projector, policy designers can survey the landscape of model input-output pairs, define custom regions (e.g., "violence"), and navigate these regions with rules that can be applied to LLM outputs (e.g., if output contains "violence" and "graphic details," then rewrite without "graphic details"). Policy Projector supports interactive policy authoring using LLM classification and steering and a map visualization reflecting the policy designer's work. In an evaluation with 12 AI safety experts, our system helps policy designers to address problematic model behaviors extending beyond an existing, comprehensive harm taxonomy.
Background: Quantum computing is a rapidly growing new programming paradigm that brings significant changes to the design and implementation of algorithms. Understanding quantum algorithms requires knowledge of physics and mathematics, which can be challenging for software developers. Aims: In this work, we provide a first analysis of how LLMs can support developers' understanding of quantum code. Method: We empirically analyse and compare the quality of explanations provided by three widely adopted LLMs (Gpt3.5, Llama2, and Tinyllama) using two different human-written prompt styles for seven state-of-the-art quantum algorithms. We also analyse how consistent LLM explanations are over multiple rounds and how LLMs can improve existing descriptions of quantum algorithms. Results: Llama2 provides the highest quality explanations from scratch, while Gpt3.5 emerged as the LLM best suited to improve existing explanations. In addition, we show that adding a small amount of context to the prompt significantly improves the quality of explanations. Finally, we observe how explanations are qualitatively and syntactically consistent over multiple rounds. Conclusions: This work highlights promising results, and opens challenges for future research in the field of LLMs for quantum code explanation. Future work includes refining the methods through prompt optimisation and parsing of quantum code explanations, as well as carrying out a systematic assessment of the quality of explanations.
The recently developed retrieval-augmented generation (RAG) technology has enabled the efficient construction of domain-specific applications. However, it also has limitations, including the gap between vector similarity and the relevance of knowledge reasoning, as well as insensitivity to knowledge logic, such as numerical values, temporal relations, expert rules, and others, which hinder the effectiveness of professional knowledge services. In this work, we introduce a professional domain knowledge service framework called Knowledge Augmented Generation (KAG). KAG is designed to address the aforementioned challenges with the motivation of making full use of the advantages of knowledge graph(KG) and vector retrieval, and to improve generation and reasoning performance by bidirectionally enhancing large language models (LLMs) and KGs through five key aspects: (1) LLM-friendly knowledge representation, (2) mutual-indexing between knowledge graphs and original chunks, (3) logical-form-guided hybrid reasoning engine, (4) knowledge alignment with semantic reasoning, and (5) model capability enhancement for KAG. We compared KAG with existing RAG methods in multihop question answering and found that it significantly outperforms state-of-theart methods, achieving a relative improvement of 19.6% on 2wiki and 33.5% on hotpotQA in terms of F1 score. We have successfully applied KAG to two professional knowledge Q&A tasks of Ant Group, including E-Government Q&A and E-Health Q&A, achieving significant improvement in professionalism compared to RAG methods.
Large language models (LLMs) with long-context processing are still challenging because of their implementation complexity, training efficiency and data sparsity. To address this issue, a new paradigm named Online Long-context Processing (OLP) is proposed when we process a document of unlimited length, which typically occurs in the information reception and organization of diverse streaming media such as automated news reporting, live e-commerce, and viral short videos. Moreover, a dilemma was often encountered when we tried to select the most suitable LLM from a large number of LLMs amidst explosive growth aiming for outstanding performance, affordable prices, and short response delays. In view of this, we also develop Role Reinforcement Learning (Role-RL) to automatically deploy different LLMs in their respective roles within the OLP pipeline according to their actual performance. Extensive experiments are conducted on our OLP-MINI dataset and it is found that OLP with Role-RL framework achieves OLP benchmark with an average recall rate of 93.2% and the LLM cost saved by 79.4%. The code and dataset are publicly available at: https://anonymous.4open.science/r/Role-RL.
We initiate a formal investigation into the design and analysis of LLM-based algorithms, i.e. algorithms that contain one or multiple calls of large language models (LLMs) as sub-routines and critically rely on the capabilities of LLMs. While LLM-based algorithms, ranging from basic LLM calls with prompt engineering to complicated LLM-powered agent systems and compound AI systems, have achieved remarkable empirical success, the design and optimization of them have mostly relied on heuristics and trial-and-errors, which is largely due to a lack of formal and analytical study for these algorithms. To fill this gap, we start by identifying the computational-graph representation of LLM-based algorithms, the design principle of task decomposition, and some key abstractions, which then facilitate our formal analysis for the accuracy and efficiency of LLM-based algorithms, despite the black-box nature of LLMs. Through extensive analytical and empirical investigation in a series of case studies, we demonstrate that the proposed framework is broadly applicable to a wide range of scenarios and diverse patterns of LLM-based algorithms, such as parallel, hierarchical and recursive task decomposition. Our proposed framework holds promise for advancing LLM-based algorithms, by revealing the reasons behind curious empirical phenomena, guiding the choices of hyperparameters, predicting the empirical performance of algorithms, and inspiring new algorithm design. To promote further study of LLM-based algorithms, we release our source code at https://github.com/modelscope/agentscope/tree/main/examples/paper_llm_based_algorithm.
The increasing demand for intelligent assistants in human-populated environments has motivated significant research in autonomous robotic systems. Traditional service robots and virtual assistants, however, struggle with real-world task execution due to their limited capacity for dynamic reasoning and interaction, particularly when human collaboration is required. Recent developments in Large Language Models have opened new avenues for improving these systems, enabling more sophisticated reasoning and natural interaction capabilities. In this paper, we introduce AssistantX, an LLM-powered proactive assistant designed to operate autonomously in a physical office environment. Unlike conventional service robots, AssistantX leverages a novel multi-agent architecture, PPDR4X, which provides advanced inference capabilities and comprehensive collaboration awareness. By effectively bridging the gap between virtual operations and physical interactions, AssistantX demonstrates robust performance in managing complex real-world scenarios. Our evaluation highlights the architecture's effectiveness, showing that AssistantX can respond to clear instructions, actively retrieve supplementary information from memory, and proactively seek collaboration from team members to ensure successful task completion. More details and videos can be found at https://assistantx-agent.github.io/AssistantX/.
Training large language models (LLMs) for external tool usage is a rapidly expanding field, with recent research focusing on generating synthetic data to address the shortage of available data. However, the absence of systematic data quality checks poses complications for properly training and testing models. To that end, we propose two approaches for assessing the reliability of data for training LLMs to use external tools. The first approach uses intuitive, human-defined correctness criteria. The second approach uses a model-driven assessment with in-context evaluation. We conduct a thorough evaluation of data quality on two popular benchmarks, followed by an extrinsic evaluation that showcases the impact of data quality on model performance. Our results demonstrate that models trained on high-quality data outperform those trained on unvalidated data, even when trained with a smaller quantity of data. These findings empirically support the significance of assessing and ensuring the reliability of training data for tool-using LLMs.
LLM text decoding is key component for perceived LLM quality. We demonstrate two experiments showing that decoding methods could be improved by manipulation of token probabilities. First, we test few LLM on SummEval summary scoring dataset, to measure reading comprehension. We compare scores from greedy decoding to expected values over the next token distribution. We scale logits by large temperature to increase the entropy of scores. This allows strong improvement of performance on SummEval (in terms of correlations to human judgement). We see improvement from 6-8% to 13-28% for 7B Mistral and from 20%-46% to 37%-56% for Mixtral, beating GPT 4 0314 result on two metrics. Part of the gain seems related to positional bias. Secondly, we use probability-based tree sampling algorithm, to examine all most probable generations for given prompt.
In response to the increasing mental health challenges faced by college students, we sought to understand their perspectives on how AI applications, particularly Large Language Models (LLMs), can be leveraged to enhance their mental well-being. Through pilot interviews with ten diverse students, we explored their opinions on the use of LLMs across five fictional scenarios: General Information Inquiry, Initial Screening, Reshaping Patient-Expert Dynamics, Long-term Care, and Follow-up Care. Our findings revealed that students' acceptance of LLMs varied by scenario, with participants highlighting both potential benefits, such as proactive engagement and personalized follow-up care, and concerns, including limitations in training data and emotional support. These insights inform how AI technology should be designed and implemented to effectively support and enhance students' mental well-being, particularly in scenarios where LLMs can complement traditional methods, while maintaining empathy and respecting individual preferences.
The surge in applications of large language models (LLMs) has prompted concerns about the generation of misleading or fabricated information, known as hallucinations. Therefore, detecting hallucinations has become critical to maintaining trust in LLM-generated content. A primary challenge in learning a truthfulness classifier is the lack of a large amount of labeled truthful and hallucinated data. To address the challenge, we introduce HaloScope, a novel learning framework that leverages the unlabeled LLM generations in the wild for hallucination detection. Such unlabeled data arises freely upon deploying LLMs in the open world, and consists of both truthful and hallucinated information. To harness the unlabeled data, we present an automated membership estimation score for distinguishing between truthful and untruthful generations within unlabeled mixture data, thereby enabling the training of a binary truthfulness classifier on top. Importantly, our framework does not require extra data collection and human annotations, offering strong flexibility and practicality for real-world applications. Extensive experiments show that HaloScope can achieve superior hallucination detection performance, outperforming the competitive rivals by a significant margin. Code is available at https://github.com/deeplearningwisc/haloscope.
Adopting human and large language models (LLM) as judges (a.k.a human- and LLM-as-a-judge) for evaluating the performance of LLMs has recently gained attention. Nonetheless, this approach concurrently introduces potential biases from human and LLMs, questioning the reliability of the evaluation results. In this paper, we propose a novel framework that is free from referencing groundtruth annotations for investigating Misinformation Oversight Bias, Gender Bias, Authority Bias and Beauty Bias on LLM and human judges. We curate a dataset referring to the revised Bloom's Taxonomy and conduct thousands of evaluations. Results show that human and LLM judges are vulnerable to perturbations to various degrees, and that even the cutting-edge judges possess considerable biases. We further exploit these biases to conduct attacks on LLM judges. We hope that our work can notify the community of the bias and vulnerability of human- and LLM-as-a-judge, as well as the urgency of developing robust evaluation systems.
Despite recent advancements in large language models (LLMs), their performance on complex reasoning problems requiring multi-step thinking and combining various skills is still limited. To address this, we propose a novel framework HDFlow for complex reasoning with LLMs that combines fast and slow thinking modes in an adaptive manner. Our approach consists of two key components: 1) a new approach for slow, deliberate reasoning called Dynamic Workflow, which automatically decomposes complex problems into more manageable sub-tasks and dynamically designs a workflow to assemble specialized LLM or symbolic reasoning tools to solve sub-tasks; 2) Hybrid Thinking, a general framework that dynamically combines fast and slow thinking based on problem complexity. Finally, we propose an easy-to-scale method for automatically synthesizing a large-scale dataset of 27K challenging reasoning problems for complex reasoning and a hybrid thinking tuning method that trains smaller LLMs on this dataset to internalize the fast/slow hybrid reasoning strategies. Experiments on four reasoning benchmark datasets demonstrate that our slow thinking with dynamic workflows significantly outperforms Chain-of-Thought, and hybrid thinking achieves the highest accuracy while providing an effective balance between computational efficiency and performance. Fine-tuning using our hybrid thinking approach also significantly boosts the complex reasoning capabilities of open-source language models. The results showcase the promise of slow thinking, dynamic workflows, and hybrid thinking in expanding the frontier of complex problem-solving with LLMs\footnote{Code and data will be released at \url{https://github.com/wenlinyao/HDFlow}.}.
Large Language Models (LLMs) have demonstrated remarkable capabilities in handling long context inputs, but this comes at the cost of increased computational resources and latency. Our research introduces a novel approach for the long context bottleneck to accelerate LLM inference and reduce GPU memory consumption. Our research demonstrates that LLMs can identify relevant tokens in the early layers before generating answers to a query. Leveraging this insight, we propose an algorithm that uses early layers of an LLM as filters to select and compress input tokens, significantly reducing the context length for subsequent processing. Our method, GemFilter, demonstrates substantial improvements in both speed and memory efficiency compared to existing techniques, such as standard attention and SnapKV/H2O. Notably, it achieves a 2.4$\times$ speedup and 30\% reduction in GPU memory usage compared to SOTA methods. Evaluation on the Needle in a Haystack task shows that GemFilter significantly outperforms standard attention, SnapKV and demonstrates comparable performance on the LongBench challenge. GemFilter is simple, training-free, and broadly applicable across different LLMs. Crucially, it provides interpretability by allowing humans to inspect the selected input sequence. These findings not only offer practical benefits for LLM deployment, but also enhance our understanding of LLM internal mechanisms, paving the way for further optimizations in LLM design and inference. Our code is available at \url{https://github.com/SalesforceAIResearch/GemFilter}.
Large Language Models (LLMs) are increasingly integrated into various medical fields, including mental health support systems. However, there is a gap in research regarding the effectiveness of LLMs in non-English mental health support applications. To address this problem, we present a novel multilingual adaptation of widely-used mental health datasets, translated from English into six languages (Greek, Turkish, French, Portuguese, German, and Finnish). This dataset enables a comprehensive evaluation of LLM performance in detecting mental health conditions and assessing their severity across multiple languages. By experimenting with GPT and Llama, we observe considerable variability in performance across languages, despite being evaluated on the same translated dataset. This inconsistency underscores the complexities inherent in multilingual mental health support, where language-specific nuances and mental health data coverage can affect the accuracy of the models. Through comprehensive error analysis, we emphasize the risks of relying exclusively on large language models (LLMs) in medical settings (e.g., their potential to contribute to misdiagnoses). Moreover, our proposed approach offers significant cost savings for multilingual tasks, presenting a major advantage for broad-scale implementation.
Large Language Models (LLMs) have been widely applied in summarization due to their speedy and high-quality text generation. Summarization for sensemaking involves information compression and insight extraction. Human guidance in sensemaking tasks can prioritize and cluster relevant information for LLMs. However, users must translate their cognitive thinking into natural language to communicate with LLMs. Can we use more readable and operable visual representations to guide the summarization process for sensemaking? Therefore, we propose introducing an intermediate step--a schematic visual workspace for human sensemaking--before the LLM generation to steer and refine the summarization process. We conduct a series of proof-of-concept experiments to investigate the potential for enhancing the summarization by GPT-4 through visual workspaces. Leveraging a textual sensemaking dataset with a ground truth summary, we evaluate the impact of a human-generated visual workspace on LLM-generated summarization of the dataset and assess the effectiveness of space-steered summarization. We categorize several types of extractable information from typical human workspaces that can be injected into engineered prompts to steer the LLM summarization. The results demonstrate how such workspaces can help align an LLM with the ground truth, leading to more accurate summarization results than without the workspaces.
As large language models (LLMs) evolve to handle increasingly longer contexts, serving inference requests for context lengths in the range of millions of tokens presents unique challenges. While existing techniques are effective for training, they fail to address the unique challenges of inference, such as varying prefill and decode phases and their associated latency constraints - like Time to First Token (TTFT) and Time Between Tokens (TBT). Furthermore, there are no long context inference solutions that allow batching requests to increase the hardware utilization today. In this paper, we propose three key innovations for efficient interactive long context LLM inference, without resorting to any approximation: adaptive chunking to reduce prefill overheads in mixed batching, Sequence Pipeline Parallelism (SPP) to lower TTFT, and KV Cache Parallelism (KVP) to minimize TBT. These contributions are combined into a 3D parallelism strategy, enabling Mnemosyne to scale interactive inference to context lengths at least up to 10 million tokens with high throughput enabled with batching. To our knowledge, Mnemosyne is the first to be able to achieve support for 10 million long context inference efficiently, while satisfying production-grade SLOs on TBT (30ms) on contexts up to and including 10 million.
Multimodal large language models (MLLMs) have enabled LLM-based agents to directly interact with application user interfaces (UIs), enhancing agents' performance in complex tasks. However, these agents often suffer from high latency and low reliability due to the extensive sequential UI interactions. To address this issue, we propose AXIS, a novel LLM-based agents framework prioritize actions through application programming interfaces (APIs) over UI actions. This framework also facilitates the creation and expansion of APIs through automated exploration of applications. Our experiments on Office Word demonstrate that AXIS reduces task completion time by 65%-70% and cognitive workload by 38%-53%, while maintaining accuracy of 97%-98% compare to humans. Our work contributes to a new human-agent-computer interaction (HACI) framework and a fresh UI design principle for application providers in the era of LLMs. It also explores the possibility of turning every applications into agents, paving the way towards an agent-centric operating system (Agent OS).
One of the key issues contributing to inefficiency in Puskesmas is the time-consuming nature of doctor-patient interactions. Doctors need to conduct thorough consultations, which include diagnosing the patient's condition, providing treatment advice, and transcribing detailed notes into medical records. In regions with diverse linguistic backgrounds, doctors often have to ask clarifying questions, further prolonging the process. While diagnosing is essential, transcription and summarization can often be automated using AI to improve time efficiency and help doctors enhance care quality and enable early diagnosis and intervention. This paper proposes a solution using a localized large language model (LLM) to transcribe, translate, and summarize doctor-patient conversations. We utilize the Whisper model for transcription and GPT-3 to summarize them into the ePuskemas medical records format. This system is implemented as an add-on to an existing web browser extension, allowing doctors to fill out patient forms while talking. By leveraging this solution for real-time transcription, translation, and summarization, doctors can improve the turnaround time for patient care while enhancing the quality of records, which become more detailed and insightful for future visits. This innovation addresses challenges like overcrowded facilities and the administrative burden on healthcare providers in Indonesia. We believe this solution will help doctors save time, provide better care, and produce more accurate medical records, representing a significant step toward modernizing healthcare and ensuring patients receive timely, high-quality care, even in resource-constrained settings.
Large Language Models (LLMs) are widely used in both industry and academia for various tasks, yet evaluating the consistency of generated text responses continues to be a challenge. Traditional metrics like ROUGE and BLEU show a weak correlation with human judgment. More sophisticated metrics using Natural Language Inference (NLI) have shown improved correlations but are complex to implement, require domain-specific training due to poor cross-domain generalization, and lack explainability. More recently, prompt-based metrics using LLMs as evaluators have emerged; while they are easier to implement, they still lack explainability and depend on task-specific prompts, which limits their generalizability. This work introduces Automated eXplainable Consistency Evaluation using LLMs (AXCEL), a prompt-based consistency metric which offers explanations for the consistency scores by providing detailed reasoning and pinpointing inconsistent text spans. AXCEL is also a generalizable metric which can be adopted to multiple tasks without changing the prompt. AXCEL outperforms both non-prompt and prompt-based state-of-the-art (SOTA) metrics in detecting inconsistencies across summarization by 8.7%, free text generation by 6.2%, and data-to-text conversion tasks by 29.4%. We also evaluate the influence of underlying LLMs on prompt based metric performance and recalibrate the SOTA prompt-based metrics with the latest LLMs for fair comparison. Further, we show that AXCEL demonstrates strong performance using open source LLMs.
Large language models (LLMs) have revolutionized how we interact with technology, but their personalization to individual user preferences remains a significant challenge, particularly in on-device applications. Traditional methods often depend heavily on labeled datasets and can be resource-intensive. To address these issues, we present Adaptive Self-Supervised Learning Strategies (ASLS), which utilizes self-supervised learning techniques to personalize LLMs dynamically. The framework comprises a user profiling layer for collecting interaction data and a neural adaptation layer for real-time model fine-tuning. This innovative approach enables continuous learning from user feedback, allowing the model to generate responses that align closely with user-specific contexts. The adaptive mechanisms of ASLS minimize computational demands and enhance personalization efficiency. Experimental results across various user scenarios illustrate the superior performance of ASLS in boosting user engagement and satisfaction, highlighting its potential to redefine LLMs as highly responsive and context-aware systems on-device.
In this paper, we present an effective data augmentation framework leveraging the Large Language Model (LLM) and Diffusion Model (DM) to tackle the challenges inherent in data-scarce scenarios. Recently, DMs have opened up the possibility of generating synthetic images to complement a few training images. However, increasing the diversity of synthetic images also raises the risk of generating samples outside the target distribution. Our approach addresses this issue by embedding novel semantic information into text prompts via LLM and utilizing real images as visual prompts, thus generating semantically rich images. To ensure that the generated images remain within the target distribution, we dynamically adjust the guidance weight based on each image's CLIPScore to control the diversity. Experimental results show that our method produces synthetic images with enhanced diversity while maintaining adherence to the target distribution. Consequently, our approach proves to be more efficient in the few-shot setting on several benchmarks. Our code is available at https://github.com/kkyuhun94/dalda .
The increasing capability and widespread usage of large language models (LLMs) highlight the desirability of automatic detection of LLM-generated text. Zero-shot detectors, due to their training-free nature, have received considerable attention and notable success. In this paper, we identify a new feature, token cohesiveness, that is useful for zero-shot detection, and we demonstrate that LLM-generated text tends to exhibit higher token cohesiveness than human-written text. Based on this observation, we devise TOCSIN, a generic dual-channel detection paradigm that uses token cohesiveness as a plug-and-play module to improve existing zero-shot detectors. To calculate token cohesiveness, TOCSIN only requires a few rounds of random token deletion and semantic difference measurement, making it particularly suitable for a practical black-box setting where the source model used for generation is not accessible. Extensive experiments with four state-of-the-art base detectors on various datasets, source models, and evaluation settings demonstrate the effectiveness and generality of the proposed approach. Code available at: \url{https://github.com/Shixuan-Ma/TOCSIN}.
The fusion of Large Language Models (LLMs) and robotic systems has led to a transformative paradigm in the robotic field, offering unparalleled capabilities not only in the communication domain but also in skills like multimodal input handling, high-level reasoning, and plan generation. The grounding of LLMs knowledge into the empirical world has been considered a crucial pathway to exploit the efficiency of LLMs in robotics. Nevertheless, connecting LLMs' representations to the external world with multimodal approaches or with robots' bodies is not enough to let them understand the meaning of the language they are manipulating. Taking inspiration from humans, this work draws attention to three necessary elements for an agent to grasp and experience the world. The roadmap for LLMs grounding is envisaged in an active bodily system as the reference point for experiencing the environment, a temporally structured experience for a coherent, self-related interaction with the external world, and social skills to acquire a common-grounded shared experience.
When operating in human environments, robots need to handle complex tasks while both adhering to social norms and accommodating individual preferences. For instance, based on common sense knowledge, a household robot can predict that it should avoid vacuuming during a social gathering, but it may still be uncertain whether it should vacuum before or after having guests. In such cases, integrating common-sense knowledge with human preferences, often conveyed through human explanations, is fundamental yet a challenge for existing systems. In this paper, we introduce GRACE, a novel approach addressing this while generating socially appropriate robot actions. GRACE leverages common sense knowledge from Large Language Models (LLMs), and it integrates this knowledge with human explanations through a generative network architecture. The bidirectional structure of GRACE enables robots to refine and enhance LLM predictions by utilizing human explanations and makes robots capable of generating such explanations for human-specified actions. Our experimental evaluations show that integrating human explanations boosts GRACE's performance, where it outperforms several baselines and provides sensible explanations.
The move toward Sixth-Generation (6G) networks relies on open interfaces and protocols for seamless interoperability across devices, vendors, and technologies. In this context, open 6G development involves multiple disciplines and requires advanced simulation approaches for testing. In this demo paper, we propose a generative simulation approach based on a multi-agent Large Language Model (LLM) and Network Simulator 3 (ns-3), called Generative Open xG Network Simulation (GenOnet), to effectively generate, debug, execute, and interpret simulated Open Fifth-Generation (5G) environments. The first version of GenOnet application represents a specialized adaptation of the OpenAI GPT models. It incorporates supplementary tools, agents, 5G standards, and seamless integration with ns-3 simulation capabilities, supporting both C++ variants and Python implementations. This release complies with the latest Open Radio Access Network (O-RAN) and 3GPP standards.
Large Language Models (LLMs) create exciting possibilities for powerful language processing tools to accelerate research in materials science. While LLMs have great potential to accelerate materials understanding and discovery, they currently fall short in being practical materials science tools. In this position paper, we show relevant failure cases of LLMs in materials science that reveal current limitations of LLMs related to comprehending and reasoning over complex, interconnected materials science knowledge. Given those shortcomings, we outline a framework for developing Materials Science LLMs (MatSci-LLMs) that are grounded in materials science knowledge and hypothesis generation followed by hypothesis testing. The path to attaining performant MatSci-LLMs rests in large part on building high-quality, multi-modal datasets sourced from scientific literature where various information extraction challenges persist. As such, we describe key materials science information extraction challenges which need to be overcome in order to build large-scale, multi-modal datasets that capture valuable materials science knowledge. Finally, we outline a roadmap for applying future MatSci-LLMs for real-world materials discovery via: 1. Automated Knowledge Base Generation; 2. Automated In-Silico Material Design; and 3. MatSci-LLM Integrated Self-Driving Materials Laboratories.
Peer review is an essential process to determine the quality of papers submitted to scientific conferences or journals. However, it is subjective and prone to biases. Several studies have been conducted to apply techniques from NLP to support peer review, but they are based on black-box techniques and their outputs are difficult to interpret and trust. In this paper, we propose a novel pipeline to support and understand the reviewing and decision-making processes of peer review: the PeerArg system combining LLMs with methods from knowledge representation. PeerArg takes in input a set of reviews for a paper and outputs the paper acceptance prediction. We evaluate the performance of the PeerArg pipeline on three different datasets, in comparison with a novel end-2-end LLM that uses few-shot learning to predict paper acceptance given reviews. The results indicate that the end-2-end LLM is capable of predicting paper acceptance from reviews, but a variant of the PeerArg pipeline outperforms this LLM.
Test smells arise from poor design practices and insufficient domain knowledge, which can lower the quality of test code and make it harder to maintain and update. Manually refactoring test smells is time-consuming and error-prone, highlighting the necessity for automated approaches. Current rule-based refactoring methods often struggle in scenarios not covered by predefined rules and lack the flexibility needed to handle diverse cases effectively. In this paper, we propose a novel approach called UTRefactor, a context-enhanced, LLM-based framework for automatic test refactoring in Java projects. UTRefactor extracts relevant context from test code and leverages an external knowledge base that includes test smell definitions, descriptions, and DSL-based refactoring rules. By simulating the manual refactoring process through a chain-of-thought approach, UTRefactor guides the LLM to eliminate test smells in a step-by-step process, ensuring both accuracy and consistency throughout the refactoring. Additionally, we implement a checkpoint mechanism to facilitate comprehensive refactoring, particularly when multiple smells are present. We evaluate UTRefactor on 879 tests from six open-source Java projects, reducing the number of test smells from 2,375 to 265, achieving an 89% reduction. UTRefactor outperforms direct LLM-based refactoring methods by 61.82% in smell elimination and significantly surpasses the performance of a rule-based test smell refactoring tool. Our results demonstrate the effectiveness of UTRefactor in enhancing test code quality while minimizing manual involvement.
The profound impact of food on health necessitates advanced nutrition-oriented food recommendation services. Conventional methods often lack the crucial elements of personalization, explainability, and interactivity. While Large Language Models (LLMs) bring interpretability and explainability, their standalone use falls short of achieving true personalization. In this paper, we introduce ChatDiet, a novel LLM-powered framework designed specifically for personalized nutrition-oriented food recommendation chatbots. ChatDiet integrates personal and population models, complemented by an orchestrator, to seamlessly retrieve and process pertinent information. The personal model leverages causal discovery and inference techniques to assess personalized nutritional effects for a specific user, whereas the population model provides generalized information on food nutritional content. The orchestrator retrieves, synergizes and delivers the output of both models to the LLM, providing tailored food recommendations designed to support targeted health outcomes. The result is a dynamic delivery of personalized and explainable food recommendations, tailored to individual user preferences. Our evaluation of ChatDiet includes a compelling case study, where we establish a causal personal model to estimate individual nutrition effects. Our assessments, including a food recommendation test showcasing a 92\% effectiveness rate, coupled with illustrative dialogue examples, underscore ChatDiet's strengths in explainability, personalization, and interactivity.
While many online services provide privacy policies for end users to read and understand what personal data are being collected, these documents are often lengthy and complicated. As a result, the vast majority of users do not read them at all, leading to data collection under uninformed consent. Several attempts have been made to make privacy policies more user friendly by summarising them, providing automatic annotations or labels for key sections, or by offering chat interfaces to ask specific questions. With recent advances in Large Language Models (LLMs), there is an opportunity to develop more effective tools to parse privacy policies and help users make informed decisions. In this paper, we propose an entailment-driven LLM based framework to classify paragraphs of privacy policies into meaningful labels that are easily understood by users. The results demonstrate that our framework outperforms traditional LLM methods, improving the F1 score in average by 11.2%. Additionally, our framework provides inherently explainable and meaningful predictions.
Conversational Health Agents (CHAs) are interactive systems that provide healthcare services, such as assistance and diagnosis. Current CHAs, especially those utilizing Large Language Models (LLMs), primarily focus on conversation aspects. However, they offer limited agent capabilities, specifically lacking multi-step problem-solving, personalized conversations, and multimodal data analysis. Our aim is to overcome these limitations. We propose openCHA, an open-source LLM-powered framework, to empower conversational agents to generate a personalized response for users' healthcare queries. This framework enables developers to integrate external sources including data sources, knowledge bases, and analysis models, into their LLM-based solutions. openCHA includes an orchestrator to plan and execute actions for gathering information from external sources, essential for formulating responses to user inquiries. It facilitates knowledge acquisition, problem-solving capabilities, multilingual and multimodal conversations, and fosters interaction with various AI platforms. We illustrate the framework's proficiency in handling complex healthcare tasks via two demonstrations and four use cases. Moreover, we release openCHA as open source available to the community via GitHub.
Despite significant progress in applying large language models (LLMs) to the medical domain, several limitations still prevent them from practical applications. Among these are the constraints on model size and the lack of cohort-specific labeled datasets. In this work, we investigated the potential of improving a lightweight LLM, such as Llama 3.1-8B, through fine-tuning with datasets using synthetic labels. Two tasks are jointly trained by combining their respective instruction datasets. When the quality of the task-specific synthetic labels is relatively high (e.g., generated by GPT4- o), Llama 3.1-8B achieves satisfactory performance on the open-ended disease detection task, with a micro F1 score of 0.91. Conversely, when the quality of the task-relevant synthetic labels is relatively low (e.g., from the MIMIC-CXR dataset), fine-tuned Llama 3.1-8B is able to surpass its noisy teacher labels (micro F1 score of 0.67 v.s. 0.63) when calibrated against curated labels, indicating the strong inherent underlying capability of the model. These findings demonstrate the potential of fine-tuning LLMs with synthetic labels, offering a promising direction for future research on LLM specialization in the medical domain.
Large language models (LLMs) have shown outstanding performance across numerous real-world tasks. However, the autoregressive nature of these models makes the inference process slow and costly. Speculative decoding has emerged as a promising solution, leveraging a smaller auxiliary model to draft future tokens, which are then validated simultaneously by the larger model, achieving a speed-up of 1-2x. Although speculative decoding matches the same distribution as multinomial sampling, multinomial sampling itself is prone to suboptimal outputs, whereas beam sampling is widely recognized for producing higher-quality results by maintaining multiple candidate sequences at each step. This paper explores the novel integration of speculative decoding with beam sampling. However, there are four key challenges: (1) how to generate multiple sequences from the larger model's distribution given drafts sequences from the small model; (2) how to dynamically optimize the number of beams to balance efficiency and accuracy; (3) how to efficiently verify the multiple drafts in parallel; and (4) how to address the extra memory costs inherent in beam sampling. To address these challenges, we propose dynamic-width speculative beam decoding (DSBD). Specifically, we first introduce a novel draft and verification scheme that generates multiple sequences following the large model's distribution based on beam sampling trajectories from the small model. Then, we introduce an adaptive mechanism to dynamically tune the number of beams based on the context, optimizing efficiency and effectiveness. Besides, we extend tree-based parallel verification to handle multiple trees simultaneously, accelerating the verification process. Finally, we illustrate a simple modification to our algorithm to mitigate the memory overhead of beam sampling...
With the advancements of Large Language Models (LLMs), an increasing number of open-source software projects are using LLMs as their core functional component. Although research and practice on LLMs are capturing considerable interest, no dedicated studies explored the challenges faced by practitioners of LLM open-source projects, the causes of these challenges, and potential solutions. To fill this research gap, we conducted an empirical study to understand the issues that practitioners encounter when developing and using LLM open-source software, the possible causes of these issues, and potential solutions.We collected all closed issues from 15 LLM open-source projects and labelled issues that met our requirements. We then randomly selected 994 issues from the labelled issues as the sample for data extraction and analysis to understand the prevalent issues, their underlying causes, and potential solutions. Our study results show that (1) Model Issue is the most common issue faced by practitioners, (2) Model Problem, Configuration and Connection Problem, and Feature and Method Problem are identified as the most frequent causes of the issues, and (3) Optimize Model is the predominant solution to the issues. Based on the study results, we provide implications for practitioners and researchers of LLM open-source projects.
This study introduces a comprehensive benchmark designed to evaluate the performance of large language models (LLMs) in understanding and processing cultural knowledge, with a specific focus on Hakka culture as a case study. Leveraging Bloom's Taxonomy, the study develops a multi-dimensional framework that systematically assesses LLMs across six cognitive domains: Remembering, Understanding, Applying, Analyzing, Evaluating, and Creating. This benchmark extends beyond traditional single-dimensional evaluations by providing a deeper analysis of LLMs' abilities to handle culturally specific content, ranging from basic recall of facts to higher-order cognitive tasks such as creative synthesis. Additionally, the study integrates Retrieval-Augmented Generation (RAG) technology to address the challenges of minority cultural knowledge representation in LLMs, demonstrating how RAG enhances the models' performance by dynamically incorporating relevant external information. The results highlight the effectiveness of RAG in improving accuracy across all cognitive domains, particularly in tasks requiring precise retrieval and application of cultural knowledge. However, the findings also reveal the limitations of RAG in creative tasks, underscoring the need for further optimization. This benchmark provides a robust tool for evaluating and comparing LLMs in culturally diverse contexts, offering valuable insights for future research and development in AI-driven cultural knowledge preservation and dissemination.
The reliance of popular programming languages such as Python and JavaScript on centralized package repositories and open-source software, combined with the emergence of code-generating Large Language Models (LLMs), has created a new type of threat to the software supply chain: package hallucinations. These hallucinations, which arise from fact-conflicting errors when generating code using LLMs, represent a novel form of package confusion attack that poses a critical threat to the integrity of the software supply chain. This paper conducts a rigorous and comprehensive evaluation of package hallucinations across different programming languages, settings, and parameters, exploring how a diverse set of models and configurations affect the likelihood of generating erroneous package recommendations and identifying the root causes of this phenomenon. Using 16 popular LLMs for code generation and two unique prompt datasets, we generate 576,000 code samples in two programming languages that we analyze for package hallucinations. Our findings reveal that that the average percentage of hallucinated packages is at least 5.2% for commercial models and 21.7% for open-source models, including a staggering 205,474 unique examples of hallucinated package names, further underscoring the severity and pervasiveness of this threat. To overcome this problem, we implement several hallucination mitigation strategies and show that they are able to significantly reduce the number of package hallucinations while maintaining code quality. Our experiments and findings highlight package hallucinations as a persistent and systemic phenomenon while using state-of-the-art LLMs for code generation, and a significant challenge which deserves the research community's urgent attention.
Logical reasoning is a fundamental task in natural language processing that presents significant challenges to Large Language Models (LLMs). The inherent characteristics of logical reasoning makes it well-suited for symbolic representations such as first-order logic (FOL). Research in symbolic logical reasoning explored FOL generation using state-of-the-art LLMs (i.e., GPT-4) to produce FOL translations of natural language (NL) statements, but errors in translation are usually not the focus. We address this by categorizing the translation errors in FOL statements generated by LLMs. To make progress towards improving the quality of FOL translations for smaller language models such as LLaMA-2 13B and Mistral 7B, we create ProofFOL, a high-quality FOL-annotated subset of ProofWriter dataset using GPT-4o. The models fine-tuned on this silver standard data achieve a significant gain in performance when compared to larger language models such as LLaMA-2 70B. In addition to improving the model using large data, we also tackle the issue of data scarcity and introduce an incremental framework encompassing of data augmentation and verification steps. In the augmentation process, a single pair of (premises, conclusion) is split into multiple new instances based on the predicates and FOLs. This data is used for fine-tuning, and the inference on this model generates FOLs with fewer errors over the model trained on the original data. Our investigation on the translation errors leads to generation of a perturbation dataset, which is used to train a verifier that corrects potential syntactic and semantic FOL translation errors. We demonstrate an efficient method for making the most of a limited existing human-annotated dataset. Our results show state-of-the-art performance for ProofWriter and ProntoQA datasets using ProofFOL on LLaMA-2 and Mistral models.
LLMs have shown promising results in task planning due to their strong natural language understanding and reasoning capabilities. However, issues such as hallucinations, ambiguities in human instructions, environmental constraints, and limitations in the executing agent's capabilities often lead to flawed or incomplete plans. This paper proposes MultiTalk, an LLM-based task planning methodology that addresses these issues through a framework of introspective and extrospective dialogue loops. This approach helps ground generated plans in the context of the environment and the agent's capabilities, while also resolving uncertainties and ambiguities in the given task. These loops are enabled by specialized systems designed to extract and predict task-specific states, and flag mismatches or misalignments among the human user, the LLM agent, and the environment. Effective feedback pathways between these systems and the LLM planner foster meaningful dialogue. The efficacy of this methodology is demonstrated through its application to robotic manipulation tasks. Experiments and ablations highlight the robustness and reliability of our method, and comparisons with baselines further illustrate the superiority of MultiTalk in task planning for embodied agents.
Large Language Models(LLMs) have revolutionized various applications in natural language processing (NLP) by providing unprecedented text generation, translation, and comprehension capabilities. However, their widespread deployment has brought to light significant concerns regarding biases embedded within these models. This paper presents a comprehensive survey of biases in LLMs, aiming to provide an extensive review of the types, sources, impacts, and mitigation strategies related to these biases. We systematically categorize biases into several dimensions. Our survey synthesizes current research findings and discusses the implications of biases in real-world applications. Additionally, we critically assess existing bias mitigation techniques and propose future research directions to enhance fairness and equity in LLMs. This survey serves as a foundational resource for researchers, practitioners, and policymakers concerned with addressing and understanding biases in LLMs.
Medication errors significantly threaten patient safety, leading to adverse drug events and substantial economic burdens on healthcare systems. Clinical Decision Support Systems (CDSSs) aimed at mitigating these errors often face limitations, including reliance on static databases and rule-based algorithms, which can result in high false alert rates and alert fatigue among clinicians. This paper introduces HELIOT, an innovative CDSS for drug allergy management, integrating Large Language Models (LLMs) with a comprehensive pharmaceutical data repository. HELIOT leverages advanced natural language processing capabilities to interpret complex medical texts and synthesize unstructured data, overcoming the limitations of traditional CDSSs. An empirical evaluation using a synthetic patient dataset and expert-verified ground truth demonstrates HELIOT's high accuracy, precision, recall, and F1 score, uniformly reaching 100\% across multiple experimental runs. The results underscore HELIOT's potential to enhance decision support in clinical settings, offering a scalable, efficient, and reliable solution for managing drug allergies.
Large Language Models (LLMs) have the promise to revolutionize computing broadly, but their complexity and extensive training data also expose significant privacy vulnerabilities. One of the simplest privacy risks associated with LLMs is their susceptibility to membership inference attacks (MIAs), wherein an adversary aims to determine whether a specific data point was part of the model's training set. Although this is a known risk, state of the art methodologies for MIAs rely on training multiple computationally costly shadow models, making risk evaluation prohibitive for large models. Here we adapt a recent line of work which uses quantile regression to mount membership inference attacks; we extend this work by proposing a low-cost MIA that leverages an ensemble of small quantile regression models to determine if a document belongs to the model's training set or not. We demonstrate the effectiveness of this approach on fine-tuned LLMs of varying families (OPT, Pythia, Llama) and across multiple datasets. Across all scenarios we obtain comparable or improved accuracy compared to state of the art shadow model approaches, with as little as 6% of their computation budget. We demonstrate increased effectiveness across multi-epoch trained target models, and architecture miss-specification robustness, that is, we can mount an effective attack against a model using a different tokenizer and architecture, without requiring knowledge on the target model.
Multi-human multi-robot teams combine the complementary strengths of humans and robots to tackle complex tasks across diverse applications. However, the inherent heterogeneity of these teams presents significant challenges in initial task allocation (ITA), which involves assigning the most suitable tasks to each team member based on their individual capabilities before task execution. While current learning-based methods have shown promising results, they are often computationally expensive to train, and lack the flexibility to incorporate user preferences in multi-objective optimization and adapt to last-minute changes in real-world dynamic environments. To address these issues, we propose REBEL, an LLM-based ITA framework that integrates rule-based and experience-enhanced learning. By leveraging Retrieval-Augmented Generation, REBEL dynamically retrieves relevant rules and past experiences, enhancing reasoning efficiency. Additionally, REBEL can complement pre-trained RL-based ITA policies, improving situational awareness and overall team performance. Extensive experiments validate the effectiveness of our approach across various settings. More details are available at https://sites.google.com/view/ita-rebel .
In this paper we explore evaluation of LLM capabilities. We present measurements of GPT-4 performance on several deterministic tasks; each task involves a basic calculation and takes as input parameter some element drawn from a large well-defined population (e.g., count elements in a list, multiply two k-digit numbers, etc). We examine several conditions per-task and perform enough trials so that statistically significant differences can be detected. This allows us to investigate the sensitivity of task-accuracy both to query phrasing and input parameter population. We find that seemingly trivial modifications in the task-prompt or input population can yield differences far larger than can be explained by sampling effects. For example, performance on a simple list-counting task varies with query-phrasing and list-length, but also with list composition (i.e., the thing-to-be-counted) and object frequency (e.g., success when an element accounts for $\approx$ 50\% of a list is different from when it accounts for $\approx$ 70\% etc). We conclude that efforts to quantify LLM capabilities easily succumb to the language-as-fixed-effect fallacy, where experimental observations are improperly generalized beyond what the data supports. A consequence appears to be that intuitions that have been formed based on interactions with humans form a very unreliable guide as to which input modifications should ``make no difference'' to LLM performance.
Modern software systems like the Linux kernel are among the world's largest and most intricate codebases, continually evolving with new features and increasing complexity. Understanding these systems poses significant challenges due to their scale and the unstructured nature of development artifacts such as commits and mailing list discussions. We introduce Code-Survey, the first LLM-driven methodology designed to systematically explore and analyze large-scale codebases. The central principle behind Code-Survey is to treat LLMs as human participants, acknowledging that software development is also a social activity and thereby enabling the application of established social science techniques. By carefully designing surveys, Code-Survey transforms unstructured data, such as commits, emails, into organized, structured, and analyzable datasets. This enables quantitative analysis of complex software evolution and uncovers valuable insights related to design, implementation, maintenance, reliability, and security. To demonstrate the effectiveness of Code-Survey, we apply it to the Linux kernel's eBPF subsystem. We construct the Linux-bpf dataset, comprising over 670 features and 16,000 commits from the Linux community. Our quantitative analysis uncovers important insights into the evolution of eBPF, such as development patterns, feature interdependencies, and areas requiring attention for reliability and security. The insights have been initially validated by eBPF experts. Furthermore, Code-Survey can be directly applied to other subsystems within Linux and to other large-scale software projects. By providing a versatile tool for systematic analysis, Code-Survey facilitates a deeper understanding of complex software systems, enabling improvements across a variety of domains and supporting a wide range of empirical studies. The code and dataset is open-sourced.
Recent advancements have showcased the capabilities of Large Language Models like GPT4 and Llama2 in tasks such as summarization, translation, and content review. However, their widespread use raises concerns, particularly around the potential for LLMs to spread persuasive, humanlike misinformation at scale, which could significantly influence public opinion. This study examines these risks, focusing on LLMs ability to propagate misinformation as factual. To investigate this, we built the LLM Echo Chamber, a controlled digital environment simulating social media chatrooms, where misinformation often spreads. Echo chambers, where individuals only interact with like minded people, further entrench beliefs. By studying malicious bots spreading misinformation in this environment, we can better understand this phenomenon. We reviewed current LLMs, explored misinformation risks, and applied sota finetuning techniques. Using Microsoft phi2 model, finetuned with our custom dataset, we generated harmful content to create the Echo Chamber. This setup, evaluated by GPT4 for persuasiveness and harmfulness, sheds light on the ethical concerns surrounding LLMs and emphasizes the need for stronger safeguards against misinformation.
Large language models are transforming the creative process by offering unprecedented capabilities to algorithmically generate ideas. While these tools can enhance human creativity when people co-create with them, it's unclear how this will impact unassisted human creativity. We conducted two large pre-registered parallel experiments involving 1,100 participants attempting tasks targeting the two core components of creativity, divergent and convergent thinking. We compare the effects of two forms of large language model (LLM) assistance -- a standard LLM providing direct answers and a coach-like LLM offering guidance -- with a control group receiving no AI assistance, and focus particularly on how all groups perform in a final, unassisted stage. Our findings reveal that while LLM assistance can provide short-term boosts in creativity during assisted tasks, it may inadvertently hinder independent creative performance when users work without assistance, raising concerns about the long-term impact on human creativity and cognition.